Vitamin D Status and the Risk of Lung Cancer: A Cohort Study in Finland

Lung cancer is the most common cancer worldwide, accounting for 1.2 million new cases annually and responsible for 18% of all cancer deaths (1). Although considerable effort has been devoted to elucidation of the etiology of lung cancer, besides giving up, or preventing smoking (2), the means to prevent it are limited. Recently, the possible beneficial suppressing effect of vitamin D on the development of several cancers, including lung cancer, has received attention (3).

Vitamin D, a steroid hormone that has long been known for its importance for bone health, may be converted from 7-dehydrocholesterol by UV radiation in the skin or ingested through diet and supplements. It is then hydroxylated in the liver to the major circulating form, 25-hydroxyvitamin D [25(OH)D] and further converted into a biologically active form, 1,25(OH)₂D (3).

Increasing evidence from a diversity of scientific approaches suggests that vitamin D have a suppressing effect on the development of various cancers. Animal and in vitro studies have shown that vitamin D can suppress tumor progression by reducing cell proliferation, invasiveness, and angiogenesis and stimulating apoptosis and cell differentiation (3-5). It has also reported to protect against metastasis in various tumor models, including cancer of the lung (3, 5). Ecologic studies have shown lower incidence and mortality rates of certain cancers in populations with high sunlight exposure and thus higher vitamin D levels compared with those with lower ambient sunlight (6). However, epidemiologic evidence of an association between vitamin D and cancer incidence is inconclusive and mainly limited to breast, prostate, and colon cancers (3). The present study extends previous research by evaluating whether serum vitamin D level is associated with the incidence of lung cancer in a cohort study including 7,000 subjects in Finland.

The Mini-Finland Health Survey was carried out over the period 1978 to 1980 by the Mobile Clinic of the Social Insurance Institution in 40 areas of Finland (7). A two-stage random sample (n = 8,000) was drawn from the population register to represent Finnish men and women ages ≥30 years. A total of 7,217 subjects (participation rate, 90%) participated in a health examination. In the present study on vitamin D and lung cancer, eligibility was restricted to those who had no history of cancer at baseline and had a blood sample available for vitamin D analysis, resulting in a cohort of 6,937 subjects.

In the health examination, height and weight were measured and participants were categorized into three groups (<23, 23.0-29.9, and ≥30 kg/m²) according to their body mass index (BMI). Blood samples were also taken and kept frozen at -20°C until 2003, when 25(OH)D concentrations were determined using RIA (DiaSorin) in an accredited laboratory (Population Research Laboratory of National Public Health Institute). The interassay coefficient of the variation of the 25(OH)D determination was 7.80% at the mean level of 47.3 nmol/L (n = 167) and 9.12% at the level of 131.3 nmol/L (n = 135). The proportion of quality-control samples was 13.5%, and the control samples were analyzed in the beginning, middle, and end of the daily analysis series. Serum cotinine concentrations were determined with a method employing a modification of the Nicotine Metabolite RIA kit (Diagnostic Products).

Information on socioeconomic background, symptoms, diseases, medications, and lifestyle was collected via questionnaires and interviews. Educational level was categorized into two groups (low and high) and marital status into four groups (unmarried, married and those who were in a common law marriage, widowed, and divorced). Alcohol consumption categories (0, 1-14.9, and ≥15 g/wk ethanol) were derived from responses to questionnaire items that asked average weekly consumption of beer, wine, and liquor during the preceding month. Both self-reported information on smoking habits and serum cotinine concentration were used to generate the smoking variable. Subjects who reported that they had never smoked or had quit smoking and had a serum cotinine level of ≤100 ng/mL were categorized as nonsmokers. The rest of the subjects were divided into tertiles (cutoff points, 405 and 764 ng/mL) based on their serum cotinine level. As baseline examinations were conducted in different seasons, subjects were referred to as two groups, Summer (June-September) and non-Summer (October-May).

Incident cases of lung cancer were identified through linkage with the Finnish Cancer Registry using a nationwide individual identification number as the identity link. The registry identifies all cancer cases in the country and contains information on the primary site of the tumor, time of diagnosis, malignancy, and histology (8). During the follow-up, which lasted until December 21, 2002 (a maximum of 24 years), 122 primary lung cancers occurred, of which 97 were male and 25 were female lung cancer cases.

The association between serum 25(OH)D level and lung cancer risk was estimated by Cox's proportional hazards model using relative risk (RR) with 95% confidence interval (95% CI). Subjects were stratified into tertiles based on their serum vitamin D level, and a test for trend was done across the tertiles using the likelihood ratio test. The cutoff points for serum D level were 34.0 and 51.0 nmol/L for men and 30.0 and 46.0 nmol/L for women. Marital status, educational level, BMI, alcohol consumption, smoking, age, and season of baseline measurement, as well as sex (when appropriate), were considered as potential confounders. Additionally, to control for reverse causality due to undiagnosed disease at baseline, cancer cases that occurred within the first 4 years of follow-up were excluded from further analyses. Season-specific tertile cutoff points for serum 25(OH)D were also used, as the baseline examinations were conducted in different seasons and vitamin D displays seasonal variability. Effect modification by sex, smoking, age, BMI, season of blood collection, and alcohol consumption was evaluated by including each potential effect modifier in a separate multivariable model. When significant interactions were found, analyses were repeated and stratified at the median for the effect modifier.

The mean (SD) serum 25(OH)D level was 42.9 (19.6) nmol/L, 45.2 (20.4) nmol/L in men and 41.0 (18.7) nmol/L in women. The distribution of 25(OH)D measurement varied according to the background characteristics presented in Table 1. Married and more highly educated subjects were more likely to have higher vitamin D status compared with unmarried, widowed, or divorced and less educated subjects, respectively. Low alcohol consumption and high BMI were associated with a low 25(OH)D level. Smoking was also related to low vitamin D status among men but not among women. Compared with those who had participated in the study during the summer season, those who participated in the study between October and May had a lower serum 25(OH)D level [39.2 (SD, 17.2) versus 59.1 (SD, 21.2) nmol/L].

Serum 25(OH)D level was inversely associated with lung cancer incidence when adjusted with age and sex only (RR for the highest versus lowest tertile, 0.55; 95% CI, 0.35-0.87; P_trend = 0.01; Table 2). Further adjustment for potential confounders, including marital status, educational level, BMI, alcohol consumption, smoking, age, sex, and season of baseline measurement, attenuated the observed association (RR, 0.72; 95% CI, 0.43-1.19; P_trend = 0.22). Exclusion of cases diagnosed within the first 4 years of follow-up did not notably change the results (RR, 0.61; 95% CI, 0.35-1.08; P_trend = 0.10). Results were also similar when season-specific cutoff points were used (data not shown). The association between serum vitamin D and lung cancer occurrence was not significantly modified by smoking (P_interaction = 0.60), BMI (P = 0.74), alcohol consumption (P = 0.09), or season of blood collection (P = 0.98).

There was a statistically significant interaction between vitamin D and the sex of participant (P_interaction = 0.02). Among women, serum 25(OH)D level was inversely associated with lung cancer incidence (multiple-adjusted RR for the highest versus lowest tertile, 0.16; 95% CI, 0.04-0.59; P_trend < 0.001; Table 2). For men, there was no significant association (RR, 1.03; 95% CI 0.59-1.82; P_trend = 0.81). Exclusion of cases diagnosed within the first 4 years of follow-up did not notably change the results (data not shown).

We also observed a significant interaction (P = 0.02) between vitamin D and age at baseline. In age-stratified analysis, a high vitamin D status was associated with a lower risk of lung cancer in participants ages ≤50 years (multiple-adjusted RR for the highest versus lowest tertile, 0.34; 95% CI, 0.13-0.90; P_trend = 0.04) but not among older participants (RR, 0.92; 95% CI, 0.50-1.70; P_trend = 0.79; Table 2).

Findings from the present cohort study suggest that although there is no overall association between 25(OH)D level and lung cancer risk, women and younger participants with higher circulating levels of vitamin D may be at a lower risk of lung cancer. Women in the highest tertile of 25(OH)D had an over 80% lower risk of lung cancer compared with those in the lowest tertile. Correspondingly, younger participants in the top tertile of vitamin D had less than half the risk of lung cancer compared with those who were in the lowest tertile. There was no association between vitamin D and lung cancer risk among men or older participants.

To our knowledge, only one previous study has examined the association between vitamin D status and lung cancer risk (9). In this study, conducted among participants of NHANES III, serum 25(OH)D level was not associated with the risk of lung cancer mortality (RR for the highest versus lowest quartile, 1.14; 95% CI, 0.60-2.18). Previously, Giovannucci et al. (10) estimated that a 25 nmol/L difference in predicted 25(OH)D level was associated with a 20%, although not statistically significant, reduction in lung cancer mortality risk in men.

The present data highlighted a potential interaction between vitamin D and age and sex. The apparent protective association of high serum vitamin D level observed for women and younger participants may have been a chance finding or may have some basis in lung cancer biology/histology and/or reflect differences in the metabolism of vitamin D. The observations that lung cancer incidence is higher among female never smokers compared with male never smokers (11) and the lung cancer risk associated with smoking is higher among women compared with men (12) suggest a possible role for gender-dependent hormones in the development of the disease. Indeed, there is evidence that estrogens may play a role in the development of lung adenocarcinoma in women (13). The findings that estrogen receptors are present in lung tumors and that estrogens may play a role in the activation of vitamin D and expression of the vitamin D receptor are provocative and clearly warrant further exploration (14).

Alternatively, these divergent results may reflect differences in lifestyle related to sex and age that are associated with the risk of lung cancer. Compared with nonsmokers, lower vitamin D status has been observed in smokers (15) and there is some evidence that the anticarcinogenic effect of 25(OH)D is limited to never smokers (16). It is, therefore, possible that the overwhelming contribution of smoking as a cause of lung cancer (2) poses a challenge to detecting the vitamin D-lung cancer association, especially among men and older individuals. The low number of lung cancer cases who did not smoke limited our ability to conduct subgroup analyses. However, no interaction was observed between vitamin D and smoking.

Mean vitamin D values in our study were approximately the same order of magnitude as those reported previously in Finland (17, 18) but somewhat lower than generally observed in other European (19) and American (20) populations. It is estimated that optimal serum 25(OH)D concentrations would be ≥83 nmol/L for colorectal cancer prevention (21) and ≥121 nmol/L for breast cancer prevention (22). These values are substantially higher than cutoff points for the highest vitamin D tertile in our cohort (51 nmol/L for men and 46 nmol/L for women).

An important strength of our study is its prospective design. Serum 25(OH)D level measurements were obtained from a study cohort entirety that derived from a representative sample of the Finnish population. Blood samples were collected up to 24 years before the diagnosis of lung cancer, making the presence of cancer at the time of blood donation improbable. Cancer diagnosis was derived from the nationwide registry, which has shown to have high validity (8). Furthermore, as information on several known and potential risk factors of lung cancer was collected, we were able to account for potential confounding.

The main limitation of the present study is that the vitamin D level was assessed from a single blood sample, which may reflect only recent rather than long-term exposure. The correlation coefficient between two measurements of vitamin D done 3 years apart appears to be moderately high (0.70; ref. 23), suggesting that single serum measurements of this compound could be a useful tool in epidemiologic studies. It is, however, likely that the season of blood collection introduced some measurement error, which we were not able to control, as we did not have a measure of intraindividual variation in serum vitamin D level by season. Including the season of blood sampling as a potential confounder in the model, using it as an effect-modifying factor, or using season-specific cutoff points for vitamin D did not substantially change the results. Despite the fact that vitamin D is relatively stable during storage (24), we cannot rule out the possibility that concentrations might have changed during storage. Finally, due to the low number of female lung cancer cases, caution must be taken in interpreting the results.

In conclusion, although there was no overall association between 25(OH)D level and lung cancer risk, women and younger participants with a higher circulating level of vitamin D were observed to have a lower risk of lung cancer. Although there are experimental data supporting the suppressing effect of vitamin D on the development of lung cancer, large epidemiologic studies from different populations with repeated measurements of vitamin D are warranted to confirm this finding.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.