Fiber Intake and Survival After Colorectal Cancer Diagnosis

Key Points

Question  Is fiber intake after colorectal cancer diagnosis associated with mortality?

Findings  In this prospective cohort study that included 1575 patients with stage I to III colorectal cancer, higher intake of fiber, especially from cereals, was associated with a lower risk of colorectal cancer–specific and overall mortality. Patients who increased their fiber intake after diagnosis from levels before diagnosis showed better survival; higher intake of whole grains was also associated with favorable survival.

Meaning  Higher fiber intake after the diagnosis of nonmetastatic colorectal cancer may reduce the risk of colorectal cancer–specific and overall mortality.

Importance  Although high dietary fiber intake has been associated with a lower risk of colorectal cancer (CRC), it remains unknown whether fiber benefits CRC survivors.

Objective  To assess the association of postdiagnostic fiber intake with mortality.

Design, Setting, and Participants  A total of 1575 health care professionals with stage I to III CRC were evaluated in 2 prospective cohorts, Nurses’ Health Study and Health Professionals Follow-up Study. Colorectal cancer–specific and overall mortality were determined after adjusting for other potential predictors for cancer survival. The study was conducted from December 23, 2016, to August 23, 2017.

Exposures  Consumption of total fiber and different sources of fiber and whole grains assessed by a validated food frequency questionnaire between 6 months and 4 years after CRC diagnosis.

Main Outcomes and Measures  Hazard ratios (HRs) and 95% CIs of CRC-specific and overall mortality after adjusting for other potential predictors for cancer survival.

Results  Of the 1575 participants, 963 (61.1%) were women; mean (SD) age was 68.6 (8.9) years. During a median of 8 years of follow-up, 773 deaths were documented, including 174 from CRC. High intake of total fiber after diagnosis was associated with lower mortality. The multivariable HR per each 5-g increment in intake per day was 0.78 (95% CI, 0.65-0.93; P = .006) for CRC-specific mortality and 0.86 (95% CI, 0.79-0.93; P < .001) for all-cause mortality. Patients who increased their fiber intake after diagnosis from levels before diagnosis had a lower mortality, and each 5-g/d increase in intake was associated with 18% lower CRC-specific mortality (95% CI, 7%-28%; P = .002) and 14% lower all-cause mortality (95% CI, 8%-19%; P < .001). According to the source of fiber, cereal fiber was associated with lower CRC-specific mortality (HR per 5-g/d increment, 0.67; 95% CI, 0.50-0.90; P = .007) and all-cause mortality (HR, 0.78; 95% CI, 0.68-0.90; P < .001); vegetable fiber was associated with lower all-cause mortality (HR, 0.83; 95% CI, 0.72-0.96; P = .009) but not CRC-specific mortality (HR, 0.82; 95% CI, 0.60-1.13; P = .22); no association was found for fruit fiber. Whole grain intake was associated with lower CRC-specific mortality (HR per 20-g/d increment, 0.72; 95% CI, 0.59-0.88; P = .002), and this beneficial association was attenuated after adjusting for fiber intake (HR, 0.77; 95% CI, 0.62-0.96; P = .02).

Conclusions and Relevance  Higher fiber intake after the diagnosis of nonmetastatic CRC is associated with lower CRC-specific and overall mortality. Increasing fiber consumption after diagnosis may confer additional benefits to patients with CRC.

Colorectal cancer (CRC) is the third most common cancer and third leading cause of cancer death in the United States.¹ Given advances in early detection and treatment, the number of CRC survivors is estimated to exceed 1.4 million in 2016 and expected to grow over the coming decades.² Many cancer survivors are motivated to seek self-care strategies, particularly dietary counseling, to facilitate their treatment and recovery.³ However, due to lack of data on postdiagnostic diet and CRC survival, most dietary recommendations for CRC survivors are primarily based on incidence studies.^4,5 Therefore, identifying prognostic dietary factors is needed to improve CRC survivorship.

Fiber has been associated with lower risk of CRC in many—but not all—studies.⁶ The most recent expert report concludes that consumption of foods containing dietary fiber probably protects against CRC.⁷ Fiber helps to minimize exposure to intestinal carcinogens by diluting fecal content and decreasing transit time⁸ and also has systemic benefits on insulin sensitivity and metabolic regulation,⁹ which have been linked to CRC prognosis.¹⁰ Moreover, fiber can be fermented by the gut bacteria into short-chain fatty acids, such as butyrate, acetate, and propionate, that possess a diversity of tumor-suppressive effects.^11,12 Preclinical studies have indicated the potential of butyrate and its analogs as chemotherapeutic agents in several tumor models,^13,14 including CRC.¹⁵

Despite these data, to our knowledge, no study has yet examined the association between fiber intake and survival of patients with CRC. Therefore, we used data from 2 large prospective cohorts—the Nurses’ Health Study (NHS) and the Health Professionals Follow-up Study (HPFS)—to test the hypothesis that high consumption of fiber and its major food sources after CRC diagnosis might be associated with lower mortality.

The NHS enrolled 121 700 US registered female nurses who were aged 30 to 55 years in 1976. The HPFS enrolled 51 529 US male health care professionals who were aged 40 to 75 years in 1986. Details about the 2 cohorts have been described elsewhere.^16,17 Briefly, participants were mailed a questionnaire inquiring about their medical history and lifestyle factors at baseline and every 2 years thereafter. Dietary data were collected and updated every 4 years using food frequency questionnaires (FFQs). In the present analysis, we used 1980 for the NHS and 1986 for the HPFS as baseline when we first collected detailed data on fiber intake. The follow-up rates have been 95.4% in the NHS and 95.9% in the HPFS for each of the questionnaires through 2010. The present study was conducted from December 23, 2016, to August 23, 2017. This study was approved by the institutional review board at the Brigham and Women’s Hospital. We obtained written informed consent from all participants.

On each biennial follow-up questionnaire, participants were asked whether they had received a diagnosis of CRC during the previous 2 years. For participants who reported CRC diagnosis, we asked for their written informed consent to acquire medical records and reports of pathologic tests. Study physicians, blinded to exposure data, reviewed all medical records to confirm CRC diagnosis and record the disease stage, histologic findings, and tumor location.¹⁸ In the present analysis, we included a total of 1575 participants who were diagnosed with stage I to III CRC throughout follow-up and completed the FFQ after diagnosis (963 in the NHS and 612 in the HPFS) (eFigure in the Supplement).

Deaths were identified through review of the National Death Index, family members, or the postal system in response to the follow-up questionnaires. The cause of death was assigned by study physicians based on all available data, including medical records. More than 96% of the deaths have been identified using these methods.¹⁹

Dietary intake data were collected repeatedly by FFQs in which participants were asked how often, on average, they consumed each food of a standard portion size during the previous year. We calculated the daily intake for each nutrient by multiplying the reported frequency of consumption of each item by its nutrient content and then summing across all foods. Fiber intake was calculated using the Association of Official Analytical Chemists method (accepted by the US Food and Drug Administration and the Food and Agriculture Organization of the World Health Organization for nutrition labeling purposes).²⁰ We adjusted fiber intake for total caloric intake using the nutrient residual method.²¹ Food frequency questionnaires have shown good reproducibility and validity for assessing fiber intake (eMethods in the Supplement).²²

Fiber intake from major food sources, including cereals, vegetables, and fruits, was considered separately. We also assessed whole grain consumption from all grain-containing foods (rice, bread, pasta, and breakfast cereals) according to the dry weight of whole grain ingredients in each food, as previously described.²³

For postdiagnostic intake, the first FFQ collected at least 6 months but no more than 4 years after diagnosis (median, 2.2 years) was used to avoid assessment during the period of active treatment.²⁴ In a sensitivity analysis, we also examined the cumulative mean intake of fiber throughout the entire postdiagnostic period (eMethods in the Supplement). Prediagnostic intake was based on the last FFQ reported before CRC diagnosis.

We calculated person-time of follow-up from the return date of the FFQ that was used for postdiagnostic assessment to death or the end of the study period (June 1, 2012, for the NHS and January 31, 2012, for the HPFS), whichever came first. In the main analysis, death from CRC was the primary end point, and deaths from other causes were censored. In secondary analyses, death from any cause was the end point.

We used cause-specific Cox proportional hazards regression models with time since diagnosis as the time scale, accounting for left truncation due to differences between patients in the timing of postdiagnostic assessment.²⁵ We calculated hazard ratios (HRs) and 95% CIs of death, adjusted for prediagnostic fiber intake and other potential predictors for cancer survival (eMethods in the Supplement). The Wald test for trend was performed using the median for each quartile of fiber intake as a continuous variable. We tested proportional hazards assumption by including the interaction term between fiber intake and time into the model and did not find statistical evidence for violation of this assumption.

To reduce residual confounding, we further adjusted for a propensity score that reflected associations of fiber consumption with potential confounding covariates.²⁶ To minimize any bias resulting from the availability of postdiagnostic questionnaire data, we applied the inverse probability weighting method to all survival analyses.²⁷ We examined the dose-response relationship between fiber intake and mortality using the restricted cubic spline analysis.²⁸ More details about these analyses are provided in the eMethods in the Supplement.

We also calculated the change in fiber intake by subtracting the prediagnostic intake from the postdiagnostic intake and examined the association with mortality. Moreover, we assessed consumption of whole grains and different sources of fiber. Finally, we conducted the stratified analysis by clinicopathologic and lifestyle factors. P value for interaction was calculated using the likelihood ratio test. All statistical tests were 2-sided, and P < .05 was considered statistically significant. We used SAS, version 9.4 for all analyses (SAS Institute).

Among 1575 eligible patients diagnosed with stage I to III CRC, 963 (61.1%) were women; mean (SD) age was 68.6 (8.9) years. We documented 773 deaths, of which 174 (22.5%) were classified as CRC-specific deaths over a median of 8 years of follow-up. Other major causes of death included cardiovascular diseases (n = 168) and cancers other than CRC (n = 121). The overall 5-year survival rates were 83% (95% CI, 79%-86%) for stage I cancer, 82% (95% CI, 78%-86%) for stage II cancer, and 72% (95% CI, 67%-76%) for stage III cancer. These rates appeared to be comparable to national estimates.²⁹

Participants with higher fiber intake tended to have a healthier lifestyle (eTable 1 in the Supplement). Anatomic subsite and grade of differentiation did not differ across quartiles of fiber intake, whereas patients with the highest fiber intake were more likely to have stage I cancer than stage II and III cancers. To stringently control for any confounding effect by stage, we performed stage-stratified Cox regression for all the association analyses.

As reported in Table 1, fiber intake was inversely associated with CRC-specific mortality after adjusting for other potential determinants for cancer prognosis. The multivariable HR per each 5-g increase in intake per day was 0.78 (95% CI, 0.65-0.93; P = .006) for CRC-specific mortality and 0.86 (95% CI, 0.79-0.93; P < .001) for all-cause mortality. Similar findings were obtained when cumulative mean fiber consumption after diagnosis was used for the analysis (eTable 2 in the Supplement).

The spline analysis showed a linear relationship between fiber intake after diagnosis and CRC-specific mortality. For all-cause mortality, a statistically significant nonlinear relationship was observed. The benefit associated with increasing fiber intake achieved its maximum at approximately 24 g/d, and no further reduction in mortality was found beyond this level of intake (Figure).

To minimize bias associated with occult recurrences or other undiagnosed major illnesses that could influence dietary intake, we excluded the first year of follow-up in a sensitivity analysis. The results remained essentially unchanged, with the multivariable HR per 5-g/d increment for CRC-specific mortality of 0.80 (95% CI, 0.66-0.97; P = .02 for trend) and the HR for all-cause mortality of 0.85 (95% CI, 0.78-0.93; P = .002 for trend). Similar results were also obtained in the propensity score analysis, with the corresponding HRs of 0.80 (95% CI, 0.67-0.96; P = .02 for trend) and 0.86 (95% CI, 0.79-0.93; P < .001 for trend), respectively, indicating the robustness of our findings to confounding.

No statistically significant interaction was detected between fiber intake and tumor subsite or stage (P > .05 for interaction) (eTable 3 in the Supplement). In a subset of patients with chemotherapy data (n = 375), fiber intake was similar among those who received chemotherapy (median, 19.5 g/d) and those who did not (median, 19.1 g/d) (P = .13 for Wilcoxon test).

Prediagnostic and postdiagnostic intake of fiber were modestly correlated (Spearman correlation coefficient r = 0.58). We assessed whether changing fiber intake after diagnosis was associated with mortality. As reported in Table 2, patients who increased their fiber intake after diagnosis from levels before diagnosis had a lower mortality rate, and each 5-g/d increase in fiber intake was associated with 18% lower CRC-specific mortality (95% CI, 7%-28%; P = .002) and 14% lower all-cause mortality (95% CI, 8%-19%; P < .001).

Next, we examined associations by major dietary sources of fiber, including those from cereals, vegetables, and fruits. Fiber intakes from these sources were weakly correlated (Spearman correlation coefficient r<0.25) (eTable 4 in the Supplement). As noted in Table 3, fiber from cereals showed an inverse association with lower mortality after mutual adjustment for fruit and vegetable fiber. The HR associated with 5-g/d increments in cereal fiber was 0.67 (95% CI, 0.50-0.90; P = .007 for trend) for CRC-specific mortality and 0.78 (95% CI, 0.68-0.90; P < .001 for trend) for all-cause mortality. In contrast, no association was found for fruit fiber. Vegetable fiber was associated with lower all-cause mortality (HR per 5-g/d increment, 0.83; 95% CI, 0.72-0.96; P = .009 for trend) but not CRC-specific mortality (HR, 0.82; 95% CI, 0.60-1.13; P = .22).

Whole grain consumption was associated with lower CRC-specific mortality, with the multivariable HR per 20-g/d increment of 0.72 (95% CI, 0.59-0.88; P = .002 for trend) (Table 4). This association was attenuated after adjusting for total fiber intake (HR, 0.77; 95% CI, 0.62-0.96; P = .02 for trend). Similar, but weaker, attenuation was observed for all-cause mortality, with the HR changing from 0.88 (95% CI, 0.80-0.97; P = .008 for trend) to 0.91 (95% CI, 0.83-1.01; P = .08 for trend) after including fiber in the multivariable model.

In an exploratory analysis, we examined whether the association between total fiber intake and mortality differed by other predictors of cancer prognosis, including sex, age, smoking status, alcohol consumption, body mass index, physical activity, aspirin use, and dietary glycemic load (eTable 5 in the Supplement). We observed a statistically significant interaction with alcohol consumption for CRC-specific mortality (for patients with alcohol intake <7 g/day: HR, 0.73; 95% CI, 0.57-0.94; for patients with alcohol intake ≥7 g/day: HR, 1.07; 95% CI, 0.75-1.52; P = .05) and with regular aspirin use for all-cause mortality (for patients who did not regularly use aspirin: HR, 0.81; 95% CI, 0.73-0.91; for regular aspirin users: HR, 0.99; 95% CI, 0.86-1.13; P = .01). However, given the multiple testing and limited event numbers, these findings should be interpreted cautiously. No other statistically significant interaction was detected.

To our knowledge, this is the first prospective study examining the prognostic influence of fiber intake among patients with CRC. We found that patients with a higher intake of fiber, especially that from cereals, had a lower rate of CRC-specific and all-cause mortality. Patients who increased their intake from their levels before diagnosis experienced a modest reduction in mortality. Higher consumption of whole grains was also associated with better survival, and this beneficial association was partly mediated by fiber. Our findings provide novel evidence for the potential benefit of increasing fiber and whole grain consumption among patients with CRC.

Substantial evidence supports the protective effect of high fiber intake for CRC prevention. According to a meta-analysis of 25 prospective studies, each 10-g increment in daily intake of total and cereal fiber was associated with an approximately 10% lowered risk of developing CRC.³⁰ This finding agrees with the well-established data from animal studies that high-fiber diets promote apoptosis and suppress colorectal tumor development.^31-33 Our present study adds to the existing literature and suggests that the effect of high fiber intake may extend beyond protection against cancer incidence and contribute to better prognosis after cancer is established.

Higher intake of fiber, especially cereal fiber, has been linked to improved insulin sensitivity,³⁴ lipid profile,³⁵ endothelial function, and reduced inflammation.³⁶ Emerging, albeit limited, evidence suggests that hyperinsulinemia and markers of insulin resistance and inflammation predict worse survival in patients with CRC^37,38 and may mediate the adverse metabolic effect of obesity and physical inactivity on CRC recurrence and death.^24,39 Therefore, higher fiber intake after CRC diagnosis may improve patients’ survival by mitigating the tumor-promoting effect of hyperinsulinemia and inflammation.

Bacterial fermentation of fiber also produces butyrate, which has been increasingly implicated in modulation of the tumor microenvironment.¹² Although butyrate is a major energy source for normal colonocytes, it is metabolized to a lesser extent in cancer cells due to the Warburg effect and accumulates in the nucleus of cancerous colonocytes in which it functions as an inhibitor of histone deacetylase to epigenetically regulate expression of numerous genes responsible for tumor growth, angiogenesis, migration, and chemoresistance.¹² Moreover, butyrate may influence CRC prognosis by modulating the function of tumor-infiltrating immune cells, including regulatory T cells^40,41 and macrophages,⁴² which have been increasingly recognized for their critical roles in tumor-host interactions and cancer prognosis.^43,44

Consistent with these data, we found that patients with CRC who consumed higher levels of fiber after diagnosis had substantially lower rates of death. The beneficial association appeared to differ by fiber sources, with cereal fiber showing the strongest association. These findings are further supported by the favorable survival rates in association with high consumption of whole grain, a rich source of cereal fiber. Concordant with our findings, previous studies support the importance of fiber sources. Compared with fiber from other sources, cereal fiber and whole grains have been most consistently associated with a lower incidence of colorectal neoplasia,^30,45-48 type 2 diabetes,^49,50 cardiovascular disease,^50,51 and total mortality.^50,52 Although the exact reasons remain unclear, it is possible that the generally high fiber content of cereals, especially whole grains, may contribute to the more pronounced benefit, whereas the amount of fiber consumed from fruits may be too low for an association with health outcomes to be observed. Alternatively, other components in cereals and whole grains may contribute to their favorable effects, such as vitamins, minerals, phenols, and phytoestrogens.^53-55 However, in the present study, adjusting for fiber intake attenuated the association between whole grain and lower mortality, suggesting that fiber may be an important contributor for the protective effect of whole grains.⁵⁶

Advantages of the present study include the prospective design, detailed collection of prediagnostic and postdiagnostic diet and lifestyle information, standardized medical record review of self-reported CRC and deaths, and long-term follow-up. Moreover, the detailed covariate data collected in parallel with fiber intake allowed for rigorous control for confounding by various predictors of cancer survival.

There are limitations that are worth noting. First, detailed treatment data were largely unavailable in the cohorts. However, among a subset of patients with chemotherapy data, fiber intake did not differ by the use of chemotherapy. Moreover, during the time of the study, adjuvant treatment was largely standardized and primarily correlated with disease stage. Thus, our ability to control for stage in our analyses minimized any potential confounding by treatment. In addition, because all participants were health care professionals, any difference in access to adjuvant chemotherapy is minimized. Second, beyond cause of mortality, data on cancer recurrences were unavailable in our cohorts. Nevertheless, because the median survival for metastatic CRC was approximately 10 to 12 months during much of this study,⁵⁷ CRC-specific mortality should be a reasonable surrogate for cancer-specific outcomes. Third, the number of CRC-specific deaths is relatively small; therefore, further large-scale studies are needed.

Finally, as an observational study, residual confounding cannot be excluded, although we observed similar results through multivariable adjustment and propensity score analysis. Our findings need to be validated by further studies, including possible clinical trials. Previous observational data regarding the favorable influence of other lifestyle factors on CRC survival have been subsequently confirmed in randomized trials. For example, in support of the beneficial association with CRC survival for physical activity⁵⁸ and high intake of vitamin D⁵⁹ and marine ω-3 fatty acid⁶⁰ in prospective cohort studies, randomized clinical trials have documented a positive influence of exercise intervention on patients’ quality of life and functional capacity,^61-64 the benefit of preoperative ω-3 fatty acid therapy on reducing tumor vascularity and prolonging patients’ survival,⁶⁵ and the effect of high-dose vitamin D supplementation on improved survival in patients with metastatic CRC.⁶⁶ These robust data indicate the critical role of prospective observational studies in identification of modifiable lifestyle factors for improvement of cancer survival.

Higher intake of fiber and whole grains after CRC diagnosis is associated with a lower rate of death from that disease and other causes. Our findings provide support for the nutritional recommendations of maintaining sufficient fiber intake among CRC survivors.

Corresponding Author: Andrew T. Chan, MD, MPH, Clinical and Translational Epidemiology Unit, Massachusetts General Hospital and Harvard Medical School, 55 Fruit St, GRJ-825C, Boston, MA 02114 (achan@partners.org).

Published Online: November 2, 2017. doi:10.1001/jamaoncol.2017.3684

Correction: This article was corrected on March 7, 2019, to correct omissions in the Conflict of Interest Disclosures.

Accepted for Publication: August 25, 2017.

Author Contributions: Drs Song and Chan had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Song, Fuchs, Giovannucci, Chan.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Song, Chan.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Song, Wang, Chan.

Obtained funding: Ogino, Fuchs, Giovannucci, Chan.

Administrative, technical, or material support: Fuchs, Chan.

Study supervision: Fuchs, Giovannucci, Chan.

Conflict of Interest Disclosures: Dr Chan previously served as a consultant for Bayer Healthcare, Aralez Pharmaceuticals, and Pfizer Inc for work unrelated to the topic of this article. Dr Fuchs has been a consultant and/or a scientific advisor for Eli Lilly, Entrinsic Health, Pfizer, Merck, Sanofi, Roche, Genentech, Merrimack Pharmaceuticals, Dicerna, Bayer, Celgene, Agios, Gilead Sciences, Five Prime Therapeutics, Taiho, and KEW. No other disclosures are reported.

Funding/Support: This work was supported by US National Institutes of Health grants P01 CA87969 (Meir Stampfer, MD, DrPH); UM1 CA186107 (Dr Stampfer); P01 CA55075 (Walter Willett, MD, DrPH); UM1 CA167552 (Dr Willett); P50 CA127003 (Dr Fuchs); K24 DK098311, R01 CA137178, R01 CA202704, and R01 CA176726 (Dr Chan); and R01 CA151993 and R35 CA197735 (Dr Ogino); the 2017 American Association for Cancer Research-AstraZeneca Fellowship in Immuno-oncology Research grant 17-40-12-SONG (Dr Song); and by grants from the American Institute for Cancer Research (Dr Wu), the Project P Fund for Colorectal Cancer Research, The Friends of the Dana-Farber Cancer Institute, Bennett Family Fund, and the Entertainment Industry Foundation through the National Colorectal Cancer Research Alliance.

Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We thank the participants and staff of the Nurses’ Health Study and the Health Professionals Follow-up Study for their valuable contributions as well as the following state cancer registries for their help: Alabama, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida, Georgia, Idaho, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Nebraska, New Hampshire, New Jersey, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, Tennessee, Texas, Virginia, Washington, and Wyoming.