Main

Colorectal cancer (CRC) is one of the most common cancers globally with considerable geographic variation (Bonithon-Kopp and Benhamiche, 1999; Ferlay et al, 2010). Environmental factors such as physical inactivity, body and abdominal fatness, red and processed meat and excess alcohol consumption may contribute to CRC incidence (Wynder et al, 1969; Haenszel et al, 1973; Takata et al, 2004; WCRF, 2007; Akhter et al, 2008). Some researchers speculate that the recent high CRC rates in Japan and Korea may be explained by a genetic predisposition, even when consuming a similar diet, caused by Asian’s higher sensitivity and susceptibility to gastric and colon cancer (McMichael and Potter, 1985a; Le Marchand et al, 2002).

The increasing trend of CRC in Japan might also be attributed to the common lifestyle, including the consumption of cereals such as rice, which Japanese living overseas presumably continue to consume. Rice is a staple food in Japan, typically consumed in the form of white rice (Kenko Eiyo Joho Kenkyukai, 2009; Nanri et al, 2010; Science and Technology Agency, 2005; Uchida et al, 2010). There is also some, albeit limited, evidence from the 1970s suggesting large quantities of rice may increase the risk of CRC (Burkitt, 1971; Haenszel et al, 1973). However, the evidence for rice and CRC risk in the general population has been sparse. Therefore, we for the first time analysed a large-scale prospective population-based cohort to understand the association between CRC and rice intake in the Japanese population.

Materials and methods

Study population

The Japan Public Health Center-based prospective Study (JPHC Study) is a 30-year on-going cohort study focusing on cancer, cardiovascular and other lifestyle-related diseases, consisting of ∼140 000 Japanese citizens, aged 40–69 years at baseline (1990–1994), in 11 public health centres throughout Japan. The details of the study design have been described elsewhere (Tsugane and Sobue, 2001). Study participants responding to the 5-year follow-up questionnaire 1995–1999, aged 45–74 years were included in the present study. One public health centre area was excluded because cancer incidence information was not available. We started with a population-based cohort of 133 323 participants. We excluded 245 disqualified subjects, 11 583 subjects who had died, moved out of the study area or were lost to follow-up before the starting point. From these, 98 505 responded to the questionnaire, (response rate 81.1%). The study received approval from the institutional review board of the National Cancer Center, Tokyo, Japan (No. 13-021).

Dietary assessment

In the validated FFQ (Ishihara et al, 2003; Sasaki et al, 2003; Ishihara et al, 2006; Nanri et al, 2010), participants were asked about the frequency of consumption of individual food items as well as representative sizes relative to standard portions. Details regarding carbohydrate-related food items collected in the JPHC Study were described previously (Nanri et al, 2010). Briefly, for rice (japonica rice; round and short grain), participants selected from nine options ranging from <1 bowl per day to ⩾10 bowls per day and rice-bowl portion size: small (110 g of rice), medium (140 g of rice) and large (170 g of rice).

Follow-up and identification of CRC cases

Subjects were followed for move-out, death and occurrence of cancer from the 5-year follow-up survey (around 1995 and 1998) through 31 December 2008.

We identified CRC incidence by hospital records and population-based cancer registries in the study areas. We coded CRC cases as C18-C20 according to the International Classification of Diseases for Oncology, third edition. We further divided CRC into colon (C18.0–C18.5 for proximal colon and C18.6–C18.7 for distal colon) and rectal (C19 and C20) cancer.

Statistical analysis

We excluded participants who had been diagnosed with or reported as having cancer before the starting point (n=4086) or subjects with missing information for main exposure variables: that is, rice, bread, noodle intake (n=1030) or who reported extreme total energy intakes (upper 2.5% or lower 2.5%; n=4666) or missing data. A total of 88 722 participants were enrolled in our analysis. Subjects with missing values for covariates were excluded (n=15 221), resulting in a final sample of 73 501 men (n=34 559) and women (n=38 942).

We calculated person-years of follow-up for each participant from the starting point to the date of CRC diagnosis, date of emigration from the study area, date of death or end of the follow-up (31 December 2008), whichever came first. Using Cox proportional hazards models, we calculated the risk for developing CRC and its anatomic subsites for rice, bread, noodle and cereal categories in quartiles and per 100 g increase by sex, with the lowest consumption category as the reference, adjusted for potential confounding variables (indicated in Tables 2 and 3). The median value of each quartile was included in the trend analysis. Covariates were included based on associations found in our previous studies on CRC. Dietary factors were adjusted by total energy using the residual method (Kipnis et al, 1993; Brown et al, 1994; Willett et al, 1997; Akhter et al, 2008). All analyses were performed with Stata SE 12.1 (StataCorp, College Station, TX, USA).

Table 2 Hazard ratio (HR) and 95% CI of colorectal cancer according to quartiles of rice, bread, noodle and cereal intake (g per day) in the JPHC Study
Full size table
Table 3 Hazard ratio (HR) and 95% CI of colon, rectal, proximal and distal colon cancer according to quartiles of rice intake (g per day) in the JPHC Study
Full size table

Results

During 801 937 person-years of follow-up, we identified 1276 incident cases of CRC (777 for men and 499 for women) (Table 1). Age-adjusted CRC incidence ranged from 20.62 (highest quarter of rice) to 24.10 (lowest) in men and from 11.20 to 12.05 in women. Table 2 presents Hazard ratios (HRs) and 95% confidence intervals (CIs) for CRC incidence according to quartile of rice, bread, noodle and cereal intake among Japanese men and women. The multivariate HRs for the highest compared with the lowest quartile of rice in men was 0.77 (95% CI, 0.56–1.07). The trend analysis, analysed by scores, was not statistically significant. In women, no association was found for rice intake and CRC risk; however, a non-significant trend between increased cereal intake and the risk of CRC was found.

Table 1 Age-adjusted characteristics of study participants at the 5-year follow-up survey according to quartiles of rice intake in the JPHC Study
Full size table

Further stratified analyses showed site-specific results: colon, rectum, proximal and distal colon cancer (Table 3). We noted a non-significant inverse association between the quartiles of rice intake and the risk of rectal cancer (Table 3) in men. In women, a non-significant trend of risk increase by quartile of rice intake and proximal colon cancer (Table 3) was found but not in men. Distal colon cancer showed no association with rice in both sexes (Table 3).

A non-significant risk trend was observed for bread, noodle and cereal intake (see Supplementary Tables).

Sensitivity analyses were performed excluding cases from the first 3 years of observation; because of the outcome, dietary habits might have been influenced by preclinical symptoms; however, results were not substantially different.

Discussion

This is the first population-based study on the association between rice intake and CRC in Japan. Overall, we found that rice intake was not associated with CRC in men and women. Results were similar for CRC subsites.

The effect of rice on CRC is inconclusive, especially in Asian populations, consuming a diet high in white rice. The Shanghai Women’s Health Study observed no association between rice intake and the risk of CRC (Li et al, 2011), which accords with the present result. In the case of Japan, descriptive data regarding the association between rice and CRC are inconsistent. Rice and CRC may be inversely associated as suggested by a decreasing trend of rice intake in Japan in recent decades (Kenko Eiyo Joho Kenkyukai, 2009) concurrent with increasing trends in CRC incidence. However, a positive link is implied by high mortality rates in prefectures where large amounts of rice are consumed (Tanaka et al, 2004; Ministry of Health Labour and Welfare, 2013a, 2013b). This inconsistency may confuse us to understand the association between rice intake and CRC.

Alternatively, rice intake may be a surrogate of a traditional Japanese diet and thus may represent specific dietary patterns associated with CRC risk. In particular, rice, not specifically white rice, may be an important source of dietary fibre, and a surrogate of starch. Fibre may have a protective role (Wakai et al, 2006, 2007; WCRF, 2007); however, results remain elusive. Refined carbohydrates are a possible aetiological factor related to CRC (Cleave, 1956) through two pathways: deficient fibre or bacterial changes – degradation of bile salts to carcinogens and induction of tumours (Burkitt, 1971). Starch may be important in the production of short-chain fatty acids and stool quality (Uchida et al, 2010), possibly affecting the colon and rectum. Starch may enhance colorectal carcinogenesis via hyperinsulinaemia, assumed to be a mechanism in obesity-related carcinogenesis (Giovannucci and Michaud, 2007). Resistant starch may be protective (Uchida et al, 2010).

Some studies indicated that diet-associated risk was more prominent in proximal (right) colon cancer in women (McMichael and Potter, 1983, 1985a, 1985b), whereas rectal cancer was more common among men (Wynder et al, 1969; Bonithon-Kopp and Benhamiche, 1999). Our non-significant results hint at an inverse trend between rice intake and rectal cancer among men, mirror the findings of a Japanese case–control study in the southern part of the country suggesting that rice consumption is inversely associated with distal colon and rectal cancers (Uchida et al, 2010). The difference in risk by sex has also been indicated in previous studies where rice intake was found to be protective regarding risk of colon cancer in men (Wynder et al, 1969; Correa, 1981). In-depth sex- and CRC subsite-specific research is needed to draw more precise conclusions.

The JPHC Study has several strengths such as including a large general population sample with high response and low loss-to-follow-up rate, and exposure data selection before cancer diagnosis to exclude the recall bias. However, there are several limitations in the present study including some measurement error of long-term dietary intake by our FFQ (Ishihara et al, 2003), residual confounding effect that we failed to adjust and influence of improvements in CRC diagnostics on CRC incidence (Holford, 1991; Dubrow et al, 1993; Huang et al, 1999; Minami et al, 2006). In addition, we built quartiles of rice that were imbalanced in the number of subjects owing to several big peaks in the distribution.

Despite such limitations, the prospective JPHC Study suggests that the consumption of rice does not have a substantial impact on the risk of CRC in Japanese men and women. Further large prospective studies, especially in the Asian context are needed to confidently interpret these results.