Repeat polymorphisms in ESR2 and ARand colorectal cancer risk and prognosis: results from a German population-based case-control study

The DACHS study is an ongoing population-based case-control study conducted in southwest Germany, which has previously been described in detail [20, 21]. Briefly, cases were recruited from patients who received in-patient treatment in a hospital of the study region due to a first diagnosis of colorectal cancer. To be eligible, participants had to be at least 30 years old and capable to complete the interview. Controls were randomly selected from lists of population registries and matched according to gender, 5-year age groups and county of residence. Individuals with a history of colorectal cancer were excluded from the study. The present study comprised 746 female and 1052 male incident colorectal cancer patients as well as 732 female and 1078 male controls recruited between January 01, 2003 and December 31, 2007. Ancestry of the participants was homogenous with about 1% being of non-European descent. Patients diagnosed with any other cancer except squamous and basal cell skin cancer before their first diagnosis of colorectal cancer (N =160), patients who died within 30 days after diagnosis and whose death may be related to surgery (N =7), and patients without follow-up information (N =7) were excluded from the survival analyses, which comprised 665 female and 959 male cases. Written informed consent was obtained from all study participants and the study was approved by the ethics committees of the University of Heidelberg and the State Medical Boards of Baden-Wuerttemberg and Rhineland-Palatinate, Germany.

Patients and controls were interviewed in person by trained interviewers using standardized questionnaires. In the interview, information on sociodemographic factors, previous health examinations, medication such as the use of menopausal hormone therapy and non-steroidal anti-inflammatory drugs (NSAIDs), family history of colorectal cancer, and life-style related factors was collected. Additionally, pathology reports and discharge letters were collected. Self-reported use of menopausal hormone therapy was validated for women entering the study before December 31, 2006 [22]. The study participants were asked to provide either a blood sample or a mouthwash sample.

On average three years after diagnosis, a questionnaire was sent to the treating physicians of the patients to collect information on therapy, and newly diagnosed concomitant diseases. A second follow-up questionnaire was mailed to the patients about five years after diagnosis. Vital status and date of death were obtained from the population registries and the cause of death was verified by death certificates obtained from the health authorities in the Rhein-Neckar-Odenwald region. New diagnoses and cancer recurrences were verified through medical records of the attending physicians. In total 665 female and 959 male cases were included in the survival analysis.

Genomic DNA was extracted from blood (98%) or mouthwash samples (2%) using Flexigene Kit 250 (Qiagen, Valencia, CA, USA) and Qiagen Mini Kit (Qiagen, Valencia, CA, USA), respectively. Genomic regions containing the AR and ESR2 microsatellite markers were amplified by polymerase chain reaction (PCR). We used previously reported primers [9]. The PCR reaction mixture consisted of 4 ng genomic DNA in a 5 μl reaction volume containing 1.5 mM magnesium chloride, 1x reaction buffer, 0.20 μM deoxynucleoside triphosphates mixture, 0.18U Platinum-Taq DNA polymerase (Invitrogen, Darmstadt, Germany) and 0.5 μM of each primer. About 2 μl of the 1/10 diluted PCR product was added to 0.5 μl size marker and denaturized for 3 minutes at 95°C. The detection was done using the ABI 3130XL Genetic Analyzer (Applied Biosystems, Carlsbad, CA, USA) and the fluorescently labelled DNA fragments were analysed by size using the GeneMapper 4.0 software (Applied Biosystems, Carlsbad, CA, USA). A random sample of 6.6% was genotyped twice for quality control.

The repeat numbers in AR and ESR2 were averaged for each individual, assuming that each increase in number is related to a constant proportional change in relative risk. For better illustration of the associations with differing levels of repeat number, we categorized the continuous variable of average repeat number into quartiles, according to the distribution in controls. The genotypes were also dichotomized in order to report results comparable to previous studies, based on the median repeat number in controls (<22 repeats/≥22 for AR CAG repeats and <24 repeats/≥24 for ESR2 CA repeats).

To evaluate the ESR2-status of colorectal tumours, ESR2 expression was measured immunohistochemically in tissue microarrays [19]. For this study, we classified samples with less than 10% of the cell nuclei showing strong positive staining or with less than 50% of the nuclei showing weak positive staining as ESR2 negative. ESR2 positivity was defined as weak staining of more than 50% of the cell nuclei or strong positive staining in at least 10% of the cell nuclei.

For survival analysis, follow-up time was calculated as the time between the date of diagnosis and the date of event or censoring. Events of interest were death from any cause (overall survival) and death due to colorectal cancer (disease-specific survival).

All statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, NC, USA). Statistical significance was determined according to the conventional significance-level of α =5%.

Genotype frequencies were assessed in cases and controls separately and tested for deviation from Hardy-Weinberg equilibrium (HWE) in controls using a one degree of freedom Chi-square test. Unconditional logistic regression was used to calculate odds ratios (ORs) as well as confidence intervals (CIs) for colorectal cancer risk associated with genotypes. To test whether gender specific associations are statistically different, we built a multiplicative interaction term between the respective genotype variable and gender and performed a log likelihood ratio test. The models were adjusted for age and county of residence. The inclusion of additional colorectal cancer risk factors did not change the OR estimates substantially (changes <10% in all cases). The following factors were assessed: having a first degree relative diagnosed with colorectal cancer, ever regular use of NSAIDs (2+ times/week, ≥1 year), pack-years of smoking (in categories of 10 pack-years), average lifetime alcohol consumption (g/day in quartiles), average physical activity in the 12 months before diagnosis (in metabolic equivalent of task (MET) hours/week quartiles), ever colorectal endoscopy, ever diagnosis of diabetes, consumption of red meat in last 12 months (low, moderate, high) and body mass index (BMI) ≥5 years before diagnosis/date of interview (in five categories, <23 kg/m², ≥23 to <25 kg/m², ≥25 to <27 kg/m², ≥27 to <30 kg/m², ≥30 kg/m²). In secondary analysis, we evaluated risk associations according to ESR2-status using multinomial logistic regression. Heterogeneity between the risk estimates was assessed using unadjusted logistic regression models in case-case analyses.

Median follow-up time of cases after diagnosis was computed using the reverse Kaplan-Meier method [23]. Regression analyses based on the Cox proportional hazards models were applied to evaluate associations of the polymorphisms with overall and disease-specific survival. The models were determined using backward selection, retaining variables with a p-value of ≤0.2. Validity of the proportional hazards assumption was assessed by including a time-dependent component for each explanatory variable. The models were adjusted for tumour extent (T1, T2, T3, T4), nodal status (N0, N1, N2), distant metastasis (M0, M1), screening detection of colorectal cancer (yes/no), treatment with adjuvant chemotherapy (yes/no), BMI at diagnosis (kg/m², continuous), diagnosis of diabetes after colorectal cancer diagnosis (yes/no), diagnosis of other cancer after colorectal cancer diagnosis (yes/no), age at diagnosis and year of diagnosis. The models were additionally stratified for grade of differentiation (well/moderate, poor/undifferentiated) as this variable showed a time-dependent effect on overall survival. We accounted for left truncation of the follow-up period. The association of the ESR2 and AR repeat polymorphisms with survival according to tumoural ESR2-status was assessed using subgroup analysis. Heterogeneity of ESR2-specific hazard ratios was evaluated using an interaction term between ESR2-status and genotype.