A mitochondrial quality regulation gene signature for prognosis and tumor microenvironment characterization in breast cancer: an integrative analysis with experimental validation

A mitochondrial quality regulation gene signature for prognosis and tumor microenvironment characterization in breast cancer: an integrative analysis with experimental validation | 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 A mitochondrial quality regulation gene signature for prognosis and tumor microenvironment characterization in breast cancer: an integrative analysis with experimental validation Huaiwen Pu, Tingjing Li, Renji Liang, Zhongxiang Fan, Bowen Tang, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8995736/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 10 You are reading this latest preprint version Abstract Background: Breast cancer (BC) remains a leading cause of female cancer mortality, necessitating the identification of novel biomarkers for precise prognostic stratification. Mitochondrial quality-related genes (MQRGs) are critical players in tumorigenesis; however, their specific role in BC remains insufficiently characterized. Methods: We integrated transcriptomic and clinical datasets from TCGA, GTEx, and GEO to identify differentially expressed MQRGs. A 7-gene prognostic signature, comprising SLC45A1 , HPN , CHAD , CXCL9 , GLYATL2 , KRT14 , and IGLV6-57 , was developed using LASSO and multivariate Cox regression. Its predictive robustness was rigorously assessed via Kaplan-Meier and time-dependent ROC analyses.We further characterized the tumor microenvironment (TME) and validated our findings through quantitative real-time PCR（qRT-PCR） and multiplex immunohistochemistry (mIHC). Results: Unsupervised consensus clustering identified two distinct molecular subtypes related to MQRG (Clusters A and B), which exhibited significantly different clinical outcomes and tumor microenvironment (TME) characteristics. Utilizing subtype-specific differentially expressed genes, we developed a 7-gene prognostic signature (including SLC45A1 , HPN , CHAD , CXCL9 , GLYATL2 , KRT14 , and IGLV6-57 ) through LASSO-Cox regression. The signature demonstrated robust prognostic reliability across cohorts. High-risk group were distinguished by an immunosuppressive TME architecture, augmented TMB, and divergent therapeutic sensitivities, contrasting sharply with the low-risk group.Experimental validation via qRT-PCR confirmed the dysregulation of these critical genes in breast cancer cells. Notably, mIHC revealed the spatial distribution of CXCL9 , highlighting its predominant expression in tumor-associated macrophages and its significant positive correlation with M1 (iNOS+) rather than M2 (CD206+) polarization, indicating its role in modulating anti-tumor immunity. Conclusion: This study establishes a robust Mitochondrial Quality Regulation Gene Signature as an independent prognostic biomarker for BC. By elucidating the spatial dynamics of CXCL9 in promoting M1 macrophage recruitment, our findings provide an integrative framework for risk stratification and personalized therapeutic interventions, particularly for immunotherapy. Mitochondrial quality regulation genes Breast cancer Tumor microenvironment prognostic model Multiplex immunohistochemistry drug sensitivity Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 1. Introduction Breast cancer (BC) remains a substantial health concern worldwide, with its complex molecular mechanisms and heterogeneous biological behavior posing challenges to the development of optimal therapeutic and preventive strategies[1-3]. Comprehensive gene expression profiling has led to BC classification into five distinct subtypes: human epidermal growth factor receptor 2 (HER2)-enriched, luminal A/B, basal-like, and the later-identified claudin-low variant. Clinical decision-making and prognostic assessment are primarily guided by immunohistochemical evaluation of HER2 status, together with hormone receptor [estrogen and progesterone receptors] expression patterns and detailed disease staging. At present, traditional treatment modalities for BC include surgery, radiotherapy, chemotherapy, targeted therapy, and endocrine therapy, among others[4, 5]. Advanced-stage disease is uniformly associated with poor clinical outcomes, necessitating continuous therapeutic intervention, with triple-negative BC (TNBC) demonstrating a dismal prognosis in metastatic settings, exhibiting median survival durations below 24 months[6]. These difficulties underscore the pressing need for identifying and validating novel biomarkers and therapeutic targets to enhance diagnostic precision, prognostic accuracy, and treatment efficacy in BC management. Mitochondria, double-membrane-bound eukaryotic organelles, are the primary site of oxidative phosphorylation, the fundamental process for intracellular adenosine triphosphate (ATP) biosynthesis[7]. In addition to energy production, mitochondria are integral to lipid biosynthesis, calcium buffering, and iron-sulfur cluster assembly. Their pivotal involvement in innate immunity and the precise execution of cell death pathways further underscores their indispensable role in maintaining tissue-level homeostasis.To preserve mitochondrial and cellular integrity, eukaryotic cells employ a sophisticated and dynamic mitochondrial quality control (MQC) system, jointly regulated by the nuclear and mitochondrial genomes[8]. MQC mechanisms, including mitochondrial dynamics (fusion and fission), mitophagy (selective autophagic clearance of damaged mitochondria), and biogenesis, collectively govern mitochondrial quality and quantity[9]. In cancer, metabolic reprogramming enables tumor cells to divert metabolites toward biosynthetic pathways, aiming at supporting rapid proliferation and accumulating macromolecular precursors necessary for tumor growth[10]. MQC has been found to be crucial in various diseases[11] and in dysregulation in the pathogenesis of lung cancer[12], colon cancer[13], BC[14], melanoma[15], ovarian carcinoma[16], prostate adenocarcinoma[17], and pancreatic cancer[18]. Given the limited research on the link between MQRGs and BC, elucidating MQC dysfunction is paramount for therapeutic innovation. In this study, we leveraged the MitoCarta 3.0 mammalian mitochondrial proteome to identify 20 core MQRGs involved in mitochondrial dynamics, mitophagy, and biogenesis (ranging from PPARGC1A to MAP1LC3C ; detailed in Table S1). These markers provide a robust framework for deciphering the molecular drivers of mitochondrial dysregulation in BC [19]. This study aimed to stratify BC patients into molecular subtypes relying upon MQRG expression profiles and explore the interaction between these subtypes and clinical attributes, tumor microenvironment (TME), immune status, and drug sensitivity. Through analysis, seven key biomarkers ( SLC45A1 , HPN , CHAD , CXCL9 , GLYATL2 , KRT14 , and IGLV6-57 ) associated with BC prognosis were identified, from which a BC prognostic model was developed and further validated for its accuracy in predicting BC prognosis. These findings may provide new insights for personalized therapies and improve patient outcomes. 2. Method 2.1 Dataset curation and preprocessing The comprehensive research workflow and analytical pipeline of this study are delineated in Figure 1. RNA-seq data were acquired from 1,097 BC specimens in the TCGA(https://portal.gdc.cancer.gov) and 327 BC specimens from the GEO (https://www.ncbi.nlm.nih.gov/geo/), encompassing somatic mutation profiles and associated clinical data: Vital status, age, gender, tumor grade, and pathological stage. Additionally, we validated model-related gene expression patterns using the GSE20685 database acquired from the GEO database. Additional datasets were retrieved by accessing the UNCAN (https://uncan.eu/) and HPA (https://www.proteinatlas.org/). Prior to analysis, all datasets underwent rigorous preprocessing, including format conversion and standardization. Based on a systematic literature review, 20 MQRGs were identified for inclusion. As all data were derived from publicly available repositories, no informed consent or ethics committee approval was required. 2.2 Consensus clustering analysis of MQRGs Consensus clustering analysis was carried out via the R package "ConsensusClusterPlus". Cluster assignment revealed enhanced intra-subtype correlations, as demonstrated by the cumulative distribution function (CDF) curve displaying a relatively flat slope relative to the steeper slopes characteristic of robust within-group homogeneity. Conversely, inter-subtype correlations were significantly weaker, confirming distinct molecular profiles between clusters. Based on prognostic MQRG expression patterns, we stratified tumor specimens into distinct MQRG subtypes and utilized principal component analysis (PCA) to validate the classification, with the resulting dimensionality reduction effectively capturing inter-subgroup variations. Finally, Kaplan-Meier (KM) survival analysis was performed using the "survival" package, aiming at comparing clinical outcomes across MQRG-based subtypes. 2.3 Identification of DEGs and functional enrichment analysis Applying an adj-p-value of 0.05 alongside a fold-change of 2.0, 125 DEGs were identified across diverse MQRG subtypes using the "limma". Both KEGG[20] and GO[21] enrichment analyses were applied to elucidate key molecular pathways and classify DEGs based on molecular function (MF), cellular component (CC), and biological process (BP) categories, respectively. The "clusterProfiler" was used to conduct functional enrichment investigations, while gene Set Variation Analysis (GSVA) was implemented to quantify pathway-level enrichment scores, thereby identifying distinct signaling perturbations across the cohorts.l[22]. 2.4 Construction of the MQRG RS model The MQRG scoring system was formulated to determine tumor-specific molecular patterns of metabolic-quiescence-associated genes. Initially, univariate Cox regression analysis was carried out on the transcriptomic dataset to recognize BC overall survival (OS) significantly related to differentially expressed genes (DEGs). Subsequently, we deployed consensus clustering analysis to delineate two distinct molecular subtypes (MQRG-Clusters A/B) based on prognostic MQRG expression profiles. A 1:1 randomization protocol was implemented to allocate BC patients into training and testing cohorts with balanced baseline characteristics. The MQRG-based scoring system was subsequently developed via the training dataset. To optimize model generalizability, Lasso-Cox regression analysis was applied on MQRG candidate genes via the "glmnet", with penalty parameters optimized through 10-fold cross-validation to avoid overfitting. Finally, multivariate Cox proportional hazards modeling was employed to assess MQRG's independent prognostic significance in the training cohort. The training cohort was stratified into low- (LRG) and high-risk (HRG) groups using the median RS as the cutoff (LRG: scores ≤ median; HRG: scores > median), then KM analysis was performed. The same RS cutoff was utilized for the testing cohort, and model performance was further evaluated using KM and time-dependent ROC curves. 2.5 Nomogram establishment and validation A predictive nomogram was generated using the "rms," incorporating clinical features and RSs. Each variable contributed to a score, and the sum of these scores generated a total score for each patient, followed by calculating AUC values for 1-, 3-, and 5-year OS, and plotting ROC curves to assess nomogram performance. Furthermore, we evaluated the nomogram performance by comparing observed and predicted survival probabilities at the pre-determined time points by generating a calibration plot. 2.6 Analysis of prognostic genes The expression profiles and prognostic utility of the seven signature genes were ascertained leveraging the Xiantao Academic platform(https://www.xiantao.love/). We utilized the TCGA-BRCA dataset (XENA/TCGA_GTEx-BRCA), where RNA-seq data were harmonized through the TOIL pipeline and represented in TPM units. To maintain cohort integrity, our study population was refined by filtering out normal controls and cases with missing survival or pathological records.The association between gene expression and patient survival was assessed using Kaplan-Meier (K-M) survival analysis. 2.7 Cell culture with qRT-PCR analysis The frozen batch of the human normal mammary epithelium MCF-10A, human BC cells HCC1937, and MDA-MB-453 (Procell, Wuhan, China) was defrosted by being immersed in a water bath at 37 ℃. The MCF-10A cells were cultured in DMEM/F12 medium that contained 5% horse serum, 20 ng/mL epidermal growth factor, 0.5 μg/mL cortisol, 10 μg/mL recombinant human insulin, 1% non-essential amino acid solution, and 1% penicillin-streptomycin (P/S). Meanwhile, the HCC1937 and MDA-MB-453 cells were cultured on 10 cm plates using RPMI-1640 medium that contained 10% FBS and 1% P/S. Subsequently, the specimens were incubated at 37 °C and 5% CO 2 . After isolating ribosomal RNA with TRIzol reagent (Invitrogen, Carlsbad, CA, USA), cDNA was synthesized by mixing total RNA with a Takara PrimeScript RT reagent kit. The RT‑qPCR was performed on a CFX96™ Real-Time PCR (Bio-Rad Laboratories, Inc., USA) using Takara SYBR® Green assays with 2× SYBR Green qPCR Mix (SparkJade), and all samples were analyzed in triplicate. Gene expression levels were quantified by the 2 -ΔΔC t approach, using GAPDH for normalization. Table S2 lists the genes ( SLC45A1 , HPN , CHAD , CXCL9 , GLYATL2 , KRT14 , and IGLV6-57 ) primer sequences used. 2.8 Evaluation of immune cells Infiltration This molecular classification therapeutic relevance was systematically validated through Spearman correlation analyses between MQRG subtypes and clinicopathological parameters, including age, sex, TNM staging, and survival outcomes. The ssGSEA was deployed to estimate the relative infiltration levels of 23 distinct immune cell types in each BC specimen. Immune and stromal infiltration indices were estimated via the ESTIMATE tool, while the CIBERSORT algorithm was deployed to quantify the landscape of tumor-infiltrating ICs. These analyses allowed for a meticulous evaluation of the cellular heterogeneity and immunological characteristics across the heterogeneous specimens of both risk strata. 2.9 Multiplex immunohistochemistry staining The tumor tissue microarrays(Brcsur2205) was purchased from Hunan AiFang Biological Technology Co., Ltd. and contained 120 cases of breast cancer tissue samples with clinical pathological characteristics and survival information.Clinical information is provided in Table S3.Tissue sections were deparaffinized in xylene and rehydrated through a descending ethanol gradient. Multiplex immunohistochemistry staining (mIHC) was executed using a seven-color mIHC kit (AFIHC027, AiFang Biologcal, China), and the antigens were exposed to microwave radiation per protocol. To block endogenous peroxidase activity, sections were incubated in 3% hydrogen peroxide at room temperature for 15 min, followed by a 15-min incubation with 10% goat serum for nonspecific binding blockade. Primary antibody A was introduced and incubated at 4 °C overnight, followed by three washes with PBST. A polymer-HRP-conjugated universal secondary antibody (anti-mouse/rabbit IgG; AFIHC001, AiFang Biological) was then added and incubated at room temperature for half an hour. After PBST washing, a TYR-based fluorescent dye was applied for 8 min, and washed thrice with PBST. Antibody A was subsequently stripped via microwave treatment, and the sections were washed three times with PBST. The staining cycle was repeated iteratively for each subsequent primary antibody B, with blocking using goat serum before each new primary antibody application. The panel of antigens visualized included: CK-Pan (AF20164, 1:200), CD206 (AFRM0009, 1:100), iNOS (AFRP0001, 1:200), Vimentin (AF20105, 1:1500), Ki67 (AF20068, 1:200); all from AiFang Biological), and CXCL9 (YT6008, 1:200, Immunoway). Following completion of all staining cycles, nuclei were counterstained with DAPI for 10 min at room temperature in darkness, followed by three PBST washes. Slides were mounted with an anti-fade fluorescence mounting medium and imaged via an eight-channel fluorescence digital slide scanner (model AF-KL-20-8, AiFang Biological, China). During the image acquisition stage, the eight-channel fluorescence digital slide scanner was used to capture the mIHC slide, saving the acquired images in the kfbf format. In the image data analysis stage, custom algorithms were developed in the VISIOPHARM software (VISIOPHARM, Hoersholm, Denmark) to conduct data analysis on the images. To ensure result consistency, fixed and uniform threshold parameters were set through the VISIOPHARM application to identify and quantitatively analyze the association between CXCL9 and immune cells (ICs), particularly macrophages. After the analysis was completed, the results were exported, and SPSS was used for the statistical analysis.The entire core was categorized into tumor and stromal areas based on the expression of PanCK and vimentin. 2.10 Assessment of the TMB, CSC, and mutation Using the "maftools"[23], we generated a mutation annotation format from TCGA data to systematically identify somatic mutations among BC patients in both groups. Additionally, potential associations between CSC, TMB, and the identified HRG/LRG classifications were investigated. 2.11 Drug susceptibility analysis The "pRRophetic"[24] software was utilized to determine the half-maximal inhibitory concentrations (IC50) of conventional chemotherapeutic agents utilized in BC to evaluate potential differences in therapeutic susceptibility between the two groups. 2.12 Statistical analysis Data were generated via the Perl programming language (v5.32.1) while performing data processing, analysis, and visualization via R (v4.3.3). Statistical significance across all analyses was categorized as follows: ns ( p ≥ 0.05), *( p < 0.05), **( p < 0.01), ***( p < 0.001), and****( p < 0.0001). 3. Results 3.1 Genomic landscapes, expression profiles, and chromosomal mapping of MQRGs To begin, 20 MQRG expression levels were detected in healthy samples and BC samples from the TCGA (Figure 2A), revealing a significant overexpression of three MQRGs in BC samples: ESRRA , FIS1 , and SQSTM1. Conversely, ten MQRGs were significantly downregulated: PPARGC1A , PPARA , PPARG , NFE2L2 , MFN2, MIEF1 , PINK1 , PARK2 , MAP1LC3B , and MAP1LC3C . Subsequently, CNV mutation frequency was investigated. The majority of these mutations manifested as copy number amplifications. In contrast, genes such as PRKN , PPARA , MAP1LC3B , FIS1 , MFF , PINK1 , and MIEF2 exhibited scattered CNV deletions (Figure 2B). Additionally, the chromosomal locations of CNV alterations for these 20 MQRGs were delineated, as displayed in Figure 2C. The interactions among BC MQRGs and their prognostic implications were systematically investigated using a network (Figure 2D). This analysis identified all MQRGs as risk factors, with the exception of PPARGC1A , PPARG , NRF1 , MFN2 , FIS1 , MIEF2 , MAP1LC3A , and MAP1LC3C , which were regarded as favorable factors. Assessment of gene connectivity exposed a tightly-knit network of MQRGs within the breast cancer microenvironment (Figure 2E). Most gene-to-gene associations exhibited strong positive co-expression (red lines), with especially prominent links among factors controlling mitochondrial turnover and dynamics.Regarding somatic mutation incidence in these 20 MQRGs, among 991 BC samples, 55 samples (5.55%) harbored mutations in MQRGs (Figure 2F). A total of six genes, namely PPARGC1A , OPA1 , PPARG , PINK1 , NFE2L2 , and MFN1 , were identified with mutations. Furthermore, 16 MQRGs were significantly related to OS. Specifically, BC patients with overexpressed MAP1LC3A , MAP1LC3C , FIS1 , MIEF2, MFN2 , and PPARG exhibited more favorable OS outcomes (Figure S1). Based on these findings, a remarkable divergence was observed between healthy and BC samples in MQRG expression and genetic landscapes. 3.2 Identification of MQRG-related BC subtypes Unsupervised consensus clustering of MQRG expression profiles identified two robust molecular subtypes (k = 2), designated MQRG Cluster A (n = 794) and MQRG Cluster B (n = 626) (Figures 3A-D). As anticipated, PCA demonstrated significant variations in transcription patterns between the two MQRG subtypes (Figure 3E). Finally, KM curves revealed that OS was better in patients in Cluster A than Cluster B ( p =0.006, Figure 3F). 3.3 TME features in distinct MQRG molecular subtypes The clinicopathologic attributes of distinct BC subtypes were meticulously compared using the TCGA and GSE20685 databases, unveiling significant differences in MQRG expression patterns and clinicopathologic characteristics (Figure 4A). Comparative analysis revealed differential expression patterns of MQRGs between prognostic clusters. For instance, cluster B demonstrated significant upregulation of mitochondrial fusion regulators ( TFAM , MFN1 , and OPA1 ) and oxidative stress response genes ( NFE2L2, MIEF1 ) compared to Cluster A. Conversely, autophagy-related genes ( MAP1LC3A , FIS1 , MIEF2 ) and mitochondrial biogenesis markers ( ESRRA , SQSTM1 ) were markedly downregulated in the poor-prognosis Cluster B. These diametrically opposing expression profiles suggest that impaired MQC mechanisms, characterized by dysfunctional fusion-fission homeostasis and compromised autophagic clearance, are mechanistically linked to unfavorable clinical outcomes in BC patients. Mitochondrial fission is mainly governed by Drp1, while fusion is mediated by MFN1/2 and OPA1 [25]. The Drp1 is overexpressed in human invasive BC and lymph node metastases. Functional experiments revealed that Drp1 knockdown or MFN1 overexpression induced mitochondrial elongation or perinuclear clustering, respectively, and both significantly attenuated the metastatic potential of BC cells. Conversely, silencing MFN1/2 triggered mitochondrial fragmentation and enhanced BC cell metastasis[26]. Furthermore, OPA1 -mediated mitochondrial fusion in endothelial cells has been reported to promote tumor angiogenesis, thereby enhancing tumor growth and metastasis[27]. Notably, targeted inhibition of OPA1 suppresses TNBC progression, highlighting OPA1 as a promising therapeutic target in this aggressive BC subtype[28]. TFAM and PGC-1α serve as central regulatory factors in mitochondrial biogenesis[29]. Notably, TFAM is significantly overexpressed in BC, which likely contributes to tumor growth by enhancing mitochondrial DNA replication and transcription, thereby potentially supplying the energy and metabolic resources essential for the rapid cancer cell proliferation[30]. MAP1LC3A , a pivotal gene in autophagy, exerts a dual and complex role in BC[31]. In the early stages of BC, autophagy maintains genomic stability by removing damaged organelles and proteins, thereby effectively inhibiting tumorigenesis. However, during the progression of BC, autophagy paradoxically switches its function to provide energy and metabolic sustenance for tumor cells, enabling them to survive under hypoxic, nutrient-deprived, or chemotherapeutic stress conditions, consequently promoting tumor growth and metastasis[32]. MAP1LC3A overexpression in BC tissues is significantly related to tumor invasiveness, metastasis, and a poor prognosis[33]. Further in-depth investigations into the molecular mechanisms underlying MQC and its multifaceted function in tumorigenesis are warranted to facilitate the development of effective anti-cancer therapies[34]. To systematically characterize biological pathway heterogeneity between MQRG-associated subtypes, GSVA was performed. GO analysis demonstrated that cluster A was significantly upregulated in deacetylase activity, whereas cluster B exhibited pronounced activation across multiple functional categories, including cyclin-dependent serine/threonine kinase activator activity, tetrahydrofolate metabolic processes, histone kinase activity, and negative regulation of translational initiation (Figure 4B). Histone deacetylases have become promising therapeutic targets in metastatic BC owing to their ability to modulate aberrant acetylation patterns associated with disease progression[35]. Further pathway enrichment analysis revealed distinct metabolic reprogramming patterns. Specifically, cluster A was predominantly enriched in taurine/hypotaurine metabolism and ribosome biogenesis pathways, while cluster B displayed significant enrichment in folate-mediated one-carbon metabolism, cell cycle progression, DNA mismatch repair, and DNA replication pathways (Figure 4C). ICs contribute to tumor biology regulation, wherein suitable mitochondrial function is indispensable for IC phenotype, proliferation, and differentiation[36]. Numerous studies have established that the characteristics of IC infiltration signifi{Bianchini, 2022 #78}cantly differ across distinct molecular subtypes of BC[37]. Most TNBCs are generally highly immunogenic and have significantly higher stromal tumor-infiltrating lymphocytes, which is related to a favorable prognosis[6, 38]. Nevertheless, owing to the abundance of immunosuppressive factors within its TME, the anti-tumorigenic functions of these ICs are substantially impaired. In HER2-overexpressing BC, IC infiltration is also prominent and closely related to the efficacy of anti-HER2 targeted therapy[39, 40]. Consequently, the CIBERSORT algorithm showcased significant disparities in IC composition between the two molecular subtypes in BC (Figure 4D). Specifically, MQRG Cluster B displayed markedly higher infiltration of 14 IC types than Cluster A: activated B cells, activated CD4 + /CD8 + T cells, activated dendritic cells, gamma delta T cells, immature B cells, myeloid-derived suppressor cells, immature dendritic cells, macrophages, regulatory T cells, natural killers (NK), T follicular helper cells, and types 1/2 T helper cells. Given that functional cytotoxic CD8⁺ T cells are core modulators of anti-tumor immunity and are strongly related to improved patient survival[41, 42], the elevated immune infiltration in Cluster B suggests a more immunoreactive TME. 3.4 Gene subtypes based on DEGs of MQRG subtypes Here, 125 DEGs associated with MQRG subtypes were identified using the "limma" (Figure 5A). Subsequently, functional enrichment analyses were carried out. Genes associated with MQRG subtypes were significantly enriched in BPs: cell division and the regulation of humoral immune response, suggesting their potential involvement in regulating key PBs such as disease progression and immune evasion (Figure 5B). At the same time, KEGG pathway analysis revealed enrichment in the cell cycle and estrogen pathways (Figure 5C). To further validate these findings, consensus clustering was applied, and the results demonstrated that categorizing patients into two genomic subtypes based on prognostic DEG yielded the most optimal grouping outcomes (Figures S2A-F). Regarding clinical characteristics, gene cluster A showed overexpression of most prognostic DEGs compared to gene cluster B (Figure 5D). Kaplan-Meier survival analysis revealed that patients in gene cluster A exhibited a notably superior OS compared to those in cluster B(Figure 5E). There were significant differences in MQRG expression profiles between the two clusters, with most MQRGs being differentially expressed (Figure 5F). 3.5 Identification of prognostic DEGs and construction of predictive models Based on the MQRG subtype-related DEGs, a predictive RS model was developed. Initially, the "caret" was used to allocate patients in a random manner into training (n=712) and testing cohorts (n=711) at a 1:1 ratio. From an initial set of 125 MQRG subtype-related prognostic DEGs, LASSO Cox regression, guided by the minimum partial likelihood deviance criterion, identified 15 OS-related genes (Figures 6A-B). Subsequently, the 15 OS-related genes were evaluated using multivariate Cox regression analysis, thereby identifying seven genes: SLC45A1 , HPN , CHAD , CXCL9 , GLYATL2 , KRT14 , and IGLV6-57 , and displaying the corresponding gene Coefficients (Figure 6C). RSs were calculated using the following formula derived from multivariate Cox regression analysis: Comparative analysis revealed that both gene and MQRG clusters B exhibited significantly higher RSs than their respective Cluster A counterparts (Figures 6D,E). 3.6 Development and validation of predictive risk scores In the training cohort, individual RSs were calculated using the previously defined seven-gene signature (Figures 7A, D). The optimal cutoff for risk stratification, determined as the median RS using the "survminer", was used to classify patients into HRG and LRG. RSs showed an inverse relation with survival duration and a positive relation with mortality. Figure S3A depicts the distribution of BC specimens according to different classification methods. Furthermore, differential expression analysis of MQRGs across different risk groups manifested that 14 genes displayed differential expression among 20 MQRGs, with most genes being overexpressed in the LRG (Figure S3B). Figure 7G shows a heatmap of seven prognostic genes between both risk groups. KM survival analysis showed that the LRG had substantially higher survival rates than the HRG ( p < 0.001; Figure 7J). To validate the prognosis-related gene scores (PRG_scores) prognostic performance, PRG_scores were calculated across the testing and entire sets. Following the formula applied in the training set, patients were allocated into LRG or HRG. Figures 7B, E, H, and 7C, F, I depict the PRG_score distribution, patient survival status, and the seven prognostic gene expression patterns across the LRG and HRG, respectively. Survival analysis showcased that LRG exhibited significantly better prognosis than the HRG ( p < 0.001; Figures 7K,L). 3.7 Development of a nomogram for survival prediction As illustrated by the time-dependent ROC analysis, the model yielded AUCs of 0.736, 0.755, and 0.684 for 1, 3, and 5 -year survival prediction in the training group (Figure 8A). Notably, these prognostic accuracies remained consistent across the testing and comprehensive cohorts (Figures 8,C), substantiating the PRG_score as a potent and stable indicator for the survival outcomes of individuals with BC.Then, a prognostic nomogram was developed by integrating PRG scores with key clinical attributes to enable accurate individualized prediction of BC patient outcomes (Figure 8D). The calibration curve closely aligned with the ideal 45° diagonal line, indicating high agreement between predicted and observed survival probabilities. Furthermore, the nomogram exhibited strong predictive discrimination, as evidenced by a robust Concordance Index (C-index) (Figure 8E). 3.8 qRT-PCR validation and Kaplan−Meier analysis of hub genes included in prognostic model In BC tissues, CHAD , CXCL9 , IGLV6-57 , and HPN mRNA levels were significantly overexpressed relative to those in adjacent normal breast tissues; conversely, SLC45A1 , GLYATL2 , and KRT14 expressions were markedly suppressed (Figures 9A-G). Kaplan-Meier (K-M) survival analysis demonstrated that high expression of SLC45A1 , HPN , CHAD , CXCL9 , KRT14 , and IGLV6-57 correlated with improved OS, whereas GLYATL2 overexpression was related to poorer OS (all p < 0.05) (Figures S4A-G). The qRT-PCR analysis of the seven prognostic genes in the BC cell lines HCC1937 and MDA-MB-453, unlike MCF-10A, revealed a consistent dysregulation pattern. Specifically, CHAD , CXCL9 , HPN , and IGLV6-57 were significantly upregulated in both BC cell lines relative to MCF-10A, whereas SLC45A1 , GLYATL2 , and KRT14 were markedly downregulated (Figures 9H-N). Notably, these in vitro expression patterns closely mirrored the mRNA expression profiles of the corresponding genes in clinical BC tissues relative to normal breast tissues. 3.9 TME across risk groups The CIBERSORT was deployed to investigate the interaction between RSs and IC abundance. Scatter plot analysis revealed that RSs were positively associated with the abundance of memory B cells, activated dendritic cells, M0/M2 macrophages, activated mast cells, and resting NKs (Figure 10A-F), and negatively related to the abundance of M1 macrophages,CD4 + resting dendritic cells, naive B cells, monocytes, plasma cells, regulatory T cells, resting mast cells, CD8 + T cells, follicular helper T cells, and resting memory T cells (Figures 10G-P). Furthermore, the relation between the seven prognostic genes in the established model and IC abundance was examined. Notably, significant correlations were observed between most ICs and these seven genes (Figure 10Q). Figure 10R illustrates a robust association between high stromal/immune scores and low RSs. 3.10 Association of CXCL9 mRNA with M1/M2 macrophage infiltration CXCL9 has been confirmed to be expressed in some CD68+ tumor-associated macrophages(TAMs) within the stroma of BC[43], but the association between CXCL9 expression levels and M1/M2 macrophage phenotypic balance remains unvalidated. To investigate the clinical relevance of CXCL9 within BC microenvironment, we performed mIHC staining on a cohort of BC tissues and their corresponding adjacent normal tissues. The representative mIHC images indicated that CXCL9 , in addition to the macrophage markers iNOS (M1) and CD206 (M2), was markedly elevated in tumor tissues (Figure 11A), while adjacent normal tissues showed negligible expression (Figure 11B). These findings indicate that CXCL9 expression escalates significantly alongside the advancement of BC, being intimately linked to the progressive infiltration of tumor-associated macrophages (TAMs). Moreover, CXCL9 exhibited a positive correlation with the infiltration of both M1 and M2 macrophage subtypes. We further quantified the relationship between CXCL9 expression and macrophage polarization using integrated optical density measurements and cell counts. Our analysis revealed a significant positive correlation between CXCL9 levels and the infiltration of iNOS+ (M1-type) macrophages ( R = 0.3203, p < 0.0001) (Figure 11C), as well as with CD206+ (M2-type) macrophages ( R =0.052, p =0.5581)(Figure 11D). To enhance our understanding of the spatial architecture, we generated heatmaps for CXCL9+ cells. The analysis revealed a non-uniform distribution of CXCL9 , with distinct regional enrichments (Figure 11E).Using a tissue segmentation algorithm, we classified the tumor microenvironment into tumor parenchyma (yellow) and stromal compartments (red) (Figure 11F). Our findings indicate that CXCL9+ cells are strategically localized within the stroma, suggesting that their spatial arrangement may influence the recruitment and positioning of immune cells.Figure 11G illustrates the results of triple co-staining with DAPI, CXCL9, CK-pan, and Vimentin, demonstrating partial colocalization of CXCL9 with both CK-pan and Vimentin. The CXCL9+CK-pan+ double-positive cells were predominantly localized in the tumor epithelial region, whereas CXCL9+Vimentin+ double-positive cells were primarily found in the tumor stromal area or at the epithelial-stromal interface. Observations at high magnification (100 μm) revealed that CXCL9 was expressed in a punctate or small cluster-like pattern, with colocalized regions alongside CK-pan and Vimentin mostly surrounding tumor cell nests. Additionally, CXCL9 may also be expressed in stromal cells, such as fibroblasts, thereby indirectly contributing to the remodeling of the tumor microenvironment and being potentially linked to the Epithelial-Mesenchymal Transition (EMT) process in breast cancer cells. 3.11 Connection of RSs with TMB and CSC index High TMB patients are more likely to gain immunotherapy benefit, which can be attributed to the increased neoantigen load generated by elevated TMB[44]. Analyzing the mutation profiles from the TCGA-BC cohort revealed that the HRG exhibited significantly higher TMB than the LRG (Figure 12A), signaling that HRG patients may derive substantial benefits from immunotherapeutic interventions. The Spearman correlation analysis manifested a positive connection between TMB and RSs (Figure 12B). Furthermore, cancer stem cell (CSC) index values were integrated with RSs to explore their potential correlation in BC. A linear relationship was noted between the CSC index and RSs (Figure 12C). Quantitative analysis showcased a significant positive correlation (R = 0.29, p < 2.2e-16) between these parameters, implying that BC cells with heightened MQRG scores display enhanced stem-cell-like properties and minimized cellular differentiation potential. Thereafter, somatic mutation distribution between the two groups classified according to RSs within the TCGA-BC cohort was analyzed. The analysis demonstrated a significantly higher mutation frequency in the HRG in contrast to the LRG (Figures 12D-E). The top ten mutated genes were TP53 , PIK3CA , TTN , MUC16 , KMT2C , GATA3 , MAP3K1 , USH2A , CDH1 , and FLG in the HRG , and PIK3CA , TP53 , CDH1 , TTN , GATA3 , MUC16 , MAP3K1 , KMT2C , HMCN1 , and SYNE1 in the LRG . More importantly, patients in the HRG demonstrated remarkably higher TP53 , TTN , and MUC16 mutation frequencies than those in the LRG. However, PIK3CA , CDH1 , and GATA3 mutation frequencies were higher in the LRG. 3.12 Drug sensitivity analysis in different RSs Patient response to pharmacological treatment is often mirrored in their drug sensitivity. Thus, a panel of drugs typically administered for BC treatment was selected to evaluate the sensitivity levels of patients in both risk groups to these agents. Notably, patients with low RSs had lower IC50 values for all-trans retinoic acid, axitinib, bexarotene, bleomycin, bosutinib, cytarabine, dimethyloxalylglycine, gemcitabine, imatinib, lenalidomide, methotrexate, nilotinib, Nutlin, obatoclax mesylate, and rapamycin. However, patients with high RSs displayed significantly decreased IC50 values for therapeutic drugs such as A.443654, bicalutamide, CGP.082996, cisplatin, CMK, docetaxel, parthenolide, thapsigargin, RO.3306, and VX.680 (Figure S5). 4. Discussion On a global scale, BC is still a major public health challenge. Its sophisticated pathogenesis and heterogeneous clinical manifestations pose substantial obstacles to the development of effective therapeutic and preventive strategies[45]. Tumors are characterized by the ungoverned proliferation and unrestricted abnormal cell expansion, causing neoplastic masses characterized by disrupted tissue architecture, cellular pleomorphism, increased mitotic activity, and invasive capacity. Mitochondrial quality imbalance is pivotal in cancer progression[46], as dysregulated mitochondrial energy metabolism constitutes a hallmark of cancer[47]. Moreover, mitochondria critically regulate biosynthetic pathways, signal transduction, apoptosis, cellular differentiation, and cell cycle and growth control, processes that are intricately intertwined with tumorigenesis and malignant progression. Mitochondria serve as the primary site of ATP production, the central energy currency essential for cell survival and fundamental cellular processes, earning them the designation as the cell's "powerhouse." However, their implication in oxidative phosphorylation renders them particularly vulnerable to damage, as it generates high ROS levels as a byproduct. ROS can impair protein folding and structure and induce mutations in mitochondrial DNA[48]. Compounded by constant exposure to diverse environmental stressors, this susceptibility heightens the risk of mitochondrial dysfunction. To counteract these threats, eukaryotic cells have evolved a sophisticated MQC system that continuously monitors and preserves mitochondrial network integrity and functionality[49]. Core MQC mechanisms comprise mitophagy, mitochondrial fission/fusion dynamics, mitochondrial biogenesis, and proteostasis-mediated quality control of the mitochondrial proteome[7, 50-52]. Tumor cells undergo metabolic reprogramming driven by mutations, resulting in altered metabolic flux through conventional pathways used by normal cells, with increases or decreases relative to their premalignant tissue of origin[53]. In the BC microenvironment, the unlimited proliferative ability of cancer cells drives mitochondrial metabolic reprogramming, with MQC dysfunction serving as a key driver. Mutations or epigenetic silencing of core MQC genes (e.g., PINK1 , Parkin ) impair mitophagy, accumulating damaged mitochondria and excessive ROS release, which in turn induce DNA damage and genomic instability, thereby accelerating BC progression[54]. Notably, MQC abnormalities exhibit marked molecular heterogeneity across BC subtypes, with a paucity of MQC-related gene-based prognostic models. Deciphering subtype-specific MQC regulatory networks can lead to the identification of metabolic targets for personalized therapy and novel prognostic markers, representing a critical future direction. Herein, we systematically investigated the role of MQRGs in BC, unraveling their critical value in molecular subtyping, prognostic prediction, and therapeutic response through multi-omics analysis. MQC dysregulation has been reported to be pivotal in tumor progression. Relying upon MQRG expression profiles, we categorized BC patients into MQRG-Clusters A/B molecular subtypes, showing significant differences in clinicopathological features, IC infiltration of TME, and prognosis. The differential TME characteristics associated with distinct MQRG subtypes suggest that tailored therapeutic strategies can be designed to target specific TME components. Of note, cluster B patients exhibited upregulated mitochondrial fusion-related genes (e.g., TFAM , MFN1 ) and suppressed autophagy-related genes (e.g., MAP1LC3A , FIS1 ), suggesting that disrupted mitochondrial dynamic balance drives BC malignancy[7]. This finding aligns with prior studies reporting that aberrant mitochondrial fission enhances cancer cell invasiveness by remodeling metabolic pathways, while autophagic dysfunction promotes damaged mitochondria accumulation and genomic instability[9]. For instance, Drp1-mediated mitochondrial fission promotes epithelial-mesenchymal transition in BC cells, whilst MAP1LC3A deficiency accelerates tumor growth by activating the mTOR pathway[26]. The development of an MQRG-based prognostic scoring model, incorporating seven key genes ( SLC45A1 , HPN , CHAD , CXCL9 , GLYATL2 , KRT14 , and IGLV6-57 ), along with a nomogram, offers a valuable tool for clinical decision-making. Emerging data suggest that CXCL9 expression by TAMs governs CXCR3-expressing stem-like CD8 + T cell recruitment and positioning that mediate clinical responses to anti-PD(L)-1 therapy[55]. CXCL9 can form a functional synergistic network with the receptor CXCR3 and contributes to tumor immune regulation. Especially in malignant tumors such as BC, it profoundly influences tumor progression and prognosis by governing IC recruitment and differentiation besides TME remodeling[56]. Unlike healthy tissues, KRT14 is downregulated in BC tissues, which is related to poor prognosis[57], possibly due to elevated KRT14 methylation levels in BC. KRT14 is implicated in tumor cells' invasive and migratory capabilities and is associated with the TME; thus, it may be a reliable prognostic biomarker[58]. Furthermore, earlier studies have concluded that KRT14 may be a candidate metastasis regulator in TNBC, with its upregulated expression promoting the peritoneal metastasis of TNBC[59]. Notably, the RS effectively discriminated between HRG and LRG, with a clear separation in OS outcomes. The nomogram, with its relatively high predictive accuracy for 1-, 3-, and 5-year survival rates, offers a more comprehensive and accurate prognostic tool. This can aid in stratifying patients according to their risk of disease recurrence and mortality, enabling the implementation of more appropriate therapeutic strategies. High-risk patients could potentially benefit from more aggressive treatment regimens, including intensified chemotherapy or early-stage immunotherapy, whereas low-risk patients might avoid unnecessary treatment-related toxicities. Notably, the HRG showed significantly higher TMB than the LRG, with a positive relation between TMB and RSs, aligning with findings of KEYNOTE-522 that demonstrated that high-TMB TNBC patients derive greater benefit from immunotherapy, suggesting that HRG patients may be optimal candidates for immune checkpoint inhibitor (ICI) treatment[60]. The ICI pembrolizumab integrated with chemotherapy had FDA approval for PD-L1-positive metastatic and early-stage TNBC[61]. Immune infiltration analysis revealed that HRG patients had increased infiltration levels of memory B cells and M0 macrophages but a reduced level of M1 macrophages and resting CD8 + T cells, indicating an imbalanced immune microenvironment wherein tumor cells may induce immunogenicity via high TMB while upregulating immune checkpoint molecules[6]. Emerging evidence has confirmed that the chemokines CXCL9/10 are crucial for vigorous responses to ICIs (anti-PD-1 and anti-CTLA-4) and, particularly, that CXCL9/10-secreting macrophages are vital for their therapeutic efficiency[62]. The spatiotemporal orchestration of the tumor microenvironment (TME) by chemokines is a fundamental determinant of BC progression and therapeutic response. In the present study, we characterized the expression landscape of CXCL9 and its complex relationship with macrophage polarization, revealing that CXCL9 is not merely a bystander in tumor progression but acts as a potential immunological rheostat that modulates the recruitment and phenotypic balance of tumor-associated macrophages (TAMs). While previous research identified CXCL9 expression in stromal CD68+ TAMs in BC, the functional link between CXCL9 levels and M1/M2 phenotypic equilibrium was not well understood. Our quantitative multiplex immunohistochemistry (mIHC) analysis provides robust evidence for the phenotypic preference of CXCL9, demonstrating a significant positive correlation with iNOS+ M1-type macrophages ( R = 0.3203, p < 0.0001) , while its correlation with CD206+ M2-type macrophages was minimal( R =0.052, p =0.5581). This differential association suggests that CXCL9 serves as a selective chemoattractant or paracrine inducer for pro-inflammatory, anti-tumorigenic M1 macrophages rather than a general driver of myeloid infiltration. Given the essential role of M1 macrophages in Th1-mediated anti-tumor immunity[63], the CXCL9-M1 axis may be a key mechanism for converting "cold" tumors into "hot," immune-active environments[64]. Additionally, our findings reveal the dual cellular origin of CXCL9 within the BC microenvironment. Its partial colocalization with CK-pan and Vimentin indicates production by both malignant epithelial cells and mesenchymal stromal components. The presence of CXCL9+CK-pan+ cells within tumor nests, along with CXCL9+Vimentin+ clusters at the epithelial-stromal interface, points to a collaborative secretory network. High-magnification imaging shows CXCL9 forming clusters around tumor nests, implying the presence of "chemotactic hubs" that attract CXCR3-expressing immune cells from the stroma into the tumor. Moreover, the correlation between CXCL9 and Vimentin+ cells associates CXCL9 with the EMT process, where Vimentin marks mesenchymal traits often observed in cancer-associated fibroblasts (CAFs). The localization of CXCL9 at the tumor's invasive front may reflect a compensatory immune response to counteract immunosuppressive signals linked to EMT-driven stromal changes.Clinical Implications for Prognostic StratificationFrom a clinical perspective, the association between high CXCL9 expression and prolonged overall survival (OS) can be attributed to the "normalization" of the immune landscape. By fostering an environment rich in M1-type macrophages, CXCL9 promotes a "hot" tumor phenotype characterized by enhanced antigen presentation and T-cell activation. This molecular milieu not only naturally suppresses tumor growth but also potentially heightens the sensitivity to immune checkpoint inhibitors (ICIs). Consequently, the Mitochondrial Quality Regulation Gene Signature, with CXCL9 as a core component, offers a dual-purpose tool: it stratifies patients by risk while simultaneously identifying those who may possess a more responsive, immune-active microenvironment.In conclusion, our study positions CXCL9 as a pivotal orchestrator of the immune landscape in breast cancer, promoting the recruitment of M1 macrophages. The positive correlation between CSC index and RS (R = 0.29, p < 2.2e-16) suggests that BC with high MQRG scores exhibit enhanced stemness, potentially mediating resistance to conventional chemotherapy. It is worthwhile emphasizing that patients in the HRG had significantly increased TP53 , TTN, and MUC16 mutation frequencies more than those in the LRG. Among these, TP53 , mutated in approximately 30% of BC cases, is the most frequently altered gene in this malignancy[65]. TP53 encodes the p53 tumor suppressor, a transcription factor that activates genes governing cell cycle arrest, apoptosis, DNA repair, metabolic reprogramming, and senescence in response to genotoxic stressors such as radiation and chemotherapy[66]. Accumulating evidence underscores p53 as a therapeutically relevant target in BC, particularly in TNBC and HER2-positive subtypes characterized by a high TP53 mutation burden[67]. Drug sensitivity analysis further revealed distinct responses to cisplatin and docetaxel between the HRG and LRG, providing a basis for personalized chemotherapy. Indeed, the observed differential drug sensitivity between both risk groups holds significant implications for personalized chemotherapy. The disparities in IC50 values between the risk groups provide a theoretical reference for tailoring chemotherapy regimens. These findings can be utilized to select drugs that are more likely to be effective for individual patients, minimizing the use of ineffective drugs and reducing the risk of associated toxicities. This personalized approach to chemotherapy can enhance treatment efficacy and patient quality of life. Notwithstanding, some limitations of this study cannot be overlooked. To begin, the molecular mechanisms behind MQRG's effects on BC development remain incompletely understood. While this study established associations between MQRGs and various aspects of BC, the precise signaling pathways and regulatory networks through which these genes exert their effects necessitate further investigation. For example, the role of MQRGs in mitochondrial dynamics, bioenergetics, and their crosstalk with other cellular processes, such as DNA damage response and epigenetic regulation, needs to be elucidated. Elucidating these mechanisms is paramount for developing more targeted and effective therapies. Another limitation is the reliance on data from public databases. Although these databases provide extensive data resources, the lack of validation in independent clinical cohorts may limit generalizability. Variations in patient populations, data collection methods, and treatment protocols across different studies can introduce biases. Future research should validate these findings in large-scale, well-characterized clinical cohorts to confirm the robustness of the identified associations and prognostic models. Furthermore, the scope of drug sensitivity analysis was limited to a subset of chemotherapeutic agents. With the rapid advancement of novel immunotherapies and targeted therapies in oncology, investigating the sensitivity of different MQRG subtypes to these new treatment modalities is crucial. Future studies should include more drugs, including emerging targeted agents and immunotherapeutic drugs, to provide more comprehensive insights to inform personalized therapeutic strategies. 5. Conclusion This study establishes a robust 7-gene Mitochondrial Quality Regulation Gene Signature as an independent prognostic indicator for breast cancer. Through the integration of multi-dimensional omics data and experimental assays, we elucidated the role of CXCL9 as a spatial orchestrator that promotes pro-inflammatory M1 macrophage recruitment and polarization. This signature provides a comprehensive framework for risk stratification and the prediction of therapeutic responses, including chemotherapy and immunotherapy. Our findings suggest that targeting mitochondrial-immune crosstalk represents a promising strategy for remodeling the tumor microenvironment and improving patient outcomes in breast cancer. Abbreviations BC CAFs Breast Cancer cancer-associated fibroblasts CDF Cumulative Distribution Function CNV Copy Number Variant CSC Cancer Stem Cell DEGs EMT Differentially Expressed Genes Epithelial-Mesenchymal Transition GO Gene Ontology GSVA Gene Set Variation Analysis HRG High-Risk Group IC Immune Cell KEGG Kyoto Encyclopedia Of Genes And Genomes KM Kaplan-Meier LRG mIHC Low-Risk Group multiplex immunohistochemistry MQRGs Mitochondrial Quality-Related Genes OS Overall Survival PCA qRT-PCR Principal Components Analysi Quantitative real-time PCR ROC Receiver Operating Characteristic RS Risk Score TCGA TAMs The Cancer Genome Atlas tumor-associated macrophages TMB Tumor Mutation Burden TME Tumor Microenvironment Declarations Acknowledgements We extend our gratitude to GEO and TCGA database, and all contributors who generously shared their data on these platforms.This study would like to extend its gratitude to Hunan AiFang Biologcal Co., Ltd. for providing relevant antibodies, multiplex fluorescence staining kits, as well as staining, scanning, and data analysis services. Author contributions Huaiwen Pu: Conceptualization, Formal analysis, Visualization, Writing – original draft. Tingjing Li: Formal analysis, Visualization, Writing – original draft. Renji Liang: Visualization, Writing – original draft. Zhongxiang Fan: Software, Formal analysis, Writing – original draft. Bowen Tang: Data curation, Formal analysis, Visualization, Writing – original draft. Yongmei Luo: Data curation, Validation, Writing – review & editing. Xinyu Yi: Data curation, Validation, Writing – review & editing. Liming Xie: Project administration, Validation，Writing – review & editing. Yuehua Li: Conceptualization, Methodology, Supervision, Writing – review & editing. Funding This research was funded by the Interdisciplinary Research Program in Medicine and Engineering, the First Affiliated Hospital of University of South China (No.IRP-M&E-2025-05) and Clinical Medical Research 4310 Program of the University of South China (20224310NHYCG07). Data availability All data generated or analysed during this study are included in this published article and its supplementary information files. The source data utilized for the development of the prognostic model are accessible via public databases. Further technical inquiries should be addressed to the corresponding authors. Declarations Ethics approval and consent to participate The use of human tumor tissue microarray samples in this study was approved by the Ethics Committee on Biological Science and Technology of Hunan Aifang Biological Co., Ltd. (Approval No. HN20250401). All procedures were performed in accordance with the Declaration of Helsinki, and written informed consent was obtained from all participants or their legal guardians. Consent for publication All the authors give the consent for the publication of identifiable details, which can include the text, figures and other materials in this manuscript. Competing interests The authors declare that they have no competing interests. References Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin. 2021;71(3):209-49. Giaquinto AN, Sung H, Newman LA, Freedman RA, Smith RA, Star J, et al. Breast cancer statistics 2024. CA Cancer J Clin. 2024;74(6):477-95. Loibl S, Poortmans P, Morrow M, Denkert C, Curigliano G. Breast cancer. Lancet. 2021;397(10286):1750-69. Waks AG, Winer EP. Breast Cancer Treatment: A Review. Jama. 2019;321(3):288-300. Xiong X, Zheng LW, Ding Y, Chen YF, Cai YW, Wang LP, et al. Breast cancer: pathogenesis and treatments. Signal Transduct Target Ther. 2025;10(1):49. Bianchini G, De Angelis C, Licata L, Gianni L. Treatment landscape of triple-negative breast cancer - expanded options, evolving needs. Nat Rev Clin Oncol. 2022;19(2):91-113. Song J, Herrmann JM, Becker T. Quality control of the mitochondrial proteome. Nat Rev Mol Cell Biol. 2021;22(1):54-70. Choong CJ, Okuno T, Ikenaka K, Baba K, Hayakawa H, Koike M, et al. Alternative mitochondrial quality control mediated by extracellular release. Autophagy. 2021;17(10):2962-74. Luo Y, Ma J, Lu W. The Significance of Mitochondrial Dysfunction in Cancer. Int J Mol Sci. 2020;21(16). Fontana F, Limonta P. The multifaceted roles of mitochondria at the crossroads of cell life and death in cancer. Free Radic Biol Med. 2021;176:203-21. Liu BH, Xu CZ, Liu Y, Lu ZL, Fu TL, Li GR, et al. Mitochondrial quality control in human health and disease. Mil Med Res. 2024;11(1):32. Rehman J, Zhang HJ, Toth PT, Zhang Y, Marsboom G, Hong Z, et al. Inhibition of mitochondrial fission prevents cell cycle progression in lung cancer. Faseb j. 2012;26(5):2175-86. Xiong X, Hasani S, Young LEA, Rivas DR, Skaggs AT, Martinez R, et al. Activation of Drp1 promotes fatty acids-induced metabolic reprograming to potentiate Wnt signaling in colon cancer. Cell Death Differ. 2022;29(10):1913-27. Kannan A, Wells RB, Sivakumar S, Komatsu S, Singh KP, Samten B, et al. Mitochondrial Reprogramming Regulates Breast Cancer Progression. Clin Cancer Res. 2016;22(13):3348-60. Serasinghe MN, Wieder SY, Renault TT, Elkholi R, Asciolla JJ, Yao JL, et al. Mitochondrial division is requisite to RAS-induced transformation and targeted by oncogenic MAPK pathway inhibitors. Mol Cell. 2015;57(3):521-36. Gao T, Zhang X, Zhao J, Zhou F, Wang Y, Zhao Z, et al. SIK2 promotes reprogramming of glucose metabolism through PI3K/AKT/HIF-1α pathway and Drp1-mediated mitochondrial fission in ovarian cancer. Cancer Lett. 2020;469:89-101. Lee YG, Nam Y, Shin KJ, Yoon S, Park WS, Joung JY, et al. Androgen-induced expression of DRP1 regulates mitochondrial metabolic reprogramming in prostate cancer. Cancer Lett. 2020;471:72-87. Nagdas S, Kashatus JA, Nascimento A, Hussain SS, Trainor RE, Pollock SR, et al. Drp1 Promotes KRas-Driven Metabolic Changes to Drive Pancreatic Tumor Growth. Cell Rep. 2019;28(7):1845-59.e5. Rath S, Sharma R, Gupta R, Ast T, Chan C, Durham TJ, et al. MitoCarta3.0: an updated mitochondrial proteome now with sub-organelle localization and pathway annotations. Nucleic Acids Res. 2021;49(D1):D1541-d7. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000;28(1):27-30. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25-9. Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7. Mayakonda A, Lin DC, Assenov Y, Plass C, Koeffler HP. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res. 2018;28(11):1747-56. Yan C, Niu Y, Ma L, Tian L, Ma J. System analysis based on the cuproptosis-related genes identifies LIPT1 as a novel therapy target for liver hepatocellular carcinoma. J Transl Med. 2022;20(1):452. Tábara LC, Segawa M, Prudent J. Molecular mechanisms of mitochondrial dynamics. Nat Rev Mol Cell Biol. 2025;26(2):123-46. Zhao J, Zhang J, Yu M, Xie Y, Huang Y, Wolff DW, et al. Mitochondrial dynamics regulates migration and invasion of breast cancer cells. Oncogene. 2013;32(40):4814-24. Herkenne S, Ek O, Zamberlan M, Pellattiero A, Chergova M, Chivite I, et al. Developmental and Tumor Angiogenesis Requires the Mitochondria-Shaping Protein Opa1. Cell Metab. 2020;31(5):987-1003.e8. Zamberlan M, Boeckx A, Muller F, Vinelli F, Ek O, Vianello C, et al. Inhibition of the mitochondrial protein Opa1 curtails breast cancer growth. J Exp Clin Cancer Res. 2022;41(1):95. Itoh Y, Khawaja A, Laptev I, Cipullo M, Atanassov I, Sergiev P, et al. Mechanism of mitoribosomal small subunit biogenesis and preinitiation. Nature. 2022;606(7914):603-8. Gao W, Wu M, Wang N, Zhang Y, Hua J, Tang G, et al. Increased expression of mitochondrial transcription factor A and nuclear respiratory factor-1 predicts a poor clinical outcome of breast cancer. Oncol Lett. 2018;15(2):1449-58. Othman EQ, Kaur G, Mutee AF, Muhammad TS, Tan ML. Immunohistochemical expression of MAP1LC3A and MAP1LC3B protein in breast carcinoma tissues. J Clin Lab Anal. 2009;23(4):249-58. Wu Q, Sharma D. Autophagy and Breast Cancer: Connected in Growth, Progression, and Therapy. Cells. 2023;12(8). Sivridis E, Koukourakis MI, Zois CE, Ledaki I, Ferguson DJ, Harris AL, et al. LC3A-positive light microscopy detected patterns of autophagy and prognosis in operable breast carcinomas. Am J Pathol. 2010;176(5):2477-89. Zong Y, Li H, Liao P, Chen L, Pan Y, Zheng Y, et al. Mitochondrial dysfunction: mechanisms and advances in therapy. Signal Transduct Target Ther. 2024;9(1):124. Guadagni A, Barone S, Alfano AI, Pelliccia S, Bello I, Panza E, et al. Tackling triple negative breast cancer with HDAC inhibitors: 6 is the isoform! Eur J Med Chem. 2024;279:116884. Bai R, Cui J. Mitochondrial immune regulation and anti-tumor immunotherapy strategies targeting mitochondria. Cancer Lett. 2023;564:216223. Onkar SS, Carleton NM, Lucas PC, Bruno TC, Lee AV, Vignali DAA, et al. The Great Immune Escape: Understanding the Divergent Immune Response in Breast Cancer Subtypes. Cancer Discov. 2023;13(1):23-40. Keenan TE, Tolaney SM. Role of Immunotherapy in Triple-Negative Breast Cancer. J Natl Compr Canc Netw. 2020;18(4):479-89. Bianchini G, Gianni L. The immune system and response to HER2-targeted treatment in breast cancer. Lancet Oncol. 2014;15(2):e58-68. Griguolo G, Pascual T, Dieci MV, Guarneri V, Prat A. Interaction of host immunity with HER2-targeted treatment and tumor heterogeneity in HER2-positive breast cancer. J Immunother Cancer. 2019;7(1):90. Zhong X, Wu H, Ouyang C, Zhang W, Shi Y, Wang YC, et al. Ncoa2 Promotes CD8+ T cell-Mediated Antitumor Immunity by Stimulating T-cell Activation via Upregulation of PGC-1α Critical for Mitochondrial Function. Cancer Immunol Res. 2023;11(10):1414-31. Zhang L, Romero P. Metabolic Control of CD8(+) T Cell Fate Decisions and Antitumor Immunity. Trends Mol Med. 2018;24(1):30-48. DU Shaoqian TM, CAO Yuan, WANG Hongxia, HU Xiaoqu, FAN Guangjian, ZANG Lijuan. . CXCL9 expression in breast cancer and its correlation with the characteristics of tumor immunoinfiltration. Journal of Shanghai Jiao Tong University (Medical Science). 2023;43(7):860-72. Chan TA, Yarchoan M, Jaffee E, Swanton C, Quezada SA, Stenzinger A, et al. Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic. Ann Oncol. 2019;30(1):44-56. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229-63. Zong WX, Rabinowitz JD, White E. Mitochondria and Cancer. Mol Cell. 2016;61(5):667-76. Hanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov. 2022;12(1):31-46. Pickles S, Vigié P, Youle RJ. Mitophagy and Quality Control Mechanisms in Mitochondrial Maintenance. Curr Biol. 2018;28(4):R170-r85. Ni HM, Williams JA, Ding WX. Mitochondrial dynamics and mitochondrial quality control. Redox Biol. 2015;4:6-13. Adebayo M, Singh S, Singh AP, Dasgupta S. Mitochondrial fusion and fission: The fine-tune balance for cellular homeostasis. Faseb j. 2021;35(6):e21620. Scarpulla RC. Transcriptional paradigms in mammalian mitochondrial biogenesis and function. Physiol Rev. 2008;88(2):611-38. Gustafsson Å B, Dorn GW, 2nd. Evolving and Expanding the Roles of Mitophagy as a Homeostatic and Pathogenic Process. Physiol Rev. 2019;99(1):853-92. DeBerardinis RJ, Chandel NS. Fundamentals of cancer metabolism. Sci Adv. 2016;2(5):e1600200. Li Q, Chu Y, Li S, Yu L, Deng H, Liao C, et al. The oncoprotein MUC1 facilitates breast cancer progression by promoting Pink1-dependent mitophagy via ATAD3A destabilization. Cell Death Dis. 2022;13(10):899. Marcovecchio PM, Thomas G, Salek-Ardakani S. CXCL9-expressing tumor-associated macrophages: new players in the fight against cancer. J Immunother Cancer. 2021;9(2). Pan M, Wei X, Xiang X, Liu Y, Zhou Q, Yang W. Targeting CXCL9/10/11-CXCR3 axis: an important component of tumor-promoting and antitumor immunity. Clin Transl Oncol. 2023;25(8):2306-20. Fridman WH, Pagès F, Sautès-Fridman C, Galon J. The immune contexture in human tumours: impact on clinical outcome. Nat Rev Cancer. 2012;12(4):298-306. Liao S, Zhang X, Chen L, Zhang J, Lu W, Rao M, et al. KRT14 is a promising prognostic biomarker of breast cancer related to immune infiltration. Mol Immunol. 2025;180:55-73. Verma A, Singh A, Singh MP, Nengroo MA, Saini KK, Satrusal SR, et al. EZH2-H3K27me3 mediated KRT14 upregulation promotes TNBC peritoneal metastasis. Nat Commun. 2022;13(1):7344. Schmid P, Cortes J, Pusztai L, McArthur H, Kümmel S, Bergh J, et al. Pembrolizumab for Early Triple-Negative Breast Cancer. N Engl J Med. 2020;382(9):810-21. Heater NK, Warrior S, Lu J. Current and future immunotherapy for breast cancer. J Hematol Oncol. 2024;17(1):131. House IG, Savas P, Lai J, Chen AXY, Oliver AJ, Teo ZL, et al. Macrophage-Derived CXCL9 and CXCL10 Are Required for Antitumor Immune Responses Following Immune Checkpoint Blockade. Clin Cancer Res. 2020;26(2):487-504. Mantovani A, Marchesi F, Malesci A, Laghi L, Allavena P. Tumour-associated macrophages as treatment targets in oncology. Nat Rev Clin Oncol. 2017;14(7):399-416. Liang B, Duan Z, Long S, Zhou P. Infiltration of CXCL9+ macrophages confers a favorable prognosis in breast cancer: Insights from an integrated single-cell RNA and bulk RNA sequencing study. PLoS One. 2025;20(12):e0337175. Comprehensive molecular portraits of human breast tumours. Nature. 2012;490(7418):61-70. Kastenhuber ER, Lowe SW. Putting p53 in Context. Cell. 2017;170(6):1062-78. Marvalim C, Datta A, Lee SC. Role of p53 in breast cancer progression: An insight into p53 targeted therapy. Theranostics. 2023;13(4):1421-42. Additional Declarations No competing interests reported. Supplementary Files SupplementaryMaterial.pdf Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 07 May, 2026 Reviews received at journal 01 May, 2026 Reviewers agreed at journal 18 Apr, 2026 Reviews received at journal 12 Apr, 2026 Reviewers agreed at journal 09 Apr, 2026 Reviewers agreed at journal 19 Mar, 2026 Reviewers invited by journal 05 Mar, 2026 Editor assigned by journal 03 Mar, 2026 Submission checks completed at journal 03 Mar, 2026 First submitted to journal 28 Feb, 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-8995736","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":603150775,"identity":"dd4cad7a-bc75-49d4-9f6c-b00505bc483a","order_by":0,"name":"Huaiwen Pu","email":"","orcid":"","institution":"First Affiliated Hospital of University of South China","correspondingAuthor":false,"prefix":"","firstName":"Huaiwen","middleName":"","lastName":"Pu","suffix":""},{"id":603150776,"identity":"946ab127-3bd9-4beb-b510-1ce8d820fefb","order_by":1,"name":"Tingjing Li","email":"","orcid":"","institution":"First Affiliated Hospital of University of South China","correspondingAuthor":false,"prefix":"","firstName":"Tingjing","middleName":"","lastName":"Li","suffix":""},{"id":603150777,"identity":"ee0cff4f-a2c9-46a4-ada3-3a8d42636071","order_by":2,"name":"Renji Liang","email":"","orcid":"","institution":"First Affiliated Hospital of University of South China","correspondingAuthor":false,"prefix":"","firstName":"Renji","middleName":"","lastName":"Liang","suffix":""},{"id":603150778,"identity":"701c2e96-5f4a-4efc-8e5f-e544ead64b53","order_by":3,"name":"Zhongxiang Fan","email":"","orcid":"","institution":"First Affiliated Hospital of University of South China","correspondingAuthor":false,"prefix":"","firstName":"Zhongxiang","middleName":"","lastName":"Fan","suffix":""},{"id":603150779,"identity":"5a8a5235-f7ec-482c-99ef-7681ab48758c","order_by":4,"name":"Bowen Tang","email":"","orcid":"","institution":"First Affiliated Hospital of University of South China","correspondingAuthor":false,"prefix":"","firstName":"Bowen","middleName":"","lastName":"Tang","suffix":""},{"id":603150780,"identity":"3e71530d-1051-4298-9199-5d425f28068e","order_by":5,"name":"Yongmei Luo","email":"","orcid":"","institution":"First Affiliated Hospital of University of South China","correspondingAuthor":false,"prefix":"","firstName":"Yongmei","middleName":"","lastName":"Luo","suffix":""},{"id":603150781,"identity":"8fa3ee7a-2b85-4a3a-822b-aee1e45bb1a9","order_by":6,"name":"Xinyu Yi","email":"","orcid":"","institution":"First Affiliated Hospital of University of South China","correspondingAuthor":false,"prefix":"","firstName":"Xinyu","middleName":"","lastName":"Yi","suffix":""},{"id":603150782,"identity":"ffcd7b4d-b458-4a0c-8d4a-d2460e93b644","order_by":7,"name":"Liming Xie","email":"","orcid":"","institution":"First Affiliated Hospital of University of South China","correspondingAuthor":false,"prefix":"","firstName":"Liming","middleName":"","lastName":"Xie","suffix":""},{"id":603150783,"identity":"fca25301-7a07-4022-a17d-08152caf593d","order_by":8,"name":"Yuehua Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABEElEQVRIie3RMUvEMBTA8RcCjUO0a46gfoVKoChV/CoJgl3uXFxuuKEiPBcP1/ox3NwsBJzqfqNdnBzqplDFtHgI0uqNDvkP5ZHwKyEB8Pn+YRvUfUgGEFJaFPX3Bh0kwZKMLtBUOUQrEPgiUVkqxVcijMfi5dZuw0LHcr9pTCbSqoZpYjL2UPQfjMej69LunOX6WE4wcmSsBJSpyfiJHiJyHS05F/peTrKOBEDQuoFHg+Qd7SEKg3K3aUn6BOTjD9L+85JbqiBoiY7dhfxGgtO9OaZHOUNSzVEp5M/KHTJ1w7iXhKG9WbxhcnBnw7p4bbY2r5i7sXqWuKHsJQBrP9a7l9LLoTf2OLjl8/l8vq5PM59ZR26RmnMAAAAASUVORK5CYII=","orcid":"","institution":"First Affiliated Hospital of University of South China","correspondingAuthor":true,"prefix":"","firstName":"Yuehua","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2026-02-28 13:38:38","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8995736/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8995736/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104784219,"identity":"47791594-218f-42ee-9208-502186b0e1c7","added_by":"auto","created_at":"2026-03-17 08:05:50","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":3684566,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe research methodology.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/0727fbae3d788577596aa43d.png"},{"id":104785650,"identity":"20038062-a137-4bf2-a6d8-169ceedd290a","added_by":"auto","created_at":"2026-03-17 08:12:35","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":8553146,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGenetic and transcriptional alterations of 20 MQRGs in BC.\u003c/strong\u003e (A) Differential expression of the 20 MQRGs between BC and normal breast tissues. (B) CNV frequency across the 20 MQRGs. (C) Chromosomal localization of the 20 MQRGs, displayed in a circular layout. (D) Interactions between MQRGs in BC; edge thickness reflects interaction strength, with pink and blue lines denoting positive and negative correlations, respectively.(E) Correlation analysis for the MQRGs. (F) Oncoplot displaying the somatic mutation landscape of MQRGs in 991 breast cancer samples.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/9d922f1f17b6a914da28a3af.png"},{"id":104784249,"identity":"0bfd272c-da01-4084-a594-7f5010a70aad","added_by":"auto","created_at":"2026-03-17 08:06:21","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1900873,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of two molecular BC subtypes based on MQRG expression via consensus clustering. \u003c/strong\u003e(A) Consensus matrix heatmap showing sample clustering for k = 2, revealing two stable MQRG subtypes (MQRG Cluster A and B) with corresponding consensus correlation blocks. (B) CDF curves for consensus indices across increasing numbers of clusters (k), representing the proportion of sample pairs consistently co-clustered (1.0 = co-clustered 100% of the time). (C) Delta area plot illustrating the change in area under the CDF curve relative to k − 1, used to determine optimal cluster number; the x-axis denotes cluster number (k), and the y-axis reflects the relative increase in consensus area. (D) Tracking plot. (E) PCA plots based on prognostic MQRG expression, clearly separating patients into two subtypes; blue and yellow dots represent MQRG Clusters A and B, respectively. (F) KM curves for OS of both MQRG subtypes (chi-square test, \u003cem\u003ep\u003c/em\u003e =0.006).\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/3ba8ed6597cff486d10e39de.png"},{"id":104784232,"identity":"024394bc-3356-4c95-9dd2-ad1aa390771d","added_by":"auto","created_at":"2026-03-17 08:05:58","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":3004180,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eClinicopathological features and immune microenvironment between the two MQRG subtypes. \u003c/strong\u003e(A) Heatmap depicting differences in clinical features and MQRG expression profiles between MQRG Cluster A and B. (B) GSVA of molecular functions across subtypes; red mirrors upregulated (activated) functions, and blue reflects downregulated (suppressed) functions. (C) GSVA of biological pathways, with red and blue denoting pathway activation and suppression, respectively, in the two subtypes. (D) Relative abundances of 23 immune cells (ICs) populations in the TME, comparing infiltration levels between MQRG Cluster A and B.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/26f6329147d64e6b0516c741.png"},{"id":104785569,"identity":"dd80e881-97f6-4106-ad30-70dbb2a728fd","added_by":"auto","created_at":"2026-03-17 08:12:19","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":6320872,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDerivation of gene subtypes based on MQRG-associated DEGs. \u003c/strong\u003e(A) Venn diagram identifying 125 DEGs significantly associated with the two MQRG subtypes. (B-C) GO and KEGG analysis of DEGs. (D) Heatmap showing differences in clinical features and expressions of the prognostic DEGs between gene Cluster A and B. (E) KM OS curves for both gene subtypes, demonstrating a significant survival difference (log-rank tests, \u003cem\u003ep\u003c/em\u003e\u0026lt; 0.001). (F) Expression patterns of the 20 MQRGs across the two gene subtypes, highlighting subtype-specific regulation.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/0b95cad0b14eaad54ca057f5.png"},{"id":104785815,"identity":"7d9544e2-07de-4c47-b2ce-ccb4353811c0","added_by":"auto","created_at":"2026-03-17 08:13:07","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":2816277,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDevelopment of the RS model using LASSO regression. \u003c/strong\u003e(A-B) LASSO regression: Potential prognostic genes discovered together with the partial likelihood deviation on these genes. (C) Names and Coefficients of Key Prognostic Genes. (D) Disparity in RSs between the two gene subtypes and (E) MQRG subtypes.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/7eacebd8024c2f99782963b1.png"},{"id":104788195,"identity":"50a5f389-cad2-4bd1-9abd-f909f7098d9a","added_by":"auto","created_at":"2026-03-17 08:24:08","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":3712082,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRS evaluation and patient stratification across cohorts. \u003c/strong\u003eIn the training, testing, and entire cohorts: (A-C) Distribution of RSs and their association with patient survival status, (D-F) Scatter plots depicting the relation between survival time and RS, (G-I) Expression profiles of the seven prognostic genes, and (J-L) KM survival curve.\u003c/p\u003e","description":"","filename":"Figure7.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/bc516f1602b704e4d84251cf.png"},{"id":104786354,"identity":"e9e7348d-0e37-4d02-a315-fbe77c63fdde","added_by":"auto","created_at":"2026-03-17 08:16:46","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":1771581,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDevelopment and validation of a prognostic BC nomogram. \u003c/strong\u003e(A-C) Time-dependent ROC curves assessing RS predictive accuracy for 1-, 3-, and 5-year OS in all cohorts. (D) Prognostic nomogram integrating clinical variables and the RS to predict individual survival probability. (B) Calibration plot evaluating the agreement between predicted and observed survival probabilities.\u003c/p\u003e","description":"","filename":"Figure8.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/e393a6575cb26676b9c28e3c.png"},{"id":104785682,"identity":"38d08ec7-a109-416c-90f8-271039578db7","added_by":"auto","created_at":"2026-03-17 08:12:45","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":2394842,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eExpression landscape and experimental validation of the seven model-related genes in BC. \u003c/strong\u003e(A-G) Expression differences of 7 hub genes (\u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eGLYATL2\u003c/em\u003e, \u003cem\u003eKRT14\u003c/em\u003e, \u003cem\u003eHPN\u003c/em\u003e, \u003cem\u003eIGLV6-57\u003c/em\u003e, and \u003cem\u003eSLC45A1\u003c/em\u003e) in normal and breast cancer tissues in the TCGA and GTEx datasets.(H-N) Validation of the expression of 7 hub genes by qRT-qPCR in cell lines (MCF10A, HCC1937, and MDA-MB-453).ns (\u003cem\u003ep \u003c/em\u003e≥ 0.05), *(\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.05), **(\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01), ***(\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.001), and****(\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.0001)\u003c/p\u003e","description":"","filename":"Figure9.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/ebd9af812eb75d0f6983b4b9.png"},{"id":104785513,"identity":"269a232a-ff17-4eb2-8051-062463788b30","added_by":"auto","created_at":"2026-03-17 08:12:06","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":5450426,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTME and immune checkpoint profiles across risk groups. \u003c/strong\u003e(A-P) IC types and RSs connection. (Q) Correlation heatmap of gene models and IC contents. (R) Violin diagram: TME scores in HRG/LRG.\u003c/p\u003e","description":"","filename":"Figure10.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/ecd04a046d6ecca96c6eb0d9.png"},{"id":104785658,"identity":"4a621a27-356b-459d-bced-9aa5bc3139eb","added_by":"auto","created_at":"2026-03-17 08:12:37","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":13269946,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSpatial distribution and correlation analysis of \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eCXCL9\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003eand macrophage polarization in tumor tissues. \u003c/strong\u003e(A, B) Representative multiplex mIHC images illustrating the expression of \u003cem\u003eCXCL9\u003c/em\u003e(green), iNOS (M1 marker, red), and CD206 (M2 marker, orange/yellow) in tumor (A) and adjacent normal tissues (B). Nuclei are stained with DAPI (blue). Scale bars are provided in each panel (100 μm and 1 mm).(C, D) Spearman correlation analysis depicting the relationship between the expression levels of \u003cem\u003eCXCL9\u003c/em\u003e and the infiltration density of iNOS+ M1-like macrophages as well as CD206+ M2-like macrophages.(E) Spatial density heatmap depicting the distribution and infiltration intensity of CXCL9+ cells within the tumor tissue. (F) Segmentation map of the tumor microenvironment, distinguishing regions with the tumor parenchyma (nests) highlighted in yellow and the stromal compartments marked in red.(G) Representative mIHCimages showcasing the spatial colocalization of \u003cem\u003eCXCL9\u003c/em\u003e (green) with the epithelial marker CK-pan (cyan) and mesenchymal marker Vimentin (red). Nuclei are counterstained with DAPI (blue). Scale bars = 100 μm and 500 μm.\u003c/p\u003e","description":"","filename":"Figure11.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/f5eadf91b2d1f8d7f0898c2d.png"},{"id":104785646,"identity":"5247ce28-b31c-4d81-b673-a0f6acb0617a","added_by":"auto","created_at":"2026-03-17 08:12:33","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":2412194,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTumor mutational landscape across risk groups. \u003c/strong\u003e(A) TMB expression in different RSs. (B) Spearman correlation between TMB and RS. (C) CSC index-RS association. (D) Somatic mutation features result in high- and (E) low RSs; each column represents a single patient, the top barplot indicates TMB, the right barplot shows mutation type distribution, and the numerical values on the right denote gene-specific mutation frequencies.\u003c/p\u003e","description":"","filename":"Figure12.png","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/67c9e18637d34f355df18348.png"},{"id":104808878,"identity":"de0dd838-fa08-4a82-a864-2d0b4e030be9","added_by":"auto","created_at":"2026-03-17 12:40:21","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":52564332,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/15e6550f-ec5b-4230-a3f5-8960a7161370.pdf"},{"id":104785506,"identity":"2ae386dd-1e1d-4c6f-90da-40ee2ba21b23","added_by":"auto","created_at":"2026-03-17 08:12:03","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":765442,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8995736/v1/5f77572a5b7beb358b4eeb04.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A mitochondrial quality regulation gene signature for prognosis and tumor microenvironment characterization in breast cancer: an integrative analysis with experimental validation","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eBreast cancer (BC) remains a substantial health concern worldwide, with its complex molecular mechanisms and heterogeneous biological behavior posing challenges to the development of optimal therapeutic and preventive strategies[1-3]. Comprehensive gene expression profiling has led to BC classification into five distinct subtypes: human epidermal growth factor receptor 2 (HER2)-enriched, luminal A/B, basal-like, and the later-identified claudin-low variant. Clinical decision-making and prognostic assessment are primarily guided by immunohistochemical evaluation of HER2 status, together with hormone receptor [estrogen and progesterone receptors] expression patterns and detailed disease staging. At present, traditional treatment modalities for BC include surgery, radiotherapy, chemotherapy, targeted therapy, and endocrine therapy, among others[4, 5]. Advanced-stage disease is uniformly associated with poor clinical outcomes, necessitating continuous therapeutic intervention, with triple-negative BC (TNBC) demonstrating a dismal prognosis in metastatic settings, exhibiting median survival durations below 24 months[6]. These difficulties underscore the pressing need for identifying and validating novel biomarkers and therapeutic targets to enhance diagnostic precision, prognostic accuracy, and treatment efficacy in BC management.\u003c/p\u003e\n\u003cp\u003eMitochondria, double-membrane-bound eukaryotic organelles, are the primary site of oxidative phosphorylation, the fundamental process for intracellular adenosine triphosphate (ATP) biosynthesis[7]. In addition to energy production, mitochondria are integral to lipid biosynthesis, calcium buffering, and iron-sulfur cluster assembly. Their pivotal involvement in innate immunity and the precise execution of cell death pathways further underscores their indispensable role in maintaining tissue-level homeostasis.To preserve mitochondrial and cellular integrity, eukaryotic cells employ a sophisticated and dynamic mitochondrial quality control (MQC) system, jointly regulated by the nuclear and mitochondrial genomes[8]. MQC mechanisms, including mitochondrial dynamics (fusion and fission), mitophagy (selective autophagic clearance of damaged mitochondria), and biogenesis, collectively govern mitochondrial quality and quantity[9]. In cancer, metabolic reprogramming enables tumor cells to divert metabolites toward biosynthetic pathways, aiming at supporting rapid proliferation and accumulating macromolecular precursors necessary for tumor growth[10]. MQC has been found to be crucial in various diseases[11] and in dysregulation in the pathogenesis of lung cancer[12], colon cancer[13], BC[14], melanoma[15], ovarian carcinoma[16], prostate adenocarcinoma[17], and pancreatic cancer[18]. \u003c/p\u003e\n\u003cp\u003eGiven the limited research on the link between MQRGs and BC, elucidating MQC dysfunction is paramount for therapeutic innovation. In this study, we leveraged the MitoCarta 3.0 mammalian mitochondrial proteome to identify 20 core MQRGs involved in mitochondrial dynamics, mitophagy, and biogenesis (ranging from \u003cem\u003ePPARGC1A\u003c/em\u003e to \u003cem\u003eMAP1LC3C\u003c/em\u003e; detailed in Table S1). These markers provide a robust framework for deciphering the molecular drivers of mitochondrial dysregulation in BC [19].\u003c/p\u003e\n\u003cp\u003eThis study aimed to stratify BC patients into molecular subtypes relying upon MQRG expression profiles and explore the interaction between these subtypes and clinical attributes, tumor microenvironment (TME), immune status, and drug sensitivity. Through analysis, seven key biomarkers (\u003cem\u003eSLC45A1\u003c/em\u003e, \u003cem\u003eHPN\u003c/em\u003e, \u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eGLYATL2\u003c/em\u003e, \u003cem\u003eKRT14\u003c/em\u003e, and \u003cem\u003eIGLV6-57\u003c/em\u003e) associated with BC prognosis were identified, from which a BC prognostic model was developed and further validated for its accuracy in predicting BC prognosis. These findings may provide new insights for personalized therapies and improve patient outcomes.\u003c/p\u003e"},{"header":"2. Method","content":"\u003ch2\u003e2.1 Dataset curation and preprocessing\u003c/h2\u003e\n\u003cp\u003eThe comprehensive research workflow and analytical pipeline of this study are delineated in Figure 1. RNA-seq data were acquired from 1,097 BC specimens in the TCGA(https://portal.gdc.cancer.gov) and 327 BC specimens from the GEO (https://www.ncbi.nlm.nih.gov/geo/), encompassing somatic mutation profiles and associated clinical data: Vital status, age, gender, tumor grade, and pathological stage. Additionally, we validated model-related gene expression patterns using the GSE20685 database acquired from the GEO database. Additional datasets were retrieved by accessing the UNCAN (https://uncan.eu/) and HPA (https://www.proteinatlas.org/). Prior to analysis, all datasets underwent rigorous preprocessing, including format conversion and standardization. Based on a systematic literature review, 20 MQRGs were identified for inclusion. As all data were derived from publicly available repositories, no informed consent or ethics committee approval was required.\u003c/p\u003e\n\u003ch2\u003e2.2 Consensus clustering analysis of MQRGs\u003c/h2\u003e\n\u003cp\u003eConsensus clustering analysis was carried out via the R package \"ConsensusClusterPlus\". Cluster assignment revealed enhanced intra-subtype correlations, as demonstrated by the cumulative distribution function (CDF) curve displaying a relatively flat slope relative to the steeper slopes characteristic of robust within-group homogeneity. Conversely, inter-subtype correlations were significantly weaker, confirming distinct molecular profiles between clusters. Based on prognostic MQRG expression patterns, we stratified tumor specimens into distinct MQRG subtypes and utilized principal component analysis (PCA) to validate the classification, with the resulting dimensionality reduction effectively capturing inter-subgroup variations. Finally, Kaplan-Meier (KM) survival analysis was performed using the \"survival\" package, aiming at comparing clinical outcomes across MQRG-based subtypes.\u003c/p\u003e\n\u003ch2\u003e2.3 Identification of DEGs and functional enrichment analysis\u003c/h2\u003e\n\u003cp\u003eApplying an adj-p-value of 0.05 alongside a fold-change of 2.0, 125 DEGs were identified across diverse MQRG subtypes using the \"limma\". Both KEGG[20] and GO[21] enrichment analyses were applied to elucidate key molecular pathways and classify DEGs based on molecular function (MF), cellular component (CC), and biological process (BP) categories, respectively. The \"clusterProfiler\" was used to conduct functional enrichment investigations, while gene Set Variation Analysis (GSVA) was implemented to quantify pathway-level enrichment scores, thereby identifying distinct signaling perturbations across the cohorts.l[22].\u003c/p\u003e\n\u003ch2\u003e2.4 Construction of the MQRG RS model\u003c/h2\u003e\n\u003cp\u003eThe MQRG scoring system was formulated to determine tumor-specific molecular patterns of metabolic-quiescence-associated genes. Initially, univariate Cox regression analysis was carried out on the transcriptomic dataset to recognize BC overall survival (OS) significantly related to differentially expressed genes (DEGs). Subsequently, we deployed consensus clustering analysis to delineate two distinct molecular subtypes (MQRG-Clusters A/B) based on prognostic MQRG expression profiles. A 1:1 randomization protocol was implemented to allocate BC patients into training and testing cohorts with balanced baseline characteristics. The MQRG-based scoring system was subsequently developed via the training dataset. To optimize model generalizability, Lasso-Cox regression analysis was applied on MQRG candidate genes via the \"glmnet\", with penalty parameters optimized through 10-fold cross-validation to avoid overfitting. Finally, multivariate Cox proportional hazards modeling was employed to assess MQRG's independent prognostic significance in the training cohort. The training cohort was stratified into low- (LRG) and high-risk (HRG) groups using the median RS as the cutoff (LRG: scores ≤ median; HRG: scores \u0026gt; median), then KM analysis was performed. The same RS cutoff was utilized for the testing cohort, and model performance was further evaluated using KM and time-dependent ROC curves.\u003c/p\u003e\n\u003ch2\u003e2.5 Nomogram establishment and validation\u003c/h2\u003e\n\u003cp\u003eA predictive nomogram was generated using the \"rms,\" incorporating clinical features and RSs. Each variable contributed to a score, and the sum of these scores generated a total score for each patient, followed by calculating AUC values for 1-, 3-, and 5-year OS, and plotting ROC curves to assess nomogram performance. Furthermore, we evaluated the nomogram performance by comparing observed and predicted survival probabilities at the pre-determined time points by generating a calibration plot.\u003c/p\u003e\n\u003ch2\u003e2.6 Analysis of prognostic genes\u003c/h2\u003e\n\u003cp\u003eThe expression profiles and prognostic utility of the seven signature genes were ascertained leveraging the Xiantao Academic platform(https://www.xiantao.love/). We utilized the TCGA-BRCA dataset (XENA/TCGA_GTEx-BRCA), where RNA-seq data were harmonized through the TOIL pipeline and represented in TPM units. To maintain cohort integrity, our study population was refined by filtering out normal controls and cases with missing survival or pathological records.The association between gene expression and patient survival was assessed using Kaplan-Meier (K-M) survival analysis.\u003c/p\u003e\n\u003ch2\u003e2.7 Cell culture with qRT-PCR analysis\u003c/h2\u003e\n\u003cp\u003eThe frozen batch of the human normal mammary epithelium MCF-10A, human BC cells HCC1937, and MDA-MB-453 (Procell, Wuhan, China) was defrosted by being immersed in a water bath at 37 ℃. The MCF-10A cells were cultured in DMEM/F12 medium that contained 5% horse serum, 20 ng/mL epidermal growth factor, 0.5 μg/mL cortisol, 10 μg/mL recombinant human insulin, 1% non-essential amino acid solution, and 1% penicillin-streptomycin (P/S). Meanwhile, the HCC1937 and MDA-MB-453 cells were cultured on 10 cm plates using RPMI-1640 medium that contained 10% FBS and 1% P/S. Subsequently, the specimens were incubated at 37 °C and 5% CO\u003csub\u003e2\u003c/sub\u003e. After isolating ribosomal RNA with TRIzol reagent (Invitrogen, Carlsbad, CA, USA), cDNA was synthesized by mixing total RNA with a Takara PrimeScript RT reagent kit. The RT‑qPCR was performed on a CFX96™ Real-Time PCR (Bio-Rad Laboratories, Inc., USA) using Takara SYBR® Green assays with 2× SYBR Green qPCR Mix (SparkJade), and all samples were analyzed in triplicate. Gene expression levels were quantified by the 2\u003csup\u003e-ΔΔC t\u003c/sup\u003e approach, using GAPDH for normalization. Table S2 lists the genes (\u003cem\u003eSLC45A1\u003c/em\u003e, \u003cem\u003eHPN\u003c/em\u003e, \u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eGLYATL2\u003c/em\u003e, \u003cem\u003eKRT14\u003c/em\u003e, and \u003cem\u003eIGLV6-57\u003c/em\u003e) primer sequences used.\u003c/p\u003e\n\u003ch2\u003e2.8 Evaluation of immune cells Infiltration\u003c/h2\u003e\n\u003cp\u003eThis molecular classification therapeutic relevance was systematically validated through Spearman correlation analyses between MQRG subtypes and clinicopathological parameters, including age, sex, TNM staging, and survival outcomes. The ssGSEA was deployed to estimate the relative infiltration levels of 23 distinct immune cell types in each BC specimen. Immune and stromal infiltration indices were estimated via the ESTIMATE tool, while the CIBERSORT algorithm was deployed to quantify the landscape of tumor-infiltrating ICs. These analyses allowed for a meticulous evaluation of the cellular heterogeneity and immunological characteristics across the heterogeneous specimens of both risk strata.\u003c/p\u003e\n\u003ch2\u003e2.9 Multiplex immunohistochemistry staining\u003c/h2\u003e\n\u003cp\u003eThe tumor tissue microarrays(Brcsur2205) was purchased from Hunan AiFang Biological Technology Co., Ltd. and contained 120 cases of breast cancer tissue samples with clinical pathological characteristics and survival information.Clinical information is provided in Table S3.Tissue sections were deparaffinized in xylene and rehydrated through a descending ethanol gradient. Multiplex immunohistochemistry staining (mIHC) was executed using a seven-color mIHC kit (AFIHC027, AiFang Biologcal, China), and the antigens were exposed to microwave radiation per protocol. To block endogenous peroxidase activity, sections were incubated in 3% hydrogen peroxide at room temperature for 15 min, followed by a 15-min incubation with 10% goat serum for nonspecific binding blockade. Primary antibody A was introduced and incubated at 4 °C overnight, followed by three washes with PBST. A polymer-HRP-conjugated universal secondary antibody (anti-mouse/rabbit IgG; AFIHC001, AiFang Biological) was then added and incubated at room temperature for half an hour. After PBST washing, a TYR-based fluorescent dye was applied for 8 min, and washed thrice with PBST. Antibody A was subsequently stripped via microwave treatment, and the sections were washed three times with PBST. The staining cycle was repeated iteratively for each subsequent primary antibody B, with blocking using goat serum before each new primary antibody application. The panel of antigens visualized included: CK-Pan (AF20164, 1:200), CD206 (AFRM0009, 1:100), iNOS (AFRP0001, 1:200), Vimentin (AF20105, 1:1500), Ki67 (AF20068, 1:200); all from AiFang Biological), and CXCL9 (YT6008, 1:200, Immunoway). Following completion of all staining cycles, nuclei were counterstained with DAPI for 10 min at room temperature in darkness, followed by three PBST washes. Slides were mounted with an anti-fade fluorescence mounting medium and imaged via an eight-channel fluorescence digital slide scanner (model AF-KL-20-8, AiFang Biological, China).\u003c/p\u003e\n\u003cp\u003eDuring the image acquisition stage, the eight-channel fluorescence digital slide scanner was used to capture the mIHC slide, saving the acquired images in the kfbf format. In the image data analysis stage, custom algorithms were developed in the VISIOPHARM software (VISIOPHARM, Hoersholm, Denmark) to conduct data analysis on the images. To ensure result consistency, fixed and uniform threshold parameters were set through the VISIOPHARM application to identify and quantitatively analyze the association between \u003cem\u003eCXCL9\u003c/em\u003e and immune cells (ICs), particularly macrophages. After the analysis was completed, the results were exported, and SPSS was used for the statistical analysis.The entire core was categorized into tumor and stromal areas based on the expression of PanCK and vimentin.\u003c/p\u003e\n\u003ch2\u003e2.10 Assessment of the TMB, CSC, and mutation\u003c/h2\u003e\n\u003cp\u003eUsing the \"maftools\"[23], we generated a mutation annotation format from TCGA data to systematically identify somatic mutations among BC patients in both groups. Additionally, potential associations between CSC, TMB, and the identified HRG/LRG classifications were investigated.\u003c/p\u003e\n\u003ch2\u003e2.11 Drug susceptibility analysis\u003c/h2\u003e\n\u003cp\u003eThe \"pRRophetic\"[24] software was utilized to determine the half-maximal inhibitory concentrations (IC50) of conventional chemotherapeutic agents utilized in BC to evaluate potential differences in therapeutic susceptibility between the two groups.\u003c/p\u003e\n\u003ch2\u003e2.12 Statistical analysis\u003c/h2\u003e\n\u003cp\u003eData were generated via the Perl programming language (v5.32.1) while performing data processing, analysis, and visualization via R (v4.3.3). Statistical significance across all analyses was categorized as follows: ns (\u003cem\u003ep \u003c/em\u003e≥ 0.05), *(\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.05), **(\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01), ***(\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.001), and****(\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.0001).\u003c/p\u003e"},{"header":"3. Results","content":"\u003cp\u003e3.1 Genomic landscapes, expression profiles, and chromosomal mapping of\u003cstrong\u003e MQRGs\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo begin, 20 MQRG expression levels were detected in healthy samples and BC samples from the TCGA (Figure 2A), revealing a significant overexpression of three MQRGs in BC samples: \u003cem\u003eESRRA\u003c/em\u003e, \u003cem\u003eFIS1\u003c/em\u003e, and \u003cem\u003eSQSTM1.\u003c/em\u003e Conversely, ten MQRGs were significantly downregulated: \u003cem\u003ePPARGC1A\u003c/em\u003e, \u003cem\u003ePPARA\u003c/em\u003e, \u003cem\u003ePPARG\u003c/em\u003e, \u003cem\u003eNFE2L2\u003c/em\u003e, \u003cem\u003eMFN2,\u003c/em\u003e \u003cem\u003eMIEF1\u003c/em\u003e, \u003cem\u003ePINK1\u003c/em\u003e, \u003cem\u003ePARK2\u003c/em\u003e,\u003cem\u003e MAP1LC3B\u003c/em\u003e, and \u003cem\u003eMAP1LC3C\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003eSubsequently, CNV mutation frequency was investigated. The majority of these mutations manifested as copy number amplifications. In contrast, genes such as \u003cem\u003ePRKN\u003c/em\u003e, \u003cem\u003ePPARA\u003c/em\u003e, \u003cem\u003eMAP1LC3B\u003c/em\u003e, \u003cem\u003eFIS1\u003c/em\u003e, \u003cem\u003eMFF\u003c/em\u003e, \u003cem\u003ePINK1\u003c/em\u003e, and \u003cem\u003eMIEF2 \u003c/em\u003eexhibited scattered CNV deletions (Figure 2B). Additionally, the chromosomal locations of CNV alterations for these 20 MQRGs were delineated, as displayed in Figure 2C. The interactions among BC MQRGs and their prognostic implications were systematically investigated using a network (Figure 2D). This analysis identified all MQRGs as risk factors, with the exception of \u003cem\u003ePPARGC1A\u003c/em\u003e, \u003cem\u003ePPARG\u003c/em\u003e, \u003cem\u003eNRF1\u003c/em\u003e, \u003cem\u003eMFN2\u003c/em\u003e, \u003cem\u003eFIS1\u003c/em\u003e, \u003cem\u003eMIEF2\u003c/em\u003e, \u003cem\u003eMAP1LC3A\u003c/em\u003e, and \u003cem\u003eMAP1LC3C\u003c/em\u003e, which were regarded as favorable factors. Assessment of gene connectivity exposed a tightly-knit network of MQRGs within the breast cancer microenvironment (Figure 2E). Most gene-to-gene associations exhibited strong positive co-expression (red lines), with especially prominent links among factors controlling mitochondrial turnover and dynamics.Regarding somatic mutation incidence in these 20 MQRGs, among 991 BC samples, 55 samples (5.55%) harbored mutations in MQRGs (Figure 2F). A total of six genes, namely \u003cem\u003ePPARGC1A\u003c/em\u003e, \u003cem\u003eOPA1\u003c/em\u003e, \u003cem\u003ePPARG\u003c/em\u003e, \u003cem\u003ePINK1\u003c/em\u003e,\u003cem\u003e NFE2L2\u003c/em\u003e, and \u003cem\u003eMFN1\u003c/em\u003e, were identified with mutations. \u003c/p\u003e\n\u003cp\u003eFurthermore, 16 MQRGs were significantly related to OS. Specifically, BC patients with overexpressed \u003cem\u003eMAP1LC3A\u003c/em\u003e,\u003cem\u003e MAP1LC3C\u003c/em\u003e, \u003cem\u003eFIS1\u003c/em\u003e, \u003cem\u003eMIEF2,\u003c/em\u003e \u003cem\u003eMFN2\u003c/em\u003e, and \u003cem\u003ePPARG \u003c/em\u003eexhibited more favorable OS outcomes (Figure S1). Based on these findings, a remarkable divergence was observed between healthy and BC samples in MQRG expression and genetic landscapes.\u003c/p\u003e\n\u003ch2\u003e3.2 Identification of MQRG-related BC subtypes\u003c/h2\u003e\n\u003cp\u003eUnsupervised consensus clustering of MQRG expression profiles identified two robust molecular subtypes (k = 2), designated MQRG Cluster A (n = 794) and MQRG Cluster B (n = 626) (Figures 3A-D). As anticipated, PCA demonstrated significant variations in transcription patterns between the two MQRG subtypes (Figure 3E). Finally, KM curves revealed that OS was better in patients in Cluster A than Cluster B (\u003cem\u003ep\u003c/em\u003e=0.006, Figure 3F).\u003c/p\u003e\n\u003ch2\u003e3.3 TME features in distinct MQRG molecular subtypes\u003c/h2\u003e\n\u003cp\u003eThe clinicopathologic attributes of distinct BC subtypes were meticulously compared using the TCGA and GSE20685 databases, unveiling significant differences in MQRG expression patterns and clinicopathologic characteristics (Figure 4A). Comparative analysis revealed differential expression patterns of MQRGs between prognostic clusters. For instance, cluster B demonstrated significant upregulation of mitochondrial fusion regulators (\u003cem\u003eTFAM\u003c/em\u003e, \u003cem\u003eMFN1\u003c/em\u003e, and \u003cem\u003eOPA1\u003c/em\u003e) and oxidative stress response genes (\u003cem\u003eNFE2L2,\u003c/em\u003e \u003cem\u003eMIEF1\u003c/em\u003e) compared to Cluster A. Conversely, autophagy-related genes (\u003cem\u003eMAP1LC3A\u003c/em\u003e, \u003cem\u003eFIS1\u003c/em\u003e, \u003cem\u003eMIEF2\u003c/em\u003e) and mitochondrial biogenesis markers (\u003cem\u003eESRRA\u003c/em\u003e, \u003cem\u003eSQSTM1\u003c/em\u003e) were markedly downregulated in the poor-prognosis Cluster B. These diametrically opposing expression profiles suggest that impaired MQC mechanisms, characterized by dysfunctional fusion-fission homeostasis and compromised autophagic clearance, are mechanistically linked to unfavorable clinical outcomes in BC patients.\u003c/p\u003e\n\u003cp\u003eMitochondrial fission is mainly governed by Drp1, while fusion is mediated by MFN1/2 and \u003cem\u003eOPA1\u003c/em\u003e[25]. The Drp1 is overexpressed in human invasive BC and lymph node metastases. Functional experiments revealed that Drp1 knockdown or \u003cem\u003eMFN1\u003c/em\u003e overexpression induced mitochondrial elongation or perinuclear clustering, respectively, and both significantly attenuated the metastatic potential of BC cells. Conversely, silencing MFN1/2 triggered mitochondrial fragmentation and enhanced BC cell metastasis[26]. Furthermore, \u003cem\u003eOPA1\u003c/em\u003e-mediated mitochondrial fusion in endothelial cells has been reported to promote tumor angiogenesis, thereby enhancing tumor growth and metastasis[27]. Notably, targeted inhibition of \u003cem\u003eOPA1\u003c/em\u003e suppresses TNBC progression, highlighting \u003cem\u003eOPA1\u003c/em\u003e as a promising therapeutic target in this aggressive BC subtype[28].\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eTFAM\u003c/em\u003e and PGC-1\u0026alpha; serve as central regulatory factors in mitochondrial biogenesis[29]. Notably, \u003cem\u003eTFAM\u003c/em\u003e is significantly overexpressed in BC, which likely contributes to tumor growth by enhancing mitochondrial DNA replication and transcription, thereby potentially supplying the energy and metabolic resources essential for the rapid cancer cell proliferation[30]. \u003cem\u003eMAP1LC3A\u003c/em\u003e, a pivotal gene in autophagy, exerts a dual and complex role in BC[31]. In the early stages of BC, autophagy maintains genomic stability by removing damaged organelles and proteins, thereby effectively inhibiting tumorigenesis. However, during the progression of BC, autophagy paradoxically switches its function to provide energy and metabolic sustenance for tumor cells, enabling them to survive under hypoxic, nutrient-deprived, or chemotherapeutic stress conditions, consequently promoting tumor growth and metastasis[32]. \u003cem\u003eMAP1LC3A\u003c/em\u003e overexpression in BC tissues is significantly related to tumor invasiveness, metastasis, and a poor prognosis[33]. Further in-depth investigations into the molecular mechanisms underlying MQC and its multifaceted function in tumorigenesis are warranted to facilitate the development of effective anti-cancer therapies[34].\u003c/p\u003e\n\u003cp\u003eTo systematically characterize biological pathway heterogeneity between MQRG-associated subtypes, GSVA was performed. GO analysis demonstrated that cluster A was significantly upregulated in deacetylase activity, whereas cluster B exhibited pronounced activation across multiple functional categories, including cyclin-dependent serine/threonine kinase activator activity, tetrahydrofolate metabolic processes, histone kinase activity, and negative regulation of translational initiation (Figure 4B). Histone deacetylases have become promising therapeutic targets in metastatic BC owing to their ability to modulate aberrant acetylation patterns associated with disease progression[35]. Further pathway enrichment analysis revealed distinct metabolic reprogramming patterns. Specifically, cluster A was predominantly enriched in taurine/hypotaurine metabolism and ribosome biogenesis pathways, while cluster B displayed significant enrichment in folate-mediated one-carbon metabolism, cell cycle progression, DNA mismatch repair, and DNA replication pathways (Figure 4C).\u003c/p\u003e\n\u003cp\u003eICs contribute to tumor biology regulation, wherein suitable mitochondrial function is indispensable for IC phenotype, proliferation, and differentiation[36]. Numerous studies have established that the characteristics of IC infiltration signifi{Bianchini, 2022 #78}cantly differ across distinct molecular subtypes of BC[37]. Most TNBCs are generally highly immunogenic and have significantly higher stromal tumor-infiltrating lymphocytes, which is related to a favorable prognosis[6, 38]. Nevertheless, owing to the abundance of immunosuppressive factors within its TME, the anti-tumorigenic functions of these ICs are substantially impaired. In HER2-overexpressing BC, IC infiltration is also prominent and closely related to the efficacy of anti-HER2 targeted therapy[39, 40]. Consequently, the CIBERSORT algorithm showcased significant disparities in IC composition between the two molecular subtypes in BC (Figure 4D). Specifically, MQRG Cluster B displayed markedly higher infiltration of 14 IC types than Cluster A: activated B cells, activated CD4\u003csup\u003e+\u003c/sup\u003e/CD8\u003csup\u003e+\u003c/sup\u003eT cells, activated dendritic cells, gamma delta T cells, immature B cells, myeloid-derived suppressor cells, immature dendritic cells, macrophages, regulatory T cells, natural killers (NK), T follicular helper cells, and types 1/2 T helper cells. Given that functional cytotoxic CD8⁺ T cells are core modulators of anti-tumor immunity and are strongly related to improved patient survival[41, 42], the elevated immune infiltration in Cluster B suggests a more immunoreactive TME.\u003c/p\u003e\n\u003ch2\u003e3.4 Gene subtypes based on DEGs of MQRG subtypes\u003c/h2\u003e\n\u003cp\u003eHere, 125 DEGs associated with MQRG subtypes were identified using the \u0026quot;limma\u0026quot; (Figure 5A). Subsequently, functional enrichment analyses were carried out. Genes associated with MQRG subtypes were significantly enriched in BPs: cell division and the regulation of humoral immune response, suggesting their potential involvement in regulating key PBs such as disease progression and immune evasion (Figure 5B). At the same time, KEGG pathway analysis revealed enrichment in the cell cycle and estrogen pathways (Figure 5C). To further validate these findings, consensus clustering was applied, and the results demonstrated that categorizing patients into two genomic subtypes based on prognostic DEG yielded the most optimal grouping outcomes (Figures S2A-F).\u003c/p\u003e\n\u003cp\u003eRegarding clinical characteristics, gene cluster A showed overexpression of most prognostic DEGs compared to gene cluster B (Figure 5D). Kaplan-Meier survival analysis revealed that patients in gene cluster A exhibited a notably superior OS compared to those in cluster B(Figure 5E). There were significant differences in MQRG expression profiles between the two clusters, with most MQRGs being differentially expressed (Figure 5F).\u003c/p\u003e\n\u003ch2\u003e3.5 Identification of prognostic DEGs and construction of predictive models\u003c/h2\u003e\n\u003cp\u003eBased on the MQRG subtype-related DEGs, a predictive RS model was developed. Initially, the \u0026quot;caret\u0026quot; was used to allocate patients in a random manner into training (n=712) and testing cohorts (n=711) at a 1:1 ratio. From an initial set of 125 MQRG subtype-related prognostic DEGs, LASSO Cox regression, guided by the minimum partial likelihood deviance criterion, identified 15 OS-related genes (Figures 6A-B). Subsequently, the 15 OS-related genes were evaluated using multivariate Cox regression analysis, thereby identifying seven genes: \u003cem\u003eSLC45A1\u003c/em\u003e, \u003cem\u003eHPN\u003c/em\u003e, \u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eGLYATL2\u003c/em\u003e, \u003cem\u003eKRT14\u003c/em\u003e, and \u003cem\u003eIGLV6-57\u003c/em\u003e, and displaying the corresponding gene Coefficients (Figure 6C). RSs were calculated using the following formula derived from multivariate Cox regression analysis: \u003cimg width=\"175\" height=\"38\" src=\"data:image/png;base64,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\" alt=\"image\"\u003e\u003c/p\u003e\n\u003cp\u003eComparative analysis revealed that both gene and MQRG clusters B exhibited significantly higher RSs than their respective Cluster A counterparts (Figures 6D,E). \u003c/p\u003e\n\u003ch2\u003e3.6 Development and validation of predictive risk scores\u003c/h2\u003e\n\u003cp\u003eIn the training cohort, individual RSs were calculated using the previously defined seven-gene signature\u003cspan dir=\"RTL\"\u003e \u003c/span\u003e(Figures 7A, D). The optimal cutoff for risk stratification, determined as the median RS using the \u0026quot;survminer\u0026quot;, was used to classify patients into HRG and LRG. RSs showed an inverse relation with survival duration and a positive relation with mortality. Figure S3A depicts the distribution of BC specimens according to different classification methods. Furthermore, differential expression analysis of MQRGs across different risk groups manifested that 14 genes displayed differential expression among 20 MQRGs, with most genes being overexpressed in the LRG (Figure S3B). Figure 7G shows a heatmap of seven prognostic genes between both risk groups. KM survival analysis showed that the LRG had substantially higher survival rates than the HRG (\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001; Figure 7J). \u003c/p\u003e\n\u003cp\u003eTo validate the prognosis-related gene scores (PRG_scores) prognostic performance, PRG_scores were calculated across the testing and entire sets. Following the formula applied in the training set, patients were allocated into LRG or HRG. Figures 7B, E, H, and 7C, F, I depict the PRG_score distribution, patient survival status, and the seven prognostic gene expression patterns across the LRG and HRG, respectively. Survival analysis showcased that LRG exhibited significantly better prognosis than the HRG (\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.001; Figures 7K,L). \u003c/p\u003e\n\u003ch2\u003e3.7 Development of a nomogram for survival prediction\u003c/h2\u003e\n\u003cp\u003eAs illustrated by the time-dependent ROC analysis, the model yielded AUCs of 0.736, 0.755, and 0.684 for 1, 3, and 5 -year survival prediction in the training group (Figure 8A). Notably, these prognostic accuracies remained consistent across the testing and comprehensive cohorts (Figures 8,C), substantiating the PRG_score as a potent and stable indicator for the survival outcomes of individuals with BC.Then, a prognostic nomogram was developed by integrating PRG scores with key clinical attributes to enable accurate individualized prediction of BC patient outcomes (Figure 8D). The calibration curve closely aligned with the ideal 45\u0026deg; diagonal line, indicating high agreement between predicted and observed survival probabilities. Furthermore, the nomogram exhibited strong predictive discrimination, as evidenced by a robust Concordance Index (C-index) (Figure 8E). \u003c/p\u003e\n\u003ch2\u003e3.8 qRT-PCR validation and Kaplan\u0026minus;Meier analysis of hub genes included in prognostic model\u003c/h2\u003e\n\u003cp\u003eIn BC tissues, \u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eIGLV6-57\u003c/em\u003e, and \u003cem\u003eHPN\u003c/em\u003e mRNA levels were significantly overexpressed relative to those in adjacent normal breast tissues; conversely, \u003cem\u003eSLC45A1\u003c/em\u003e, \u003cem\u003eGLYATL2\u003c/em\u003e, and \u003cem\u003eKRT14\u003c/em\u003e expressions were markedly suppressed (Figures 9A-G). Kaplan-Meier (K-M) survival analysis demonstrated that high expression of \u003cem\u003eSLC45A1\u003c/em\u003e, \u003cem\u003eHPN\u003c/em\u003e, \u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eKRT14\u003c/em\u003e, and \u003cem\u003eIGLV6-57\u003c/em\u003e correlated with improved OS, whereas \u003cem\u003eGLYATL2\u003c/em\u003e overexpression was related to poorer OS (all \u003cem\u003ep \u003c/em\u003e\u0026lt; 0.05) (Figures S4A-G).\u003c/p\u003e\n\u003cp\u003eThe qRT-PCR analysis of the seven prognostic genes in the BC cell lines HCC1937 and MDA-MB-453, unlike MCF-10A, revealed a consistent dysregulation pattern. Specifically, \u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eHPN\u003c/em\u003e, and \u003cem\u003eIGLV6-57\u003c/em\u003e were significantly upregulated in both BC cell lines relative to MCF-10A, whereas \u003cem\u003eSLC45A1\u003c/em\u003e, \u003cem\u003eGLYATL2\u003c/em\u003e, and \u003cem\u003eKRT14\u003c/em\u003e were markedly downregulated (Figures 9H-N). Notably, these in vitro expression patterns closely mirrored the mRNA expression profiles of the corresponding genes in clinical BC tissues relative to normal breast tissues.\u003c/p\u003e\n\u003ch2\u003e3.9 TME across risk groups\u003c/h2\u003e\n\u003cp\u003eThe CIBERSORT was deployed to investigate the interaction between RSs and IC abundance. Scatter plot analysis revealed that RSs were positively associated with the abundance of memory B cells, activated dendritic cells, M0/M2 macrophages, activated mast cells, and resting NKs (Figure 10A-F), and negatively related to the abundance of M1 macrophages,CD4\u003csup\u003e+\u003c/sup\u003e resting dendritic cells, naive B cells, monocytes, plasma cells, regulatory T cells, resting mast cells, CD8\u003csup\u003e+\u003c/sup\u003e T cells, follicular helper T cells, and resting memory T cells (Figures 10G-P). Furthermore, the relation between the seven prognostic genes in the established model and IC abundance was examined. Notably, significant correlations were observed between most ICs and these seven genes (Figure 10Q). Figure 10R illustrates a robust association between high stromal/immune scores and low RSs.\u003c/p\u003e\n\u003ch2\u003e3.10 Association of \u003cem\u003eCXCL9\u003c/em\u003e mRNA with M1/M2 macrophage infiltration\u003c/h2\u003e\n\u003cp\u003e\u003cem\u003eCXCL9\u003c/em\u003e has been confirmed to be expressed in some CD68+ tumor-associated macrophages(TAMs) within the stroma of BC[43], but the association between \u003cem\u003eCXCL9\u003c/em\u003e expression levels and M1/M2 macrophage phenotypic balance remains unvalidated.\u003c/p\u003e\n\u003cp\u003eTo investigate the clinical relevance of \u003cem\u003eCXCL9\u003c/em\u003e within BC microenvironment, we performed mIHC staining on a cohort of BC tissues and their corresponding adjacent normal tissues. The representative mIHC images indicated that \u003cem\u003eCXCL9\u003c/em\u003e, in addition to the macrophage markers iNOS (M1) and CD206 (M2), was markedly elevated in tumor tissues (Figure 11A), while adjacent normal tissues showed negligible expression (Figure 11B). These findings indicate that CXCL9 expression escalates significantly alongside the advancement of BC, being intimately linked to the progressive infiltration of tumor-associated macrophages (TAMs).\u003c/p\u003e\n\u003cp\u003eMoreover, \u003cem\u003eCXCL9\u003c/em\u003e exhibited a positive correlation with the infiltration of both M1 and M2 macrophage subtypes. We further quantified the relationship between \u003cem\u003eCXCL9\u003c/em\u003e expression and macrophage polarization using integrated optical density measurements and cell counts. Our analysis revealed a significant positive correlation between \u003cem\u003eCXCL9\u003c/em\u003e levels and the infiltration of iNOS+ (M1-type) macrophages (\u003cem\u003eR\u003c/em\u003e = 0.3203, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.0001) (Figure 11C), as well as with CD206+ (M2-type) macrophages (\u003cem\u003eR\u003c/em\u003e =0.052, \u003cem\u003ep\u003c/em\u003e=0.5581)(Figure 11D). To enhance our understanding of the spatial architecture, we generated heatmaps for CXCL9+ cells. The analysis revealed a non-uniform distribution of \u003cem\u003eCXCL9\u003c/em\u003e, with distinct regional enrichments (Figure 11E).Using a tissue segmentation algorithm, we classified the tumor microenvironment into tumor parenchyma (yellow) and stromal compartments (red) (Figure 11F). Our findings indicate that CXCL9+ cells are strategically localized within the stroma, suggesting that their spatial arrangement may influence the recruitment and positioning of immune cells.Figure 11G illustrates the results of triple co-staining with DAPI, CXCL9, CK-pan, and Vimentin, demonstrating partial colocalization of CXCL9 with both CK-pan and Vimentin. The CXCL9+CK-pan+ double-positive cells were predominantly localized in the tumor epithelial region, whereas CXCL9+Vimentin+ double-positive cells were primarily found in the tumor stromal area or at the epithelial-stromal interface. Observations at high magnification (100 \u0026mu;m) revealed that CXCL9 was expressed in a punctate or small cluster-like pattern, with colocalized regions alongside CK-pan and Vimentin mostly surrounding tumor cell nests. Additionally, CXCL9 may also be expressed in stromal cells, such as fibroblasts, thereby indirectly contributing to the remodeling of the tumor microenvironment and being potentially linked to the Epithelial-Mesenchymal Transition (EMT) process in breast cancer cells.\u003c/p\u003e\n\u003ch2\u003e3.11 Connection of RSs with TMB and CSC index\u003c/h2\u003e\n\u003cp\u003eHigh TMB patients are more likely to gain immunotherapy benefit, which can be attributed to the increased neoantigen load generated by elevated TMB[44]. Analyzing the mutation profiles from the TCGA-BC cohort revealed that the HRG exhibited significantly higher TMB than the LRG (Figure 12A), signaling that HRG patients may derive substantial benefits from immunotherapeutic interventions. The Spearman correlation analysis manifested a positive connection between TMB and RSs (Figure 12B).\u003c/p\u003e\n\u003cp\u003eFurthermore, cancer stem cell (CSC) index values were integrated with RSs to explore their potential correlation in BC. A linear relationship was noted between the CSC index and RSs (Figure 12C). Quantitative analysis showcased a significant positive correlation (R = 0.29, \u003cem\u003ep\u003c/em\u003e \u0026lt; 2.2e-16) between these parameters, implying that BC cells with heightened MQRG scores display enhanced stem-cell-like properties and minimized cellular differentiation potential.\u003c/p\u003e\n\u003cp\u003eThereafter, somatic mutation distribution between the two groups classified according to RSs within the TCGA-BC cohort was analyzed. The analysis demonstrated a significantly higher mutation frequency in the HRG in contrast to the LRG (Figures 12D-E). The top ten mutated genes were \u003cem\u003eTP53\u003c/em\u003e,\u003cem\u003ePIK3CA\u003c/em\u003e, \u003cem\u003eTTN\u003c/em\u003e, \u003cem\u003eMUC16\u003c/em\u003e, \u003cem\u003eKMT2C\u003c/em\u003e, \u003cem\u003eGATA3\u003c/em\u003e, \u003cem\u003eMAP3K1\u003c/em\u003e, \u003cem\u003eUSH2A\u003c/em\u003e, \u003cem\u003eCDH1\u003c/em\u003e, and \u003cem\u003eFLG \u003c/em\u003ein the\u003cem\u003e HRG\u003c/em\u003e, and\u003cem\u003e PIK3CA\u003c/em\u003e, \u003cem\u003eTP53\u003c/em\u003e, \u003cem\u003eCDH1\u003c/em\u003e, \u003cem\u003eTTN\u003c/em\u003e, \u003cem\u003eGATA3\u003c/em\u003e, \u003cem\u003eMUC16\u003c/em\u003e, \u003cem\u003eMAP3K1\u003c/em\u003e, \u003cem\u003eKMT2C\u003c/em\u003e, \u003cem\u003eHMCN1\u003c/em\u003e, and \u003cem\u003eSYNE1 \u003c/em\u003ein the LRG\u003cem\u003e.\u003c/em\u003e More importantly, patients in the HRG demonstrated remarkably higher \u003cem\u003eTP53\u003c/em\u003e, \u003cem\u003eTTN\u003c/em\u003e, and\u003cem\u003e MUC16\u003c/em\u003e mutation frequencies than those in the\u003cem\u003e \u003c/em\u003eLRG. However, \u003cem\u003ePIK3CA\u003c/em\u003e, \u003cem\u003eCDH1\u003c/em\u003e, and \u003cem\u003eGATA3\u003c/em\u003e mutation frequencies were higher in the LRG.\u003c/p\u003e\n\u003ch2\u003e3.12 Drug sensitivity analysis in different RSs\u003c/h2\u003e\n\u003cp\u003ePatient response to pharmacological treatment is often mirrored in their drug sensitivity. Thus, a panel of drugs typically administered for BC treatment was selected to evaluate the sensitivity levels of patients in both risk groups to these agents. Notably, patients with low RSs had lower IC50 values for all-trans retinoic acid, axitinib, bexarotene, bleomycin, bosutinib, cytarabine, dimethyloxalylglycine, gemcitabine, imatinib, lenalidomide, methotrexate, nilotinib, Nutlin, obatoclax mesylate, and rapamycin. However, patients with high RSs displayed significantly decreased IC50 values for therapeutic drugs such as A.443654, bicalutamide, CGP.082996, cisplatin, CMK, docetaxel, parthenolide, thapsigargin, RO.3306, and VX.680 (Figure S5).\u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eOn a global scale, BC is still a major public health challenge. Its sophisticated pathogenesis and heterogeneous clinical manifestations pose substantial obstacles to the development of effective therapeutic and preventive strategies[45].\u003c/p\u003e\n\u003cp\u003eTumors are characterized by the ungoverned proliferation and unrestricted abnormal cell expansion, causing neoplastic masses characterized by disrupted tissue architecture, cellular pleomorphism, increased mitotic activity, and invasive capacity. Mitochondrial quality imbalance is pivotal in cancer progression[46], as dysregulated mitochondrial energy metabolism constitutes a hallmark of cancer[47]. Moreover, mitochondria critically regulate biosynthetic pathways, signal transduction, apoptosis, cellular differentiation, and cell cycle and growth control, processes that are intricately intertwined with tumorigenesis and malignant progression.\u003c/p\u003e\n\u003cp\u003eMitochondria serve as the primary site of ATP production, the central energy currency essential for cell survival and fundamental cellular processes, earning them the designation as the cell's \"powerhouse.\" However, their implication in oxidative phosphorylation renders them particularly vulnerable to damage, as it generates high ROS levels as a byproduct. ROS can impair protein folding and structure and induce mutations in mitochondrial DNA[48]. Compounded by constant exposure to diverse environmental stressors, this susceptibility heightens the risk of mitochondrial dysfunction. To counteract these threats, eukaryotic cells have evolved a sophisticated MQC system that continuously monitors and preserves mitochondrial network integrity and functionality[49]. Core MQC mechanisms comprise mitophagy, mitochondrial fission/fusion dynamics, mitochondrial biogenesis, and proteostasis-mediated quality control of the mitochondrial proteome[7, 50-52]. Tumor cells undergo metabolic reprogramming driven by mutations, resulting in altered metabolic flux through conventional pathways used by normal cells, with increases or decreases relative to their premalignant tissue of origin[53]. In the BC microenvironment, the unlimited proliferative ability of cancer cells drives mitochondrial metabolic reprogramming, with MQC dysfunction serving as a key driver. Mutations or epigenetic silencing of core MQC genes (e.g., \u003cem\u003ePINK1\u003c/em\u003e, \u003cem\u003eParkin\u003c/em\u003e) impair mitophagy, accumulating damaged mitochondria and excessive ROS release, which in turn induce DNA damage and genomic instability, thereby accelerating BC progression[54]. Notably, MQC abnormalities exhibit marked molecular heterogeneity across BC subtypes, with a paucity of MQC-related gene-based prognostic models. Deciphering subtype-specific MQC regulatory networks can lead to the identification of metabolic targets for personalized therapy and novel prognostic markers, representing a critical future direction.\u003c/p\u003e\n\u003cp\u003eHerein, we systematically investigated the role of MQRGs in BC, unraveling their critical value in molecular subtyping, prognostic prediction, and therapeutic response through multi-omics analysis. MQC dysregulation has been reported to be pivotal in tumor progression. Relying upon MQRG expression profiles, we categorized BC patients into MQRG-Clusters A/B molecular subtypes, showing significant differences in clinicopathological features, IC infiltration of TME, and prognosis. The differential TME characteristics associated with distinct MQRG subtypes suggest that tailored therapeutic strategies can be designed to target specific TME components. Of note, cluster B patients exhibited upregulated mitochondrial fusion-related genes (e.g., \u003cem\u003eTFAM\u003c/em\u003e, \u003cem\u003eMFN1\u003c/em\u003e) and suppressed autophagy-related genes (e.g., \u003cem\u003eMAP1LC3A\u003c/em\u003e, \u003cem\u003eFIS1\u003c/em\u003e), suggesting that disrupted mitochondrial dynamic balance drives BC malignancy[7]. This finding aligns with prior studies reporting that aberrant mitochondrial fission enhances cancer cell invasiveness by remodeling metabolic pathways, while autophagic dysfunction promotes damaged mitochondria accumulation and genomic instability[9]. For instance, Drp1-mediated mitochondrial fission promotes epithelial-mesenchymal transition in BC cells, whilst \u003cem\u003eMAP1LC3A\u003c/em\u003e deficiency accelerates tumor growth by activating the mTOR pathway[26].\u003c/p\u003e\n\u003cp\u003eThe development of an MQRG-based prognostic scoring model, incorporating seven key genes (\u003cem\u003eSLC45A1\u003c/em\u003e, \u003cem\u003eHPN\u003c/em\u003e, \u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eGLYATL2\u003c/em\u003e, \u003cem\u003eKRT14\u003c/em\u003e, and \u003cem\u003eIGLV6-57\u003c/em\u003e), along with a nomogram, offers a valuable tool for clinical decision-making. Emerging data suggest that \u003cem\u003eCXCL9\u003c/em\u003e expression by TAMs governs CXCR3-expressing stem-like CD8\u003csup\u003e+\u003c/sup\u003e T cell recruitment and positioning that mediate clinical responses to anti-PD(L)-1 therapy[55]. \u003cem\u003eCXCL9\u003c/em\u003e can form a functional synergistic network with the receptor CXCR3 and contributes to tumor immune regulation. Especially in malignant tumors such as BC, it profoundly influences tumor progression and prognosis by governing IC recruitment and differentiation besides TME remodeling[56]. Unlike healthy tissues, \u003cem\u003eKRT14\u003c/em\u003e is downregulated in BC tissues, which is related to poor prognosis[57], possibly due to elevated \u003cem\u003eKRT14\u003c/em\u003e methylation levels in BC. \u003cem\u003eKRT14\u003c/em\u003e is implicated in tumor cells' invasive and migratory capabilities and is associated with the TME; thus, it may be a reliable prognostic biomarker[58]. Furthermore, earlier studies have concluded that \u003cem\u003eKRT14\u003c/em\u003e may be a candidate metastasis regulator in TNBC, with its upregulated expression promoting the peritoneal metastasis of TNBC[59].\u003c/p\u003e\n\u003cp\u003eNotably, the RS effectively discriminated between HRG and LRG, with a clear separation in OS outcomes. The nomogram, with its relatively high predictive accuracy for 1-, 3-, and 5-year survival rates, offers a more comprehensive and accurate prognostic tool. This can aid in stratifying patients according to their risk of disease recurrence and mortality, enabling the implementation of more appropriate therapeutic strategies. High-risk patients could potentially benefit from more aggressive treatment regimens, including intensified chemotherapy or early-stage immunotherapy, whereas low-risk patients might avoid unnecessary treatment-related toxicities.\u003c/p\u003e\n\u003cp\u003eNotably, the HRG showed significantly higher TMB than the LRG, with a positive relation between TMB and RSs, aligning with findings of KEYNOTE-522 that demonstrated that high-TMB TNBC patients derive greater benefit from immunotherapy, suggesting that HRG patients may be optimal candidates for immune checkpoint inhibitor (ICI) treatment[60]. The ICI pembrolizumab integrated with chemotherapy had FDA approval for PD-L1-positive metastatic and early-stage TNBC[61].\u003c/p\u003e\n\u003cp\u003eImmune infiltration analysis revealed that HRG patients had increased infiltration levels of memory B cells and M0 macrophages but a reduced level of M1 macrophages and resting CD8\u003csup\u003e+\u003c/sup\u003e T cells, indicating an imbalanced immune microenvironment wherein tumor cells may induce immunogenicity via high TMB while upregulating immune checkpoint molecules[6]. \u003c/p\u003e\n\u003cp\u003eEmerging evidence has confirmed that the chemokines CXCL9/10 are crucial for vigorous responses to ICIs (anti-PD-1 and anti-CTLA-4) and, particularly, that CXCL9/10-secreting macrophages are vital for their therapeutic efficiency[62]. The spatiotemporal orchestration of the tumor microenvironment (TME) by chemokines is a fundamental determinant of BC progression and therapeutic response. In the present study, we characterized the expression landscape of \u003cem\u003eCXCL9\u003c/em\u003e and its complex relationship with macrophage polarization, revealing that \u003cem\u003eCXCL9\u003c/em\u003e is not merely a bystander in tumor progression but acts as a potential immunological rheostat that modulates the recruitment and phenotypic balance of tumor-associated macrophages (TAMs). While previous research identified CXCL9 expression in stromal CD68+ TAMs in BC, the functional link between CXCL9 levels and M1/M2 phenotypic equilibrium was not well understood. Our quantitative multiplex immunohistochemistry (mIHC) analysis provides robust evidence for the phenotypic preference of CXCL9, demonstrating a significant positive correlation with iNOS+ M1-type macrophages (\u003cem\u003eR\u003c/em\u003e = 0.3203, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.0001) , while its correlation with CD206+ M2-type macrophages was minimal(\u003cem\u003eR\u003c/em\u003e =0.052, \u003cem\u003ep\u003c/em\u003e=0.5581). This differential association suggests that CXCL9 serves as a selective chemoattractant or paracrine inducer for pro-inflammatory, anti-tumorigenic M1 macrophages rather than a general driver of myeloid infiltration. Given the essential role of M1 macrophages in Th1-mediated anti-tumor immunity[63], the CXCL9-M1 axis may be a key mechanism for converting \"cold\" tumors into \"hot,\" immune-active environments[64].\u003c/p\u003e\n\u003cp\u003eAdditionally, our findings reveal the dual cellular origin of CXCL9 within the BC microenvironment. Its partial colocalization with CK-pan and Vimentin indicates production by both malignant epithelial cells and mesenchymal stromal components. The presence of CXCL9+CK-pan+ cells within tumor nests, along with CXCL9+Vimentin+ clusters at the epithelial-stromal interface, points to a collaborative secretory network. High-magnification imaging shows \u003cem\u003eCXCL9\u003c/em\u003e forming clusters around tumor nests, implying the presence of \"chemotactic hubs\" that attract CXCR3-expressing immune cells from the stroma into the tumor. Moreover, the correlation between CXCL9 and Vimentin+ cells associates CXCL9 with the EMT process, where Vimentin marks mesenchymal traits often observed in cancer-associated fibroblasts (CAFs). The localization of CXCL9 at the tumor's invasive front may reflect a compensatory immune response to counteract immunosuppressive signals linked to EMT-driven stromal changes.Clinical Implications for Prognostic StratificationFrom a clinical perspective, the association between high CXCL9 expression and prolonged overall survival (OS) can be attributed to the \"normalization\" of the immune landscape. By fostering an environment rich in M1-type macrophages, CXCL9 promotes a \"hot\" tumor phenotype characterized by enhanced antigen presentation and T-cell activation. This molecular milieu not only naturally suppresses tumor growth but also potentially heightens the sensitivity to immune checkpoint inhibitors (ICIs). Consequently, the Mitochondrial Quality Regulation Gene Signature, with CXCL9 as a core component, offers a dual-purpose tool: it stratifies patients by risk while simultaneously identifying those who may possess a more responsive, immune-active microenvironment.In conclusion, our study positions CXCL9 as a pivotal orchestrator of the immune landscape in breast cancer, promoting the recruitment of M1 macrophages.\u003c/p\u003e\n\u003cp\u003eThe positive correlation between CSC index and RS (R = 0.29,\u003cem\u003ep\u003c/em\u003e \u0026lt; 2.2e-16) suggests that BC with high MQRG scores exhibit enhanced stemness, potentially mediating resistance to conventional chemotherapy. It is worthwhile emphasizing that patients in the HRG had significantly increased \u003cem\u003eTP53\u003c/em\u003e, \u003cem\u003eTTN,\u003c/em\u003e and \u003cem\u003eMUC16\u003c/em\u003e mutation frequencies more than those in the LRG. Among these, \u003cem\u003eTP53\u003c/em\u003e, mutated in approximately 30% of BC cases, is the most frequently altered gene in this malignancy[65]. \u003cem\u003eTP53\u003c/em\u003e encodes the \u003cem\u003ep53\u003c/em\u003e tumor suppressor, a transcription factor that activates genes governing cell cycle arrest, apoptosis, DNA repair, metabolic reprogramming, and senescence in response to genotoxic stressors such as radiation and chemotherapy[66]. Accumulating evidence underscores \u003cem\u003ep53\u003c/em\u003e as a therapeutically relevant target in BC, particularly in TNBC and HER2-positive subtypes characterized by a high \u003cem\u003eTP53\u003c/em\u003e mutation burden[67].\u003c/p\u003e\n\u003cp\u003eDrug sensitivity analysis further revealed distinct responses to cisplatin and docetaxel between the HRG and LRG, providing a basis for personalized chemotherapy. Indeed, the observed differential drug sensitivity between both risk groups holds significant implications for personalized chemotherapy. The disparities in IC50 values between the risk groups provide a theoretical reference for tailoring chemotherapy regimens. These findings can be utilized to select drugs that are more likely to be effective for individual patients, minimizing the use of ineffective drugs and reducing the risk of associated toxicities. This personalized approach to chemotherapy can enhance treatment efficacy and patient quality of life.\u003c/p\u003e\n\u003cp\u003eNotwithstanding, some limitations of this study cannot be overlooked. To begin, the molecular mechanisms behind MQRG's effects on BC development remain incompletely understood. While this study established associations between MQRGs and various aspects of BC, the precise signaling pathways and regulatory networks through which these genes exert their effects necessitate further investigation. For example, the role of MQRGs in mitochondrial dynamics, bioenergetics, and their crosstalk with other cellular processes, such as DNA damage response and epigenetic regulation, needs to be elucidated. Elucidating these mechanisms is paramount for developing more targeted and effective therapies. Another limitation is the reliance on data from public databases. Although these databases provide extensive data resources, the lack of validation in independent clinical cohorts may limit generalizability. Variations in patient populations, data collection methods, and treatment protocols across different studies can introduce biases. Future research should validate these findings in large-scale, well-characterized clinical cohorts to confirm the robustness of the identified associations and prognostic models.\u003c/p\u003e\n\u003cp\u003eFurthermore, the scope of drug sensitivity analysis was limited to a subset of chemotherapeutic agents. With the rapid advancement of novel immunotherapies and targeted therapies in oncology, investigating the sensitivity of different MQRG subtypes to these new treatment modalities is crucial. Future studies should include more drugs, including emerging targeted agents and immunotherapeutic drugs, to provide more comprehensive insights to inform personalized therapeutic strategies.\u003c/p\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eThis study establishes a robust 7-gene Mitochondrial Quality Regulation Gene Signature as an independent prognostic indicator for breast cancer. Through the integration of multi-dimensional omics data and experimental assays, we elucidated the role of \u003cem\u003eCXCL9\u003c/em\u003e as a spatial orchestrator that promotes pro-inflammatory M1 macrophage recruitment and polarization. This signature provides a comprehensive framework for risk stratification and the prediction of therapeutic responses, including chemotherapy and immunotherapy. Our findings suggest that targeting mitochondrial-immune crosstalk represents a promising strategy for remodeling the tumor microenvironment and improving patient outcomes in breast cancer.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eBC\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eCAFs\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eBreast Cancer\u003c/p\u003e\n \u003cp\u003ecancer-associated fibroblasts\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eCDF\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCumulative Distribution Function\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eCNV\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCopy Number Variant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eCSC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCancer Stem Cell\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eDEGs\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eEMT\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eDifferentially Expressed Genes\u003c/p\u003e\n \u003cp\u003eEpithelial-Mesenchymal Transition\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eGO\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eGene Ontology\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eGSVA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eGene Set Variation Analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eHRG\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eHigh-Risk Group\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eIC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eImmune Cell\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eKEGG\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eKyoto Encyclopedia Of Genes And Genomes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eKM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eKaplan-Meier\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eLRG\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003emIHC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eLow-Risk Group\u003c/p\u003e\n \u003cp\u003emultiplex immunohistochemistry\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMQRGs\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMitochondrial Quality-Related Genes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eOS\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eOverall Survival\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003ePCA\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eqRT-PCR\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003ePrincipal Components Analysi\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;Quantitative real-time PCR\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eROC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eReceiver Operating Characteristic\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eRS\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eRisk Score\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTCGA\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eTAMs\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eThe Cancer Genome Atlas\u003c/p\u003e\n \u003cp\u003etumor-associated macrophages\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTMB\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTumor Mutation Burden\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTME\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTumor Microenvironment\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003cp\u003eAcknowledgements\u003c/p\u003e\n\u003cp\u003eWe extend our gratitude to GEO and TCGA database, and all contributors who generously shared their data on these platforms.This study would like to extend its gratitude to Hunan AiFang\u0026nbsp;Biologcal Co., Ltd. for providing relevant antibodies, multiplex fluorescence staining kits, as well as staining, scanning, and data analysis services.\u003c/p\u003e\n\u003cp\u003eAuthor contributions\u003c/p\u003e\n\u003cp\u003eHuaiwen Pu: Conceptualization, Formal analysis, Visualization, Writing \u0026ndash; original draft. Tingjing Li: Formal analysis, Visualization, Writing \u0026ndash; original draft. Renji Liang: Visualization, Writing \u0026ndash; original draft. Zhongxiang Fan: Software, Formal analysis, Writing \u0026ndash; original draft. Bowen Tang: Data curation, Formal analysis, Visualization, Writing \u0026ndash; original draft. Yongmei Luo: Data curation, Validation, Writing \u0026ndash; review \u0026amp; editing. Xinyu Yi: Data curation, Validation, Writing \u0026ndash; review \u0026amp; editing. Liming Xie: Project administration, Validation，Writing \u0026ndash; review \u0026amp; editing. Yuehua Li: Conceptualization, Methodology, Supervision, Writing \u0026ndash; review \u0026amp; editing.\u003c/p\u003e\n\u003cp\u003eFunding\u003c/p\u003e\n\u003cp\u003eThis research was funded by the Interdisciplinary Research Program in Medicine and Engineering, the First Affiliated Hospital of University of South China (No.IRP-M\u0026amp;E-2025-05) and Clinical Medical Research 4310 Program of the University of South China (20224310NHYCG07).\u003c/p\u003e\n\u003cp\u003eData availability\u003c/p\u003e\n\u003cp\u003eAll data generated or analysed during this study are included in this published article and its supplementary information files. The source data utilized for the development of the prognostic model are accessible via public databases. Further technical inquiries should be addressed to the corresponding authors.\u003c/p\u003e\n\u003cp\u003eDeclarations\u003c/p\u003e\n\u003cp\u003eEthics approval and consent to participate\u003c/p\u003e\n\u003cp\u003eThe use of human tumor tissue microarray samples in this study was approved by the Ethics Committee on Biological Science and Technology of Hunan Aifang Biological Co., Ltd. (Approval No. HN20250401). All procedures were performed in accordance with the Declaration of Helsinki, and written informed consent was obtained from all participants or their legal guardians.\u003c/p\u003e\n\u003cp\u003eConsent for publication\u003c/p\u003e\n\u003cp\u003eAll the authors give the consent for the publication of identifiable details, which can include the text, figures and other materials in this manuscript.\u003c/p\u003e\n\u003cp\u003eCompeting interests\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eSung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin. 2021;71(3):209-49.\u003c/li\u003e\n \u003cli\u003eGiaquinto AN, Sung H, Newman LA, Freedman RA, Smith RA, Star J, et al. Breast cancer statistics 2024. CA Cancer J Clin. 2024;74(6):477-95.\u003c/li\u003e\n \u003cli\u003eLoibl S, Poortmans P, Morrow M, Denkert C, Curigliano G. Breast cancer. Lancet. 2021;397(10286):1750-69.\u003c/li\u003e\n \u003cli\u003eWaks AG, Winer EP. Breast Cancer Treatment: A Review. Jama. 2019;321(3):288-300.\u003c/li\u003e\n \u003cli\u003eXiong X, Zheng LW, Ding Y, Chen YF, Cai YW, Wang LP, et al. Breast cancer: pathogenesis and treatments. Signal Transduct Target Ther. 2025;10(1):49.\u003c/li\u003e\n \u003cli\u003eBianchini G, De Angelis C, Licata L, Gianni L. Treatment landscape of triple-negative breast cancer - expanded options, evolving needs. Nat Rev Clin Oncol. 2022;19(2):91-113.\u003c/li\u003e\n \u003cli\u003eSong J, Herrmann JM, Becker T. Quality control of the mitochondrial proteome. Nat Rev Mol Cell Biol. 2021;22(1):54-70.\u003c/li\u003e\n \u003cli\u003eChoong CJ, Okuno T, Ikenaka K, Baba K, Hayakawa H, Koike M, et al. Alternative mitochondrial quality control mediated by extracellular release. Autophagy. 2021;17(10):2962-74.\u003c/li\u003e\n \u003cli\u003eLuo Y, Ma J, Lu W. The Significance of Mitochondrial Dysfunction in Cancer. Int J Mol Sci. 2020;21(16).\u003c/li\u003e\n \u003cli\u003eFontana F, Limonta P. The multifaceted roles of mitochondria at the crossroads of cell life and death in cancer. Free Radic Biol Med. 2021;176:203-21.\u003c/li\u003e\n \u003cli\u003eLiu BH, Xu CZ, Liu Y, Lu ZL, Fu TL, Li GR, et al. Mitochondrial quality control in human health and disease. Mil Med Res. 2024;11(1):32.\u003c/li\u003e\n \u003cli\u003eRehman J, Zhang HJ, Toth PT, Zhang Y, Marsboom G, Hong Z, et al. Inhibition of mitochondrial fission prevents cell cycle progression in lung cancer. Faseb j. 2012;26(5):2175-86.\u003c/li\u003e\n \u003cli\u003eXiong X, Hasani S, Young LEA, Rivas DR, Skaggs AT, Martinez R, et al. Activation of Drp1 promotes fatty acids-induced metabolic reprograming to potentiate Wnt signaling in colon cancer. Cell Death Differ. 2022;29(10):1913-27.\u003c/li\u003e\n \u003cli\u003eKannan A, Wells RB, Sivakumar S, Komatsu S, Singh KP, Samten B, et al. Mitochondrial Reprogramming Regulates Breast Cancer Progression. Clin Cancer Res. 2016;22(13):3348-60.\u003c/li\u003e\n \u003cli\u003eSerasinghe MN, Wieder SY, Renault TT, Elkholi R, Asciolla JJ, Yao JL, et al. Mitochondrial division is requisite to RAS-induced transformation and targeted by oncogenic MAPK pathway inhibitors. Mol Cell. 2015;57(3):521-36.\u003c/li\u003e\n \u003cli\u003eGao T, Zhang X, Zhao J, Zhou F, Wang Y, Zhao Z, et al. SIK2 promotes reprogramming of glucose metabolism through PI3K/AKT/HIF-1\u0026alpha; pathway and Drp1-mediated mitochondrial fission in ovarian cancer. Cancer Lett. 2020;469:89-101.\u003c/li\u003e\n \u003cli\u003eLee YG, Nam Y, Shin KJ, Yoon S, Park WS, Joung JY, et al. Androgen-induced expression of DRP1 regulates mitochondrial metabolic reprogramming in prostate cancer. Cancer Lett. 2020;471:72-87.\u003c/li\u003e\n \u003cli\u003eNagdas S, Kashatus JA, Nascimento A, Hussain SS, Trainor RE, Pollock SR, et al. Drp1 Promotes KRas-Driven Metabolic Changes to Drive Pancreatic Tumor Growth. Cell Rep. 2019;28(7):1845-59.e5.\u003c/li\u003e\n \u003cli\u003eRath S, Sharma R, Gupta R, Ast T, Chan C, Durham TJ, et al. MitoCarta3.0: an updated mitochondrial proteome now with sub-organelle localization and pathway annotations. Nucleic Acids Res. 2021;49(D1):D1541-d7.\u003c/li\u003e\n \u003cli\u003eKanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000;28(1):27-30.\u003c/li\u003e\n \u003cli\u003eAshburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25-9.\u003c/li\u003e\n \u003cli\u003eH\u0026auml;nzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7.\u003c/li\u003e\n \u003cli\u003eMayakonda A, Lin DC, Assenov Y, Plass C, Koeffler HP. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res. 2018;28(11):1747-56.\u003c/li\u003e\n \u003cli\u003eYan C, Niu Y, Ma L, Tian L, Ma J. System analysis based on the cuproptosis-related genes identifies LIPT1 as a novel therapy target for liver hepatocellular carcinoma. J Transl Med. 2022;20(1):452.\u003c/li\u003e\n \u003cli\u003eT\u0026aacute;bara LC, Segawa M, Prudent J. Molecular mechanisms of mitochondrial dynamics. Nat Rev Mol Cell Biol. 2025;26(2):123-46.\u003c/li\u003e\n \u003cli\u003eZhao J, Zhang J, Yu M, Xie Y, Huang Y, Wolff DW, et al. Mitochondrial dynamics regulates migration and invasion of breast cancer cells. Oncogene. 2013;32(40):4814-24.\u003c/li\u003e\n \u003cli\u003eHerkenne S, Ek O, Zamberlan M, Pellattiero A, Chergova M, Chivite I, et al. Developmental and Tumor Angiogenesis Requires the Mitochondria-Shaping Protein Opa1. Cell Metab. 2020;31(5):987-1003.e8.\u003c/li\u003e\n \u003cli\u003eZamberlan M, Boeckx A, Muller F, Vinelli F, Ek O, Vianello C, et al. Inhibition of the mitochondrial protein Opa1 curtails breast cancer growth. J Exp Clin Cancer Res. 2022;41(1):95.\u003c/li\u003e\n \u003cli\u003eItoh Y, Khawaja A, Laptev I, Cipullo M, Atanassov I, Sergiev P, et al. Mechanism of mitoribosomal small subunit biogenesis and preinitiation. Nature. 2022;606(7914):603-8.\u003c/li\u003e\n \u003cli\u003eGao W, Wu M, Wang N, Zhang Y, Hua J, Tang G, et al. Increased expression of mitochondrial transcription factor A and nuclear respiratory factor-1 predicts a poor clinical outcome of breast cancer. Oncol Lett. 2018;15(2):1449-58.\u003c/li\u003e\n \u003cli\u003eOthman EQ, Kaur G, Mutee AF, Muhammad TS, Tan ML. Immunohistochemical expression of MAP1LC3A and MAP1LC3B protein in breast carcinoma tissues. J Clin Lab Anal. 2009;23(4):249-58.\u003c/li\u003e\n \u003cli\u003eWu Q, Sharma D. Autophagy and Breast Cancer: Connected in Growth, Progression, and Therapy. Cells. 2023;12(8).\u003c/li\u003e\n \u003cli\u003eSivridis E, Koukourakis MI, Zois CE, Ledaki I, Ferguson DJ, Harris AL, et al. LC3A-positive light microscopy detected patterns of autophagy and prognosis in operable breast carcinomas. Am J Pathol. 2010;176(5):2477-89.\u003c/li\u003e\n \u003cli\u003eZong Y, Li H, Liao P, Chen L, Pan Y, Zheng Y, et al. Mitochondrial dysfunction: mechanisms and advances in therapy. Signal Transduct Target Ther. 2024;9(1):124.\u003c/li\u003e\n \u003cli\u003eGuadagni A, Barone S, Alfano AI, Pelliccia S, Bello I, Panza E, et al. Tackling triple negative breast cancer with HDAC inhibitors: 6 is the isoform! Eur J Med Chem. 2024;279:116884.\u003c/li\u003e\n \u003cli\u003eBai R, Cui J. Mitochondrial immune regulation and anti-tumor immunotherapy strategies targeting mitochondria. Cancer Lett. 2023;564:216223.\u003c/li\u003e\n \u003cli\u003eOnkar SS, Carleton NM, Lucas PC, Bruno TC, Lee AV, Vignali DAA, et al. The Great Immune Escape: Understanding the Divergent Immune Response in Breast Cancer Subtypes. Cancer Discov. 2023;13(1):23-40.\u003c/li\u003e\n \u003cli\u003eKeenan TE, Tolaney SM. Role of Immunotherapy in Triple-Negative Breast Cancer. J Natl Compr Canc Netw. 2020;18(4):479-89.\u003c/li\u003e\n \u003cli\u003eBianchini G, Gianni L. The immune system and response to HER2-targeted treatment in breast cancer. Lancet Oncol. 2014;15(2):e58-68.\u003c/li\u003e\n \u003cli\u003eGriguolo G, Pascual T, Dieci MV, Guarneri V, Prat A. Interaction of host immunity with HER2-targeted treatment and tumor heterogeneity in HER2-positive breast cancer. J Immunother Cancer. 2019;7(1):90.\u003c/li\u003e\n \u003cli\u003eZhong X, Wu H, Ouyang C, Zhang W, Shi Y, Wang YC, et al. Ncoa2 Promotes CD8+ T cell-Mediated Antitumor Immunity by Stimulating T-cell Activation via Upregulation of PGC-1\u0026alpha; Critical for Mitochondrial Function. Cancer Immunol Res. 2023;11(10):1414-31.\u003c/li\u003e\n \u003cli\u003eZhang L, Romero P. Metabolic Control of CD8(+) T Cell Fate Decisions and Antitumor Immunity. Trends Mol Med. 2018;24(1):30-48.\u003c/li\u003e\n \u003cli\u003eDU Shaoqian TM, CAO Yuan, WANG Hongxia, HU Xiaoqu, FAN Guangjian, ZANG Lijuan. . CXCL9 expression in breast cancer and its correlation with the characteristics of tumor immunoinfiltration. Journal of Shanghai Jiao Tong University (Medical Science). 2023;43(7):860-72.\u003c/li\u003e\n \u003cli\u003eChan TA, Yarchoan M, Jaffee E, Swanton C, Quezada SA, Stenzinger A, et al. Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic. Ann Oncol. 2019;30(1):44-56.\u003c/li\u003e\n \u003cli\u003eBray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229-63.\u003c/li\u003e\n \u003cli\u003eZong WX, Rabinowitz JD, White E. Mitochondria and Cancer. Mol Cell. 2016;61(5):667-76.\u003c/li\u003e\n \u003cli\u003eHanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov. 2022;12(1):31-46.\u003c/li\u003e\n \u003cli\u003ePickles S, Vigi\u0026eacute; P, Youle RJ. Mitophagy and Quality Control Mechanisms in Mitochondrial Maintenance. Curr Biol. 2018;28(4):R170-r85.\u003c/li\u003e\n \u003cli\u003eNi HM, Williams JA, Ding WX. Mitochondrial dynamics and mitochondrial quality control. Redox Biol. 2015;4:6-13.\u003c/li\u003e\n \u003cli\u003eAdebayo M, Singh S, Singh AP, Dasgupta S. Mitochondrial fusion and fission: The fine-tune balance for cellular homeostasis. Faseb j. 2021;35(6):e21620.\u003c/li\u003e\n \u003cli\u003eScarpulla RC. Transcriptional paradigms in mammalian mitochondrial biogenesis and function. Physiol Rev. 2008;88(2):611-38.\u003c/li\u003e\n \u003cli\u003eGustafsson \u0026Aring; B, Dorn GW, 2nd. Evolving and Expanding the Roles of Mitophagy as a Homeostatic and Pathogenic Process. Physiol Rev. 2019;99(1):853-92.\u003c/li\u003e\n \u003cli\u003eDeBerardinis RJ, Chandel NS. Fundamentals of cancer metabolism. Sci Adv. 2016;2(5):e1600200.\u003c/li\u003e\n \u003cli\u003eLi Q, Chu Y, Li S, Yu L, Deng H, Liao C, et al. The oncoprotein MUC1 facilitates breast cancer progression by promoting Pink1-dependent mitophagy via ATAD3A destabilization. Cell Death Dis. 2022;13(10):899.\u003c/li\u003e\n \u003cli\u003eMarcovecchio PM, Thomas G, Salek-Ardakani S. CXCL9-expressing tumor-associated macrophages: new players in the fight against cancer. J Immunother Cancer. 2021;9(2).\u003c/li\u003e\n \u003cli\u003ePan M, Wei X, Xiang X, Liu Y, Zhou Q, Yang W. Targeting CXCL9/10/11-CXCR3 axis: an important component of tumor-promoting and antitumor immunity. Clin Transl Oncol. 2023;25(8):2306-20.\u003c/li\u003e\n \u003cli\u003eFridman WH, Pag\u0026egrave;s F, Saut\u0026egrave;s-Fridman C, Galon J. The immune contexture in human tumours: impact on clinical outcome. Nat Rev Cancer. 2012;12(4):298-306.\u003c/li\u003e\n \u003cli\u003eLiao S, Zhang X, Chen L, Zhang J, Lu W, Rao M, et al. KRT14 is a promising prognostic biomarker of breast cancer related to immune infiltration. Mol Immunol. 2025;180:55-73.\u003c/li\u003e\n \u003cli\u003eVerma A, Singh A, Singh MP, Nengroo MA, Saini KK, Satrusal SR, et al. EZH2-H3K27me3 mediated KRT14 upregulation promotes TNBC peritoneal metastasis. Nat Commun. 2022;13(1):7344.\u003c/li\u003e\n \u003cli\u003eSchmid P, Cortes J, Pusztai L, McArthur H, K\u0026uuml;mmel S, Bergh J, et al. Pembrolizumab for Early Triple-Negative Breast Cancer. N Engl J Med. 2020;382(9):810-21.\u003c/li\u003e\n \u003cli\u003eHeater NK, Warrior S, Lu J. Current and future immunotherapy for breast cancer. J Hematol Oncol. 2024;17(1):131.\u003c/li\u003e\n \u003cli\u003eHouse IG, Savas P, Lai J, Chen AXY, Oliver AJ, Teo ZL, et al. Macrophage-Derived CXCL9 and CXCL10 Are Required for Antitumor Immune Responses Following Immune Checkpoint Blockade. Clin Cancer Res. 2020;26(2):487-504.\u003c/li\u003e\n \u003cli\u003eMantovani A, Marchesi F, Malesci A, Laghi L, Allavena P. Tumour-associated macrophages as treatment targets in oncology. Nat Rev Clin Oncol. 2017;14(7):399-416.\u003c/li\u003e\n \u003cli\u003eLiang B, Duan Z, Long S, Zhou P. Infiltration of CXCL9+ macrophages confers a favorable prognosis in breast cancer: Insights from an integrated single-cell RNA and bulk RNA sequencing study. PLoS One. 2025;20(12):e0337175.\u003c/li\u003e\n \u003cli\u003eComprehensive molecular portraits of human breast tumours. Nature. 2012;490(7418):61-70.\u003c/li\u003e\n \u003cli\u003eKastenhuber ER, Lowe SW. Putting p53 in Context. Cell. 2017;170(6):1062-78.\u003c/li\u003e\n \u003cli\u003eMarvalim C, Datta A, Lee SC. Role of p53 in breast cancer progression: An insight into p53 targeted therapy. Theranostics. 2023;13(4):1421-42.\u003c/li\u003e\n\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":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Mitochondrial quality regulation genes, Breast cancer, Tumor microenvironment, prognostic model, Multiplex immunohistochemistry, drug sensitivity","lastPublishedDoi":"10.21203/rs.3.rs-8995736/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8995736/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eBreast cancer (BC) remains a leading cause of female cancer mortality, necessitating the identification of novel biomarkers for precise prognostic stratification. Mitochondrial quality-related genes (MQRGs) are critical players in tumorigenesis; however, their specific role in BC remains insufficiently characterized.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eWe integrated transcriptomic and clinical datasets from TCGA, GTEx, and GEO to identify differentially expressed MQRGs. A 7-gene prognostic signature, comprising \u003cem\u003eSLC45A1\u003c/em\u003e, \u003cem\u003eHPN\u003c/em\u003e, \u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eGLYATL2\u003c/em\u003e, \u003cem\u003eKRT14\u003c/em\u003e, and \u003cem\u003eIGLV6-57\u003c/em\u003e, was developed using LASSO and multivariate Cox regression. Its predictive robustness was rigorously assessed via Kaplan-Meier and time-dependent ROC analyses.We further characterized the tumor microenvironment (TME) and validated our findings through quantitative real-time PCR（qRT-PCR） and multiplex immunohistochemistry (mIHC).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eUnsupervised consensus clustering identified two distinct molecular subtypes related to MQRG (Clusters A and B), which exhibited significantly different clinical outcomes and tumor microenvironment (TME) characteristics. Utilizing subtype-specific differentially expressed genes, we developed a 7-gene prognostic signature (including \u003cem\u003eSLC45A1\u003c/em\u003e, \u003cem\u003eHPN\u003c/em\u003e, \u003cem\u003eCHAD\u003c/em\u003e, \u003cem\u003eCXCL9\u003c/em\u003e, \u003cem\u003eGLYATL2\u003c/em\u003e, \u003cem\u003eKRT14\u003c/em\u003e, and \u003cem\u003eIGLV6-57\u003c/em\u003e) through LASSO-Cox regression. The signature demonstrated robust prognostic reliability across cohorts. High-risk group were distinguished by an immunosuppressive TME architecture, augmented TMB, and divergent therapeutic sensitivities, contrasting sharply with the low-risk group.Experimental validation via qRT-PCR confirmed the dysregulation of these critical genes in breast cancer cells. Notably, mIHC revealed the spatial distribution of \u003cem\u003eCXCL9\u003c/em\u003e, highlighting its predominant expression in tumor-associated macrophages and its significant positive correlation with M1 (iNOS+) rather than M2 (CD206+) polarization, indicating its role in modulating anti-tumor immunity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eThis study establishes a robust Mitochondrial Quality Regulation Gene Signature as an independent prognostic biomarker for BC. By elucidating the spatial dynamics of \u003cem\u003eCXCL9\u003c/em\u003e in promoting M1 macrophage recruitment, our findings provide an integrative framework for risk stratification and personalized therapeutic interventions, particularly for immunotherapy.\u003c/p\u003e","manuscriptTitle":"A mitochondrial quality regulation gene signature for prognosis and tumor microenvironment characterization in breast cancer: an integrative analysis with experimental validation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-11 12:28:45","doi":"10.21203/rs.3.rs-8995736/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-05-07T10:07:44+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-02T03:50:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"189109365550979345779647118812856165335","date":"2026-04-19T03:38:55+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-12T08:04:56+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"50222467576687326232900559473825446201","date":"2026-04-09T21:52:53+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"73320213224546383056970409086608502651","date":"2026-03-19T15:14:19+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-03-05T16:02:51+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-03T06:25:49+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-03T06:24:13+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Cancer","date":"2026-02-28T13:35:34+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"d4ac0552-4b8d-4dd5-8f17-5eb17a4e3ad6","owner":[],"postedDate":"March 11th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Revision requested","date":"2026-05-07T10:07:44+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-02T03:50:03+00:00","index":70,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[],"tags":[],"updatedAt":"2026-05-07T10:23:39+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-11 12:28:45","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8995736","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8995736","identity":"rs-8995736","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}