Frequent Whole-Genome Duplication Events Drive the Genomic Evolution of Triple-Negative Breast Cancer During Neoadjuvant Chemotherapy

Frequent Whole-Genome Duplication Events Drive the Genomic Evolution of Triple-Negative Breast Cancer During Neoadjuvant Chemotherapy | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Frequent Whole-Genome Duplication Events Drive the Genomic Evolution of Triple-Negative Breast Cancer During Neoadjuvant Chemotherapy Devin P Bendixsen, Fiona Semple, Alastair Ironside, Mustafa İsmail Özkaraca, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9212020/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 11 You are reading this latest preprint version Abstract Background Triple-negative breast cancer (TNBC) is associated with poor survival rate and high genomic instability, generating complex tumour genomes. However, the processes that generate this complexity are poorly studied in longitudinal samples. Here, we study the temporal dynamics of TNBC somatic mutations, revealing major transitions in tumour genome evolution, from diagnostic biopsies, through treatment, to cancer remission or recurrence. Methods Deep whole exome sequencing and CUTseq, a reduced representation whole genome sequencing approach, were performed in parallel, to comprehensively identify short nucleotide variants (SNVs), copy number alterations (CNAs) and aneuploidies. Tumour samples (N=74) from 22 patients were profiled before and after neoadjuvant chemotherapy (NACT), and encompassed spatially diverse samples from multiple primary breast tumours, to allow tracking of the gain and loss of candidate driver variants over time. Results Genome-wide SNV mutational burden remained stable across disease progression and RCB classes. However, recurrent SNVs were identified in several known TNBC driver genes in response to treatment, with TP53, MICA, CYP2D6, BRCA1, and BRCA2 being frequently altered. The candidate driver variants in these genes frequently exhibited dynamic changes throughout the course of a patient’s treatment, with the original SNVs in pre-treatment samples often lost, while novel variants in the same genes emerged at subsequent time points. In contrast to the stable genome-wide SNV burdens, dramatic changes in chromosome structure were seen in all tumours, with abundant CNAs and chromosome arm aneuploidies seen in pre-treatment samples, followed by frequent loss of these alterations post-treatment, and their re-emergence at recurrence. Whole genome duplication (WGD) events appear to drive these dynamics, with a higher frequency of pre-treatment WGD seen in patients with the best response to NACT. WGD also frequently appears late in the disease at recurrence, suggesting potential roles in the evolution resistance. Conclusions Comprehensive longitudinal profiling of the TNBC genome demonstrates the complex interplay of SNVs and structural alterations during tumour progression, leading to diverse evolutionary trajectories impacting patient outcomes. Complex mutational patterns encompassing entire chromosomes emerge during progression, with WGD events making major contributions to intra-tumour heterogeneity, and emerging as a potential candidate biomarker of response at both tumour establishment and recurrence. Triple negative breast cancer tumour evolution copy number variants aneuploidy whole genome duplication neoadjuvant chemotherapy Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Background Triple-negative breast cancer (TNBC) accounts for 15–20% of breast cancer cases, and is associated with a high rate of relapse and the worst five-year survival rate of all breast cancer subtypes. For many years, targeted therapeutic options for TNBC patients were limited as TNBCs lack expression of the oestrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2), which are essential targets for endocrine and biological therapies. However, more recently TNBC treatments have evolved significantly to include the use of immunotherapy, antibody-drug conjugates and PARP inhibitors combined with traditional chemotherapy, thereby increasing the treatment options beyond targeted treagtments. Collectively, TNBCs are characterised by high levels of genetic instability resulting in complex patterns of copy number alterations (CNAs) and structural rearrangements [ 1 ]. Whole genome duplication (WGD) has been shown to be an independent predictor for poor survival in many cancers including estrogen receptor-positive breast cancers [ 2 ]. Although WGD has been shown to occur in TNBC [ 3 ], the consequences for tumour evolution and patient outcomes are essentially unstudied. Increased tumour infiltrating lymphocytes (TILs), higher PD-L1 (programmed death ligand 1) expression levels and deficiency in homologous recombination (HRD) are more commonly seen in TNBC compared with other breast cancer subtypes [ 4 ]. Differences in molecular subtypes, immune composition and mutational landscapes, underlie the heterogeneity of TNBC. Four TNBC subtypes with unique biology and transcriptional diversity have been identified; basal-like (BL), mesenchymal (MES), luminal androgen receptor (LAR) and immunomodulatory (IM) [ 5 – 8 ]. Each subtype varies in immune composition with improved outcomes reported for TNBC patients with increased expression of immune gene signatures and increased TILs [ 9 – 11 ]. For the TNBC patients with known germline BRCA1/2 mutations (10%), additional treatment with poly ADP ribose polymerase (PARP) inhibitors can be beneficial [ 12 ]. Immunotherapy targeting immune checkpoint inhibitors (ICIs) has revolutionised the treatment of TNBC patients, but response is suboptimal in some patients, therefore the search for further biomarkers of response beyond PD-L1 is ongoing [ 13 ]. TNBC patients often respond well to neoadjuvant chemotherapy (NACT), with 34% of patients achieving pathological complete response (pCR) which is subsequently associated with 75% lower risk of recurrence, compared to TNBC patients not achieving pCR [ 14 , 15 ]. The recent Keynote-522 trial demonstrated that combining the ICI Pembrolizumab with conventional chemotherapy significantly increased pCR rates with 64.8% of patients acheiving pCR in the experimental arm compared to 51.2% in the placebo arm. Furthermore, the addition of Pembrolizumab, increased event-free survival rates to 84.5% compared to 76.8% in the control arm, with significant effects on overall survival [ 16 , 17 ]. High pCR rates have been associated with higher tumour mutational burden (TMB), elevated levels of PD-L1 expression and increased levels of TILs in the tumour microenvironment [ 18 ]. Defining the landscape of TNBC and the molecular changes that occur in response to NACT, will allow stratification of patients for directed treatment. However, this is confounded by the high genomic instability and extensive intra-tumour heterogeneity (ITH) that are thought to drive the pathogenesis, treatment resistance and metastatic spread of TNBC [ 19 ]. When and how these genetic alterations emerge is not known, but WGD is thought to be common in TNBC [ 3 , 20 ], while aneuploidies and copy number alterations (CNAs) have been implicated as drivers of the adaptive response to chemotherapy and metastasis [ 21 , 22 ]. Studies including multiple samples over time have the potential to reveal new tumour biology, but can also entail large investments of time, effort and resources. There is consequently a pressing need for accessible, affordable assays with relatively modest requirements for computationally intensive analysis. For example, although the mutational landscape of TNBC has been examined in previous studies (e.g. Lehmann et al, 2021), the temporal dynamics of CNA and short nucleotide variants (SNVs including single nucleotide substitutions and indels) and WGD during disease progression are poorly studied. Consequently, we lack a detailed account of tumour evolution as patients are diagnosed, undergo NACT, and their disease enters remission or recurs. Here, we employ deep (~ 100X) Whole Exome Sequencing (WES) to assay recurrent SNVs and CNAs in patient TNBC samples to identify variants that play a role in disease progression and response to treatment. In parallel, we use CUTseq [ 23 ], a reduced representation Whole Genome Sequencing (WGS) approach, to comprehensively identify copy number alterations genome-wide. In addition to recurrent SNVs in known TNBC driver genes we demonstrate major shifts in CNAs throughout the genome following NACT, suggesting new biomarkers of treatment response and revealing critical roles for WGD in tumour evolution. Methods Patient and Tumour samples Samples from TNBC patients treated at the Western General Hospital Cancer Centre, enrolled in the NEO and metcfDNA studies (references: 2014/0005 and SR680) at Edinburgh University were acquired through the NRS BioResource and Tissue Governance Unit. Patients with TNBC were determined from pathology records and only samples with matched tumour and blood samples were included. One sample with no matched blood was paired with the patient’s normal breast sample. Collection of human tissues and blood and relevant clinicopathological data were approved by the South East Scotland Research Ethics Committee 01 with REC references: 13/SS/0236 and 15/ES/0094, and all patients consented to sample collection and use of their data in research. Patients and their clinical data (Table S1 ) were anonymised. All patients underwent NACT with anthracyclines followed by taxanes in combination with Carboplatin, and response was assessed using the internationally recognised residual cancer burden (RCB) scoring index. This system provides a continuous score (RCB score) and categorises patients into four discrete groups (RCB 0, RCB I, RCB II, RCB III) based on the amount of residual disease following NACT. RCB-0 corresponds to no residual disease, also known as pathological complete response (pCR), RCB-1 denotes a small amount of residual invasive cancer, RCB-II a moderate amount of residual invasive cancer and RCB-III indicates substantial residual disease [ 24 ]. Samples from before treatment with NACT (PreT), at the end of NACT after surgery (PostT), where available from recurrent tumours (R) and from selected mid NACT biopsies (at completion of anthacyclines before commencing taxanes/carboplatin) (MidT) were collected. Tumour infiltrating lymphocytes (TILs) were assessed by a breast cancer consultant pathologist (AI) using light microscopy on routine diagnostic H&E-stained glass slides. The number of TILs were quantified as a percentage of the tumour-associated stroma occupied by lymphocytes, as per the recommendations of The International Immuno-Oncology Biomarker Working Group on Breast Cancer. Three distinct regions of predefined size for each tumour were evaluated for TILs using a visual reference image index of pre-determined densities [ 25 , 26 ]. DNA extraction Formalin fixed paraffin embedded (FFPE) samples were sectioned onto slides using a cryostat. An initial 4µm section was stained with hematoxylin and eosin (H&E) and reviewed by a breast cancer consultant pathologist (AI) to determine sample tumour cellularity and mark up the tumour regions (Table S2). DNA was extracted (using Qiagen QIAamp DNA FFPE Advanced UNG Kit #56704) from 10µm sections, 2 to 5 sections per sample depending on tumour cellularity. Extraction was carried out on marked up tumour enriched areas and where possible, multiregional sampling was carried out by extracting DNA from spatially distinct tumour enriched areas on each slide, or from slides prepared from different FFPE blocks from the same tumour sample. Germline DNA was extracted from whole blood collected before commencement of NAC using the Qiagen DNeasy Blood & Tissue Kit (#69504). DNA was quantified using the Qubit dsDNA HS Assay Kit (#Q32851). Whole Exome Sequencing The quality of the DNA extracted from the FFPE samples was assessed with the Illumina FFPE QC Kit (Illumina Inc, #15013664) according to the manufacturer’s protocol. The required ΔCq ≤ 5 was achieved by all samples determining suitability for library preparation. For each sample, 50ng of DNA sample was used for library preparation with the Illumina DNA Prep with Enrichment, (S) Tagmentation Kit (Illumina #20025524) using IDT® for Illumina® DNA/RNA UD Indexes Set A, Tagmentation (Illumina #20027213) according to the manufacturer's protocol with modifications for FFPE DNA where appropriate. Using the Twist Comprehensive Exome panel (Twist Biosciences #102032), 500ng of each pre-enrichment gDNA library was pooled in sets of 12 and hybridised to the targeted regions during an overnight incubation, followed by capture onto magnetic beads and amplification of the exome enriched library pools, which were then purified with Agencourt AMPure XP beads (Beckman Coulter, #A36881). Libraries were quantified using Qubit dsDNA HS assay and quality assessed using Agilent Bioanalysis with the DNA HS Kit (#5067 − 4626). Fragment size and quantity measurements were used to calculate molarity for each enrichment library pool. Sequencing was performed on the NextSeq 2000 platform (Illumina Inc, #20038897) using NextSeq 2000 P3 Reagents (200 Cycles) (#20040560). A single pool containing the 8 enrichment pools was sequenced on 3 separate P3 flow cells to achieve a minimum of 100X coverage. PhiX Control v3 (Illumina, # FC-110- 3001) was spiked into each run at a concentration of 1%. Library preparation and sequencing were carried out at the Clinical Research Facility (CRF), Edinburgh, UK. Reduced Representation Whole Genome Sequencing – CUTseq Multiplexed DNA sequencing libraries were generated using CUTseq [ 23 ] following the published protocol ( https://protocolexchange.researchsquare.com/article/pex-25/v3 ) with several modifications. Firstly, adapters (A1-A48) specific to the HindIII cut site which incorporate sample-specific barcodes, the RA5 Illumina sequencing adapter and the T7 promoter sequence were purchased from Integrated DNA Technologies (IDT) (Table S3). Briefly, forward strand barcode adapters were 5’ phosphorylated then annealed to the reverse strand barcode adapters by incubation at 95°C with step down cooling to 25°C over the course of 45min to form double-stranded(ds)-barcoded HindIII-specific adapters. DNA (10-50ng) from patient samples was fragmented by enzymatic digestion with HindIII-HF® (NEB, # R3104L) for 4hr at 37°C in a 10µl volume of CutSmart® Buffer (NEB, #B6004). Generally, 48 samples were prepared at the same time. Digested DNA samples were barcoded using T4 DNA ligase (NEB, # M0202T) by incubating with ds-barcoded HindIII adapters at 16°C for 18hr. Samples were combined into 3 pools of 16 samples, each containing ~ 300-500ng starting DNA, and purified using Agencourt AMPure XP (Beckman Coulter, #A63880) at a ratio of 0.9X beads to sample. The 3 pools were combined and fragmented with target peak size 200bp using the Covaris E220 Evolution. Sample concentration was determined using Agilent Technologies DNA HS Kit (#5067 − 4626), followed by evaporation to ~ 20ng/µl. In vitro transcription was carried out with 8µl of sample (~ 160ng) using the MEGAscript® T7 Transcription Kit (ThermoFisher Scientific, #AM1334-5) following the manufacturer’s protocol, and the resulting RNA was purified using Agencourt RNAClean XP beads (Beckman Coulter, #A63987) at a ratio of 1.8X beads to sample. For RNA sequencing library preparation, primers (RTP, RP1 and RPIndex) and adapter (RA3) were purchased as standard desalted oligos from IDT (Table S3). RNA was quantified using the NanoDrop Spectrophotometer (Thermo Scientific) subsequently 1µg of RNA was used for library prep. Briefly, the RA3 adapter was ligated using T4 RNA ligase 2, deletion mutant (Cambio Ltd, #LR2D1132K) at 25°C for 2hr, followed by reverse transcription using the RTP primer and SuperScript IV RT (Invitrogen, #18090050) at 50°C for 1hr. Each library was then amplified using NEBNext® UltraTM II PCR master mix (#M0544L) and indexed using Illumina primers (RPI2 and RPI3) to allow simultaneous sequencing of two 48 sample libraries. After double-sided size selection purification using AMPure XP beads, final library quantification and size distribution were carried out by Qubit dsDNA HS assay and Agilent Bioanalysis (DNA HS Kit, #5067 − 4626). Sequencing (1x131bp) was performed on the NextSeq 2000 platform (Illumina Inc, #SY- 415–1002) at the CRF, Edinburgh, using the NextSeq 1000/2000 P3 Reagents (100 cycles) v3 Kit (#20040599) with PhiX Control v3 (Illumina, #FC-110-3001) spiked in at a concentration of 5%. Primary processing of WES and CUTseq sequencing data The raw sequencing reads for the WES and CUTseq datasets were assessed and processed independently. Sequencing reads generated by CUTseq were pre-processed using an established pipeline ( github.com/ljwharbers/copynumber-calling ). Briefly, reads were demultiplexed into individual samples based on the internal CUTseq 8 bp barcodes allowing for a single mismatch. The adapters were trimmed using fastp v0.23.2 [ 27 ] and reads were aligned to the GRCh38 reference genome using BWA mem v0.7.17 [ 28 ] and SAMtools v1.16.1 [ 29 ]. Duplicate marking and mapping and alignment quality were assessed using Picard and Alfred v0.2.6 [ 30 ]. The resulting mapped reads provided a genome-wide readout (at 3,422bp resolution) across all HindIII restriction enzyme cut sites. The raw sequencing reads from WES were processed using the nf-core [ 31 ] pipeline Sarek v3.2 for identification of germline and somatic variants [ 32 , 33 ]. Reads were assessed for quality using FastQC v0.11.9 [ 34 ] and MultiQC v1.17 [ 35 ] and adapter and quality-trimmed using fastp v0.23.2 [ 27 ] then aligned to the GRCh38 reference genome using BWA mem v0.7.17 [ 28 ] and SAMtools v1.16.1 [ 29 ]. Base recalibration and duplicate marking was performed using GATK4 v4.3.0.0 [ 36 ] and exome-wide read depth was assessed using mosdepth v0.3.6 [ 37 ]. Variant calling and filtering To assess the mutational landscape within our samples, SNVs and small insertions and deletions (indels) were identified from the WES samples. To limit bias in the identification of somatic variants and to rescue potential false negative mutations, SNVs and indels were called as a majority vote between Mutect2 (GATK v4.3.0.0) [ 36 , 38 ], Strelka2 v2.9.10 [ 39 ] and VarScan v2.4.6 [ 40 ]. Somatic variants were annotated using Ensembl Variant Effect Predictor (VEP) v107 [ 41 ]. CaMutQC v1.1.0 [ 42 ] was used to filter somatic variants based on sequencing quality thresholds, strand bias or the presence of an adjacent indel tag, applying the following parameters: tumour total depth ≥ 40, tumour variant depth ≥ 5, normal total depth ≥ 10, Variant Allele Frequency (VAF) ≥ 0.1 and strand bias score ≥ 2. Variants occurring in the 1000 Genomes Project dataset [ 41 ] were removed as these were highly likely to be germline. To be retained, a consensus somatic variant (identified by ≥ 2/3 variant callers) had to be in a protein coding region, pass CaMutQC filtering thresholds, have an allele frequency ≥ 0.1, and be estimated by VEP to have high or moderate impact. For each patient, germline variants were identified from matched blood or normal tissue samples using VarScan v2.4.6 [ 40 ] and annotated using Ensembl VEP v107 [ 41 ] then checked against the ClinVar database v20240107 to identify clinically relevant mutational variants [ 43 ]. Tumour mutational burden was calculated as the number of consensus mutations per Mb and microsatellite instability (MSI) was assessed using MSIsensor-pro v1.3.0 [ 44 ]. Genes identified as a source of artefacts [ 45 – 47 ] were removed, and mutational loads in TNBC relevant genes [ 7 , 8 , 48 – 51 ] were assessed in detail (Table S4). The number of alterations in oncogenic signalling pathways was determined using Maftools and OncogenicPathways [ 52 , 53 ]. SNV mutational signatures were extracted using the mutSignatures pipeline (v2.1.5) [ 54 ]. Preliminary mutational process assessment suggested that two mutational signatures reconstruct the SNV landscape with minimal error (compared to median mutation counts) and was further corroborated by a silhouette plot after extraction. The de novo signatures were then compared to the COSMIC SBS Mutational Signatures v3.4 [ 55 , 56 ]. To infer candidate driver variants two driver prediction methods with different approaches were applied, dNdScv [ 57 ] and OncodriveFML [ 58 ], on SNVs with variant allele frequency (VAF) > 0.1 in PreT samples. OncodriveFML v2.2.0 was used with default parameters to identify candidate variants based on significant bias in functional impact of mutations (Q-value < 0.05). dNdScv was used with default parameters and identified candidate driver variants based on the ratio of synonymous to non-synonymous mutations (qglobal_cv < 0.05). To detect structural variants, Manta v1.6.0 [ 59 ] was used on the WES samples. Five categories of structural variants were assessed: inversions (INV), insertions (INS), deletions (DEL), duplications (DUP), and breakend (BND) variants. To limit the number of false positives, ‘micro-inversions’ less than 400bp were removed from the analysis. Each WES and CUTseq sample was assessed for somatic CNA using ASCAT.sc v0.1 [ 60 ]. This approach can be employed on either shallow-coverage or targeted sequencing and uses average read depth over various bin sizes as opposed to per base pair read depth. Crosetto et al have shown that CUTseq with HindIII digestion yields reproducible copy number profiles at 100Kb resolution but we conservatively adopted a resolution of 300Kb guaranteeing a larger number of CUTseq sites in each bin (the average distance between 2 consecutive HindIII sites is 3.4kb) [ 23 ]. CNA calls based upon WES or CUTseq were removed if they overlapped a blacklisted region [ 61 ], known to produce mapping artefacts, by > 25% of the CNA length. The union of CNA calls from WES and CUTseq was used for all subsequent analyses. CNA calls were intersected with a list of genes (Table S9) implicated in TNBC progression compiled from the literature [ 8 , 49 , 62 ]. CNA deletion/amplification hotspots were determined by GISTIC v2.0.23 [ 63 ] based on WES and CUTseq data for PreT samples. Regions were considered significantly enriched for alterations at a q-value < 0.05 [ 63 ]. Homologous recombination deficiency (HRD) was predicted using scarHRD v0.1.1 [ 64 ] based on loss of heterozygosity (LOH), large scale transitions (LST) and telomeric allelic imbalances. WES sequencing copy number profiles from Sequenza [ 65 ] were used as input to scarHRD. WGD events were called where ASCAT.sc estimated ploidy for either WES or CUTseq were ≥ 3, which is a more conservative approach than a previous TNBC study using > 2.7 [ 7 ]. Analysis of co-occurrence or mutual exclusivity Linear relationships among genomic metrics (i.e. TMB, CNA burden, ploidy, etc) and clinical variables (TIL, RCB, OS), were determined using the R package ‘corrplot’ [ 66 ] using a confidence level of 0.95, adjusted for multiple testing using the Benjamini-Hochberg procedure. Patterns of pairwise co-occurrence or mutual exclusivity were tested using SELECT v1.6.3 [ 67 , 68 ]. SELECT was run 10 times using 10 different random seeds and the median select score for each pairwise interaction was used. The effects of variables on overall survival were modelled with Cox proportional hazards regression models using the coxph method in the R package ‘survival’ v3.8.3 [ 69 , 70 ] and Kaplan-Meier curves were plotted using R package ‘ggsurvfit’ v1.1.0 [ 71 ]. Analysis of mutational patterns across time We employed two strategies to examine mutational patterns over time, one based upon differences between mean values at each timepoint, and another considering entire temporal trajectories for each patient, based on linear mixed models (LMM). In the first strategy, for each genomic metric (i.e. TMB, CNA burden, ploidy, etc), significant differences between sample means at each timepoint (PreT, PostT, R) or across RCB classes were assessed using Kruskal-Wallis rank sum tests. In cases of presence/absence (i.e. SNV prevalence or copy number status) a Fisher’s exact test was used for two variables and a Chi-squared test for three or more. Pairwise comparisons amongst groups were made using Wilcoxon rank sum tests with Bonferonni corrections for multiple testing. In the second strategy we applied LMM (for continuous variables) or Generalised Linear Mixed Models (GLMM) (for binary dependent variables), treating patients as random effects, to examine complete temporal trajectories. All linear mixed models were implemented in Python using the mixedlm function from the statsmodels package [ 72 ], which supports the random-effects structure required for repeated measures data. For generalised linear mixed models, we used the pymer4 package [ 73 ], which interfaces with the R lme4 framework [ 74 ] to support non-Gaussian outcomes such as binary variables. This approach allowed us to accurately evaluate whether any of the molecular features changed significantly over time, taking the entire temporal trend into account for each patient, and adjusting for within patient correlations between timepoints. The variables analysed and the corresponding summary results from the LMM/GLMM models are provided in Table S14. Results Tumour samples were collected from 22 TNBC patients over the course of NACT and progression of the disease. Samples collected at the following timepoints, PreT (before NACT), PostT (end of NACT after surgery), R (from recurrent tumours) and MidT (at the midpoint of NACT) are shown in Fig. 1 A. Response to NACT was assessed by a breast cancer consultant pathologist and the residual cancer burden (RCB) score and RCB class were determined for each patient (Table S1 ). Patients were ranked from low to high RCB (Fig. 1 B and Table S1 ). PreT samples were available for all patients while PostT samples only exist for patients where there was not pathological complete response to NACT (RCB = I-III). Where available, multiple regions (n = 2–4) of the PostT specimen were obtained and analysed. Ten patients had recurrence following completion of NACT and subsequent adjuvant treatment. Recurrence samples (R) were attainable for 7 of these patients, and it was possible to sample multiple regions (n = 2–4) in 3 patients (Table S1 ). The aim was to establish the most complete sampling over time for each patient, supported by multi-region sampling at timepoints where this was feasible. The numbers of samples sequenced and analysed for each patient at each time point are shown in Fig. 1 B and the full clinical and pathological characteristics of TNBC patients in the study is described in Table S1 . The level of tumour infiltrating lymphocytes (TILs) was measured in all samples and varied widely, from 1–90% of the stromal area (Fig. 1 C, Table S2). Highest infiltration was observed in PreT samples in patients with a good response to NACT and was negatively correlated with RCB score as expected (Figure S1 ). Accordingly, tumours recurred in 4/5 patients with PreT TIL levels of < 10%, consistent with previous reports [ 75 ] and increased TIL levels in PostT in our study was not indicative of recurrence as previously reported [ 76 ]. In the patients with recurrence samples, TILs significantly decreased from a PreT average of 33% to an R average of 5% (Figure S1 , p = 0.03). SNV mutational burden remains stable between RCB classes and across disease progression Tumour mutational burden (TMB) was assessed for each tumour sample (n = 74) using the total burden of consensus SNV (short nucleotide variant) calls from deep whole exome sequencing (WES) (Fig. 1 D, Table S5, Methods). TMB for PreT samples ranged from 1.1 to 19 SNV/Mb, with most patients ranging from 7.1 to 15.6 SNV/Mb. As might be expected, TMB and patient age at diagnosis were positively correlated (R 2 = 0.22, p = 0.027). Two notable outlier patients exhibited significantly lower TMB (1.1 and 1.4 SNV/Mb) and were found to harbour germline mutations in BRCA1 and RAD51C (TNBC6 and TNBC21 respectively) predicted to cause homologous recombination repair deficiency (HRD) (Figure S2). In general, across each patient’s longitudinal samples the mean TMB did not vary significantly across the course of the disease, based upon pairwise comparisons between timepoints (Fig. 1 D); though analysis of complete longitudinal profiles using linear mixed modelling (LMM) revealed a modest decline in TMB for R samples (p = 0.03) compared to PreT samples (Table S14). Overall, TMB did not differ significantly between different RCB classes (Figure S3). Mean genomic SV burden also showed no significant relationship with sampling time (PreT, MidT, PostT or R) or RCB class (Figure S4, Table S8). Microsatellite instability (MSI) was largely stable across time (Fig. 1 E, Table S5), with slight decreases in PostT (p = 0.026) and R samples (p = 0.018) compared to PreT detected in longitudinal LLM modelling (Table S14). We exploited our deep WES data to study high confidence SNVs (supported by > 40 reads) in detail, providing a window on variants lost or newly emerging during disease progression. The fate of SNVs (single nucleotide substitutions and indels) in each patient was followed using a novel metric: variant retention (Fig. 1 D and Table S5). Variant retention is defined as the proportion of SNVs found in common between PreT and subsequent timepoints from the same patient. Between PreT and PostT the retention of SNVs in the majority of patient PostT samples was generally low, ranging from 5–56%, suggesting frequent replacements of one dominant clone by another within tumours in response to treatment. In patients with both PostT and R samples the variant retention at recurrence generally approximated PostT values (on average ± 1.12%). Additional samples midway through NACT (MidT) in a fraction of patients (n = 4) allowed an insight into how quickly variants can be altered during treatment. The variant retention for the MidT samples were equivalent or higher than paired PostT samples for those patients, consistent with rapid responses to treatment. Mutational signatures were extracted de novo from the SNV mutational data (Methods), and the prevalence of known COSMIC SBS mutational signatures was also assessed (Fig. 1 F, Table S7). Derived signature 1 (Sig 1) closely matched COSMIC SBS1 (0.89 cosine similarity) and derived signature 2 (Sig 2) most closely matched SBS13 (0.70 cosine similarity). SBS1 and SBS5 signatures were prevalent in most patients, with minor influences by SBS3, SBS6 and SBS13. SBS1 and SBS5 have been shown to be clock-like signatures and increase with patient age. In patients harbouring germline disruptive SNVs in BRCA1 and RAD51C (TNBC6 and TNBC21), SBS3 was the most dominant signature, which is associated with defective homologous recombination repair. Overall, the stable SNV mutational burden over the progression of TNBC was matched by consistent mutational signatures, suggesting that the driver SNVs present in each tumour may also show stability over time. Loss and emergence of SNVs in known driver genes is common during TNBC treatment Analysis of SNVs disrupting protein coding regions revealed many candidate driver genes found in previous studies. After filtering to limit false positive SNV calls (Methods), we identified SNVs disrupting 57 of 98 candidate genes previously associated with TNBC progression in the literature (Figure S2, Table S5). Those candidates disrupted most frequently by SNVs found in the 22 patients here are shown in Fig. 2 A. A combination of driver gene prediction algorithms (Methods) identified TP53 , CYP2D6 and a novel TNBC-associated gene MICA as candidate driver genes (Table S6), in spite of the limited statistical power of this modest cohort. Variants in one or more of these 3 genes were observed in 17/22 patients. The most frequent variants found in the PreT samples were in TP53 and MICA , both genes being altered in 9/22 patients (41%), followed by CYP2D6 with variants in 7/22 patients (32%). BRCA1 and BRCA2 variants were mutually exclusive (Figure S5), with BRCA1 or BRCA2 being detected in 10/22 patients (45%). Pairwise analysis in the PreT samples determined disruptive BRCA2 variants (in 6 patients) and disruptive TP53 variants (in 9 patients) to also be mutually exclusive events (Figure S5). Overall, 91% (20/22) of patients had PreT samples with somatic or germline disruptions of TP53 , BRCA1 , BRCA2 , PRKDC or ATM, consistent with the early development of genomic instability. The remaining two patients (TNBC9, TNBC15) developed disruptive variants in these genes at PostT, suggesting ongoing selection for genomic instability via alterations to these same 5 genes across the cohort. We employed a conservative consensus SNV calling (sequencing read depth > 40, variant allele frequency (VAF) > 0.1) strategy to ensure confidence in the detection of both ‘lost’ and ‘novel’ variants in later (PostT and/or R) samples, relative to the initial PreT biopsy samples. Losses and gains of SNVs were frequently observed over time, even in the 5 most frequently altered driver genes TP53, MICA, CYP2D6, BRCA1 and BRCA2 (Fig. 2 A). For example, TP53 variants detected in the PreT samples of 7 patients were no longer observed in the paired PostT or R samples of these patients. This included a patient (TNBC10), where three PostT samples were sequenced and the TP53 variant identified in the PreT sample of the patient was lost in all three samples. Whereas, in patient TNBC22, four PostT samples were sequenced and the corresponding PreT TP53 variant was only identified in one sample suggesting intra-tumour heterogeneity. In addition, TP53 variants arose in the PostT or R samples of 5 patients where these variants were not detected in the paired PreT samples. Remarkably, in some patients, variants were lost concurrently with new variants arising in the same gene. This was seen with MICA SNVs in 4 patients, where later (PostT or R) gains and losses of variants were observed relative to the paired PreT samples. Some genes, i.e. BCL9 and PIKC3A were subject only to loss of the PreT variants across the time course of the study, with 3/4 patients’ (BCL9) and 2/2 patients’ (PIK3CA) variants undetectable after NACT. PostT and R samples also gained variants in a number of TNBC-related genes, in particular ERBB2 variants appeared in 4 patients, ALK variants in 3 patients and EGFR variants in 2 patients, with additional genes obtaining PostT variants in single patients (Figure S2). Thus, the consistent SNV burdens seen genome-wide (Fig. 1 ) obscure frequent turnover at the level of particular driver variants in particular genes, likely reflecting clonal competition and replacement within tumours over time. The prevalence of genes altered by SNVs were largely consistent over time and across RCB classes (Fig. 2 B). A notable deviation is the prevalence of BRCA1 variants (3/7 patients) in the PreT samples of patients in RCB 0, which exhibited the best response to NACT. The locations of the SNVs occurring within each of the 6 most altered driver genes show similar patterns over time (Fig. 2 C, and Table S5). In general, it seems that both lost and novel SNVs disrupt similar regions of the proteins encoded by these driver genes, and this is also true of SNVs maintained across all timepoints in the same genes. For example, across both PreT and PostT samples, 10 different TP53 variants, including 8 missense and 2 nonsense variants, were detected. Multiple lost and novel SNVs were found, and on 4 occasions the same variant was observed in both PreT and PostT, but regardless of timing the vast majority were present in the DNA binding domain (Fig. 2 C). Other driver genes failed to show strong biases in the locations of lost and novel variants, suggesting similar disruptions of gene functions over time. There were no statistically significant associations between candidate driver variants and patient survival, though such analyses are under-powered with the current data (Figure S6). Shifts in altered oncogenic pathways were also evident, across time and across RCB classes, based upon changes in the proteins predicted to be altered by SNVs (Fig. 2 D). Patients with poorer responses to NACT (RCB 2–3) show more frequent alteration of the NOTCH (+ 18%, OR = 1.3) and TP53 (+ 18%, OR = 1.5) pathways than patients with better response to NACT (RCB 0–1). Better responding patients (RCB 0–1) showed higher frequencies of altered WNT (+ 9%, OR = 1.2), PI3K (+ 27%, OR = 1.6) and RTK-RAS (+ 27%, OR = 1.6) pathways (Fig. 2 D). Comprehensive detection of copy number alterations over disease progression Copy number alterations (CNAs) were detected in our cohort using two different but complementary assays, WES and CUTSeq [ 23 ] (Fig. 3 A). WES is known to provide a detailed interrogation of mutations within the fraction of the genome occupied by exons, while CUTSeq provides a broader picture of CNAs that are distributed throughout the human genome. For example, a closer examination of chromosome 8 in a patient, revealed that both approaches identified an amplification of chromosome arm 8q (Fig. 3 B). However, as expected the read depth of WES closely mirrors the density of genes, whereas CUTseq shows a more uniform distribution across the chromosome arm. Decreasing the genomic range and examining a more focal region around a single gene ( MYC ) further exemplifies the high concentration of reads within exons in WES, while CUTseq shows a more even distribution. In contrast, a closer examination of chromosome X revealed a deletion near the centromere that was detected using CUTseq, but not by WES (Fig. 3 B). The detection of such alterations in gene-poor regions highlights the potential of CUTseq to assess genomic instability more comprehensively. CNA patterns over the course of treatment showed differences between patients (all CNA profiles are provided in Figure S7). Patient TNBC18 (Fig. 3 C) provides an archetypal example of stability over time, where amplified and deleted regions present at PreT, were also found in all 4 tumour regions sampled at PostT, and at recurrence (with similar profiles over time seen in both CUTSeq and WES). In contrast, CNA profiles for many patients, such as TNBC11, were dynamic over time (Fig. 3 C), with widespread amplifications seen across most chromosomes in the PreT sample becoming completely absent in PostT samples, then re-appearing in the R tumour sample. Importantly, this pattern of alteration loss and recurrence through longitudinal time is consistent in both CUTseq and WES data for TNBC11, demonstrating broadly consistent results between both assays. Across all samples, there was a high level of agreement between the two approaches (Fig. 3 D), as indicated with significant correlations of estimated tumour ploidy (p = 0.032) and the proportion of each genome either amplified (p = 0.014) or deleted (p = 0.004). However, the amount of sequencing reads required to assess CNA profiles using each approach differed, with CUTseq on average requiring only 10.5% of the sequencing reads needed in the WES analysis (Fig. 3 D). The regions of the genome enriched for amplifications or deletions across the PreT samples revealed similar patterns between CUTseq and WES CNA calls ( Figure S7 , Table S10 ). Two regions were jointly identified as hotspots for amplifications: 5p15.1 and 8q. The 5p15.1 hotspot encompassed 36 amplified genes including, MYO10 (myosin X) which has been reported to be overexpressed and to drive genomic instability in cancers [ 77 ] and contributes to breast cancer aggressiveness and metastasis [ 78 ]. Large regions of 8q were enriched for amplifications in both CUTseq and WES analyses (Table S10). The MYC gene is located on this chromosome arm and is commonly amplified in TNBC and promotes immune suppression [ 79 ]. Despite the high level of agreement between the two approaches, there were regions of enrichment which were only identified using one approach and not the other. WES identified 3q26.33 as enriched for amplifications, which notably contains PIK3CA which is commonly amplified in TNBC and potentially linked to response to targeted treatment strategies [ 80 ]. Alternatively, CUTseq identified a focal amplification of 19q12 harbouring CCNE1 which may confer chemotherapy resistance and has been linked to poor prognosis in TNBC [ 81 ], as well as other genes (POP4, PLEKHF1, and TSHZ3) reported to be amplified in cancer [ 82 ]. This region of chromosome 19 is adjacent to the centromere which may explain why it was only identified using CUTseq. Similarly, CUTseq identified Xq13.2 as a hotspot for deletions which contains XIST , the loss of which has been linked to accelerated tumour growth and brain metastases in breast cancer [ 83 ]. Overall, comparisons of WES and CUTseq estimates of CNA, arm level events and ploidy (Fig. 3 D) suggest that these complementary assays provide similar results where they can be compared, while each may still contribute novel information ( Figure S7 ). This supports our use of a union of WES/CUTseq CNA calls in the following analyses. Pervasive aneuploidies and whole genome duplication drives tumour evolution Though genome-wide SNV loads show broad stability over time (Fig. 1 ), chromosome arm aneuploidies are dynamic, seen abundantly across PreT samples, but often showing dramatic losses in PostT samples (Fig. 4 A, Table S13). These aneuploidies were frequent across all RCB classes and affected almost all chromosomes, with gains of 1q, 8q, 19q and 22q most frequent at PreT (Fig. 4 A, 4 B). Dramatic decreases in arm alterations were observed between PreT and PostT samples across most patients, consistent with a disproportionate impact of NACT on tumour lineages carrying aneuploidies. These arm level alterations then often reappeared in later (R) samples but usually involved novel CNA events, suggesting the dominance of distinct lineages in late disease (Fig. 4 A). In general, R samples are notably enriched for these novel events that mimic PreT arm gains. Consistent with timepoint specific advantages to particular tumour clones, certain arm gain events, such as 19q gain seen at PreT, appear to be replaced at later timepoints by 19q arm loss at PostT in multiple patients (Fig. 4 A). This significant pattern of decreased CNA burden at PostT as compared to PreT and R was seen when comparing the mean proportion of the genome altered by amplifications (Fig. 4 C). The prevalence of gain or loss for certain chromosome arms (2p, 7p, 10p, and Xp, Chi-squared p < 0.05 with Bonferonni correction) was significantly related to a specific sampling time point (Fig. 4 B). In contrast, certain aneuploidies, particularly gain of 1q and 8q, showed convincing evidence of persistence across the course of treatment, even in PostT samples where most other aneuploidies disappeared (Fig. 4 B). This may suggest ‘aneuploidy addiction’ - where particular aneuploidies are required for tumourigenesis - which has recently been demonstrated in vitro in breast cancer cell lines [ 84 ]. Strikingly, tumour cell ‘addictions’ to both 1q and 8q gains were found to confer strong advantages in this study, enhancing tumour cell growth, and 1q gains were found to be among the earliest alterations to occur in breast tumour development [ 84 ]. However, in this in vitro study the mutational mechanisms underlying these aneuploidies remained unknown. The 1q and 8q gains seen here, in TNBC in vivo , often occur on a background of pervasive CNA gains seen across the vast majority of chromosomes (Fig. 4 B, 4 C, Table S11), suggesting they may be a result of WGD. Indeed, predicting WGD using a conservative approach (Methods) we found evidence for WGD in 36% (8/22) of patients at PreT, 21% (3/14) at PostT, and in 57% (4/7) at R. All tumours with WGD carry 8q gains, and all but one carry 1q gains at PreT (Fig. 4 A), suggesting that WGD may be exploited to generate advantageous aneuploidies during earlier stages, but also later at recurrence (Fig. 4 D). Gains of 1q and 8q were also seen in 27% (6/22) of PreT samples lacking evidence for WGD, suggesting other mutational mechanisms also produce these alterations. The re-emergence of WGD may reflect the rise in frequency of pre-existing clones with WGD, or further de novo WGD events in later disease. This aligns with recent work in ovarian cancer proposing that WGD is an ongoing mutational process promoting evolvability [ 85 ]. Importantly, the prevalence of WGD was not uniformly distributed across RCB classes, with the majority (5/7) of patients with the best response to NACT (RCB 0) having tumours with WGD at PreT (Fig. 4 D). One of the drugs included in NACT (docetaxel) has been shown to induce polyploidization and aberrant aneuploidies by stabilizing microtubules [ 86 ]. The increased chromosome instability introduced by NACT may cause tumour cells that have previously undergone WGD to become unviable leading to improved treatment response and reduced residual cancer burden. Given the large changes in ploidy and frequent WGD seen during TNBC evolution, many genes undergo recurrent alterations (Fig. 5 ). Among the genes amplified are many that have previously been reported to undergo alterations during progression (Fig. 5 A) in prior studies of the TNBC genome [ 8 , 49 , 62 ]. Among the most recurrently amplified are MYC (located at 8q24.21), AKT3 (1q43) and MDM4 (1q32.1). All three amplifications are present in the majority of PreT samples, are very rarely deleted, and also show evidence of persistence in multiple patients throughout treatment and into recurrence (Fig. 5 A, B), consistent with the 1q and 8q chromosome arm gains that also persist through treatment (Fig. 4 A). All three have been specifically linked to TNBC via their upregulation and amplification. MYC amplification is commonly observed in TNBC and is thought to suppress innate immunity and allow tumour immune escape [ 79 ]. AKT3 amplification has been observed in multiple breast cancer subtypes, including TNBC, and is thought to mediate resistance to tamoxifen [ 87 ]. MDM4 has emerged as an important factor in promoting TNBC tumour formation and metastasis, via its role as a TP53 inhibitor [ 62 ]. It therefore seems plausible that the frequent and persistent 1q and 8q arm gains seen in these tumours (Fig. 4 A) result in increases in the expression of one or more of these genes, though we lack direct evidence for this. This would also offer a potential explanation for tumour cell ‘addiction’ to 1q and 8q gains seen in recent in vitro studies [ 84 ]. Analysis of altered oncogenic pathways revealed an intricate pattern of synergy between genes in the RTK/ RAS pathway and 1q and 8q gains. Amplifications of eight genes (Fig. 5 A, B) in the RTK/ RAS pathway ( EGFR , IGF1R , KRAS , KIT , NF1 , FGFR1 , PDGFRA and MAP3K1 ) showed co-occurrence with gains in 1q (12/16 patients) or 8q (14/15). Other frequently altered genes suggest a more complex mutational landscape. For example, BRD4 (19p13.12), NOTCH3 (19p13.2) and CCNE1 (19q12) undergo amplification and deletion in different samples and at different timepoints (Fig. 5 A, B). This volatile state may suggest the presence of complex rearrangements such as chromothripsis, which has been reported to be particularly enriched on chromosome 19 in high grade serous ovarian cancer [ 88 ]. We also observe frequent amplifications and deletions of KDM6A , with amplifications common in PreT but deletions at later stages (Fig. 5 A, B), consistent with reports that KDM6A loss can reduce sensitivity to the therapeutic agents used to treat TNBC patients [ 89 ]. A closer examination of the relationship between gene copy number alterations and WGD status in PreT samples revealed distinct patterns (Fig. 5 C). In samples that had undergone WGD, amplifications of the top 15 most commonly altered genes were highly prevalent (87.5%). In contrast, non-WGD samples showed a lower frequency of amplification (38.6%). Deletions were relatively rare in WGD samples; however, in non-WGD samples, deletions affecting BRD4 and NOTCH3 at 19p13.12 were common, occurring in 57.1% of cases compared to just 12.5% in WGD samples. These findings suggest that WGD shapes genome-wide mutational patterns and drives distinct combinations of candidate driver gene alterations. Clinical correlates of mutational loads and aneuploidy A number of clinically relevant features were found to be inter-correlated (Fig. 6 A). TMB in PreT samples positively correlated with metrics of genomic instability across various scales: CUTseq derived ploidy (R = 0.59, p = 0.04), total structural variant burden (R = 0.73, p = 0.0001) and microsatellite instability (R = 0.73, p = 0.009). As anticipated, the amount of the genome amplified by CNAs correlated with CUTseq (R = 0.71, p = 0.004) and WES derived ploidy estimates (Fig. 6 A, R = 0.73, p = 0.0009), and TIL infiltration and RCB score were negatively correlated (R=-0.56, p = 0.045) (Fig. 6 A, Figure S1 ). Investigating relationships between mutations, clinical features and overall patient overall survival (OS) showed that patient OS varied significantly, with patients in RCB class II/III having poorer OS than RCB 0/I patients (p = 0.039) (Fig. 6 B). Although HRD scores did not differ significantly between RCB classes (Figure S8), as expected, patients with tumours showing evidence for HRD in PreT samples (defined as scarHRD scores > 70) had significantly better OS (Fig. 6 B, p = 0.01). This is in agreement with previous studies demonstrating higher HRD scores are associated with a better OS and better sensitivity to treatments such as PARP inhibitors and platinum-based chemotherapy [ 90 , 91 ]. Using a Bayesian approach (Methods) we looked for pairwise co-occurrence or mutual exclusivity among mutational events, including the presence or absence of SNVs in driver genes (Fig. 2 ), chromosome arm aneuploidies (Fig. 4 ) or CNAs recurrently impacting genes (Fig. 5 ) (see Figure S9). The output was dominated by pairwise co-occurrence events between amplifications in genes and chromosome arms, as expected with frequent WGD events (Figure S1 0). To analyse longitudinal patterns within each individual patient, we employed Linear Mixed Model (LMM) and Generalised LMM (for binary dependent variables) modelling approaches for all features. The modelling indicates that several molecular and clinical features change significantly across time points (Fig. 6 C, Table S14), showing patterns that are broadly consistent with the analyses based upon pairwise comparisons between timepoints (Fig. 4 ). For instance, from PreT to R, we observe a decrease in TILs ( \(\:\beta\:=-22.03\) , SE = 5.62, p < 0.0001) and a reduction of SV burden ( \(\:\beta\:=-9.24\) , SE = 4.38, p = 0.035). Similarly, the odds of WGD shrinks to 18% of the initial odds from PreT to PostT (p = 0.038), indicating that WGD is significantly less likely to occur at the PostT time point compared to PreT (Table S14). WGD is also commonly seen in R samples (Fig. 4 D), which is likely to reflect the resurgence of WGD bearing clones in these recurrent tumours. In fact, even though median ploidy is greater than 2 in PreT samples, median (WES or CUTseq) ploidy is higher at recurrence, suggesting multiple WGD events in some samples (Fig. 6 D). As the threshold for WGD (mean fraction of genome amplified over WES and CUTseq datasets) is increased from 20% to 70%, PreT and R samples were consistently higher than PostT samples, demonstrating that this trend is not dependent on the threshold by which WGD was determined (Fig. 6 E). We conclude that WGD events are an under-appreciated mutational mechanism underlying the evolution of TNBC during treatment. These events appear to drive the acquisition of advantageous aneuploidies, particularly 1q and 8q gains, and are predicted to impact known cancer driver genes. WGD bearing tumours are common at PreT and presumably confer proliferative advantages before treatment, but these tumours also appear to be unusually sensitive to NACT, resulting in a strong decline in the ploidy of PostT samples. At later stages higher ploidy mediated by WGD, again appears to offer advantages to dominant clones within R samples. Discussion Using two complementary sequencing approaches (WES and CUTseq), we have reconstructed the mutational landscape of TNBC and tracked its dynamics during NACT. The genomic and transcriptomic landscape of TNBC has been examined in several previous studies [ 6 – 8 , 51 ], but TNBC temporal mutational dynamics during chemotherapy treatment is poorly studied. Here, we describe a cohort with longitudinal sampling, providing an unusual opportunity to determine how the mutational landscape responds to treatment, including SNVs, CNAs and aneuploidies. We found that although the genome-wide SNV mutational burden remained stable during disease progression, the loss and emergence of SNVs in known TNBC driver genes was common in response to treatment. These complex dynamics were seen in TP53 , MICA , CYP2D6 , BRCA1 and BRCA2 , suggesting clonal loss and replacement events over time, consistent with selection favouring different variants at different times in the same driver gene. Dramatic genome-wide structural shifts were seen in all samples, reflected in CNA and chromosome arm aneuploidies, with a substantial bias to duplications. Overall, the abundant structural variation seen in PreT samples was followed by frequent loss of these alterations after treatment, but the re-emergence of similar alterations at recurrence. These alterations include advantageous aneuploidies increasing tumour growth in vitro , such as 1q gain, that has been proposed to drive tumourigenesis via an increase in MDM4 expression, which phenocopies the effects of TP53 inactivating mutations to suppress TP53 signalling [ 84 ]. Such aneuploidies have also been demonstrated to create collateral therapeutic vulnerabilities via the toxic effects of duplicating many 1q genes, which can represent novel drug targets [ 84 ]. We also see positive correlations between ploidy and TMB, suggesting tumours undergoing WGD may present higher neoantigen loads, targetable by neoantigen vaccines, underlining the promise of such emerging approaches in TNBC [ 92 ]. Relatively early, and frequent (36% of PreT samples) WGD events appear to drive or at least co-occur with the extensive aneuploidies seen pre-treatment, consistent with a recent study of TNBC patient-derived xenograft samples that lacked longitudinal sampling [ 20 ]. Notably a higher frequency of PreT WGD was seen in patients with the best response to NACT (RCB0/1: 55%; RCB2/3: 18%), suggesting that early WGD may confer susceptibility to treatment. WGD was also frequently found at recurrence (58% of R samples), either reflecting the re-emergence of proliferative WGD bearing clones, or late, de novo WGD events. To our knowledge this is the first evidence for WGD as a common event in both pre-treatment and recurrent TNBC tumours. Subsequent studies with larger sample sizes and generating WGS data will be better placed to make more detailed estimates of the impact of WGD on clinical outcomes. These unusual, longitudinally sampled data suggest the presence of WGD itself may be a useful prognostic biomarker of NACT response, associated with better or worse outcomes, depending on the time of sampling. Emerging studies also suggest that WGD may be a targetable vulnerability [ 93 , 94 ]. Detecting WGD and genome-wide CNA accurately in a FFPE tumour sample is not possible using the most widely used profiling technologies (targeted sequencing panels) while WES and WGS are not financially viable options for many healthcare providers. However, we have demonstrated that CUTseq represents an affordable assay that provides accurate results in the setting of structurally complex tumour genomes sampled using FFPE blocks and primary tumour sections. Conclusions Determining the origins and dynamics of structural complexity in tumourigenesis over time is critical to understand the genomic heterogeneity between patients and within tumours. In this study we reveal the genome-wide shifts in structural variation that underlie TNBC evolution during treatment. We present evidence that WGD in pre-treatment biopsies is associated with better response to NACT, but also appears to confer advantages to tumours at recurrence. Validation in larger, temporally sampled cohorts is required to confirm these results, but the analysis of large numbers of samples will rely on the development of affordable sequencing strategies that can be used to profile the (FFPE) samples that are generally available from healthcare providers. We have demonstrated that CUTseq and WES can be combined to that end, and readily provide new insights into tumour evolution over time. Declarations Acknowledgements The authors thank the patients and their families for their participation in this academic study. We are also grateful for the time and effort involved in sample collection, of the breast radiologists, surgeons and nurses at the Edinburgh Breast Unit. We would like to acknowledge the Edinburgh Clinical Research Facility for DNA sequencing and support, and the NRS Lothian Bioresource for additional tissue samples. Some of the results published here are in part based upon data generated by the TCGA Research Network: https://www.cancer.gov/tcga. Authors’ contributions O.O., F.S., A.E. and C.A.S. designed the study and supervised analysis of the cohort. F.S. performed laboratory work and D.P.B. performed the bioinformatic analyses. A.M., M.I.O. and A.E. provided additional analysis. A.I. performed histopathological review. O.O designed the NEO and metcfDNA studies, arranged relevant ethical approvals and recruited patients (under her care throughout their breast cancer journey). N.W collected tissue samples and gathered patient clinicopathological characteristics, from national health electronic patient records. D.P.B., F.S., O.O. and C.A.S. wrote the manuscript, which was read and approved by all authors. Funding This study was supported by funding and donations from Make 2nds Count, John Henry Cugley and Gillian Hunt and the Breast Cancer Institute. Salary for F.S. was supported by the Susan Melville Fund. C.A.S., D.B. and A.M. were supported by MRC core funding to the MRC Human Genetics Unit, University of Edinburgh and MRC Programme funding MC_UU_00035/1. AE was supported by a University of Edinburgh Chancellor’s Fellowship, a Langmuir Talent Development Fellowship and core funding from the MRC Human Genetics Unit. Data availability Genomic data generated in this study have been submitted to the European Genome/ Phenome Archive, with accession numbers EGAD00001015684 (CUTseq data) and EGAD00001015687 (WES data). Ethics approval and consent to participate Collection of human tissues and relevant clinicopathological data were approved by the South East Scotland Research Ethics Committee 01 with REC references: 13/SS/0236 and 15/ES/0094, and all patients consented to sample collection and use of their data in research Consent for publication Not applicable Competing interests None References Nik-Zainal S, Davies H, Staaf J, Ramakrishna M, Glodzik D, Zou X, et al. Landscape of somatic mutations in 560 breast cancer whole-genome sequences. Nat Nat Publishing Group. 2016;534:47–54. https://doi.org/10.1038/nature17676 . Bielski CM, Zehir A, Penson AV, Donoghue MTA, Chatila W, Armenia J, et al. Genome doubling shapes the evolution and prognosis of advanced cancers. Nat Genet. 2018;50:1189–95. https://doi.org/10.1038/s41588-018-0165-1 . López S, Lim EL, Horswell S, Haase K, Huebner A, Dietzen M, et al. Interplay between whole-genome doubling and the accumulation of deleterious alterations in cancer evolution. Nat Genet Nat Publishing Group. 2020;52:283–93. https://doi.org/10.1038/s41588-020-0584-7 . Kang K, Wu Y, Han C, Wang L, Wang Z, Zhao A. Homologous recombination deficiency in triple-negative breast cancer: Multi-scale transcriptomics reveals distinct tumor microenvironments and limitations in predicting immunotherapy response. Comput Biol Med. 2023;158:106836. https://doi.org/10.1016/j.compbiomed.2023.106836 . Burstein MD, Tsimelzon A, Poage GM, Covington KR, Contreras A, Fuqua SAW, et al. Comprehensive Genomic Analysis Identifies Novel Subtypes and Targets of Triple-Negative Breast Cancer. Clin Cancer Res. 2015;21:1688–98. https://doi.org/10.1158/1078-0432.CCR-14-0432 . Lehmann BD, Jovanović B, Chen X, Estrada MV, Johnson KN, Shyr Y, et al. Refinement of Triple-Negative Breast Cancer Molecular Subtypes: Implications for Neoadjuvant Chemotherapy Selection. PLOS ONE Public Libr Sci. 2016;11:e0157368. https://doi.org/10.1371/journal.pone.0157368 . Jiang Y-Z, Ma D, Suo C, Shi J, Xue M, Hu X, et al. Genomic and Transcriptomic Landscape of Triple-Negative Breast Cancers: Subtypes and Treatment Strategies. Cancer Cell. 2019;35:428–e4405. https://doi.org/10.1016/j.ccell.2019.02.001 . Lehmann BD, Colaprico A, Silva TC, Chen J, An H, Ban Y, et al. Multi-omics analysis identifies therapeutic vulnerabilities in triple-negative breast cancer subtypes. Nat Commun Nat Publishing Group. 2021;12:6276. https://doi.org/10.1038/s41467-021-26502-6 . Zhu S, Wu Y, Song B, Yi M, Yan Y, Mei Q, et al. Recent advances in targeted strategies for triple-negative breast cancer. J Hematol OncolJ Hematol Oncol. 2023;16:100. https://doi.org/10.1186/s13045-023-01497-3 . Denkert C, von Minckwitz G, Darb-Esfahani S, Lederer B, Heppner BI, Weber KE, et al. Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy. Lancet Oncol. 2018;19:40–50. https://doi.org/10.1016/S1470-2045(17)30904-X . Salgado R, Denkert C, Demaria S, Sirtaine N, Klauschen F, Pruneri G, et al. The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs Working Group 2014. Ann Oncol Off J Eur Soc Med Oncol. 2015;26:259–71. https://doi.org/10.1093/annonc/mdu450 . Menezes MCS, Raheem F, Mina L, Ernst B, Batalini F. PARP Inhibitors for Breast Cancer: Germline BRCA1/2 and Beyond. Cancers. 2022;14:4332. https://doi.org/10.3390/cancers14174332 . Bianchini G, De Angelis C, Licata L, Gianni L. Treatment landscape of triple-negative breast cancer — expanded options, evolving needs. Nat Rev Clin Oncol Nat Publishing Group. 2022;19:91–113. https://doi.org/10.1038/s41571-021-00565-2 . Cortazar P, Zhang L, Untch M, Mehta K, Costantino JP, Wolmark N, et al. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet Elsevier. 2014;384:164–72. https://doi.org/10.1016/S0140-6736(13)62422-8 . Fitzpatrick A, Tutt A. Controversial issues in the neoadjuvant treatment of triple-negative breast cancer. Ther Adv Med Oncol. 2019;11:1758835919882581. https://doi.org/10.1177/1758835919882581 . Schmid P, Cortes J, Dent R, McArthur H, Pusztai L, Kümmel S, et al. Overall Survival with Pembrolizumab in Early-Stage Triple-Negative Breast Cancer. N Engl J Med. 2024;391:1981–91. https://doi.org/10.1056/NEJMoa2409932 . Schmid P, Cortes J, Dent R, Pusztai L, McArthur H, Kümmel S, et al. Event-free Survival with Pembrolizumab in Early Triple-Negative Breast Cancer. N Engl J Med Mass Med Soc. 2022;386:556–67. https://doi.org/10.1056/NEJMoa2112651 . Valencia GA, Rioja P, Morante Z, Ruiz R, Fuentes H, Castaneda CA, et al. Immunotherapy in triple-negative breast cancer: A literature review and new advances. World J Clin Oncol. 2022;13:219–36. https://doi.org/10.5306/wjco.v13.i3.219 . Xu Q, Kaur J, Wylie D, Mittal K, Li H, Kolachina R, et al. A Case Series Exploration of Multi-Regional Expression Heterogeneity in Triple-Negative Breast Cancer Patients. Int J Mol Sci. 2022;23:13322. https://doi.org/10.3390/ijms232113322 . Kuzmin E, Baker TM, Lesluyes T, Monlong J, Abe KT, Coelho PP, et al. Evolution of chromosome-arm aberrations in breast cancer through genetic network rewiring. Cell Rep. 2024;43:113988. https://doi.org/10.1016/j.celrep.2024.113988 . Karaayvaz M, Cristea S, Gillespie SM, Patel AP, Mylvaganam R, Luo CC, et al. Unravelling subclonal heterogeneity and aggressive disease states in TNBC through single-cell RNA-seq. Nat Commun Nat Publishing Group. 2018;9:3588. https://doi.org/10.1038/s41467-018-06052-0 . Kim C, Gao R, Sei E, Brandt R, Hartman J, Hatschek T, et al. Chemoresistance Evolution in Triple-Negative Breast Cancer Delineated by Single-Cell Sequencing. Cell. 2018;173:879–e89313. https://doi.org/10.1016/j.cell.2018.03.041 . Zhang X, Garnerone S, Simonetti M, Harbers L, Nicoś M, Mirzazadeh R, et al. CUTseq is a versatile method for preparing multiplexed DNA sequencing libraries from low-input samples. Nat Commun Nat Publishing Group. 2019;10:4732. https://doi.org/10.1038/s41467-019-12570-2 . Bossuyt V, Provenzano E, Symmans WF, Boughey JC, Coles C, Curigliano G, et al. Recommendations for standardized pathological characterization of residual disease for neoadjuvant clinical trials of breast cancer by the BIG-NABCG collaboration. Ann Oncol Off J Eur Soc Med Oncol. 2015;26:1280–91. https://doi.org/10.1093/annonc/mdv161 . Dieci MV, Radosevic-Robin N, Fineberg S, van den Eynden G, Ternes N, Penault-Llorca F, et al. Update on tumor-infiltrating lymphocytes (TILs) in breast cancer, including recommendations to assess TILs in residual disease after neoadjuvant therapy and in carcinoma in situ: A report of the International Immuno-Oncology Biomarker Working Group on Breast Cancer. Semin Cancer Biol. 2018;52:16–25. https://doi.org/10.1016/j.semcancer.2017.10.003 . Denkert C, Wienert S, Poterie A, Loibl S, Budczies J, Badve S, et al. Standardized evaluation of tumor-infiltrating lymphocytes in breast cancer: results of the ring studies of the international immuno-oncology biomarker working group. Mod Pathol Off J U S Can Acad Pathol Inc. 2016;29:1155–64. https://doi.org/10.1038/modpathol.2016.109 . Chen S, Zhou Y, Chen Y, Gu J. fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics. 2018;34:i884–90. https://doi.org/10.1093/bioinformatics/bty560 . Li H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM [Internet]. arXiv; 2013 [cited 2023 Oct 31]. https://doi.org/10.48550/arXiv.1303.3997 Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinforma Oxf Engl. 2009;25:2078–9. https://doi.org/10.1093/bioinformatics/btp352 . Rausch T, Hsi-Yang Fritz M, Korbel JO, Benes V. Alfred: interactive multi-sample BAM alignment statistics, feature counting and feature annotation for long- and short-read sequencing. Bioinformatics. 2019;35:2489–91. https://doi.org/10.1093/bioinformatics/bty1007 . Ewels PA, Peltzer A, Fillinger S, Patel H, Alneberg J, Wilm A, et al. The nf-core framework for community-curated bioinformatics pipelines. Nat Biotechnol. 2020;38:276–8. https://doi.org/10.1038/s41587-020-0439-x . Garcia M, Juhos S, Larsson M, Olason PI, Martin M, Eisfeldt J, et al. Sarek: A portable workflow for whole-genome sequencing analysis of germline and somatic variants. F1000Research. 2020;9:63. https://doi.org/10.12688/f1000research.16665.2 . Hanssen F, Garcia MU, Folkersen L, Pedersen AS, Lescai F, Jodoin S et al. Scalable and efficient DNA sequencing analysis on different compute infrastructures aiding variant discovery [Internet]. bioRxiv; 2024 [cited 2024 May 9]. p. 2023.07.19.549462. https://doi.org/10.1101/2023.07.19.549462 Andrews S. FastQC: A Quality Control tool for High Throughput Sequence Data [Internet]. 2010 [cited 2022 Dec 12]. https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ . Accessed 12 Dec 2022. Ewels P, Magnusson M, Lundin S, Käller M. MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics. 2016;32:3047–8. https://doi.org/10.1093/bioinformatics/btw354 . McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The Genome Analysis Toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010;20:1297–303. https://doi.org/10.1101/gr.107524.110 . Pedersen BS, Quinlan AR. Mosdepth: quick coverage calculation for genomes and exomes. Bioinformatics. 2018;34:867–8. https://doi.org/10.1093/bioinformatics/btx699 . Benjamin D, Sato T, Cibulskis K, Getz G, Stewart C, Lichtenstein L. Calling Somatic SNVs and Indels with Mutect2 [Internet]. bioRxiv; 2019 [cited 2024 May 9]. p. 861054. https://doi.org/10.1101/861054 Kim S, Scheffler K, Halpern AL, Bekritsky MA, Noh E, Källberg M, et al. Strelka2: fast and accurate calling of germline and somatic variants. Nat Methods. 2018;15:591–4. https://doi.org/10.1038/s41592-018-0051-x . Koboldt DC, Zhang Q, Larson DE, Shen D, McLellan MD, Lin L, et al. VarScan 2: Somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Res. 2012;22:568–76. https://doi.org/10.1101/gr.129684.111 . McLaren W, Gil L, Hunt SE, Riat HS, Ritchie GRS, Thormann A, et al. The Ensembl Variant Effect Predictor. Genome Biol. 2016;17:122. https://doi.org/10.1186/s13059-016-0974-4 . Wang X, Jiang T, Shen A, Chen Y, Zhou Y, Liu J et al. CaMutQC: An R Package for Integrative Quality Control of Cancer Somatic Mutations [Internet]. bioRxiv; 2024 [cited 2024 Aug 15]. p. 2024.08.12.606123. https://doi.org/10.1101/2024.08.12.606123 Landrum MJ, Lee JM, Riley GR, Jang W, Rubinstein WS, Church DM, et al. ClinVar: public archive of relationships among sequence variation and human phenotype. Nucleic Acids Res. 2014;42:D980–5. https://doi.org/10.1093/nar/gkt1113 . Jia P, Yang X, Guo L, Liu B, Lin J, Liang H, et al. MSIsensor-Pro: Fast, Accurate, and Matched-Normal-Sample-Free Detection of Microsatellite Instability. Genomics Proteom Bioinf. 2020;18:65–71. https://doi.org/10.1016/j.gpb.2020.02.001 . Jamuar SS, Kuan JL, Brett M, Tiang Z, Tan WLW, Lim JY et al. Incidentalome from Genomic Sequencing: A Barrier to Personalized Medicine? EBioMedicine. 2016;5:211–6. https://doi.org/10.1016/j.ebiom.2016.01.030 Shyr C, Tarailo-Graovac M, Gottlieb M, Lee JJ, van Karnebeek C, Wasserman WW. FLAGS, frequently mutated genes in public exomes. BMC Med Genomics. 2014;7:64. https://doi.org/10.1186/s12920-014-0064-y . Lawrence MS, Stojanov P, Polak P, Kryukov GV, Cibulskis K, Sivachenko A, et al. Mutational heterogeneity in cancer and the search for new cancer-associated genes. Nat Nat Publishing Group. 2013;499:214–8. https://doi.org/10.1038/nature12213 . Chen H, Chong W, Yang X, Zhang Y, Sang S, Li X, et al. Age-related mutational signature negatively associated with immune activity and survival outcome in triple-negative breast cancer. Oncoimmunology. 2020;9:1788252. https://doi.org/10.1080/2162402X.2020.1788252 . Derakhshan F, Reis-Filho JS. Pathogenesis of Triple-Negative Breast Cancer. Annu Rev Pathol Mech Dis. 2022;17:181–204. https://doi.org/10.1146/annurev-pathol-042420-093238 . Shah SP, Roth A, Goya R, Oloumi A, Ha G, Zhao Y, et al. The clonal and mutational evolution spectrum of primary triple-negative breast cancers. Nat Nat Publishing Group. 2012;486:395–9. https://doi.org/10.1038/nature10933 . Li RQ, Yan L, Zhang L, Ma HX, Wang HW, Bu P, et al. Genomic characterization reveals distinct mutational landscapes and therapeutic implications between different molecular subtypes of triple-negative breast cancer. Sci Rep Nat Publishing Group. 2024;14:12386. https://doi.org/10.1038/s41598-024-62991-3 . Mayakonda A, Lin D-C, Assenov Y, Plass C, Koeffler HP. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res. 2018;28:1747–56. https://doi.org/10.1101/gr.239244.118 . Sanchez-Vega F, Mina M, Armenia J, Chatila WK, Luna A, La KC, et al. Oncogenic Signaling Pathways in The Cancer Genome Atlas. Cell. 2018;173:321–e33710. https://doi.org/10.1016/j.cell.2018.03.035 . Fantini D, Vidimar V, Yu Y, Condello S, Meeks JJ. MutSignatures: an R package for extraction and analysis of cancer mutational signatures. Sci Rep Nat Publishing Group. 2020;10:18217. https://doi.org/10.1038/s41598-020-75062-0 . Alexandrov LB, Kim J, Haradhvala NJ, Huang MN, Tian Ng AW, Wu Y, et al. The repertoire of mutational signatures in human cancer. Nat Nat Publishing Group. 2020;578:94–101. https://doi.org/10.1038/s41586-020-1943-3 . Sondka Z, Dhir NB, Carvalho-Silva D, Jupe S, Madhumita, McLaren K, et al. COSMIC: a curated database of somatic variants and clinical data for cancer. Nucleic Acids Res. 2023;52:D1210–7. https://doi.org/10.1093/nar/gkad986 . Martincorena I, Raine KM, Gerstung M, Dawson KJ, Haase K, Loo PV, et al. Universal Patterns of Selection in Cancer and Somatic Tissues. Cell Elsevier. 2017;171:1029–e104121. https://doi.org/10.1016/j.cell.2017.09.042 . Mularoni L, Sabarinathan R, Deu-Pons J, Gonzalez-Perez A, López-Bigas N. OncodriveFML: a general framework to identify coding and non-coding regions with cancer driver mutations. Genome Biol. 2016;17:128. https://doi.org/10.1186/s13059-016-0994-0 . Chen X, Schulz-Trieglaff O, Shaw R, Barnes B, Schlesinger F, Källberg M, et al. Manta: rapid detection of structural variants and indels for germline and cancer sequencing applications. Bioinformatics. 2016;32:1220–2. https://doi.org/10.1093/bioinformatics/btv710 . Van Loo P, Nordgard SH, Lingjærde OC, Russnes HG, Rye IH, Sun W et al. Allele-specific copy number analysis of tumors. Proc Natl Acad Sci. Proceedings of the National Academy of Sciences; 2010;107:16910–5. https://doi.org/10.1073/pnas.1009843107 Amemiya HM, Kundaje A, Boyle AP. The ENCODE Blacklist: Identification of Problematic Regions of the Genome. Sci Rep Nat Publishing Group. 2019;9:9354. https://doi.org/10.1038/s41598-019-45839-z . Gao C, Xiao G, Bargonetti J. Contemplations on MDMX (MDM4) driving triple negative breast cancer circulating tumor cells and metastasis. Oncotarget. 2019;10:5007–10. https://doi.org/10.18632/oncotarget.27134 . Mermel CH, Schumacher SE, Hill B, Meyerson ML, Beroukhim R, Getz G. GISTIC2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers. Genome Biol. 2011;12:R41. https://doi.org/10.1186/gb-2011-12-4-r41 . Sztupinszki Z, Diossy M, Krzystanek M, Reiniger L, Csabai I, Favero F, et al. Migrating the SNP array-based homologous recombination deficiency measures to next generation sequencing data of breast cancer. Npj Breast Cancer Nat Publishing Group. 2018;4:1–4. https://doi.org/10.1038/s41523-018-0066-6 . Favero F, Joshi T, Marquard AM, Birkbak NJ, Krzystanek M, Li Q, et al. Sequenza: allele-specific copy number and mutation profiles from tumor sequencing data. Ann Oncol. 2015;26:64–70. https://doi.org/10.1093/annonc/mdu479 . Wei T, Simko V. R package corrplot: Visualization of a Correlation Matrix [Internet]. 2024. https://github.com/taiyun/corrplot Mina M, Raynaud F, Tavernari D, Battistello E, Sungalee S, Saghafinia S, et al. Conditional Selection of Genomic Alterations Dictates Cancer Evolution and Oncogenic Dependencies. Cancer Cell Elsevier. 2017;32:155–e1686. https://doi.org/10.1016/j.ccell.2017.06.010 . Mina M, Iyer A, Tavernari D, Raynaud F, Ciriello G. Discovering functional evolutionary dependencies in human cancers. Nat Genet Nat Publishing Group. 2020;52:1198–207. https://doi.org/10.1038/s41588-020-0703-5 . Therneau TM. A Package for Survival Analysis in R [Internet]. 2024. https://CRAN.R-project.org/package=survival Therneau TM, Grambsch PM. Modeling survival data: extending the Cox model. New York: Springer; 2001. Sjoberg DD, Baillie M, Fruechtenicht C, Haesendonckx S, Treis T. ggsurvfit: Flexible Time-to-Event Figures [Internet]. 2025. https://github.com/pharmaverse/ggsurvfit Seabold S, Perktold J, statsmodels. Econometric and statistical modeling with python. 9th Python Sci Conf. 2010. Jolly E. Pymer4: Connecting R and Python for Linear Mixed Modeling. J Open Source Softw. 2018;3:862. https://doi.org/10.21105/joss.00862 . Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw. 2015;67:1–48. https://doi.org/10.18637/jss.v067.i01 . Leon-Ferre RA, Jonas SF, Salgado R, Loi S, de Jong V, Carter JM, et al. Tumor-Infiltrating Lymphocytes in Triple-Negative Breast Cancer. JAMA. 2024;331:1135–44. https://doi.org/10.1001/jama.2024.3056 . Giacchetti S, Faucheux L, Gardair C, Cuvier C, de Roquancourt A, Campedel L, et al. Negative Relationship between Post-Treatment Stromal Tumor-Infiltrating Lymphocyte (TIL) and Survival in Triple-Negative Breast Cancer Patients Treated with Dose-Dense Dose-Intense NeoAdjuvant Chemotherapy. Cancers. 2022;14:1331. https://doi.org/10.3390/cancers14051331 . Mayca Pozo F, Geng X, Tamagno I, Jackson MW, Heimsath EG, Hammer JA, et al. MYO10 drives genomic instability and inflammation in cancer. Sci Adv Am Association Advancement Sci. 2021;7:eabg6908. https://doi.org/10.1126/sciadv.abg6908 . Cao R, Chen J, Zhang X, Zhai Y, Qing X, Xing W, et al. Elevated expression of myosin X in tumours contributes to breast cancer aggressiveness and metastasis. Br J Cancer Nat Publishing Group. 2014;111:539–50. https://doi.org/10.1038/bjc.2014.298 . Zimmerli D, Brambillasca CS, Talens F, Bhin J, Linstra R, Romanens L, et al. MYC promotes immune-suppression in triple-negative breast cancer via inhibition of interferon signaling. Nat Commun Nat Publishing Group. 2022;13:6579. https://doi.org/10.1038/s41467-022-34000-6 . Amato O, Buisseret L, Gebhart G, Plouznikoff N, Larsimont D, Awada A, et al. PIK3CA copy-number gain and inhibitors of the PI3K/AKT/mTOR pathway in triple-negative breast cancer. Cold Spring Harb Mol Case Stud. 2023;9:a006255. https://doi.org/10.1101/mcs.a006255 . Zhao Z-M, Yost SE, Hutchinson KE, Li SM, Yuan Y-C, Noorbakhsh J, et al. CCNE1 amplification is associated with poor prognosis in patients with triple negative breast cancer. BMC Cancer. 2019;19:96. https://doi.org/10.1186/s12885-019-5290-4 . Natrajan R, Mackay A, Wilkerson PM, Lambros MB, Wetterskog D, Arnedos M, et al. Functional characterization of the 19q12 amplicon in grade III breast cancers. Breast Cancer Res. 2012;14:R53. https://doi.org/10.1186/bcr3154 . Xing F, Liu Y, Wu S-Y, Wu K, Sharma S, Mo Y-Y, et al. Loss of XIST in breast cancer activates MSN-c-Met and reprograms microglia via exosomal microRNA to promote brain metastasis. Cancer Res. 2018;78:4316–30. https://doi.org/10.1158/0008-5472.CAN-18-1102 . Girish V, Lakhani AA, Thompson SL, Scaduto CM, Brown LM, Hagenson RA, et al. Oncogene-like addiction to aneuploidy in human cancers. Sci Am Association Advancement Sci. 2023;0:eadg4521. https://doi.org/10.1126/science.adg4521 . McPherson A, Vázquez-García I, Myers MA, Zatzman M, Al-Rawi D, Weiner A et al. Ongoing genome doubling promotes evolvability and immune dysregulation in ovarian cancer [Internet]. bioRxiv; 2024 [cited 2025 May 15]. p. 2024.07.11.602772. https://doi.org/10.1101/2024.07.11.602772 Ogden A, Rida PCG, Knudsen BS, Kucuk O, Aneja R. Docetaxel-induced polyploidization may underlie chemoresistance and disease relapse. Cancer Lett. 2015;367:89–92. https://doi.org/10.1016/j.canlet.2015.06.025 . Hinz N, Jücker M. Distinct functions of AKT isoforms in breast cancer: a comprehensive review. Cell Commun Signal. 2019;17:154. https://doi.org/10.1186/s12964-019-0450-3 . Ewing A, Meynert A, Silk R, Aitken S, Bendixsen DP, Churchman M, et al. Divergent trajectories to structural diversity impact patient survival in high grade serous ovarian cancer. Nat Commun Nat Publishing Group. 2025;16:5586. https://doi.org/10.1038/s41467-025-60655-y . Xiao J-F, Kua L-F, Ding L-W, Sun Q-Y, Myint KN, Chia X-R, et al. KDM6A Depletion in Breast Epithelial Cells Leads to Reduced Sensitivity to Anticancer Agents and Increased TGFβ Activity. Mol Cancer Res MCR. 2022;20:637–49. https://doi.org/10.1158/1541-7786.MCR-21-0402 . Belli C, Duso BA, Ferraro E, Curigliano G. Homologous recombination deficiency in triple negative breast cancer. Breast. 2019;45:15–21. https://doi.org/10.1016/j.breast.2019.02.007 . Sharma P, Barlow WE, Godwin AK, Pathak H, Isakova K, Williams D, et al. Impact of homologous recombination deficiency biomarkers on outcomes in patients with triple-negative breast cancer treated with adjuvant doxorubicin and cyclophosphamide (SWOG S9313). Ann Oncol. 2018;29:654–60. https://doi.org/10.1093/annonc/mdx821 . Zhang X, Goedegebuure SP, Chen MY, Mishra R, Zhang F, Yu YY, et al. Neoantigen DNA vaccines are safe, feasible, and induce neoantigen-specific immune responses in triple-negative breast cancer patients. Genome Med. 2024;16:131. https://doi.org/10.1186/s13073-024-01388-3 . Cohen-Sharir Y, McFarland JM, Abdusamad M, Marquis C, Bernhard SV, Kazachkova M, et al. Aneuploidy renders cancer cells vulnerable to mitotic checkpoint inhibition. Nature. 2021;590:486–91. https://doi.org/10.1038/s41586-020-03114-6 . Quinton RJ, DiDomizio A, Vittoria MA, Kotýnková K, Ticas CJ, Patel S, et al. Whole-genome doubling confers unique genetic vulnerabilities on tumour cells. Nature. 2021;590:492–7. https://doi.org/10.1038/s41586-020-03133-3 . Additional Declarations No competing interests reported. Supplementary Files SuppTablesBendixsenetal2025.xlsx SuppFiguresBendixsenetal2025.docx Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 07 May, 2026 Reviews received at journal 30 Apr, 2026 Reviews received at journal 13 Apr, 2026 Reviewers agreed at journal 13 Apr, 2026 Reviewers agreed at journal 13 Apr, 2026 Reviews received at journal 07 Apr, 2026 Reviewers agreed at journal 02 Apr, 2026 Reviewers invited by journal 02 Apr, 2026 Editor assigned by journal 31 Mar, 2026 Submission checks completed at journal 25 Mar, 2026 First submitted to journal 24 Mar, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9212020","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":618272081,"identity":"e0114b61-0440-46e3-a01a-dc6ca9ec58c7","order_by":0,"name":"Devin P Bendixsen","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Devin","middleName":"P","lastName":"Bendixsen","suffix":""},{"id":618272082,"identity":"4d3b7cb6-ec1d-46e5-8e99-235d589306a3","order_by":1,"name":"Fiona Semple","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Fiona","middleName":"","lastName":"Semple","suffix":""},{"id":618272083,"identity":"cb579e78-62f9-4ce8-aeac-075eea7dbe6d","order_by":2,"name":"Alastair Ironside","email":"","orcid":"","institution":"Western General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Alastair","middleName":"","lastName":"Ironside","suffix":""},{"id":618272084,"identity":"e0e2229e-34a4-435b-b7af-90c791fc8a7d","order_by":3,"name":"Mustafa İsmail Özkaraca","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Mustafa","middleName":"İsmail","lastName":"Özkaraca","suffix":""},{"id":618272085,"identity":"c0f82f7f-72eb-4846-afdd-bbce72014572","order_by":4,"name":"Natalie Wilson","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Natalie","middleName":"","lastName":"Wilson","suffix":""},{"id":618272086,"identity":"d411a7ef-e4d5-47ff-a845-5f3a6a76218c","order_by":5,"name":"Alison Meynert","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Alison","middleName":"","lastName":"Meynert","suffix":""},{"id":618272087,"identity":"56334b94-4c0f-49cf-bd8c-3b934ec70872","order_by":6,"name":"Ailith Ewing","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Ailith","middleName":"","lastName":"Ewing","suffix":""},{"id":618272088,"identity":"a0a9685f-42f2-4c12-8598-c62016480494","order_by":7,"name":"Colin A Semple","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Colin","middleName":"A","lastName":"Semple","suffix":""},{"id":618272089,"identity":"d01e6e12-6493-4b58-abcc-e1f16b98d088","order_by":8,"name":"Olga Oikonomidou","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBUlEQVRIiWNgGAWjYDACdjApYQCmEtgYGPgZGCAcCVxamNG1SDYQpwWqigGoxeAAAS38zOwXPzBUWBjzSzcf3fCgzCZx8/HmDQw/ahgSZzZg1yLZzFMswXBGwkxyzrG0Gwnn0hK3nTlWwNhzjCFxNg5bDA7zJEgwtknYGNzIMbuR2HY4cduNHAMG3gaGxHm4tST/QNGyef4bA8a/eLWwHwPZYgbXskGCx4AZZAsuhwH9wmaRcEbCWHJGGtgvxjPOpBUcljkmYYzL+/zs7Y9vfKioM+yXSD5280eZjWx/++GND9/U2MjOOIDDGgYeA4YEdLEDeCISCNgf4JEcBaNgFIyCUQAEAMPKWwVsVw6DAAAAAElFTkSuQmCC","orcid":"","institution":"University of Edinburgh","correspondingAuthor":true,"prefix":"","firstName":"Olga","middleName":"","lastName":"Oikonomidou","suffix":""}],"badges":[],"createdAt":"2026-03-24 12:40:28","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9212020/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9212020/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106534458,"identity":"24f8607d-671d-4260-bc09-39aa98f51345","added_by":"auto","created_at":"2026-04-09 15:03:46","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1093852,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStudy overview and tumour mutational burden in triple-negative breast cancer. (A) \u003c/strong\u003eLongitudinal samples were collected over several years for each patient, including a blood sample for germline DNA, a PreT biopsy before commencement of NACT and a MidT biopsy (in 4 patients). Multiple regions (n=1-4) of the PostT and R tumour samples were sequenced.\u003cstrong\u003e (B) \u003c/strong\u003ePatients (n=22) were selected based on response to NACT; good response = residual cancer burden score of RCB 0/I, poor response = RCB II/III. Samples (n=96) are represented by coloured rectangles, yellow = germline, orange = PreT, orange with grey triangle =MidT, blue = PostT, light blue = R. The number of regions sampled per timepoint per patient are indicated in the rectangles (n=0 means no sample was available due to complete response to therapy or unattainability at recurrence sites). \u003cstrong\u003e(C)\u003c/strong\u003e The percentage (0-100%) TILs found in each tumour sample, at each timepoint, orange = PreT, grey = MidT, blue = PostT, light blue = R, is shown for each patient with TNBC patient number indicated at the top. A patient number on top of a T-bar indicates where multiregional samples were assayed from that patient. Vertical dotted lines within each timepoint separate RCB classes. \u003cstrong\u003eD) \u003c/strong\u003eTMB calculated as SNVs per Mb for each patient across timepoints, orange = PreT, grey = MidT, blue = PostT, light blue = R. The proportion of SNVs retained from PreT across subsequent timepoints is shown. Increasing variant retention percentages are indicated on a scale of white to black (0-60%). \u003cstrong\u003e(E) \u003c/strong\u003eMSI scores for each sample across the time course, dots representing scores between 0 and 15 are coloured, as above to indicate different timepoints. \u003cstrong\u003e(F)\u003c/strong\u003eMutational signatures found using \u003cem\u003ede novo\u003c/em\u003e extraction (top) or by estimating the contribution of known COSMIC SBS mutational signatures (bottom).\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9212020/v1/725e384ab04e339b70b7b4f7.jpeg"},{"id":106726389,"identity":"61f6bae3-b44d-42f0-ad45-a04a0f0a2e1e","added_by":"auto","created_at":"2026-04-12 18:35:58","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1036314,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLongitudinal SNV landscape of TNBC during NACT. (A) \u003c/strong\u003eOncoplot indicating the presence of deleterious variants predicted to disrupt TNBC gene function. Variants detected for each TNBC patient (patient number indicated at the top) are shown at each timepoint, with variant type indicated by a coloured square (see key, bottom). For PostT and R samples a patient number on top of a T-bar shows the number of multiregional samples assayed from that patient. Variants lost between PreT and PostT or between PreT and R, are denoted by a square with a black outline. Variants not found in PreT that are detected in PostT or R, are denoted by an asterisk. Orange dots indicate genes identified as driver candidates and germline variants are indicated with a solid diagonal line. Where multi-region samples exist for the same timepoint they generally support the existence of the same driver mutation.\u003cstrong\u003e (B)\u003c/strong\u003eSNV prevalence through longitudinal time (PreT, PostT, R) and across RCB classes (0, I, II, III). \u003cstrong\u003e(C)\u003c/strong\u003e Gene diagrams of the six most commonly altered genes comparing the location and prevalence of mutations in the PreT and PostT samples. Variants type is indicated and coloured as in panel A. White squares indicate lost variants and asterisks indicate newly emerging variants occurring between PreT and PostT timepoints. Orange and blue numbers indicate the proportion of patients with altered genes in PreT and PostT respectively. Percentages beneath each diagram indicate the proportion of PostT patients with ≥1 sample with a novel (asterisk), or lost (square) variant, as well as variants observed to be both novel and lost in the cohort (asterisk in square). \u003cstrong\u003e(D)\u003c/strong\u003eProportion of altered oncogenic pathways across RCB classes in PreT samples (top) and through time (bottom).\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9212020/v1/237a26270ef9871f5e3026dd.jpeg"},{"id":106724911,"identity":"8d1455f5-72c5-42b0-ad17-4345080a8679","added_by":"auto","created_at":"2026-04-12 18:30:23","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1129308,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComparison of whole exome sequencing (WES) and CUTseq in detecting CNAs. (A) \u003c/strong\u003eGraphical representation of WES and CUTseq approaches. \u003cstrong\u003e(B)\u003c/strong\u003eExamples of sequencing read depth and coverage of chr8 and chrX for a single patient. Chromosome karyotypes and gene density are shown for context. \u003cstrong\u003e(C) \u003c/strong\u003eLongitudinal CNA calls for two patients (TNBC18 and TNBC11) indicating the consistent patterns exhibited between WES and CUTseq. Red indicates amplified genomics regions, blue indicates deleted regions, grey indicates diploid regions, black indicates regions with estimated copy number \u0026gt;6, and white indicates regions where copy number was unobtainable. The dashed black line indicates the estimated tumour ploidy. The 4 plots show the estimated copy number (y-axis) for each chromosome (x-axis) for the PreT sample of the 2 selected patients. Below this, amplifications and deletions are summarised for all timepoints, including multiregional samples, as indicated on the left. Estimated tumour ploidy, and the percentage of the genome amplified or deleted from each approach is shown on the right. \u003cstrong\u003e(D) \u003c/strong\u003ePairwise sample comparisons of CUTseq and WES for number of sequencing reads, estimated tumour ploidy, percentage of genome amplified, deleted and the total altered by CNAs.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9212020/v1/e1ca8bc9a531c498974fca25.jpeg"},{"id":106725891,"identity":"a629cbed-053b-4a92-800a-454350353502","added_by":"auto","created_at":"2026-04-12 18:34:19","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1688435,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe fate of chromosome arm aneuploidies during NACT. (A) \u003c/strong\u003eThe prevalence of aneuploidy for each chromosome arm (rows) longitudionally (four columns representing timepoints: PreT, MidT, PostT and R) are shown for each patient. For PostT and R samples a patient number on top of a T-bar shows the number of multiregional samples assayed from that patient. A chromosome arm was determined to be aneuploid if \u0026gt;60% of the total chromosome arm length was amplified (red) or deleted (blue) by the union CNA calls of WES and CUTseq. Where methods made different calls, the square contains both red and blue. Percentage of the genome with aneuploidies is indicated by bar charts above each column. Aneuploidies lost between PreT and subsequent time points are denoted by a square with a black outline, and newly emerging aneuploidies in MidT, PostT and R, that are not detected in PreT, are indicated with asterisks. Samples estimated to have undergone WGD are indicated with a grey circle. \u003cstrong\u003e(B) \u003c/strong\u003eThe prevalence of chromosome arm aneuploidies at each timepoint (PreT, PostT, R) and for each RCB class (0, I, II III). Size of circles indicate proportion of patients in each category with amplifications (red) and deletions (blue). Grey background and asterisk indicate significant variation in prevalence (amplification or deletion) between time points using Chi-squared tests (p≤0.05 using Bonferonni correction for multiple testing). \u003cstrong\u003e(C) \u003c/strong\u003eGenomic instability measured as the proportion of each genome amplified (AMP), deleted (DEL) or combination of amplified and deleted (Total) across the time course (left) and RCB class (right). Statistical significance was determined using the Kruskal-Wallis rank sum test, and pairwise Wilcoxon rank sum test. Significance levels are indicated as follows: *p \u0026lt; 0.05; **p \u0026lt; 0.01. \u003cstrong\u003e(D) \u003c/strong\u003eProportion of patients with WGD defined as ploidy ≥3. The left panel indicates proportion of patients at each timepoint(PreT, PostT and R) with WGD. The right panel indicates the proportions of patients with WGD at PreT for RCB classes (0, I, II, III).\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9212020/v1/7947e52e2efc64f3f60e786b.jpeg"},{"id":106959675,"identity":"ce0f8e20-f020-459a-90bf-881c60e1f89c","added_by":"auto","created_at":"2026-04-15 09:13:25","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":400720,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGenes with recurrent CNAs during NACT (A) \u003c/strong\u003eOncoplot indicating prevalence of amplification (red) and deletion (blue) of genes previously associated with TNBC progression. Alterations lost between PreT and subsequent time points are denoted by a square with a black outline and newly emerging alterations, in MidT, PostT and R, that are not detected in PreT, are indicated with asterisks. Squares showing both red and blue indicate where WES and CUTSeq disagree over the presence of a newly emerging CNA. RCB class and TNBC patient number are shown along the top, a patient number on top of a T-bar shows the number of multiregional samples assayed from that patient. Orange dots in \u003cstrong\u003e(A)\u003c/strong\u003e, \u003cstrong\u003e(B) \u003c/strong\u003eand\u003cstrong\u003e (C) \u003c/strong\u003eindicate genes identified in CNA hotspots in enrichment analysis (Table S10). \u003cstrong\u003e(B) \u003c/strong\u003ePrevalence of gene alterations at each timepoint (PreT, PostT, R) and across RCB classes (0, I, II III). Size of circles indicate proportion of patients in each category with gene amplifications (red) and deletions (blue) and a grey background and asterisk indicate significant variation in prevalence (gene amplification or deletion) between time points using Chi-squared tests (p≤0.05 using Bonferonni correction for multiple testing) \u003cstrong\u003e(C) \u003c/strong\u003ePrevalence of amplification (red) and deletion (blue) in the top 15 altered genes identified in panel A, split according to WGD status of PreT samples.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-9212020/v1/40fc9c6ac53ec4499dd5d9b0.png"},{"id":106534466,"identity":"33d09378-c607-4501-b97a-e5da16e13333","added_by":"auto","created_at":"2026-04-09 15:03:46","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":667854,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCorrelations among mutational and clinical variables. (A) \u003c/strong\u003ePairwise correlations between genomic alterations and other features of interest in PreT samples, indicated by an all-against-all correlation matrix with strength of correlation indicated by size of coloured circles (bottom left). Additional scatter plots depict the relationships between selected variables showing significant correlations. \u003cstrong\u003e(B) \u003c/strong\u003eKaplan-Meier curves depicting relationship between RCB class, HRD score and patient OS. Hazard ratios (HR) and p-values are derived from Cox Proportional-Hazard models.\u003cstrong\u003e (C) \u003c/strong\u003eForest plot of coefficients derived from LMM (for continuous variables) or GLMM (for binary variables) for the complete trajectory of each molecular or clinical feature over time, indicating features increasing (positive coefficients) or decreasing (negative) over time. Coefficients for PostT (blue) and recurrence (light blue) are shown relative to PreT samples. \u003cstrong\u003e(D) \u003c/strong\u003ePloidy estimates from ASCAT.sc for each sample across time using WES and CUTseq. Median ploidy estimates for WES and CUTseq respectively, were PreT: 2.05 and 2.1, PostT: 2.0 and 2.0 and R: 2.45 and 3.0. Asterisks indicate statistical significance using Pairwise Wilcoxon Rank Sum Tests (adjusted p-values \u0026lt;0.01=**, \u0026lt;0.05=*). \u003cstrong\u003e(E) \u003c/strong\u003eThe proportions of samples (PreT, PostT or R) showing evidence of WGD, based upon various thresholds for the mean fraction of genome amplified across WES and CUTseq approaches.\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9212020/v1/250bfcfb4568c0f0d17e230e.jpeg"},{"id":106994282,"identity":"9fc6790e-502f-4747-8280-f0c7f695220b","added_by":"auto","created_at":"2026-04-15 15:07:17","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":7314715,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9212020/v1/032897f9-61f9-4bdd-b877-d05936f33c96.pdf"},{"id":106725126,"identity":"434cd72f-255a-49a8-8d17-e6f937fdc163","added_by":"auto","created_at":"2026-04-12 18:31:30","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":149492,"visible":true,"origin":"","legend":"","description":"","filename":"SuppTablesBendixsenetal2025.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-9212020/v1/d602cff79bab225d8766c971.xlsx"},{"id":106534460,"identity":"061edb82-457c-42e0-b900-6a1acb62e368","added_by":"auto","created_at":"2026-04-09 15:03:46","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":6327032,"visible":true,"origin":"","legend":"","description":"","filename":"SuppFiguresBendixsenetal2025.docx","url":"https://assets-eu.researchsquare.com/files/rs-9212020/v1/164a678945649d776738fd2a.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Frequent Whole-Genome Duplication Events Drive the Genomic Evolution of Triple-Negative Breast Cancer During Neoadjuvant Chemotherapy","fulltext":[{"header":"Background","content":"\u003cp\u003eTriple-negative breast cancer (TNBC) accounts for 15\u0026ndash;20% of breast cancer cases, and is associated with a high rate of relapse and the worst five-year survival rate of all breast cancer subtypes. For many years, targeted therapeutic options for TNBC patients were limited as TNBCs lack expression of the oestrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2), which are essential targets for endocrine and biological therapies. However, more recently TNBC treatments have evolved significantly to include the use of immunotherapy, antibody-drug conjugates and PARP inhibitors combined with traditional chemotherapy, thereby increasing the treatment options beyond targeted treagtments.\u003c/p\u003e \u003cp\u003eCollectively, TNBCs are characterised by high levels of genetic instability resulting in complex patterns of copy number alterations (CNAs) and structural rearrangements [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Whole genome duplication (WGD) has been shown to be an independent predictor for poor survival in many cancers including estrogen receptor-positive breast cancers [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Although WGD has been shown to occur in TNBC [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e], the consequences for tumour evolution and patient outcomes are essentially unstudied. Increased tumour infiltrating lymphocytes (TILs), higher PD-L1 (programmed death ligand 1) expression levels and deficiency in homologous recombination (HRD) are more commonly seen in TNBC compared with other breast cancer subtypes [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Differences in molecular subtypes, immune composition and mutational landscapes, underlie the heterogeneity of TNBC. Four TNBC subtypes with unique biology and transcriptional diversity have been identified; basal-like (BL), mesenchymal (MES), luminal androgen receptor (LAR) and immunomodulatory (IM) [\u003cspan additionalcitationids=\"CR6 CR7\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Each subtype varies in immune composition with improved outcomes reported for TNBC patients with increased expression of immune gene signatures and increased TILs [\u003cspan additionalcitationids=\"CR10\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFor the TNBC patients with known germline \u003cem\u003eBRCA1/2\u003c/em\u003e mutations (10%), additional treatment with poly ADP ribose polymerase (PARP) inhibitors can be beneficial [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Immunotherapy targeting immune checkpoint inhibitors (ICIs) has revolutionised the treatment of TNBC patients, but response is suboptimal in some patients, therefore the search for further biomarkers of response beyond PD-L1 is ongoing [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. TNBC patients often respond well to neoadjuvant chemotherapy (NACT), with 34% of patients achieving pathological complete response (pCR) which is subsequently associated with 75% lower risk of recurrence, compared to TNBC patients not achieving pCR [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. The recent Keynote-522 trial demonstrated that combining the ICI Pembrolizumab with conventional chemotherapy significantly increased pCR rates with 64.8% of patients acheiving pCR in the experimental arm compared to 51.2% in the placebo arm. Furthermore, the addition of Pembrolizumab, increased event-free survival rates to 84.5% compared to 76.8% in the control arm, with significant effects on overall survival [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. High pCR rates have been associated with higher tumour mutational burden (TMB), elevated levels of PD-L1 expression and increased levels of TILs in the tumour microenvironment [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Defining the landscape of TNBC and the molecular changes that occur in response to NACT, will allow stratification of patients for directed treatment. However, this is confounded by the high genomic instability and extensive intra-tumour heterogeneity (ITH) that are thought to drive the pathogenesis, treatment resistance and metastatic spread of TNBC [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. When and how these genetic alterations emerge is not known, but WGD is thought to be common in TNBC [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], while aneuploidies and copy number alterations (CNAs) have been implicated as drivers of the adaptive response to chemotherapy and metastasis [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eStudies including multiple samples over time have the potential to reveal new tumour biology, but can also entail large investments of time, effort and resources. There is consequently a pressing need for accessible, affordable assays with relatively modest requirements for computationally intensive analysis. For example, although the mutational landscape of TNBC has been examined in previous studies (e.g. Lehmann et al, 2021), the temporal dynamics of CNA and short nucleotide variants (SNVs including single nucleotide substitutions and indels) and WGD during disease progression are poorly studied. Consequently, we lack a detailed account of tumour evolution as patients are diagnosed, undergo NACT, and their disease enters remission or recurs. Here, we employ deep (~\u0026thinsp;100X) Whole Exome Sequencing (WES) to assay recurrent SNVs and CNAs in patient TNBC samples to identify variants that play a role in disease progression and response to treatment. In parallel, we use CUTseq [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e], a reduced representation Whole Genome Sequencing (WGS) approach, to comprehensively identify copy number alterations genome-wide. In addition to recurrent SNVs in known TNBC driver genes we demonstrate major shifts in CNAs throughout the genome following NACT, suggesting new biomarkers of treatment response and revealing critical roles for WGD in tumour evolution.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003ePatient and Tumour samples\u003c/h2\u003e \u003cp\u003eSamples from TNBC patients treated at the Western General Hospital Cancer Centre, enrolled in the NEO and metcfDNA studies (references: 2014/0005 and SR680) at Edinburgh University were acquired through the NRS BioResource and Tissue Governance Unit. Patients with TNBC were determined from pathology records and only samples with matched tumour and blood samples were included. One sample with no matched blood was paired with the patient\u0026rsquo;s normal breast sample. Collection of human tissues and blood and relevant clinicopathological data were approved by the South East Scotland Research Ethics Committee 01 with REC references: 13/SS/0236 and 15/ES/0094, and all patients consented to sample collection and use of their data in research. Patients and their clinical data (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) were anonymised. All patients underwent NACT with anthracyclines followed by taxanes in combination with Carboplatin, and response was assessed using the internationally recognised residual cancer burden (RCB) scoring index. This system provides a continuous score (RCB score) and categorises patients into four discrete groups (RCB 0, RCB I, RCB II, RCB III) based on the amount of residual disease following NACT. RCB-0 corresponds to no residual disease, also known as pathological complete response (pCR), RCB-1 denotes a small amount of residual invasive cancer, RCB-II a moderate amount of residual invasive cancer and RCB-III indicates substantial residual disease [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Samples from before treatment with NACT (PreT), at the end of NACT after surgery (PostT), where available from recurrent tumours (R) and from selected mid NACT biopsies (at completion of anthacyclines before commencing taxanes/carboplatin) (MidT) were collected. Tumour infiltrating lymphocytes (TILs) were assessed by a breast cancer consultant pathologist (AI) using light microscopy on routine diagnostic H\u0026amp;E-stained glass slides. The number of TILs were quantified as a percentage of the tumour-associated stroma occupied by lymphocytes, as per the recommendations of The International Immuno-Oncology Biomarker Working Group on Breast Cancer. Three distinct regions of predefined size for each tumour were evaluated for TILs using a visual reference image index of pre-determined densities [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eDNA extraction\u003c/h3\u003e\u003cp\u003eFormalin fixed paraffin embedded (FFPE) samples were sectioned onto slides using a cryostat. An initial 4\u0026micro;m section was stained with hematoxylin and eosin (H\u0026amp;E) and reviewed by a breast cancer consultant pathologist (AI) to determine sample tumour cellularity and mark up the tumour regions (Table S2). DNA was extracted (using Qiagen QIAamp DNA FFPE Advanced UNG Kit #56704) from 10\u0026micro;m sections, 2 to 5 sections per sample depending on tumour cellularity. Extraction was carried out on marked up tumour enriched areas and where possible, multiregional sampling was carried out by extracting DNA from spatially distinct tumour enriched areas on each slide, or from slides prepared from different FFPE blocks from the same tumour sample. Germline DNA was extracted from whole blood collected before commencement of NAC using the Qiagen DNeasy Blood \u0026amp; Tissue Kit (#69504). DNA was quantified using the Qubit dsDNA HS Assay Kit (#Q32851).\u003c/p\u003e\n\u003ch3\u003eWhole Exome Sequencing\u003c/h3\u003e\n\u003cp\u003eThe quality of the DNA extracted from the FFPE samples was assessed with the Illumina FFPE QC Kit (Illumina Inc, #15013664) according to the manufacturer\u0026rsquo;s protocol. The required ΔCq\u0026thinsp;\u0026le;\u0026thinsp;5 was achieved by all samples determining suitability for library preparation. For each sample, 50ng of DNA sample was used for library preparation with the Illumina DNA Prep with Enrichment, (S) Tagmentation Kit (Illumina #20025524) using IDT\u0026reg; for Illumina\u0026reg; DNA/RNA UD Indexes Set A, Tagmentation (Illumina #20027213) according to the manufacturer's protocol with modifications for FFPE DNA where appropriate. Using the Twist Comprehensive Exome panel (Twist Biosciences #102032), 500ng of each pre-enrichment gDNA library was pooled in sets of 12 and hybridised to the targeted regions during an overnight incubation, followed by capture onto magnetic beads and amplification of the exome enriched library pools, which were then purified with Agencourt AMPure XP beads (Beckman Coulter, #A36881). Libraries were quantified using Qubit dsDNA HS assay and quality assessed using Agilent Bioanalysis with the DNA HS Kit (#5067\u0026thinsp;\u0026minus;\u0026thinsp;4626). Fragment size and quantity measurements were used to calculate molarity for each enrichment library pool. Sequencing was performed on the NextSeq 2000 platform (Illumina Inc, #20038897) using NextSeq 2000 P3 Reagents (200 Cycles) (#20040560). A single pool containing the 8 enrichment pools was sequenced on 3 separate P3 flow cells to achieve a minimum of 100X coverage. PhiX Control v3 (Illumina, # FC-110- 3001) was spiked into each run at a concentration of 1%. Library preparation and sequencing were carried out at the Clinical Research Facility (CRF), Edinburgh, UK.\u003c/p\u003e\n\u003ch3\u003eReduced Representation Whole Genome Sequencing – CUTseq\u003c/h3\u003e\n\u003cp\u003eMultiplexed DNA sequencing libraries were generated using CUTseq [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] following the published protocol (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://protocolexchange.researchsquare.com/article/pex-25/v3\u003c/span\u003e\u003cspan address=\"https://protocolexchange.researchsquare.com/article/pex-25/v3\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e)\u003c/span\u003e with several modifications. Firstly, adapters (A1-A48) specific to the HindIII cut site which incorporate sample-specific barcodes, the RA5 Illumina sequencing adapter and the T7 promoter sequence were purchased from Integrated DNA Technologies (IDT) (Table S3). Briefly, forward strand barcode adapters were 5\u0026rsquo; phosphorylated then annealed to the reverse strand barcode adapters by incubation at 95\u0026deg;C with step down cooling to 25\u0026deg;C over the course of 45min to form double-stranded(ds)-barcoded HindIII-specific adapters. DNA (10-50ng) from patient samples was fragmented by enzymatic digestion with HindIII-HF\u0026reg; (NEB, # R3104L) for 4hr at 37\u0026deg;C in a 10\u0026micro;l volume of CutSmart\u0026reg; Buffer (NEB, #B6004). Generally, 48 samples were prepared at the same time. Digested DNA samples were barcoded using T4 DNA ligase (NEB, # M0202T) by incubating with ds-barcoded HindIII adapters at 16\u0026deg;C for 18hr. Samples were combined into 3 pools of 16 samples, each containing\u0026thinsp;~\u0026thinsp;300-500ng starting DNA, and purified using Agencourt AMPure XP (Beckman Coulter, #A63880) at a ratio of 0.9X beads to sample. The 3 pools were combined and fragmented with target peak size 200bp using the Covaris E220 Evolution. Sample concentration was determined using Agilent Technologies DNA HS Kit (#5067\u0026thinsp;\u0026minus;\u0026thinsp;4626), followed by evaporation to ~\u0026thinsp;20ng/\u0026micro;l. In vitro transcription was carried out with 8\u0026micro;l of sample (~\u0026thinsp;160ng) using the MEGAscript\u0026reg; T7 Transcription Kit (ThermoFisher Scientific, #AM1334-5) following the manufacturer\u0026rsquo;s protocol, and the resulting RNA was purified using Agencourt RNAClean XP beads (Beckman Coulter, #A63987) at a ratio of 1.8X beads to sample. For RNA sequencing library preparation, primers (RTP, RP1 and RPIndex) and adapter (RA3) were purchased as standard desalted oligos from IDT (Table S3). RNA was quantified using the NanoDrop Spectrophotometer (Thermo Scientific) subsequently 1\u0026micro;g of RNA was used for library prep. Briefly, the RA3 adapter was ligated using T4 RNA ligase 2, deletion mutant (Cambio Ltd, #LR2D1132K) at 25\u0026deg;C for 2hr, followed by reverse transcription using the RTP primer and SuperScript IV RT (Invitrogen, #18090050) at 50\u0026deg;C for 1hr. Each library was then amplified using NEBNext\u0026reg; UltraTM II PCR master mix (#M0544L) and indexed using Illumina primers (RPI2 and RPI3) to allow simultaneous sequencing of two 48 sample libraries. After double-sided size selection purification using AMPure XP beads, final library quantification and size distribution were carried out by Qubit dsDNA HS assay and Agilent Bioanalysis (DNA HS Kit, #5067\u0026thinsp;\u0026minus;\u0026thinsp;4626). Sequencing (1x131bp) was performed on the NextSeq 2000 platform (Illumina Inc, #SY- 415\u0026ndash;1002) at the CRF, Edinburgh, using the NextSeq 1000/2000 P3 Reagents (100 cycles) v3 Kit (#20040599) with PhiX Control v3 (Illumina, #FC-110-3001) spiked in at a concentration of 5%.\u003c/p\u003e\n\u003ch3\u003ePrimary processing of WES and CUTseq sequencing data\u003c/h3\u003e\n\u003cp\u003eThe raw sequencing reads for the WES and CUTseq datasets were assessed and processed independently. Sequencing reads generated by CUTseq were pre-processed using an established pipeline (\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003egithub.com/ljwharbers/copynumber-calling\u003c/span\u003e). Briefly, reads were demultiplexed into individual samples based on the internal CUTseq 8 bp barcodes allowing for a single mismatch. The adapters were trimmed using fastp v0.23.2 [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e] and reads were aligned to the GRCh38 reference genome using BWA mem v0.7.17 [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] and SAMtools v1.16.1 [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. Duplicate marking and mapping and alignment quality were assessed using Picard and Alfred v0.2.6 [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. The resulting mapped reads provided a genome-wide readout (at 3,422bp resolution) across all HindIII restriction enzyme cut sites.\u003c/p\u003e \u003cp\u003eThe raw sequencing reads from WES were processed using the nf-core [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e] pipeline Sarek v3.2 for identification of germline and somatic variants [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. Reads were assessed for quality using FastQC v0.11.9 [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e] and MultiQC v1.17 [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e] and adapter and quality-trimmed using fastp v0.23.2 [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e] then aligned to the GRCh38 reference genome using BWA mem v0.7.17 [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] and SAMtools v1.16.1 [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. Base recalibration and duplicate marking was performed using GATK4 v4.3.0.0 [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e] and exome-wide read depth was assessed using mosdepth v0.3.6 [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e].\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eVariant calling and filtering\u003c/h2\u003e \u003cp\u003eTo assess the mutational landscape within our samples, SNVs and small insertions and deletions (indels) were identified from the WES samples. To limit bias in the identification of somatic variants and to rescue potential false negative mutations, SNVs and indels were called as a majority vote between Mutect2 (GATK v4.3.0.0) [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e], Strelka2 v2.9.10 [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e] and VarScan v2.4.6 [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Somatic variants were annotated using Ensembl Variant Effect Predictor (VEP) v107 [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. CaMutQC v1.1.0 [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e] was used to filter somatic variants based on sequencing quality thresholds, strand bias or the presence of an adjacent indel tag, applying the following parameters: tumour total depth\u0026thinsp;\u0026ge;\u0026thinsp;40, tumour variant depth\u0026thinsp;\u0026ge;\u0026thinsp;5, normal total depth\u0026thinsp;\u0026ge;\u0026thinsp;10, Variant Allele Frequency (VAF)\u0026thinsp;\u0026ge;\u0026thinsp;0.1 and strand bias score\u0026thinsp;\u0026ge;\u0026thinsp;2. Variants occurring in the 1000 Genomes Project dataset [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e] were removed as these were highly likely to be germline. To be retained, a consensus somatic variant (identified by \u0026ge;\u0026thinsp;2/3 variant callers) had to be in a protein coding region, pass CaMutQC filtering thresholds, have an allele frequency\u0026thinsp;\u0026ge;\u0026thinsp;0.1, and be estimated by VEP to have high or moderate impact. For each patient, germline variants were identified from matched blood or normal tissue samples using VarScan v2.4.6 [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e] and annotated using Ensembl VEP v107 [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e] then checked against the ClinVar database v20240107 to identify clinically relevant mutational variants [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Tumour mutational burden was calculated as the number of consensus mutations per Mb and microsatellite instability (MSI) was assessed using MSIsensor-pro v1.3.0 [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eGenes identified as a source of artefacts [\u003cspan additionalcitationids=\"CR46\" citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e] were removed, and mutational loads in TNBC relevant genes [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan additionalcitationids=\"CR49 CR50\" citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e] were assessed in detail (Table S4). The number of alterations in oncogenic signalling pathways was determined using Maftools and OncogenicPathways [\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e]. SNV mutational signatures were extracted using the mutSignatures pipeline (v2.1.5) [\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e]. Preliminary mutational process assessment suggested that two mutational signatures reconstruct the SNV landscape with minimal error (compared to median mutation counts) and was further corroborated by a silhouette plot after extraction. The de novo signatures were then compared to the COSMIC SBS Mutational Signatures v3.4 [\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTo infer candidate driver variants two driver prediction methods with different approaches were applied, dNdScv [\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e] and OncodriveFML [\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e], on SNVs with variant allele frequency (VAF)\u0026thinsp;\u0026gt;\u0026thinsp;0.1 in PreT samples. OncodriveFML v2.2.0 was used with default parameters to identify candidate variants based on significant bias in functional impact of mutations (Q-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05). dNdScv was used with default parameters and identified candidate driver variants based on the ratio of synonymous to non-synonymous mutations (qglobal_cv\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e \u003cp\u003eTo detect structural variants, Manta v1.6.0 [\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e] was used on the WES samples. Five categories of structural variants were assessed: inversions (INV), insertions (INS), deletions (DEL), duplications (DUP), and breakend (BND) variants. To limit the number of false positives, \u0026lsquo;micro-inversions\u0026rsquo; less than 400bp were removed from the analysis.\u003c/p\u003e \u003cp\u003eEach WES and CUTseq sample was assessed for somatic CNA using ASCAT.sc v0.1 [\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e]. This approach can be employed on either shallow-coverage or targeted sequencing and uses average read depth over various bin sizes as opposed to per base pair read depth. Crosetto et al have shown that CUTseq with HindIII digestion yields reproducible copy number profiles at 100Kb resolution but we conservatively adopted a resolution of 300Kb guaranteeing a larger number of CUTseq sites in each bin (the average distance between 2 consecutive HindIII sites is 3.4kb) [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. CNA calls based upon WES or CUTseq were removed if they overlapped a blacklisted region [\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e], known to produce mapping artefacts, by \u0026gt;\u0026thinsp;25% of the CNA length. The union of CNA calls from WES and CUTseq was used for all subsequent analyses. CNA calls were intersected with a list of genes (Table S9) implicated in TNBC progression compiled from the literature [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e]. CNA deletion/amplification hotspots were determined by GISTIC v2.0.23 [\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e] based on WES and CUTseq data for PreT samples. Regions were considered significantly enriched for alterations at a q-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 [\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eHomologous recombination deficiency (HRD) was predicted using scarHRD v0.1.1 [\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e] based on loss of heterozygosity (LOH), large scale transitions (LST) and telomeric allelic imbalances. WES sequencing copy number profiles from Sequenza [\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e] were used as input to scarHRD. WGD events were called where ASCAT.sc estimated ploidy for either WES or CUTseq were \u0026ge;\u0026thinsp;3, which is a more conservative approach than a previous TNBC study using\u0026thinsp;\u0026gt;\u0026thinsp;2.7 [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eAnalysis of co-occurrence or mutual exclusivity\u003c/h3\u003e\n\u003cp\u003eLinear relationships among genomic metrics (i.e. TMB, CNA burden, ploidy, etc) and clinical variables (TIL, RCB, OS), were determined using the R package \u0026lsquo;corrplot\u0026rsquo; [\u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e] using a confidence level of 0.95, adjusted for multiple testing using the Benjamini-Hochberg procedure. Patterns of pairwise co-occurrence or mutual exclusivity were tested using SELECT v1.6.3 [\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e, \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e]. SELECT was run 10 times using 10 different random seeds and the median select score for each pairwise interaction was used. The effects of variables on overall survival were modelled with Cox proportional hazards regression models using the coxph method in the R package \u0026lsquo;survival\u0026rsquo; v3.8.3 [\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e, \u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e] and Kaplan-Meier curves were plotted using R package \u0026lsquo;ggsurvfit\u0026rsquo; v1.1.0 [\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003eAnalysis of mutational patterns across time\u003c/h3\u003e\n\u003cp\u003eWe employed two strategies to examine mutational patterns over time, one based upon differences between mean values at each timepoint, and another considering entire temporal trajectories for each patient, based on linear mixed models (LMM). In the first strategy, for each genomic metric (i.e. TMB, CNA burden, ploidy, etc), significant differences between sample means at each timepoint (PreT, PostT, R) or across RCB classes were assessed using Kruskal-Wallis rank sum tests. In cases of presence/absence (i.e. SNV prevalence or copy number status) a Fisher\u0026rsquo;s exact test was used for two variables and a Chi-squared test for three or more. Pairwise comparisons amongst groups were made using Wilcoxon rank sum tests with Bonferonni corrections for multiple testing. In the second strategy we applied LMM (for continuous variables) or Generalised Linear Mixed Models (GLMM) (for binary dependent variables), treating patients as random effects, to examine complete temporal trajectories. All linear mixed models were implemented in Python using the mixedlm function from the statsmodels package [\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e], which supports the random-effects structure required for repeated measures data. For generalised linear mixed models, we used the pymer4 package [\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e], which interfaces with the R lme4 framework [\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e] to support non-Gaussian outcomes such as binary variables. This approach allowed us to accurately evaluate whether any of the molecular features changed significantly over time, taking the entire temporal trend into account for each patient, and adjusting for within patient correlations between timepoints. The variables analysed and the corresponding summary results from the LMM/GLMM models are provided in Table S14.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eTumour samples were collected from 22 TNBC patients over the course of NACT and progression of the disease. Samples collected at the following timepoints, PreT (before NACT), PostT (end of NACT after surgery), R (from recurrent tumours) and MidT (at the midpoint of NACT) are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003eA. Response to NACT was assessed by a breast cancer consultant pathologist and the residual cancer burden (RCB) score and RCB class were determined for each patient (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Patients were ranked from low to high RCB (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003eB and Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). PreT samples were available for all patients while PostT samples only exist for patients where there was not pathological complete response to NACT (RCB\u0026thinsp;=\u0026thinsp;I-III). Where available, multiple regions (n\u0026thinsp;=\u0026thinsp;2\u0026ndash;4) of the PostT specimen were obtained and analysed. Ten patients had recurrence following completion of NACT and subsequent adjuvant treatment. Recurrence samples (R) were attainable for 7 of these patients, and it was possible to sample multiple regions (n\u0026thinsp;=\u0026thinsp;2\u0026ndash;4) in 3 patients (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). The aim was to establish the most complete sampling over time for each patient, supported by multi-region sampling at timepoints where this was feasible. The numbers of samples sequenced and analysed for each patient at each time point are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003eB and the full clinical and pathological characteristics of TNBC patients in the study is described in Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e. The level of tumour infiltrating lymphocytes (TILs) was measured in all samples and varied widely, from 1\u0026ndash;90% of the stromal area (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003eC, Table S2). Highest infiltration was observed in PreT samples in patients with a good response to NACT and was negatively correlated with RCB score as expected (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Accordingly, tumours recurred in 4/5 patients with PreT TIL levels of \u0026lt;\u0026thinsp;10%, consistent with previous reports [\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e] and increased TIL levels in PostT in our study was not indicative of recurrence as previously reported [\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e76\u003c/span\u003e]. In the patients with recurrence samples, TILs significantly decreased from a PreT average of 33% to an R average of 5% (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, p\u0026thinsp;=\u0026thinsp;0.03).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eSNV mutational burden remains stable between RCB classes and across disease progression\u003c/h2\u003e \u003cp\u003eTumour mutational burden (TMB) was assessed for each tumour sample (n\u0026thinsp;=\u0026thinsp;74) using the total burden of consensus SNV (short nucleotide variant) calls from deep whole exome sequencing (WES) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003eD, Table S5, Methods). TMB for PreT samples ranged from 1.1 to 19 SNV/Mb, with most patients ranging from 7.1 to 15.6 SNV/Mb. As might be expected, TMB and patient age at diagnosis were positively correlated (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.22, p\u0026thinsp;=\u0026thinsp;0.027). Two notable outlier patients exhibited significantly lower TMB (1.1 and 1.4 SNV/Mb) and were found to harbour germline mutations in \u003cem\u003eBRCA1\u003c/em\u003e and \u003cem\u003eRAD51C\u003c/em\u003e (TNBC6 and TNBC21 respectively) predicted to cause homologous recombination repair deficiency (HRD) (Figure S2). In general, across each patient\u0026rsquo;s longitudinal samples the mean TMB did not vary significantly across the course of the disease, based upon pairwise comparisons between timepoints (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003eD); though analysis of complete longitudinal profiles using linear mixed modelling (LMM) revealed a modest decline in TMB for R samples (p\u0026thinsp;=\u0026thinsp;0.03) compared to PreT samples (Table S14). Overall, TMB did not differ significantly between different RCB classes (Figure S3). Mean genomic SV burden also showed no significant relationship with sampling time (PreT, MidT, PostT or R) or RCB class (Figure S4, Table S8). Microsatellite instability (MSI) was largely stable across time (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003eE, Table S5), with slight decreases in PostT (p\u0026thinsp;=\u0026thinsp;0.026) and R samples (p\u0026thinsp;=\u0026thinsp;0.018) compared to PreT detected in longitudinal LLM modelling (Table S14).\u003c/p\u003e \u003cp\u003eWe exploited our deep WES data to study high confidence SNVs (supported by \u0026gt;\u0026thinsp;40 reads) in detail, providing a window on variants lost or newly emerging during disease progression. The fate of SNVs (single nucleotide substitutions and indels) in each patient was followed using a novel metric: variant retention (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003eD and Table S5). Variant retention is defined as the proportion of SNVs found in common between PreT and subsequent timepoints from the same patient. Between PreT and PostT the retention of SNVs in the majority of patient PostT samples was generally low, ranging from 5\u0026ndash;56%, suggesting frequent replacements of one dominant clone by another within tumours in response to treatment. In patients with both PostT and R samples the variant retention at recurrence generally approximated PostT values (on average\u0026thinsp;\u0026plusmn;\u0026thinsp;1.12%). Additional samples midway through NACT (MidT) in a fraction of patients (n\u0026thinsp;=\u0026thinsp;4) allowed an insight into how quickly variants can be altered during treatment. The variant retention for the MidT samples were equivalent or higher than paired PostT samples for those patients, consistent with rapid responses to treatment.\u003c/p\u003e \u003cp\u003eMutational signatures were extracted \u003cem\u003ede novo\u003c/em\u003e from the SNV mutational data (Methods), and the prevalence of known COSMIC SBS mutational signatures was also assessed (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003eF, Table S7). Derived signature 1 (Sig 1) closely matched COSMIC SBS1 (0.89 cosine similarity) and derived signature 2 (Sig 2) most closely matched SBS13 (0.70 cosine similarity). SBS1 and SBS5 signatures were prevalent in most patients, with minor influences by SBS3, SBS6 and SBS13. SBS1 and SBS5 have been shown to be clock-like signatures and increase with patient age. In patients harbouring germline disruptive SNVs in BRCA1 and RAD51C (TNBC6 and TNBC21), SBS3 was the most dominant signature, which is associated with defective homologous recombination repair. Overall, the stable SNV mutational burden over the progression of TNBC was matched by consistent mutational signatures, suggesting that the driver SNVs present in each tumour may also show stability over time.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eLoss and emergence of SNVs in known driver genes is common during TNBC treatment\u003c/h2\u003e \u003cp\u003eAnalysis of SNVs disrupting protein coding regions revealed many candidate driver genes found in previous studies. After filtering to limit false positive SNV calls (Methods), we identified SNVs disrupting 57 of 98 candidate genes previously associated with TNBC progression in the literature (Figure S2, Table S5). Those candidates disrupted most frequently by SNVs found in the 22 patients here are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eA. A combination of driver gene prediction algorithms (Methods) identified \u003cem\u003eTP53\u003c/em\u003e, \u003cem\u003eCYP2D6\u003c/em\u003e and a novel TNBC-associated gene \u003cem\u003eMICA\u003c/em\u003e as candidate driver genes (Table S6), in spite of the limited statistical power of this modest cohort. Variants in one or more of these 3 genes were observed in 17/22 patients. The most frequent variants found in the PreT samples were in \u003cem\u003eTP53\u003c/em\u003e and \u003cem\u003eMICA\u003c/em\u003e, both genes being altered in 9/22 patients (41%), followed by \u003cem\u003eCYP2D6\u003c/em\u003e with variants in 7/22 patients (32%). \u003cem\u003eBRCA1\u003c/em\u003e and \u003cem\u003eBRCA2\u003c/em\u003e variants were mutually exclusive (Figure S5), with \u003cem\u003eBRCA1\u003c/em\u003e or \u003cem\u003eBRCA2\u003c/em\u003e being detected in 10/22 patients (45%). Pairwise analysis in the PreT samples determined disruptive \u003cem\u003eBRCA2\u003c/em\u003e variants (in 6 patients) and disruptive \u003cem\u003eTP53\u003c/em\u003e variants (in 9 patients) to also be mutually exclusive events (Figure S5). Overall, 91% (20/22) of patients had PreT samples with somatic or germline disruptions of \u003cem\u003eTP53\u003c/em\u003e, \u003cem\u003eBRCA1\u003c/em\u003e, \u003cem\u003eBRCA2\u003c/em\u003e, \u003cem\u003ePRKDC\u003c/em\u003e or ATM, consistent with the early development of genomic instability. The remaining two patients (TNBC9, TNBC15) developed disruptive variants in these genes at PostT, suggesting ongoing selection for genomic instability via alterations to these same 5 genes across the cohort.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe employed a conservative consensus SNV calling (sequencing read depth\u0026thinsp;\u0026gt;\u0026thinsp;40, variant allele frequency (VAF)\u0026thinsp;\u0026gt;\u0026thinsp;0.1) strategy to ensure confidence in the detection of both \u0026lsquo;lost\u0026rsquo; and \u0026lsquo;novel\u0026rsquo; variants in later (PostT and/or R) samples, relative to the initial PreT biopsy samples. Losses and gains of SNVs were frequently observed over time, even in the 5 most frequently altered driver genes TP53, MICA, CYP2D6, BRCA1 and BRCA2 (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). For example, TP53 variants detected in the PreT samples of 7 patients were no longer observed in the paired PostT or R samples of these patients. This included a patient (TNBC10), where three PostT samples were sequenced and the TP53 variant identified in the PreT sample of the patient was lost in all three samples. Whereas, in patient TNBC22, four PostT samples were sequenced and the corresponding PreT TP53 variant was only identified in one sample suggesting intra-tumour heterogeneity. In addition, TP53 variants arose in the PostT or R samples of 5 patients where these variants were not detected in the paired PreT samples. Remarkably, in some patients, variants were lost concurrently with new variants arising in the same gene. This was seen with MICA SNVs in 4 patients, where later (PostT or R) gains and losses of variants were observed relative to the paired PreT samples. Some genes, i.e. BCL9 and PIKC3A were subject only to loss of the PreT variants across the time course of the study, with 3/4 patients\u0026rsquo; (BCL9) and 2/2 patients\u0026rsquo; (PIK3CA) variants undetectable after NACT. PostT and R samples also gained variants in a number of TNBC-related genes, in particular ERBB2 variants appeared in 4 patients, ALK variants in 3 patients and EGFR variants in 2 patients, with additional genes obtaining PostT variants in single patients (Figure S2). Thus, the consistent SNV burdens seen genome-wide (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003e) obscure frequent turnover at the level of particular driver variants in particular genes, likely reflecting clonal competition and replacement within tumours over time.\u003c/p\u003e \u003cp\u003eThe prevalence of genes altered by SNVs were largely consistent over time and across RCB classes (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). A notable deviation is the prevalence of BRCA1 variants (3/7 patients) in the PreT samples of patients in RCB 0, which exhibited the best response to NACT. The locations of the SNVs occurring within each of the 6 most altered driver genes show similar patterns over time (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eC, and Table S5). In general, it seems that both lost and novel SNVs disrupt similar regions of the proteins encoded by these driver genes, and this is also true of SNVs maintained across all timepoints in the same genes. For example, across both PreT and PostT samples, 10 different TP53 variants, including 8 missense and 2 nonsense variants, were detected. Multiple lost and novel SNVs were found, and on 4 occasions the same variant was observed in both PreT and PostT, but regardless of timing the vast majority were present in the DNA binding domain (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). Other driver genes failed to show strong biases in the locations of lost and novel variants, suggesting similar disruptions of gene functions over time. There were no statistically significant associations between candidate driver variants and patient survival, though such analyses are under-powered with the current data (Figure S6). Shifts in altered oncogenic pathways were also evident, across time and across RCB classes, based upon changes in the proteins predicted to be altered by SNVs (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). Patients with poorer responses to NACT (RCB 2\u0026ndash;3) show more frequent alteration of the NOTCH (+\u0026thinsp;18%, OR\u0026thinsp;=\u0026thinsp;1.3) and TP53 (+\u0026thinsp;18%, OR\u0026thinsp;=\u0026thinsp;1.5) pathways than patients with better response to NACT (RCB 0\u0026ndash;1). Better responding patients (RCB 0\u0026ndash;1) showed higher frequencies of altered WNT (+\u0026thinsp;9%, OR\u0026thinsp;=\u0026thinsp;1.2), PI3K (+\u0026thinsp;27%, OR\u0026thinsp;=\u0026thinsp;1.6) and RTK-RAS (+\u0026thinsp;27%, OR\u0026thinsp;=\u0026thinsp;1.6) pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eD).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eComprehensive detection of copy number alterations over disease progression\u003c/h2\u003e \u003cp\u003eCopy number alterations (CNAs) were detected in our cohort using two different but complementary assays, WES and CUTSeq [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). WES is known to provide a detailed interrogation of mutations within the fraction of the genome occupied by exons, while CUTSeq provides a broader picture of CNAs that are distributed throughout the human genome. For example, a closer examination of chromosome 8 in a patient, revealed that both approaches identified an amplification of chromosome arm 8q (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). However, as expected the read depth of WES closely mirrors the density of genes, whereas CUTseq shows a more uniform distribution across the chromosome arm. Decreasing the genomic range and examining a more focal region around a single gene (\u003cem\u003eMYC\u003c/em\u003e) further exemplifies the high concentration of reads within exons in WES, while CUTseq shows a more even distribution. In contrast, a closer examination of chromosome X revealed a deletion near the centromere that was detected using CUTseq, but not by WES (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). The detection of such alterations in gene-poor regions highlights the potential of CUTseq to assess genomic instability more comprehensively.\u003c/p\u003e \u003cp\u003eCNA patterns over the course of treatment showed differences between patients (all CNA profiles are provided in Figure S7). Patient TNBC18 (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e3\u003c/span\u003eC) provides an archetypal example of stability over time, where amplified and deleted regions present at PreT, were also found in all 4 tumour regions sampled at PostT, and at recurrence (with similar profiles over time seen in both CUTSeq and WES). In contrast, CNA profiles for many patients, such as TNBC11, were dynamic over time (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e3\u003c/span\u003eC), with widespread amplifications seen across most chromosomes in the PreT sample becoming completely absent in PostT samples, then re-appearing in the R tumour sample. Importantly, this pattern of alteration loss and recurrence through longitudinal time is consistent in both CUTseq and WES data for TNBC11, demonstrating broadly consistent results between both assays. Across all samples, there was a high level of agreement between the two approaches (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e3\u003c/span\u003eD), as indicated with significant correlations of estimated tumour ploidy (p\u0026thinsp;=\u0026thinsp;0.032) and the proportion of each genome either amplified (p\u0026thinsp;=\u0026thinsp;0.014) or deleted (p\u0026thinsp;=\u0026thinsp;0.004). However, the amount of sequencing reads required to assess CNA profiles using each approach differed, with CUTseq on average requiring only 10.5% of the sequencing reads needed in the WES analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e3\u003c/span\u003eD).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe regions of the genome enriched for amplifications or deletions across the PreT samples revealed similar patterns between CUTseq and WES CNA calls (\u003cb\u003eFigure S7\u003c/b\u003e, \u003cb\u003eTable S10\u003c/b\u003e). Two regions were jointly identified as hotspots for amplifications: 5p15.1 and 8q. The 5p15.1 hotspot encompassed 36 amplified genes including, MYO10 (myosin X) which has been reported to be overexpressed and to drive genomic instability in cancers [\u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e77\u003c/span\u003e] and contributes to breast cancer aggressiveness and metastasis [\u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e78\u003c/span\u003e]. Large regions of 8q were enriched for amplifications in both CUTseq and WES analyses (Table S10). The MYC gene is located on this chromosome arm and is commonly amplified in TNBC and promotes immune suppression [\u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e79\u003c/span\u003e]. Despite the high level of agreement between the two approaches, there were regions of enrichment which were only identified using one approach and not the other. WES identified 3q26.33 as enriched for amplifications, which notably contains PIK3CA which is commonly amplified in TNBC and potentially linked to response to targeted treatment strategies [\u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e80\u003c/span\u003e]. Alternatively, CUTseq identified a focal amplification of 19q12 harbouring CCNE1 which may confer chemotherapy resistance and has been linked to poor prognosis in TNBC [\u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e81\u003c/span\u003e], as well as other genes (POP4, PLEKHF1, and TSHZ3) reported to be amplified in cancer [\u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e82\u003c/span\u003e]. This region of chromosome 19 is adjacent to the centromere which may explain why it was only identified using CUTseq.\u0026nbsp;Similarly, CUTseq identified Xq13.2 as a hotspot for deletions which contains \u003cem\u003eXIST\u003c/em\u003e, the loss of which has been linked to accelerated tumour growth and brain metastases in breast cancer [\u003cspan citationid=\"CR83\" class=\"CitationRef\"\u003e83\u003c/span\u003e]. Overall, comparisons of WES and CUTseq estimates of CNA, arm level events and ploidy (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e3\u003c/span\u003eD) suggest that these complementary assays provide similar results where they can be compared, while each may still contribute novel information (\u003cb\u003eFigure S7\u003c/b\u003e). This supports our use of a union of WES/CUTseq CNA calls in the following analyses.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003ePervasive aneuploidies and whole genome duplication drives tumour evolution\u003c/h2\u003e \u003cp\u003eThough genome-wide SNV loads show broad stability over time (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e1\u003c/span\u003e), chromosome arm aneuploidies are dynamic, seen abundantly across PreT samples, but often showing dramatic losses in PostT samples (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eA, Table S13). These aneuploidies were frequent across all RCB classes and affected almost all chromosomes, with gains of 1q, 8q, 19q and 22q most frequent at PreT (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eA, \u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). Dramatic decreases in arm alterations were observed between PreT and PostT samples across most patients, consistent with a disproportionate impact of NACT on tumour lineages carrying aneuploidies. These arm level alterations then often reappeared in later (R) samples but usually involved novel CNA events, suggesting the dominance of distinct lineages in late disease (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). In general, R samples are notably enriched for these novel events that mimic PreT arm gains. Consistent with timepoint specific advantages to particular tumour clones, certain arm gain events, such as 19q gain seen at PreT, appear to be replaced at later timepoints by 19q arm loss at PostT in multiple patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). This significant pattern of decreased CNA burden at PostT as compared to PreT and R was seen when comparing the mean proportion of the genome altered by amplifications (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). The prevalence of gain or loss for certain chromosome arms (2p, 7p, 10p, and Xp, Chi-squared p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 with Bonferonni correction) was significantly related to a specific sampling time point (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). In contrast, certain aneuploidies, particularly gain of 1q and 8q, showed convincing evidence of persistence across the course of treatment, even in PostT samples where most other aneuploidies disappeared (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). This may suggest \u0026lsquo;aneuploidy addiction\u0026rsquo; - where particular aneuploidies are required for tumourigenesis - which has recently been demonstrated \u003cem\u003ein vitro\u003c/em\u003e in breast cancer cell lines [\u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e84\u003c/span\u003e]. Strikingly, tumour cell \u0026lsquo;addictions\u0026rsquo; to both 1q and 8q gains were found to confer strong advantages in this study, enhancing tumour cell growth, and 1q gains were found to be among the earliest alterations to occur in breast tumour development [\u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e84\u003c/span\u003e]. However, in this \u003cem\u003ein vitro\u003c/em\u003e study the mutational mechanisms underlying these aneuploidies remained unknown. The 1q and 8q gains seen here, in TNBC \u003cem\u003ein vivo\u003c/em\u003e, often occur on a background of pervasive CNA gains seen across the vast majority of chromosomes (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eB, \u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eC, Table S11), suggesting they may be a result of WGD. Indeed, predicting WGD using a conservative approach (Methods) we found evidence for WGD in 36% (8/22) of patients at PreT, 21% (3/14) at PostT, and in 57% (4/7) at R. All tumours with WGD carry 8q gains, and all but one carry 1q gains at PreT (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eA), suggesting that WGD may be exploited to generate advantageous aneuploidies during earlier stages, but also later at recurrence (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eD). Gains of 1q and 8q were also seen in 27% (6/22) of PreT samples lacking evidence for WGD, suggesting other mutational mechanisms also produce these alterations. The re-emergence of WGD may reflect the rise in frequency of pre-existing clones with WGD, or further \u003cem\u003ede novo\u003c/em\u003e WGD events in later disease. This aligns with recent work in ovarian cancer proposing that WGD is an ongoing mutational process promoting evolvability [\u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e85\u003c/span\u003e]. Importantly, the prevalence of WGD was not uniformly distributed across RCB classes, with the majority (5/7) of patients with the best response to NACT (RCB 0) having tumours with WGD at PreT (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eD). One of the drugs included in NACT (docetaxel) has been shown to induce polyploidization and aberrant aneuploidies by stabilizing microtubules [\u003cspan citationid=\"CR86\" class=\"CitationRef\"\u003e86\u003c/span\u003e]. The increased chromosome instability introduced by NACT may cause tumour cells that have previously undergone WGD to become unviable leading to improved treatment response and reduced residual cancer burden.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eGiven the large changes in ploidy and frequent WGD seen during TNBC evolution, many genes undergo recurrent alterations (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Among the genes amplified are many that have previously been reported to undergo alterations during progression (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e5\u003c/span\u003eA) in prior studies of the TNBC genome [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e]. Among the most recurrently amplified are \u003cem\u003eMYC\u003c/em\u003e (located at 8q24.21), \u003cem\u003eAKT3\u003c/em\u003e (1q43) and \u003cem\u003eMDM4\u003c/em\u003e (1q32.1). All three amplifications are present in the majority of PreT samples, are very rarely deleted, and also show evidence of persistence in multiple patients throughout treatment and into recurrence (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e5\u003c/span\u003eA, B), consistent with the 1q and 8q chromosome arm gains that also persist through treatment (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). All three have been specifically linked to TNBC via their upregulation and amplification. \u003cem\u003eMYC\u003c/em\u003e amplification is commonly observed in TNBC and is thought to suppress innate immunity and allow tumour immune escape [\u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e79\u003c/span\u003e]. \u003cem\u003eAKT3\u003c/em\u003e amplification has been observed in multiple breast cancer subtypes, including TNBC, and is thought to mediate resistance to tamoxifen [\u003cspan citationid=\"CR87\" class=\"CitationRef\"\u003e87\u003c/span\u003e]. \u003cem\u003eMDM4\u003c/em\u003e has emerged as an important factor in promoting TNBC tumour formation and metastasis, via its role as a \u003cem\u003eTP53\u003c/em\u003e inhibitor [\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e]. It therefore seems plausible that the frequent and persistent 1q and 8q arm gains seen in these tumours (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eA) result in increases in the expression of one or more of these genes, though we lack direct evidence for this. This would also offer a potential explanation for tumour cell \u0026lsquo;addiction\u0026rsquo; to 1q and 8q gains seen in recent \u003cem\u003ein vitro studies\u003c/em\u003e [\u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e84\u003c/span\u003e]. Analysis of altered oncogenic pathways revealed an intricate pattern of synergy between genes in the RTK/ RAS pathway and 1q and 8q gains. Amplifications of eight genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e5\u003c/span\u003eA, B) in the RTK/ RAS pathway (\u003cem\u003eEGFR\u003c/em\u003e, \u003cem\u003eIGF1R\u003c/em\u003e, \u003cem\u003eKRAS\u003c/em\u003e, \u003cem\u003eKIT\u003c/em\u003e, \u003cem\u003eNF1\u003c/em\u003e, \u003cem\u003eFGFR1\u003c/em\u003e, \u003cem\u003ePDGFRA\u003c/em\u003e and \u003cem\u003eMAP3K1\u003c/em\u003e) showed co-occurrence with gains in 1q (12/16 patients) or 8q (14/15).\u003c/p\u003e \u003cp\u003eOther frequently altered genes suggest a more complex mutational landscape. For example, \u003cem\u003eBRD4\u003c/em\u003e (19p13.12), \u003cem\u003eNOTCH3\u003c/em\u003e (19p13.2) and \u003cem\u003eCCNE1\u003c/em\u003e (19q12) undergo amplification and deletion in different samples and at different timepoints (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e5\u003c/span\u003eA, B). This volatile state may suggest the presence of complex rearrangements such as chromothripsis, which has been reported to be particularly enriched on chromosome 19 in high grade serous ovarian cancer [\u003cspan citationid=\"CR88\" class=\"CitationRef\"\u003e88\u003c/span\u003e]. We also observe frequent amplifications and deletions of \u003cem\u003eKDM6A\u003c/em\u003e, with amplifications common in PreT but deletions at later stages (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e5\u003c/span\u003eA, B), consistent with reports that \u003cem\u003eKDM6A\u003c/em\u003e loss can reduce sensitivity to the therapeutic agents used to treat TNBC patients [\u003cspan citationid=\"CR89\" class=\"CitationRef\"\u003e89\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eA closer examination of the relationship between gene copy number alterations and WGD status in PreT samples revealed distinct patterns (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e5\u003c/span\u003eC). In samples that had undergone WGD, amplifications of the top 15 most commonly altered genes were highly prevalent (87.5%). In contrast, non-WGD samples showed a lower frequency of amplification (38.6%). Deletions were relatively rare in WGD samples; however, in non-WGD samples, deletions affecting BRD4 and NOTCH3 at 19p13.12 were common, occurring in 57.1% of cases compared to just 12.5% in WGD samples. These findings suggest that WGD shapes genome-wide mutational patterns and drives distinct combinations of candidate driver gene alterations.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eClinical correlates of mutational loads and aneuploidy\u003c/h2\u003e \u003cp\u003eA number of clinically relevant features were found to be inter-correlated (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e6\u003c/span\u003eA). TMB in PreT samples positively correlated with metrics of genomic instability across various scales: CUTseq derived ploidy (R\u0026thinsp;=\u0026thinsp;0.59, p\u0026thinsp;=\u0026thinsp;0.04), total structural variant burden (R\u0026thinsp;=\u0026thinsp;0.73, p\u0026thinsp;=\u0026thinsp;0.0001) and microsatellite instability (R\u0026thinsp;=\u0026thinsp;0.73, p\u0026thinsp;=\u0026thinsp;0.009). As anticipated, the amount of the genome amplified by CNAs correlated with CUTseq (R\u0026thinsp;=\u0026thinsp;0.71, p\u0026thinsp;=\u0026thinsp;0.004) and WES derived ploidy estimates (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e6\u003c/span\u003eA, R\u0026thinsp;=\u0026thinsp;0.73, p\u0026thinsp;=\u0026thinsp;0.0009), and TIL infiltration and RCB score were negatively correlated (R=-0.56, p\u0026thinsp;=\u0026thinsp;0.045) (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e6\u003c/span\u003eA, Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Investigating relationships between mutations, clinical features and overall patient overall survival (OS) showed that patient OS varied significantly, with patients in RCB class II/III having poorer OS than RCB 0/I patients (p\u0026thinsp;=\u0026thinsp;0.039) (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e6\u003c/span\u003eB). Although HRD scores did not differ significantly between RCB classes (Figure S8), as expected, patients with tumours showing evidence for HRD in PreT samples (defined as scarHRD scores\u0026thinsp;\u0026gt;\u0026thinsp;70) had significantly better OS (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e6\u003c/span\u003eB, p\u0026thinsp;=\u0026thinsp;0.01). This is in agreement with previous studies demonstrating higher HRD scores are associated with a better OS and better sensitivity to treatments such as PARP inhibitors and platinum-based chemotherapy [\u003cspan citationid=\"CR90\" class=\"CitationRef\"\u003e90\u003c/span\u003e, \u003cspan citationid=\"CR91\" class=\"CitationRef\"\u003e91\u003c/span\u003e]. Using a Bayesian approach (Methods) we looked for pairwise co-occurrence or mutual exclusivity among mutational events, including the presence or absence of SNVs in driver genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003e), chromosome arm aneuploidies (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003e) or CNAs recurrently impacting genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e5\u003c/span\u003e) (see Figure S9). The output was dominated by pairwise co-occurrence events between amplifications in genes and chromosome arms, as expected with frequent WGD events (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e0).\u003c/p\u003e \u003cp\u003eTo analyse longitudinal patterns within each individual patient, we employed Linear Mixed Model (LMM) and Generalised LMM (for binary dependent variables) modelling approaches for all features. The modelling indicates that several molecular and clinical features change significantly across time points (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e6\u003c/span\u003eC, Table S14), showing patterns that are broadly consistent with the analyses based upon pairwise comparisons between timepoints (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003e). For instance, from PreT to R, we observe a decrease in TILs (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\beta\\:=-22.03\\)\u003c/span\u003e\u003c/span\u003e, SE\u0026thinsp;=\u0026thinsp;5.62, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) and a reduction of SV burden (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\beta\\:=-9.24\\)\u003c/span\u003e\u003c/span\u003e, SE\u0026thinsp;=\u0026thinsp;4.38, p\u0026thinsp;=\u0026thinsp;0.035). Similarly, the odds of WGD shrinks to 18% of the initial odds from PreT to PostT (p\u0026thinsp;=\u0026thinsp;0.038), indicating that WGD is significantly less likely to occur at the PostT time point compared to PreT (Table S14). WGD is also commonly seen in R samples (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e4\u003c/span\u003eD), which is likely to reflect the resurgence of WGD bearing clones in these recurrent tumours. In fact, even though median ploidy is greater than 2 in PreT samples, median (WES or CUTseq) ploidy is higher at recurrence, suggesting multiple WGD events in some samples (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e6\u003c/span\u003eD). As the threshold for WGD (mean fraction of genome amplified over WES and CUTseq datasets) is increased from 20% to 70%, PreT and R samples were consistently higher than PostT samples, demonstrating that this trend is not dependent on the threshold by which WGD was determined (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e6\u003c/span\u003eE).\u003c/p\u003e \u003cp\u003eWe conclude that WGD events are an under-appreciated mutational mechanism underlying the evolution of TNBC during treatment. These events appear to drive the acquisition of advantageous aneuploidies, particularly 1q and 8q gains, and are predicted to impact known cancer driver genes. WGD bearing tumours are common at PreT and presumably confer proliferative advantages before treatment, but these tumours also appear to be unusually sensitive to NACT, resulting in a strong decline in the ploidy of PostT samples. At later stages higher ploidy mediated by WGD, again appears to offer advantages to dominant clones within R samples.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eUsing two complementary sequencing approaches (WES and CUTseq), we have reconstructed the mutational landscape of TNBC and tracked its dynamics during NACT. The genomic and transcriptomic landscape of TNBC has been examined in several previous studies [\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e], but TNBC temporal mutational dynamics during chemotherapy treatment is poorly studied. Here, we describe a cohort with longitudinal sampling, providing an unusual opportunity to determine how the mutational landscape responds to treatment, including SNVs, CNAs and aneuploidies. We found that although the genome-wide SNV mutational burden remained stable during disease progression, the loss and emergence of SNVs in known TNBC driver genes was common in response to treatment. These complex dynamics were seen in \u003cem\u003eTP53\u003c/em\u003e, \u003cem\u003eMICA\u003c/em\u003e, \u003cem\u003eCYP2D6\u003c/em\u003e, \u003cem\u003eBRCA1\u003c/em\u003e and \u003cem\u003eBRCA2\u003c/em\u003e, suggesting clonal loss and replacement events over time, consistent with selection favouring different variants at different times in the same driver gene.\u003c/p\u003e \u003cp\u003eDramatic genome-wide structural shifts were seen in all samples, reflected in CNA and chromosome arm aneuploidies, with a substantial bias to duplications. Overall, the abundant structural variation seen in PreT samples was followed by frequent loss of these alterations after treatment, but the re-emergence of similar alterations at recurrence. These alterations include advantageous aneuploidies increasing tumour growth \u003cem\u003ein vitro\u003c/em\u003e, such as 1q gain, that has been proposed to drive tumourigenesis via an increase in \u003cem\u003eMDM4\u003c/em\u003e expression, which phenocopies the effects of \u003cem\u003eTP53\u003c/em\u003e inactivating mutations to suppress \u003cem\u003eTP53\u003c/em\u003e signalling [\u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e84\u003c/span\u003e]. Such aneuploidies have also been demonstrated to create collateral therapeutic vulnerabilities via the toxic effects of duplicating many 1q genes, which can represent novel drug targets [\u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e84\u003c/span\u003e]. We also see positive correlations between ploidy and TMB, suggesting tumours undergoing WGD may present higher neoantigen loads, targetable by neoantigen vaccines, underlining the promise of such emerging approaches in TNBC [\u003cspan citationid=\"CR92\" class=\"CitationRef\"\u003e92\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eRelatively early, and frequent (36% of PreT samples) WGD events appear to drive or at least co-occur with the extensive aneuploidies seen pre-treatment, consistent with a recent study of TNBC patient-derived xenograft samples that lacked longitudinal sampling [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Notably a higher frequency of PreT WGD was seen in patients with the best response to NACT (RCB0/1: 55%; RCB2/3: 18%), suggesting that early WGD may confer susceptibility to treatment. WGD was also frequently found at recurrence (58% of R samples), either reflecting the re-emergence of proliferative WGD bearing clones, or late, \u003cem\u003ede novo\u003c/em\u003e WGD events. To our knowledge this is the first evidence for WGD as a common event in both pre-treatment and recurrent TNBC tumours. Subsequent studies with larger sample sizes and generating WGS data will be better placed to make more detailed estimates of the impact of WGD on clinical outcomes.\u003c/p\u003e \u003cp\u003eThese unusual, longitudinally sampled data suggest the presence of WGD itself may be a useful prognostic biomarker of NACT response, associated with better or worse outcomes, depending on the time of sampling. Emerging studies also suggest that WGD may be a targetable vulnerability [\u003cspan citationid=\"CR93\" class=\"CitationRef\"\u003e93\u003c/span\u003e, \u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e94\u003c/span\u003e]. Detecting WGD and genome-wide CNA accurately in a FFPE tumour sample is not possible using the most widely used profiling technologies (targeted sequencing panels) while WES and WGS are not financially viable options for many healthcare providers. However, we have demonstrated that CUTseq represents an affordable assay that provides accurate results in the setting of structurally complex tumour genomes sampled using FFPE blocks and primary tumour sections.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eDetermining the origins and dynamics of structural complexity in tumourigenesis over time is critical to understand the genomic heterogeneity between patients and within tumours. In this study we reveal the genome-wide shifts in structural variation that underlie TNBC evolution during treatment. We present evidence that WGD in pre-treatment biopsies is associated with better response to NACT, but also appears to confer advantages to tumours at recurrence. Validation in larger, temporally sampled cohorts is required to confirm these results, but the analysis of large numbers of samples will rely on the development of affordable sequencing strategies that can be used to profile the (FFPE) samples that are generally available from healthcare providers. We have demonstrated that CUTseq and WES can be combined to that end, and readily provide new insights into tumour evolution over time.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eThe authors thank the patients and their families for their participation in this academic study. We are also grateful for the time and effort involved in sample collection, of the breast radiologists, surgeons and nurses at the Edinburgh Breast Unit. We would like to acknowledge the Edinburgh Clinical Research Facility for DNA sequencing and support, and the NRS Lothian Bioresource for additional tissue samples. Some of the results published here are in part based upon data generated by the TCGA Research Network: https://www.cancer.gov/tcga.\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eO.O., F.S., A.E. and C.A.S. designed the study and supervised analysis of the cohort. F.S. performed laboratory work and D.P.B. performed the bioinformatic analyses. A.M., M.I.O. and A.E. provided additional analysis. A.I. performed histopathological review. O.O designed the NEO and metcfDNA studies, arranged relevant ethical approvals and recruited patients (under her care throughout their breast cancer journey). N.W collected tissue samples and gathered patient clinicopathological characteristics, from national health electronic patient records. D.P.B., F.S., O.O. and C.A.S. wrote the manuscript, which was read and approved by all authors.\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eThis study was supported by funding and donations from Make 2nds Count, John Henry Cugley and Gillian Hunt and the Breast Cancer Institute. Salary for F.S. was supported by the Susan Melville Fund. C.A.S., D.B. and A.M. were supported by MRC core funding to the MRC Human Genetics Unit, University of Edinburgh and MRC Programme funding MC_UU_00035/1. AE was supported by a University of Edinburgh Chancellor\u0026rsquo;s Fellowship, a Langmuir Talent Development Fellowship and core funding from the MRC Human Genetics Unit.\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eGenomic data generated in this study have been submitted to the European Genome/ Phenome Archive, with accession numbers EGAD00001015684 (CUTseq data) and EGAD00001015687 (WES data).\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eCollection of human tissues and relevant clinicopathological data were approved by the South East Scotland Research Ethics Committee 01 with REC references: 13/SS/0236 and 15/ES/0094, and all patients consented to sample collection and use of their data in research\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eNone\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eNik-Zainal S, Davies H, Staaf J, Ramakrishna M, Glodzik D, Zou X, et al. Landscape of somatic mutations in 560 breast cancer whole-genome sequences. Nat Nat Publishing Group. 2016;534:47\u0026ndash;54. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/nature17676\u003c/span\u003e\u003cspan address=\"10.1038/nature17676\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBielski CM, Zehir A, Penson AV, Donoghue MTA, Chatila W, Armenia J, et al. Genome doubling shapes the evolution and prognosis of advanced cancers. Nat Genet. 2018;50:1189\u0026ndash;95. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41588-018-0165-1\u003c/span\u003e\u003cspan address=\"10.1038/s41588-018-0165-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL\u0026oacute;pez S, Lim EL, Horswell S, Haase K, Huebner A, Dietzen M, et al. Interplay between whole-genome doubling and the accumulation of deleterious alterations in cancer evolution. Nat Genet Nat Publishing Group. 2020;52:283\u0026ndash;93. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41588-020-0584-7\u003c/span\u003e\u003cspan address=\"10.1038/s41588-020-0584-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKang K, Wu Y, Han C, Wang L, Wang Z, Zhao A. Homologous recombination deficiency in triple-negative breast cancer: Multi-scale transcriptomics reveals distinct tumor microenvironments and limitations in predicting immunotherapy response. Comput Biol Med. 2023;158:106836. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compbiomed.2023.106836\u003c/span\u003e\u003cspan address=\"10.1016/j.compbiomed.2023.106836\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurstein MD, Tsimelzon A, Poage GM, Covington KR, Contreras A, Fuqua SAW, et al. Comprehensive Genomic Analysis Identifies Novel Subtypes and Targets of Triple-Negative Breast Cancer. Clin Cancer Res. 2015;21:1688\u0026ndash;98. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/1078-0432.CCR-14-0432\u003c/span\u003e\u003cspan address=\"10.1158/1078-0432.CCR-14-0432\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLehmann BD, Jovanović B, Chen X, Estrada MV, Johnson KN, Shyr Y, et al. Refinement of Triple-Negative Breast Cancer Molecular Subtypes: Implications for Neoadjuvant Chemotherapy Selection. PLOS ONE Public Libr Sci. 2016;11:e0157368. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0157368\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0157368\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang Y-Z, Ma D, Suo C, Shi J, Xue M, Hu X, et al. Genomic and Transcriptomic Landscape of Triple-Negative Breast Cancers: Subtypes and Treatment Strategies. Cancer Cell. 2019;35:428\u0026ndash;e4405. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.ccell.2019.02.001\u003c/span\u003e\u003cspan address=\"10.1016/j.ccell.2019.02.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLehmann BD, Colaprico A, Silva TC, Chen J, An H, Ban Y, et al. Multi-omics analysis identifies therapeutic vulnerabilities in triple-negative breast cancer subtypes. Nat Commun Nat Publishing Group. 2021;12:6276. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41467-021-26502-6\u003c/span\u003e\u003cspan address=\"10.1038/s41467-021-26502-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu S, Wu Y, Song B, Yi M, Yan Y, Mei Q, et al. Recent advances in targeted strategies for triple-negative breast cancer. J Hematol OncolJ Hematol Oncol. 2023;16:100. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13045-023-01497-3\u003c/span\u003e\u003cspan address=\"10.1186/s13045-023-01497-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDenkert C, von Minckwitz G, Darb-Esfahani S, Lederer B, Heppner BI, Weber KE, et al. Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy. Lancet Oncol. 2018;19:40\u0026ndash;50. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S1470-2045(17)30904-X\u003c/span\u003e\u003cspan address=\"10.1016/S1470-2045(17)30904-X\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSalgado R, Denkert C, Demaria S, Sirtaine N, Klauschen F, Pruneri G, et al. The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs Working Group 2014. Ann Oncol Off J Eur Soc Med Oncol. 2015;26:259\u0026ndash;71. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/annonc/mdu450\u003c/span\u003e\u003cspan address=\"10.1093/annonc/mdu450\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMenezes MCS, Raheem F, Mina L, Ernst B, Batalini F. PARP Inhibitors for Breast Cancer: Germline BRCA1/2 and Beyond. Cancers. 2022;14:4332. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/cancers14174332\u003c/span\u003e\u003cspan address=\"10.3390/cancers14174332\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBianchini G, De Angelis C, Licata L, Gianni L. Treatment landscape of triple-negative breast cancer \u0026mdash; expanded options, evolving needs. Nat Rev Clin Oncol Nat Publishing Group. 2022;19:91\u0026ndash;113. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41571-021-00565-2\u003c/span\u003e\u003cspan address=\"10.1038/s41571-021-00565-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCortazar P, Zhang L, Untch M, Mehta K, Costantino JP, Wolmark N, et al. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet Elsevier. 2014;384:164\u0026ndash;72. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0140-6736(13)62422-8\u003c/span\u003e\u003cspan address=\"10.1016/S0140-6736(13)62422-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFitzpatrick A, Tutt A. Controversial issues in the neoadjuvant treatment of triple-negative breast cancer. Ther Adv Med Oncol. 2019;11:1758835919882581. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1758835919882581\u003c/span\u003e\u003cspan address=\"10.1177/1758835919882581\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchmid P, Cortes J, Dent R, McArthur H, Pusztai L, K\u0026uuml;mmel S, et al. Overall Survival with Pembrolizumab in Early-Stage Triple-Negative Breast Cancer. N Engl J Med. 2024;391:1981\u0026ndash;91. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1056/NEJMoa2409932\u003c/span\u003e\u003cspan address=\"10.1056/NEJMoa2409932\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchmid P, Cortes J, Dent R, Pusztai L, McArthur H, K\u0026uuml;mmel S, et al. Event-free Survival with Pembrolizumab in Early Triple-Negative Breast Cancer. N Engl J Med Mass Med Soc. 2022;386:556\u0026ndash;67. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1056/NEJMoa2112651\u003c/span\u003e\u003cspan address=\"10.1056/NEJMoa2112651\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eValencia GA, Rioja P, Morante Z, Ruiz R, Fuentes H, Castaneda CA, et al. Immunotherapy in triple-negative breast cancer: A literature review and new advances. World J Clin Oncol. 2022;13:219\u0026ndash;36. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5306/wjco.v13.i3.219\u003c/span\u003e\u003cspan address=\"10.5306/wjco.v13.i3.219\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu Q, Kaur J, Wylie D, Mittal K, Li H, Kolachina R, et al. A Case Series Exploration of Multi-Regional Expression Heterogeneity in Triple-Negative Breast Cancer Patients. Int J Mol Sci. 2022;23:13322. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ijms232113322\u003c/span\u003e\u003cspan address=\"10.3390/ijms232113322\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKuzmin E, Baker TM, Lesluyes T, Monlong J, Abe KT, Coelho PP, et al. Evolution of chromosome-arm aberrations in breast cancer through genetic network rewiring. Cell Rep. 2024;43:113988. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.celrep.2024.113988\u003c/span\u003e\u003cspan address=\"10.1016/j.celrep.2024.113988\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaraayvaz M, Cristea S, Gillespie SM, Patel AP, Mylvaganam R, Luo CC, et al. Unravelling subclonal heterogeneity and aggressive disease states in TNBC through single-cell RNA-seq. Nat Commun Nat Publishing Group. 2018;9:3588. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41467-018-06052-0\u003c/span\u003e\u003cspan address=\"10.1038/s41467-018-06052-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim C, Gao R, Sei E, Brandt R, Hartman J, Hatschek T, et al. Chemoresistance Evolution in Triple-Negative Breast Cancer Delineated by Single-Cell Sequencing. Cell. 2018;173:879\u0026ndash;e89313. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.cell.2018.03.041\u003c/span\u003e\u003cspan address=\"10.1016/j.cell.2018.03.041\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang X, Garnerone S, Simonetti M, Harbers L, Nicoś M, Mirzazadeh R, et al. CUTseq is a versatile method for preparing multiplexed DNA sequencing libraries from low-input samples. Nat Commun Nat Publishing Group. 2019;10:4732. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41467-019-12570-2\u003c/span\u003e\u003cspan address=\"10.1038/s41467-019-12570-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBossuyt V, Provenzano E, Symmans WF, Boughey JC, Coles C, Curigliano G, et al. Recommendations for standardized pathological characterization of residual disease for neoadjuvant clinical trials of breast cancer by the BIG-NABCG collaboration. Ann Oncol Off J Eur Soc Med Oncol. 2015;26:1280\u0026ndash;91. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/annonc/mdv161\u003c/span\u003e\u003cspan address=\"10.1093/annonc/mdv161\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDieci MV, Radosevic-Robin N, Fineberg S, van den Eynden G, Ternes N, Penault-Llorca F, et al. Update on tumor-infiltrating lymphocytes (TILs) in breast cancer, including recommendations to assess TILs in residual disease after neoadjuvant therapy and in carcinoma in situ: A report of the International Immuno-Oncology Biomarker Working Group on Breast Cancer. Semin Cancer Biol. 2018;52:16\u0026ndash;25. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.semcancer.2017.10.003\u003c/span\u003e\u003cspan address=\"10.1016/j.semcancer.2017.10.003\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDenkert C, Wienert S, Poterie A, Loibl S, Budczies J, Badve S, et al. Standardized evaluation of tumor-infiltrating lymphocytes in breast cancer: results of the ring studies of the international immuno-oncology biomarker working group. Mod Pathol Off J U S Can Acad Pathol Inc. 2016;29:1155\u0026ndash;64. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/modpathol.2016.109\u003c/span\u003e\u003cspan address=\"10.1038/modpathol.2016.109\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen S, Zhou Y, Chen Y, Gu J. fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics. 2018;34:i884\u0026ndash;90. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/bioinformatics/bty560\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/bty560\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM [Internet]. arXiv; 2013 [cited 2023 Oct 31]. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.1303.3997\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.1303.3997\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinforma Oxf Engl. 2009;25:2078\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/bioinformatics/btp352\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/btp352\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRausch T, Hsi-Yang Fritz M, Korbel JO, Benes V. Alfred: interactive multi-sample BAM alignment statistics, feature counting and feature annotation for long- and short-read sequencing. Bioinformatics. 2019;35:2489\u0026ndash;91. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/bioinformatics/bty1007\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/bty1007\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEwels PA, Peltzer A, Fillinger S, Patel H, Alneberg J, Wilm A, et al. The nf-core framework for community-curated bioinformatics pipelines. Nat Biotechnol. 2020;38:276\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41587-020-0439-x\u003c/span\u003e\u003cspan address=\"10.1038/s41587-020-0439-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGarcia M, Juhos S, Larsson M, Olason PI, Martin M, Eisfeldt J, et al. Sarek: A portable workflow for whole-genome sequencing analysis of germline and somatic variants. F1000Research. 2020;9:63. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.12688/f1000research.16665.2\u003c/span\u003e\u003cspan address=\"10.12688/f1000research.16665.2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHanssen F, Garcia MU, Folkersen L, Pedersen AS, Lescai F, Jodoin S et al. Scalable and efficient DNA sequencing analysis on different compute infrastructures aiding variant discovery [Internet]. bioRxiv; 2024 [cited 2024 May 9]. p. 2023.07.19.549462. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/2023.07.19.549462\u003c/span\u003e\u003cspan address=\"10.1101/2023.07.19.549462\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAndrews S. FastQC: A Quality Control tool for High Throughput Sequence Data [Internet]. 2010 [cited 2022 Dec 12]. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.bioinformatics.babraham.ac.uk/projects/fastqc/\u003c/span\u003e\u003cspan address=\"https://www.bioinformatics.babraham.ac.uk/projects/fastqc/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Accessed 12 Dec 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEwels P, Magnusson M, Lundin S, K\u0026auml;ller M. MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics. 2016;32:3047\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/bioinformatics/btw354\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/btw354\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The Genome Analysis Toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010;20:1297\u0026ndash;303. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/gr.107524.110\u003c/span\u003e\u003cspan address=\"10.1101/gr.107524.110\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePedersen BS, Quinlan AR. Mosdepth: quick coverage calculation for genomes and exomes. Bioinformatics. 2018;34:867\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/bioinformatics/btx699\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/btx699\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBenjamin D, Sato T, Cibulskis K, Getz G, Stewart C, Lichtenstein L. Calling Somatic SNVs and Indels with Mutect2 [Internet]. bioRxiv; 2019 [cited 2024 May 9]. p. 861054. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/861054\u003c/span\u003e\u003cspan address=\"10.1101/861054\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim S, Scheffler K, Halpern AL, Bekritsky MA, Noh E, K\u0026auml;llberg M, et al. Strelka2: fast and accurate calling of germline and somatic variants. Nat Methods. 2018;15:591\u0026ndash;4. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41592-018-0051-x\u003c/span\u003e\u003cspan address=\"10.1038/s41592-018-0051-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKoboldt DC, Zhang Q, Larson DE, Shen D, McLellan MD, Lin L, et al. VarScan 2: Somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Res. 2012;22:568\u0026ndash;76. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/gr.129684.111\u003c/span\u003e\u003cspan address=\"10.1101/gr.129684.111\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcLaren W, Gil L, Hunt SE, Riat HS, Ritchie GRS, Thormann A, et al. The Ensembl Variant Effect Predictor. Genome Biol. 2016;17:122. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13059-016-0974-4\u003c/span\u003e\u003cspan address=\"10.1186/s13059-016-0974-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang X, Jiang T, Shen A, Chen Y, Zhou Y, Liu J et al. CaMutQC: An R Package for Integrative Quality Control of Cancer Somatic Mutations [Internet]. bioRxiv; 2024 [cited 2024 Aug 15]. p. 2024.08.12.606123. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/2024.08.12.606123\u003c/span\u003e\u003cspan address=\"10.1101/2024.08.12.606123\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLandrum MJ, Lee JM, Riley GR, Jang W, Rubinstein WS, Church DM, et al. ClinVar: public archive of relationships among sequence variation and human phenotype. Nucleic Acids Res. 2014;42:D980\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/nar/gkt1113\u003c/span\u003e\u003cspan address=\"10.1093/nar/gkt1113\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJia P, Yang X, Guo L, Liu B, Lin J, Liang H, et al. MSIsensor-Pro: Fast, Accurate, and Matched-Normal-Sample-Free Detection of Microsatellite Instability. Genomics Proteom Bioinf. 2020;18:65\u0026ndash;71. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.gpb.2020.02.001\u003c/span\u003e\u003cspan address=\"10.1016/j.gpb.2020.02.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJamuar SS, Kuan JL, Brett M, Tiang Z, Tan WLW, Lim JY et al. Incidentalome from Genomic Sequencing: A Barrier to Personalized Medicine? EBioMedicine. 2016;5:211\u0026ndash;6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.ebiom.2016.01.030\u003c/span\u003e\u003cspan address=\"10.1016/j.ebiom.2016.01.030\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShyr C, Tarailo-Graovac M, Gottlieb M, Lee JJ, van Karnebeek C, Wasserman WW. FLAGS, frequently mutated genes in public exomes. BMC Med Genomics. 2014;7:64. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12920-014-0064-y\u003c/span\u003e\u003cspan address=\"10.1186/s12920-014-0064-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLawrence MS, Stojanov P, Polak P, Kryukov GV, Cibulskis K, Sivachenko A, et al. Mutational heterogeneity in cancer and the search for new cancer-associated genes. Nat Nat Publishing Group. 2013;499:214\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/nature12213\u003c/span\u003e\u003cspan address=\"10.1038/nature12213\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen H, Chong W, Yang X, Zhang Y, Sang S, Li X, et al. Age-related mutational signature negatively associated with immune activity and survival outcome in triple-negative breast cancer. Oncoimmunology. 2020;9:1788252. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/2162402X.2020.1788252\u003c/span\u003e\u003cspan address=\"10.1080/2162402X.2020.1788252\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDerakhshan F, Reis-Filho JS. Pathogenesis of Triple-Negative Breast Cancer. Annu Rev Pathol Mech Dis. 2022;17:181\u0026ndash;204. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1146/annurev-pathol-042420-093238\u003c/span\u003e\u003cspan address=\"10.1146/annurev-pathol-042420-093238\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShah SP, Roth A, Goya R, Oloumi A, Ha G, Zhao Y, et al. The clonal and mutational evolution spectrum of primary triple-negative breast cancers. Nat Nat Publishing Group. 2012;486:395\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/nature10933\u003c/span\u003e\u003cspan address=\"10.1038/nature10933\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi RQ, Yan L, Zhang L, Ma HX, Wang HW, Bu P, et al. Genomic characterization reveals distinct mutational landscapes and therapeutic implications between different molecular subtypes of triple-negative breast cancer. Sci Rep Nat Publishing Group. 2024;14:12386. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41598-024-62991-3\u003c/span\u003e\u003cspan address=\"10.1038/s41598-024-62991-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMayakonda A, Lin D-C, Assenov Y, Plass C, Koeffler HP. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res. 2018;28:1747\u0026ndash;56. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/gr.239244.118\u003c/span\u003e\u003cspan address=\"10.1101/gr.239244.118\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSanchez-Vega F, Mina M, Armenia J, Chatila WK, Luna A, La KC, et al. Oncogenic Signaling Pathways in The Cancer Genome Atlas. Cell. 2018;173:321\u0026ndash;e33710. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.cell.2018.03.035\u003c/span\u003e\u003cspan address=\"10.1016/j.cell.2018.03.035\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFantini D, Vidimar V, Yu Y, Condello S, Meeks JJ. MutSignatures: an R package for extraction and analysis of cancer mutational signatures. Sci Rep Nat Publishing Group. 2020;10:18217. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41598-020-75062-0\u003c/span\u003e\u003cspan address=\"10.1038/s41598-020-75062-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlexandrov LB, Kim J, Haradhvala NJ, Huang MN, Tian Ng AW, Wu Y, et al. The repertoire of mutational signatures in human cancer. Nat Nat Publishing Group. 2020;578:94\u0026ndash;101. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41586-020-1943-3\u003c/span\u003e\u003cspan address=\"10.1038/s41586-020-1943-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSondka Z, Dhir NB, Carvalho-Silva D, Jupe S, Madhumita, McLaren K, et al. COSMIC: a curated database of somatic variants and clinical data for cancer. Nucleic Acids Res. 2023;52:D1210\u0026ndash;7. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/nar/gkad986\u003c/span\u003e\u003cspan address=\"10.1093/nar/gkad986\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartincorena I, Raine KM, Gerstung M, Dawson KJ, Haase K, Loo PV, et al. Universal Patterns of Selection in Cancer and Somatic Tissues. Cell Elsevier. 2017;171:1029\u0026ndash;e104121. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.cell.2017.09.042\u003c/span\u003e\u003cspan address=\"10.1016/j.cell.2017.09.042\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMularoni L, Sabarinathan R, Deu-Pons J, Gonzalez-Perez A, L\u0026oacute;pez-Bigas N. OncodriveFML: a general framework to identify coding and non-coding regions with cancer driver mutations. Genome Biol. 2016;17:128. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13059-016-0994-0\u003c/span\u003e\u003cspan address=\"10.1186/s13059-016-0994-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen X, Schulz-Trieglaff O, Shaw R, Barnes B, Schlesinger F, K\u0026auml;llberg M, et al. Manta: rapid detection of structural variants and indels for germline and cancer sequencing applications. Bioinformatics. 2016;32:1220\u0026ndash;2. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/bioinformatics/btv710\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/btv710\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Loo P, Nordgard SH, Lingj\u0026aelig;rde OC, Russnes HG, Rye IH, Sun W et al. Allele-specific copy number analysis of tumors. Proc Natl Acad Sci. Proceedings of the National Academy of Sciences; 2010;107:16910\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1073/pnas.1009843107\u003c/span\u003e\u003cspan address=\"10.1073/pnas.1009843107\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAmemiya HM, Kundaje A, Boyle AP. The ENCODE Blacklist: Identification of Problematic Regions of the Genome. Sci Rep Nat Publishing Group. 2019;9:9354. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41598-019-45839-z\u003c/span\u003e\u003cspan address=\"10.1038/s41598-019-45839-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGao C, Xiao G, Bargonetti J. Contemplations on MDMX (MDM4) driving triple negative breast cancer circulating tumor cells and metastasis. Oncotarget. 2019;10:5007\u0026ndash;10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18632/oncotarget.27134\u003c/span\u003e\u003cspan address=\"10.18632/oncotarget.27134\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMermel CH, Schumacher SE, Hill B, Meyerson ML, Beroukhim R, Getz G. GISTIC2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers. Genome Biol. 2011;12:R41. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/gb-2011-12-4-r41\u003c/span\u003e\u003cspan address=\"10.1186/gb-2011-12-4-r41\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSztupinszki Z, Diossy M, Krzystanek M, Reiniger L, Csabai I, Favero F, et al. Migrating the SNP array-based homologous recombination deficiency measures to next generation sequencing data of breast cancer. Npj Breast Cancer Nat Publishing Group. 2018;4:1\u0026ndash;4. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41523-018-0066-6\u003c/span\u003e\u003cspan address=\"10.1038/s41523-018-0066-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFavero F, Joshi T, Marquard AM, Birkbak NJ, Krzystanek M, Li Q, et al. Sequenza: allele-specific copy number and mutation profiles from tumor sequencing data. Ann Oncol. 2015;26:64\u0026ndash;70. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/annonc/mdu479\u003c/span\u003e\u003cspan address=\"10.1093/annonc/mdu479\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWei T, Simko V. R package corrplot: Visualization of a Correlation Matrix [Internet]. 2024. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/taiyun/corrplot\u003c/span\u003e\u003cspan address=\"https://github.com/taiyun/corrplot\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMina M, Raynaud F, Tavernari D, Battistello E, Sungalee S, Saghafinia S, et al. Conditional Selection of Genomic Alterations Dictates Cancer Evolution and Oncogenic Dependencies. Cancer Cell Elsevier. 2017;32:155\u0026ndash;e1686. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.ccell.2017.06.010\u003c/span\u003e\u003cspan address=\"10.1016/j.ccell.2017.06.010\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMina M, Iyer A, Tavernari D, Raynaud F, Ciriello G. Discovering functional evolutionary dependencies in human cancers. Nat Genet Nat Publishing Group. 2020;52:1198\u0026ndash;207. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41588-020-0703-5\u003c/span\u003e\u003cspan address=\"10.1038/s41588-020-0703-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTherneau TM. A Package for Survival Analysis in R [Internet]. 2024. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://CRAN.R-project.org/package=survival\u003c/span\u003e\u003cspan address=\"https://CRAN.R-project.org/package=survival\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTherneau TM, Grambsch PM. Modeling survival data: extending the Cox model. New York: Springer; 2001.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSjoberg DD, Baillie M, Fruechtenicht C, Haesendonckx S, Treis T. ggsurvfit: Flexible Time-to-Event Figures [Internet]. 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/pharmaverse/ggsurvfit\u003c/span\u003e\u003cspan address=\"https://github.com/pharmaverse/ggsurvfit\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSeabold S, Perktold J, statsmodels. Econometric and statistical modeling with python. 9th Python Sci Conf. 2010.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJolly E. Pymer4: Connecting R and Python for Linear Mixed Modeling. J Open Source Softw. 2018;3:862. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.21105/joss.00862\u003c/span\u003e\u003cspan address=\"10.21105/joss.00862\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBates D, M\u0026auml;chler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw. 2015;67:1\u0026ndash;48. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18637/jss.v067.i01\u003c/span\u003e\u003cspan address=\"10.18637/jss.v067.i01\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeon-Ferre RA, Jonas SF, Salgado R, Loi S, de Jong V, Carter JM, et al. Tumor-Infiltrating Lymphocytes in Triple-Negative Breast Cancer. JAMA. 2024;331:1135\u0026ndash;44. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1001/jama.2024.3056\u003c/span\u003e\u003cspan address=\"10.1001/jama.2024.3056\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGiacchetti S, Faucheux L, Gardair C, Cuvier C, de Roquancourt A, Campedel L, et al. Negative Relationship between Post-Treatment Stromal Tumor-Infiltrating Lymphocyte (TIL) and Survival in Triple-Negative Breast Cancer Patients Treated with Dose-Dense Dose-Intense NeoAdjuvant Chemotherapy. Cancers. 2022;14:1331. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/cancers14051331\u003c/span\u003e\u003cspan address=\"10.3390/cancers14051331\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMayca Pozo F, Geng X, Tamagno I, Jackson MW, Heimsath EG, Hammer JA, et al. MYO10 drives genomic instability and inflammation in cancer. Sci Adv Am Association Advancement Sci. 2021;7:eabg6908. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1126/sciadv.abg6908\u003c/span\u003e\u003cspan address=\"10.1126/sciadv.abg6908\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCao R, Chen J, Zhang X, Zhai Y, Qing X, Xing W, et al. Elevated expression of myosin X in tumours contributes to breast cancer aggressiveness and metastasis. Br J Cancer Nat Publishing Group. 2014;111:539\u0026ndash;50. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/bjc.2014.298\u003c/span\u003e\u003cspan address=\"10.1038/bjc.2014.298\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZimmerli D, Brambillasca CS, Talens F, Bhin J, Linstra R, Romanens L, et al. MYC promotes immune-suppression in triple-negative breast cancer via inhibition of interferon signaling. Nat Commun Nat Publishing Group. 2022;13:6579. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41467-022-34000-6\u003c/span\u003e\u003cspan address=\"10.1038/s41467-022-34000-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAmato O, Buisseret L, Gebhart G, Plouznikoff N, Larsimont D, Awada A, et al. PIK3CA copy-number gain and inhibitors of the PI3K/AKT/mTOR pathway in triple-negative breast cancer. Cold Spring Harb Mol Case Stud. 2023;9:a006255. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/mcs.a006255\u003c/span\u003e\u003cspan address=\"10.1101/mcs.a006255\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao Z-M, Yost SE, Hutchinson KE, Li SM, Yuan Y-C, Noorbakhsh J, et al. CCNE1 amplification is associated with poor prognosis in patients with triple negative breast cancer. BMC Cancer. 2019;19:96. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12885-019-5290-4\u003c/span\u003e\u003cspan address=\"10.1186/s12885-019-5290-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNatrajan R, Mackay A, Wilkerson PM, Lambros MB, Wetterskog D, Arnedos M, et al. Functional characterization of the 19q12 amplicon in grade III breast cancers. Breast Cancer Res. 2012;14:R53. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/bcr3154\u003c/span\u003e\u003cspan address=\"10.1186/bcr3154\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXing F, Liu Y, Wu S-Y, Wu K, Sharma S, Mo Y-Y, et al. Loss of XIST in breast cancer activates MSN-c-Met and reprograms microglia via exosomal microRNA to promote brain metastasis. Cancer Res. 2018;78:4316\u0026ndash;30. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/0008-5472.CAN-18-1102\u003c/span\u003e\u003cspan address=\"10.1158/0008-5472.CAN-18-1102\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGirish V, Lakhani AA, Thompson SL, Scaduto CM, Brown LM, Hagenson RA, et al. Oncogene-like addiction to aneuploidy in human cancers. Sci Am Association Advancement Sci. 2023;0:eadg4521. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1126/science.adg4521\u003c/span\u003e\u003cspan address=\"10.1126/science.adg4521\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcPherson A, V\u0026aacute;zquez-Garc\u0026iacute;a I, Myers MA, Zatzman M, Al-Rawi D, Weiner A et al. Ongoing genome doubling promotes evolvability and immune dysregulation in ovarian cancer [Internet]. bioRxiv; 2024 [cited 2025 May 15]. p. 2024.07.11.602772. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/2024.07.11.602772\u003c/span\u003e\u003cspan address=\"10.1101/2024.07.11.602772\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOgden A, Rida PCG, Knudsen BS, Kucuk O, Aneja R. Docetaxel-induced polyploidization may underlie chemoresistance and disease relapse. Cancer Lett. 2015;367:89\u0026ndash;92. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.canlet.2015.06.025\u003c/span\u003e\u003cspan address=\"10.1016/j.canlet.2015.06.025\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHinz N, J\u0026uuml;cker M. Distinct functions of AKT isoforms in breast cancer: a comprehensive review. Cell Commun Signal. 2019;17:154. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12964-019-0450-3\u003c/span\u003e\u003cspan address=\"10.1186/s12964-019-0450-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEwing A, Meynert A, Silk R, Aitken S, Bendixsen DP, Churchman M, et al. Divergent trajectories to structural diversity impact patient survival in high grade serous ovarian cancer. Nat Commun Nat Publishing Group. 2025;16:5586. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41467-025-60655-y\u003c/span\u003e\u003cspan address=\"10.1038/s41467-025-60655-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiao J-F, Kua L-F, Ding L-W, Sun Q-Y, Myint KN, Chia X-R, et al. KDM6A Depletion in Breast Epithelial Cells Leads to Reduced Sensitivity to Anticancer Agents and Increased TGFβ Activity. Mol Cancer Res MCR. 2022;20:637\u0026ndash;49. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/1541-7786.MCR-21-0402\u003c/span\u003e\u003cspan address=\"10.1158/1541-7786.MCR-21-0402\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBelli C, Duso BA, Ferraro E, Curigliano G. Homologous recombination deficiency in triple negative breast cancer. Breast. 2019;45:15\u0026ndash;21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.breast.2019.02.007\u003c/span\u003e\u003cspan address=\"10.1016/j.breast.2019.02.007\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSharma P, Barlow WE, Godwin AK, Pathak H, Isakova K, Williams D, et al. Impact of homologous recombination deficiency biomarkers on outcomes in patients with triple-negative breast cancer treated with adjuvant doxorubicin and cyclophosphamide (SWOG S9313). Ann Oncol. 2018;29:654\u0026ndash;60. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/annonc/mdx821\u003c/span\u003e\u003cspan address=\"10.1093/annonc/mdx821\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang X, Goedegebuure SP, Chen MY, Mishra R, Zhang F, Yu YY, et al. Neoantigen DNA vaccines are safe, feasible, and induce neoantigen-specific immune responses in triple-negative breast cancer patients. Genome Med. 2024;16:131. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13073-024-01388-3\u003c/span\u003e\u003cspan address=\"10.1186/s13073-024-01388-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCohen-Sharir Y, McFarland JM, Abdusamad M, Marquis C, Bernhard SV, Kazachkova M, et al. Aneuploidy renders cancer cells vulnerable to mitotic checkpoint inhibition. Nature. 2021;590:486\u0026ndash;91. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41586-020-03114-6\u003c/span\u003e\u003cspan address=\"10.1038/s41586-020-03114-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQuinton RJ, DiDomizio A, Vittoria MA, Kot\u0026yacute;nkov\u0026aacute; K, Ticas CJ, Patel S, et al. Whole-genome doubling confers unique genetic vulnerabilities on tumour cells. Nature. 2021;590:492\u0026ndash;7. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41586-020-03133-3\u003c/span\u003e\u003cspan address=\"10.1038/s41586-020-03133-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"genome-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Genome Medicine](https://genomemedicine.biomedcentral.com/)","snPcode":"13073","submissionUrl":"https://submission.springernature.com/new-submission/13073/3","title":"Genome Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Triple negative breast cancer, tumour evolution, copy number variants, aneuploidy, whole genome duplication, neoadjuvant chemotherapy","lastPublishedDoi":"10.21203/rs.3.rs-9212020/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9212020/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBackground\u003c/p\u003e\n\u003cp\u003eTriple-negative breast cancer (TNBC) is associated with poor survival rate and high genomic instability, generating complex tumour genomes. However, the processes that generate this complexity are poorly studied in longitudinal samples. Here, we study the temporal dynamics of TNBC somatic mutations, revealing major transitions in tumour genome evolution, from diagnostic biopsies, through treatment, to cancer remission or recurrence.\u003c/p\u003e\n\u003cp\u003eMethods\u003c/p\u003e\n\u003cp\u003eDeep whole exome sequencing and CUTseq, a reduced representation whole genome sequencing approach, were performed in parallel, to comprehensively identify short nucleotide variants (SNVs), copy number alterations (CNAs) and aneuploidies. Tumour samples (N=74) from 22 patients were profiled before and after neoadjuvant chemotherapy (NACT), and encompassed spatially diverse samples from multiple primary breast tumours, to allow tracking of the gain and loss of candidate driver variants over time.\u003c/p\u003e\n\u003cp\u003eResults\u003c/p\u003e\n\u003cp\u003eGenome-wide SNV mutational burden remained stable across disease progression and RCB classes. However, recurrent SNVs were identified in several known TNBC driver genes in response to treatment, with TP53, MICA, CYP2D6, BRCA1, and BRCA2 being frequently altered. The candidate driver variants in these genes frequently exhibited dynamic changes throughout the course of a patient’s treatment, with the original SNVs in pre-treatment samples often lost, while novel variants in the same genes emerged at subsequent time points.\u003c/p\u003e\n\u003cp\u003eIn contrast to the stable genome-wide SNV burdens, dramatic changes in chromosome structure were seen in all tumours, with abundant CNAs and chromosome arm aneuploidies seen in pre-treatment samples, followed by frequent loss of these alterations post-treatment, and their re-emergence at recurrence. Whole genome duplication (WGD) events appear to drive these dynamics, with a higher frequency of pre-treatment WGD seen in patients with the best response to NACT. WGD also frequently appears late in the disease at recurrence, suggesting potential roles in the evolution resistance.\u003c/p\u003e\n\u003cp\u003eConclusions\u003c/p\u003e\n\u003cp\u003eComprehensive longitudinal profiling of the TNBC genome demonstrates the complex interplay of SNVs and structural alterations during tumour progression, leading to diverse evolutionary trajectories impacting patient outcomes. Complex mutational patterns encompassing entire chromosomes emerge during progression, with WGD events making major contributions to intra-tumour heterogeneity, and emerging as a potential candidate biomarker of response at both tumour establishment and recurrence.\u003c/p\u003e","manuscriptTitle":"Frequent Whole-Genome Duplication Events Drive the Genomic Evolution of Triple-Negative Breast Cancer During Neoadjuvant Chemotherapy","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-09 15:03:40","doi":"10.21203/rs.3.rs-9212020/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-05-07T14:38:56+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-30T22:28:23+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-13T19:08:56+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"52456595086046286807497134091524260056","date":"2026-04-13T18:47:04+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"175555140052275160652557815212545857548","date":"2026-04-13T16:45:03+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-08T01:11:28+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"310818271941822220059164468537704959044","date":"2026-04-02T17:34:27+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-02T14:20:13+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-01T03:03:27+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-25T06:22:54+00:00","index":"","fulltext":""},{"type":"submitted","content":"Genome Medicine","date":"2026-03-24T12:29:37+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"genome-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Genome Medicine](https://genomemedicine.biomedcentral.com/)","snPcode":"13073","submissionUrl":"https://submission.springernature.com/new-submission/13073/3","title":"Genome Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"63c6c9d8-21fe-4515-9f36-147dd3a5e0e7","owner":[],"postedDate":"April 9th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Revision requested","date":"2026-05-07T14:38:56+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-30T22:28:23+00:00","index":25,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[],"tags":[],"updatedAt":"2026-05-07T14:53:52+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-09 15:03:40","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9212020","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9212020","identity":"rs-9212020","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}