Molecular analysis of endometrioid and non-endometrioid uterine cancer and response to treatment (246)

Abstract

<b>Objectives:</b> Carboplatin and Paclitaxel are the recommended first-line regimens for uterine cancer patients requiring systemic treatment. In light of promising new alternative systemic therapies, we aimed to identify biomarkers that may predict response to carboplatin. <b>Methods:</b> Uterine cancer tumor samples were analyzed using next-generation sequencing (NGS: NextSeq, 592 Genes) and immunohistochemistry (IHC) (Caris Life Sciences, Phoenix, AZ). Real-world overall survival (OS) was obtained from insurance claims data, and Kaplan-Meier estimates were calculated for molecularly defined patient cohorts from the first treatment of carboplatin to the last day of contact. The median survival of carboplatin-associated survival (CAS) was determined after the first carboplatin treatment. Two cohorts were created and compared: <i>1)</i> OS after carboplatin-based treatment that was less than the median CAS of the whole cohort (Non-Responders, "NR"); and <i>2)</i> OS after carboplatin-based treatment that was longer than the median CAS (Responders, "R"). NGS and IHC results of the tumor samples of these cohorts were compared. Statistical significance was calculated using Mann-Whitney U and Chi-square test and adjusted for multiple comparisons (adjusted p-value = q-value). <b>Results:</b> Total 2436 uterine cancer patients who received carboplatin treatment with molecular data were included in this analysis: 578 (23.7%) were of endometrioid histology, and 1051 (43.1%) serous, carcinosarcoma, or clear cell histology. Irrespective of histology, 20 genes were significantly altered (q<0.05) between R and NR, including genes related to the pathways involved in Base/Nucleotide Excision Repair, Cell Cycle Control, Chromatin Remodeling, RTK RAS, HR, PI3K, VEGF, and WNT signaling (Table 1). Of note, <i>KMT2C,</i> a histone methyltransferase, was more frequently mutated in Rs compared to NRs (8.44% vs 3.73% mutated, p=0.01). But <i>HOOK3</i> and <i>NSD3</i> were more frequently amplified in NRs compared to Rs (1.19% vs 0.12%, q=0.047; 2.14% vs 0.73%, q=0.11). <i>NSD3,</i> a gene involved in the regulation of tumorigenesis, was amplified in 1.07% of patients and was associated with a median difference in survival of 661 fewer days (HR: 2.75, 95% CI: 1.70-4.45, p<0.00001). Among endometrioid and serous uterine histology, ER positivity was seen significantly more frequently in Rs compared to NRs (83% vs 66.2%, q=0.03; 60.1% vs 45.8%, q=0.03). However, <i>CCNE1</i> amplification was less frequent in Rs compared to NRs (4.80% vs 16.4%, q=0.01). There were no molecular differences seen between Rs and NRs in carcinosarcoma and clear cell histologies treated with carboplatin. <b>Conclusions:</b> There are significant differences in the molecular profiles of endometrial cancers that have longer median CAS compared to those with shorter median CAS. Further investigation into biomarkers that may predict response to carboplatin-based therapy in endometrial cancer is needed.