Study Population
This study initially consisted of 213 consecutively recruited women with histologically confirmed endometrial cancer who presented for treatment at the Hunter Centre for Gynaecological Cancer, John Hunter Hospital, Newcastle, New South Wales, Australia between the years 1992 and 2005. Women that had additionally been diagnosed with breast cancer were excluded from this study.
The final analysis included 191 endometrial cancer patients. Data on reproductive and environmental risk factors including ethnicity, body mass index (BMI), diabetes, high blood pressure (HBP), age of diagnosis of endometrial cancer, age of menarche, age of menopause, other personal cancer history, family history of cancer (defined as cancer in the index patient plus one or more first or second degree relatives diagnosed with cancer), parity, breastfeeding, oral contraceptive use, chemotherapy, radiotherapy, hormone therapy (HT), smoking and alcohol use was collected using self reported questionnaires. Information regarding recurrence, stage, grade and histology of endometrial cancer was collected from the medical records.
The control population consisted of 291 women who were recruited between the years 2004 and 2005 for the Hunter Community Study. This study aims to identify genetic and environmental factors associated with ageing in a cohort of individuals obtained from the Hunter region, Newcastle, New South Wales, Australia. Any control that had a prior diagnosis of either breast or endometrial cancer was excluded from the study. Controls were matched to cases by sex and age.
All participants provided informed written consent prior to participation in this study. Ethics approval was obtained from the Human Research Ethics Committee, University of Newcastle and the Hunter Area Research Ethics Committee, Hunter New England Health Service, Newcastle, New South Wales, Australia.
Statistical Analysis
Power calculations were performed using Quanto (Version 1.2.3, May 2007,
http://hydra.usc.edu/GxE. The number of cases and controls were chosen to detect a 2-fold increased risk, assuming a dominant genetic model, minor allele frequency of 6.5%, p = 0.05, 80% power and 1.52 control/case ratio. For each polymorphism, Hardy-Weinberg Equilibrium (HWE) was calculated in the control groups to check for compliance using the Institute for Human Genetics, statistics website,
http://ihg2.helmholtz-muenchen.de/ihg/snps.html (Munich, Germany). To determine differences in genotype frequencies and environmental and reproductive risk factors between the cases and controls, chi-squared (χ
2) statistics, odds ratios (ORs) and 95% confidence intervals (CI) were calculated using unconditional logistic regression. Multivariate unconditional logistic regression was performed to determine if any risk factors altered the significance of the genotype frequency results. The risk factors taken into account were: BMI (<25 and ≥ 25 kg/m
2) diabetes (yes/no), HBP (yes/no), HT (yes/no), personal history of cancer (yes/no), smoking (ever/never) and alcohol consumption (ever/never). Other risk factors such as age of menopause were not included in the analysis since this information was not available for the controls.
The genotype frequencies of the cyclin D1 870 G>A polymorphism was compared in the case group stratified for the following environmental and reproductive risk factors; BMI (<25 versus >= 25), age of menarche (<12 v >= 12), age of menopause (<50 versus >= 50), parity (yes/no), oral contraceptive use (ever/never), HBP (yes/no), diabetes (yes/no), personal history of ovarian, colorectal, and/or cervical cancer (yes/no), radiotherapy (yes/no), chemotherapy (yes/no), hormone therapy (yes/no), family history of uterine, breast, colorectal and/or ovarian cancer (yes/no, defined as one first or second degree relative with cancer), smoking (ever/never), alcohol (ever/never), stage of cancer, grade of cancer, histology and cancer recurrence. This analysis was performed by using chi-squared (χ2) analysis and ORs and 95% CIs were calculated using unconditional logistic regression.
T-tests were used to determine differences in the age of diagnosis of endometrial cancer by genotype. Kaplan Meier survival analysis was used to plot the cumulative survival versus the patient's age of diagnosis of endometrial cancer. By comparing the Kaplan-Meier survival curves for each genotype, we tested if there were differences in the age of diagnosis of endometrial cancer by genotype. The Wilcoxon's test was used to determine the significance of observations from early ages of diagnosis, log-rank test, which gives more weight to later ages and Tarone-Ware test, which is an intermediate of the two other tests were used to examine the homogeneity of the survival curves. The polymorphisms that showed a statistically significant difference between the genotypes and the age of diagnosis of endometrial cancer for all three statistical tests were further examined by a multivariate Cox regression model where a number of specific risk factors were incorporated into the analysis.
The significance levels of all tests were set at p < 0.05 and were two-sided. All statistical analysis was performed with Intercooled STATA 8.2 (Stata Corp., College Station, TX, USA), SPSS Version 15 (SPSS Inc. Chicago, IL, USA) and GraphPad Instat version 3.06 (GraphPad Software, San Diego, CA, USA).