Ultrasensitive circulating tumor DNA (ctDNA) detection for prognostication in triple-negative breast cancer (TNBC) post-neoadjuvant chemotherapy (NAC).

552 Background: NAC +/- immune checkpoint inhibitors is the standard of care for most early-stage TNBC patients. After NAC and breast surgery, adjuvant treatment decisions rely on pathological complete response (pCR) status, which does not inform on the presence of distant micrometastases. Blood-based assays for ctDNA enable non-invasive monitoring of residual disease levels at sensitivities down to 1 part-per-million (PPM). We report the prognostic value of ultrasensitive ctDNA detection during NAC in patients with TNBC. Methods: Early-stage TNBC patients treated with NAC in the SCANDARE prospective study were evaluated for ctDNA before, during, after NAC/pre-surgery and during post-surgical follow-up. Plasma ctDNA was profiled using NeXT Personal (Personalis), an ultrasensitive tumor-informed MRD assay that tracks up to 1,800 patient tumor-specific variants based on whole genome sequencing to attain sensitivity down to 1-3 PPM with >99.9% specificity. Results: Plasma ctDNA was analyzed for 279 samples from 84 TNBC patients (Stage I: 2%, II: 77%, III: 20%), of whom 35 (42%) achieved pCR after NAC. After a median follow-up of 53 months, 16 patients (19%) developed distant metastases, and 18 patients (21%) died. ctDNA was detected before NAC in all 82 patients with an available sample (median=3461 PPM, IQR: 1168-22078), with pretreatment levels in the ultrasensitive range (<100 PPM) in 12%. Most ctDNA detections during (51%) and post-NAC (55%) were <100 PPM. Patients with rates of early on-treatment ctDNA reduction faster than the median, or clearance, had significantly improved distant relapse-free interval (DRFI, log-rank P=7.7x10 -3 ) and overall survival (OS, log-rank P=7.9x10 -3 ). Patients with post-NAC, pre-surgery ctDNA clearance had significantly improved DRFI (log-rank P=3.5x10 -7 ) and OS (log-rank P=1.4x10 -11 ), and were enriched for pCR (59% vs 9%, OR=13.9, 95% CI [2.8,137.2], Fisher’s exact P=1.4x10 -4 ). Multivariable Cox models including pCR and ctDNA detection post-NAC, pre-surgery performed significantly better than models including only pCR for predicting DRFI (LRT P=8.0x10 -4 ) and OS (LRT P=6.6x10 -4 ). ctDNA detection significantly stratified survival outcomes among patients without pCR (DRFI HR=8.2, 95% CI [1.8,37.0], Cox P=6x10 -3 ), with 60% of ctDNA-positive patients developing distant metastases vs 10% of ctDNA-negative patients. During follow-up, all plasma samples from non-relapsing patients were ctDNA-negative. In relapsed patients, ctDNA was detected in 95% of samples collected at relapse or tumor progression. Conclusions: Ultrasensitive ctDNA detection informs on the outcome of early TNBC treated by NAC, independently of pCR status. These results, obtained with samples taken post-NAC but before-surgery warrant investigating the benefit of implementing ctDNA detection in an interventional setting.