Association between Coffee Consumption and Risk of Prostate Cancer in Japanese Men: A Population-Based Cohort Study in Japan

Coffee, one of the world's most consumed beverages, contains many substances with potential anti-mutagenic and antioxidant properties, including phenolic acids and diterpenes (kahweol and cafestol). These substances may play a protective role against cancer (1). Previous epidemiological studies have suggested that coffee has a favorable effect on the development of some site-specific cancers, such as liver cancer (2), endometrial cancer (3), and oral/pharyngeal cancer (4). Although numerous epidemiological studies have demonstrated that higher coffee consumption is associated with a reduced incidence of prostate cancer, the evidence remains controversial. Meta-analyses of prospective cohort studies have drawn conflicting conclusions, finding either inverse associations (5–9) or no association (10–12).

According to global cancer statistics, prostate cancer is the second most commonly occurring cancer (after lung cancer) among men globally, with 1,414,259 new incident cases diagnosed in 2020 (13). Prostate cancer is well known to be more prevalent in Western countries than Asian countries (14–16). Reflecting this, most epidemiological studies have been performed in Western countries, especially the United States. However, the etiology is not well understood, and few epidemiological studies have examined the impact of coffee consumption on the risk of prostate cancer among Asians (17–19). Given the differences in dietary and lifestyle factors and the incidence of prostate cancer between Asian and Western populations, it is important to conduct epidemiological studies in Asian countries.

This study aimed to elucidate the association between coffee consumption and prostate cancer risk in a Japanese large-scale population-based cohort study.

The Japan Public Health Center-based Prospective Study (JPHC study) is an ongoing Japanese population-based prospective study, as described in detail elsewhere (20), that constitutes two cohorts, Cohort I, which began in 1990, and Cohort II, which began in 1993. Cohort I consists of 61,595 participants (40 to 59 years at baseline) from five public health center areas, namely Ninohe (Iwate), Yokote (Akita), Saku (Nagano), Chubu (Okinawa), and Katsushika (Tokyo), and Cohort II of 78,825 participants (40 to 69 years at baseline) from six public health center areas, namely Mito (Ibaraki), Nagaoka (Niigata), Chuo-higashi (Kochi), Kamigoto (Nagasaki), Miyako (Okinawa), and Suita (Osaka). The study areas covered, while mainly restricted to rural areas, are distributed throughout the north to south of Japan. In total, 140,420 residents from 11 public health center areas participated in the study.

In the present analysis, we identified 68,722 men. Of these, we excluded subjects for the following reasons: Participants in a public health center area (Katsushika, Tokyo) where no cancer incidence data were available (n = 2,919), non-Japanese nationality (n = 31), pre-commencement emigration (n = 115), duplicate enrollment (n = 7), incorrect date of birth (n = 2), and not providing responses to the self-reported questionnaire (n = 15,210). Moreover, those with prostate cancer incidence before follow-up (n = 6), without follow-up information (n = 30), and had missing coffee consumption information (n = 574) and missing covariate information (n = 1,606) were also excluded. Finally, 48,222 subjects were included in this analysis. A flowchart of the subject selection process is provided in Fig. 1. 

This study was approved by the National Cancer Center's institutional review boards (approval number: 2001–021).

The follow-up period for each subject started on the baseline survey response date and continued until one of the following, whichever occurred first: Prostate cancer diagnosis, emigration from the study area, death, or December 31, 2015 (end of follow-up was December 31, 2012 for Osaka and December 31, 2013 for Kochi and Nagasaki). Changes in residence status, for reasons such as survival or emigration from the study area, were identified annually through the residential registry. When participants were lost to follow-up, we censored the latest date on which they were confirmed to be living in the study area.

In the questionnaire, coffee consumption was grouped into the following categories: Almost none, 1 to 2 days a week, 3 to 4 days a week, 1 to 2 cups a day, 3 to 4 cups a day and more than 5 cups a day. For analysis, coffee consumption was regrouped into four categories: Almost none, less than 1 cup a day, 1 to 2 cups a day, and ≥3 cups a day. Validity was assessed using 28- or 14-day dietary records. The correlation coefficient for coffee consumption between the self-administered questionnaire and dietary record was 0.59 among men (21).

Information on lifestyle factors, including coffee and green tea consumption, alcohol drinking status, smoking status, medical history, and family history were obtained from a self-administered questionnaire administered in 1990 to Cohort I and in 1993 to 1994 to Cohort II.

A total of 50,438 subjects returned valid responses (response rate 77%).

Prostate cancer incidence was identified through population-based cancer registries, which link to information on active patient notifications from major local hospitals. Death certificates were used to supplement information on cancer incidence. Information on the means by which cancer was detected, degree of differentiation and local staging were obtained from cancer registries and clinical information. The Gleason score was extracted from medical records into cohort-specific registration forms by either physicians in the hospital or researchers in JPHC Study, who were blinded to subjects' lifestyle information. Cases of prostate cancer were classified according to the International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3). Cancer was detected in patients with prostate cancer by screening (n = 517), subjective symptoms (n = 353), and incidentally while at the hospital for another condition (n = 386).

A total of 1,617 new prostate cancer cases were diagnosed during the study period. We divided subjects' diagnoses into advanced and localized prostate cancer based on local staging, the Gleason score or degree of differentiation (22). Cases in which the cancer had spread from the prostate to other tissues, such as lymph nodes or bone, were classified as advanced cases. Prostate cancer cases in which the cancer was confined within the prostate were defined as localized cases. Among cases that could not be categorized using local staging, those with a Gleason score of 8 to 10 or with poorly differentiated tumors were categorized as advanced, whereas those with a Gleason score ≤7 or with highly or moderately differentiated tumors were categorized as localized cases. We identified 1,099 localized (68.0%), 461 advanced (28.5%), and 57 undetermined (3.5%) cases.

Participant's baseline characteristics are expressed as mean and standard deviation for continuous variables and the percentage for categorical variables. To control for confounding, we directly standardized statistics to the age distribution of our study subjects.

The follow-up time was from the baseline survey response date until censoring (death, relocation from the study area, or the end of the study period) or prostate cancer diagnosis, whichever occurred first. For participants lost to follow-up, the last confirmed date of their presence in the study area was used as the date of censor. Among the study subjects, 5,965 (12.4%) emigrated from the study area and 20 (0.1%) were lost to follow-up during the follow-up period. To estimate the impact of coffee consumption on prostate cancer risk, crude and adjusted hazard ratios (HR) and 95% confidence intervals (CI) were calculated using Cox regression analysis. All models were stratified by participant area at recruitment. Multivariable models were adjusted for subjects' 5-year age group at baseline (40 to 44, 45 to 49, 50 to 54, 55 to 59, 60 to 64, and 65 to 69 years), smoking status (never, former, and current), alcohol drinking status (never, occasional, and regular), body mass index (less than 21, 21 to 22.9, 23 to 24.9, and ≥25 kg/m²), green tea consumption (less than 1 cup a day, 1 to 2 cups a day, 3 to 4 cups a day, and ≥5 cups a day), leisure-time physical activity (almost none, 1 to 3 times a month, 1 to 2 times a week, 3 to 4 times a week, and almost every day), the presence of a medical history of diabetes mellitus (yes or no), the presence of a family history of prostate cancer (yes or no), and a history of screening (by any of the following: blood pressure, electrocardiogram, chest X-ray, gastric photofluorography, and fecal occult blood test) in the past year (yes or no). These variables are either known or suspected risk factors for cancer or have been found to be associated with the risk of prostate cancer (23).

We calculated P values for trend by assigning ordinal variables to the coffee consumption categories. In analyses of localized cancers, we assumed that advanced cancer was censored, and vice versa. To reduce the impact of selection bias, we separately analyzed the effect of coffee intake in subjects whose prostate cancer was detected on the basis of subjective symptoms, screening or incidentally while at the hospital for another condition. Moreover, we conducted sensitivity analysis, in which we excluded prostate cancer cases diagnosed within 3 years after recruitment, to reduce the possibility of reverse causality. Because smoking and alcohol drinking status are some of the most common risk factors for cancer and our previous study demonstrated that alcohol consumption and smoking increase the risk of advanced prostate cancer (24), we also performed subgroup analyses based on smoking status (never or ever smoker) and alcohol drinking status (never or occasional and regular alcohol drinker).

A P value less than 0.05 was considered statistically significant. All statistical analyses were conducted using SAS software 9.4 (SAS Institute, Inc.).

We compared the age-standardized baseline characteristics of subjects among the coffee consumption groups (Table 1). Subjects with higher coffee consumption were younger, more likely to be a smoker and to participate in leisure-time physical activity, and less likely to consume alcohol and develop diabetes mellitus. During a mean follow-up period of 18.8 years, 1,617 participants were newly diagnosed with prostate cancer. Of these, 1,099 and 461 patients had localized and advanced prostate cancer, respectively.

The crude and adjusted HRs for prostate cancer in association with coffee consumption are shown in Table 2. We found no significant associations. The adjusted HRs for the highest compared with lowest coffee consumption group were 1.08 (95% CI, 0.90–1.30), 1.08 (95% CI, 0.84–1.38), and 1.00 (95% CI, 0.67–1.47) for risk of total, localized and advanced prostate cancer, respectively. There was no dose–response relationship.

We also estimated the impact of coffee consumption on prostate cancer risk in subjects whose cancer was detected on the basis of subjective symptoms to limit the effects of selection bias (Table 3). We found no association between coffee consumption and prostate cancer risk (almost none vs. ≥3 cups/d: adjusted HR, 1.19; 95% CI, 0.81–1.74). Likewise, there were no significant associations between coffee consumption and localized or advanced prostate cancer risk (almost none vs. ≥3 cups/d: adjusted HR for localized 1.01; 95% CI, 0.59–1.73; adjusted HR for advanced 1.47; 95% CI, 0.84–2.56). No significant trends were observed. Moreover, analyses that excluded prostate cancer cases diagnosed within the first 3 years after recruitment showed a null association, similar to the main analysis.

We conducted additional analysis of subjects in whom cancer was detected by screening and incidentally while at the hospital for another condition (Supplementary Table S1). There was no significant association between coffee consumption and prostate cancer risk in these groups (almost none vs. ≥3 cups/d: adjusted HR for all prostate cancer 1.10; 95% CI, 0.85–1.41; adjusted HR for localized 1.18; 95% CI, 0.89–1.55; adjusted HR for advanced 0.82; 95% CI, 0.43–1.56). Next, we stratified the data by smoking (never or ever smoker) and alcohol drinking status (never or occasional and regular alcohol drinker; Supplementary Table S2). The results remained similar to our main findings.

Moreover, we evaluated the association between coffee consumption and fatal prostate cancer (defined as death from prostate cancer). In our study, 220 subjects died of prostate cancer. The risk of fatal prostate cancer appeared to decrease as coffee consumption increased. However, no significant association was observed for any coffee consumption category when compared with almost no consumption (less than 1 cup a day, adjusted HR, 0.89; 95% CI, 0.65–1.21; 1 to 2 cups a day, adjusted HR, 0.77; 95% CI, 0.53–1.11; ≥3 cups a day, adjusted HR, 0.72; 95% CI, 0.42–1.24).

We evaluated the impact of coffee intake on prostate cancer incidence in the JPHC study. There were no associations between coffee consumption and prostate cancer incidence among Japanese men. Moreover, we observed no association by prostate cancer stage.

Recent studies have reported that coffee may play a protective role against several types of cancer, including oral and neck, colorectal, breast, and liver cancer (2, 4, 25). Although the majority of meta-analyses of cohort studies have reported an inverse association between coffee intake and prostate cancer risk (5–9), three meta-analyses reported a null association (10–12); the latter result is consistent with the findings of the present cohort study. A large proportion of the studies included in previous meta-analyses were conducted in Western countries, where prostate cancer incidence and mortality are high, and lifestyle differs from that in Asian countries. In contrast, the prostate cancer incidence in Asian countries is relatively low and the effect size is small. Hence, it may be difficult to detect a significant effect of coffee consumption in Asian countries. Four previous meta-analyses that reported that there is sufficient evidence of a protective effect of coffee intake conducted additional analyses by geographic region. Three of these found no significant association in the Asian region (9, 11, 12). Furthermore, epidemiological studies on the effects of coffee on risk of developing prostate cancer in Asian populations are limited. Prospective studies conducted by Iso and colleagues (17) and Allen and colleagues (26) suggest that coffee consumption has no impact on the risk of developing prostate cancer among Japanese men. These findings support our results.

Japanese subjects who drank less coffee tended to consume more tea. Our previous study concluded that green tea consumption dose-dependently decreased the risk of advanced prostate cancer (27). Meta-analyses have demonstrated that higher green tea consumption is linked to reduced prostate cancer risk (28, 29). Hence, the impact of green tea consumption on prostate cancer risk may affect the association between coffee intake and prostate cancer. However, even after adjusting for green tea consumption, we found no significant association between coffee consumption and prostate cancer risk.

Our previous study demonstrated that alcohol consumption and smoking increase the risk of advanced prostate cancer (24). Furthermore, smoking and alcohol consumption are related to coffee consumption. Hence, we further analyzed the association between coffee intake and prostate cancer risk by stratifying the data by smoking or alcohol drinking status to completely remove any potential confounding effects. The results of the stratified analyses were similar to our main findings.

Coffee contains numerous bioactive substances (mainly caffeine, chlorogenic acids and diterpenes) that are associated with various potential health benefits, such as anti-inflammation, anti-diabetes, and anticarcinogenesis activity (30). Several studies have reported that caffeic acid, chologenic acid, and trigonelline have biological activity, including anticancer and anticarcinogenesis activity (31–33). Iwamoto and colleagues (34) evaluated the effect of six major coffee compounds (kahweol acetate, cafestol, caffeine, caffeic acid, chlorogenic acid, and trigonelline hydrochloride) on four prostate cancer cell lines and found that kahweol and cafestol inhibited prostate cancer cell proliferation and migration. Interestingly, however, kahweol and cafestol in coffee have been shown to have strong anticarcinogenic properties in the liver and kidneys, but not in other organs (35). Hence, the protective effects of coffee on prostate cancer risk may be smaller than that against cancers in other organs.

The amount and type of coffee consumed may affect the relationship between coffee and cancer. Zhong and colleagues (6) reported that a 2 cup/d increment in coffee consumption was significantly associated with a 0.93-fold reduction in prostate cancer risk in their meta-analysis. A recent dose–response meta-analysis reported a marked decrease in the risk of prostate cancer among men who consume six or more cups of coffee a day (10). The risk of prostate cancer may thus decrease with increasing coffee intake. Although a Japanese prospective study by Li and colleagues (18) also found a dose–response relationship, we did not observe such a relationship in our study. In fact, we found no significant association between subjects who consumed more than 5 cups of coffee per day and prostate cancer risk in our study (crude HR, 1.02; 95% CI, 0.91–1.15; adjusted HR, 1.03; 95% CI, 0.92–1.17). The average daily intake of coffee was 4.3 cups/d among the UK general population between 1995 and 2000 (36). According to CoffeeResearch.org, Americans consumed 3.1 cups of coffee per day (37). In contrast, Japanese consumed 11.04 cups of coffee per week (equivalent to 1.58 cups per day) in 2000 (38), which is considerably lower than that in the US or UK. Because Asian populations consume less coffee than Western populations, some Asian studies use ≥3 cups of coffee a day as the cutoff value for the highest category (17, 18). Further studies are needed to evaluate the dose–response relationship between coffee consumption and prostate cancer risk in Asians. Moreover, the amount of active compounds in coffee depends on the variety, degree of roasting, brewing method, and serving size (39). Although coffee brew strength can differ substantially between countries, the results of a population-based prospective study showed that there is no association between coffee brewing method and prostate cancer risk (40). Previous Asian studies that reported an association between coffee consumption and prostate cancer risk did not consider the type of coffee or preparation method (17–19). According to the latest statistics from the All Japan Coffee Association, the Japanese population consumed 3.28 cups of roasted and ground coffee, and 5.01 cups of instant coffee a week in 1990 (38). Instant coffee has reportedly low levels of chlorogenic acids, a bioactive component of coffee (41).

Several cohort studies have investigated the relationship between coffee consumption and prostate cancer risk according to cancer stage (18, 42–45). A recent meta-analysis reported the protective effects of coffee consumption on localized prostate cancer (9). IGF-1 and sex hormone-binding globulin (SHBG) are known prognostic parameters of prostate cancer, with a direct association having been observed between IGF-1 or SHBG levels and prostate cancer risk, in particular for localized and low-grade cases (46, 47). Previous studies have reported that caffeine reduces IGF-1 levels (48) and affects the metabolism and levels of SHBG (49). Furthermore, a large epidemiological study reported that coffee consumption may directly influence IGF-1 and SHBG levels (50). Thus, coffee may lead to a lower risk of prostate cancer through its mediation of IGF-1 and SHBG levels. However, the JPHC Study previously reported that SHBG was not associated with prostate cancer in a nested case–control study (51). One possible reason for the discrepancy in findings is that IGF-1 levels are not associated with the risk of prostate cancer in Asian populations (52, 53). Moreover, there is no association between coffee consumption and IGF-1 among middle-aged Japanese (54). Coffee also reportedly has anti-inflammatory properties. A Western study found that elevated high-sensitivity C-reactive protein (CRP) levels are associated with increased prostate cancer risk (55). To our knowledge, no studies have shown an association between CRP and prostate cancer risk in Asian populations. However, Japanese men have significantly lower levels of inflammatory markers than men from the United States (56). Therefore, the anti-inflammatory effect of coffee may be weaker in Asian than in Western populations.

Meanwhile, we observed inverse associations between coffee consumption and fatal prostate cancer, although these were not significant. A previous meta-analysis also showed an inverse association of coffee with fatal and high-grade prostate cancer (11). Given the small number of fatal prostate cancer cases and the large confidence intervals in our study, further investigations with larger sample sizes are needed to obtain more stable estimates for fatal prostate cancer in Asian populations.

This study has several strengths. The study design was prospective and population-based, and our data contained information on prostate cancer stages. Furthermore, the response rate to our questionnaires was relatively high (response rate, 77%). Nevertheless, the study has some limitations. First, data on coffee consumption were obtained via a self-reported questionnaire conducted at baseline. Second, although the study areas examined were distributed throughout the north to south of Japan, our study subjects may not be representative of the entire Japanese population because they mainly lived in rural areas. Coffee consumption may also differ between our study subjects and the overall Japanese population. By assuming that the coffee consumption categories 1–2 cups/d, 3–4 cups/d or more than 5 cups/d are equivalent to 1.5, 3.5 or 5 cups/d, respectively, we calculated that the average amount of coffee consumed was 0.95 cups/d in our population, which is lower than that reported by the Japanese coffee association (1.58 cups/d; ref. 38). In addition, the average proportion of subjects who identified as a current smoker and regular alcohol drinker, and reported leisure-time physical activity ≥1 day per week in our study was 52.6%, 69.0%, and 19.1%, respectively. Compared with the 1993 National Nutrition Survey by the Ministry of Health, Labor, and Welfare, which reported that the average proportion of Japanese who identified as a current smoker and regular alcohol drinker, and reported leisure-time physical activity ≥2 times per week was 44.8%, 46.3%, and 24.3%, respectively, our population comprised a greater proportion of regular alcohol drinkers and current smokers, and lower proportion of those reporting leisure-time physical activity. Regarding incidence of prostate cancer, our study subjects tend to be more likely to develop prostate cancer than the overall Japanese population in 2000 (prostate cancer incidence in this study vs. that in the Japanese population by age group, 55–59 years: 29.4 vs. 14.3, 60–64 years: 22.6 vs. 39.0, 65–69 years: 167.3 vs. 95.3 per 100,000 people, respectively; ref. 57). A Japanese study from 2000 reported the following clinical T staging distributions: T1c, 20.3%; T2a, 21.8%; and T2b, 17.3%, and that nearly 60% of prostate cancer cases were diagnosed with localized prostate cancer (58). Hence, the proportion of patients with localized prostate cancer in our study is higher than that in the overall Japanese population. Third, we did not obtain information on PSA screening from our subjects. A previous study reported that never smokers were more likely to have undergone PSA screening than current smokers (59). In addition, we found that never smokers did not drink more coffee than regular smokers. Therefore, PSA screening may affect both coffee intake and detection of prostate cancer. Fourth, we did not take into account possible changes in coffee consumption over time or misclassification, with non-differential misclassification potentially leading to an underestimation of the results. Finally, we cannot exclude the possibility of potentially unmeasured confounding factors, such as dietary factors.

In conclusion, our findings suggest that increased coffee consumption is not significantly associated with a reduced risk of prostate cancer in Japanese men. Likewise, we observed no association between coffee consumption and prostate cancer stage. Thus, coffee may have no protective effects against prostate cancer among Japanese men.

T. Yamaji reports grants from Ministry of Health, Labor, and Welfare of Japan during the conduct of the study. No disclosures were reported by the other authors.

T. Imatoh: Formal analysis, writing–original draft. N. Sawada: Conceptualization, supervision, writing–review and editing. T. Yamaji: Supervision, writing–review and editing. M. Iwasaki: Supervision, writing–review and editing. M. Inoue: Supervision, writing–review and editing. S. Tsugane: Resources, supervision, project administration, writing–review and editing.

This study was supported by a National Cancer Center Research and Development Fund [23-A-31 (toku), 26-A-2, 29-A-4, 2020-J-4; since 2011)]; grant-in-aid for Cancer Research from the Ministry of Health, Labor and Welfare of Japan (from 1989 to 2010; 19shi-2); and the Ministry of Agriculture, Fishery and Forestry, Japan (JPJ005336). We are indebted to the Aomori, Akita, Iwate, Niigata, Nagano, Ibaraki, Osaka, Kochi, Nagasaki, and Okinawa Cancer Registries for providing their incidence data. JPHC members are listed at the following site (as of April 2020): https://epi.ncc.go.jp/en/jphc/781/8510.html.

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.