In this cohort of patients with TNBC, clearance of ctDNA at mid-NAC timepoint was significantly associated with pCR. Additionally, detection of ctDNA at this timepoint was associated with reduced overall survival and recurrence-free survival compared to patients with negative ctDNA. Our results demonstrate the utility of ctDNA measurement early in treatment course among patients with TNBC, an aggressive subtype with a high risk of relapse. Importantly, surveillance of ctDNA at the mid-treatment timepoint creates a window of opportunity to transition to alternative therapies in a subset of patients with poor response to traditional chemotherapy.
Our findings here are in line with current studies in the literature (20, 24–26). Cailleux et al. collected serial plasma samples from 44 patients with early-stage breast cancer and found that detection of ctDNA at baseline prior to surgery and at last follow-up was associated with shorter event-free survival (27). While their baseline ctDNA detection rate was considerably lower than that in our study (58% vs 90%), this discrepancy may be explained by the higher percentage of HR-positive/HER2-positive disease in their study cohort. Prior studies have demonstrated differences in ctDNA levels by receptor subtypes, with triple negative disease exhibiting relatively higher ctDNA levels and HR-positive/HER2-negative disease with lower ctDNA levels (28). Likewise, Garcia-Murillas et al. assessed ctDNA levels in a prospective cohort of 55 patients with early breast cancer using digital droplet PCR (20). Their results demonstrated that detection of ctDNA in serial postsurgical samples was associated with early relapse. However, they also found that baseline ctDNA detection and abundance was not associated with disease-free survival (DFS) or early relapse. In one of the largest studies focused on triple negative disease, secondary analysis of the BRE12-158 randomized clinical trial demonstrated that presence of ctDNA in 196 patients with early-stage TNBC after neoadjuvant chemotherapy and surgical resection was significantly associated with worse distant disease-free survival, DFS, and OS (29). While the combination of ctDNA and CTC positivity was prognostic, the authors found that presence of CTCs alone was not. Similarly, in our cohort, there was no difference in OS or DFS based on CTC positivity at baseline or mid-treatment.
Collectively, studies in the literature have brought forth convincing evidence demonstrating the prognostic value of ctDNA in TNBC (20, 24, 25, 29). However, the predictive value of liquid biopsy remains a field under active investigation (22, 23). In our study, assessment of serial ctDNA and CTC status early in treatment course prior to surgical resection allows for evaluation of its use as a blood-based predictive biomarker. Of the 19 patients who achieved ctDNA clearance at mid-treatment, almost all had evidence of partial or complete response on US post-NAC and more than half achieved pCR. Importantly, none of the patients with detectable mid-treatment ctDNA achieved pCR (15/15, 100% specificity). However, of the 27 patients without recurrence, only three remained ctDNA positive at mid-treatment (24/27, 89% specificity). Since these patients would continue with the remaining NAC regimen, and in the event of unfavorable response at surgery would likely receive adjuvant therapy, ctDNA-positivity at mid-treatment may indicate a high probability of recurrence that could be mitigated with targeted or experimental therapies. As such, the clinical utility in mid-treatment ctDNA may lie in its ability to rule out residual disease and possible recurrence. For the 30% of patients in our study cohort who did not achieve ctDNA clearance at mid-NAC timepoint, transitioning to second-line or investigative therapies could maximize their oncologic outcomes. Importantly, these patients may be classified as poor responders, whereby serial monitoring of ctDNA provides lead time to modify treatments in the neoadjuvant setting.
It is worth noting that our results emphasize the importance of serial collection of samples throughout the disease course. Whether ctDNA and CTCs are analyzed in the neoadjuvant or adjuvant setting, a single timepoint at pre-treatment baseline or post-surgical surveillance offers limited information. The predictive value of liquid biopsy could be maximized by monitoring dynamic changes over time as patients undergo their treatment course. In a subset analysis of the I-SPY2 trial, the authors found that within the TNBC cohort, early clearance of ctDNA at three-weeks after NAC initiation was a significant predictor of pCR and residual cancer burden (RCB), compared to clearance at the 12-week mark (23). These results suggest that monitoring ctDNA at even earlier timepoints during NAC can help identify an optimal window with the maximal amount of lead time for intervention. Furthermore, it is interesting in our study that CTCs failed to offer predictive or prognostic value in this cohort of TNBC patients. Several larger studies including breast cancer of all subtypes have previously demonstrated that presence of CTCs was an independent prognostic factor for overall survival (30, 31). However, in this small subset of early-stage triple negative patients, the predictive and prognostic value of CTCs failed to reach significance. This finding may be due to smaller cohort size or the fact that majority of our patients had early stage T1-2 (78%) and N0 disease (56.8%). Previous studies have similarly demonstrated that ctDNA might have greater sensitivity and correlation with changes in tumor burden compared to CTCs (32). From a logistical standpoint, monitoring treatment response and disease burden may be more feasible with ctDNA as sample collection of plasma is easily attainable and does not require isolation of a specific cell population. Nevertheless, more work is needed to further characterize use of ctDNA and CTCs both alone and together as synergistic biomarkers throughout the neoadjuvant period in future prospective studies.
As with any study, there are limitations that warrant further discussion. For one, this study was conducted at a single-institution and thus has limited generalizability. However, few studies have evaluated the role of serial liquid biopsy in the neoadjuvant setting in patients with TNBC and thus, we believe our findings hold important implications for patients with this aggressive tumor subtype. Second, as this study was designed as an observational trial, we were unable to control for impact of confounders, such as variations in systemic treatment regimens, particularly those deemed chemo-insensitive and subsequently received experimental targeted therapies, on ctDNA and CTC clearance. Several randomized prospective trials currently in progress will provide more clarity on clinical utility of ctDNA in disease monitoring and tailoring of treatment regimens (33). Third, given that our follow-up period was a median of 44 months, there may be missed future relapses beyond this time period. However, studies indicate that risk of recurrence in TNBC peaks at approximately three-years and drops precipitously thereafter (3). Lastly, our data had occasional missing timepoints and as such, we were only able to perform analysis of patients with matched samples. Future studies with a larger cohort and more complete sample collections across timepoints will help further validate our work.
This study demonstrates that serial monitoring of ctDNA in the neoadjuvant setting might be useful as both a predictive and prognostic marker in patients with TNBC. ctDNA serves as a surrogatefor residual molecular disease and provides lead time for clinicians to personalize treatment regimens based on patient response. In particular, patients who are at high risk for disease progression and/or poor responders to standard systemic therapies may benefit the most from use of liquid biopsy in the neoadjuvant setting.