<b>Objectives:</b> To describe endometrial carcinoma (EC) molecular classification utilizing clinical next-generation sequencing (NGS) and immunohistochemistry (IHC) and to assess the clinical, pathologic, and survival outcomes based on molecular data. <b>Methods:</b> We identified all primary and recurrent EC samples of 341-505 cancer-related genes undergoing NGS (MSK-Integrated Mutation Profiling of Actionable Cancer Targets; MSK-IMPACT) from test inception in 2014 through 2020. Samples were classified into the Cancer Genome Atlas (TCGA) molecular subtypes utilizing MSK-IMPACT results for <i>POLE</i> hotspot mutations, <i>TP53</i> mutations, and MSIsensor score; and mismatch repair (MMR) and p53 IHC results collected from pathology reports. Tumors not classified per TCGA based on available NGS and IHC data were excluded from the analysis. Clinical, pathologic, and survival data were obtained for primary ECs undergoing initial staging surgery or treatment at our institution after staging surgery. Only NGS performed prior to documented recurrence was used for survival analysis, to avoid ascertainment bias. Appropriate statistical tests were applied. <b>Results:</b> Clinical MSK-IMPACT sequencing was performed on 2174 ECs. Except for 40 unclassifiable cases, 2134 tumors corresponded to 2066 unique patients, including 1712 (80%) primary and 409 (19%) recurrent tumors. Molecular classification was distributed as 5% <i>POLE</i> (<i>n</i>=113), 22% microsatellite instability-high (MSI-H; <i>n</i>=479), 38% copy-number high (CN-H) (<i>n</i>=818), and 34% CN-low (<i>n</i>=724). MMR-deficiency by IHC was found in 16% of <i>POLE</i>, and IHC-based aberrant p53 expression in 12% of <i>POLE</i> and 2% of MSI-H ECs. Due to the hierarchy of our hybrid classification algorithm, this did not affect molecular designation. Otherwise, all available MMR and p53 IHC results were concordant with NGS results to determine classification. CN-H was 91% non-endometrioid with 67% FIGO G3 within the endometrioid type, while <i>POLE</i>, MSI, and CN-L were 10-32% and 17-21%, respectively (p<0.05). Criteria for survival analysis using primary tumor NGS results were met by 1118 patients. The median follow-up was 22.5 months (0.5-214 months). Three-year progression-free survival was 97% for <i>POLE</i> (SE 2.0), 90% CN-L (SE 2.1), 80% MSI-H (SE 3.1), and 46% CN-H (SE 3.5), (p<0.01). Three-year overall survival was 100% for <i>POLE</i>, 94% MSI-H (SE 2.1), 93% CN-L (SE 2.2), and 71% CN-H (SE 3.6), (p<0.01). CN-H classification remained prognostic for recurrence outcomes when adjusting for age, stage, and histology (HR: 6.2, 95% CI: 1.5-26.6; p=0.04). <b>Conclusions:</b> Our approach to molecular classification involves a hybrid strategy utilizing combined NGS and IHC analysis as a surrogate for whole-exome sequencing employed by TCGA. Our current database is mostly CN-H, which may reflect the referral-based nature of our institution. When applying appropriate criteria for survival analysis, CN-H portends the worst outcomes. Future studies should consider the histological heterogeneity of CN-H when considering treatment options.Fig. 1
Read full abstract