ABSTRACT
Background
Polyphenols are antioxidants with promising anticancer properties, but few studies have examined the associations of specific dietary polyphenols with colorectal cancer (CRC) risks or among Black individuals in the United States.
Objectives
We examined the associations between dietary polyphenols and CRC and assessed differences in these associations or polyphenol intakes by subgroups, including race (Black and White), that may contribute to cancer disparities.
Methods
The Southern Community Cohort Study prospectively enrolled individuals from the southeastern United States during 2002–2009, most of whom had a low income or are Black. Validated FFQ data and polyphenol databases were used to estimate polyphenol intakes. Cox proportional hazards models were used to obtain HRs and 95% CIs for the highest compared to the lowest intake quintiles (Qs) of specific polyphenols. Median intakes of quintiles were used to obtain linear trends, and restricted cubic splines were used to obtain nonlinear trends. Subgroup analyses were conducted by cancer site, sex, race, household income, and BMI-defined obesity status.
Results
Among 71,599 participants, the median polyphenol intake was lower for Black individuals (452 mg/day; IQR, 277–672 mg/day) than White individuals (958 mg/day; IQR, 587–1597 mg/day). A significant, inverse, nonlinear association was observed for total polyphenol intake with the CRC risk (HR, 0.57; 95% CI, 0.38–0.86; P = 0.008 comparing 650 mg/day of intake to 0 mg/day). In addition, inverse linear associations were observed for tyrosols and the CRC risk (HRQ5vsQ1, 0.68; 95% CI, 0.50–0.91; P = 0.0014) and for hydroxybenzoic acids and the rectal cancer risk (HRQ5vsQ1, 0.49; 95% CI, 0.29–0.82; P = 0.0007). Associations were consistent by sex, race, income, and BMI.
Conclusions
Increasing intakes of total polyphenols, tyrosols, and hydroxybenzoic acids were associated with decreased CRC or rectal cancer risks, and associations were consistent across subgroups. Differences in polyphenol intakes may contribute to the increased CRC incidence among Black US individuals.
Keywords: polyphenols, diet, colorectal cancer, disparity, SCCS, prospective cohort, chemoprevention
Introduction
Colorectal cancer (CRC) is currently the third most common cancer in the United States. Within the United States, the CRC incidence is higher among older ages, males, some racial/ethnic groups, and in the South and Midwest regions. Notably, non-Hispanic Black individuals bear a disproportionate burden of CRC, with incidence and mortality rates that are roughly 20% and 40% higher, respectively, than those for non-Hispanic White individuals (1). The majority of the CRC risk in adults has been associated with modifiable risk factors, including a suboptimal diet (2,3). The diet-associated CRC burden may be higher among some subgroups, including Black individuals, suggesting that a portion of the cancer disparity could be mitigated through dietary intervention (3). Therefore, there is significant interest in identifying dietary components that may reduce CRC risks and therefore be targeted by interventions.
Polyphenols are the most abundant dietary antioxidants and are found mainly in fruits, vegetables, and beverages such as wine, tea, and coffee. The major polyphenol classes include flavonoids, phenolic acids, stilbenes, and lignans. They are of interest as a means of dietary cancer prevention (4) via multiple mechanisms, including anti-inflammatory properties, free radical scavenging, altering biotransformation of carcinogens, interfering with tumor-signaling pathways, inducing apoptosis of tumor cells, and altering the gut microbiome (5, 6).
Despite basic research that suggests chemopreventive potential, previous cohort studies have generally observed no association between total polyphenol intake or intake of its major classes and CRC risks (7–9). Conversely, some case-control studies have found significant inverse associations with flavonoids and lignans (9). However, few studies have examined the associations of other classes of polyphenols with CRC risks (7, 10, 11). The European Prospective Investigation Into Cancer and Nutrition (EPIC) found an inverse association of phenolic acids and colon cancer in men and a positive association with rectal cancer in women, but no association was found with other polyphenol classes (7). In a Japanese case-control study, coffee polyphenols—predominantly phenolic acids—were inversely associated with the CRC risk (10). To our knowledge, no study has examined the association of polyphenol intake and CRC among non-Hispanic US Black participants.
The Southern Community Cohort Study (SCCS) is an ongoing, prospective cohort study designed to study the causes of cancer disparities and represents a predominantly low-income population at high risk of a suboptimal diet. Additionally, the cohort is predominantly Black. We aimed to evaluate the association between dietary intakes of total, classes, and major subclasses of polyphenols in a predominantly low-income and Black population in the SCCS. Additionally, we evaluated differences in these associations by racial group, household income, sex, and obesity status.
Methods
Study population
The SCCS is a prospective cohort study including roughly 85,000 adults aged 40–79 years at enrollment across 12 southeastern US states. The SCCS was designed to address racial and socioeconomic cancer health disparities. The cohort methods are described in detail in a previous publication (12). Briefly, participants were enrolled between March 2002 and September 2009 either in-person at community health centers (86%) or by mail-based population sampling (14%). All participants provided written informed consent, and the study was approved by the Institutional Review Boards at Vanderbilt University and Meharry Medical College.
Data collection and outcome ascertainment
At enrollment, SCCS participants completed a survey to elicit information on lifestyle, sociodemographic, and health history factors, including racial and ethnic groups. A comorbidity index was derived as a modified version of the Charlson index, and was calculated as the sum of scores for chronic diseases self-reported on the SCCS enrollment questionnaire (13). Derivation of the index, including a list of the diseases and the scores assigned to each disease, has been described previously (14). Participants also completed a validated semi-quantitative FFQ designed to be sensitive to cultural dietary differences in this cohort (15). Estimates of the amount of food in grams per day were calculated for each FFQ food item, and the distribution of foods that contributed to the calculation for each FFQ item was obtained using race-, sex-, and geographic region–specific food intakes (16). The percent weights of each of these foods were calculated for each FFQ food item using USDA recipe databases that contain the percent weights of 1878 unique ingredients contributing to each food variable. Each ingredient was then linked to the USDA Expanded Flavonoid Database (v. 1.1), which contains data from the USDA Database for the Flavonoid Content of Selected Foods (v. 3.2) and the USDA Database for the Isoflavone Content of Selected Foods (v. 2.1) (17). Each ingredient was also merged with the USDA Database for the Proanthocyanidin Content of Selected Foods (v. 2), largely using a manual review to assign each ingredient, taking into account the percent weights of ingredients and retention factors applied in the Expanded Flavonoid Database (18). Likewise, to obtain data on other polyphenol classes, the ingredients were manually assigned to foods within Phenol-Explorer using standard recipes, ingredient lists, and patent information for certain ingredients (19).
USDA polyphenol data are presented in aglycone equivalents and account for food processing. Phenol-Explorer data are generally available as conjugated polyphenols. To combine the data from these 2 sources, Phenol-Explorer data were first converted to aglycone equivalents. Next, if the aglycone was already represented in the USDA data, as was the case for most flavonoids, the USDA values were used in place of Phenol-Explorer data, as USDA polyphenol data are more likely to represent the foods consumed by US participants. Some polyphenols in Phenol-Explorer have values obtained by multiple methods. Some foods contain polyphenols linked to the food matrix itself that are released after hydrolysis (20). For these polyphenols, the values obtained by chromatography after hydrolysis were used. Next, retention factors were used when appropriate to account for changes during the cooking process. Lastly, Phenol-Explorer and USDA data were merged to produce an SCCS-specific database for each demographic stratum and food variable. This allowed for estimation of the polyphenol intake for each participant.
Intake levels were assessed for major classes and subclasses, including total polyphenols, flavonoids [anthocyanins, chalcones, dihydrochalcones, dihydroflavonols, flavanols (including monomers, proanthocyanidins, and flavanol derivatives, such as theaflavins and thearubigins), flavanones, flavones, flavonols, and isoflavonoids], phenolic acids (hydroxybenzoic acids, hydroxycinnamic acids, hydroxyphenylacetic acids, and hydroxyphenylpropanoic acids), lignans, stilbenes, alkylmethoxyphenols, alkylphenols, curcuminoids, furanocoumarins, hydroxybenzaldehydes, hydroxybenzoketones, hydroxycoumarins, hydroxyphenylpropenes, methoxyphenols, naphtoquinones, phenolic terpenes, and tyrosols. Several subclasses made a negligible contribution to polyphenol intake and would be less likely to have plausible biological effects given the low concentrations in the body (Supplemental Table 1). Thus, for this analysis, we focused on total polyphenols, major classes, and subclasses with a median intake in this cohort of 1 mg/day or greater [i.e., flavanols (monomers, proanthocyanidins, and flavanol derivatives, including theaflavins and thearubigins), flavonols, flavanones, anthocyanins, flavones, hydroxycinnamic acids, hydroxybenzoic acids, alkylphenols, alkylmethoxyphenols, and tyrosols].
Incident CRC and vital status data were obtained through linkage with the 12 SCCS state cancer registries and the National Death Index. Participants were followed from enrollment in the cohort through the date of CRC diagnosis, death, loss to follow-up, or 31 December 2016, the date through which all cancer registries reported having complete data. Information on the CRC site is available for a subset of participants in these data sources.
Statistical methods
Polyphenol intakes were log-transformed, adjusted for total energy intake using the residual energy adjustment described previously, and then back-transformed (21). Energy-adjusted polyphenol intake was then grouped into quintiles of intake based on the distribution within the whole cohort. Baseline characteristics of the cohort were compared across the quintiles of total polyphenols and major classes and with incident cancer using chi-square tests and ANOVAs. Factors were considered to be confounders if they were associated with both polyphenol intake and incident cancer or changed point estimates by 10% or more after inclusion in models. HRs and 95% CIs were estimated using Cox proportional hazards models adjusted for age (continuous), sex, race, menopausal status, annual household income (<$15,000, $15,000–$24,999, $25,000–$49,999, ≥$50,000), enrollment source (Community Health Center, general public), BMI (<25, 25–29.9, ≥30 kg/m2), physical activity (continuous, metabolic equivalent hours per day), smoking status and pack years (never smokers, former <20 pack years, former ≥20 pack years, current <20 pack years, current ≥20 pack years), comorbidity index (continuous), total energy intake (continuous, kcal/day), overall diet quality (Healthy Eating Index 2010: ≤50, >50), alcohol intake (nondrinkers; moderate drinkers, defined by the USDA Dietary Guidelines as ≤1 drink/day for women and ≤2 drinks per day for men; more than moderate drinkers), and family history of CRC (no known family history, known family history) (22, 23). The proportional hazards assumption was confirmed for the Cox models by plotting Schoenfield residuals against time and log time and analyzing for flat lines around 0. The linearity assumption was checked by plotting the Martingale residuals against each covariate. Tests for linear trends were assessed using the median value of the quintile as a continuous variable in the Cox models. To assess potential nonlinear relationships, a restricted cubic splines analysis was carried out utilizing a complete case analysis to generate graphs of the HRs at different levels of polyphenol intake. Four knots were utilized and placed at the 5th, 35th, 65th, and 95th percentiles. Stratified analyses by sex, race, household income, and BMI category were conducted to assess potential effect modifications. CRC subsites of the colon and rectum were analyzed individually. A sensitivity analysis was also conducted by limiting the analysis to those who were followed up for more than 2 years after baseline.
To address missingness in covariates, which ranged from 0% to 3%, multiple imputation with 5 iterations of imputation was used. For the imputation step, covariates, polyphenol intake, follow-up time, outcome, and cancer stage (local/advanced) were included. Nonnormal variables with missing data (e.g., physical activity and the comorbidity index) were first log-transformed for the imputation step, then back-transformed. For categorical and binary variables, dummy variables were utilized and left unrounded after imputation, as this is less likely to lead to bias than rounding (24). This was performed using the SAS statistical software (PROC MI). All analyses were performed using SAS OnDemand for Academics (release 9.04) with an alpha level of 0.05.
Results
From the cohort of 84,508, 1023 participants were excluded because they did not have valid baseline data, 5663 were excluded because they had no nutrient data calculated (which includes individuals falling outside of 600–8000 kcal/day) and 6223 were excluded because of incident cancer at baseline, leaving 71,599 participants (Supplemental Figure 1). The median follow-up time was 11.3 years and there were 787 incident cases of CRC, including 574 colon cancers and 181 rectal cancers, with the remaining cases having an unspecified site.
The baseline characteristics of individuals by quintile of total polyphenol intake are shown in Table 1 . In comparison to those with the lowest intakes, individuals with higher intakes were more likely to be older, be White, have a higher income, have a healthier overall diet, and be a person who smokes.
TABLE 1.
Baseline characteristics of participants according to quintiles of energy-adjusted total polyphenol intake in the Southern Community Cohort Study1
| Characteristics | Quintile 1 (Low) | Quintile 2 | Quintile 3 | Quintile 4 | Quintile 5 (High) | P value2 |
|---|---|---|---|---|---|---|
| n | 14,319 | 14,320 | 14,320 | 14,320 | 14,320 | |
| Polyphenol range, mg/day | <278 | 278–453 | 454–661 | 662–1055 | >1055 | |
| Age, y (SD) | 49.7 (7.7) | 50.5 (8.0) | 52.1 (8.6) | 53.6 (9.0) | 53.9 (8.9) | <0.0001 |
| Female, n (%) | 8240 (57.5) | 8158 (57.0) | 8562 (59.8) | 8628 (60.3) | 8672 (60.6) | <0.0001 |
| Premenopausal, n (%) | 3442 (41.8) | 3045 (37.3) | 2716 (31.7) | 2367 (27.4) | 2186 (25.2) | <0.0001 |
| Postmenopausal, n (%) | 4784 (58.1) | 5098 (62.5) | 5837 (68.2) | 6247 (72.4) | 6473 (74.6) | <0.0001 |
| Unknown, n (%) | 14 (0.2) | 15 (0.2) | 9 (0.1) | 14 (0.2) | 13 (0.1) | |
| Race/ethnic group, n (%) | — | — | — | — | — | <0.0001 |
| Black | 12,016 (83.9) | 12,080 (84.4) | 11,467 (80.1) | 8443 (59.0) | 4002 (27.9) | |
| White | 1962 (13.7) | 1826 (12.8) | 2311 (16.1) | 5077 (35.5) | 9309 (65.0) | |
| Other | 305 (2.1) | 357 (2.5) | 481 (3.4) | 730 (5.1) | 914 (6.4) | |
| Unknown | 36 (0.3) | 57 (0.4) | 61 (0.4) | 70 (0.5) | 95 (0.7) | |
| Income, n (%) | — | — | — | — | — | <0.0001 |
| <$15,000/year | 8912 (62.2) | 8378 (58.5) | 7839 (54.7) | 7321 (51.1) | 6446 (45.0) | |
| $15,000–$24,999/year | 3054 (21.3) | 3196 (22.3) | 3188 (22.3) | 2922 (20.4) | 2784 (19.4) | |
| $25,000–$49,999/year | 1567 (10.9) | 1734 (12.1) | 2057 (14.4) | 2184 (15.3) | 2454 (17.1) | |
| >$50,000/year | 606 (4.2) | 853 (6.0) | 1067 (7.5) | 1685 (11.8) | 2413 (16.9) | |
| Unknown | 180 (1.3) | 159 (1.1) | 169 (1.2) | 208 (1.5) | 223 (1.6) | |
| Enrollment site, n (%) | — | — | — | — | — | <0.0001 |
| Community Health Center | 13201 (92.2) | 13009 (90.8) | 12875 (89.9) | 12314 (86.0) | 11190 (78.1) | |
| General public | 1118 (7.8) | 1311 (9.2) | 1445 (10.1) | 2006 (14.0) | 3130 (21.9) | |
| BMI, kg/m2, n (%) | — | — | — | — | — | <0.0001 |
| <25 | 3468 (24.2) | 3729 (26.0) | 3411 (23.8) | 3520 (24.6) | 3818 (26.7) | |
| 25–29.9 | 4036 (28.2) | 4138 (28.9) | 4184 (29.2) | 4266 (29.8) | 4495 (31.4) | |
| ≥30 | 6625 (46.3) | 6287 (43.9) | 6565 (45.8) | 6355 (44.4) | 5854 (40.9) | |
| Unknown | 190 (1.3) | 166 (1.2) | 160 (1.1) | 179 (1.3) | 153 (1.1) | |
| Activity, MET-hr/day, n (SD) | 22.1 (19.1) | 24.0 (20.1) | 23.2 (19.1) | 22.6 (18.7) | 22.0 (17.9) | <0.0001 |
| Unknown, n (%) | 210 (1.5) | 218 (1.5) | 211 (1.5) | 237 (1.7) | 310 (2.2) | |
| Smoking status, n (%) | ||||||
| Never smoker | 5474 (38.2) | 5441 (38.0) | 5335 (37.3) | 5293 (37.0) | 4321 (30.2) | <0.0001 |
| Former, <20 pack years | 1595 (11.1) | 1703 (11.9) | 1928 (13.5) | 2074 (14.5) | 1987 (13.9) | |
| Former, >20 pack years | 882 (6.2) | 874 (6.1) | 1057 (7.4) | 1431 (10.0) | 1786 (12.5) | |
| Current, <20 pack years | 3847 (26.9) | 3810 (26.6) | 3388 (23.7) | 2639 (18.4) | 2127 (14.9) | |
| Current, >20 pack years | 2295 (16.0) | 2206 (15.4) | 2349 (16.4) | 2577 (18.0) | 3719 (26.0) | |
| Unknown | 226 (1.6) | 286 (2.0) | 263 (1.8) | 306 (2.1) | 380 (2.7) | |
| Comorbidity Index, n (SD) | 1.7 (1.4) | 1.7 (1.4) | 1.8 (1.4) | 1.8 (1.4) | 1.9 (1.4) | <0.0001 |
| Unknown, n (%) | 259 (1.8) | 288 (2.0) | 277 (1.9) | 314 (2.2) | 434 (3.0) | |
| Energy intake, kcal, n (SD) | 2250 (1506) | 2921 (1651) | 2792 (1508) | 2607 (1361) | 2302 (1065) | <0.0001 |
| Healthy Eating Index, n (SD) | 51.9 (10.3) | 57.4 (10.8) | 59.0 (11.4) | 61.0 (12.3) | 59.1 (12.9) | <0.0001 |
| Alcohol intake, n (%) | — | — | — | — | — | <0.0001 |
| Nondrinker | 6339 (44.3) | 6071 (42.4) | 6715 (46.9) | 6993 (48.8) | 6845 (47.8) | |
| Moderate drinker | 5255 (36.7) | 4837 (33.8) | 4951 (34.6) | 5074 (35.4) | 5568 (38.9) | |
| More than moderate drinker | 2621 (18.3) | 3302 (23.1) | 2554 (17.8) | 2121 (14.8) | 1747 (12.2) | |
| Unknown | 104 (0.7) | 110 (0.8) | 100 (0.7) | 132 (0.9) | 160 (1.1) | |
Percentages show the percent of observations in that quintile. Abbreviation: MET-hr/day, metabolic equivalent hours per day.
P values were obtained using chi-square tests for categorical variables and ANOVAs for continuous variables.
Median polyphenol intakes, stratified by sex and race, are shown in Supplemental Table 1. In general, the polyphenol intake was half as much for Black individuals (452 mg/day; IQR, 277–672 mg/day) as for White individuals (958 mg/day; IQR, 587–1597 mg/day). The greatest contributors to the total polyphenol intake were phenolic acids (53%) and flavonoids (43%; Supplemental Figure 2), and the greatest sources were coffee (38%) and tea (20%; Supplemental Figures 3 and 4). While the major contributors were relatively constant across racial and sex strata, White participants obtained significantly higher median polyphenol intakes from coffee (455 mg/day; IQR, 11–1357 mg/day) and tea (94 mg/day; IQR, 6–532 mg/day) compared with Black participants [coffee, 25 mg/day (IQR, 0–304 mg/day); tea, 14 mg/day (IQR, 5–149 mg/day); P < 0.0001].
Several measures of polyphenol intake were associated with decreased CRC risks, although the relationships were not always linear. Total polyphenol intake was associated with a significantly decreased CRC risk [Table 2; P nonlinear trend = 0.007; 3rd quintile (HR, 0.73; 95% CI, 0.58–0.91) and 4th quintile (HR, 0.72; 95% CI, 0.57–0.91)] compared to the lowest quintile of intake. The U shape of this significant nonlinear trend is shown in Figure 1, where an intermediate intake of roughly 650 mg/day was associated with a decreased risk (HR, 0.57; 95% CI, 0.38–0.86; P = 0.008 comparing 650 mg/day of intake to 0 mg/day) that returned to a null association with a higher intake. Similar statistically significant, nonlinear trends were observed for phenolic acids (P nonlinear trend = 0.015) and hydroxycinnamic acids (P nonlinear trend = 0.04), where an intermediate intake was associated with a decreased risk and increasing intake beyond this had a null association. Flavonols were associated with a decreased CRC risk (P nonlinear trend = 0.04; 2nd quintile HR, 0.75; 95% CI, 0.60–0.93). The nonlinear trend of flavonols was L-shaped, with the initial increasing intake associated with a decreased risk that plateaued with further increases in intake. Hydroxybenzoic acids were associated with a decreased risk (3rd quintile HR, 0.80; 95% CI, 0.64–1.00; P linear trend = 0.049). Tyrosols were strongly associated with a decreased risk [P nonlinear trend = 0.04; 5th quintile (HR, 0.68; 95% CI, 0.50–0.91; P linear trend = 0.0014)]. The nonlinear trend of tyrosols was L-shaped, with an initial increasing intake associated with a decreased risk that plateaued with further increases in intake. No other polyphenols were associated with the CRC risk. In a sensitivity analysis excluding individuals with less than 2 years of follow-up, the relationships were mostly similar. The exception was anthocyanin intake, which became statistically significantly associated with an approximately 30% reduced CRC risk (for quintile 5 compared with quintile 1: P linear trend = 0.03; HR, 0.73; 95% CI, 0.56–0.94).
TABLE 2.
Associations between polyphenol intakes and colorectal cancer risk in the Southern Community Cohort Study1
| Quintile 1 (Lowest) | Quintile 2 | Quintile 3 | Quintile 4 | Quintile 5 (Highest) | P linear trend2 | P nonlinear trend3 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Polyphenol | HR | Cases | HR (95% CI) | Cases | HR (95% CI) | Cases | HR (95% CI) | Cases | HR (95% CI) | Cases | ||
| Total polyphenols | 1.00 (ref) | 179 | 0.82 (0.66–1.02) | 155 | 0.73 (0.58–0.91) | 148 | 0.72 (0.57–0.91) | 145 | 0.88 (0.69–1.12) | 160 | 0.65 | 0.007 |
| Flavonoids | 1.00 (ref) | 157 | 1.03 (0.82–1.28) | 169 | 0.88 (0.70–1.11) | 151 | 0.89 (0.70–1.13) | 155 | 0.95 (0.75–1.21) | 155 | 0.63 | 0.18 |
| Flavanols | 1.00 (ref) | 161 | 0.97 (0.78–1.21) | 160 | 0.94 (0.75–1.18) | 160 | 0.88 (0.70–1.11) | 152 | 0.96 (0.76–1.21) | 154 | 0.77 | 0.40 |
| Proanthocyanidins | 1.00 (ref) | 161 | 0.99 (0.79–1.23) | 161 | 0.87 (0.69–1.10) | 145 | 0.91 (0.73–1.15) | 157 | 0.89 (0.71–1.13) | 163 | 0.35 | 0.38 |
| Flavanol monomers | 1.00 (ref) | 154 | 1.06 (0.85–1.32) | 171 | 0.93 (0.74–1.17) | 155 | 0.97 (0.77–1.22) | 161 | 0.95 (0.75–1.21) | 146 | 0.60 | 0.38 |
| Flavanol derivatives | 1.00 (ref) | 169 | 0.90 (0.72–1.12) | 154 | 0.97 (0.78–1.21) | 161 | 0.95 (0.76–1.18) | 159 | 0.92 (0.73–1.15) | 144 | 0.65 | 0.47 |
| Flavonols | 1.00 (ref) | 182 | 0.75 (0.60–0.93) | 139 | 0.82 (0.66–1.03) | 155 | 0.85 (0.68–1.06) | 162 | 0.82 (0.65–1.03) | 149 | 0.38 | 0.04 |
| Flavanones | 1.00 (ref) | 152 | 0.90 (0.72–1.14) | 137 | 1.06 (0.84–1.32) | 166 | 0.96 (0.76–1.21) | 159 | 0.94 (0.75–1.18) | 173 | 0.65 | 0.40 |
| Anthocyanins | 1.00 (ref) | 167 | 0.91 (0.73–1.13) | 163 | 0.83 (0.66–1.05) | 156 | 0.78 (0.62–1.00) | 149 | 0.78 (0.61–1.00) | 152 | 0.08 | 0.15 |
| Flavones | 1.00 (ref) | 161 | 0.90 (0.72–1.13) | 156 | 0.80 (0.63–1.02) | 145 | 0.78 (0.61–0.99) | 145 | 0.90 (0.70–1.14) | 180 | 0.51 | 0.13 |
| Phenolic acids | 1.00 (ref) | 176 | 0.82 (0.66–1.03) | 149 | 0.85 (0.68–1.06) | 158 | 0.76 (0.61–0.96) | 153 | 0.85 (0.67–1.08) | 151 | 0.46 | 0.015 |
| Hydroxycinnamic acids | 1.00 (ref) | 164 | 0.91 (0.72–1.14) | 156 | 0.90 (0.72–1.13) | 162 | 0.83 (0.66–1.04) | 159 | 0.86 (0.67–1.10) | 146 | 0.31 | 0.04 |
| Hydroxybenzoic acids | 1.00 (ref) | 179 | 0.96 (0.77–1.18) | 172 | 0.80 (0.64–1.00) | 144 | 0.84 (0.67–1.05) | 151 | 0.80 (0.64–1.01) | 141 | 0.049 | 0.21 |
| Lignans | 1.00 (ref) | 160 | 0.90 (0.72–1.13) | 157 | 0.96 (0.76–1.21) | 162 | 1.04 (0.82–1.32) | 170 | 0.90 (0.69–1.17) | 138 | 0.73 | 0.22 |
| Stilbenes | 1.00 (ref) | 149 | 0.90 (0.70–1.16) | 151 | 0.98 (0.74–1.30) | 172 | 0.99 (0.74–1.33) | 171 | 0.93 (0.70–1.23) | 144 | 0.70 | 0.10 |
| Other polyphenols | ||||||||||||
| Alkylphenols | 1.00 (ref) | 155 | 0.94 (0.75–1.18) | 147 | 0.95 (0.75–1.20) | 150 | 0.92 (0.73–1.17) | 148 | 1.11 (0.88–1.40) | 187 | 0.09 | 0.06 |
| Alkylmethoxyphenols | 1.00 (ref) | 166 | 1.06 (0.85–1.32) | 168 | 0.95 (0.76–1.19) | 157 | 0.83 (0.66–1.04) | 143 | 0.97 (0.76–1.23) | 153 | 0.55 | 0.06 |
| Tyrosols | 1.00 (ref) | 174 | 1.09 (0.88–1.35) | 192 | 0.99 (0.79–1.25) | 163 | 0.87 (0.69–1.11) | 143 | 0.68 (0.50–0.91) | 115 | 0.0014 | 0.04 |
The models were adjusted for sex, race, menopausal status, household income, enrollment site, BMI category, physical activity, smoking status, pack years, comorbidity index, energy intake, healthy eating index, alcohol intake, and family history of colorectal cancer.
P linear trend values were obtained using the medians of quintiles as a continuous variable in the Cox proportional hazards models.
P nonlinear trend values were obtained from a likelihood ratio test by comparing the overall fit of models where polyphenol intake was modeled with restricted cubic splines to the fit of models where polyphenol intake was a continuous variable.
FIGURE 1.
Associations between colorectal cancer risk and energy-adjusted polyphenol intake compared with minimal intake using restricted cubic splines analysis (n = 65,915), where nonlinearity was significant for intakes of: (A) total polyphenol (P nonlinear trend = 0.007), (B) phenolic acid (P nonlinear trend = 0.015), (C) hydroxycinnamic acid (P nonlinear trend = 0.04), (D) flavanol (P nonlinear trend = 0.04), and (E) tyrosol (P nonlinear trend = 0.04) with 95% CIs demonstrated by the shaded area.
Stratified analysis
The relationship between polyphenol intakes and the cancer risk varied somewhat by colon or rectal cancer (Table 3). The colon cancer risk was inversely and significantly associated with tyrosol intake (HR, 0.63; 95% CI, 0.45–0.89; P linear trend = 0.002; P nonlinear trend = 0.04), total polyphenols [3rd quintile (HR, 0.74; 95% CI, 0.57–0.96) and 4th quintile (HR, 0.74; 95% CI, 0.56–0.97); P nonlinear trend = 0.01], and phenolic acids (4th quintile HR, 0.75; 95% CI, 0.58–0.98; borderline significant P nonlinear trend = 0.06). For rectal cancer, hydroxybenzoic acids were strongly related to a 50% reduced risk (P linear trend = 0.0007), and flavonoids were related to a 40% reduced risk (P-trend 0.04).
TABLE 3.
Associations between polyphenol intakes and colorectal cancer risk in the Southern Community Cohort Study stratified by cancer site, racial group, and income1
| Quintile 1 (lowest) | Quintile 2 | Quintile 3 | Quintile 4 | Quintile 5 (highest) | P linear trend2 | P non-linear trend3 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Stratum and Polyphenol | HR (95% CI) | Cases | HR (95% CI) | Cases | HR (95% CI) | Cases | HR (95% CI) | Cases | HR (95% CI) | Cases | ||
| Colon | ||||||||||||
| Total polyphenols | 1.00 (ref) | 125 | 0.81 (0.63–1.06) | 110 | 0.74 (0.57–0.96) | 108 | 0.74 (0.56–0.97) | 107 | 0.96 (0.72–1.27) | 124 | 0.75 | 0.010 |
| Flavonoids | 1.00 (ref) | 105 | 1.10 (0.84–1.44) | 122 | 0.90 (0.68–1.20) | 105 | 0.99 (0.75–1.31) | 118 | 1.11 (0.84–1.47) | 124 | 0.47 | 0.17 |
| Flavonols | 1.00 (ref) | 120 | 0.85 (0.65–1.10) | 105 | 0.91 (0.70–1.19) | 115 | 0.92 (0.71–1.20) | 119 | 0.93 (0.71–1.22) | 115 | 0.94 | 0.13 |
| Anthocyanins | 1.00 (ref) | 125 | 0.82 (0.63–1.07) | 112 | 0.79 (0.60–1.04) | 114 | 0.73 (0.55–0.97) | 108 | 0.75 (0.56–0.99) | 115 | 0.13 | 0.23 |
| Flavones | 1.00 (ref) | 115 | 0.88 (0.67–1.15) | 108 | 0.79 (0.60–1.05) | 102 | 0.84 (0.63–1.11) | 112 | 0.94 (0.71–1.25) | 137 | 0.95 | 0.45 |
| Phenolic acids | 1.00 (ref) | 126 | 0.77 (0.59–1.01) | 103 | 0.85 (0.66–1.11) | 119 | 0.75 (0.58–0.98) | 112 | 0.87 (0.66–1.16) | 114 | 0.78 | 0.06 |
| Hydroxybenzoic acids | 1.00 (ref) | 120 | 0.95 (0.73–1.23) | 115 | 0.88 (0.68–1.15) | 108 | 0.95 (0.73–1.24) | 117 | 0.94 (0.72–1.22) | 114 | 0.78 | 0.24 |
| Lignans4 | 1.00 (ref) | 115 | 0.90 (0.69–1.18) | 114 | 0.96 (0.74–1.26) | 120 | 1.02 (0.77–1.34) | 125 | 0.85 (0.62–1.15) | 100 | 0.47 | 0.23 |
| Stilbenes | 1.00 (ref) | 110 | 0.88 (0.65–1.19) | 107 | 1.00 (0.73–1.39) | 128 | 1.02 (0.73–1.43) | 129 | 0.85 (0.61–1.18) | 100 | 0.30 | 0.07 |
| Other polyphenols | ||||||||||||
| Tyrosols | 1.00 (ref) | 127 | 1.07 (0.83–1.37) | 139 | 0.97 (0.74–1.26) | 118 | 0.86 (0.65–1.14) | 106 | 0.63 (0.45–0.89) | 84 | 0.002 | 0.04 |
| Rectum | ||||||||||||
| Total polyphenols | 1.00 (ref) | 46 | 0.75 (0.48–1.18) | 35 | 0.66 (0.41–1.05) | 32 | 0.74 (0.47–1.18) | 35 | 0.77 (0.47–1.28) | 33 | 0.53 | 0.18 |
| Flavonoids | 1.00 (ref) | 42 | 0.97 (0.62–1.49) | 43 | 0.78 (0.49–1.25) | 36 | 0.74 (0.46–1.19) | 34 | 0.61 (0.36–1.02) | 26 | 0.04 | 0.50 |
| Flavonols | 1.00 (ref) | 54 | 0.53 (0.33–0.84) | 29 | 0.56 (0.35–0.88) | 30 | 0.65 (0.42–1.02) | 35 | 0.66 (0.42–1.05) | 33 | 0.30 | 0.07 |
| Anthocyanins | 1.00 (ref) | 36 | 0.97 (0.60–1.55) | 37 | 1.03 (0.63–1.67) | 40 | 0.91 (0.55–1.51) | 35 | 0.89 (0.53–1.51) | 33 | 0.59 | 0.30 |
| Flavones | 1.00 (ref) | 36 | 1.08 (0.69–1.70) | 44 | 0.81 (0.49–1.32) | 34 | 0.58 (0.34–1.00) | 25 | 0.92 (0.56–1.51) | 42 | 0.42 | 0.04 |
| Phenolic acids | 1.00 (ref) | 42 | 0.86 (0.55–1.37) | 35 | 0.85 (0.53–1.37) | 34 | 0.79 (0.49–1.26) | 33 | 1.02 (0.62–1.67) | 37 | 0.73 | 0.06 |
| Hydroxybenzoic acids | 1.00 (ref) | 49 | 1.06 (0.71–1.58) | 52 | 0.65 (0.41–1.03) | 31 | 0.58 (0.36–0.95) | 27 | 0.49 (0.29–0.82) | 22 | 0.0007 | 0.21 |
| Lignans | 1.00 (ref) | 35 | 0.89 (0.54–1.45) | 33 | 1.14 (0.70–1.84) | 39 | 1.36 (0.84–2.22) | 43 | 1.13 (0.65–1.95) | 31 | 0.35 | 0.36 |
| Stilbenes | 1.00 (ref) | 31 | 1.06 (0.62–1.81) | 37 | 1.07 (0.59–1.94) | 38 | 1.05 (0.56–1.97) | 35 | 1.52 (0.85–2.70) | 40 | 0.07 | 0.20 |
| Other polyphenols | ||||||||||||
| Tyrosols | 1.00 (ref) | 42 | 1.03 (0.66–1.62) | 42 | 1.10 (0.69–1.74) | 41 | 0.98 (0.60–1.60) | 35 | 0.66 (0.34–1.27) | 21 | 0.16 | 0.45 |
| Black | ||||||||||||
| Total polyphenols | 1.00 (ref) | 154 | 0.79 (0.62–1.01) | 130 | 0.76 (0.60–0.97) | 132 | 0.73 (0.56–0.95) | 98 | 0.94 (0.69–1.28) | 59 | 0.67 | 0.013 |
| Flavonoids | 1.00 (ref) | 125 | 0.97 (0.75–1.24) | 130 | 0.81 (0.62–1.06) | 113 | 0.91 (0.70–1.19) | 121 | 0.99 (0.74–1.33) | 84 | 0.86 | 0.11 |
| Flavonols | 1.00 (ref) | 145 | 0.77 (0.60–0.99) | 115 | 0.79 (0.61–1.01) | 118 | 0.80 (0.62–1.03) | 115 | 0.84 (0.63–1.11) | 80 | 0.36 | 0.012 |
| Anthocyanins | 1.00 (ref) | 120 | 0.92 (0.71–1.19) | 117 | 0.87 (0.66–1.14) | 115 | 0.73 (0.55–0.98) | 99 | 0.88 (0.66–1.16) | 122 | 0.45 | 0.17 |
| Flavones | 1.00 (ref) | 117 | 0.80 (0.61–1.05) | 105 | 0.75 (0.57–0.99) | 103 | 0.76 (0.58–1.01) | 108 | 0.84 (0.64–1.11) | 140 | 0.57 | 0.18 |
| Phenolic acids | 1.00 (ref) | 146 | 0.85 (0.66–1.08) | 128 | 0.86 (0.67–1.10) | 128 | 0.79 (0.62–1.02) | 119 | 0.89 (0.64–1.22) | 52 | 0.53 | 0.08 |
| Hydroxybenzoic acids | 1.00 (ref) | 130 | 0.97 (0.76–1.23) | 136 | 0.83 (0.64–1.08) | 116 | 0.83 (0.64–1.08) | 110 | 0.80 (0.60–1.06) | 81 | 0.08 | 0.33 |
| Lignans | 1.00 (ref) | 148 | 0.92 (0.72–1.17) | 140 | 0.93 (0.72–1.20) | 119 | 0.98 (0.75–1.28) | 101 | 0.94 (0.69–1.28) | 65 | 0.88 | 0.29 |
| Stilbenes | 1.00 (ref) | 117 | 0.86 (0.64–1.16) | 107 | 0.98 (0.71–1.35) | 124 | 1.04 (0.74–1.44) | 123 | 0.99 (0.72–1.35) | 102 | 0.83 | 0.32 |
| Other polyphenols | ||||||||||||
| Tyrosols | 1.00 (ref) | 149 | 1.11 (0.88–1.41) | 153 | 0.94 (0.72–1.22) | 106 | 0.84 (0.63–1.11) | 91 | 0.69 (0.48–0.99) | 74 | 0.013 | 0.11 |
| White | ||||||||||||
| Total polyphenols | 1.00 (ref) | 19 | 1.12 (0.60–2.09) | 22 | 0.52 (0.25–1.06) | 13 | 0.67 (0.38–1.18) | 41 | 0.84 (0.50–1.40) | 95 | 0.99 | 0.19 |
| Flavonoids | 1.00 (ref) | 27 | 1.25 (0.75–2.08) | 34 | 1.14 (0.68–1.94) | 34 | 0.78 (0.45–1.35) | 29 | 0.98 (0.62–1.56) | 66 | 0.49 | 0.40 |
| Flavonols | 1.00 (ref) | 29 | 0.75 (0.43–1.31) | 23 | 1.07 (0.64–1.78) | 33 | 1.16 (0.71–1.89) | 44 | 0.88 (0.56–1.39) | 61 | 0.77 | 0.25 |
| Anthocyanins | 1.00 (ref) | 42 | 0.83 (0.53–1.29) | 41 | 0.72 (0.45–1.16) | 38 | 0.83 (0.52–1.34) | 44 | 0.47 (0.27–0.81) | 25 | 0.013 | 0.27 |
| Flavones | 1.00 (ref) | 38 | 1.13 (0.72–1.77) | 43 | 1.09 (0.68–1.74) | 41 | 0.84 (0.51–1.39) | 34 | 1.01 (0.60–1.70) | 34 | 0.67 | 0.38 |
| Phenolic acids | 1.00 (ref) | 23 | 0.79 (0.43–1.46) | 19 | 0.86 (0.48–1.54) | 25 | 0.70 (0.40–1.23) | 29 | 0.82 (0.51–1.32) | 94 | 0.84 | 0.09 |
| Hydroxybenzoic acids | 1.00 (ref) | 44 | 0.93 (0.59–1.49) | 32 | 0.63 (0.37–1.07) | 22 | 0.87 (0.55–1.38) | 37 | 0.82 (0.54–1.23) | 55 | 0.49 | 0.14 |
| Lignans4 | 1.00 (ref) | 13 | 0.99 (0.51–1.90) | 32 | 1.00 (0.54–1.88) | 57 | 1.02 (0.54–1.93) | 88 | 0.87 | 0.20 | ||
| Stilbenes | 1.00 (ref) | 26 | 1.05 (0.60–1.84) | 43 | 0.95 (0.50–1.78) | 44 | 0.84 (0.43–1.63) | 42 | 0.70 (0.36–1.36) | 35 | 0.15 | 0.07 |
| Other polyphenols | ||||||||||||
| Tyrosols | 1.00 (ref) | 23 | 0.91 (0.53–1.55) | 35 | 1.00 (0.60–1.66) | 53 | 0.80 (0.47–1.35) | 44 | 0.56 (0.30–1.02) | 35 | 0.02 | 0.43 |
| <$15,000/y | ||||||||||||
| Total polyphenols | 1.00 (ref) | 109 | 0.89 (0.67–1.18) | 95 | 0.79 (0.59–1.06) | 88 | 0.82 (0.61–1.11) | 87 | 0.86 (0.62–1.19) | 75 | 0.46 | 0.09 |
| Flavonoids | 1.00 (ref) | 98 | 1.08 (0.82–1.44) | 103 | 0.79 (0.58–1.08) | 76 | 1.01 (0.75–1.36) | 93 | 1.03 (0.76–1.40) | 84 | 0.78 | 0.20 |
| Flavonols | 1.00 (ref) | 114 | 0.69 (0.51–0.92) | 75 | 0.82 (0.61–1.09) | 88 | 0.91 (0.69–1.21) | 97 | 0.82 (0.60–1.11) | 80 | 0.68 | 0.02 |
| Anthocyanins | 1.00 (ref) | 114 | 0.91 (0.69–1.20) | 101 | 0.70 (0.51–0.95) | 76 | 0.71 (0.52–0.97) | 76 | 0.84 (0.62–1.14) | 87 | 0.43 | 0.23 |
| Flavones | 1.00 (ref) | 105 | 0.87 (0.65–1.16) | 92 | 0.82 (0.61–1.11) | 87 | 0.79 (0.58–1.08) | 83 | 0.76 (0.55–1.04) | 87 | 0.10 | 0.11 |
| Phenolic acids | 1.00 (ref) | 104 | 0.87 (0.65–1.16) | 88 | 0.93 (0.70–1.24) | 97 | 0.80 (0.60–1.07) | 90 | 0.84 (0.61–1.17) | 75 | 0.37 | 0.19 |
| Hydroxybenzoic acids | 1.00 (ref) | 110 | 0.93 (0.71–1.23) | 95 | 0.80 (0.60–1.08) | 81 | 0.86 (0.64–1.15) | 86 | 0.89 (0.67–1.20) | 82 | 0.50 | 0.33 |
| Lignans | 1.00 (ref) | 104 | 0.93 (0.70–1.23) | 100 | 0.93 (0.69–1.26) | 90 | 1.00 (0.73–1.36) | 89 | 0.93 (0.66–1.31) | 71 | 0.87 | 0.23 |
| Stilbenes | 1.00 (ref) | 89 | 0.96 (0.69–1.34) | 97 | 1.01 (0.70–1.44) | 103 | 1.01 (0.69–1.48) | 95 | 1.00 (0.70–1.43) | 70 | 0.95 | 0.43 |
| Other polyphenols | ||||||||||||
| Tyrosols | 1.00 (ref) | 108 | 1.14 (0.87–1.50) | 117 | 1.01 (0.75–1.36) | 89 | 0.95 (0.70–1.29) | 82 | 0.62 (0.41–0.93) | 58 | 0.006 | 0.15 |
| $15,000–$24,999/y | ||||||||||||
| Total polyphenols | 1.00 (ref) | 31 | 1.03 (0.63–1.70) | 35 | 1.00 (0.61–1.64) | 35 | 0.93 (0.55–1.57) | 29 | 1.00 (0.57–1.76) | 27 | 0.91 | 0.49 |
| Flavonoids | 1.00 (ref) | 31 | 0.85 (0.51–1.43) | 29 | 1.02 (0.62–1.70) | 35 | 0.93 (0.55–1.56) | 32 | 1.02 (0.60–1.74) | 30 | 0.78 | 0.33 |
| Flavonols | 1.00 (ref) | 35 | 0.74 (0.44–1.22) | 27 | 0.79 (0.48–1.31) | 29 | 0.98 (0.60–1.59) | 35 | 0.98 (0.59–1.63) | 31 | 0.62 | 0.33 |
| Anthocyanins | 1.00 (ref) | 28 | 0.95 (0.56–1.62) | 29 | 1.43 (0.85–2.38) | 43 | 0.95 (0.53–1.68) | 28 | 1.10 (0.62–1.95) | 29 | 1.00 | 0.05 |
| Flavones | 1.00 (ref) | 27 | 1.25 (0.75–2.11) | 34 | 0.78 (0.44–1.40) | 23 | 0.99 (0.56–1.74) | 29 | 1.49 (0.87–2.57) | 44 | 0.13 | 0.47 |
| Phenolic acids | 1.00 (ref) | 34 | 0.98 (0.60–1.60) | 34 | 1.12 (0.69–1.80) | 38 | 0.79 (0.47–1.31) | 28 | 0.77 (0.43–1.38) | 23 | 0.24 | 0.35 |
| Hydroxybenzoic acids | 1.00 (ref) | 36 | 0.94 (0.59–1.51) | 35 | 0.81 (0.49–1.34) | 30 | 0.90 (0.55–1.48) | 31 | 0.79 (0.47–1.33) | 25 | 0.40 | 0.34 |
| Lignans | 1.00 (ref) | 36 | 0.62 (0.36–1.05) | 23 | 1.05 (0.65–1.69) | 39 | 1.09 (0.66–1.79) | 37 | 0.79 (0.44–1.43) | 22 | 0.99 | 0.35 |
| Stilbenes | 1.00 (ref) | 33 | 1.16 (0.67–1.99) | 36 | 1.18 (0.64–2.16) | 34 | 1.11 (0.58–2.11) | 30 | 1.02 (0.55–1.90) | 24 | 0.83 | 0.25 |
| Other polyphenols | ||||||||||||
| Tyrosols | 1.00 (ref) | 42 | 0.88 (0.55–1.42) | 35 | 0.68 (0.40–1.15) | 26 | 0.90 (0.54–1.49) | 32 | 0.74 (0.39–1.40) | 22 | 0.49 | 0.008 |
| ≥$25,000/y | ||||||||||||
| Total polyphenols | 1.00 (ref) | 35 | 0.51 (0.30–0.86) | 24 | 0.43 (0.26–0.73) | 25 | 0.39 (0.23–0.66) | 27 | 0.74 (0.46–1.20) | 69 | 0.74 | 0.001 |
| Flavonoids | 1.00 (ref) | 25 | 1.00 (0.59–1.69) | 35 | 0.92 (0.55–1.56) | 39 | 0.57 (0.33–1.00) | 29 | 0.72 (0.42–1.23) | 61 | 0.11 | 0.13 |
| Flavonols | 1.00 (ref) | 31 | 0.93 (0.57–1.52) | 35 | 0.87 (0.53–1.42) | 37 | 0.59 (0.34–1.00) | 26 | 0.73 (0.44–1.20) | 60 | 0.11 | 0.22 |
| Anthocyanins | 1.00 (ref) | 23 | 0.85 (0.48–1.48) | 31 | 0.75 (0.42–1.32) | 35 | 0.81 (0.46–1.42) | 44 | 0.51 (0.28–0.91) | 48 | 0.02 | 0.46 |
| Flavones | 1.00 (ref) | 27 | 0.69 (0.40–1.19) | 29 | 0.72 (0.42–1.24) | 32 | 0.60 (0.34–1.04) | 32 | 0.80 (0.47–1.36) | 74 | 0.82 | 0.24 |
| Phenolic acids | 1.00 (ref) | 35 | 0.58 (0.34–0.98) | 25 | 0.45 (0.26–0.78) | 22 | 0.62 (0.38–1.00) | 34 | 0.81 (0.50–1.32) | 69 | 0.42 | 0.007 |
| Hydroxybenzoic acids | 1.00 (ref) | 30 | 1.02 (0.63–1.65) | 39 | 0.76 (0.46–1.26) | 33 | 0.73 (0.44–1.22) | 33 | 0.62 (0.37–1.05) | 52 | 0.03 | 0.32 |
| Lignans | 1.00 (ref) | 19 | 1.13 (0.63–2.03) | 31 | 1.00 (0.55–1.80) | 32 | 1.17 (0.65–2.09) | 43 | 0.96 (0.52–1.78) | 59 | 0.75 | 0.50 |
| Stilbenes | 1.00 (ref) | 26 | 0.48 (0.25–0.92) | 18 | 0.74 (0.39–1.40) | 33 | 0.78 (0.40–1.51) | 43 | 0.67 (0.35–1.27) | 60 | 0.58 | 0.37 |
| Other polyphenols | ||||||||||||
| Tyrosols | 1.00 (ref) | 24 | 1.27 (0.75–2.15) | 39 | 1.37 (0.81–2.32) | 44 | 0.74 (0.41–1.35) | 26 | 0.82 (0.43–1.57) | 62 | 0.14 | 0.44 |
The models were adjusted for sex, race, menopausal status, household income, enrollment site, BMI category, physical activity, smoking status, pack years, comorbidity index, energy intake, healthy eating index, alcohol intake, and family history of colorectal cancer, unless a particular variable was being stratified. Only polyphenol groups that were significantly associated with the colorectal cancer risk in the overall models of Table 2 were included in the stratified analysis. Participants with missing values for the strata variable were not included in that stratified analysis.
P linear trend values were obtained using the medians of quintiles as a continuous variable in the Cox proportional hazards models.
P nonlinear trend values were obtained from a likelihood ratio test by comparing the overall fit of models where polyphenol intake was modeled with restricted cubic splines to the fit of models where polyphenol intake was a continuous variable.
Lignans were analyzed by quartiles for White individuals instead of quintiles due to low colorectal cancer case numbers in the lowest quintiles of intake among White individuals.
No striking differences in the associations between polyphenol intakes and the CRC risk were observed by racial group, alcohol use, household income, or smoking status (data not shown), nor for BMI or sex (Supplemental Table 2), nor were most tests for interaction statistically significant, except for stilbenes and race (P interaction = 0.049); however, no significant association for stilbene intake with the CRC risk was observed for either Black or White individuals. Notably, the statistically significant inverse relationship with tyrosol intake was present among both men and women, Black and White participants, and overweight participants. In an exploratory analysis, interactions were tested between polyphenols and menopausal status, and significant interactions were observed for flavanol monomers (P interaction = 0.03), flavanol derivatives (P interaction = 0.03), and alkylphenols (P interaction = 0.02; data not shown in table). For postmenopausal women, alkylphenols were strongly associated with a 50% increased CRC risk (P linear trend = 0.0019), flavanol monomers were associated with a decreased CRC risk with a nonsignificant trend [5th quintile (HR, 0.67; 95% CI, 0.47–0.96); P linear trend = 0.07], and flavanol derivatives were not significantly associated with the CRC risk. For premenopausal women, alkylphenols and flavanol monomers were not significantly associated with the CRC risk, but flavanol derivatives were associated with an increased CRC risk [5th quintile (HR, 2.03; 95% CI, 1.06–3.89); P linear trend = 0.03], although the number of CRC cases across quintiles of flavanol derivatives for premenopausal women was low (15–26).
Discussion
This large, prospective cohort study, conducted in a predominantly Black US population, found significant, inverse, nonlinear associations of total polyphenol and phenolic acids intakes with CRC risks. In addition, intakes of tyrosols were associated with substantial reductions in CRC risks, regardless of the cancer location, race, sex, income, and BMI.
To our knowledge, this is the first observational study to observe an inverse association of tyrosols and the CRC risk (7). After adjusting for alcohol, the primary contributors to tyrosol intake were olives and olive oil (19). Tyrosols have antioxidant, antiproliferative, and proapoptotic effects on cancer cells; can reach high concentrations in the colon; and may contribute to decreased CRC risks in diets supplemented with olive oil (25–27). The observed nonlinear association may explain the null associations reported in studies like EPIC, where the median tyrosol intake was higher than that in SCCS (3.5 mg/day compared with 1 mg/day, respectively), as the CRC risk plateaus with increasing intake (7). In the SCCS, Black individuals had 30% less tyrosol intake than White individuals; this difference was associated with a 6.5% higher CRC risk. The tyrosols in 2 teaspoons of extra-virgin olive oil would be sufficient to achieve the levels associated with 24% and 19% reductions in CRC risks for Black and White individuals, respectively.
The lack of a linear association of total polyphenol intake in this study is consistent with the results from EPIC; however, unlike EPIC, an increased rectal cancer risk in women was not observed (7). However, we found significant nonlinear associations with intermediate intakes for CRC risks, which have been previously observed (10). The association with total polyphenols is likely driven by phenolic acids, which accounted for 53% of total polyphenols. Phenolic acids were also related to CRC risks in a U-shaped manner. In this study, coffee intake accounted for 72% of phenolic acid intake. Thus, we would expect similar associations between coffee intake and CRC; however, there was no association between coffee intake and the CRC risk in an SCCS nested case-control study (28). A meta-analysis of prospective cohort studies only found an inverse association between CRC risks and coffee consumption in US studies (29). A U-shaped relationship with coffee polyphenols was reported in the Fukuoka study, similar to the phenolic acid relationship in our study (10). This differs from the EPIC results, which reported a suggestive association with rectal cancer in women and an inverse association with colon cancer in men (7). Phenolic acids have anticancer properties, including their antioxidant capacity, antiproliferative effects, proapoptotic effects, and interference with vascular endothelial growth factor-induced angiogenesis (25, 26).
Hydroxybenzoic acids were associated with decreased risks for rectal cancer and CRC. This differs from the suggestive increased rectal cancer risk found among women in the EPIC study (7). While few epidemiologic studies have examined the association of hydroxybenzoic acids and CRC risk, an in vitro study suggested hydroxybenzoic acids may slow cell proliferation and serve as a common pathway for reduced CRC risk with aspirin and flavonoids slowing cell proliferation (27).
The null associations for stilbenes, lignans, and most flavonoid subclasses are largely consistent with results from prior studies (7–9, 30). While stilbenes have chemopreventive effects in vivo, the null association in SCCS may be due to low intakes (31). Our finding of a decreased rectal cancer risk with total flavonoids differs from the increased risk in women and null association in men participating in EPIC (7). Among flavonoid subclasses, anthocyanins were associated with a decreased CRC risk after excluding cases diagnosed within 2 years of enrollment. This is consistent with findings from recent meta-analyses (7, 8, 32). The stronger association after excluding early cases may be explained by the exclusion of cases related to factors prior to enrollment or may indicate a latency effect.
On exploratory analysis, flavanol derivatives (theaflavins and thearubigins) were associated with an increased CRC risk among premenopausal women, which has not been evaluated in previous studies. This finding is limited by small case numbers and is contrary to in vitro evidence of anticancer properties of flavanols (7, 8, 33, 34). Alkylphenols (mostly from grains) showed a strong trend for an increased CRC risk among postmenopausal women. This was surprising given the evidence that alkylphenols exhibit anticancer properties in vitro (30). The findings in postmenopausal women may derive from alkylphenol's hormonal effects, as some have been found to suppress estrogen biosynthesis and bind to the estrogen receptor (35, 36). Reduced estrogen synthesis could lead to an increased CRC risk, given the known protective association between estrogen and CRC in postmenopausal patients (37). Alternatively, these findings may be due to chance since they were observed in exploratory analyses.
The strengths of this study include the prospective cohort design, robust sample size, and ability to represent a large geographic area in the United States with some of the highest rates of CRC. In addition, this study was conducted among a predominantly low-income population, with substantial inclusion of Black participants. A major advantage of this study was the matching of nearly 2000 foods to polyphenol entries that captured both gender and racial intake differences; additionally, we used aglycone equivalents in the analysis, which may be more representative of the polyphenol content after digestion (38). Lastly, linkage of the cohort allowed for nearly complete follow-up, and those who were lost to follow-up were accounted for by censoring at the age of last follow-up. Limitations include the possibility of misclassification due to the use of an FFQ (39), lack of assessment for changes in diet or potential confounders over time, and residual confounding. It is possible that differences in genetic variants, gut microbiota, environment, or other dietary factors that were not measured in this study may contribute to differences in polyphenol absorption or metabolism, and thus explain some of the observed differences between racial/ethnic groups in this study (40). There is potential for confounding by other bioactive compounds, such as coffee bioactives (e.g., caffeine, chlorogenic acids) (41). However, hydroxycinnamic acids, which were more closely related to coffee consumption than total phenolic acids, were more weakly associated with CRC than total phenolic acids. Additionally, we found significant associations with polyphenol groups derived from only a few sources (e.g., tyrosols), as well as from many sources (e.g., hydroxybenzoic acids), so it is not possible to draw conclusions about individual foods having greater importance for these associations. Polyphenols within a subclass may have varied biochemical properties, such that analyses by classes and subclasses of polyphenols could overlook the associations of particularly bioactive compounds; however, this would be expected to attenuate the associations.
In conclusion, we found relationships between intakes of several polyphenols and reduced CRC risks, particularly for tyrosol intake. For all polyphenol classes for which an association—linear or nonlinear—was observed, except flavones, Black participants had lower intakes than White participants. The differences between sexes were less drastic, but males had a roughly 5% lower total polyphenol intake. These findings may contribute to the disparity of CRC risks among Black and male individuals in the United States compared to other racial groups or females, respectively. As the lowest quintiles of polyphenol intake had higher percentages of Black individuals than the higher quintiles, modest increases in polyphenol intakes may be sufficient to reduce CRC risks for these individuals by achieving intake levels associated with decreased CRC risks.
Supplementary Material
Acknowledgments
The authors’ responsibilities were as follows – LTF, QD, MJS: designed the research; LTF, HM: analyzed data or performed the statistical analysis; and all authors: conducted the research, wrote the paper, and read and approved the final manuscript. The authors report no conflicts of interest.
Notes
Research reported in this publication was supported by the National Cancer Institute of the NIH under award number U01CA202979. Additional support was provided by the Vanderbilt University School of Medicine Project Funding.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Supplemental Tables 1 and 2 and Supplemental Figures 1–4 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.
Abbreviations used: CRC, colorectal cancer; EPIC, European Prospective Investigation Into Cancer and Nutrition; Q, quintile; SCCS, Southern Community Cohort Study.
Contributor Information
Landon T Fike, Vanderbilt University School of Medicine, Nashville, TN, USA.
Heather Munro, International Epidemiology Field Station, Vanderbilt University Medical Center, Nashville, TN, USA.
Danxia Yu, Vanderbilt University School of Medicine, Nashville, TN, USA; Department of Medicine, Division of Epidemiology, Vanderbilt Epidemiology Center, Vanderbilt University, Nashville, TN, USA; Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN, USA.
Qi Dai, Vanderbilt University School of Medicine, Nashville, TN, USA; Department of Medicine, Division of Epidemiology, Vanderbilt Epidemiology Center, Vanderbilt University, Nashville, TN, USA; Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN, USA.
Martha J Shrubsole, Vanderbilt University School of Medicine, Nashville, TN, USA; International Epidemiology Field Station, Vanderbilt University Medical Center, Nashville, TN, USA; Department of Medicine, Division of Epidemiology, Vanderbilt Epidemiology Center, Vanderbilt University, Nashville, TN, USA; Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN, USA.
Data Availability
Data described in the manuscript, code book, and analytic code will be made available upon request pending application to and approval by the Southern Community Cohort Study.
References
- 1. Siegel RL, Miller KD, Sauer AG, Fedewa SA, Butterly LF, Anderson JC, Cercek A, Smith RA, Jemal A. Colorectal cancer statistics, 2020. CA Cancer J Clin. 2020;70(3):145–64. [DOI] [PubMed] [Google Scholar]
- 2. Islami F, Goding Sauer A, Miller KD, Siegel RL, Fedewa SA, Jacobs EJ, McCullough ML, Patel AV, Ma J, Soerjomataram Iet al. Proportion and number of cancer cases and deaths attributable to potentially modifiable risk factors in the United States. CA Cancer J Clin. 2018;68(1):31–54. [DOI] [PubMed] [Google Scholar]
- 3. Zhang FF, Cudhea F, Shan Z, Michaud DS, Imamura F, Eom H, Ruan M, Rehm CD, Liu J, Du Met al. Preventable cancer burden associated with poor diet in the United States. JNCI Cancer Spectrum. 2019;3(2):pkz034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Koch W. Dietary polyphenols—Important non-nutrients in the prevention of chronic noncommunicable diseases. A systematic review. Nutrients. 2019;11(5):1039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Alam MN, Almoyad M, Huq F. Polyphenols in colorectal cancer: Current state of knowledge including clinical trials and molecular mechanism of action. Biomed Res Int. 2018;2018:1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ding S, Xu S, Fang J, Jiang H. The protective effect of polyphenols for colorectal cancer. Front Immunol. 2020;11:1407. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Zamora-Ros R, Cayssials V, Jenab M, Rothwell JA, Fedirko V, Aleksandrova K, Tjønneland A, Kyrø C, Overvad K, Boutron-Ruault Met al. Dietary intake of total polyphenol and polyphenol classes and the risk of colorectal cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Eur J Epidemiol. 2018;33(11):1063–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Chang H, Lei L, Zhou Y, Ye F, Zhao G. Dietary flavonoids and the risk of colorectal cancer: An updated meta-analysis of epidemiological studies. Nutrients. 2018;10(7):950. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Grosso G, Godos J, Lamuela-Raventos R, Ray S, Micek A, Pajak A, Sciacca S, D'Orazio N, Rio DD, Galvano F. A comprehensive meta-analysis on dietary flavonoid and lignan intake and cancer risk: Level of evidence and limitations. Mol Nutr Food Res. 2017;61(4):1600930. [DOI] [PubMed] [Google Scholar]
- 10. Wang ZJ, Ohnaka K, Morita M, Toyomura K, Kono S, Ueki T, Tanaka M, Kakeji Y, Maehara Y, Okamura Tet al. Dietary polyphenols and colorectal cancer risk: The Fukuoka colorectal cancer study. World J Gastroenterol. 2013;19(17):2683–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Bahrami A, Jafari S, Rafiei P, Beigrezaei S, Sadeghi A, Hekmatdoost A, Rashidkhani B, Hejazi E. Dietary intake of polyphenols and risk of colorectal cancer and adenoma–A case-control study from Iran. Complement Ther Med. 2019;45:269–74. [DOI] [PubMed] [Google Scholar]
- 12. Signorello LB, Hargreaves MK, Steinwandel MD, Zheng W, Cai Q, Schlundt DG, Buchowski MS, Arnold CW, McLaughlin JK, Blot WJ. Southern community cohort study: Establishing a cohort to investigate health disparities. J Natl Med Assoc. 2005;97(7):972–9. [PMC free article] [PubMed] [Google Scholar]
- 13. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 1987;40(5):373–83. [DOI] [PubMed] [Google Scholar]
- 14. Sonderman JS, Munro HM, Blot WJ, Tarone RE, McLaughlin JK. Suicides, homicides, accidents, and other external causes of death among Blacks and Whites in the Southern Community Cohort Study. PLoS One. 2014;9(12):e114852. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Buchowski MS, Schlundt DG, Hargreaves MK, Hankin JH, Signorello LB, Blot WJ. Development of a culturally sensitive food frequency questionnaire for use in the Southern Community Cohort Study. Cell Mol Biol (Noisy-le-grand). 2003;49(8):1295–304. [PubMed] [Google Scholar]
- 16. Signorello LB, Munro HM, Buchowski MS, Schlundt DG, Cohen SS, Hargreaves MK, Blot WJ. Estimating nutrient intake from a food frequency questionnaire: incorporating the elements of race and geographic region. Am J Epidemiol. 2009;170(1):104–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Bhagwat S, Haytowitz DB, Wasswa-Kintu S. USDA's expanded flavonoid database for the assessment of dietary intakes, release 1.1 - December 2015. Nutrient Data Laboratory, Beltsville Human Nutrition Research Center, ARS, USDA; 2015. Available from[Internet]: https://data.nal.usda.gov/dataset/usdas-expanded-flavonoid-database-assessment-dietary-intakes-release-11-december-2015. [DOI] [PubMed] [Google Scholar]
- 18. Bhagwat S, Haytowitz DB. USDA database for the proanthocyanidin content of selected foods, release 2 (2015). Nutrient Data Laboratory, Beltsville Human Nutrition Research Center, ARS, USDA; 2015. Available from[Internet]: https://data.nal.usda.gov/dataset/usda-database-proanthocyanidin-content-selected-foods-release-2-2015 [Google Scholar]
- 19. Neveu V, Perez-Jiménez J, Vos F, Crespy V, du Chaffaut L, Mennen L, Knox C, Eisner R, Cruz J, Wishart Det al. Phenol-Explorer: An online comprehensive database on polyphenol contents in foods. Database (Oxford). 2010;2010:bap024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Pérez-Jiménez J, Neveu V, Vos F, Scalbert A. Systematic analysis of the content of 502 polyphenols in 452 foods and beverages: An application of the Phenol-Explorer database. J Agric Food Chem. 2010;58(8):4959–69. [DOI] [PubMed] [Google Scholar]
- 21. Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake in epidemiologic studies. Am J Clin Nutr. 1997;65(4):1220S–8S. [DOI] [PubMed] [Google Scholar]
- 22. Guenther PM, Casavale KO, Reedy J, Kirkpatrick SI, Hiza HAB, Kuczynski KJ, Kahle LL, Krebs-Smith SM. Update of the Healthy Eating Index: HEI-2010. J Acad Nutr Diet. 2013;113(4):569–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. U.S. Department of Agriculture and U.S. Department of Health and Human Services . Dietary Guidelines for Americans, 2020-2025. 9th Edition. 2020; Available at: DietaryGuidelines.gov. [Google Scholar]
- 24. Horton NJ, Lipsitz SR, Parzen M. A potential for bias when rounding in multiple imputation. Am Stat. 2003;57(4):229–32. [Google Scholar]
- 25. Rosa L, NJA S. Anticancer properties of phenolic acids in colon cancer–A review. J Nutr Food Sci. 2016;6:2. [Google Scholar]
- 26. Srinivasulu C, Ramgopal M, Ramanjaneyulu G, Anuradha CM, Suresh Kumar C. Syringic acid (SA)‒A review of its occurrence, biosynthesis, pharmacological and industrial importance. Biomed Pharmacother. 2018;108:547–57. [DOI] [PubMed] [Google Scholar]
- 27. Sankaranarayanan R, Kumar DR, Patel J, Bhat GJ. Do aspirin and flavonoids prevent cancer through a common mechanism involving hydroxybenzoic acids?—The metabolite hypothesis. Molecules. 2020;25(9):2243. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Schmit SL, Nwogu O, Matejcic M, DeRenzis A, Lipworth L, Blot WJ, Raskin L. Coffee consumption and cancer risk in African Americans from the Southern Community Cohort Study. Sci Rep. 2020;10(1):17907. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Sartini M, Bragazzi NL, Spagnolo AM, Schinca E, Ottria G, Dupont C, Cristina ML. Coffee consumption and risk of colorectal cancer: A systematic review and meta-analysis of prospective studies. Nutrients. 2019;11(3):694. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Jiang R, Botma A, Rudolph A, Hüsing A, Chang-Claude J. Phyto-oestrogens and colorectal cancer risk: A systematic review and dose–response meta-analysis of observational studies. Br J Nutr. 2016;116(12):2115–28. [DOI] [PubMed] [Google Scholar]
- 31. Sirerol JA, Rodríguez ML, Mena S, Asensi MA, Estrela JM, Ortega AL. Role of natural stilbenes in the prevention of cancer. Oxid Med Cell Longev. 2015;2016:e3128951. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Wang X, Yang DY, Yang LQ, Zhao WZ, Cai LY, Shi HP. Anthocyanin consumption and risk of colorectal cancer: A meta-analysis of observational studies. J Am Coll. Nutr. 2019;38(5):470–7. [DOI] [PubMed] [Google Scholar]
- 33. Lambert JD, Hong J, Yang G, Liao J, Yang CS. Inhibition of carcinogenesis by polyphenols: Evidence from laboratory investigations. Am J Clin Nutr. 2005;81(1):284S–91S. [DOI] [PubMed] [Google Scholar]
- 34. Lei L, Yang Y, He H, Chen E, Du L, Dong J, Yang J. Flavan-3-ols consumption and cancer risk: A meta-analysis of epidemiologic studies. Oncotarget. 2016;7(45):73573–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Oskarsson A, Ohlsson Andersson Å. Suppressed sex hormone biosynthesis by alkylresorcinols: A possible link to chemoprevention. Nutr Cancer. 2016;68(6):978–87. [DOI] [PubMed] [Google Scholar]
- 36. Kochukov MY, Jeng YJ, Watson CS. Alkylphenol xenoestrogens with varying carbon chain lengths differentially and potently activate signaling and functional responses in GH3/B6/F10 somatomammotropes. Environ Health Perspect. 2009;117(5):723–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Caiazza F, Ryan EJ, Doherty G, Winter DC, Sheahan K. Estrogen receptors and their implications in colorectal carcinogenesis. Front Oncol. 2015;5:19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Manach C, Scalbert A, Morand C, Rémésy C, Jiménez L. Polyphenols: Food sources and bioavailability. Am J Clin Nutr. 2004;79(5):727–47. [DOI] [PubMed] [Google Scholar]
- 39. Naska A, Lagiou A, Lagiou P. Dietary assessment methods in epidemiological research: Current state of the art and future prospects. F1000Res. 2017;6:926. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Alkhaldy A. Inter-individual variability of polyphenol metabolism and colonic health. Ph.D. Thesis, University of Glasgow, Glasgow, UK, 2014. [Google Scholar]
- 41. Bułdak RJ, Hejmo T, Osowski M, Bułdak Ł, Kukla M, Polaniak R, Birkner E. The impact of coffee and its selected bioactive compounds on the development and progression of colorectal cancer in vivo and in vitro. Molecules. 2018;23(12):3309. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data described in the manuscript, code book, and analytic code will be made available upon request pending application to and approval by the Southern Community Cohort Study.