De-escalated Neoadjuvant Chemotherapy in Early Triple-Negative Breast Cancer (TNBC): Impact of Molecular Markers and Final Survival Analysis of the WSG-ADAPT-TN Trial

Although optimal treatment in early triple-negative breast cancer (TNBC) remains unclear, de-escalated chemotherapy appears to be an option in selected patients within this aggressive subtype. Previous studies have identified several pro-immune factors as prognostic markers in TNBC, but their predictive impact regarding different chemotherapy strategies is still controversial.ADAPT-TN is a randomized neoadjuvant multicenter phase II trial in early patients with TNBC (n = 336) who were randomized to 12 weeks of nab-paclitaxel 125 mg/m2 + gemcitabine or carboplatin d 1,8 q3w. Omission of further (neo-) adjuvant chemotherapy was allowed only in patients with pathological complete response [pCR, primary endpoint (ypT0/is, ypN0)]. Secondary invasive/distant disease-free and overall survival (i/dDFS, OS) and translational research objectives included quantification of a predictive impact of markers regarding selection for chemotherapy de-escalation, measured by gene expression of 119 genes (including PAM50 subtype) by nCounter platform and stromal tumor-infiltrating lymphocytes (sTIL).After 60 months of median follow-up, 12-week-pCR was favorably associated (HR, 0.24; P = 0.001) with 5y-iDFS of 90.6% versus 62.8%. No survival advantage of carboplatin use was observed, despite a higher pCR rate [HR, 1.04; 95% confidence interval (CI), 0.68-1.59]. Additional anthracycline-containing chemotherapy was not associated with a significant iDFS advantage in pCR patients (HR, 1.29; 95% CI, 0.41-4.02). Beyond pCR rate, nodal status and high sTILs were independently associated with better iDFS, dDFS, and OS by multivariable analysis.Short de-escalated neoadjuvant taxane/platinum-based combination therapy appears to be a promising strategy in early TNBC for using pCR rate as an early decision point for further therapy (de-) escalation together with node-negative status and high sTILs. See related commentary by Sharma, p. 4840.