Molecular and pathologic data can guide selection of endometrioid endometrial cancer patients for ovarian preservation (239)

Abstract

<b>Objectives:</b> To investigate the association of The Cancer Genome Atlas (TCGA) molecular subtype, <i>CTNNB1</i> (B-catenin) hotspot mutation, and pathologic factors with concurrent or recurrent ovarian disease in endometrial cancer (EC) patients (pts). <b>Methods:</b> EC pts aged ≤50 years were grouped by those electing oophorectomy (BO) versus ovarian preservation (OP) at staging surgery (January 2010-June 2021). ECs were classified as TCGA subtypes using a clinical tumor-normal next-generation sequencing (NGS) assay and immunohistochemistry data: <i>POLE</i> hotspot mutated, microsatellite instability-high (MSI-H), copy number high (CN-high), or copy number low (CN-low). <i>CTNNB1</i> hotspot mutations were classified. Germline data identified Lynch syndrome pts. Associations between molecular/pathologic features and concurrent ovarian disease (synchronous or metastatic) in BO pts were compared with Wilcoxon rank-sum and Fisher's exact tests. Associations with isolated ovarian recurrences in OP pts were examined with time to event outcomes. <b>Results:</b> Of 420 pts, 317 underwent BO, and 103 OP followed institutional guidelines (Table 1). Among BO pts with NGS (<i>n</i>=90), 14 (16%) had <i>POLE,</i> 22 (24%) MSI-H, six (7%) CN-high and 48 (53%) CN-low EC. Pathogenic <i>CTNNB1</i> mutations were present in 30/90 (33%) ECs. Of 102 pts with germline testing, 28 (28%) had Lynch syndrome. TCGA classification for OP pts with NGS (<i>n</i>=37) included six (16%) <i>POLE</i>, one (3%) MSI-H, one (3%) CN-high and 29 (78%) CN-low EC. Pathogenic <i>CTNNB1</i> mutations were present in 19/37 (51%) ECs. One of 23 (4%) pts with germline testing had Lynch syndrome. Of 317 BO pts, 27 (9%) had a concurrent ovarian disease. Among BO pts with NGS, concurrent ovarian tumors were diagnosed in 0/14 (0%) <i>POLE</i>, 2/48 (4%) CN-low, 10/22 (45%) MSI-H and 3/6 (50%) CN-high ECs (p<0.001). Concurrent ovarian disease was present in 1/37 (3%) hotspot <i>CTNNB1</i> mutated versus 9/58 (16%) wildtype/<i>CTNNB1</i> non-hotspot mutated ECs (p=0.15). Seven of 28 (25%) Lynch versus 7/74 (10%) non-Lynch syndrome pts had concurrent ovarian tumors (p=0.06). In BO pts, concurrent ovarian disease was associated with FIGO grade 2-3 versus grade 1 EC (20% vs 5%; p<0.001), positive versus negative lymphovascular space invasion (LVSI) (20% vs 7%; p=0.004), positive versus negative washings (28% vs 7%; p=0.016) and ≥50% versus <50% myoinvasion (24% vs 7%; p=0.006). Of 103 OP pts, four (4%) had isolated ovarian recurrences over a median follow-up of 17 months. Two pts had high-risk pathologic features and two had high-risk molecular features (CN-high EC and Lynch syndrome). <b>Conclusions:</b> Molecular and pathologic data improve risk stratification of pre-menopausal EC pts and enhance patient selection for ovarian preservation. In addition to Lynch syndrome, concurrent ovarian disease was seen more frequently in CN-high and MSI-H tumors or those with grade 2-3, LVSI, deep myoinvasion, and positive washings. Isolated ovarian recurrences may be independently associated with high-risk molecular features.