Combining molecular subtype and traditional histo-pathologic tumor classifications to predict survival in patients with endometrioid or serous endometrial cancers (221)

Abstract

Objectives: We sought to determine the interaction between pathologic and histologic features and impact on overall survival (OS) among patients with endometrial cancer classified by the four molecular subtypes, identified by The Cancer Genome Atlas (TCGA). Methods: Using TCGA’s Pan-Cancer Atlas accessed via the cBioPortal, we identified patients with endometrioid or serous endometrial cancer. Patients with available molecular data were included in our analysis. We recorded available outcomes data, and FIGO stage and pathologic factors were manually extracted from surgical pathology reports. OS was compared among groups using Kaplan-Meier curves. Results: A total of 507 patients with molecularly subtyped endometrioid or serous endometrial cancers were identified. FIGO stage was inferred using surgical pathology reports, and data regarding LVSI and histologic grade were available in 467 patients. Tumors were categorized as POLE hypermutated (43/467; 9.2%), microsatellite instable (MSI) hypermutated (139/467; 29.8%), copy number low (CN-low) (142/467; 30.4%), and CN-high (143/467; 30.6%). While molecular subtype was significantly correlated with OS among the entire study population (p <0.0001), this correlation was only significant among stage III tumors when divided by FIGO stages (n=100, p = 0.012). Increasing histologic grade was associated with worse OS among patients with CN-low tumors (5-year survival: G1 100%, G2 88%, G3 59.4%; p = 0.0051) but not among patients with other TCGA molecular subtypes or low-grade tumors. Of the 73 patients with CN-high tumors with LVSI, 32 (43.8%) were node-positive. Of the 35 patients with CN-low tumors with LVSI, four (11.4%) were node-positive. The presence of LVSI was associated with worse OS than those without LVSI among patients with CN-high (3-year survival rate: 56.9% vs 83.1%, p = 0.0048) and CN-low (3-year survival rate: 88.5% vs 94.4%, p = 0.0024) tumors. Conclusions: In this cohort of patients with endometrioid and serous endometrial cancer from the TCGA, the presence of LVSI was associated with worse OS among patients with CN-low and CN-high tumors. Patients with CN-low tumors with LVSI were less likely to be node positive than patients with CN-high tumors with LVSI. Increasing histologic grade was associated with worse OS only among patients with CN-low tumors. These data support a combined molecular histopathologic classification for prognosis in endometrial cancer. Objectives: We sought to determine the interaction between pathologic and histologic features and impact on overall survival (OS) among patients with endometrial cancer classified by the four molecular subtypes, identified by The Cancer Genome Atlas (TCGA). Methods: Using TCGA’s Pan-Cancer Atlas accessed via the cBioPortal, we identified patients with endometrioid or serous endometrial cancer. Patients with available molecular data were included in our analysis. We recorded available outcomes data, and FIGO stage and pathologic factors were manually extracted from surgical pathology reports. OS was compared among groups using Kaplan-Meier curves. Results: A total of 507 patients with molecularly subtyped endometrioid or serous endometrial cancers were identified. FIGO stage was inferred using surgical pathology reports, and data regarding LVSI and histologic grade were available in 467 patients. Tumors were categorized as POLE hypermutated (43/467; 9.2%), microsatellite instable (MSI) hypermutated (139/467; 29.8%), copy number low (CN-low) (142/467; 30.4%), and CN-high (143/467; 30.6%). While molecular subtype was significantly correlated with OS among the entire study population (p <0.0001), this correlation was only significant among stage III tumors when divided by FIGO stages (n=100, p = 0.012). Increasing histologic grade was associated with worse OS among patients with CN-low tumors (5-year survival: G1 100%, G2 88%, G3 59.4%; p = 0.0051) but not among patients with other TCGA molecular subtypes or low-grade tumors. Of the 73 patients with CN-high tumors with LVSI, 32 (43.8%) were node-positive. Of the 35 patients with CN-low tumors with LVSI, four (11.4%) were node-positive. The presence of LVSI was associated with worse OS than those without LVSI among patients with CN-high (3-year survival rate: 56.9% vs 83.1%, p = 0.0048) and CN-low (3-year survival rate: 88.5% vs 94.4%, p = 0.0024) tumors. Conclusions: In this cohort of patients with endometrioid and serous endometrial cancer from the TCGA, the presence of LVSI was associated with worse OS among patients with CN-low and CN-high tumors. Patients with CN-low tumors with LVSI were less likely to be node positive than patients with CN-high tumors with LVSI. Increasing histologic grade was associated with worse OS only among patients with CN-low tumors. These data support a combined molecular histopathologic classification for prognosis in endometrial cancer.