br disease specific survival for
disease specific survival, for obese patients on univariate analysis but that it was not significant after multivariable control for age, FIGO Stage, subtype and grade [8,9]. A recent study of 154 women with high-risk endometrial cancer, failed to show an association with BMI, progression free survival, or overall survival, in a population of women characterized by 78% with BMI N 25 kg/m2 .
Conversely several studies have reported 1.6–2.9 increased hazard of death, including disease specific and overall for patients with elevated BMI verses patients of normal weight with endometrial cancer . Additionally, endometrial cancer survivors with BMI N 40 were reported to have decreased overall survival as compared to BMI b30 (HR 2.77; 95% CI 1.21–6.36 p = 0.16) in a Gynecologic Group Study . In a pooled analysis of patients enrolled in 5 Gynecologic Oncology Group trials receiving cisplatin and doxorubicin there was no associa-tion of BMI with progression free survival for women receiving doxoru-bicin and platinum as part of GOG trials overall, but decreased overall survival for women with stage 3 or 4 disease and higher BMI . Sur-vivors with BMI ≥ 30 kg/m2 were reported to have higher all cause (HR 1.6, 95% CI 1.0–2.5) and disease specific (HR 2.0, 95% CI 0.8–5.1) mortal-ity as compared to non-obese survivors by Chia et al. . After stratifi-cation for poor risk factor including high stage, high grade, deep myometrial invasion as defined by N50% invasion, cervical involvement, Olaparib node metastases, positive cytology, and lymphvascular space in-vasion Gates et al. also reported poorer survival for overweight verses normal weight survivors . Felix et al. from a study of 4609 women enrolled in the GOG 210 trial reported decreased survival for women with Class II obesity to normal BMI (HR2.29, 95% CI = 1.06–4.98, p = 0.01) . In these studies the majority of participants were of white race and non-Hispanic ethnicity.
The purpose of this study was to report the association of obesity with survival of women with both endometrioid and non-endometrioid cancers in a racially and ethnically diverse population. Secondarily, the goal of this study was to evaluate the association of race, and the use of antihyperlipidemic/antihypglycemic/anti-inflammatory medications with obesity and survival.
After Internal Review Board approval all women treated for endo-metrial cancer from 1999 to 2016 at Montefiore Medical Center were reviewed for histopathological, treatment and demographic data. Data collected included age at histopathological diagnosis, stage, grade, his-tology, self-identified race/ethnicity, medical comorbidities, medication use, treatment, and date of last contact. Body mass index (BMI) was measured in kilograms per meter squared and categories of BMI were designated as underweight b 18.5, normal 18.5–24.9, overweight 25–29.9, class I obese 30–34.9, class II obese 35–39.9, class III obesity
≥ 40. Type 2 or non-endometrioid cancers were defined as serous, carci-nosarcomas, clear cell, mixed and poorly differentiated. Any percentage of serous, carcinosarcoma, clear cell, mixed or poorly differentiated tumor qualified as classification into that respective histological cate-gory. Death was confirmed by query of the Social Security Death Index. Disease specific survival (DSS) was defined as the interval from time of histopathologically confirmed cancer from biopsy or surgical resection to death attributable to uterine malignancy. Overall survival
(OS) was defined as the interval from diagnosis to death by any cause. For patients who were still alive at the time of analysis, DSS and OS were censored at the date of follow up. Demographic, histopathologic and medical data were presented as face valid terms and univariable analysis was performed to identify differences between obesity categories. Linear regression modeling was performed for BMI categories and associated age at diagnosis. DSS and OS by BMI category were summarized for the entire cohort, endometrioid, and non-endometrioid patients separately using Kaplan-Meier survival curves. Univariable, multivariable and stratified Cox regression modeling were performed to assess the independent
association of BMI on survival. The cohort was analyzed as a whole and then separate sub-cohorts of endometrioid histology only, non-endometrioid or Type II histology for distribution of characteristics including race and obesity and then univariable and multivariable Cox regression modeling was conducted for the entire cohort and histologic sub-cohorts. All possible factors associated with both BMI and disease-specific mortality and those found to have p b 0.10 in univariable anal-ysis were included in the multivariable model. Variables with evidence of collinearity were excluded from the model as described.
In a sub-cohort of patients with endometrioid (Type I) lesions con-fined to the uterus, evaluation of high risk factors including depth of invasion, lymph vascular space invasion, and grade of tumor were eval-uated for their association with obesity. Additionally, estrogen receptor and progesterone receptor status were obtained when available from the pathology report and univariable and multivariable modeling was performed to identify any association of obesity as measured in kilo-grams per meter-squared with high risk factors and with race. Proges-terone and estrogen receptor positive was defined by institutional pathology standard of ≥20%. Testing for Lynch syndrome and microsat-ellite stability was also obtained from the pathology record. All analyses were performed using Stata version 14.0 (StataCorp. 2015. Stata Statis-tical Software: Release 14. College Station, TX: StataCorp LP.)