Urinary Total Isothiocyanates and Colorectal Cancer: A Prospective Study of Men in Shanghai, China

Epidemiologic evidence suggests an inverse association between consumption of cruciferous (or Brassica) vegetables and risk of cancer (1). Cruciferous vegetables such as broccoli, cabbage, bok choy, and watercress contain high concentrations of glucosinolates, the precursors of indoles and isothiocyanates (ITCs), which may exert chemoprotective effects on cancer in animals and possibly in humans (2-4).

ITCs influence the metabolism of procarcinogens by both inhibiting phase I (activation) enzymes and inducing phase II (detoxification) enzymes (4-6). Although their exact mechanisms of chemoprotection are not fully understood, ITCs are believed to enhance the detoxification of carcinogens through transcriptional activation of phase II enzymes via the antioxidant/electrophile response element, thereby reducing the exposure of tissues to reactive intermediates of procarcinogens, blocking the initiation of chemical carcinogenesis (4-6). It is speculated that the mechanism for the combination of phase I enzyme inhibition and phase II enzyme activation depends on the particular ITC and carcinogen (2, 7).

Laboratory experiments have shown the chemopreventive effects of cruciferous vegetables in various organ sites and some have shown that ITCs have chemopreventive properties. Tan et al. (8) showed that dietary intake of Chinese cabbage, a cruciferous vegetable, significantly reduced the level of 2-amino-1-methyl-6-phenylimidazo [4,5-b]pyridine-DNA adducts in the rat colon. Chung et al. (9) showed that phenethyl ITC and sulforaphane, major ITCs in watercress and broccoli, respectively, significantly reduced the formation of chemical-induced colonic aberrant crypt foci in rats at the initiation stage. Although the laboratory data about the chemopreventive effects of ITCs on colorectal cancer have been compelling, there are sparse epidemiologic data to attribute the chemopreventive effects of cruciferous vegetables to ITCs.

Earlier, within the prospective Singapore Chinese Health Study cohort, we noted a statistically significant, inverse association between intake of dietary total ITCs based on food frequency questionnaire and colorectal cancer, especially among subjects with absence of glutathione S-transferase M1 and glutathione S-transferase T1 genes (10). In the present study, we examined the relation between total ITCs in urine at baseline (i.e., before cancer diagnosis) and subsequent risk of colorectal cancer in the Shanghai Cohort Study, a prospective cohort investigation of 18,244 men in Shanghai, China, with 16 years of follow-up information. We also examined the association between total urinary ITC levels and risk of colorectal cancer stratified by the time interval from the assessment of baseline ITCs to cancer diagnosis with an attempt to evaluate the potential differential effect of ITCs at different stages of carcinogenesis.

The design of the Shanghai Cohort Study has been previously described in detail (11, 12). Between January 1, 1986 and September 30, 1989, 18,244 men between the ages of 45 and 64 years with no prior history of cancer were enrolled in a prospective study of diet and cancer (representing ∼80% of the eligible subjects). At enrollment, each participant was interviewed in person by a trained nurse using a structured questionnaire asking for information on demographic characteristics, history of tobacco and alcohol use, usual adult dietary habits, and medical history. The Institutional Review Boards at the University of Minnesota and the Shanghai Cancer Institute had approved this study.

At recruitment, each participant was asked if they had ever smoked at least one cigarette per day for 6 months or more. If he answered yes, he was classified as a smoker. Information about current smoking status (yes, no), the number of cigarettes smoked per day, and number of years of smoking over lifetime was obtained from all smokers. For those who quit smoking, the number of years since quitting smoking was recorded. We also asked each participant if he had ever drunk alcoholic beverages at least once weekly for 6 months or more. If he answered yes, we further asked for the typical amount of beer, wine, or spirits consumed per week, separately. One drink was defined as 360 g of beer (12.6 g of ethanol), 103 g of wine (12.3 g of ethanol), or 30 g of spirits (12.9 g of ethanol; ref. 13). Current diet was assessed through a food frequency questionnaire that included 45 food groups/items that represented commonly consumed local foods. The baseline questionnaire did not list any specific cruciferous vegetables (14).

Following the completion of the interview, participants were asked to provide a 10-mL nonfasting blood sample and a single-void urine sample. Biospecimen collection usually occurred between the hours of 5 p.m. and 9 p.m., on average ∼3 h after the last meal. Urine samples were immediately placed in an ice box and transported on the same day to the processing laboratory to be stored at 4°C. The following morning, the urine sample was aliquoted into two vials each with 10 mL and an additional vial with 25 mL. All aliquots of urine samples were originally stored at −20°C until 2001 when the samples were transferred to −70°C freezers until analysis.

Follow-up of the cohort for incident cancers and deaths has been ongoing and involves multiple methods. In addition to annual in-person visits to surviving cohort members since the inception of the cohort in 1986, there are continual, periodic reviews of records of the Shanghai Cancer Registry and Shanghai Municipal Vital Statistics Office, and electronic linkage of these databases with the cohort data set. As of June 30, 2002, the cutoff date for case ascertainment of the present study, the cumulative losses to follow-up were 479 (2.6%) subjects (i.e., their vital status was unable to be determined via routine ascertainment methods).

As of June 30, 2002, a total of 242 incident colorectal cancer cases had been identified (121 cases of colon cancer and 121 cases of rectal cancer). All but 9 of the 242 cases were histopathologically confirmed (96%). The 9 cases (5 colon cancer and 4 rectal cancer) that were not histopathologically confirmed were diagnosed by cytology/surgery and clinical symptoms, or radiology and clinical symptoms.

Control subjects were randomly selected from cohort members who were free of cancer and alive at the time of the cancer diagnosis of the index case patient. For each case, five controls were randomly selected. All five controls were matched to the index case on date of birth (within 2 years), date of biospecimen collection (within 1 month), and neighborhood of residence at recruitment.

Total urinary ITCs was measured as a cyclocondensation product of reaction with 1,2-benzenedithiol by high-performance liquid chromatography as described previously (15, 16). The intraassay coefficient of variation within the batch was <5% and the limit of detection for the urinary ITCs was 0.1 μmol/L. Values at or above 0.1 μmol/L were reported as positive, whereas values below this concentration were reported as negative. Urinary ITC concentrations are stable on long-term storage at −20°C (15). Urinary creatinine concentration was determined on each sample using a modified method as described previously (17). Urine samples on 17 cases and 6 control subjects were depleted after measurements of other urinary biomarkers. Thus, the present study included 225 cases and 1,119 of their matched controls with known ITC values.

The χ² test and the t test were used to compare the distributions of selected demographic, dietary, and other lifestyle factors between cases and controls. Urinary ITC levels were expressed in units of urinary creatinine (μmol/g creatinine) to correct for the varying water contents of individual spot urine samples. We used standard statistical methods for case-control studies. Conditional logistic regression models were used to calculate matched odds ratios (ORs), their corresponding 95% confidence intervals (95% CIs), and P values. Study subjects were grouped by quartiles of total ITCs defined according to the distribution in all controls. Subjects also were categorized into low (the lowest quartile) versus high (the upper three quartiles) levels of urinary total ITCs due to comparable ORs noted between subjects in the upper three quartiles of ITCs. To adjust for potential confounding effects of other covariates, we used conditional logistic regression models that included the following variables as covariates: level of education (no formal school or primary school, junior middle school, senior high school, college and above), body mass index (<18.5, 18.5 to <21.0, 21.0-23.5, 23.5 to <26, 26+ kg/m²), which best characterized the study population (18), number of years of smoking (nonsmokers, <30, 30+), and number of alcoholic drinks per day (nondrinkers, <2, 2 to <4, 4+). Further adjustment for dietary factors, including consumption of total fat, protein, and carbohydrate (percent of total calorie), dietary fiber (g/1,000 kcal), and urinary levels of epigallocatechin and methylated epigallocatechin (in quartile; ref. 19), did not alter the association between urinary total ITCs and colorectal cancer risk; therefore, the dietary factor–adjusted results were not presented in this report.

To examine the potential differential effect of ITCs at different stages of colorectal carcinogenesis, we examined the association between urinary ITC and risk of colorectal cancer stratified by the number of years between study enrollment (i.e., baseline assessment of ITC exposure) and cancer diagnosis. Unconditional logistic regression models were used to estimate ORs and their corresponding 95% CIs. In addition to all covariates listed above, the unconditional logistic regression models included all matching variables, which were age, number of years of sample storage, and neighborhood of residence at recruitment.

Statistical analysis was carried out using the Statistical Analysis System software version 9.1 (SAS Institute). All P values reported are two sided, and those that were <0.05 were considered to be statistically significant.

The present study included 225 colorectal cancer cases and 1,119 matched controls. Of the 225 colorectal cancer cases, 117 (52.00%) were colon cancer cases and 108 (48.00%) were rectal cancer cases. The mean age (±SD) of case patients at cancer diagnosis was 66.37 (±5.85) years. The corresponding figure of matched control subjects at cancer diagnosis of index cases was 66.25 (±5.78). The average time interval between baseline interview (i.e., biospecimen collection) and cancer diagnosis was 8.70 (±3.82) years, ranging from 3 months to 16 years.

Table 1 shows the distributions of selected demographic and lifestyle factors in cases and controls. Cases and controls were comparable in terms of body mass index (kg/m²), level of education, alcohol consumption, and cigarette smoking.

Urinary concentration of ITCs ranged from undetectable to 66.00 μmol/g, with a mean of 2.75 μmol/g in cases, and from undetectable to 57.25 μmol/g, with a mean of 2.32 μmol/g in controls. Overall, increased levels of urinary total ITCs were associated with a modest, statistically nonsignificant reduction in risk of colorectal cancer among total subjects (Table 2). Results were comparable across subsites (colon versus rectum).

Among colorectal cancer cases (n = 47) diagnosed within 5 years of enrollment, 15% were in the lowest quartile of urinary total ITCs, whereas 28% were in the highest quartile of ITCs (Table 3). The comparable figures among colorectal cancer cases (n = 99) diagnosed after 10 or more years after enrollment were 37% and 18%, respectively. Consequently, for colorectal cancer cases diagnosed within 5 years of enrollment, higher (Q2-4) versus lowest (Q1) quartiles of urinary ITCs were associated with a statistically nonsignificant increase in risk of colorectal cancer compared with the lowest quartile of urinary total ITCs (OR, 1.93; 95% CI, 0.85-4.39; P = 0.12); exclusion of the six cases diagnosed within 12 months after enrollment resulted in a stronger positive association between ITCs and cancer risk (OR, 2.40; 95% CI, 0.93-6.21; P = 0.07). In contrast, for colorectal cancer cases diagnosed 5 or more years after enrollment, increased level of ITCs was associated with a statistically significant reduction in risk of colorectal cancer, and this inverse ITC-colorectal cancer association became even stronger for cases occurring 10 or more years after enrollment (P for trend = 0.006). The ORs for urinary ITC levels by duration of follow-up (<5 versus 5+ years) were statistically different from each other (P = 0.02).

We also examined the association between urinary total ITCs and risk of colorectal cancer according to subjects' status in use of tobacco/alcohol. There were no discernible differences in the ITC-colorectal cancer association between these subgroups (data not shown). Urinary green tea polyphenols were inversely and statistically significantly related to colon cancer risk in this Shanghai cohort (19). Further adjustment for these variables did not alter the association between urinary levels of ITCs and colon cancer risk (data not shown). We further examined whether the inverse ITC-colorectal cancer association could be explained by other dietary factors. Further adjustment for dietary total fat, protein, carbohydrate, and fiber did not change the association between urinary total ITCs and colorectal cancer risk (data not shown).

The present study showed a statistically significant, inverse association between urinary total ITC level and colorectal cancer for cases diagnosed at least 5 years after enrollment. This inverse ITC-colorectal cancer association became stronger with longer time intervals between exposure assessment (baseline urine collection) and diagnosis of colorectal cancer. Interestingly, high level of ITCs was associated with a statistically nonsignificant increase in risk of colorectal cancer diagnosed within 5 years of enrollment and this positive ITC-cancer risk association was statistically significantly different from the negative ITC-cancer risk association noted among cases with a longer time period of follow-up. One might argue that baseline measurements in cancer cases diagnosed within 1 year of enrollment were noninformative due to possible dietary changes as a result of clinical symptoms. Thus, we repeated the analysis with exclusion of subjects with less than 1 year of follow-up time; the positive ITC-cancer association in cases diagnosed within 5 years of enrollment actually became stronger [from OR of 1.93 (P = 0.12) to OR of 2.40 (P = 0.07)]. These results suggest that, in a given individual, timing of the ITC exposure may be an important codeterminant of the ITC-colorectal cancer risk association.

The present study confirms our earlier observation in Singapore (10) and supports the hypothesis that ITC protects against colorectal cancer in humans. Experimental studies have shown that dietary intake of Chinese cabbage significantly reduced adduct formation in the colon of rats treated with a known colon carcinogen (8). Dietary ITCs significantly reduced the formation of chemical-induced colonic aberrant crypt foci in rats (9). In humans, high consumption of cruciferous vegetables, the primary source of ITC exposure, has been linked to reduced risk of colorectal cancer (10, 20) and colon adenoma, a recognized precancerous lesion of colorectal cancer (21, 22).

Humans are exposed to ITCs through the consumption of cruciferous vegetables, which are rich in glucosinolates, precursors of ITCs. Different cruciferous vegetables contain different types and levels of glucosinolates (23). Therefore, exposure levels to specific types of ITCs for a given population depend on both the quantities and types of cruciferous vegetables being consumed. Subjects of the present study, on average, consumed 50% more cruciferous vegetables than Singapore Chinese (43 g/1,000 kcal versus 29 g/1,000 kcal per day) and their frequency of consumption is thrice that of Americans in Los Angeles (8 servings versus 2.6 servings per week; ref. 24). The most commonly consumed cruciferous vegetables in Shanghai and Singapore Chinese are bok choy, choy sum, cabbage/wong nga pak, and kai lan (16), whereas for whites in the United States broccoli is by far the most commonly consumed cruciferous vegetable (20). Thus, results on consumption of total cruciferous vegetables in relation to risk of colorectal cancer across different study populations should be interpreted cautiously.

This is the first prospective study to examine the association between a validated biomarker of ITC exposure and colorectal cancer risk. Strengths of this study include the prospective study design. The biospecimens for ITC measurements were collected before the occurrence of colorectal cancer, avoiding the direct effect of disease on assessment of exposure. The long duration of follow-up (up to 16 years) allowed the present study to examine the long-term protective effect of ITC against the development of colorectal cancer. The almost complete follow-up for incident cancer and death minimized the potential bias on results due to the loss to follow-up. The present study also has several potential limitations. It cannot be assumed that ITC levels in a randomly timed, single-void urine sample correlate with usual intake of dietary ITC. However, we had earlier shown among Chinese in Singapore, who share a similar cultural and dietary heritage as our study population in Shanghai, a close and statistically significant correlation between dietary ITCs ascertained from a validated food frequency questionnaire and total ITC levels in a randomly timed spot urine (16). The absence of multiple time point samples does not allow for the correction of regression dilution resulting from the within-person variation in ITC exposure across a wide spectrum of time, which was up to 16 years in the present study. Therefore, the reported inverse ITC-colorectal cancer association could well be underestimated. We did not collect information on intake of specific cruciferous vegetables at baseline, thus prohibiting the examination of dietary ITC in relation to colorectal cancer risk in this report. Although the inverse association between urinary ITC levels and colorectal cancer risk remained statistically significant after adjustment for multiple covariates, the possibility of residual confounding cannot be ruled out completely. Further studies are required to confirm the findings of the present study.

In summary, the present study shows a statistically significant inverse association between levels of total ITCs in urine, collected 10 years before cancer diagnosis, and the risk of developing colorectal cancer. The present study confirms our earlier observation in Singapore Chinese and implicates ITCs as potential chemopreventive agents against the development of colorectal cancer in humans.

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.

We thank Xue-Li Wang, Yue-Lan Zhang, and Jia-Rong Cheng of the Shanghai Cancer Institute for their assistance in data collection and management and the staff of the Shanghai Cancer Registry for their assistance in verifying cancer diagnoses in study participants.