End-to-end deep learning method for predicting hormonal treatment response in women with atypical endometrial hyperplasia or endometrial cancer.

Purpose: Endometrial cancer (EC) is the most common gynecologic malignancy in the United States, and atypical endometrial hyperplasia (AEH) is considered a high-risk precursor to EC. Hormone therapies and hysterectomy are practical treatment options for AEH and early-stage EC. Some patients prefer hormone therapies for reasons such as fertility preservation or being poor surgical candidates. However, accurate prediction of an individual patient's response to hormonal treatment would allow for personalized and potentially improved recommendations for these conditions. This study aims to explore the feasibility of using deep learning models on whole slide images (WSI) of endometrial tissue samples to predict the patient's response to hormonal treatment. Approach: We curated a clinical WSI dataset of 112 patients from two clinical sites. An expert pathologist annotated these images by outlining AEH/EC regions. We developed an end-to-end machine learning model with mixed supervision. The model is based on image patches extracted from pathologist-annotated AEH/EC regions. Either an unsupervised deep learning architecture (Autoencoder or ResNet50), or non-deep learning (radiomics feature extraction) is used to embed the images into a low-dimensional space, followed by fully connected layers for binary prediction, which was trained with binary responder/non-responder labels established by pathologists. We used stratified sampling to partition the dataset into a development set and a test set for internal validation of the performance of our models. Results: The autoencoder model yielded an AUROC of 0.80 with 95% CI [0.63, 0.95] on the independent test set for the task of predicting a patient with AEH/EC as a responder vs non-responder to hormonal treatment. Conclusions: These findings demonstrate the potential of using mixed supervised machine learning models on WSIs for predicting the response to hormonal treatment in AEH/EC patients.

TP53 mutation and abnormal p53 expression in endometrial cancer: Associations with race and outcomes.

OBJECTIVE: This multi-center cohort study assessed associations between race, TP53 mutations, p53 expression, and histology to investigate racial survival disparities in endometrial cancer (EC). METHODS: Black and White patients with advanced or recurrent EC with Next Generation Sequencing data in the Endometrial Cancer Molecularly Targeted Therapy Consortium database were identified. Clinicopathologic and treatment variables were summarized by race and compared. Overall survival (OS) and progression-free survival (PFS) among all patients were estimated by the Kaplan-Meier method. Cox proportional hazards models estimated the association between race, TP53 status, p53 expression, histology, and survival outcomes. RESULTS: Black patients were more likely than White patients to have TP53-mutated (N = 727, 71.7% vs 49.7%, p < 0.001) and p53-abnormal (N = 362, 71.1% vs 53.2%, p = 0.003) EC. Patients with TP53-mutated EC had worse PFS (HR 2.73 (95% CI 1.88-3.97)) and OS (HR 2.20 (95% CI 1.77-2.74)) compared to those with TP53-wildtype EC. Patients with p53-abnormal EC had worse PFS (HR 2.01 (95% CI 1.22-3.32)) and OS (HR 1.61 (95% CI 1.18-2.19)) compared to those with p53-wildtype EC. After adjusting for TP53 mutation and p53 expression, race was not associated with survival outcomes. The most frequent TP53 variants were at nucleotide positions R273 (n = 54), R248 (n = 38), and R175 (n = 23), rates of which did not differ by race. CONCLUSIONS: Black patients are more likely to have TP53-mutated and p53-abnormal EC, which are associated with worse survival outcomes than TP53- and p53-wildtype EC. The higher frequency of these subtypes among Black patients may contribute to survival disparities.

Identification of Candidate Biomarker and Drug Targets for Improving Endometrial Cancer Racial Disparities.

Racial disparities in incidence and survival exist for many human cancers. Racial disparities are undoubtedly multifactorial and due in part to differences in socioeconomic factors, access to care, and comorbidities. Within the U.S., fundamental causes of health inequalities, including socio-economic factors, insurance status, access to healthcare and screening and treatment biases, are issues that contribute to cancer disparities. Yet even these epidemiologic differences do not fully account for survival disparities, as for nearly every stage, grade and histologic subtype, survival among Black women is significantly lower than their White counterparts. To address this, we sought to investigate the proteomic profiling molecular features of endometrial cancer in order to detect modifiable and targetable elements of endometrial cancer in different racial groups, which could be essential for treatment planning. The majority of proteins identified to be significantly altered among the racial groups and that can be regulated by existing drugs or investigational agents are enzymes that regulate metabolism and protein synthesis. These drugs have the potential to improve the worse outcomes of endometrial cancer patients based on race.

Endometrial Cancer Characteristics and Risk of Recurrence.

OBJECTIVES: To describe clinicopathologic characteristics and survival outcomes of endometrial adenocarcinomas stratified by mismatch repair (MMR) status. METHODS: Single-institution, retrospective study of all women with endometrioid adenocarcinomas treated from January 2012 through December 2017. Patients were categorized into one of three groups based on MMR testing: intact MMR expression (MMR+), probable MMR mutation (MMR-), or MLH1 hypermethylation (hMLH1+). Demographics, pathologic characteristics, recurrence rates, and survival differences were analyzed. RESULTS: In total, 316 women were included in the analysis: 235 (74.4%) patients in the MMR+ group, 10 (3.1%) in the MMR- group, and 71 (22.5%) in the hMLH1+ group. Patients with hMLH1+ were significantly older, exhibited higher-grade histology and presence of lymphovascular space invasion, and were more likely to have received adjuvant treatment. The early stage hMLH1+ patients were more likely to recur (15.3% hMLH1+ vs 2.3% MMR+ vs 12.5% MMR-, P < .001). Hypermethylation remained a significant predictor of recurrence in multivariable analysis (odds ratio, 5.09; 95% confidence interval [CI], 1.54-16.86; P = .008). Recurrence-free survival was significantly reduced in early stage hMLH1+ (hazard ratio, 7.40; 95% CI, 2.80-21.62; P < .001). CONCLUSIONS: Women with hMLH1+ endometrial cancer have worse prognostic features and recur more frequently, even in patients traditionally considered low risk for recurrence.