Both the incidence of and mortality from prostate cancer vary substantially, from high in the USA and certain Western countries to considerably low in Asian countries such as China, Singapore and Japan(Reference Hsing, Tsao and Devesa1). Recently, however, these rates appear to have increased in Japan(Reference Sim and Cheng2, Reference Nakata, Takahashi, Ohtake, Takei and Yamanaka3). Among risk factors for prostate cancer, advanced age and family history appear to be strong determinants(Reference Whittemore, Wu, Kolonel, John, Gallagher, Howe, West, Teh and Stamey4). There is evidence that the incidence is increased notably in Japanese migrants living in Hawaii and is far higher than that among Japanese men living in Japan(Reference Kolonel, Nomura, Hinds, Hirohata, Hankin and Lee5, Reference Cook, Goldoft, Schwartz and Weiss6). Environmental factors such as dietary habits may therefore play an important role in the causation of prostate cancer.
A review article(Reference Terry, Rohan and Wolk7) recently summarized the results from cohort and case–control studies examining the association between fish intake and risk of prostate cancer incidence or mortality, but could not provide conclusive evidence of this association. However, none of the studies quoted showed an increased risk of prostate cancer with high intake of fish, and some have suggested that a high intake of fish reduces the risk of prostate cancer(Reference Augustsson, Michaud, Rimm, Leitzmann, Stampfer, Willett and Giovannucci8, Reference Terry, Lichtenstein, Feychting, Ahlbom and Wolk9). Others have reported that n-3 fatty acids, which are found mainly in fish, may reduce the risk of prostate cancer(Reference MacLean, Newberry and Mojica10, Reference Norrish, Skeaff, Arribas, Sharpe and Jackson11). These studies have been conducted mainly in American and Western countries, however, and the few studies conducted in Japan have yielded inconsistent findings(Reference Allen, Sauvaget, Roddam, Appleby, Nagano, Suzuki, Key and Koyama12, Reference Sonoda, Nagata and Mori13).
Thus we conducted a prospective study in Japan to investigate whether a high intake of fish is protective against the risk of prostate cancer mortality.
Methods
Study population
Study subjects were the participants of the Miyako Study, a cohort study conducted in four areas of Fukuoka Prefecture, Japan. Details of the present cohort study have been described previously(Reference Pham, Fujino, Ide, Kubo, Shirane, Tokui, Mizoue, Ogimoto, Matsuda and Yoshimura14). Briefly, the baseline survey was conducted from 1986 to 1989 via invitations to all inhabitants aged 30 to 79 years living in A town, B village, and selected districts of C city and D town (15 417 subjects in total) to participate in a self-administered questionnaire survey. Response rate was 86·1 %, equivalent to 13270 subjects (5927 men and 7343 women), who constituted the cohort. Baseline characteristics were obtained through the self-administered questionnaire, which included questions on health-related factors including smoking, alcohol, diet, disease history and others.
These 13 270 subjects were then followed for vital status, updated annually through the collaboration of the respective municipal office, until the end of 1999 in one of the study areas and the end of 2003 in the other three. For deaths during the study period, the underlying cause of death was ascertained from death certificates and coded according to the International Classification of Diseases and Injuries, 9th revision (ICD-9). In the present analyses, prostate cancer was defined as code 185 in the ICD-9.
The research protocol of the study was approved by the Ethics Committee of Medical Care and Research of the University of Occupational and Environmental Health, Kitakyushu, Japan.
Exposure assessment
The usual intake of food during the year preceding the start of the study was assessed with the self-administered questionnaire, which included an item on fish intake. Fish intake was assessed at the five levels of ‘twice or more per day’, ‘once a day’, ‘2–4 times per week’, ‘2–4 times per month’ and ‘seldom or never’. For the present analysis, consumption levels were converted into two groups by combining the ‘2–4 times per month’ and ‘seldom or never’ groups into a new ‘low intake’ group; and the ‘twice or more per day’, ‘once a day’ and ‘2–4 times per week’ groups into a ‘high intake’ group.
Exclusion
Among 5927 men at baseline, we excluded forty-eight subjects with any cancer diagnosed before the study baseline and 290 subjects with missing information on fish consumption, leaving 5589 men for analysis.
Statistical analysis
Baseline characteristics of all study subjects were compared according to fish intake using the χ 2 test for categorical variables and ANOVA for continuous variables. We counted the number of person-years of follow-up for each subject from baseline until the date of death, the date of migration from the study area or the end of follow-up, whichever came first. We used Cox proportional hazards regression analysis to estimate the hazard ratios (HR) and their 95 % confidence intervals of prostate cancer mortality according to category of fish consumption, with the ‘low intake’ group considered the referent group. Patterns in Schoenfeld residuals with time were considered to identify possible violation of the proportional hazards model and found to be valid in our analysis(Reference Kleinbaum15).
In addition to age, the following variables were considered as potential confounders and included in the proportional hazards model (all missing data were considered an additional category, termed the ‘unknown’ category, and included in the analyses): smoking (never smoker, ex-smoker and current smoker); alcohol (never drinker, ex-drinker and current drinker); employment status (employed, self-employed and unemployed); history of diabetes (yes or no); living with spouse (yes or no); and study area (town A, village B, city C and town D). In the additional model, we adjusted for the confounding factors above and also adjusted for intake of other food items, including vegetables, fruit and meat. The intake of these food items was categorized into two groups, namely ‘daily intake’ and ‘not daily intake’.
All P values and confidence intervals presented were based on two-sided tests. All statistical analyses were performed using the STATA statistical software package version 9·0 (Stata Corporation, College Station, TX, USA).
Results
During 75 072 person-years of follow-up in 5589 men (average follow-up 13·4 years), twenty-one deaths due to prostate cancer were recorded. Mean age at study baseline of these twenty-one subjects was 67·7 years, ranging from 47 and 79 years old. Among them, fifteen subjects (71 %) died from prostate cancer at the age 75 years or more. The mean follow-up period for these prostate cancer cases was 10·6 years. Distribution of some baseline characteristics differed somewhat between the case (n 21) and non-case subjects (n 5568). For example, proportions of current smokers as well as current drinkers were lower among case than non-case subjects (42·9 % and 66·7 % v. 50·5 % and 80·3 %, respectively). Furthermore, unemployment rate was 33·3 % v. 10·9 %; a history of diabetes was 14·2 % v. 6.5 %; and living with a spouse was 52·4 % v. 74·3 %. Moreover, case subjects reported more frequent consumption of vegetables and fruit than non-case subjects (33·3 % and 52·4 % v. 29·0 % and 42·1 %, respectively), but less frequent consumption of meat (3·1 % v. 7·6 %).
Table 1 shows the baseline characteristics of subjects by level of fish intake. Subjects with a high intake tended to be older (P < 0·01). Distribution of smoking and alcohol habits differed between the two fish intake levels (P < 0·01). Subjects who had a high intake of fish consumed more vegetables and fruits as well as meat (P < 0·01). The proportion of subjects living with a spouse was higher in the high intake group (P < 0·01), but no difference was seen for a history of diabetes (P = 0·90).
Table 1 Baseline characteristicsFootnote * of subjects according to level of fish intake: male participants of the Miyako Study cohort, Fukuoka Prefecture, Japan
* Information was obtained through self-report from the baseline questionnaire.
† P value based on the χ 2 test for categorical variables and ANOVA for continuous variables.
Multivariate-adjusted HR and 95 % CI for prostate cancer mortality by fish intake are shown in Table 2. Results showed a consistent inverse association in the high compared with low intake group, with a multivariate-adjusted HR of 0·11 (95 % CI 0·04, 0·28). Further adjustment for intake of other food items, including vegetables, fruit and meat, did not substantially alter the estimated HR. This model showed a multivariate-adjusted HR of 0·12 (95 % CI 0·05, 0·32) for the group with high fish intake.
Table 2 Hazard ratios (HR) of prostate cancer deaths by level of fish intake: male participants of the Miyako Study cohort, Fukuoka Prefecture, Japan
*Multivariate HR adjusted for age, smoking habit, alcohol habit, history of diabetes, employment status, living with spouse and study area.
†Multivariate HR adjusted for confounding factors in HR* and additionally adjusted for other food items, including vegetable, fruit and meat intakes.
Discussion
In our prospective study in a Japanese population, we found that a high intake of fish may be inversely associated with the risk of prostate cancer death. This association had also been examined by several previous studies. In the large Health Professionals’ Follow-up Study in the USA, a significant inverse association was seen between fish intake and metastatic prostate cancer incidence, but the association was not significant for all prostate cancers or for advanced prostate cancer(Reference Augustsson, Michaud, Rimm, Leitzmann, Stampfer, Willett and Giovannucci8). Further, a cohort study in Sweden showed a significant inverse association between prostate cancer incidence and mortality(Reference Terry, Lichtenstein, Feychting, Ahlbom and Wolk9), and a case–control study in Poland found a significant inverse association with the consumption of smoked and fried fish(Reference Pawlega, Rachtan and Dyba16). In contrast, cohort studies in Hawaii(Reference Severson, Nomura, Grove and Stemmermann17) and The Netherlands(Reference Schuurman, van den Brandt, Dorant and Goldbohm18) showed no preventive effect of fish consumption on the risk of prostate cancer incidence, while the few studies in Japan have been inconsistent: the Life Span Cohort Study in Hiroshima and Nagasaki found that subjects who consumed fish almost daily had a significantly increased risk of prostate cancer incidence(Reference Allen, Sauvaget, Roddam, Appleby, Nagano, Suzuki, Key and Koyama12), whereas a case–control study of 140 patients in Ibaraki and Hokkaido prefectures showed a protective effect of high fish consumption(Reference Sonoda, Nagata and Mori13).
The protective effect of fish against fatal prostate cancer in our study might be explained by several potential mechanisms. First, it may be related to long-chain n-3 fatty acids(Reference Terry, Rohan and Wolk7, Reference Rose19, Reference Larsson, Kumlin, Ingelman-Sundberg and Wolk20), including EPA, docosapentaenoic acid (DPA) and DHA, which are particularly abundant in fish. A nested case–control study in Japan identified a trend to an increase in serum levels of EPA, DPA and DHA with an increasing intake frequency of fish in both men and women(Reference Wakai, Ito, Kojima, Tokudome, Ozasa, Inaba, Yagyu and Tamakoshi21). These fatty acids have been shown to inhibit the biological activity of eicosanoids and androgens, which are known to have a stimulating effect on prostate cancer cell growth(Reference Ghosh and Myers22). Second, fish is also a source of Se(Reference Hagmar, Persson-Moschos, Akesson and Schutz23), an essential trace mineral, which behaves both as an antioxidant and an anti-inflammatory agent(Reference Rayman24). Experimental and epidemiological evidence supports the anticarcinogenic role of Se and reported inverse associations with prostate cancer(Reference Li, Stampfer, Giovannucci, Morris, Willett, Gaziano and Ma25, Reference Duffield-Lillico, Dalkin, Reid, Turnbull, Slate, Jacobs, Marshall and Clark26). Third, fish may contain alternative or additional vitamin D, vitamin E and retinol, which have been associated with a decreased risk of prostate cancer(Reference Chan, Gann and Giovannucci27). These potential factors may act together to cause the protective effect against prostate cancer among those with a high intake of fish.
Although the number of deaths from prostate cancer was relatively low, the strength of the present study is its prospective design. Information on potentially confounding variables and fish intake was collected before the subsequent diagnosis of any cancer, reducing the potential for the information bias inherent in case–control studies. An additional strength was the likely consistency of the findings, which did not change after adjustment for other food items.
Several limitations warrant mention. First, results were derived from the assessment of exposure at baseline only, although fish consumption may have changed over time. However, this problem is common to all large cohort studies. Second, the type and amount of fish consumed were not described, even though only fish high in n-3 fatty acids are likely to lower the risk of prostate cancer(Reference Rose19). Third, we did not assess the family history of prostate cancer, which is also reported to be an important risk factor(Reference Whittemore, Wu, Kolonel, John, Gallagher, Howe, West, Teh and Stamey4).
In conclusion, these results support the hypothesis that a high intake of fish may decrease the risk of prostate cancer death. Given the paucity of studies examining this association, especially in Asian men, these findings require confirmation in further cohort studies.
Acknowledgements
Source of funding:The study was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas C-1 (12218216) and C-2 (12218237) from the Ministry of Education, Science and Culture of Japan. The authors gratefully acknowledge the public health authorities in the municipal offices and public health staff of Miyako Health Center and Munakata Health Center for their valuable cooperation.
Conflict of interest:None declared.
Authorship responsibilities:T.-M.P. conceived the study hypothesis, performed data analyses and wrote the manuscript. T.K., R.I. and S.M. contributed to data analyses and writing of the manuscript. Y.F., N.T., T.M., I.O. and T.Y. were principal investigators responsible for data management of the Miyako Study. All authors contributed to discussion of content and writing of the manuscript.
