Carboplatin in BRCA1/2-mutated and triple-negative breast cancer BRCAness subgroups: the TNT Trial

Abstract

Germline mutations in BRCA1/2 predispose individuals to breast cancer (termed germline-mutated BRCA1/2 breast cancer, gBRCA-BC) by impairing homologous recombination (HR) and causing genomic instability. HR also repairs DNA lesions caused by platinum agents and PARP inhibitors. Triple-negative breast cancers (TNBCs) harbor subpopulations with BRCA1/2 mutations, hypothesized to be especially platinum-sensitive. Cancers in putative 'BRCAness' subgroups-tumors with BRCA1 methylation; low levels of BRCA1 mRNA (BRCA1 mRNA-low); or mutational signatures for HR deficiency and those with basal phenotypes-may also be sensitive to platinum. We assessed the efficacy of carboplatin and another mechanistically distinct therapy, docetaxel, in a phase 3 trial in subjects with unselected advanced TNBC. A prespecified protocol enabled biomarker-treatment interaction analyses in gBRCA-BC and BRCAness subgroups. The primary endpoint was objective response rate (ORR). In the unselected population (376 subjects; 188 carboplatin, 188 docetaxel), carboplatin was not more active than docetaxel (ORR, 31.4% versus 34.0%, respectively; P = 0.66). In contrast, in subjects with gBRCA-BC, carboplatin had double the ORR of docetaxel (68% versus 33%, respectively; biomarker, treatment interaction P = 0.01). Such benefit was not observed for subjects with BRCA1 methylation, BRCA1 mRNA-low tumors or a high score in a Myriad HRD assay. Significant interaction between treatment and the basal-like subtype was driven by high docetaxel response in the nonbasal subgroup. We conclude that patients with advanced TNBC benefit from characterization of BRCA1/2 mutations, but not BRCA1 methylation or Myriad HRD analyses, to inform choices on platinum-based chemotherapy. Additionally, gene expression analysis of basal-like cancers may also influence treatment selection.

Figure 1. Consort diagram

Flow of participants in the trial.

Figure 2. Response rates (overall and BRCA subgroups)

Absolute differences between treatment groups within biomarker subgroups are presented; p-values for the differences are calculated using a 2-sided Fisher’s exact test. P-values for interactions are based on a logistic regression model of response with terms for biomarker status, treatment group and interaction.

Figure 3. Progression-free survival (overall and BRCA subgroups)

Data presented is the difference in PFS restricted mean (95% CI). A negative value indicates a better response to docetaxel, positive values indicate better response to carboplatin. P-values are calculated using a 2-sided t-test comparing the mean survival between treatments (within biomarker groups as appropriate). C=Carboplatin; D=Docetaxel.

Figure 4. Response rates (HRD subgroups)

Absolute differences between treatment groups within HRD subgroups are presented; p-values for the differences are calculated using a 2-sided Fisher’s exact test. P-values for interactions are based on a logistic regression model of response with terms for biomarker status, treatment group and interaction.

Figure 5. Response rates (basal-like subgroups)

Absolute differences between treatment groups within basal subgroups are presented; p-values for the differences are calculated using a 2-sided Fisher’s exact test. P-values for interactions are based on a logistic regression model of response with terms for biomarker status, treatment group and interaction.

Figure 6. PFS (basal-like subgroups)

Data presented is the difference in PFS restricted mean within subgroups (95% CI). A negative value indicates a better response to docetaxel, positive values indicate better response to carboplatin. P-values are calculated using a 2-sided t-test comparing the mean survival between treatments within biomarker groups. C=Carboplatin; D=Docetaxel.