Circulating tumor DNA and Response Evaluation Criteria In Solid Tumors: ctDNA-RECIST proof-of-concept in HER2-positive metastatic breast cancer

Circulating tumor DNA and Response Evaluation Criteria In Solid Tumors: ctDNA-RECIST proof-of-concept in HER2-positive metastatic breast cancer | 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 Circulating tumor DNA and Response Evaluation Criteria In Solid Tumors: ctDNA-RECIST proof-of-concept in HER2-positive metastatic breast cancer Alessandra Fabi, Elena Giordani, Elena Ricciardi, Grazia Arpino, and 14 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7101376/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 20 Jan, 2026 Read the published version in Journal of Experimental & Clinical Cancer Research → Version 1 posted 9 You are reading this latest preprint version Abstract Background: Response Evaluation Criteria In Solid Tumors (RECIST 1.1) and circulating tumor DNA (ctDNA) recapitulate and anticipate response to treatment, respectively. However, no ctDNA-RECIST (cRECIST) guidelines have been formally implemented in clinical practice so far. Methods: For first proof-of-principle, HER2-positive metastatic breast cancer patients (n=50) were enrolled in the multi-center prospective GIM21 study to receive Trastuzumab-emtansine (T-DM1), and were monitored for Objective Responses, e.g. ORs (progressive disease, stable disease, partial response, complete response; PD/SD/PR/CR) vs ctDNA-ORs (cORs: cPD/cSD/cPR/cCR). Standard OR cut-offs (SD/PD≥20% and SD/PR≤30%) were applied to cORs by default, or tentatively relaxed. Results: Whichever the cut-off, bespoke NGS/dPCR (78 ctDNAs; 466 measurements) and CT scans (113 tumor lesions) revealed much deeper cORs than ORs, leading to RECIST 1.1/cRECIST divergence in 27 cPD-positive patients. Yet, OR/cOR integration remained feasible at default/common cut-offs, as shown by correlation between fast cPD and poor OR/PFS. Although a satisfactory coarse patient classifier, cPD was unfortunately confounded by patient-specific, post-cPD ctDNA increases/decreases (ctDNA waving). Then, to personalize outcome prediction, two-point cRECIST comparisons (response vs baseline/nadir) were replaced by a novel non-cRECIST variable measuring three-point Tr ends ( Tr ). Remarkably, the duration of the first post-cPD Tr drop correlated with the timing of PD in 15/15 evaluable patients (R 2 =0.76). Conclusions: Combined, cRECIST (cPD) and Tr may help to: (a) predict treatment efficacy during early drug development, (b) randomize for timely treatment switch in clinical trials, and (c) prevent premature treatment withdrawal in long-responders undergoing ctDNA waving. Future prospective studies are warranted for cRECIST/RECIST 1.1 integration/personalization in different tumors/settings. Trial registration: NCT05735392. ctDNA RECIST 1.1 ctDNA-RECIST HER2-positive Breast Cancer T-DM1 objective response evaluation. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 BACKGROUND Response Evaluation Criteria In Solid Tumors (RECIST) 1.1 assess clinical objective response (OR) to treatment in metastatic cancer based on dimensional changes of tumor lesions measured by medical imaging. The RECIST 1.1 scale, metrics and conventions guide oncology practice, define measurable endpoints in clinical trials, and ultimately provide an objective framework for the regulatory approval of new drugs [ 1 ]. However, the limitations of a purely anatomical assessment of response to treatment are widely recognized, and additional RECIST criteria have been proposed to evaluate Positron Emission Tomography (PET) and immune Response (PERCIST and iRECIST/imRECIST) [ 2 , 3 ]. Alternative and additional criteria apply to lymphomas [ 4 ] and organ-specific assessment, such as brain metastases [ 5 ]. The potential advantages to integrate circulating tumor DNA (ctDNA) into RECIST 1.1 (and possibly other objective response scales) have been outlined [ 6 – 8 ]. The EORTC RECIST working group ( https://recist.eortc.org ) and the European Liquid Biopsy Society (www.ELBS.eu) are at work to provide guidelines and recommendations for ctDNA-RECIST (cRECIST hitherto). However, publicly available ctDNA datasets are presently much smaller than the medical imaging datasets interrogated during successive RECIST refinements. Moreover, although changes in ctDNA levels may effectively stratify patients into good and poor responders [ 9 – 19 ], no prospective clinical studies have been reported, to our knowledge, to estimate ctDNA-adjusted risk of clinical progression in individual patients. Herein, we report on a multi-center prospective study, designed by the Gruppo Italiano Mammella (GIM) and called GIM21/LiqERBcept (NCT05735392). The primary aim of GIM21 was to assess relationships between ctDNA and medical imaging (e.g. RECIST 1.1 assessments) in HER2-positive, metastatic breast cancer patients receiving the first-in-class Antibody-Drug Conjugate (ADC) Trastuzumab emtansine (T-DM1) as standard of care second-line treatment [ 20 ]. Despite T-DM1 has now been superseded in this indication by an irinotecan-based Trastuzumab conjugate (Trastuzumab deruxtecan; T-DXd) [ 21 ], second-line T-DM1 treatment remains particularly appealing from a methodological viewpoint. In this setting, the expansion of tumor variants carrying known cancer drivers, as monitored by ctDNA, coincides with loss of HER2 amplification/addiction [ 22 ]. This crucial oncogenic remodeling/switch may be instrumental to align objective clinical response, ctDNA trajectories, T-DM1 efficacy, and outcome, e.g. to model a self-consistent RECIST/cRECIST framework. In the GIM21 study, we have investigated for the first time how changes in ctDNA levels (a continuous variable) may be translated into an objective discontinuous cRECIST scale. Similar to the 4 classical RECIST 1.1 Objective Responses (ORs), e g. Progressive Disease, Stable Disease, Partial Response and Complete Response (PD, SD, PR and CR), cRECIST was postulated to feature 4 ctDNA-ORs (cORs), e.g. cPD, cSD, cPR and cCR. cRECIST is proposed herein to be robust and intuitive enough to stratify patients by disease outcome and, with caveats and additions, also to guide personalized clinical decisions in clinical trials and everyday clinical practice. METHODS Patients GIM21 is an open-label, observational (blood drawing implemented as an additional procedure), multicenter, single-arm, prospective study (NCT05735392) conducted between July 2018 (first patient in) and December 2022 (last blood drawing). It recruited consecutive adult patients with a documented diagnosis of HER2-positive metastatic BC, as per ImmunoHistoChemistry score (IHC 3+, or IHC 2 + and HER2 amplification ratio ≥ 2.0). Eligible patients had no more than one line of previous anti-HER2 treatment for advanced disease (Trastuzumab, either alone or combined with Pertuzumab plus taxane). For additional inclusion and exclusion criteria, age, demographics and clinical information see Supplemental Methods and Table S1 . T-DM1 was administered at the standard dose of 3.6 mg/kg Q21 days until disease progression or development of unmanageable toxic effects (see Supplemental Methods). Follow-up was extended during subsequent therapy regimens: last clinical update September 1, 2024. Primary aim and pre-planned analysis The primary aim of GIM21 was to assess ctDNA changes vs tumor size (RECIST 1.1), which implies a primary endpoint of at least one detectable ctDNA species per patient. This precondition was met in 80% of the patients enrolled in a similar previous study [ 22 ], resulting in a sample size calculation of expected ctDNA-positives/enrolled patients of 36/45 (95% CI; 65.4–90.4%). Observed ctDNA positives were 38/43 (88%), meeting primary endpoint. Clinical samples and medical imaging To capture early ctDNA changes, commonly observed during T-DM1 treatment [ 22 ], blood was obtained right before treatment (baseline, T 0 ), and then at 3 progressively longer intervals: 3 weeks (T 1 to T 3 ), 9 weeks (T 4 to T 6 ), and 12 weeks until progression (T 7 to T p ). CT scans were instead obtained as per standard of care: at progression from previous treatment, and then every 12 weeks. This resulted in a ‘staggered’ (w3/w9/w12 vs w12) collection scheme until T 6 , followed by ctDNA/CT scan alignment (Fig. 1 a, compare green and red timelines). Blood was drawn immediately before (minutes) T-DM1 infusion at all time points except T p , that coincided with the last GIM21 visit (end of treatment), on average 7 days after the final CT scan documenting clinical progression. Technical details on blood drawing and pre-analytical processing are provided in supplementals and elsewhere [ 22 ]. NGS and dPCR A targeted NGS panel (Oncomine Pan-Cancer) was used detecting: (a) SNVs and INDELs from 52 major cancer drivers and tumor suppressor genes, (b) DNA copy number variations (CNV) from 12 genes, and (c) rearrangements from 12 genes. Alterations were called by NGS at 3 selected time points (T 0 , T 6 and T p ; Fig. 1 a, blue timeline), except when progression occurred at or before T 6 , which resulted in two NGS testing points only (T 0 and T p ). Alterations captured by NGS were systematically monitored (T 0 to T p ; Fig. 1 a, orange timeline) by bespoke digital PCR (dPCR) assays, either custom-designed or commercially obtained. Each dPCR assay was pre-tested on serial dilutions of tumor tissue DNAs (tDNAs) carrying (n ≥ 2) or lacking (n = 1) the alteration being assessed, and was accepted if its Limit of Detection (LoD) was at 0.01% VAF or better. All reagents, equipment, and online NGS/dPCR bioinformatic/interpretation tools were from ThermoFisher Scientific. RECIST 1.1 and ctDNA-RECIST (cRECIST) As per RECIST 1.1 [ 1 ], % change in tumor size is routinely expressed as the sum of diameters of target lesions relative to the lowest measurement on study. For symmetry, SNV and CNV measurements were made commensurable by expressing them as a single absolute value after rounding to two decimal places (e.g. VAF = 1.101% and CN = 2.229 made equal to 1.10 and 2.23). Absolute values were then summed and expressed as % change (D ctDNA , Fig. 1 b) relative to a reference time point, e.g. T n vs T ref , whereby T n is any time point and T ref is either T 0 (baseline) or T nadir (lowest ctDNA measurement) whichever the lesser. To turn D ctDNA into a discontinuous cRECIST scale, the 4 possible objective RECIST 1.1 responses (ORs) were mirrored into 4 corresponding ctDNA objective responses (cORs), e.g. PD/cPD, SD/cSD, PR/cPR and CR/cCR. The same cut-offs defining the upper and lower limits of SD (≥ 20% and ≤ 30% respectively) were applied by default to cSD (hitherto ‘default cut-offs’; Fig. 1 c), but were also the subject of extensive exploratory analysis (see below). PD/cPD and CR/cCR were similarly scored upon detection of a new tumor lesion/new ctDNA species, and loss of all detectable lesions/target ctDNAs by CT scan and dPCR respectively (for dPCR LoD, 0.01%VAF, see above). Two exceptions to RECIST 1.1/cRECIST mirroring are acknowledged: (i) the distinction between target and non-target lesions vs complete equivalence of all ‘target’ ctDNAs; (ii) the upper limit in the number of target lesions (n = 5) in RECIST 1.1 vs no pre-defined limit in target ctDNA species. For symmetry, default monitoring by CT scan and dPCR stopped at PD (end of treatment) and cPD respectively. Statistics Results were reported by descriptive statistics and 95% CI. Parametric and non-parametric tests were used to evaluate associations between variables. Progression-free survival (PFS) was calculated as the time between the first T-DM1 administration and either the first evidence of progressive disease or the time of last follow-up. Survival was evaluated by the Kaplan-Meier method. Data were elaborated by GraphPad Prism v10 (RRID:SCR_000306) and IBM SPSS statistics (RRID:SCR_002865). RESULTS GIM21 study flowchart Out of 50 patients enrolled (Fig. 2 a), 2 withdrew and 5 were excluded due to either missing blood drawings or poor cfDNA quality. Five additional patients were not evaluable because lacking detectable target ctDNAs (n = 5). The remaining 38 ctDNA-positive patients were sorted into cPD-positives (n = 27) and cPD-negatives (n = 11). cPD positives displayed at least one D ctDNA increase at or above the default ≥ 20% cSD/cPD cut-off (see Methods and Fig. 1 c). Since most D ctDNA increases exceeded the default cut-off, the number of cPD-positives was minimally affected by alternative cPD cut-offs within the ≥ 10% to ≥ 50% range (Fig. S1 ). Therefore, cRECIST allowed for wide confidence intervals in objective cPD assignment. Consistency was noted between patient features and ctDNA status. For instance, only one exceptional responder was still PD-free at last follow-up after 1000 days of treatment despite being cPD-positive. All the remaining 26 cPD positives experienced PD at extra-cranial sites. Conversely, all 5 patients undergoing PD exclusively due to brain metastases were cPD-negative, which is unsurprising given poor ctDNA release from intracranial locations [ 22 , 23 ]. Finally, the only patient discontinuing T-DM1 due to toxicity did so while in complete objective response, as concordantly assessed by her cPD/PD double-negative status. Demographics and the clinical pathological features of cPD-positive and cPD-negative patients are separately presented in Table S1 . No patient was lost to follow up. At last evaluation, 11/27 cPD-positives and 6/11 cPD-negatives were alive. Despite enrichment in cases with brain metastases, PFS and OS did not significantly differ between cPD-negatives and cPD-positives (Fig. 2 b and c). Therefore, detectable cPD did not apparently select for outcome. Circulating genomic alterations Details are provided in Figs. S2-S4 about: bespoke NGS/dPCR testing, oncoprints, distribution of genomic alterations at the different time points, actionable levels, technical NGS/dPCR validation, selection of dPCR as the standard cRECIST readout, and application of VAF as the optimal metric to measure ctDNA. Raw ctDNA measurements are provided in an annotated spreadsheet as Table S2 . Divergent trajectories: tumor lesions vs target ctDNAs For a detailed longitudinal evaluation of individual trajectories, each of 113 tumor lesions and 78 target ctDNAs was assessed as described in supplemental results (Fig. S5). Compared to lesions, ctDNAs underwent more numerous, frequent, discordant, asynchronous and extreme changes, including occasional direct switch from loss to gain at immediately consecutive time points. In summary, stable disease was rare from the ctDNA standpoint, and evolutionary divergence of ctDNA variants largely exceeded dissociated radiological responses among tumor lesions measured by CT scans. Then, it was of interest to determine whether this translates into RECIST 1.1/cRECIST divergence. RECIST 1.1 vs cRECIST To compare RECIST 1.1 and cRECIST, as per GIM21 primary aim, target lesion diameters and D ctDNA values were calculated at the pre-planned time points from the 27 cPD-positive GIM21 patients. Objective clinical and ctDNA responses (OR and cORs) were longitudinally assigned by application to both tumor lesions and ctDNAs of default RECIST 1.1 cut-offs (SD/PD ≥ 20% and SD/PR ≤ 30%, see Fig. 1 ). ORs/cORs were color-matched (SD/cSD, PR/cPR, CR/cCR, PD/cPD), and displayed as timelines until PD and cPD respectively. Despite a much longer OR observation period, assignments changed more frequently for cORs than ORs (Fig. 3 a, see color patchworks), and the total numbers of recorded response changes were similar (n = 52 vs 51; donut charts in Fig. 3 b). Altogether, cSD was under-represented compared to SD (16%vs 44%), whereas cPR and cCR were over-represented compared to PR and CR (Fig. 3 b). In other words, due to divergence between individual tumor lesions and ctDNAs (see Fig. S5), ORs and cORs also diverged. To further investigate divergence, best Objective Responses (best ORs and best cORs) were derived from the same dataset. For consistency, best OR (classically defined as the greatest recorded tumor shrinkage) was mirrored into best cOR (defined as the deepest D ctDNA drop on study). Interestingly, a Pearson’s correlation matrix measuring co-occurrence (PD/cPD, SD/cSD, PR/cPR, and CR/cCR) was significantly (McNemar-Bowker test p = 0.007) skewed off-diagonal, particularly toward a dominant best SD/best cPR-cCR combination phenotype seen in 13/27 patients (Fig. 3 c). Therefore, ctDNA responses were clearly detectable in almost half of the cPD-positive patients achieving disease stability as their best response. Similar patterns (timelines, pie charts and Pearson’s plots) were seen at upper/lower SD/cSD cut-offs, at least within the ≥ 50% to ≤ 60% range (Figs. S6-S9). Cut-off-independent cOR and best cOR assignments demonstrated that trajectory divergence between tumor size and ctDNA does not depend on how they are measured, but on an intrinsically deeper response to T-DM1 treatment of ctDNA compared to tumor lesions. cRECIST and clinical outcome Next, best objective responses (best ORs and best cORs) were correlated with the outcomes (PFS and OS) of the 27 cPD-positive patients. Unsurprisingly, tumor shrinkage/best OR correlated with PFS, and a non-significant trend was also noted for OS (Fig. 4 a and b). In contrast, deepest ctDNA drop/best cOR correlated with neither (Fig. 4 d and e). This poor correlation was puzzling, since many groups have observed that quantitative ctDNA changes do correlate with treatment efficacy (see for instance [ 9 – 19 ]). To clarify whether this discrepancy might arise from cOR being measured on a discontinuous scale, extensive simulations were carried out applying both discontinuous and conventional continuous metrics to the GIM21 dataset. However, quantitative ctDNA changes could not be linked to outcome (see supplemental results and Fig. S10). Then, ctDNA amounts were disregarded, and a time-dependent cRECIST variable was considered termed ctDNA-PFS (cPFS). Analogous to PFS, that measures the time elapsed between T 0 and PD, cPFS was defined as the time elapsed between T 0 and cPD. Interestingly, and in agreement with ctDNA amounts being poorly predictive in the GIM21 setting, cPFS did correlate with tumor shrinkage (best OR; Fig. 4 c) and not ctDNA drop (best cOR, Fig. 4 f). Moreover, since best OR correlated with PFS (see Fig. 4 a), cPFS appeared to correlate, albeit indirectly, with the outcome of broad patient subsets. Then, cPFS was assessed side-by side with two other time-dependent variables: LT and PFS. cPFS and LT define two consecutive treatment periods (T 0 to cPD, and cPD to T p ), whereas PFS measures the two periods altogether (T 0 to PD). Side-by-side plotting showed that cPFS was shortest and least variable (Fig. 4 g). It weakly correlated (by regression analysis) with PFS, and not at all with LT (Fig. 4 h and i). In contrast, LT/PFS correlations were supported by regression plots (Fig. 4 j), and roughly concordant LT/PFS patient ranking (Fig. 4 k). Moreover, Fig. 4 k showed that the gap between cPD and PD is in most cases small when PD occurs before median PFS, and then progressively expands, e.g. ctDNA is of special interest for long-responders. Altogether, the results in Fig. 4 and Fig. S10 revealed that time to cPD (cPFS) correlates with outcome much better than quantitative ctDNA changes. Unfortunately, possibly due to its limited variation ranges, cPFS distinguished the outcomes of broad patient subsets but poorly resolved the outcomes of individual patients, particularly long-responders. When more stringent cut-offs were applied, the reasons for this limitation became evident, since even minimal changes in cPFS measurements completely disrupted cPFS/best OR correlations, minimally if at all affecting LT/PFS correlations (Fig. S11). Then, it was hypothesized that pre-cPD metrics (including cPFS) are suboptimal in the GIM21 setting because T-DM1 response is mostly determined by events occurring late during treatment, particularlywithin the LT/PFS overlap, e.g. post-cPD. Combining cRECIST with non-cRECIST outcome indicators While looking for post-cPD variables specifying individual outcomes, it was inferred that it might be inappropriate to compare series of disconnected paired measurements, each referring to the same reference value (T n vs T ref , T n+1 vs T ref , etc), as per RECIST 1.1/cRECIST conventions. This might miss individual, informative multipoint ctDNA trends in complex, zig-zagging ctDNA trajectories (see for instance Fig. S4b), particularly in the post-cPD period. Then, the simplest possible multipoint series were considered, e.g. consecutive 3-point series with 2-point overlap (triplets). D ctDNA values were measured at each point (T 0 to T p ), and then averaged for each triplet, resulting in a triplet Tr end value dubbed Tr (Fig. 5 a). It was observed that none of two consecutive Tr values was identical to the second decimal digit, resulting in a simplified scoring system with two alternative trends only (regardless of Tr magnitude), e.g. either ctDNA increase ( Tr n > Tr n−1 ) or ctDNA decrease ( Tr n < Tr n−1 ). Disappearance of a previously detected ctDNA, and de novo appearance of a new ctDNA were assimilated to Tr decrease and increase respectively. Finally, swimmer plots were generated displaying ctDNA nadir , cPD, PD, and Tr on the timelines of the 27 cPD-positive patients (Fig. 5 b). Swimmer plots revealed that: (a) ctDNA nadir , cPD and PD invariably occur in this order, e.g. they behave as ‘placeholders’; (b) successive Tr increases and decreases, termed ‘ctDNA waving’, are evident in several timeline sections; (c) increase-only Tr patterns are seen in most (13/15) early progressors (PFS median PFS; 10/12); (d) at least 2 consecutive opposing trends are observed in 4/12 late progressors (pts 51, 34, 31 and 19); (e) PD occurs in most (25/27) patients during phases of Tr increase, and not decrease. Finally, and most interestingly (f), in 15 data-dense patients with at least three consecutive measurable Tr values, the duration of the first post-cPD Tr decrease significantly correlated with PFS and LT, although not with OS (Fig. 5 c). Tr (combined with cPFS) was a much better metric than cPFS alone ( Tr in Fig. 5 c vs cPFS in Fig. 4 h: R 2 = 0.76 vs 0.30). This improvement exclusively applied to Tr drops calculated after cPD, since the first Tr drop (at any time, regardless of cPD) poorly correlated with PFS (Fig. S12). In summary, to calculate outcome, cPFS (cRECIST) and Tr (non-cRECIST) had to be separately assessed and then combined, implying non-redundancy and subordinacy of the two variables. Thus, distinct events/variables underly T-DM1 efficacy in the pre-cPD and post-cPD (LT) periods. DISCUSSION RECIST 1.1 was developed to assist in the routine serial monitoring of treatment efficacy. Conversion of continuous measurements (tumor size) into a simple set of discontinuous OR metrics (PD, SD, PR, CR) is crucial to support multiple binary (go/no-go) decision nodes at patient re-evaluation in advanced cancer. The minimal expectation is that cRECIST should be equally discontinuous, binary, intuitive, and robust. In addition, should cRECIST forecast outcome, particularly for individual patients, it would meet diverse clinical needs with a common, highly personalized tool. The present GIM21 study provides an initial framework to embed cRECIST into RECIST 1.1, and proposes novel metrics to overcome the inherent limitations of RECIST-like approaches when applied to outcome prediction. Inspired by the previous LiqBreasTrack study, conducted in a similar setting [ 22 ], the GIM21 protocol was designed for staggered blood/CT scan collection and greater ctDNA data density at initial time points (w3/w9/w12 schedule; Fig. 1 a). Unlike tumor-informed NGS/PCR approaches widely used to detect minimal residual disease [ 24 ], our bespoke NGS/dPCR assay captured and monitored genomic alterations exclusively in blood, not to miss adaptively acquired target ctDNAs arising de novo in approximately half of T-DM1-treated patients [ 22 ]. Based on Poisson distribution analysis in up to 20.000 nanoliter-size partitions, bespoke dPCR effectively normalized for wild-type alleles (VAF rather than copies/mL), and compensated for pre-analytical variation (sample processing, cfDNA recovery, purity, and concentration), eliminating the need for duplicate testing (Figs. 1 , S2-S4). Robustness and low cost are key features for real-world implementation. For comprehensive overview of all ctDNAs, VAF and CNVs were computed into a single unit measuring overall % ctDNA changes (D ctDNA ; Fig. 1 b). In addition, the RECIST 1.1 scoring system was mirrored as much as possible into cRECIST, each objective tumor response having a ctDNA counterpart: SD/cSD, PD/cPD, CR/cCR and PR/cPR. Most crucial were the standard SD/PD (≥ 20%) and SD/PR (≤ 30%) cut-offs of RECIST 1.1 (Fig. 1 c), that to a first approximation were applied by default throughout the GIM21 study. In the lack of extensive retrospective ctDNA data warehouses, default cut-offs are supported by simplicity, immediacy, and mnemonic ease. Nevertheless, they may be questioned as arbitrary and/or unsupported by objective evidence. Then, cRECIST was validated across wide cut-off ranges (Figs. S1, S6-S9 and S11). Interestingly, due to the considerable magnitude of most observed ctDNA changes, cOR and best cOR assignments remained largely cut-off-independent, and cPD positioning exhibited limited variations at least within the ≥ 20% to ≥ 50% range for cSD/cPD, and the ≥ 30% to ≥ 60% range for cSD/cCR. However, at these expanded cut-offs outcome associations worsened considerably (Fig. S11c). Thus, despite ample confidence intervals, default cut-offs may be an obligate choice for outcome prediction, at least in the GIM21 setting. Altogether, the preliminary analysis conducted herein clearly points to the fine tuning of cSD cut-offs as to a major investigation goal in future cRECIST studies. Whichever the cut-off, whether default or expanded, objective tumor and ctDNA responses strikingly diverged (OR from cOR and best OR from best cOR). This was due to a far more marked and dynamic response to T-DM1 treatment of ctDNA than tumor size, best exemplified by the considerable phenotypic enrichment in cPD-positive GIM21 patients (50% approximately) experiencing SD as their best clinical response while undergoing a measurable (cPR/cCR) ctDNA drop (Fig. 3 ). In turn, OR/cOR divergence led to poor correlation between best cOR (deepest ctDNA drop) and outcome (PFS and OS, Fig. 4 d and e). However, when ctDNA amounts were disregarded and a time-dependent variable was derived (cPFS, the time elapsed between T 0 and cPD), correlations became apparent with best OR (Fig. 4 c) and hence, although indirectly, with PFS (Fig. 4 a). To our knowledge, this is the first demonstration that a discontinuous ctDNA/cRECIST variable can assist in monitoring response to treatment. In this respect, cPFS performance appears to be similar to other conventional continuous variables in many advanced settings and tumors (see for instance [ 9 – 19 ]). This is remarkable, because continuous metrics purely measuring ctDNA (both absolute amounts and relative changes) were inferior to cPFS when applied to the GIM21 dataset (Fig. S10). Whereas the reasons for this specific discrepancy with previous studies remain unclear at present, it is of note that cPFS, although primarily a time-dependent variable, implicitly incorporates cPD-compliant quantitative ctDNA changes. Therefore, a ‘hybrid’ computation of time and ctDNA amounts, as in cPFS, may be central to cRECIST implementation. A key observation of the GIM21 study is that cPFS and conventional continuous ctDNA variables have a common limitation: they are accurate enough to classify broad patient subsets, but remain crude predictors of individual outcomes, for which no metrics have so far been described to the best of our knowledge. While looking for such metrics, it was noted that ctDNA undergoes successive expansion and contraction cycles, termed ctDNA waving, particularly evident post-cPD and in long-term responders. Alone, cRECIST series of disconnected paired values were intrinsically inapplicable to track ctDNA waving, and required integration with ctDNA trends ( Tr ) measured in consecutive overlapping triplets of time points (Fig. 6 ). Empirically, an ongoing Tr increase was associated with an immediate risk of progression. More interestingly, the duration of the first Tr drop, as long as occurring post-cPD (Fig. S12), was found to be informative, since it predicted individual outcomes much better than cPFS alone (R 2 = 0.30 vs 0.76, Fig. 4 h vs Fig. 5 c). The simplest interpretation of ctDNA waving is that T-DM1 retains marginal (and decreasing) efficacy during step-wise acquisition of pharmacological resistance, and Tr reflects average tumor fitness throughout. The model presented in Fig. 6 and described in the legend highlights OR/cOR divergence, integrates cRECIST into RECIST 1.1, and proposes that ctDNA-assisted outcome prediction strategies could take ctDNA waving into account. As to possible mechanisms underlying ctDNA waving, a relevant observation has been recently made in the context of the same GIM21 study described herein. Following adaptive HER2 polypeptide loss under anti-HER2 therapeutic pressure, during T-DM1 treatment HER2 polypeptides were regained in tumor tissues and blood (sHER2) from a subset of patients with a favorable outcome [ 25 ]. Late (e.g. post-cPD) residual responsiveness and waving tumor/drug equilibria may potentially apply to other targeted agents, tumors, settings, and companion analytes. The GIM21 study design and protocols suffer from several limitations. These include a deliberately narrow focus on a single indication, low numerosity, failure to detect intra-cranial progression, and core-study inclusion of only 27 cRECIST-evaluable (cPD-positive) patients, although this selection did not apparently bias for outcome (Fig. 2 b and c). In addition, it may retrospectively be argued that real-time testing would have missed new target ctDNAs appearing de novo at T p . These account for 18.1% of all ctDNAs (Fig. S2 ). Also, a more intensive post-cPD blood drawing could have been very useful to systematically measure the first post-cPD Tr drop, that herein was measurable in 15 patients only. It is acknowledged that expanding cRECIST application beyond the present proof-of-concept will also require technical solutions capturing substantial numbers of target ctDNAs. Despite limitations, GIM21 is to our knowledge the first prospective study exploring the biological and clinical implications of cRECIST, providing initial evidence for irreducible conceptual differences between anatomical and ctDNA disease descriptions, and proposing practical computational approaches to personalize PD prediction. CONCLUSIONS To conclude, three main practical cRECIST implications may be foreseen: (a) in early drug development, cPFS and Tr may represent early-on proxies of PFS and therapeutic efficacy; (b) in ctDNA-guided clinical trials, specific cPFS and/or Tr values may be predefined and, when prospectively applied, they may adaptively randomize the intention-to-treat population for either treatment maintenance or discontinuation/switch; and (c) in clinical practice, cPFS and Tr may flag individual patients for intensified post-cPD CT scan surveillance, and at the same time prevent premature withdrawal of an effective treatment in long-term responders displaying post-cPD ctDNA response/waving. Abbreviations Response Evaluation Criteria In Solid Tumors version 1.1, RECIST 1.1; circulating tumor DNA, ctDNA; ctDNA-RECIST, cRECIST); Trastuzumab-emtansine, T-DM1; Objective Responses, clinical and ctDNA, ORs and cORs; Progressive Disease, PD, Stable Disease, SD; Partial Response, PR; Complete Response, CR; ctDNA-PD, cPD; ctDNA Stable Disease, cSD; ctDNA Partial Response, cPR; ctDNA Complete Response, cCR; Next Generation Sequencing, NGS; digital PCR, dPCR; Positron Emission Tomography (PET) Response Criteria for Solid Tumors, PERCIST; Immune Response Criteria for Solid Tumors, iRECIST/imRECIST; Gruppo Italiano Mammella, GIM; Antibody-Drug Conjugate, ADC; ImmunoHistoChemistry, IHC; Computerized Tomography, CT; Progression-free Survival, PFS; Overall Survival, OS; ctDNA-PFS, cPFS; Lead Time, LT. Declarations Ethics approval and consent to participate The study was approved by the Fondazione Bietti Ethical Review Board (RS-857/16). Written informed consent was obtained from all patients before study entry. Ethics approval and consent to participate Not applicable. Availability of data and materials All raw data (total, and organized by figures) of ctDNA measurements are provided in an annotated spreadsheet as Table S2. Competing interests A.F.: consultant or advisor for Roche, Lilly, Novartis, AstraZeneca, Pfizer, Seagen, Gilead, MSD, Menarini Stemline, Dompè Biotech; speaker honoraria from Astra Zeneca, Roche, Lilly, Novartis, Gilead, Pfizer, Daiichi Sankyo Exact Sciences; grant support (to the Institution) by Astra Zeneca, Roche. GA: personal fees from Novartis, Lilly; grants and personal fees from Roche; grants, personal fees, and nonfinancial support from Pfizer; grants, personal fees, and nonfinancial support from AstraZeneca; and personal fees from Daichi. E.B.: grants or contracts from Astra-Zeneca, Roche; honoraria for lectures from Merck-Sharp & Dome, Astra-Zeneca, Pfizer, Eli-Lilly, Bristol-Myers Squibb, Novartis, Takeda and Roche; member of Data Safety Monitoring Board or Advisory Board of Merck-Sharp & Dome, Pfizer, Novartis, Bristol-Myers Squibb, Astra-Zeneca, Roche, Amgen and Celltrion. F.C.: consultant for Exact Science and AstraZeneca. Funding Work performed with the unconditional support by Roche Pharmaceutical. Funding is acknowledged by the Italian Minister of Health, Ricerca Corrente 2022 (AF), Associazione Italiana per la Ricerca sul Cancro (AIRC, IG No. IG20583), Università Cattolica del Sacro Cuore (UCSC-project D1), and Ministero della Salute Ricerca Corrente 2024-2025 (EB). The funders had no role in the design of the study, collection, analysis, interpretation of the data and writing of the manuscript. Authors’ contribution A.F. and P.G. conceptualized the study and wrote the manuscript. E.G. and E.R. generated ctDNA data and elaborated all data. A.F. G.A. G.F., C.O., A.Z., A.B., E.B., S.G., L.C., and I.P. recruited GIM21 patients and generated/reviewed clinical data. M.A. generated ctDNA data. C.M. biobanked clinical specimens. F.C. chaired and G.S. co-chaired the study. D.G. was in charge of biostatistics. All authors helped to write/review, and approved the manuscript. Acknowledgements P.G. is grateful to the Colleagues of the European Liquid Biopsy Society (ELBS) for enlightening discussions. In memoriam This paper is dedicated to the memory of our dear colleague Giovanni Scambia, who passed away while this work was being finalized. References Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, Dancey J, Arbuck S, Gwyther S, Mooney M, Rubinstein L, Shankar L, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer. 2009;45:228–47. Wahl RL, Jacene H, Kasamon Y, Lodge MA. From RECIST to PERCIST: Evolving Considerations for PET response criteria in solid tumors. J Nucl Med. 2009;50(Suppl 1):S122–50. Hodi FS, Ballinger M, Lyons B, Soria JC, Nishino M, Tabernero J, Powles T, Smith D, Hoos A, McKenna C, Beyer U, Rhee I et al. Immune-Modified Response Evaluation Criteria In Solid Tumors (imRECIST): Refining Guidelines to Assess the Clinical Benefit of Cancer Immunotherapy. J Clin Oncol. 2018:JCO2017751644. Younes A, Hilden P, Coiffier B, Hagenbeek A, Salles G, Wilson W, Seymour JF, Kelly K, Gribben J, Pfreunschuh M, Morschhauser F, Schoder H, et al. International Working Group consensus response evaluation criteria in lymphoma (RECIL 2017). Ann Oncol. 2017;28:1436–47. Lin NU, Lee EQ, Aoyama H, Barani IJ, Barboriak DP, Baumert BG, Bendszus M, Brown PD, Camidge DR, Chang SM, Dancey J, de Vries EG, et al. Response assessment criteria for brain metastases: proposal from the RANO group. Lancet Oncol. 2015;16:e270–8. Jakobsen AKM, Spindler KG. ctDNA-Response evaluation criteria in solid tumors - a new measure in medical oncology. Eur J Cancer. 2023;180:180–83. Cohen SA, Liu MC, Aleshin A. Practical recommendations for using ctDNA in clinical decision making. Nature. 2023;619:259–68. Gouda MA, Janku F, Wahida A, Buschhorn L, Schneeweiss A, Karim NA, Perez DD, Del Re M, Russo A, Curigliano G, Rolfo C, Subbiah V. Liquid Biopsy Response Evaluation Criteria in Solid Tumors (LB-RECIST). Ann Oncol. 2024;35:267–75. O'Leary B, Hrebien S, Morden JP, Beaney M, Fribbens C, Huang X, Liu Y, Bartlett CH, Koehler M, Cristofanilli M, Garcia-Murillas I, Bliss JM et al. Early circulating tumor DNA dynamics and clonal selection with palbociclib and fulvestrant for breast cancer. Nat Commun 2018;9. Parikh AR, Mojtahed A, Schneider JL, Kanter K, Van Seventer EE, Fetter IJ, Thabet A, Fish MG, Teshome B, Fosbenner K, Nadres B, Shahzade HA, et al. Serial ctDNA Monitoring to Predict Response to Systemic Therapy in Metastatic Gastrointestinal Cancers. Clin Cancer Res. 2020;26:1877–85. Yang WY, Feng LF, Meng X, Chen R, Xu WH, Hou J, Xu T, Zhang L. Liquid biopsy in head and neck squamous cell carcinoma: circulating tumor cells, circulating tumor DNA, and exosomes. Expert Rev Mol Diagn. 2020;20:1213–27. Darrigues L, Pierga JY, Bernard-Tessier A, Bièche I, Silveira AB, Michel M, Loirat D, Cottu P, Cabel L, Dubot C, Geiss R, Ricci F et al. Circulating tumor DNA as a dynamic biomarker of response to palbociclib and fulvestrant in metastatic breast cancer patients. Breast Cancer Res 2021;23. Jakobsen A, Andersen RF, Hansen TF, Jensen LH, Faaborg L, Steffensen KD, Thomsen CB, Wen SWC. Early ctDNA response to chemotherapy. A potential surrogate marker for overall survival. Eur J Cancer. 2021;149:128–33. Olga Martínez-Sáez TP, Fara Brasó-Maristany N, Chic B, González-Farré E, Sanfeliu. Adela Rodríguez, Débora Martínez, Patricia Galván, Anna Belén Rodríguez, Francesco Schettini, Benedetta Conte, Maria Vidal, Barbara Adamo, Antoni Martínez, Montserrat Muñoz, Reinaldo Moreno, Patricia Villagrasa, Fernando Salvador, Eva M Ciruelos, Iris Faull, Justin I Odegaard, Aleix Prat. Circulating tumor DNA dynamics in advanced breast cancer treated with CDK4/6 inhibition and endocrine therapy. npj Breast Cancer 2021. Thompson JC, Carpenter EL, Silva BA, Rosenstein J, Chien AL, Quinn K, Espenschied CR, Mak A, Kiedrowski LA, Lefterova M, Nagy RJ, Katz SI, et al. Serial Monitoring of Circulating Tumor DNA by Next-Generation Gene Sequencing as a Biomarker of Response and Survival in Patients With Advanced NSCLC Receiving Pembrolizumab-Based Therapy. Jco Precision Oncol. 2021;5:510–24. Vega DM, Nishimura KK, Zariffa N, Thompson JC, Hoering A, Cilento V, Rosenthal A, Anagnostou V, Baden J, Beaver JA, Phd AACM, Chudova D et al. Changes in Circulating Tumor DNA Reflect Clinical Benefit Across Multiple Studies of Patients With Non-Small-Cell Lung Cancer Treated With Immune Checkpoint Inhibitors. JCO Precision Oncology 2022;6. Anagnostou V, Ho CRY, Nicholas G, Juergens RA, Sacher A, Fung AS, Wheatley-Price P, Laurie SA, Levy B, Brahmer JR, Balan A, Niknafs N, et al. ctDNA response after pembrolizumab in non-small cell lung cancer: phase 2 adaptive trial results. Nat Med. 2023;29:2559–. Assaf ZJF, Zou W, Fine AD, Socinski MA, Young A, Lipson D, Freidin JF, Kennedy M, Polisecki E, Nishio M, Fabrizio D, Oxnard GR et al. A longitudinal circulating tumor DNA-based model associated with survival in metastatic non-small-cell lung cancer. Nat Med. 2023;29. Thompson JC, Scholes DG, Carpenter EL, Aggarwal C. Molecular response assessment using circulating tumor DNA (ctDNA) in advanced solid tumors. Br J Cancer. 2023;129:1893–902. Verma S, Miles D, Gianni L, Krop IE, Welslau M, Baselga J, Pegram M, Oh DY, Dieras V, Guardino E, Fang L, Lu MW, et al. Trastuzumab emtansine for HER2-positive advanced breast cancer. N Engl J Med. 2012;367:1783–91. Hurvitz SA, Hegg R, Chung WP, Im SA, Jacot W, Ganju V, Chiu JWY, Xu B, Hamilton E, Madhusudan S, Iwata H, Altintas S, et al. Trastuzumab deruxtecan versus trastuzumab emtansine in patients with HER2-positive metastatic breast cancer: updated results from DESTINY-Breast03, a randomised, open-label, phase 3 trial. Lancet. 2023;401:105–17. Allegretti M, Fabi A, Giordani E, Ercolani C, Romania P, Nisticò C, Gasparro S, Barberi V, Ciolina M, Pescarmona E, Giannarelli D, Ciliberto G et al. Liquid biopsy identifies actionable dynamic predictors of resistance to Trastuzumab Emtansine (T-DM1) in advanced HER2-positive breast cancer. Mol Cancer 2021;20. De Mattos-Arruda L. Liquid biopsy for HER2-positive breast cancer brain metastasis: the role of the cerebrospinal fluid. ESMO Open. 2017;2:e000270. Abbosh C, Birkbak NJ, Wilson GA, Jamal-Hanjani M, Constantin T, Salari R, Le Quesne J, Moore DA, Veeriah S, Rosenthal R, Marafioti T, Kirkizlar E, et al. Phylogenetic ctDNA analysis depicts early-stage lung cancer evolution. Nature. 2017;545:446–51. Giordani E, Allegretti M, Sinibaldi A, Michelotti F, Ferretti G, Ricciardi E, Ziccheddu G, Valenti F, Di Martino S, Ercolani C, Giannarelli D, Arpino G et al. Monitoring changing patterns in HER2 addiction by liquid biopsy in advanced breast cancer patients. J Exp Clin Cancer Res. 2024;43. Additional Declarations Competing interest reported. A.F.: consultant or advisor for Roche, Lilly, Novartis, AstraZeneca, Pfizer, Seagen, Gilead, MSD, Menarini Stemline, Dompè Biotech; speaker honoraria from Astra Zeneca, Roche, Lilly, Novartis, Gilead, Pfizer, Daiichi Sankyo Exact Sciences; grant support (to the Institution) by Astra Zeneca, Roche. GA: personal fees from Novartis, Lilly; grants and personal fees from Roche; grants, personal fees, and nonfinancial support from Pfizer; grants, personal fees, and nonfinancial support from AstraZeneca; and personal fees from Daichi. E.B.: grants or contracts from Astra-Zeneca, Roche; honoraria for lectures from Merck-Sharp & Dome, Astra-Zeneca, Pfizer, Eli-Lilly, Bristol-Myers Squibb, Novartis, Takeda and Roche; member of Data Safety Monitoring Board or Advisory Board of Merck-Sharp & Dome, Pfizer, Novartis, Bristol-Myers Squibb, Astra-Zeneca, Roche, Amgen and Celltrion. F.C.: consultant for Exact Science and AstraZeneca. Supplementary Files GIM21JECCRsupplementaryJul11.docx GIM21TableS2V32.xlsx Cite Share Download PDF Status: Published Journal Publication published 20 Jan, 2026 Read the published version in Journal of Experimental & Clinical Cancer Research → Version 1 posted Editorial decision: Revision requested 19 Aug, 2025 Reviews received at journal 19 Aug, 2025 Reviews received at journal 04 Aug, 2025 Reviewers agreed at journal 25 Jul, 2025 Reviewers agreed at journal 25 Jul, 2025 Reviewers invited by journal 15 Jul, 2025 Editor assigned by journal 15 Jul, 2025 Submission checks completed at journal 15 Jul, 2025 First submitted to journal 11 Jul, 2025 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-7101376","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":485855187,"identity":"cd17da03-ab5b-4c96-a915-5e4f8e751eb6","order_by":0,"name":"Alessandra Fabi","email":"","orcid":"","institution":"Fondazione Policlinico Universitario Agostino Gemelli IRCCS","correspondingAuthor":false,"prefix":"","firstName":"Alessandra","middleName":"","lastName":"Fabi","suffix":""},{"id":485855188,"identity":"6dd64498-74ce-4407-bc66-92f871573aeb","order_by":1,"name":"Elena Giordani","email":"","orcid":"","institution":"IRCSS Regina Elena National Cancer Institute","correspondingAuthor":false,"prefix":"","firstName":"Elena","middleName":"","lastName":"Giordani","suffix":""},{"id":485855189,"identity":"d6879469-0238-4be0-b9c5-90c506915750","order_by":2,"name":"Elena Ricciardi","email":"","orcid":"","institution":"IRCSS Regina Elena National Cancer Institute","correspondingAuthor":false,"prefix":"","firstName":"Elena","middleName":"","lastName":"Ricciardi","suffix":""},{"id":485855190,"identity":"42b16b24-dde1-4787-852f-5fcc4c71c0de","order_by":3,"name":"Grazia Arpino","email":"","orcid":"","institution":"University Federico II","correspondingAuthor":false,"prefix":"","firstName":"Grazia","middleName":"","lastName":"Arpino","suffix":""},{"id":485855191,"identity":"c54f565b-bb11-4031-8521-8218a72e084e","order_by":4,"name":"Matteo Allegretti","email":"","orcid":"","institution":"IRCSS Regina Elena National Cancer Institute","correspondingAuthor":false,"prefix":"","firstName":"Matteo","middleName":"","lastName":"Allegretti","suffix":""},{"id":485855192,"identity":"de691de5-cc69-4236-a660-4d3db93139f0","order_by":5,"name":"Gianluigi Ferretti","email":"","orcid":"","institution":"IRCSS Regina Elena National Cancer Institute","correspondingAuthor":false,"prefix":"","firstName":"Gianluigi","middleName":"","lastName":"Ferretti","suffix":""},{"id":485855193,"identity":"2a38ac73-ef2d-484b-9031-edd899746f00","order_by":6,"name":"Claudia Omarini","email":"","orcid":"","institution":"University Hospital of Modena","correspondingAuthor":false,"prefix":"","firstName":"Claudia","middleName":"","lastName":"Omarini","suffix":""},{"id":485855194,"identity":"ea8ed38f-d15c-4cca-ab74-c6f102c51b86","order_by":7,"name":"Alberto Zambelli","email":"","orcid":"","institution":"ASST Papa Giovanni XXIII","correspondingAuthor":false,"prefix":"","firstName":"Alberto","middleName":"","lastName":"Zambelli","suffix":""},{"id":485855195,"identity":"624ebf0a-9587-4563-a276-11c9d54d5ade","order_by":8,"name":"Chiara Mandoj","email":"","orcid":"","institution":"IRCCS Regina Elena National Cancer Institute, Istituti Fisioterapici Ospitalieri (IFO)","correspondingAuthor":false,"prefix":"","firstName":"Chiara","middleName":"","lastName":"Mandoj","suffix":""},{"id":485855196,"identity":"95594075-91a9-4b99-8c81-6fc61c3da601","order_by":9,"name":"Andrea Botticelli","email":"","orcid":"","institution":"Sapienza University of Rome","correspondingAuthor":false,"prefix":"","firstName":"Andrea","middleName":"","lastName":"Botticelli","suffix":""},{"id":485855197,"identity":"b53c390c-4f67-4b06-9b82-eb3c084f1cd0","order_by":10,"name":"Emilio Bria","email":"","orcid":"","institution":"Università Cattolica del Sacro Cuore, Fondazione Policlinico Universitario Agostino Gemelli IRCCS","correspondingAuthor":false,"prefix":"","firstName":"Emilio","middleName":"","lastName":"Bria","suffix":""},{"id":485855198,"identity":"dd2ba41c-addb-4941-881d-2e7921df8e83","order_by":11,"name":"Stefania Gori","email":"","orcid":"","institution":"IRCCS-Sacro Cuore Don Calabria Hospital","correspondingAuthor":false,"prefix":"","firstName":"Stefania","middleName":"","lastName":"Gori","suffix":""},{"id":485855199,"identity":"399ac918-9549-4325-97ae-851a2df859e4","order_by":12,"name":"Luisa Carbognin","email":"","orcid":"","institution":"Fondazione Policlinico Universitario Agostino Gemelli IRCCS","correspondingAuthor":false,"prefix":"","firstName":"Luisa","middleName":"","lastName":"Carbognin","suffix":""},{"id":485855200,"identity":"498e5f64-c2a8-4505-ae45-705d6de4d351","order_by":13,"name":"Ida Paris","email":"","orcid":"","institution":"Fondazione Policlinico Universitario Agostino Gemelli IRCCS","correspondingAuthor":false,"prefix":"","firstName":"Ida","middleName":"","lastName":"Paris","suffix":""},{"id":485855201,"identity":"4cc2c563-d2f7-4fbd-95e5-3752ec20444c","order_by":14,"name":"Giovanni Scambia","email":"","orcid":"","institution":"Fondazione Policlinico Universitario Agostino Gemelli IRCCS","correspondingAuthor":false,"prefix":"","firstName":"Giovanni","middleName":"","lastName":"Scambia","suffix":""},{"id":485855202,"identity":"291d3d78-9a85-450b-ac00-c9c54d2a88f9","order_by":15,"name":"Francesco Cognetti","email":"","orcid":"","institution":"Ospedale Isola Tiberina – Gemelli Isola","correspondingAuthor":false,"prefix":"","firstName":"Francesco","middleName":"","lastName":"Cognetti","suffix":""},{"id":485855203,"identity":"6e59724b-e0db-4867-861b-df743cd8f32e","order_by":16,"name":"Diana Giannarelli","email":"","orcid":"","institution":"Fondazione Policlinico Universitario Agostino Gemelli IRCCS","correspondingAuthor":false,"prefix":"","firstName":"Diana","middleName":"","lastName":"Giannarelli","suffix":""},{"id":485855204,"identity":"cb5711af-39af-4f79-b0b8-bd2e548fee03","order_by":17,"name":"Patrizio Giacomini","email":"data:image/png;base64,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","orcid":"","institution":"Fondazione Policlinico Universitario Agostino Gemelli IRCCS","correspondingAuthor":true,"prefix":"","firstName":"Patrizio","middleName":"","lastName":"Giacomini","suffix":""}],"badges":[],"createdAt":"2025-07-11 11:53:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7101376/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7101376/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s13046-025-03605-2","type":"published","date":"2026-01-20T15:59:09+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":87047482,"identity":"e2f51117-0891-4c76-b281-f5a785e52a08","added_by":"auto","created_at":"2025-07-18 14:43:00","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1170167,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSample and data collection, cRECIST metrics and notations. \u003c/strong\u003e(a)\u003cstrong\u003e \u003c/strong\u003eTimelines of: T-DM1 administration (top), blood collection (red), medical imaging (green), and testing (NGS and dPCR, blue and yellow). T: time intervals; w: weeks. (b) Formula to calculate D\u003csub\u003ectDNA\u003c/sub\u003e, e.g. % change of all measured target ctDNAs between two points (T\u003csub\u003en\u003c/sub\u003e vs T\u003csub\u003eref\u003c/sub\u003e), whereby T\u003csub\u003en\u003c/sub\u003e is the test point in a multipoint series, and T\u003csub\u003eref \u003c/sub\u003eis typically either T\u003csub\u003e0\u003c/sub\u003e or T\u003csub\u003enadir\u003c/sub\u003e. (c) Mirroring of classical RECIST 1.1 into ctDNA-RECIST (cRECIST). The same cut-off values (≥20%, ≤30%) are applied by default to classify % dimensional changes in target lesions and ctDNA abundance (D\u003csub\u003ectDNA\u003c/sub\u003e), resulting in 4 possible objective responses (ORs/cORs: PD/cPD, SD/cSD, PR/cPR, and CR/cCR). Few differences are acknowledged between RECIST 1.1 and cRECIST (boxed).\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7101376/v1/462b8bee7a723f19ba8373a3.png"},{"id":87047484,"identity":"686448c7-c035-488e-8002-7589b8762f73","added_by":"auto","created_at":"2025-07-18 14:43:00","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":580999,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ectDNA-informed GIM21 flowchart.\u003c/strong\u003e (a) Flowchart summarizing recruitment, attrition, and cPD-based classification of GIM21 patients. (b) Kaplan-Meier curves: cPD-positive vs cPD-negative patients. PFS: Progression-Free Survival. OS: Overall Survival. cPD: time elapsed fron T\u003csub\u003e0\u003c/sub\u003e to cRECIST-compliant ctDNA increase. Median PFS and OS are indicated. cPD assigned by default cRECIST cut-off values.\u0026nbsp;\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7101376/v1/27c1421607ff18c727c6f73c.png"},{"id":87048493,"identity":"513da6b8-9464-404d-8a42-f35b70e7b8e6","added_by":"auto","created_at":"2025-07-18 14:51:00","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1342949,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ecRECIST vs RECIST 1.1: timelines, frequency and co-occurrence of objective responses\u003c/strong\u003e. (a) Patient-by-patient bar plots of Objective ctDNA and tumor Responses (cORs top and ORs bottom) as per cRECIST and RECIST 1.1 assessments of the 27 cPD-positive GIM21 patients. Four possible OR/cORs are noted at the pre-planned time points and color-coded consistently (inset showing PD/cPD, SD/cSD, PR/cPR and CR/cCR), then longitudinally displayed in the timelines until PD and cPD for ORs and cORs respectively. A change in color means change in OR/cOR. Only ctDNA time points in the w3/w9 dataset are considered, whereas for homogeneous handling of the entire dataset the time-dense w3 dataset was disregarded. (b) Donut charts displaying the observed ORs and cORs by frequency. (c) Pearson’s correlation matrix displaying the number of patients in each of the 9 possible best OR/cOR co-occurrence categories. All elaborations carried out at default cut-off values.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7101376/v1/13e99996b78cd14479ee6f24.png"},{"id":87047490,"identity":"0b25eb2f-8886-451e-9c94-a189fb55def9","added_by":"auto","created_at":"2025-07-18 14:43:00","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":979203,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eObjective responses and outcome\u003c/strong\u003e. (a-c) Kaplan-Meier curves of 27 cPD-positive patients sorted by best OR (greatest tumor shrinkage). (d-f) As above, sorted by best cOR (greatest D\u003csub\u003ectDNA\u003c/sub\u003e drop). OR: Objective Response. PFS: Progression-Free (PD-free) Survival. cPFS: ctDNA-PFS, eg. cPD-Free Survival. OS: Overall Survival. \u003cem\u003ep \u003c/em\u003evalues are from pooled analysis of PD vs SD vs PR/CR, and cPD vs cSD vs cPR/cCR. (g) Violin plots comparing days [95% CI] for the variables defined in figure, e.g.: ctDNA-PFS (cPFS, 65 [154-47]), Lead Time (LT, 105 [243-42]), and PFS (182 [315-154]). All values calculated for the 27 cPD-positive patients altogether. (h-j) linear regression curves (best fit) between any two of LT, PFS, and cPFS, with correlation coefficients. * pt#30: exceptional responder, cPD\u003csup\u003e+\u003c/sup\u003e/PD\u003csup\u003e-\u003c/sup\u003e at last follow up. (k) PFS and cPFS plotted by patient number after patient rankeing by LT. cORs assigned by default cut-off values in all panels.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7101376/v1/6484798d9d5d9672167aeb3e.png"},{"id":87047493,"identity":"7155f1a4-d095-456a-adc1-bc1e0a5249a2","added_by":"auto","created_at":"2025-07-18 14:43:00","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":803438,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCombining cRECIST with multipoint trends (\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eTr\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e). \u003c/strong\u003e(a) Schematic \u003cem\u003eTr\u003c/em\u003e calculation: three consecutive time points form a triplet, and triplets overlap by two points. \u003cem\u003eTr\u003c/em\u003e is calculated for each triplet by averaging three consecutive ctDNA measurements, each point as follows: (S[VAF\u003cem\u003e(T\u003c/em\u003e\u003csub\u003e\u003cem\u003en\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e) \u003c/em\u003e+ CN\u003cem\u003e(T\u003c/em\u003e\u003csub\u003e\u003cem\u003en\u003c/em\u003e\u003c/sub\u003e)]). Series from consecutive overlapping triplets define a ctDNA trend\u003cem\u003e \u003c/em\u003e(\u003cem\u003eTr\u003c/em\u003e), either positive or negative. (b) Swimmer plots of the 27 cPD-positive patients displaying placeholders (clinical/ctDNA events listed in the inset), scored by three different metrics: RECIST 1.1 (PD, end of timeline), cRECIST (ctDNA\u003csub\u003enadir\u003c/sub\u003e, cPD), and non-cRECIST (\u003cem\u003eTr\u003c/em\u003e). (c) linear regression plots: duration of the first post-cPD ctDNA decrease (calculated by \u003cem\u003eTr\u003c/em\u003e) vs PFS, LT and OS. * pt#30: exceptional responder, cPD\u003csup\u003e+\u003c/sup\u003e/PD\u003csup\u003e-\u003c/sup\u003e at last follow up.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7101376/v1/688729956fa9d851158d5a94.png"},{"id":87048494,"identity":"6a5c588c-c920-4a14-86f3-228662444bb1","added_by":"auto","created_at":"2025-07-18 14:51:00","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":669130,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ecRECIST and RECIST1.1: model, integration and metrics. \u003c/strong\u003eAn adaptive equilibrium is displayed between T-DM1 and tumor variants, in a hypothetical patient displaying a frequent phenotypic combination of best OR and best cOR, e.g. tumor stability combined with ctDNA response (SD/cPR; see Fig. 3c). Contraction and expansion of T-DM1-sensitive and T-DM1 resistant tumor variants (blue and multicolor dots respectively, middle) are noted by ORs (above) and cORs (below) on two distinct timelines corresponding to the applicable timeframes of RECIST 1.1 and cRECIST (double arrows), e.g. from T\u003csub\u003e0\u003c/sub\u003e to PD and T\u003csub\u003e0\u003c/sub\u003e to cPD respectively. Tumor dynamics are poorly captured by RECIST 1.1 (‘flat’ blue trajectory and unchanged OR assignments), but recapitulated by successive D\u003csub\u003ectDNA\u003c/sub\u003e decreases and increases (‘waving’ red trajectory and changing cORs). ctDNA goes from baseline down to ctDNA nadir, up until and beyond cPD, and then down and up again post-cPD, a pattern termed ctDNA waving herein. The pre-cPD and post-cPD periods are monitored by two distinct, non-redundant ctDNA metrics: cPFS and \u003cem\u003eTr\u003c/em\u003e. cPFS is a cRECIST time-dependent variable based on two-point comparisons. It discriminates patient subsets with broadly different outcomes by measuring the time elapsed between T\u003csub\u003e0\u003c/sub\u003e and cPD. \u003cem\u003eTr\u003c/em\u003e (three-point ctDNA \u003cem\u003eTr\u003c/em\u003eends) is a non-cRECIST metric only applicable beyond the cRECIST timeframe. Combined with knowledge of cPFS, the duration of the first post-cPD \u003cem\u003eTr\u003c/em\u003e drop (boxed in red) is proposed to personalize ctDNA-assisted outcome prediction.\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7101376/v1/0e1ae199513d46aeb171af20.png"},{"id":101151921,"identity":"a8908bf1-28f0-4059-99bf-46e98870ae06","added_by":"auto","created_at":"2026-01-26 16:08:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6196115,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7101376/v1/60db718e-4d1d-402b-b32b-845be86a2d04.pdf"},{"id":87048498,"identity":"36786c23-adf3-44eb-b661-83b3eb86bc11","added_by":"auto","created_at":"2025-07-18 14:51:00","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":12039326,"visible":true,"origin":"","legend":"","description":"","filename":"GIM21JECCRsupplementaryJul11.docx","url":"https://assets-eu.researchsquare.com/files/rs-7101376/v1/de6da572639715293a6e96fc.docx"},{"id":87047486,"identity":"06241edd-362a-4801-9771-1c9a0410b832","added_by":"auto","created_at":"2025-07-18 14:43:00","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":93366,"visible":true,"origin":"","legend":"","description":"","filename":"GIM21TableS2V32.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7101376/v1/01e2b8e36ddefbe9104ccf44.xlsx"}],"financialInterests":"Competing interest reported. A.F.: consultant or advisor for Roche, Lilly, Novartis, AstraZeneca, Pfizer, Seagen, Gilead, MSD, Menarini Stemline, Dompè Biotech; speaker honoraria from Astra Zeneca, Roche, Lilly, Novartis, Gilead, Pfizer, Daiichi Sankyo Exact Sciences; grant support (to the Institution) by Astra Zeneca, Roche. GA: personal fees from Novartis, Lilly; grants and personal fees from Roche; grants, personal fees, and nonfinancial support from Pfizer; grants, personal fees, and nonfinancial support from AstraZeneca; and personal fees from Daichi. E.B.: grants or contracts from Astra-Zeneca, Roche; honoraria for lectures from Merck-Sharp \u0026 Dome, Astra-Zeneca, Pfizer, Eli-Lilly, Bristol-Myers Squibb, Novartis, Takeda and Roche; member of Data Safety Monitoring Board or Advisory Board of Merck-Sharp \u0026 Dome, Pfizer, Novartis, Bristol-Myers Squibb, Astra-Zeneca, Roche, Amgen and Celltrion. F.C.: consultant for Exact Science and AstraZeneca.","formattedTitle":"Circulating tumor DNA and Response Evaluation Criteria In Solid Tumors: ctDNA-RECIST proof-of-concept in HER2-positive metastatic breast cancer","fulltext":[{"header":"BACKGROUND","content":"\u003cp\u003eResponse Evaluation Criteria In Solid Tumors (RECIST) 1.1 assess clinical objective response (OR) to treatment in metastatic cancer based on dimensional changes of tumor lesions measured by medical imaging. The RECIST 1.1 scale, metrics and conventions guide oncology practice, define measurable endpoints in clinical trials, and ultimately provide an objective framework for the regulatory approval of new drugs [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. However, the limitations of a purely anatomical assessment of response to treatment are widely recognized, and additional RECIST criteria have been proposed to evaluate Positron Emission Tomography (PET) and immune Response (PERCIST and iRECIST/imRECIST) [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Alternative and additional criteria apply to lymphomas [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] and organ-specific assessment, such as brain metastases [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThe potential advantages to integrate circulating tumor DNA (ctDNA) into RECIST 1.1 (and possibly other objective response scales) have been outlined [\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e–\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. The EORTC RECIST working group (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://recist.eortc.org\u003c/span\u003e\u003cspan address=\"https://recist.eortc.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and the European Liquid Biopsy Society (www.ELBS.eu) are at work to provide guidelines and recommendations for ctDNA-RECIST (cRECIST hitherto). However, publicly available ctDNA datasets are presently much smaller than the medical imaging datasets interrogated during successive RECIST refinements. Moreover, although changes in ctDNA levels may effectively stratify patients into good and poor responders [\u003cspan additionalcitationids=\"CR10 CR11 CR12 CR13 CR14 CR15 CR16 CR17 CR18\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e–\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], no prospective clinical studies have been reported, to our knowledge, to estimate ctDNA-adjusted risk of clinical progression in individual patients.\u003c/p\u003e\u003cp\u003eHerein, we report on a multi-center prospective study, designed by the Gruppo Italiano Mammella (GIM) and called GIM21/LiqERBcept (NCT05735392). The primary aim of GIM21 was to assess relationships between ctDNA and medical imaging (e.g. RECIST 1.1 assessments) in HER2-positive, metastatic breast cancer patients receiving the first-in-class Antibody-Drug Conjugate (ADC) Trastuzumab emtansine (T-DM1) as standard of care second-line treatment [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Despite T-DM1 has now been superseded in this indication by an irinotecan-based Trastuzumab conjugate (Trastuzumab deruxtecan; T-DXd) [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], second-line T-DM1 treatment remains particularly appealing from a methodological viewpoint. In this setting, the expansion of tumor variants carrying known cancer drivers, as monitored by ctDNA, coincides with loss of HER2 amplification/addiction [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. This crucial oncogenic remodeling/switch may be instrumental to align objective clinical response, ctDNA trajectories, T-DM1 efficacy, and outcome, e.g. to model a self-consistent RECIST/cRECIST framework.\u003c/p\u003e\u003cp\u003eIn the GIM21 study, we have investigated for the first time how changes in ctDNA levels (a continuous variable) may be translated into an objective discontinuous cRECIST scale. Similar to the 4 classical RECIST 1.1 Objective Responses (ORs), e g. Progressive Disease, Stable Disease, Partial Response and Complete Response (PD, SD, PR and CR), cRECIST was postulated to feature 4 ctDNA-ORs (cORs), e.g. cPD, cSD, cPR and cCR. cRECIST is proposed herein to be robust and intuitive enough to stratify patients by disease outcome and, with caveats and additions, also to guide personalized clinical decisions in clinical trials and everyday clinical practice.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cp\u003e\u003cb\u003ePatients\u003c/b\u003e\u003c/p\u003e\u003cp\u003eGIM21 is an open-label, observational (blood drawing implemented as an additional procedure), multicenter, single-arm, prospective study (NCT05735392) conducted between July 2018 (first patient in) and December 2022 (last blood drawing). It recruited consecutive adult patients with a documented diagnosis of HER2-positive metastatic BC, as per ImmunoHistoChemistry score (IHC 3+, or IHC 2 + and HER2 amplification ratio ≥ 2.0). Eligible patients had no more than one line of previous anti-HER2 treatment for advanced disease (Trastuzumab, either alone or combined with Pertuzumab plus taxane). For additional inclusion and exclusion criteria, age, demographics and clinical information see Supplemental Methods and Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e. T-DM1 was administered at the standard dose of 3.6 mg/kg Q21 days until disease progression or development of unmanageable toxic effects (see Supplemental Methods). Follow-up was extended during subsequent therapy regimens: last clinical update September 1, 2024.\u003c/p\u003e\u003cp\u003e\u003cb\u003ePrimary aim and pre-planned analysis\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe primary aim of GIM21 was to assess ctDNA changes vs tumor size (RECIST 1.1), which implies a primary endpoint of at least one detectable ctDNA species per patient. This precondition was met in 80% of the patients enrolled in a similar previous study [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], resulting in a sample size calculation of expected ctDNA-positives/enrolled patients of 36/45 (95% CI; 65.4–90.4%). Observed ctDNA positives were 38/43 (88%), meeting primary endpoint.\u003c/p\u003e\u003cp\u003e\u003cb\u003eClinical samples and medical imaging\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo capture early ctDNA changes, commonly observed during T-DM1 treatment [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], blood was obtained right before treatment (baseline, T\u003csub\u003e0\u003c/sub\u003e), and then at 3 progressively longer intervals: 3 weeks (T\u003csub\u003e1\u003c/sub\u003e to T\u003csub\u003e3\u003c/sub\u003e), 9 weeks (T\u003csub\u003e4\u003c/sub\u003e to T\u003csub\u003e6\u003c/sub\u003e), and 12 weeks until progression (T\u003csub\u003e7\u003c/sub\u003e to T\u003csub\u003ep\u003c/sub\u003e). CT scans were instead obtained as per standard of care: at progression from previous treatment, and then every 12 weeks. This resulted in a ‘staggered’ (w3/w9/w12 vs w12) collection scheme until T\u003csub\u003e6\u003c/sub\u003e, followed by ctDNA/CT scan alignment (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea, compare green and red timelines). Blood was drawn immediately before (minutes) T-DM1 infusion at all time points except T\u003csub\u003ep\u003c/sub\u003e, that coincided with the last GIM21 visit (end of treatment), on average 7 days after the final CT scan documenting clinical progression. Technical details on blood drawing and pre-analytical processing are provided in supplementals and elsewhere [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e].\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eNGS and dPCR\u003c/b\u003e\u003c/p\u003e\u003cp\u003eA targeted NGS panel (Oncomine Pan-Cancer) was used detecting: (a) SNVs and INDELs from 52 major cancer drivers and tumor suppressor genes, (b) DNA copy number variations (CNV) from 12 genes, and (c) rearrangements from 12 genes. Alterations were called by NGS at 3 selected time points (T\u003csub\u003e0\u003c/sub\u003e, T\u003csub\u003e6\u003c/sub\u003e and T\u003csub\u003ep\u003c/sub\u003e; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea, blue timeline), except when progression occurred at or before T\u003csub\u003e6\u003c/sub\u003e, which resulted in two NGS testing points only (T\u003csub\u003e0\u003c/sub\u003e and T\u003csub\u003ep\u003c/sub\u003e). Alterations captured by NGS were systematically monitored (T\u003csub\u003e0\u003c/sub\u003e to T\u003csub\u003ep\u003c/sub\u003e; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea, orange timeline) by bespoke digital PCR (dPCR) assays, either custom-designed or commercially obtained. Each dPCR assay was pre-tested on serial dilutions of tumor tissue DNAs (tDNAs) carrying (n ≥ 2) or lacking (n = 1) the alteration being assessed, and was accepted if its Limit of Detection (LoD) was at 0.01% VAF or better. All reagents, equipment, and online NGS/dPCR bioinformatic/interpretation tools were from ThermoFisher Scientific.\u003c/p\u003e\u003cp\u003e\u003cb\u003eRECIST 1.1 and ctDNA-RECIST (cRECIST)\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAs per RECIST 1.1 [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e], % change in tumor size is routinely expressed as the sum of diameters of target lesions relative to the lowest measurement on study. For symmetry, SNV and CNV measurements were made commensurable by expressing them as a single absolute value after rounding to two decimal places (e.g. VAF = 1.101% and CN = 2.229 made equal to 1.10 and 2.23). Absolute values were then summed and expressed as % change (D\u003csub\u003ectDNA\u003c/sub\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb) relative to a reference time point, e.g. T\u003csub\u003en\u003c/sub\u003e vs T\u003csub\u003eref\u003c/sub\u003e, whereby T\u003csub\u003en\u003c/sub\u003e is any time point and T\u003csub\u003eref\u003c/sub\u003e is either T\u003csub\u003e0\u003c/sub\u003e (baseline) or T\u003csub\u003enadir\u003c/sub\u003e (lowest ctDNA measurement) whichever the lesser. To turn D\u003csub\u003ectDNA\u003c/sub\u003e into a discontinuous cRECIST scale, the 4 possible objective RECIST 1.1 responses (ORs) were mirrored into 4 corresponding ctDNA objective responses (cORs), e.g. PD/cPD, SD/cSD, PR/cPR and CR/cCR. The same cut-offs defining the upper and lower limits of SD (≥ 20% and ≤ 30% respectively) were applied by default to cSD (hitherto ‘default cut-offs’; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec), but were also the subject of extensive exploratory analysis (see below). PD/cPD and CR/cCR were similarly scored upon detection of a new tumor lesion/new ctDNA species, and loss of all detectable lesions/target ctDNAs by CT scan and dPCR respectively (for dPCR LoD, 0.01%VAF, see above). Two exceptions to RECIST 1.1/cRECIST mirroring are acknowledged: (i) the distinction between target and non-target lesions vs complete equivalence of all ‘target’ ctDNAs; (ii) the upper limit in the number of target lesions (n = 5) in RECIST 1.1 vs no pre-defined limit in target ctDNA species. For symmetry, default monitoring by CT scan and dPCR stopped at PD (end of treatment) and cPD respectively.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStatistics\u003c/b\u003e\u003c/p\u003e\u003cp\u003eResults were reported by descriptive statistics and 95% CI. Parametric and non-parametric tests were used to evaluate associations between variables. Progression-free survival (PFS) was calculated as the time between the first T-DM1 administration and either the first evidence of progressive disease or the time of last follow-up. Survival was evaluated by the Kaplan-Meier method. Data were elaborated by GraphPad Prism v10 (RRID:SCR_000306) and IBM SPSS statistics (RRID:SCR_002865).\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003e\u003cb\u003eGIM21 study flowchart\u003c/b\u003e\u003c/p\u003e\u003cp\u003eOut of 50 patients enrolled (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea), 2 withdrew and 5 were excluded due to either missing blood drawings or poor cfDNA quality. Five additional patients were not evaluable because lacking detectable target ctDNAs (n\u0026thinsp;=\u0026thinsp;5). The remaining 38 ctDNA-positive patients were sorted into cPD-positives (n\u0026thinsp;=\u0026thinsp;27) and cPD-negatives (n\u0026thinsp;=\u0026thinsp;11). cPD positives displayed at least one D\u003csub\u003ectDNA\u003c/sub\u003e increase at or above the default\u0026thinsp;\u0026ge;\u0026thinsp;20% cSD/cPD cut-off (see Methods and Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec). Since most D\u003csub\u003ectDNA\u003c/sub\u003e increases exceeded the default cut-off, the number of cPD-positives was minimally affected by alternative cPD cut-offs within the \u0026ge;\u0026thinsp;10% to \u0026ge;\u0026thinsp;50% range (Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Therefore, cRECIST allowed for wide confidence intervals in objective cPD assignment.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eConsistency was noted between patient features and ctDNA status. For instance, only one exceptional responder was still PD-free at last follow-up after 1000 days of treatment despite being cPD-positive. All the remaining 26 cPD positives experienced PD at extra-cranial sites. Conversely, all 5 patients undergoing PD exclusively due to brain metastases were cPD-negative, which is unsurprising given poor ctDNA release from intracranial locations [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Finally, the only patient discontinuing T-DM1 due to toxicity did so while in complete objective response, as concordantly assessed by her cPD/PD double-negative status.\u003c/p\u003e\u003cp\u003eDemographics and the clinical pathological features of cPD-positive and cPD-negative patients are separately presented in Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e. No patient was lost to follow up. At last evaluation, 11/27 cPD-positives and 6/11 cPD-negatives were alive. Despite enrichment in cases with brain metastases, PFS and OS did not significantly differ between cPD-negatives and cPD-positives (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb and c). Therefore, detectable cPD did not apparently select for outcome.\u003c/p\u003e\u003cp\u003e\u003cb\u003eCirculating genomic alterations\u003c/b\u003e\u003c/p\u003e\u003cp\u003eDetails are provided in Figs. S2-S4 about: bespoke NGS/dPCR testing, oncoprints, distribution of genomic alterations at the different time points, actionable levels, technical NGS/dPCR validation, selection of dPCR as the standard cRECIST readout, and application of VAF as the optimal metric to measure ctDNA. Raw ctDNA measurements are provided in an annotated spreadsheet as Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003cb\u003eDivergent trajectories: tumor lesions vs target ctDNAs\u003c/b\u003e\u003c/p\u003e\u003cp\u003eFor a detailed longitudinal evaluation of individual trajectories, each of 113 tumor lesions and 78 target ctDNAs was assessed as described in supplemental results (Fig. S5). Compared to lesions, ctDNAs underwent more numerous, frequent, discordant, asynchronous and extreme changes, including occasional direct switch from loss to gain at immediately consecutive time points. In summary, stable disease was rare from the ctDNA standpoint, and evolutionary divergence of ctDNA variants largely exceeded dissociated radiological responses among tumor lesions measured by CT scans. Then, it was of interest to determine whether this translates into RECIST 1.1/cRECIST divergence.\u003c/p\u003e\u003cp\u003e\u003cb\u003eRECIST 1.1 vs cRECIST\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo compare RECIST 1.1 and cRECIST, as per GIM21 primary aim, target lesion diameters and D\u003csub\u003ectDNA\u003c/sub\u003e values were calculated at the pre-planned time points from the 27 cPD-positive GIM21 patients. Objective clinical and ctDNA responses (OR and cORs) were longitudinally assigned by application to both tumor lesions and ctDNAs of default RECIST 1.1 cut-offs (SD/PD\u0026thinsp;\u0026ge;\u0026thinsp;20% and SD/PR\u0026thinsp;\u0026le;\u0026thinsp;30%, see Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). ORs/cORs were color-matched (SD/cSD, PR/cPR, CR/cCR, PD/cPD), and displayed as timelines until PD and cPD respectively. Despite a much longer OR observation period, assignments changed more frequently for cORs than ORs (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea, see color patchworks), and the total numbers of recorded response changes were similar (n\u0026thinsp;=\u0026thinsp;52 vs 51; donut charts in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb). Altogether, cSD was under-represented compared to SD (16%vs 44%), whereas cPR and cCR were over-represented compared to PR and CR (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb). In other words, due to divergence between individual tumor lesions and ctDNAs (see Fig. S5), ORs and cORs also diverged.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo further investigate divergence, best Objective Responses (best ORs and best cORs) were derived from the same dataset. For consistency, best OR (classically defined as the greatest recorded tumor shrinkage) was mirrored into best cOR (defined as the deepest D\u003csub\u003ectDNA\u003c/sub\u003e drop on study). Interestingly, a Pearson\u0026rsquo;s correlation matrix measuring co-occurrence (PD/cPD, SD/cSD, PR/cPR, and CR/cCR) was significantly (McNemar-Bowker test p\u0026thinsp;=\u0026thinsp;0.007) skewed off-diagonal, particularly toward a dominant best SD/best cPR-cCR combination phenotype seen in 13/27 patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). Therefore, ctDNA responses were clearly detectable in almost half of the cPD-positive patients achieving disease stability as their best response. Similar patterns (timelines, pie charts and Pearson\u0026rsquo;s plots) were seen at upper/lower SD/cSD cut-offs, at least within the \u0026ge;\u0026thinsp;50% to \u0026le;\u0026thinsp;60% range (Figs. S6-S9).\u003c/p\u003e\u003cp\u003eCut-off-independent cOR and best cOR assignments demonstrated that trajectory divergence between tumor size and ctDNA does not depend on how they are measured, but on an intrinsically deeper response to T-DM1 treatment of ctDNA compared to tumor lesions.\u003c/p\u003e\u003cp\u003e\u003cb\u003ecRECIST and clinical outcome\u003c/b\u003e\u003c/p\u003e\u003cp\u003eNext, best objective responses (best ORs and best cORs) were correlated with the outcomes (PFS and OS) of the 27 cPD-positive patients. Unsurprisingly, tumor shrinkage/best OR correlated with PFS, and a non-significant trend was also noted for OS (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea and b). In contrast, deepest ctDNA drop/best cOR correlated with neither (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed and e). This poor correlation was puzzling, since many groups have observed that quantitative ctDNA changes do correlate with treatment efficacy (see for instance [\u003cspan additionalcitationids=\"CR10 CR11 CR12 CR13 CR14 CR15 CR16 CR17 CR18\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]). To clarify whether this discrepancy might arise from cOR being measured on a discontinuous scale, extensive simulations were carried out applying both discontinuous and conventional continuous metrics to the GIM21 dataset. However, quantitative ctDNA changes could not be linked to outcome (see supplemental results and Fig. S10). Then, ctDNA amounts were disregarded, and a time-dependent cRECIST variable was considered termed ctDNA-PFS (cPFS). Analogous to PFS, that measures the time elapsed between T\u003csub\u003e0\u003c/sub\u003e and PD, cPFS was defined as the time elapsed between T\u003csub\u003e0\u003c/sub\u003e and cPD. Interestingly, and in agreement with ctDNA amounts being poorly predictive in the GIM21 setting, cPFS did correlate with tumor shrinkage (best OR; Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec) and not ctDNA drop (best cOR, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef). Moreover, since best OR correlated with PFS (see Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea), cPFS appeared to correlate, albeit indirectly, with the outcome of broad patient subsets.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThen, cPFS was assessed side-by side with two other time-dependent variables: LT and PFS. cPFS and LT define two consecutive treatment periods (T\u003csub\u003e0\u003c/sub\u003e to cPD, and cPD to T\u003csub\u003ep\u003c/sub\u003e), whereas PFS measures the two periods altogether (T\u003csub\u003e0\u003c/sub\u003e to PD). Side-by-side plotting showed that cPFS was shortest and least variable (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eg). It weakly correlated (by regression analysis) with PFS, and not at all with LT (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eh and i). In contrast, LT/PFS correlations were supported by regression plots (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ej), and roughly concordant LT/PFS patient ranking (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ek). Moreover, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ek showed that the gap between cPD and PD is in most cases small when PD occurs before median PFS, and then progressively expands, e.g. ctDNA is of special interest for long-responders.\u003c/p\u003e\u003cp\u003eAltogether, the results in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e and Fig. S10 revealed that time to cPD (cPFS) correlates with outcome much better than quantitative ctDNA changes. Unfortunately, possibly due to its limited variation ranges, cPFS distinguished the outcomes of broad patient subsets but poorly resolved the outcomes of individual patients, particularly long-responders. When more stringent cut-offs were applied, the reasons for this limitation became evident, since even minimal changes in cPFS measurements completely disrupted cPFS/best OR correlations, minimally if at all affecting LT/PFS correlations (Fig. S11). Then, it was hypothesized that pre-cPD metrics (including cPFS) are suboptimal in the GIM21 setting because T-DM1 response is mostly determined by events occurring late during treatment, particularlywithin the LT/PFS overlap, e.g. post-cPD.\u003c/p\u003e\u003cp\u003e\u003cb\u003eCombining cRECIST with non-cRECIST outcome indicators\u003c/b\u003e\u003c/p\u003e\u003cp\u003eWhile looking for post-cPD variables specifying individual outcomes, it was inferred that it might be inappropriate to compare series of disconnected paired measurements, each referring to the same reference value (T\u003csub\u003en\u003c/sub\u003e vs T\u003csub\u003eref\u003c/sub\u003e, T\u003csub\u003en+1\u003c/sub\u003e vs T\u003csub\u003eref\u003c/sub\u003e, etc), as per RECIST 1.1/cRECIST conventions. This might miss individual, informative multipoint ctDNA trends in complex, zig-zagging ctDNA trajectories (see for instance Fig. S4b), particularly in the post-cPD period. Then, the simplest possible multipoint series were considered, e.g. consecutive 3-point series with 2-point overlap (triplets). D\u003csub\u003ectDNA\u003c/sub\u003e values were measured at each point (T\u003csub\u003e0\u003c/sub\u003e to T\u003csub\u003ep\u003c/sub\u003e), and then averaged for each triplet, resulting in a triplet \u003cem\u003eTr\u003c/em\u003eend value dubbed \u003cem\u003eTr\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). It was observed that none of two consecutive \u003cem\u003eTr\u003c/em\u003e values was identical to the second decimal digit, resulting in a simplified scoring system with two alternative trends only (regardless of \u003cem\u003eTr\u003c/em\u003e magnitude), e.g. either ctDNA increase (\u003cem\u003eTr\u003c/em\u003e\u003csub\u003en\u003c/sub\u003e\u0026gt;\u003cem\u003eTr\u003c/em\u003e\u003csub\u003en\u0026minus;1\u003c/sub\u003e) or ctDNA decrease (\u003cem\u003eTr\u003c/em\u003e\u003csub\u003en\u003c/sub\u003e\u0026lt;\u003cem\u003eTr\u003c/em\u003e\u003csub\u003en\u0026minus;1\u003c/sub\u003e). Disappearance of a previously detected ctDNA, and \u003cem\u003ede novo\u003c/em\u003e appearance of a new ctDNA were assimilated to \u003cem\u003eTr\u003c/em\u003e decrease and increase respectively. Finally, swimmer plots were generated displaying ctDNA\u003csub\u003enadir\u003c/sub\u003e, cPD, PD, and \u003cem\u003eTr\u003c/em\u003e on the timelines of the 27 cPD-positive patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eSwimmer plots revealed that: (a) ctDNA\u003csub\u003enadir\u003c/sub\u003e, cPD and PD invariably occur in this order, e.g. they behave as \u0026lsquo;placeholders\u0026rsquo;; (b) successive \u003cem\u003eTr\u003c/em\u003e increases and decreases, termed \u0026lsquo;ctDNA waving\u0026rsquo;, are evident in several timeline sections; (c) increase-only \u003cem\u003eTr\u003c/em\u003e patterns are seen in most (13/15) early progressors (PFS\u0026thinsp;\u0026lt;\u0026thinsp;median PFS), whereas \u003cem\u003eTr\u003c/em\u003e decreases (albeit transient) are preferentially seen in late progressors (PFS\u0026thinsp;\u0026gt;\u0026thinsp;median PFS; 10/12); (d) at least 2 consecutive opposing trends are observed in 4/12 late progressors (pts 51, 34, 31 and 19); (e) PD occurs in most (25/27) patients during phases of \u003cem\u003eTr\u003c/em\u003e increase, and not decrease. Finally, and most interestingly (f), in 15 data-dense patients with at least three consecutive measurable \u003cem\u003eTr\u003c/em\u003e values, the duration of the first post-cPD \u003cem\u003eTr\u003c/em\u003e decrease significantly correlated with PFS and LT, although not with OS (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec). \u003cem\u003eTr\u003c/em\u003e (combined with cPFS) was a much better metric than cPFS alone (\u003cem\u003eTr\u003c/em\u003e in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec vs cPFS in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eh: R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.76 vs 0.30). This improvement exclusively applied to \u003cem\u003eTr\u003c/em\u003e drops calculated after cPD, since the first \u003cem\u003eTr\u003c/em\u003e drop (at any time, regardless of cPD) poorly correlated with PFS (Fig. S12).\u003c/p\u003e\u003cp\u003eIn summary, to calculate outcome, cPFS (cRECIST) and \u003cem\u003eTr\u003c/em\u003e (non-cRECIST) had to be separately assessed and then combined, implying non-redundancy and subordinacy of the two variables. Thus, distinct events/variables underly T-DM1 efficacy in the pre-cPD and post-cPD (LT) periods.\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eRECIST 1.1 was developed to assist in the routine serial monitoring of treatment efficacy. Conversion of continuous measurements (tumor size) into a simple set of discontinuous OR metrics (PD, SD, PR, CR) is crucial to support multiple binary (go/no-go) decision nodes at patient re-evaluation in advanced cancer. The minimal expectation is that cRECIST should be equally discontinuous, binary, intuitive, and robust. In addition, should cRECIST forecast outcome, particularly for individual patients, it would meet diverse clinical needs with a common, highly personalized tool. The present GIM21 study provides an initial framework to embed cRECIST into RECIST 1.1, and proposes novel metrics to overcome the inherent limitations of RECIST-like approaches when applied to outcome prediction.\u003c/p\u003e\u003cp\u003eInspired by the previous LiqBreasTrack study, conducted in a similar setting [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], the GIM21 protocol was designed for staggered blood/CT scan collection and greater ctDNA data density at initial time points (w3/w9/w12 schedule; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). Unlike tumor-informed NGS/PCR approaches widely used to detect minimal residual disease [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e], our bespoke NGS/dPCR assay captured and monitored genomic alterations exclusively in blood, not to miss adaptively acquired target ctDNAs arising \u003cem\u003ede novo\u003c/em\u003e in approximately half of T-DM1-treated patients [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Based on Poisson distribution analysis in up to 20.000 nanoliter-size partitions, bespoke dPCR effectively normalized for wild-type alleles (VAF rather than copies/mL), and compensated for pre-analytical variation (sample processing, cfDNA recovery, purity, and concentration), eliminating the need for duplicate testing (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, S2-S4). Robustness and low cost are key features for real-world implementation.\u003c/p\u003e\u003cp\u003eFor comprehensive overview of all ctDNAs, VAF and CNVs were computed into a single unit measuring overall % ctDNA changes (D\u003csub\u003ectDNA\u003c/sub\u003e; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb). In addition, the RECIST 1.1 scoring system was mirrored as much as possible into cRECIST, each objective tumor response having a ctDNA counterpart: SD/cSD, PD/cPD, CR/cCR and PR/cPR. Most crucial were the standard SD/PD (\u0026ge;\u0026thinsp;20%) and SD/PR (\u0026le;\u0026thinsp;30%) cut-offs of RECIST 1.1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec), that to a first approximation were applied by default throughout the GIM21 study. In the lack of extensive retrospective ctDNA data warehouses, default cut-offs are supported by simplicity, immediacy, and mnemonic ease. Nevertheless, they may be questioned as arbitrary and/or unsupported by objective evidence. Then, cRECIST was validated across wide cut-off ranges (Figs. S1, S6-S9 and S11). Interestingly, due to the considerable magnitude of most observed ctDNA changes, cOR and best cOR assignments remained largely cut-off-independent, and cPD positioning exhibited limited variations at least within the \u0026ge;\u0026thinsp;20% to \u0026ge;\u0026thinsp;50% range for cSD/cPD, and the \u0026ge;\u0026thinsp;30% to \u0026ge;\u0026thinsp;60% range for cSD/cCR. However, at these expanded cut-offs outcome associations worsened considerably (Fig. S11c). Thus, despite ample confidence intervals, default cut-offs may be an obligate choice for outcome prediction, at least in the GIM21 setting. Altogether, the preliminary analysis conducted herein clearly points to the fine tuning of cSD cut-offs as to a major investigation goal in future cRECIST studies.\u003c/p\u003e\u003cp\u003eWhichever the cut-off, whether default or expanded, objective tumor and ctDNA responses strikingly diverged (OR from cOR and best OR from best cOR). This was due to a far more marked and dynamic response to T-DM1 treatment of ctDNA than tumor size, best exemplified by the considerable phenotypic enrichment in cPD-positive GIM21 patients (50% approximately) experiencing SD as their best clinical response while undergoing a measurable (cPR/cCR) ctDNA drop (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). In turn, OR/cOR divergence led to poor correlation between best cOR (deepest ctDNA drop) and outcome (PFS and OS, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed and e). However, when ctDNA amounts were disregarded and a time-dependent variable was derived (cPFS, the time elapsed between T\u003csub\u003e0\u003c/sub\u003e and cPD), correlations became apparent with best OR (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec) and hence, although indirectly, with PFS (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea).\u003c/p\u003e\u003cp\u003eTo our knowledge, this is the first demonstration that a discontinuous ctDNA/cRECIST variable can assist in monitoring response to treatment. In this respect, cPFS performance appears to be similar to other conventional continuous variables in many advanced settings and tumors (see for instance [\u003cspan additionalcitationids=\"CR10 CR11 CR12 CR13 CR14 CR15 CR16 CR17 CR18\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]). This is remarkable, because continuous metrics purely measuring ctDNA (both absolute amounts and relative changes) were inferior to cPFS when applied to the GIM21 dataset (Fig. S10). Whereas the reasons for this specific discrepancy with previous studies remain unclear at present, it is of note that cPFS, although primarily a time-dependent variable, implicitly incorporates cPD-compliant quantitative ctDNA changes. Therefore, a \u0026lsquo;hybrid\u0026rsquo; computation of time and ctDNA amounts, as in cPFS, may be central to cRECIST implementation.\u003c/p\u003e\u003cp\u003eA key observation of the GIM21 study is that cPFS and conventional continuous ctDNA variables have a common limitation: they are accurate enough to classify broad patient subsets, but remain crude predictors of individual outcomes, for which no metrics have so far been described to the best of our knowledge. While looking for such metrics, it was noted that ctDNA undergoes successive expansion and contraction cycles, termed ctDNA waving, particularly evident post-cPD and in long-term responders. Alone, cRECIST series of disconnected paired values were intrinsically inapplicable to track ctDNA waving, and required integration with ctDNA trends (\u003cem\u003eTr\u003c/em\u003e) measured in consecutive overlapping triplets of time points (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Empirically, an ongoing \u003cem\u003eTr\u003c/em\u003e increase was associated with an immediate risk of progression. More interestingly, the duration of the first \u003cem\u003eTr\u003c/em\u003e drop, as long as occurring post-cPD (Fig. S12), was found to be informative, since it predicted individual outcomes much better than cPFS alone (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.30 vs 0.76, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eh vs Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec). The simplest interpretation of ctDNA waving is that T-DM1 retains marginal (and decreasing) efficacy during step-wise acquisition of pharmacological resistance, and \u003cem\u003eTr\u003c/em\u003e reflects average tumor fitness throughout.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe model presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e and described in the legend highlights OR/cOR divergence, integrates cRECIST into RECIST 1.1, and proposes that ctDNA-assisted outcome prediction strategies could take ctDNA waving into account. As to possible mechanisms underlying ctDNA waving, a relevant observation has been recently made in the context of the same GIM21 study described herein. Following adaptive HER2 polypeptide loss under anti-HER2 therapeutic pressure, during T-DM1 treatment HER2 polypeptides were regained in tumor tissues and blood (sHER2) from a subset of patients with a favorable outcome [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Late (e.g. post-cPD) residual responsiveness and waving tumor/drug equilibria may potentially apply to other targeted agents, tumors, settings, and companion analytes.\u003c/p\u003e\u003cp\u003eThe GIM21 study design and protocols suffer from several limitations. These include a deliberately narrow focus on a single indication, low numerosity, failure to detect intra-cranial progression, and core-study inclusion of only 27 cRECIST-evaluable (cPD-positive) patients, although this selection did not apparently bias for outcome (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb and c). In addition, it may retrospectively be argued that real-time testing would have missed new target ctDNAs appearing \u003cem\u003ede novo\u003c/em\u003e at T\u003csub\u003ep\u003c/sub\u003e. These account for 18.1% of all ctDNAs (Fig. \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e). Also, a more intensive post-cPD blood drawing could have been very useful to systematically measure the first post-cPD \u003cem\u003eTr\u003c/em\u003e drop, that herein was measurable in 15 patients only. It is acknowledged that expanding cRECIST application beyond the present proof-of-concept will also require technical solutions capturing substantial numbers of target ctDNAs. Despite limitations, GIM21 is to our knowledge the first prospective study exploring the biological and clinical implications of cRECIST, providing initial evidence for irreducible conceptual differences between anatomical and ctDNA disease descriptions, and proposing practical computational approaches to personalize PD prediction.\u003c/p\u003e"},{"header":"CONCLUSIONS","content":"\u003cp\u003eTo conclude, three main practical cRECIST implications may be foreseen: (a) in early drug development, cPFS and \u003cem\u003eTr\u003c/em\u003e may represent early-on proxies of PFS and therapeutic efficacy; (b) in ctDNA-guided clinical trials, specific cPFS and/or \u003cem\u003eTr\u003c/em\u003e values may be predefined and, when prospectively applied, they may adaptively randomize the intention-to-treat population for either treatment maintenance or discontinuation/switch; and (c) in clinical practice, cPFS and \u003cem\u003eTr\u003c/em\u003e may flag individual patients for intensified post-cPD CT scan surveillance, and at the same time prevent premature withdrawal of an effective treatment in long-term responders displaying post-cPD ctDNA response/waving.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eResponse Evaluation Criteria In Solid Tumors version 1.1, RECIST 1.1; circulating tumor DNA, ctDNA; ctDNA-RECIST, cRECIST); Trastuzumab-emtansine, T-DM1; Objective Responses, clinical and ctDNA, ORs and cORs; Progressive Disease, PD, Stable Disease, SD; Partial Response, PR; Complete Response, CR; ctDNA-PD, cPD; ctDNA Stable Disease, cSD; ctDNA Partial Response, cPR; ctDNA Complete Response, cCR; Next Generation Sequencing, NGS; digital PCR, dPCR; Positron Emission Tomography (PET) Response Criteria for Solid Tumors, PERCIST; Immune Response Criteria for Solid Tumors, iRECIST/imRECIST; Gruppo Italiano Mammella, GIM; Antibody-Drug Conjugate, ADC; ImmunoHistoChemistry, IHC; Computerized Tomography, CT; Progression-free Survival, PFS; Overall Survival, OS; ctDNA-PFS, cPFS; Lead Time, LT.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was approved by the Fondazione Bietti Ethical Review Board (RS-857/16).\u0026nbsp;Written informed consent was obtained from all patients before study entry.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll raw data (total, and organized by figures) of ctDNA measurements are provided in an annotated spreadsheet as Table S2.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA.F.: consultant or advisor for Roche, Lilly, Novartis, AstraZeneca, Pfizer, Seagen, Gilead, MSD, Menarini Stemline, Dompè Biotech; speaker honoraria from Astra Zeneca, Roche, Lilly, Novartis, Gilead, Pfizer, Daiichi Sankyo Exact Sciences; grant support (to the Institution) by Astra Zeneca, Roche. GA: personal fees from Novartis, Lilly; grants and personal fees from Roche; grants, personal fees, and nonfinancial support from Pfizer; grants, personal fees, and nonfinancial support from AstraZeneca; and personal fees from Daichi. E.B.: grants or contracts from Astra-Zeneca, Roche; honoraria for lectures from Merck-Sharp \u0026amp; Dome, Astra-Zeneca, Pfizer, Eli-Lilly, Bristol-Myers Squibb, Novartis, Takeda and Roche; member of Data Safety Monitoring Board or Advisory Board of Merck-Sharp \u0026amp; Dome, Pfizer, Novartis, Bristol-Myers Squibb, Astra-Zeneca, Roche, Amgen and Celltrion. F.C.: consultant for Exact Science and AstraZeneca.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWork performed with the unconditional support by Roche Pharmaceutical. Funding is acknowledged by the Italian Minister of Health, Ricerca Corrente 2022 (AF), Associazione Italiana per la Ricerca sul Cancro (AIRC, IG No. IG20583), Università Cattolica del Sacro Cuore (UCSC-project D1), and Ministero della Salute Ricerca Corrente 2024-2025 (EB). The funders had no role in the design of the study, collection, analysis, interpretation of the data and writing of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors’ contribution\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA.F. and P.G. conceptualized the study and wrote the manuscript. E.G. and E.R. generated ctDNA data and elaborated all data. A.F. G.A. G.F., C.O., A.Z., A.B., E.B., S.G., L.C., and I.P. recruited GIM21 patients and generated/reviewed clinical data. M.A. generated ctDNA data. C.M. biobanked clinical specimens. F.C. chaired and G.S. co-chaired the study. D.G. was in charge of biostatistics. All authors helped to write/review, and approved the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eP.G. is grateful to the Colleagues of the European Liquid Biopsy Society (ELBS) for enlightening discussions.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIn memoriam\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis paper is dedicated to the memory of our dear colleague Giovanni Scambia, who passed away while this work was being finalized.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eEisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, Dancey J, Arbuck S, Gwyther S, Mooney M, Rubinstein L, Shankar L, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer. 2009;45:228\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWahl RL, Jacene H, Kasamon Y, Lodge MA. From RECIST to PERCIST: Evolving Considerations for PET response criteria in solid tumors. J Nucl Med. 2009;50(Suppl 1):S122\u0026ndash;50.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHodi FS, Ballinger M, Lyons B, Soria JC, Nishino M, Tabernero J, Powles T, Smith D, Hoos A, McKenna C, Beyer U, Rhee I et al. Immune-Modified Response Evaluation Criteria In Solid Tumors (imRECIST): Refining Guidelines to Assess the Clinical Benefit of Cancer Immunotherapy. J Clin Oncol. 2018:JCO2017751644.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYounes A, Hilden P, Coiffier B, Hagenbeek A, Salles G, Wilson W, Seymour JF, Kelly K, Gribben J, Pfreunschuh M, Morschhauser F, Schoder H, et al. International Working Group consensus response evaluation criteria in lymphoma (RECIL 2017). Ann Oncol. 2017;28:1436\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLin NU, Lee EQ, Aoyama H, Barani IJ, Barboriak DP, Baumert BG, Bendszus M, Brown PD, Camidge DR, Chang SM, Dancey J, de Vries EG, et al. Response assessment criteria for brain metastases: proposal from the RANO group. Lancet Oncol. 2015;16:e270\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJakobsen AKM, Spindler KG. ctDNA-Response evaluation criteria in solid tumors - a new measure in medical oncology. Eur J Cancer. 2023;180:180\u0026ndash;83.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCohen SA, Liu MC, Aleshin A. Practical recommendations for using ctDNA in clinical decision making. Nature. 2023;619:259\u0026ndash;68.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGouda MA, Janku F, Wahida A, Buschhorn L, Schneeweiss A, Karim NA, Perez DD, Del Re M, Russo A, Curigliano G, Rolfo C, Subbiah V. Liquid Biopsy Response Evaluation Criteria in Solid Tumors (LB-RECIST). Ann Oncol. 2024;35:267\u0026ndash;75.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eO'Leary B, Hrebien S, Morden JP, Beaney M, Fribbens C, Huang X, Liu Y, Bartlett CH, Koehler M, Cristofanilli M, Garcia-Murillas I, Bliss JM et al. Early circulating tumor DNA dynamics and clonal selection with palbociclib and fulvestrant for breast cancer. Nat Commun 2018;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eParikh AR, Mojtahed A, Schneider JL, Kanter K, Van Seventer EE, Fetter IJ, Thabet A, Fish MG, Teshome B, Fosbenner K, Nadres B, Shahzade HA, et al. Serial ctDNA Monitoring to Predict Response to Systemic Therapy in Metastatic Gastrointestinal Cancers. Clin Cancer Res. 2020;26:1877\u0026ndash;85.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang WY, Feng LF, Meng X, Chen R, Xu WH, Hou J, Xu T, Zhang L. Liquid biopsy in head and neck squamous cell carcinoma: circulating tumor cells, circulating tumor DNA, and exosomes. Expert Rev Mol Diagn. 2020;20:1213\u0026ndash;27.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDarrigues L, Pierga JY, Bernard-Tessier A, Bi\u0026egrave;che I, Silveira AB, Michel M, Loirat D, Cottu P, Cabel L, Dubot C, Geiss R, Ricci F et al. Circulating tumor DNA as a dynamic biomarker of response to palbociclib and fulvestrant in metastatic breast cancer patients. Breast Cancer Res 2021;23.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJakobsen A, Andersen RF, Hansen TF, Jensen LH, Faaborg L, Steffensen KD, Thomsen CB, Wen SWC. Early ctDNA response to chemotherapy. A potential surrogate marker for overall survival. Eur J Cancer. 2021;149:128\u0026ndash;33.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOlga Mart\u0026iacute;nez-S\u0026aacute;ez TP, Fara Bras\u0026oacute;-Maristany N, Chic B, Gonz\u0026aacute;lez-Farr\u0026eacute; E, Sanfeliu. Adela Rodr\u0026iacute;guez, D\u0026eacute;bora Mart\u0026iacute;nez, Patricia Galv\u0026aacute;n, Anna Bel\u0026eacute;n Rodr\u0026iacute;guez, Francesco Schettini, Benedetta Conte, Maria Vidal, Barbara Adamo, Antoni Mart\u0026iacute;nez, Montserrat Mu\u0026ntilde;oz, Reinaldo Moreno, Patricia Villagrasa, Fernando Salvador, Eva M Ciruelos, Iris Faull, Justin I Odegaard, Aleix Prat. Circulating tumor DNA dynamics in advanced breast cancer treated with CDK4/6 inhibition and endocrine therapy. npj Breast Cancer 2021.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eThompson JC, Carpenter EL, Silva BA, Rosenstein J, Chien AL, Quinn K, Espenschied CR, Mak A, Kiedrowski LA, Lefterova M, Nagy RJ, Katz SI, et al. Serial Monitoring of Circulating Tumor DNA by Next-Generation Gene Sequencing as a Biomarker of Response and Survival in Patients With Advanced NSCLC Receiving Pembrolizumab-Based Therapy. Jco Precision Oncol. 2021;5:510\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVega DM, Nishimura KK, Zariffa N, Thompson JC, Hoering A, Cilento V, Rosenthal A, Anagnostou V, Baden J, Beaver JA, Phd AACM, Chudova D et al. Changes in Circulating Tumor DNA Reflect Clinical Benefit Across Multiple Studies of Patients With Non-Small-Cell Lung Cancer Treated With Immune Checkpoint Inhibitors. JCO Precision Oncology 2022;6.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAnagnostou V, Ho CRY, Nicholas G, Juergens RA, Sacher A, Fung AS, Wheatley-Price P, Laurie SA, Levy B, Brahmer JR, Balan A, Niknafs N, et al. ctDNA response after pembrolizumab in non-small cell lung cancer: phase 2 adaptive trial results. Nat Med. 2023;29:2559\u0026ndash;.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAssaf ZJF, Zou W, Fine AD, Socinski MA, Young A, Lipson D, Freidin JF, Kennedy M, Polisecki E, Nishio M, Fabrizio D, Oxnard GR et al. A longitudinal circulating tumor DNA-based model associated with survival in metastatic non-small-cell lung cancer. Nat Med. 2023;29.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eThompson JC, Scholes DG, Carpenter EL, Aggarwal C. Molecular response assessment using circulating tumor DNA (ctDNA) in advanced solid tumors. Br J Cancer. 2023;129:1893\u0026ndash;902.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVerma S, Miles D, Gianni L, Krop IE, Welslau M, Baselga J, Pegram M, Oh DY, Dieras V, Guardino E, Fang L, Lu MW, et al. Trastuzumab emtansine for HER2-positive advanced breast cancer. N Engl J Med. 2012;367:1783\u0026ndash;91.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHurvitz SA, Hegg R, Chung WP, Im SA, Jacot W, Ganju V, Chiu JWY, Xu B, Hamilton E, Madhusudan S, Iwata H, Altintas S, et al. Trastuzumab deruxtecan versus trastuzumab emtansine in patients with HER2-positive metastatic breast cancer: updated results from DESTINY-Breast03, a randomised, open-label, phase 3 trial. Lancet. 2023;401:105\u0026ndash;17.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAllegretti M, Fabi A, Giordani E, Ercolani C, Romania P, Nistic\u0026ograve; C, Gasparro S, Barberi V, Ciolina M, Pescarmona E, Giannarelli D, Ciliberto G et al. Liquid biopsy identifies actionable dynamic predictors of resistance to Trastuzumab Emtansine (T-DM1) in advanced HER2-positive breast cancer. Mol Cancer 2021;20.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDe Mattos-Arruda L. Liquid biopsy for HER2-positive breast cancer brain metastasis: the role of the cerebrospinal fluid. ESMO Open. 2017;2:e000270.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAbbosh C, Birkbak NJ, Wilson GA, Jamal-Hanjani M, Constantin T, Salari R, Le Quesne J, Moore DA, Veeriah S, Rosenthal R, Marafioti T, Kirkizlar E, et al. Phylogenetic ctDNA analysis depicts early-stage lung cancer evolution. Nature. 2017;545:446\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGiordani E, Allegretti M, Sinibaldi A, Michelotti F, Ferretti G, Ricciardi E, Ziccheddu G, Valenti F, Di Martino S, Ercolani C, Giannarelli D, Arpino G et al. Monitoring changing patterns in HER2 addiction by liquid biopsy in advanced breast cancer patients. J Exp Clin Cancer Res. 2024;43.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-experimental-and-clinical-cancer-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jecc","sideBox":"Learn more about [Journal of Experimental \u0026 Clinical Cancer Research](http://jeccr.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/jecc/default.aspx","title":"Journal of Experimental \u0026 Clinical Cancer Research","twitterHandle":"@OncoBioMed","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"ctDNA, RECIST 1.1, ctDNA-RECIST, HER2-positive Breast Cancer, T-DM1, objective response evaluation.","lastPublishedDoi":"10.21203/rs.3.rs-7101376/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7101376/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eResponse Evaluation Criteria In Solid Tumors (RECIST 1.1) and circulating tumor DNA (ctDNA) recapitulate and anticipate response to treatment, respectively. However, no ctDNA-RECIST (cRECIST) guidelines have been formally implemented in clinical practice so far.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eFor first proof-of-principle, HER2-positive metastatic breast cancer patients (n=50) were enrolled in the multi-center prospective GIM21 study to receive Trastuzumab-emtansine (T-DM1), and were monitored for Objective Responses, e.g. ORs (progressive disease, stable disease, partial response, complete response; PD/SD/PR/CR) vs ctDNA-ORs (cORs: cPD/cSD/cPR/cCR). Standard OR cut-offs (SD/PD≥20% and SD/PR≤30%) were applied to cORs by default, or tentatively relaxed.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eWhichever the cut-off, bespoke NGS/dPCR (78 ctDNAs; 466 measurements) and CT scans (113 tumor lesions) revealed much deeper cORs than ORs, leading to RECIST 1.1/cRECIST divergence in 27 cPD-positive patients. Yet, OR/cOR integration remained feasible at default/common cut-offs, as shown by correlation between fast cPD and poor OR/PFS. Although a satisfactory coarse patient classifier, cPD was unfortunately confounded by patient-specific, post-cPD ctDNA increases/decreases (ctDNA waving). Then, to personalize outcome prediction, two-point cRECIST comparisons (response vs baseline/nadir) were replaced by a novel non-cRECIST variable measuring three-point \u003cem\u003eTr\u003c/em\u003eends (\u003cem\u003eTr\u003c/em\u003e). Remarkably, the duration of the first post-cPD \u003cem\u003eTr\u003c/em\u003e drop correlated with the timing of PD in 15/15 evaluable patients (R\u003csup\u003e2\u003c/sup\u003e=0.76).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions: \u003c/strong\u003eCombined, cRECIST (cPD) and \u003cem\u003eTr\u003c/em\u003e may help to: (a) predict treatment efficacy during early drug development, (b) randomize for timely treatment switch in clinical trials, and (c) prevent premature treatment withdrawal in long-responders undergoing ctDNA waving. Future prospective studies are warranted for cRECIST/RECIST 1.1 integration/personalization in different tumors/settings.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTrial registration: \u003c/strong\u003eNCT05735392.\u003c/p\u003e","manuscriptTitle":"Circulating tumor DNA and Response Evaluation Criteria In Solid Tumors: ctDNA-RECIST proof-of-concept in HER2-positive metastatic breast cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-18 14:42:55","doi":"10.21203/rs.3.rs-7101376/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-08-19T19:54:58+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-19T19:26:45+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-04T13:27:39+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"141478798240223713607487235267748908659","date":"2025-07-25T11:44:16+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"174541602408193648566989733875230633658","date":"2025-07-25T08:09:58+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-15T07:59:13+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-15T06:06:18+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-15T06:05:38+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Experimental \u0026 Clinical Cancer Research","date":"2025-07-11T11:39:11+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-experimental-and-clinical-cancer-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jecc","sideBox":"Learn more about [Journal of Experimental \u0026 Clinical Cancer Research](http://jeccr.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/jecc/default.aspx","title":"Journal of Experimental \u0026 Clinical Cancer Research","twitterHandle":"@OncoBioMed","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"08fed848-f8b7-4585-9d55-9c76b93607e6","owner":[],"postedDate":"July 18th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-01-26T16:04:06+00:00","versionOfRecord":{"articleIdentity":"rs-7101376","link":"https://doi.org/10.1186/s13046-025-03605-2","journal":{"identity":"journal-of-experimental-and-clinical-cancer-research","isVorOnly":false,"title":"Journal of Experimental \u0026 Clinical Cancer Research"},"publishedOn":"2026-01-20 15:59:09","publishedOnDateReadable":"January 20th, 2026"},"versionCreatedAt":"2025-07-18 14:42:55","video":"","vorDoi":"10.1186/s13046-025-03605-2","vorDoiUrl":"https://doi.org/10.1186/s13046-025-03605-2","workflowStages":[]},"version":"v1","identity":"rs-7101376","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7101376","identity":"rs-7101376","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}