Molecular and clinical insights of trastuzumab deruxtecan efficacy in advanced breast cancer (aBC).

Abstract

1040 Background: In aBC, trastuzumab deruxtecan (T-DXd) has demonstrated efficacy in HER2-positive, HER2-low, and HER2-ultralow disease subtypes. However, the impact of molecular markers on treatment outcomes requires further investigation, particularly in the context of HER2 status across biopsies, genetic alterations, and the tumor micro-environment (TME). Methods: We retrospectively analyzed 477 patients (pts) with aBC (DFCI = 369; Yale = 108) treated with T-DXd. HER2 immunohistochemistry (IHC) discordance was defined as a shift in HER2 status across the last two tumor samples prior to T-DXd (HER2 0 to 1, 2, or 3+ or vice versa). Concordance was defined as consistent HER2-0, HER2-low (1 or 2+) or HER2 3+ across samples. Outcomes included time-to-next treatment (TTNT) and overall survival (OS); multivariable Cox proportional hazards models were performed. Targeted tumor sequencing was conducted on 163 pts with aBC treated with T-DXd. TME analysis of 97 patients used machine learning on H&E slides to classify tumors as inflamed, desert, or altered. Results: Pts with discordant HER2 (n = 118, 25%) showed similar outcomes to those with concordant HER2-0 (n = 32) expression, both of which had significantly worse OS and TTNT compared to concordant HER2-low (both 1 or 2+) (n = 202) or HER2-3+ (n = 111) tumors (Table). PTEN mutations (mut; n=13) were associated with significantly shorter TTNT. ERBB2 amplifications or gains (n= 31) correlated with improved outcomes. CDK12 deletions or loss (n = 14) were linked to poorer TTNT (Table). PTEN mut and ERBB2 amplifications were predictive of outcomes with T-DXd, as neither alteration was associated with TTNT in pts receiving non-T-DXd first-line systemic therapy. Tumors with inflamed TME had the worst outcomes, followed by deserts and altered TMEs (Table). Conclusions: We identify favorable biomarkers of T-DXd efficacy in aBC, including concordant HER2-low or HER2-3+ status, absence of PTEN mut, and an altered or desert TME. These findings require validation to refine treatment strategies across HER2-driven malignancies. OS T-DXd: HR (95% CI) p/q-value TTNT T-DXd: HR (95% CI) p/q-value TTNT 1st Line: HR (95% CI) p/q-value Concordant HER2-0 (n = 32) vs discordant (n = 118) 1.07 (0.65-1.75) 0.789 1.17 (0.76-1.79) 0.474 - - Concordant HER2-low (n = 202) vs discordant 0.67 (0.49-0.92) 0.012 0.65 (0.50-0.85) 0.002 - - Concordant HER2-3+ (n = 111) vs discordant 0.23 (0.14-0.38) < 0.001 0.27 (9.18-0.4) < 0.001 - - PTEN mutations (n = 13) vs wild-type (WT) tumors (n =150) - - 2.20 (1.20-4.0) 0.068 0.99 (0.76-1.35) 0.93 ERBB2 amplifications/gains (n = 31) vs ERBB2 WT ( n = 132) 0.43 (0.26-0.72) 0.045 - - 1.1 ( 0.77 - 1.71) 0.51 CDK12 loss/deletions (n = 14) vs CDK12 WT (n = 116) - - 2.56 (1.39-4.76) 0.014 1.42 (1.04-1.86) 0.016 Altered (n = 28) vs Desert (n = 34) 0.58 (0.26-1.30) 0.18 0.97 (0.52-1.80 0.91 - - Inflamed (n = 35) vs Desert (n = 34) 2.21 (1.20-4.05) 0.011 2.13 (1.21-3.74) 0.0084 - -