Endometrial Hyperplasia

A multicenter retrospective study was conducted from four institutions (University of Southern California, Osaka University, Tokai University, and Niigata University). Institutional review board approval was obtained at each participating institution. Eligibility criteria for the study were consecutive patients with endometrial cancer who underwent hysterectomy-based surgical staging between January 1, 2000, and December 31, 2013. Patients with uterine carcinosarcoma, endometrial hyperplasia, and metastatic cancer to the endometrium as well as patients without surgical staging were excluded from the study. Among the eligible patients, information for patient demographics, tumor characteristics, treatment pattern, and survival outcome were abstracted from archived medical records. The STrengthening the Reporting of OBservational studies in Epidemiology guideline for retrospective study was consulted.¹⁶ A proportion of patients (76.7%) was included within the context of our previous studies.^{17 (link)}Patient demographics including age, ethnicity, body mass index (BMI, calculated as weight [kg]/ [height (m)]²), medical comorbidity (hypertension, diabetes mellitus, hypercholesterolemia, and cigarette use), and medication type were collected at the time of the initial endometrial cancer diagnosis. Medication history was verified through multiple sources. For tumor characteristics, histologic subtypes, grade, stage, depth of myometrial invasion, presence of lymphovascular space invasion, lymph node metastasis (pelvic, paraaortic, or both), and hormonal receptor status (estrogen receptor [ER] and progesterone receptor) were collected from records for hysterectomy-based surgical staging. For treatment pattern, postoperative chemotherapy type (carboplatin and paclitaxel compared with others) and postoperative radiotherapy type (whole pelvic radiotherapy compared with intracavitary brachytherapy) were recorded. For survival outcome, disease-free survival and disease-specific overall survival were obtained from institutional medical records. Data entry into a deidentified data sheet was performed by coinvestigators in each participating institution, and the principal investigator reviewed all the data for accuracy and consistency.
Obesity was defined as a BMI of 30 or greater. Low-dose aspirin refers to an 81–100 mg oral daily dose. Nonsteroidal anti-inflammatory drugs (NSAIDs) other than aspirin were classified as other NSAIDs. Tumor grade was based on the International Federation of Gynecology and Obstetrics: 5% or less solid component for grade 1, 6–50% solid component for grade 2, and greater than 50% solid component for grade 3.^{18 (link)} Cancer stage was reclassified based on the 2009 International Federation of Gynecology and Obstetrics staging system.^{19 (link)} Type I cancer was defined as grade 1–2 endometrioid histology, and type II cancer was defined as grade 3 endometrioid, serous, clear cell, mixed, and other histology.^{20 (link)-22 (link)} Deep myometrial invasion was defined as greater than 50% tumor invasion of the myometrium. Medication types were grouped as antihypertensive, antiglycemic, anticholesterol, analgesia, antacid, psychiatric, and other class agents. Disease-free survival was defined as the time interval between the date of hysterectomy-based surgical staging and the date of the first recurrence of endometrial cancer or the last follow-up date if there was no recurrence. Disease-specific overall survival was defined as the time interval between the date of hysterectomy-based surgical staging and the date of death resulting from endometrial cancer. Cases were censored if the patient was alive at the last follow-up date or if the patient died from other causes. Causes of death other than endometrial cancer were also recorded.
The primary aim of analysis was to determine the effects of low-dose aspirin use on the survival outcome of women with endometrial cancer. The secondary interest of analysis was to examine the subgroup of patients who benefit from low-dose aspirin use. Sample size estimation was performed by using α=0.05 and β=0.20. Based on prior studies, a hazard ratio (HR) for disease-free survival of 0.75 and a 10% low-dose aspirin use rate were approximated for our study population.^{14 (link)} By this computation, 140 cases of low-dose aspirin use and 1,440 cases of nonuse were estimated.
Continuous variables were expressed with mean±standard deviation or median (range) based on the normality examined by Kolmogorov-Smirnov test. Statistical significances of continuous variables between the two groups were examined by Student t test or Mann-Whitney U test, as appropriate. Spearman’s correlation coefficient was determined among continuous variables. Kruskal-Wallis test was used for the comparison of median in more than two groups. Statistical significance of ordinal and categorical variables was examined by χ² test or Fisher exact test as appropriate.
Log-rank test for univariate analysis and a Cox proportional hazard regression model for multivariable analysis were used for survival analysis. Covariates entered in the initial multivariable model were the statistically significant variables in univariate analysis with the cutoff value being P<.05. These included age (younger than 60 compared with 60 years or older), ethnicity (Caucasian, African American, Hispanic, and Asian), histologic subtype (endometrioid compared with nonendometrioid), grade (1–2 compared with 3), stage (I–II compared with III–IV), deep myometrial invasion (no compared with yes), lymphovascular space invasion (no compared with yes), postoperative chemotherapy (no compared with yes), postoperative radiotherapy (none, whole pelvic radiotherapy±intracavitary brachytherapy and intracavitary brachytherapy alone), low-dose aspirin (no compared with yes), acetaminophen (no compared with yes), other NSAIDs (no compared with yes), and metformin (no compared with yes) that were grouped in an a priori manner.^{19 (link),22 (link),23 (link)} Conditional backward method was then used to determine the independent prognostic factor for disease-free and disease-specific overall survival until all covariates became statistically significant in the final model, and magnitude of statistical significance was expressed with HR and 95% confidence interval (CI). Kaplan-Meier method was used to construct survival curves. All statistical tests were two-tailed, and a P<.05 was considered statistically significant. SPSS 12.0 was used for the analyses.

We conducted a retrospective observational cohort study after obtaining Institutional Review Board approval from the University of Southern California (USC); subjects with any histologic diagnosis of endometrial hyperplasia between 2003 and 2011 at Los Angeles County USC Medical Center were identified through the pathology department’s database. Archived charts were subsequently reviewed and the following parameters were abstracted: age, race/ethnicity, gravidity, parity, BMI at diagnosis (grouped as: normal weight BMI < 30 kg/m², class I and II obesity BMI 30–39.9 kg/m², and class III obesity BMI ≥ 40 kg/m² per the World Health Organization (WHO) criteria), medical comorbidities (diabetes mellitus, hypertension, hyperlipidemia, infertility, and polycystic ovarian syndrome), medications at diagnosis and initiated during the course of therapy, initial endometrial histologic categorization, and all subsequent endometrial histology.
The chronologic sequence and types of endometrial sampling performed were recorded with the corresponding histologic diagnosis. Acceptable forms of endometrial sampling included pipelle endometrial biopsy (EMB), VABRA aspiration biopsy (Berkley Medevices, Berkley, CA [17 (link)]), uterine curettage, and hysterectomy. The histologic diagnoses were categorized into: simple hyperplasia, complex hyperplasia, simple atypical hyperplasia, and CAH per the World Health Organization (WHO) criteria [18 ]. Treatment regimens along with follow-up of each regimen were recorded and then categorized into systemic versus localized (LNG-IUS). Though the LNG-IUS is commonly used to treat both CAH and endometrial cancer, it is not FDA approved for this indication.
Subject records were then screened for inclusion and exclusion criteria. Included patients had a confirmed pathological diagnosis of CAH, had been prescribed either an oral or local progestin as treatment, and had at least one documented and interpretable follow-up endometrial assessment performed after 4 or more weeks of therapy. The same forms of sampling previously mentioned were deemed reliable for assessing treatment response in addition to histology on hysterectomy, if performed after hormonal therapy.
Patients with a prior or concurrent diagnosis of endometrial adenocarcinoma and/or patients with known reversible sources of unopposed estrogen, such as granulosa cell tumors of the ovary or exogenous hormone therapy, were excluded, as were those who received progestin therapy in the month preceding the initial CAH diagnosis. We also excluded patients treated with non-progestin therapies such as aromatase inhibitors or GnRH agonists or if they underwent hysterectomy without intervening hormonal therapy. Subjects with inadequate or missing follow-up endometrial sampling were excluded as were those, whose only follow-up endometrial sampling occurred less than 4 weeks after beginning medical treatment. To assess treatment response, only treatments given between the first diagnosis of CAH and the subsequent biopsy were included.
Response to treatment was categorized as: complete regression, partial regression, or persistent/progressive disease. A complete regression was defined as resolution from CAH to benign endometrium (no residual typical or atypical hyperplasia and no progression to endometrial cancer). A partial regression occurred if the follow-up biopsy revealed non-atypical hyperplasia. Persistent/progressive disease signified either ongoing CAH or progression to cancer. We conducted two separate analyses to examine rates of complete regression and rates of overall regression (partial responders and complete responders). Rates were examined across collected variables. Logistic regression and Fisher’s exact test (two-tailed hypothesis) were used for statistical comparison. The magnitude of statistical significance was expressed with odds ratio (OR) and 95% confidence interval (CI).