Study population
Between August 1995 and June 1997, all inhabitants aged 20 years or older in Nord-Trøndelag county in Norway (n = 92,936) were invited to participate in the Nord-Trøndelag Health Survey ("Helseundersøkelsen i Nord-Trøndelag" = HUNT). In brief, two questionnaires including more than 200 health-related questions were administrated to the participants. The population in Nord-Trøndelag County was ethnically homogeneous (less than 3 percent of subjects were of non-white ethnicity), making it suitable for epidemiological genetic research [14].
Out of the 92,936 invited individuals, a total of 65,291 subjects (70 percent) answered the first questionnaire and participated in the health examination. Details of the non-participants are described elsewhere [14, 15].
In the HUNT 2 biobank a total of 60,241 DNA samples were available. Genotyping of the COMT locus was performed among 3048 individuals. Approximately 70 percent were selected completely at random, whereas the remaining 30 percent included had been randomly selected among an older group of individuals who did not have self-reported diabetes mellitus. This latter group was generated in connection with a planned genetic study on diabetes that needed age-matched controls to a diabetic population. In the present sample the prevalence of self-reported diabetes mellitus was slightly lower among individuals with Val/Val genotype than the other genotypes (2.0% versus 2.3%, p = 0.61), and also in a Finnish study individuals with Val/Val were less likely to be diabetic [16]. In the unidentified data file all participants were mixed, and separate analyses of the group selected completely by random was not possible. Thus, with intention to have a homogeneous study cohort, those with self-reported diabetes (n = 69) in the randomly selected group were eliminated [13], resulting in a population of 2979 individuals for the final analysis. All age groups were included, but because the age-matched controls to a diabetic population as a group are older than the general population, the total group of 2979 individuals without self-reported diabetes was 3.5 years older (mean 52.6 years versus 49.1 years) than the HUNT population as a whole.
Causes of death
We performed mortality follow-up by record linkage using the Norwegian 11-digit birth number (date of birth plus a 5-digit identifier), which is unique to each person residing in Norway, to obtain the date and underlying cause of death kept by Statistics Norway. The HUNT data including genotyping at the COMT locus were linked to the National Cause of Death register, and our report is based on mortality follow-up through the year 2004. We classified the 395 deaths on the basis of the underlying cause of death coded at Statistics Norway by using the 9th (death before 1996), or 10th revisions of the International Classification of Diseases (ICD). To further explore the influence of the Val158Met polymorphism on mortality from cancer and heart diseases on death, we also evaluated the immediate cause of death and up to five contributing causes of death. The underlying cause of death was defined as: a) the disease or injury that initiated the chain of morbid events leading directly to death or b) the external circumstances of the accident or violence that was the cause of the fatal injury [17]. By immediate cause of death was meant the disease, injury or condition directly leading to death and which was caused by the underlying cause of death [17]. Contributing cause of death was meant the condition that may have contributed to the death, but was not the direct causal relation with the disease or condition that has caused the death [17].
Genotyping of the COMT locus
Blood sampling was done whenever subjects attended, and details for the procedure and the HUNT 2 biobank are described elsewhere [14]. DNA for genotyping was extracted from peripheral blood leukocytes from whole blood or blood clots stored in the HUNT 2 biobank, using the Puregene kit (Gentra Systems Inc.) manually or with an Autopure LS (Gentra Systems Inc.). Laboratory technicians were blinded to the data from the National Cause of Death register. COMT genotypes were determined using the LightCycler (Roche Diagnostics Scandinavia AB, Bromma, Sweden) fluorescence resonance energy transfer method [18]. Polymerase chain reaction (PCR) amplifications were performed in 20 μl reactions on a LightCycler System, using 2 μl genomic DNA and the LightCycler-FastStart DNA Master Hybridization Probes kit (Roche Diagnostics Scandinavia AB, Bromma, Sweden). PCR primers (Eurogentec, Seraing, Belgium) and fluorescence labeled probes (PROLIGO, Paris, France) are shown in Table 1. Based on melting curve profiles, participants were classified as having Val/Val, Val/Met, or Met/Met genotypes. Details on PCR and melting curve conditions are available on request.
Ethics
The study was approved by the Regional Committee for Ethics in Medical Research, by the Norwegian Data Inspectorate, and by the Directorate for Health and Social Affairs.
Statistical analysis
Differences between continuous variables were tested with analyses of variance (one-way ANOVA) and between dichotomous variables with the chi-square test (including Fisher's exact test). Analyses used two-tailed estimation of significance, and p < 0.05 was considered to be statistically significant. A Kaplan-Meier method was performed to estimate the cumulative survival and differences between genotypes were tested with a log rank test. We used the Cox proportional hazards model to adjust for age, gender and other potential confounding factors, including smoking status, level of education, use of alcohol, blood pressure, coffee intake and body mass index.
Overall, our sample of 2979 individuals had more than 80 percent power to detect a 3 percent difference in total mortality and mortality by cancer or heart diseases between pooled genotypes with 95 percent certainty.
Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS), version 14.0 (SPSS Inc, Chicago).