Dietary Intake of Branched-Chain Amino Acids and Risk of Colorectal Cancer

Colorectal cancer is the third most common cancer and the third leading cause of death from cancer in the United States in both men and women (1).

Branched chain amino acids (BCAA), namely, leucine, isoleucine, and valine, are essential amino acids, and emerging evidence suggests that BCAAs may mediate pathways related to cancer development and progression, possibly due to their involvement in insulin metabolism (2–4). Some prospective studies have reported positive associations between plasma BCAA levels and risk of T2D (5, 6), CVD (7–9), and pancreatic cancer (10), which may share etiologic pathways (e.g., insulin resistance) with colorectal cancer (11).

Only few cross-sectional studies have examined the association between plasma BCAA levels and colorectal neoplasia, but did not provide support for a positive association between BCAAs and risk of colorectal neoplasia (12, 13). In one cross-sectional study from Japan, total plasma BCAA levels were inversely associated with risk of colorectal adenoma (a precursor for colorectal cancer; ref. 14) in men, but not in women (12).

In several studies, albeit not all (15), dietary intake was weakly correlated with plasma BCAA levels (16, 17), thus the contribution of dietary intake on plasma levels remains controversial. To our knowledge, no prospective study has examined the association between dietary BCAA intake and risk of colorectal cancer.

To test whether higher dietary intake of BCAA is associated with higher risk of colorectal cancer, we conducted a prospective analysis using data from three large prospective cohorts of US men and women, the Nurses' Health Study (NHS), the Nurses' Health Study II (NHS II), and the Health Professionals Follow-up Study (HPFS).

NHS began in 1976 in the United States when 121,700 registered female nurses ages 30 to 55 were recruited (18). The NHS II was established in 1989 and included 116,430 female registered nurses between the ages of 25 and 42 years. HPFS enrolled 51,529 male health professionals ages 40 to 75 years in 1986 (19). Questionnaires were administered at baseline, and follow-up questionnaires were administered biennially to collect updated information on lifestyle and medical history. The cumulative follow-up rates in these cohorts are over 90%.

The study protocol was approved by the institutional review boards of the Brigham and Women's Hospital and Harvard T.H. Chan School of Public Health, and those of participating registries as required.

Dietary intake was assessed repeatedly with validated semiquantitative food frequency questionnaires (FFQ) every 4 years, starting in 1980 for NHS, 1989 for NHS II, and 1986 for HPFS. Participants answered how often they, on average, consumed a specified amount of food during the previous year. The contribution of each food to nutrient intake was calculated by multiplying the frequency of consumption of each food by its nutrient content and summing up the nutrient intakes from all foods. These nutrient intakes were energy adjusted using the residual method (20). Total BCAA intake (from now on referred to as “BCAA intake”) was calculated by summing intakes of valine, leucine, and isoleucine. The reproducibility and validity of FFQs to estimate protein intake were examined previously (21–23). In brief, de-attenuated Spearman correlation coefficients between FFQ and dietary records were 0.56 for animal and 0.66 for vegetable protein (24). Although correlations are not available for BCAA intake per se, total BCAA intake was highly correlated with animal protein intake (all partial Spearman correlation coefficients adjusted for age were >0.90), but not with vegetable protein intake (correlations ranging between −0.1 and 0.1 in all three cohorts). The individual branch chain amino acids valine, leucine, and isoleucine were also highly correlated with each other (all partial Spearman correlation coefficients adjusted for age were 0.99).

For this analysis, follow-up started in 1984 for NHS and 1991 for NHS II, the first time a more detailed 131-item FFQ was administered in these cohorts. HPFS participants were mailed the detailed FFQ at baseline, thus follow-up for HPFS started in 1986. We excluded participants who left a high number of items blank (≥70) on the FFQs, those who reported unusually low or high energy intake, i.e., <800 or >4,200 kcal per day for men and <600 or >3,500 kcal per day for women, and those who had history of cancer (except nonmelanoma skin cancer) or ulcerative colitis at baseline. After these exclusions, 77,017 women from NHS, 92,984 women from NHS II, and 47,255 men were eligible for our analysis.

We collected information on age, body weight and height, smoking, physical activity, endoscopic screening, family history of colorectal cancer, use of multivitamin supplements, aspirin and nonsteroidal anti-inflammatory drugs (NSAID), and menopausal status and postmenopausal hormone use (for women). Body mass index (BMI) was calculated as body weight (kg) divided by height squared (m²).

When participants reported a new diagnosis of colorectal cancer on their biennial questionnaires, we asked the participant or the next-of-kin for permission to obtain medical records and pathologic reports. Study physicians blinded to exposure status reviewed medical records to confirm a diagnosis of colorectal cancer and extract information on stage, location, and histology. Proximal colon cancers were defined as cancers located in the cecum, ascending, and transverse colon, distal cancers as those in the descending and sigmoid colon, and rectal cancers as those located in the recto-sigmoid and rectum. Information on deaths was gathered from the National Death Index, state vital statistics records, the postal system, and the report from next-of-kin (25, 26), and cause of deaths was confirmed by study physicians who reviewed death certificates and medical records. For nonresponders who died of colorectal cancer, we also contacted the next-of-kin to obtain permission to review medical records and confirm a diagnosis of colorectal cancer as described above.

Person-time was calculated from the date of return of the baseline questionnaire to the date of diagnosis of colorectal cancer, the date of death or end of follow-up, whichever came first. End of follow-up was May 31, 2012, for NHS, May 31, 2013, for NHS II, and January 31, 2012, for HPFS. We calculated cumulative updated average of dietary intakes and lifestyle factors such as BMI or physical activity to better represent long-term patterns and reduce within-person variation. Cumulative updated exposure was estimated by calculating the average exposure from all available follow-up questionnaires up to the most recent follow-up questionnaire cycle (27). In addition to cumulative updated intake, we also examined associations for BCAA intake at baseline and after a 0- to 4-year lag (simple updated intake), which represents the intake obtained from the most recent FFQ prior to each follow-up cycle (27). Cox proportional hazard models were used to calculate hazard ratios (HR) and 95% confidence intervals (CI) for the association between BCAA intake (quintile) and colorectal cancer risk. We stratified jointly by age (in months) and 2-year questionnaire cycle to control finely for confounding by calendar time and age. In multivariable models, we additionally adjusted for other known and suspected risk factor for colorectal cancer (covariables were selected based on published data including previous findings from our cohorts; refs. 28–30) including BMI (< 23, 23 to <25, 25 to <27, 27 to <30, 30 to <35, >35 kg/m²), smoking status (never, past, current smoking with 1–14 cigarettes per day and with >15 cigarettes per day), alcohol intake (0 to <5, 5 to <10, 10 to <15, or ≥15 g/day), physical activity (quintile), history of colorectal cancer in a first-degree relative (yes or no), screening endoscopy (yes or no), multivitamin use (yes or no), regular aspirin or NSAIDs use (≥2 tablets/week vs. <2 tablets/week). For women, we also adjusted for menopausal status and use of postmenopausal hormone (premenopausal, never, past, current). Analysis was first conducted for each cohort separately and then pooled using the random-effects model (31). P for trend was calculated by including the median value of each quintile category as a continuous exposure in the models and utilizing the Wald test to assess statistical significance.

To test for potential modification by other colorectal cancer risk factors, including BMI, smoking, physical activity, alcohol intake, animal/vegetable protein ratio and high versus low fat intake, we included cross-product terms for exposure variables in the multivariable adjusted model. All analyses were performed using the SAS software (version 9.4, SAS Institute Inc.). All statistical tests were two sided, and a P value <0.05 was considered statistically significant.

During up to 28 years of follow-up, 1,660 colorectal cancer cases were documented in NHS, 306 cases in NHS II, and 1,343 cases in HPFS. The age-standardized distribution of participants' characteristics weighted by person-years according to BCAA intake in quintiles is shown in Table 1. In men, participants with higher BCAA intake were more likely to have higher BMI, but did not appear to differ appreciably with regard to physical activity. In women, those with higher intake were more likely to be physically active and had higher BMI. In all three cohorts, participants with higher BCAA intake were more likely to have higher intake of dairy products, protein, dietary fiber, folate, vitamin D, and calcium and less likely to be current smokers.

After pooling results from all three cohorts, we observed a borderline statistically significant inverse association between BCAA intake and colorectal cancer risk [model 1: Q5 vs. Q1, HR (95% CI): 0.89 (0.80–1.00), P_trend = 0.06; HR per SD increment (SD = 2.59 g/d in HPFS, 1.98 g/d in NHS and 2.47 g/d in NHS II): 0.95 (0.92–0.99); Table 2)].

To investigate whether observed associations may be explained by higher intake of major food sources of BCAA or by diabetes status, we also examined associations for colorectal cancer after including major food sources of BCAAs, i.e., dairy, red and processed meat, chicken and poultry, and fish, and other dietary factors such as intake of fiber, vitamin D and folate, separately (one-by-one) to the multivariable models. For the most part, risk estimates were similar when we adjusted for BCAA food sources and diabetes (Supplementary Table S1). However, after including total or dairy calcium to the models, BCAA intake was no longer significantly associated with risk of colorectal cancer (model 2, adjusted for dairy calcium intake; Table 2).

When we examined associations separately by colorectal cancer location (proximal, distal, and rectal cancers; Supplementary Table S2), BCAA intake was not associated with risk of proximal cancers. In multivariable models, we observed evidence for an inverse association between BCAA intake and distal colon (HPFS) and rectal cancers (NHS). Inverse associations between BCAA intake and left-sided cancers were slightly attenuated, but persisted after adding red and processed meat, chicken or turkey, or fish intake one-by-one to the models (Supplementary Table S3). After including dairy calcium intake to the multivariable models, inverse associations for distal colon cancer [HR per SD: 0.87 (0.78–0.97)] in HPFS and for rectal cancers in NHS [0.89 (0.79–1.00)] remained, but results for NHS did not reach statistical significance (Supplementary Table S2).

When we studied associations for BCAA intake at baseline or after a 0- to 4-year lag in the NHS and HPFS (for the 0–4 year lag analysis NHS II was excluded due to limited sample size), results were similar [baseline: Q5 vs. Q1 pooled HR 0.95 (0.81–1.12), P_trend = 0.51, simple updated: pooled HR 0.97 (0.85–1.11), P_trend = 0.68]. Furthermore, we did not observe evidence for a statistically significant interaction with BMI, smoking status, alcohol intake or physical activity, high versus low fat intake or animal protein/vegetable protein ratio (Supplementary Table S4).

In this large study using data from three prospective cohorts of US health professionals, we did not find evidence for a positive association between BCAA intake and risk of colorectal cancer. Although in multivariable adjusted models, we observed a weak inverse association between BCAA intake and colorectal cancer, after including dairy calcium intake to the models, BCAA intake was no longer associated with risk of colorectal cancer. Furthermore, in multivariable models that included dairy calcium, we observed an inverse association for distal colon cancers in HPFS and for rectal cancers in NHS.

BCAAs are important building blocks for proteins and function as nutrient signals mediating metabolic pathways, including those related to insulin and lipid metabolism (2, 3). Obesity and insulin resistance are established risk factors for colorectal cancer (11, 28) and findings from both animal and human intervention studies suggest that higher intake or supplementation of leucine may have a favorable effect on glucose homeostasis, insulin metabolism, and satiety (32–34). In obese mice with hyperinsulinemia, BCAA supplementation was associated with fewer premalignant lesions of the colon (aberrant crypt foci) when compared with normal diet-fed mice. BCAA-supplemented mice also exhibited lower expression of insulin like growth factor-I (IGF-I) receptor on the colonic mucosa and decreased serum insulin and IGF-I levels (35). Additionally, in both colon and hepatic cancer cell lines, higher exposure to BCAA was associated with lower insulin-induced proliferation (36).

Contrary to the mechanistic intake studies, several plasma-based prospective studies have recently reported positive associations between plasma BCAA levels and risk of T2D (5, 6) and CVD (7, 8), which may share etiologic pathways (e.g., insulin resistance) with colorectal cancer (11). Similarly, in a prospective study that included participants from several US cohorts including HPFS and NHS, plasma BCAA levels were positively associated with risk of pancreatic cancer (10).

Epidemiologic studies regarding the association between BCAAs and colorectal neoplasia have been limited to few retrospective plasma studies. Consistent with our findings on diet, these studies did not support that higher plasma levels of BCAA increase risk of colorectal neoplasia. For example, in a cross-sectional study based in Japan including 629 cases and 584 controls, total plasma BCAA levels were inversely associated with risk of colorectal adenoma [Q4 vs. Q1, OR (95% CI): 0.68 (0.48–0.98), P trend 0.10)]. However, after stratification by sex, inverse associations were observed in men, but not in women [total BCAA, Q4 vs. Q1, OR (95% CI): men 0.58 (0.37–0.93); women: 0.84 (0.45–1.57); ref. 12]. However, in that study, P for interaction by sex was not statistically significant (P interaction 0.55). Inverse associations also appeared to be most pronounced for leucine and restricted to colon (proximal and distal) adenomas, but the number of rectal adenoma cases was very small (19 cases in the highest quartile of BCAA concentration) and blood samples were collected after diagnosis, thus results may have been biased (12). In another retrospective study that used metabolomics data from 64 colorectal cancer cases and 65 controls, results differed depending on the laboratory method used to identify metabolic profiles. For example, valine and leucine serum levels were lower in cases when compared with controls using the GC-TOMS analysis method, but did not appear to differ between cases and control when using the UPLC-QTOFMS method (13). However, due to the retrospective design and the potential for reverse causation, these results need to be interpreted with caution. In addition, these two studies did not adjust for calcium or dairy calcium intake.

The reason for the differential associations observed in our study compared with previous prospective studies on plasma BCAAs and metabolic diseases, known to share pathways with colorectal cancer, is unclear. However, in several studies, dietary intake was weakly correlated with plasma BCAA levels (16, 17), which may, at least in part, explain these differential associations. Moreover, it is unclear whether higher plasma BCAA levels are causally related to risk of chronic diseases or are a marker for alterations in BCAA metabolism related to insulin resistance (2). We did not examine plasma BCAA levels, thus we were not able to assess the proportion of variability in plasma BCAA levels explained by dietary intake in our cohorts.

Findings from our study did not suggest that higher BCAA intake increases risk of colorectal cancer. Specifically, after adjusting for total or dairy calcium, BCAA intake was no longer associated with risk of colorectal cancer, suggesting that calcium in dairy products may have likely contributed to the initially observed inverse association between BCAA and colorectal cancer risk. In meta-analyses and in our cohorts, higher dairy or calcium intake has been associated with lower risk of colorectal neoplasia (28–30, 37–40).

Red meat and poultry (−30%), dairy products (−15%) and fish (−8%) are main contributors of total BCAA intake in our cohorts. Red meat is a known risk factor for colorectal cancer (41), whereas dairy products and fish are generally associated with lower risk of colorectal cancer (42). Dietary BCAA intake was highly correlated with animal protein intake; therefore, it was difficult to distinguish the effect of BCAAs per se from those of animal protein or its main food sources. It is unclear why BCAA intake was inversely associated with distal colon cancer in HPFS and rectal cancers in NHS. In our study, observed associations between BCAA intake and total or left-sided cancers were slightly attenuated, but persisted after adding red and processed meat, chicken or turkey or fish intake separately (one-by-one) to the models. Thus, considering the collinearity between BCAAs and its major food sources, we cannot exclude that observed associations may be due to higher intake of other components in BCAA-rich foods.

Our study has other limitations that warrant further discussion. First, misclassification in estimation of BCAA intake cannot be excluded. However, misclassification of exposure generally leads to a bias toward the null association and we calculated cumulative average of dietary BCAA intake to reduce random error and better estimate long-term exposure. Second, we cannot exclude the possibility that residual confounding may have affected our results; however, we have collected detailed data on numerous diet and lifestyle factors in our cohorts, which made it possible to assess and adjust for a wide range of potential confounders in our analysis. Third, the participants in the NHS and HPFS cohorts represent a more health conscious population than the general US population, which may limit generalizability to other populations. However, main sources for BCAA are dietary factors that can both increase (e.g., red meat) or decrease (e.g., fish) risk for colorectal cancer and as such can be part of “healthy dietary pattern” as well as an “unhealthy dietary pattern,” and the relative homogeneity of our cohort participants may help to reduce residual confounding. Fourth, we cannot entirely exclude reverse causation, but when we examined associations for BCAA intake at baseline or after a 0- to 4-year lag, results were similar.

One major strength of our study is the long follow-up and its prospective design. Repeated measurements of dietary exposure reduce random error and enable us to better estimate long-term exposure. Our large sample size also allowed us to perform stratified analyses by lifestyle characteristics and tumor location with sufficient power.

In conclusion, findings from this large prospective study do not support that higher BCAA intake, at least in the context of a US diet, is associated with higher risk of colorectal cancer. As this the first prospective study to examine this association, confirmation in other cohorts is warranted.

C.S. Fuchs is a consultant for Agios, Bain Capital, Merck, Merrimack Pharma, Pfizer, Sanofi, Taiho, Unum Therapeutics, Entrinsic, Bayer, Celgene, Dicerna, Five Prime Therapeutics, Gilead Sciences, Eli Lilly, Genentech, and KEW, is director for CytomX Therapeutics, and has ownership interest (including patents) in Cytomx Therapeutics and Entrinsic Health. J.A. Meyerhardt is a paid consulting for COTA Healthcare, a grant reviewer for Taiho Pharmaceutical, and is a consultant for Ignyta. No potential conflicts of interest were disclosed by the other authors.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication.

Conception and design: X. Zhang, C.S. Fuchs, W.C. Willett, E. Giovannucci, K. Wu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.S. Fuchs, A.T. Chan, E. Giovannucci

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Katagiri, M. Song, X. Zhang, F.K. Tabung, C.S. Fuchs, J.A. Meyerhardt, R. Nishihara, M. Iwasaki, S. Ogino, W.C. Willett, E. Giovannucci, K. Wu

Writing, review, and/or revision of the manuscript: R. Katagiri, M. Song, X. Zhang, D.H. Lee, F.K. Tabung, J.A. Meyerhardt, A.T. Chan, A.D. Joshi, M. Iwasaki, S. Ogino, W.C. Willett, E. Giovannucci, K. Wu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.H. Lee, J.A. Meyerhardt, A.T. Chan, W.C. Willett

Study supervision: C.S. Fuchs, A.T. Chan, K. Wu

We would like to thank the participants and staff of the Nurses' Health Study, the Nurses' Health Study II, and the Health Professionals Follow-up Study for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. The authors assume full responsibility for analyses and interpretation of these data.

This work was supported by NIH grants (P01 CA87969 to M.J. Stampfer; UM1 CA186107 to M.J. Stampfer; UM1 CA176726 to W.C. Willett; P01 CA55075 to W.C. Willett; UM1 CA167552 to W.C. Willett; U01 CA167552 to W.C. Willett and L.A. Mucci; P50 CA127003 to C.S. Fuchs; R01 CA137178 to A.T. Chan; K24 DK098311 to A.T. Chan; R35 CA197735 to S. Ogino; R01 CA151993 to S. Ogino; R21 CA222940 to K. Wu and R. Nishihara; R21 CA230873 to K. Wu and S. Ogino; K07 CA188126 to X. Zhang; and R00 CA207736 to F.K. Tabung), Nodal Award from the Dana-Farber Harvard Cancer Center (to S. Ogino), and grants from the Project P Fund, The Friends of the Dana-Farber Cancer Institute, Bennett Family Fund, and the Entertainment Industry Foundation through National Colorectal Cancer Research Alliance. This research was supported by a Stand Up To Cancer Colorectal Cancer Dream Team Translational Research Grant (grant number SU2C-AACR-DT22-17). Stand Up To Cancer is a division of the Entertainment Industry Foundation. Research grants are administered by the American Association for Cancer Research, the scientific partner of SU2C (to C.S. Fuchs). This work was also, in part, supported by an Investigator Initiated Grants from the American Institute for Cancer Research (AICR) to K. Wu.

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.