1032 Background: Endocrine therapy (ET) durability for ER+/HER2- breast cancer (BC) with CDK4/6 inhibitor (CDK4/6i) exposure is vastly heterogeneous. Aberrant activation of CDK4/6-Rb axis is mechanistically linked to DNA replication origin licensing programs, and IFNγ-mediated signaling has been observed at onset of CDK4/6i resistance (PMID 33536276). We previously reported licensing complex (LC) mRNA to be predictive of ET benefit in primary ER+/HER2- BC (SABCS24-P1-03-20). Here we assessed these findings in a large, real-world cohort of primary and metastatic ET±CDK4/6i-treated tumors. Methods: ER+/HER2- BC biopsies (N=10833 all-comers and a treatment-naïve subset of N=1960) with DNA- and RNA-seq profiling from Caris Life Sciences were reviewed. LC score was calculated as a mean z-score across 14 LC genes; IFNγ score of PMID 28650338 was used. Groupings of Q1 (low) and Q4 (high) LC and IFNγ score quartiles were compared (i.e., Q1/Q1, Q4/Q4, Q1/Q4, Q4/Q1). Endpoints inferred from claims data were time on 1 st ET (To1) and cumulative time on ET (start of 1 st ET to end of last ET; ToT). Logrank p-values and HRs were derived from Cox proportional hazard models. Immune microenvironment (TME) proportions were estimated via quanTIseq deconvolution. Corrected q-values were calculated for quanTIseq and genomic analyses. Results: For all-comers, LC score Q4 vs Q1 predicted shorter ToT for ET with or without CDK4/6i exposure (HR 1.27 p<0.0001; 1.13 p=0.0019). LC-Q1~IFNγ-Q1 tumors demonstrated longer ToT (ET HR 1.76 p=0.0015; ET+CDK4/6i HR 1.32 p=0.022) relative to Q4/Q4 when TP53 was not mutated ( mut ). In the treatment-naïve subset, the distribution of LC and IFNγ scores, PAM50 luminal subtype, and TP53 mut appeared similar to all-comers. IFNγ-Q4 vs Q1 predicted shorter To1 with CDK4/6i exposure (14.4 vs 19.9 months, HR 1.28 p=0.0074). LC/IFNγ profiles significantly stratified both To1 and ToT only with CDK4/6i exposure. Relative to Q1/Q1, all other quartile groups had shorter To1 (HR 1.53 p=0.071, 1.88*, and 2.15* *p<0.01) and shorter ToT (HR 1.89, 2.33, and 1.98; all p<0.05). Relative to Q1/Q1, the Q4/Q4 cohort was enriched for TP53 mut (44.2% vs 12.5% q=0.002). In TP53 WT tumors, Q4~Q4 trended toward shorter ToT than Q1~Q1 (42.7 vs 26.5 months, HR 1.67 p=0.075). Finally, LC-Q4/IFNγ-Q4 vs Q1/Q1 tumors exhibited a more inflamed TME with higher M2 macrophages, myeloid DCs, and Tregs (all q≤0.01). Conclusions: Real-world validation confirms the LC and IFNγ-stimulated transcriptional state jointly stratify outcomes in the CDK4/6i-treated setting, predicting shorter time on ET. Though enriched for TP53 mut , convergence of LC/IFNγ may hold potential predictive value in TP53 WT disease. The LC-Q1/IFNγ-Q1 CDK4/6i response predictor in the TP53 WT setting remains to be validated prospectively.
Raghavan et al. (Wed,) studied this question.