Circulating tumor DNA in breast cancer: updates from SABCS 2025

Abstract

Circulating tumor DNA (ctDNA) has emerged as a transformative biomarker in breast cancer, enabling sensitive assessment of tumor burden, molecular residual disease, and treatment resistance. Data presented at the 2025 San Antonio Breast Cancer Symposium (SABCS) demonstrate that ctDNA has moved beyond prognostication to guide treatment in advanced hormone receptor positive disease, most notably in the phase III SERENA-6 trial, where ctDNA-detected ESR1 mutations prospectively triggered endocrine therapy switch and significantly improved clinical outcomes. In early-stage breast cancer, converging evidence across subtypes shows that ctDNA-defined minimal residual disease (MRD) robustly identifies patients at high risk of recurrence. Collectively, these findings position ctDNA as a potential biomarker for dynamic treatment adaptation, supporting the next phase of precision oncology focused on MRD-guided escalation, de-escalation, and therapeutic interception across the breast cancer continuum.

To the Editor,

Circulating tumor DNA (ctDNA) has emerged as a transformative biomarker in breast cancer, enabling sensitive and dynamic assessment of tumor burden, clonal evolution, and treatment resistance across the disease continuum. Data presented at the 2025 San Antonio Breast Cancer Symposium (SABCS) indicate that advances in ctDNA assays sensitivity, together with results from large perspective clinical trials, have moved liquid biopsy from technical feasibility to clinical utility in breast cancer.

Has ctDNA moved beyond prognostication to guide treatment in advanced breast cancer?

Despite growing interest in ctDNA-guided strategies and emergence of prospective biomarker-driven studies enabled by highly sensitive assays, its clinical role has historically been largely confined to prognostic value, however, emerging data are beginning to suggest its predictive utility. In advanced breast cancer (ABC), the most definitive evidence supporting ctDNA as a treatment-guiding tool comes from the phase III SERENA-6 trial (Table 1). In this study, patients with HR+/HER2- ABC receiving first-line aromatase inhibitor (AI) plus CDK4/6 inhibitor underwent serial ctDNA monitoring every 2–3 months using a commercial hybrid-capture ctDNA assay with ~95% sensitivity and tumor fraction detection as low as 0.05%. Detection of ESR1 mutations triggered a switch from AI to camizestrant significantly prolonged progression-free (PFS) survival (16.6 vs. 9.2 months; HR 0.46) and time to subsequent therapy, and rapidly suppressed ESR1 mutant allele frequency [1]. These results demonstrate that serial ctDNA- monitoring can meaningfully alter disease trajectory and represent a paradigm shift toward real-time, biomarker-directed treatment adaptation.

Table 1.

Clinical applications of ctDNA in advanced breast cancer (ABC)

Trial	Study design	Population	ctDNA assay and timing	Time points	Key findings	Strengths	Limitations
SERENA-6 (RF7-03 Bidard)	Prospective	HR+/HER2 − ABC on first-line AI + CDK4/6i with emergent ESR1 mutations (n = 315)	Plasma ctDNA testing for ESR1 mutations (Guardant360 CDx);	Baseline and serial ctDNA monitoring	ctDNA-detected ESR1 mutation emergence guided early switch from AI to camizestrant significantly improving PFS (16.6 vs. 9.2 mon; HR 0.46), PFS2, time to subsequent therapy, and chemotherapy/ADC-free survival, with rapid suppression of ESR1m allele frequency, suggesting a predictive value of ESR1 mutation in response to switching treatment to camizestrant	Innovative biomarker-driven design, with therapy switched based on ctDNA-detected ESR1 mutations and subsequent randomized allocation, provide predictive value of ESR1 for treatment switch and response	The results may be subject to lead-time bias, raising uncertainty as to whether the observed improvements in PFS and PFS2 are at least partially inflated due to earlier intervention rather than true prolongation of disease control
CAPItello-291 exploratory ctDNA analyses (RF7-05 Turner)	Prospective	HR+/HER2 − ABC, including patients with PIK3CA/AKT1/PTEN alterations (n = 279)	Plasma ctDNA using Guardant Infinity™ (large genomic + epigenomic panel with methylation-based tumor fraction)	Baseline	ctDNA detected actionable PIK3CA/AKT1/PTEN alterations in ~ 10% of patients not identified by tissue alone. Patients with ctDNA-detected alterations derived substantial PFS benefit from capivasertib (HR 0.43), consistent across ESR1m and wild-type subgroups, supporting ctDNA as a treatment-directing biomarker when tissue is unavailable or discordant. This study suggests a predictive value of PIK3CA/AKT/PTEN alteration in response to the addition of capivasertib to fulvestrant	A randomized controlled trial framework incorporating a noninvasive ctDNA assay detecting additional 10% patients without tissue mutation and supporting the benefit of capivasertib in patients with PIK3CA/AKT/PTEN alterations	ctDNA testing may miss alterations in the setting of low tumor shedding
INAVO120 (PD5-09 Turner)	Prospective	ER+/HER2 − endocrine-resistant ABC w/ PIK3CA mutation (1416 screened, n = 277 for analysis)	FoundationOne® Liquid CDx plasma ctDNA profiling with circulating tumor fraction assessment	Baseline	Amongst the 99.6% of patients with detected ctDNA mPIK3CA, the addition of inavolisib to fluvestrant and palbociclib improved PFS (HR 0.43) and OS (HR 0.67), with its efficacy consistent amongst patients with various molecular changes, including ESR1 and PTEN mutation status. In addition, both TP53 mutations and high tumor fractions have prognostic values associated with worse prognosis	Well-conducted exploratory biomarker analysis demonstrating benefits in coexisting ESR1 and PTEN mutation but negative prognostic values in TP53 mutation on and high TMB	Predictive biomarkers beyond PIK3CA are lacking
EMBER-3 (PD5-08 Bidard)	Prospective	ER+/HER2 − ABC (total n = 874) receiving SOC ET (n = 330), imlunestrant (n = 331), or imlunestrant + abemaciclib (n = 213)	Guardant360 plasma ctDNA (74-gene panel); on-treatment dynamics	Baseline and C2D1 (4 weeks)	Early ctDNA molecular response (≥ 50% VAF reduction) was associated with longer PFS across all arms, suggesting its prognostic value. Imlunestrant + abemaciclib achieved the greatest ctDNA decline, particularly in ESR1-mutant disease, though treatment assignment remained the dominant determinant of outcome	Well-designed phase 3 randomized study demonstrating that early ctDNA molecular response and greater ct DNA decline associated with better outcome	Approximately 27% of patients were not evaluable for ctDNA analysis, raising concerns about missing data, and the findings lack predictive utility to guide treatment selection
MONALEESA pooled ctDNA analysis (PD5-10 Razavi)	Retrospective	HR+/HER2 − ABC treated with 1st-line ribociclib + endocrine therapy (n = 666 for pooled long-term response (LTR), n = 516 for pooled very-long-term response analysis (VLTR)	Targeted NGS panel	Baseline	Patients achieving LTR (PFS > 4 yrs) or VLTR (PFS > 5 yrs) had significantly lower baseline ctDNA fraction, fewer TP53 alterations, and lower CCNE1 expression. Baseline ctDNA burden stratified likelihood of durable benefit but did not direct treatment changes. This analysis identified characteristics of clinical and molecular features that represent as prognostic markers in this patient population	Long follow-up period providing valuable insight prognostic value of ctDNA at baseline	The rationale for selecting 4 years as the cutoff for long-term response is unclear, and this outcome-based definition introduces potential survivor bias

Additional evidence comes from exploratory analyses of CAPItello-291 and INAVO120. Both studies utilizing broad ctDNA profiling to identify actionable PIK3CA/AKT1/PTEN alterations in patients missed by tissue testing alone with ctDNA defined subgroup derived substantial benefits [2, 3]. Notably, the liquid biopsy assay demonstrated high sensitivity: ~95% at ~ 0.4% variant allele frequency for short variants; ~0.37% for rearrangements. Together, these data support broad ctDNA profiling as a clinically actionable tool in ABC, particularly when tissue is unavailable, insufficient, or discordant.

By contrast, EMBER-3 and pooled MONALEESA analyses continue to position ctDNA primarily as a prognostic and response-monitoring biomarker, with early ctDNA reduction and lower baseline burden associated with improved treatment responses [4, 5].

Can ctDNA-defined MRD identify patients with high recurrence risk?

ctDNA assays can sensitively detect molecular relapses and predict recurrence timing; however their role in guiding treatment adaption remains investigational. In the neoadjuvant setting, the PHERGain study showed baseline ctDNA correlated with tumor burden in HER2 positive breast cancer (Table 2). Although it did not predict pathologic complete response (pCR), failure to clear ctDNA was associated with non-pCR and inferior 3-year invasive disease-free survival (HR 4.5) [6]. Similarly, in the NSABP B-59/GeparDouze MRD study in triple-negative breast cancer (TNBC), ctDNA positivity declined from 96% at baseline to 9% after neoadjuvant therapy. Post-surgical MRD positivity identified an approximately 30-fold higher risk of distant recurrence, whereas nearly all MRD-negative patients remained metastasis-free at three years [7].

Table 2.

Clinical applications of ctDNA in early-stage breast cancer

Trial	Study design	Population	ctDNA assay and timing	Time points	Key findings	Strengths	Limitations
Translational analysis of PHERGain neoadjuvant tailored treatment study (GS1-06 Cussac)	Prospective	Early-stage HER2 + BC (n = 288)	Tissue-free methylation-based ctDNA detection at various time points to assess treatment response (Guardant Reveal™)	Baseline, cycle 3 day 1 (C3D1), pre-surgery	Baseline ctDNA detected in 71% and correlated with stage/nodal status but not pCR. ctDNA clearance at C3D1 is associated with higher pCR (48.4% vs. 22.0%, p = 0.003); No patients who failed to clear their ctDNA pre-surgery achieved pCR. Failure to clear ctDNA pre-surgery predicted worse 3-yr iDFS (HR 5.1, p = 0.014), suggesting a prognostic role of ctDNA clearance in patient outcome	Reinforced ctDNA as a dynamic prognostic marker of neoadjuvant treatment response	The study remains observational rather than actionable in the absence of biomarker-guided treatment adaptation
Prospective sub study of the NSABP-B-59 (RF4-03 Balic)	Prospective	Early-stage TNBC receiving neoadjuvant chemotherapy ± atezolizumab (MRD cohort n = 212)	Tumor-informed WES-based ctDNA (bespoke panel, median ~ 165 variants)	Baseline (pre-neoadjuvant treatment (NAT)), post-NAT, post-surgery, 12 months, and 24 months	ctDNA positivity decreased from 96% at baseline to 9% post-NAT. Post-surgery MRD positivity was strongly associated with distant recurrence (78% vs. 5%), with a 30-fold higher risk of distant relapse (HR 30.3). 95% of MRD-negative patients remained metastasis-free at 3 years, re-emphasizing MRD positivity as strong prognostic biomarker	Serial ctDNA assessment, demonstrating a strong prognostic association between MRD positivity and recurrence	Small sample size, particularly the limited number of ctDNA-positive patients (n = 9) and recurrence events (n = 7), somewhat limits the robustness of the findings
PALLAS MRD analysis (RF3-04 Parsons)	Prospective	HR+/HER2- early breast cancer; US biomarker cohort n = 420 (1012 plasma samples)	Tumor-informed WGS ctDNA (Signatera) measurement at various time-points	C1D1, C6D1, end of treatment (EOT)	With a similar prevalence amongst timepoints, ctDNA was detected in 8% of patients at C1D1 and EOT. ctDNA positivity strongly predicted distant recurrence: HR 15.1 at C1D1 and HR 21.5 at EOT (both p < 0.0001). Five-year DRFI was ~ 28–32% in ctDNA + vs. > 93% in ctDNA− patients	Well-designed prospective biomarker study with a large sample size, aiming to look at both the prognostic and predictive utility of MRD	Final results are still pending
c-TRAK-TN trial, ctDNA in early-stage TNBC after finishing treatment (RF3-05 Cunningham)	Prospective with retrospective analysis	High-risk early TNBC; 159 patients, 1026 surveillance samples	Tissue-free methylation-based ctDNA (Guardant Reveal™); serial post-treatment surveillance	Serial every 3 months post-treatment surveillance	Tissue-free assay detected ctDNA with high accuracy and earlier than ddPCR (95.4% concordance), suggesting its prognostic value. ctDNA was detected in more patients and at earlier timepoints; 12-mo detection rate 29.0% vs. 23.7% with ddPCR. Median lead time from ctDNA detection to relapse was longer with tissue-free assay than ddPCR (7.8 vs. 5.8 months, p = 0.07)	Hybrid prospective-retrospective design demonstrates the strong prognostic value of ctDNA in high-risk TNBC	Clinical applicability remains limited by the absence of predictive utility and lack of biomarker-driven intervention
LEADER trial: Adjuvant CDK in MRD setting and measurement of ctDNA (PD5-01 Medford)	Prospective	ER+/HER2- early breast cancer with detectable ctDNA (n = 15) after definitive therapy	Tumor-informed ctDNA (Signatera) serial post-treatment monitoring	Serial surveillance and on-treatment monitoring after ribociclib initiation	ctDNA positivity during surveillance identified patients at high recurrence risk. Among ctDNA-positive patients treated with ribociclib, 66% showed ctDNA decrease and 33% achieved clearance by 3 months. Favorable ctDNA dynamics were associated with longer DRFS (16.5 vs. 5.3 months), and ctDNA clearance correlated with delayed recurrence of more than 11 months	Prospective biomarker-driven interception study highlights the prognostic significance of ctDNA and the feasibility of therapeutic intervention	Without a control arm, it is not possible to assess predictive utility or attribute observed outcomes to treatment effect
CATER study: adjuvant capecitabine in MRD patients (PS5-09-28 Elliot)	Prospective	ER+/HER2- stage I to III breast cancer with ctDNA + MRD after 24 months of AI +/- 12 months of CDKi (n = 13)	ctDNA + MRD (Pathlight, SAGA Dx), with criteria identified through CLAIRE study (high risk for recurrence)	Serial plasma for ctDNA and biobanking, week 16 ctDNA clearance	ctDNA-guided secondary adjuvant therapy with metronomic capecitabine is feasible and can lead to ctDNA clearance at week 16 in a subset of MRD-positive ER+/HER2 − early breast cancer patients, supporting the concept of biomarker-driven therapeutic interception	Prospective, bio-marker driven intervention with a tumor-informed assay and a clear primary endpoint of ctDNA clearance with the aim to transform post-operative management	Very small sample size limited generalizability. ctDNA clearance cannot be validated for survival outcomes

Additional SABCS 2025 abstracts (PS4-08-06, PS5-11, PS1-13-11, PS2-10-12, and PS5-09-25) further support the expanding role of ctDNA in minimal residual disease detection, assay optimization, and early relapse prediction. Because of space and reference limitations, these studies are acknowledged here but were not individually included in the table

In the adjuvant setting, the PALLAS MRD analysis showed that ctDNA positivity was infrequent in HR+/HER2- disease but it strongly predicted distant recurrence (HR 15.1), with 5-year distant recurrence-free (DFS) survival was 93% in ctDNA-negative versus ~ 30% in ctDNA-positive patients [8]. The c-TRAK TN study further demonstrated that a tissue-free, methylation-based ctDNA assay detected relapse earlier and identified more high-risk TNBC patients [9]. In LEADER, among ctDNA-positive HR+/HER2- patients (Signatera, ~ 94% overall sensitivity, 100% in breast cancer), early ctDNA reduction or clearance were associated with longer DFS [10]. Finally, the CATER study suggested that ctDNA clearance with adjuvant capecitabine may serve as an early surrogate of outcome [11].

Challenges and future directions

Despite substantial advances, several critical questions remain. While SERENA-6 trial established as a treatment-guiding biomarker in advanced disease, the optimal cfDNA platforms and testing strategies for different clinical contexts remain undefined. This includes uncertainty regarding the relative roles of tumor-informed MRD assays, tumor-agnostic approaches, and serial broad cfDNA profiling in both the MRD and advanced disease settings. In early-stage breast cancer, prospective trials demonstrating that MRD-guided intervention improves long-term clinical outcomes are still lacking. In parallel, heterogeneity in assay design, analytical performance, and payer acceptance continues to limit widespread adoption, emphasizing the necessity for assay harmonization, prospective validation, and cost-effective implementation to fully realize the promise of ctDNA-driven precision oncology.

Author contributions

IZ: Conceptualization; Writing original draft and table; Writing review & editing; Supervision. XZ: Writing original draft and table; Writing; review & editing. MS: Review & editing. NC: Review & editing. NK: Review & editing. NV: Review & editing.SG: Review & editing. CM: Review & editing. All authors read and approved the final manuscript.

Funding

The authors declare that no funding was received for this work.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

Cynthia Ma: consulting fees from Merck, Regor Therapeutics, AstraZeneca, Novartis, Stemline, Eli Lilly, Delphi Diagnostics, Biovica, FoRx Therapeutics, Tempus, Bayer, Pfizer, Puma, Pathlight, Genentech, Funding: Pfizer. Shridar Ganesan: consulting fees from Merck, EMD Serano, Eikon, Foghorn Rx, Lunit, ConstInt Rx, KayoThera. SG has received research funding from Tempus, Gandeeva Rx, M2GEN. SG’s spouse is an employee of Merck and has equity in Merck. Neil Vasan: Consulting fee from Apertor Therapeutics, Avenzo Therapeutics, Expert Connect, GLG, Guidepoint Global, Horizon CME, MJH Healthcare, Regor Therapeutics; Honoraria: Gilead, Susan G. Komen, University of Texas Health Sciences Center; Royalties: Wolters Kluwer; Travel: Gilead; Scientific Advisory Board: Meliora Therapeutics. Douglas Marks: Consultant/Advisory: Astrazeneca, Biotheranostics, Gilead, Lilly, OncLive/MJH, Seattle Genetics, Stemline/Menarini, Daiichi, Ideology, Sponsored Research: Agendia, Merck. Other authors: none.