Plasma and Urine Metabolite Profiles Impacted by Increased Dietary Navy Bean Intake in Colorectal Cancer Survivors: A Randomized-Controlled Trial

Legumes are globally recognized as high-quality sources of dietary fibers with prebiotic functions and are sustainable sources of essential amino acids and many other bioactive plant secondary metabolites (1–3). Epidemiologic studies and multiple clinical cohort investigations now support regular consumption of common beans for colorectal cancer prevention (4–7), and additionally promote lower risk of chronic diseases such as obesity (8–13), cardiovascular disease (14, 15), and type II diabetes (16). Hughes and colleagues reported that 59 g of pinto beans per 100 g of diet reduced colon tumor (azomethane-induced) incidence by 50% in rats compared with 24% by casein-fed rats (17). Similar to the findings of previous studies, Hangen and Bennink found that diets fed to rats containing 75% navy beans resulted in 44% decreased colon tumor and adenocarcinoma incidence (18). The consumption of ∼½ cup of cooked dry beans per day for adults was practical and significantly increased intake of fiber, protein, and minerals such as folate, zinc, iron, and magnesium, while lowers the intake of saturated fats and total fat (19–21).

Emerging evidence exists for diet-derived small molecules to signal as modifiers of cancer risk and tumor behavior. Reliable dietary intake assessment methods are required to understand the associations between diet, nutritional status, and colon cancer risk. Nontargeted metabolomics is a well-established and sensitive tool for the comprehensive and concomitant characterization of multiple metabolites within a biological matrix, and is increasingly applied to cancer research to screen and discover shifts in metabolism (22, 23). In recent years, metabolomics has also been used for identifying putative biomarkers of food consumption and dietary patterns (24, 25). Less is known about whether bioactive food components reach effective cancer-protective concentrations in people due to extensive interindividual variability that can occur in host digestion and gut metabolism. Some mechanisms linking common dry bean consumption and cancer prevention include, but are not limited to, induction of apoptosis and modulation of cell cycle and proliferation (2, 26, 27). Other mechanisms for common bean intake to inhibit colon carcinogenesis involve changes to energy metabolism (28) and phase II detoxification (29) as shown by tissue microarray analysis in animals. Metabolomics has utility for measuring downstream products of metabolic disturbances shown by gene expression and for associations with environmental exposures involved in cancer risk (30, 31).

Plasma and urine are ideal biological fluids to examine the physiologic effects of dietary exposures and have shown sensitivity to detect subtle and major changes in metabolite abundance over time (32, 33). A nontargeted metabolite approach was applied herein to assess metabolic modulation of colorectal cancer risk following increased navy bean intake. The purpose of this study was to assess the impact of navy bean consumption on the plasma and urine metabolome for colorectal cancer prevention when compared with placebo-control foods, and to identify a suite of metabolites that originated from navy beans. It was hypothesized that increased consumption of navy beans would deliver bioactive metabolites to individuals at high risk for colorectal cancer recurrence and produce metabolic shifts in plasma and urine profiles.

Twenty overweight or obese colorectal cancer survivors that were a minimum of 4 months after cancer treatment were recruited for this single-blinded, placebo-controlled, randomized intervention trial as previously described (3, 34, 35). All the participants were blinded to the study arms, and only the study coordinator was aware of the participant study group allocation. The study design adhered to CONSORT guidelines (http://www.consort-statement.org/). The recruitment took place through the University of Colorado Health–North Cancer Center Network (Fort Collins, CO). Supplementary Fig. S1 illustrates the eligibility criteria and the allocation of 20 colorectal cancer survivors that completed a study titled “Beans/Bran Enriching Nutritional Eating for Intestinal Health Trial (BENEFIT)” (NCT01929122). Participants were randomized by body mass index (between 25 and 35), sex, and daily caloric intake (36) prior to dietary intervention. Supplementary Table S1 shows study population characteristics at baseline. In this 4-week intervention, participants assigned to the navy bean intervention consumed 35 g of cooked whole navy beans in powder form (daily intake of one meal and one snack that each contained 17.5 g of cooked navy bean powder; ref. 36). The control study arm received placebo meals and snacks without the addition of navy bean powder as previously described (34, 36). The study protocol and written informed consent was approved by the Colorado State University Research Integrity and Compliance Review Board and the University Colorado Cancer Center–North Institutional Review Board (Protocol # 09-1530H and 10-1038, respectively) in accordance with the 1964 Helsinki Declaration and its 2013 amendments. The 20 individuals completed the 4-week pilot dietary intervention trial between August 2010 and December 2014.

Fasting blood samples were collected into 4-mL ethylene-diamine-tetra-acetic acid (EDTA) blood collection tubes from the 20 participants who completed study visits at baseline, 2 weeks, and week 4 and kept immediately on ice until the centrifugation step to extract the plasma (1,500 rpm for 10 minutes). Plasma was further stored at −80°C until processed for metabolomics analysis using ultra-high-performance liquid chromatography–tandem mass spectrometry (UHPLC-MS/MS). At each study visit (baseline, 2 weeks, and 4 weeks), participants were given study-labeled containers for self-collection of first void of the morning urine. Urine was stored at −80°C until extracted for analysis by UHPLC-MS/MS. Urine samples were subjected to the osmolality normalization using a Fiske 210 Micro-Sample Osmometer to correct for the fluid intake variability between participants.

The global nontargeted metabolomics was performed through Metabolon, Inc. The extraction method has been described previously (33). Briefly, 80% methanol was added on plasma and urine in a ratio of 300 μL solvent per 100 mg plasma or urine. The samples were shaken for 2 minutes, and then centrifuged for 10 minutes at 4°C (12,000 rpm). The metabolite extracts from plasma and urine were analyzed using UHPLC-MS/MS with positive and negative ion mode electrospray ionization (ESI).

Samples were analyzed using a Waters ACQUITY UHPLC system coupled with a Thermo Scientific Q-Exactive high resolution/accurate mass spectrometer and interfaced with a heated ESI (HESI-II) source and Orbitrap mass analyzer. Prior to analysis, samples were resuspended in acidic or basic UHPLC-compatible solvents. The acidic and basic solutions were analyzed using positive and negative ESI, and both hydrophilic interaction chromatography (HILIC) and reverse phase (RP) chromatography (37). In HILIC mode, the extracts were injected into a BEH amide column (2.1 mm × 100 mm, 1.7 μm; Waters Corporation). The mobile phase delivered at 0.35 mL/minute consists of water, acetonitrile, and 10 mmol/L ammonium formate at pH 10.8 (HILIC positive), methanol + water, and 6.5 mmol/L ammonium bicarbonate at pH 8.0 (HILIC negative). In RP-positive and -negative ion-optimized conditions, extracts were gradient eluted from a C18 column (2.1 mm × 150 mm, 1.7 μm; Corporation). In RP positive, extracts were eluted from the column with water + methanol, containing 0.05% perfluoropentanoic acid and 0.1% formic acid, while in RP negative, water + methanol, containing 0.05% perfluoropentanoic, 0.01% formic acid, and acetonitrile was used as mobile phase. For the MS data acquisition of study meals and snacks, plasma and urine, an extended dynamic exclusion was used with the instrumental ion acquirement of 70 to 1,000 m/z. The peak picking process was performed by Metabolon through in-house peak detection and integration software. Compound identification (level 1 identification) was completed using library of over 3,300 purified standards and based on the experimentally MS-MS spectra matching the accurate mass the standard within 8 ppm, retention time, and m/z ratio.

The nontargeted metabolomics analysis of cooked navy bean powder has been previously described (3). The plasma and urine metabolites from participants in both dietary groups were cross-referenced with metabolites identified in the navy beans and in the one meal and three snacks metabolome after 4 weeks of either control or navy bean consumption.

Plasma and urine metabolite intensity levels underwent median-scaled normalization as previously described (33). A Welch two-sample t test and two-way ANOVA were performed for between-group comparisons. Repeated-measures ANOVA was used for comparisons at 2 and 4 weeks from baseline and within each study group. An estimate of false discovery rate (q-value) was calculated. Statistical significance was defined as P values of <0.05 and q-values below the threshold of <0.2 for within-group comparisons.

The metabolomes of one meal and three snacks used in this study comprised of 653 metabolites with confirmed identifications. Eight groups were used for the overarching metabolite classifications: amino acids, carbohydrates, cofactors and vitamins, energy metabolism, lipids, nucleotides, peptides, and xenobiotics. The meals and snacks metabolome contained 143 amino acids, 41 carbohydrates, 27 cofactors and vitamins, 14 energy metabolism-related metabolites, 226 lipids, 61 nucleotides, 32 peptide, and 111 xenobiotics. Supplementary Table S2 shows the metabolite raw abundance for one meal and three snacks. Composition of meals and snacks was analyzed by chemical classes for placebo-control (0 g/serving navy bean powder) and navy bean intervention (17.5 g/serving navy bean powder). Supplementary Fig. S2A displays the meals and snacks metabolite profiles for placebo-control and navy bean intervention, whereby each bar represents the total number of identified metabolites in the chemical class, and the shaded portion represents the number of metabolites that appeared in navy bean intervention meals and snacks but not in the control. The navy bean source of metabolites in meals and snacks was determined by the overlapping presence of metabolite in the navy bean powder metabolome. Supplementary Fig. S2B–S2D displays normalized relative abundance values in placebo-control and navy bean intervention meals and snacks for the following metabolites: pipecolate, S-methylcysteine, and N-delta-acetylornithine. Pipecolate, S-methylcysteine, and N-delta-acetylornithine had higher abundance in the bean intervention meals and snacks compared with control that was due to the high relative abundance shown for these compounds in the navy bean powder.

There were 854 identified metabolites from plasma (N = 20 participants) across all time points, including baseline, 2 weeks, and 4 weeks after intervention. Supplementary Table S3 shows the relative abundance of plasma metabolites classified into metabolic pathways for all participants. Table 1 lists 22 plasma metabolites that were significantly modulated in either the control, navy bean, or both study groups for 20 participants at 2 and/or 4 weeks compared with baseline (P < 0.05, q < 0.2). All data are presented as the mean fold change in the median-scaled relative abundance of each metabolite. After consuming navy beans for 2 weeks, 21 metabolites including two gut microbial metabolites, four amino acids, one cofactors and vitamins, 11 lipids, and three xenobiotics were significantly modulated in plasma when compared with baseline.

Navy bean powder significantly increased three amino acid metabolites in plasma of colorectal cancer survivors at week 2 and at week 4 when compared with the baseline levels (Table 1). Figure 1A–C illustrates median-scaled relative abundance for pipecolate, S-methylcysteine, and N-delta-acetylornithine. Individual responses over time are shown in Supplementary Fig. S3. Notably, these three metabolites were highlighted for abundance in navy bean powder and study meals and snacks metabolomes.

Table 2 lists 16 plasma metabolites with significant differences in the abundance between intervention and control groups for the 20 participants at 4 weeks (P < 0.05, q < 0.2). For all participants, data are presented as the mean fold difference between intervention and control groups at 4 weeks. The five plasma metabolites increased in abundance for the navy bean group were 3-(4-hydroxyphenyl)propionate (4.48-fold), S-allylcysteine (4.08-fold), pipecolate (3.88-fold), S-methylcysteine (1.92-fold), and 2,3-dihydroxy-2-methylbutyrate (1.34-fold). The 11 plasma metabolites with significant decreased levels compared with control at 4 weeks were betaine (0.84-fold), 1-arachidonoyl-GPC (20:4n6) (0.80-fold), 5alpha-pregnan-3beta-20alpha-diol monosulfate (0.12-fold), choline (0.80-fold), DHA (0.74-fold), docosahexaenoylcarnitine (C22:6) (0.63-fold), eicosapentaenoate (EPA) (0.77-fold), linoleoylcarnitine (C18:2) (0.71-fold), 5,6-dihydrothymine (0.71-fold), gamma-glutamyl-2-aminobutyrate (0.30-fild), and benzoate (0.77-fold). The eight metabolites identified in the plasma that were also present in the cooked navy beans include pipecolate, S-methylcysteine, betaine, EPA, docosahexaenoate (DHA), choline, gamma-glutamyl-2-aminobutyrate, and benzoate (see Table 2, note “a”) and 6 metabolites were identified in the study meals and snacks metabolome including 3-(4-hydroxyphenyl)propionate, betaine, pipecolate, S-methylcysteine, choline, and benzoate (see Table 2, note “b”). Figure 1D illustrates median-scaled relative abundance for 3-(4-hydroxyphenyl)propionate. This phenolic acid derivative was significantly increased after consuming navy beans for 4 weeks when compared with the control group. This phenolic acid was also found in the study meals and snacks metabolome (Table 2). Figure 1E and F shows the median-scaled relative abundances for EPA and benzoate. These endogenous and/or diet-derived metabolites were significantly decreased after consuming navy beans for 4 weeks when compared with the control group (Table 2). Both metabolites were identified in the navy bean metabolome, while benzoate was also identified in the study meals and snacks. Figure 1G shows a Venn diagram of 33 plasma metabolites with fold change or fold differences identified at 2 or 4 weeks after intervention. The center metabolites in the Venn diagram illustrate those with overlapping significance from each of the statistical comparisons applied, and were pipecolate, S-methylcyteine, and 2,3-dihydroxy-2methylbutyrate.

There were 703 identified metabolites from urine of 20 participants at all time points. The urine metabolites were classified in the metabolic pathways as shown for plasma. Supplementary Table S4 lists the relative abundance of all the identified urine metabolites and the respective metabolic pathways for the control and navy bean groups. Table 3 lists 10 urine metabolites that were significantly modulated in either the control, navy bean, or both study groups for 20 participants at 2 and/or 4 weeks compared with baseline (P < 0.05, q < 0.2). All data are presented as the mean fold change in the median-scaled relative abundance of each metabolite. Figure 2A–D illustrates median-scaled relative abundance for gentisate, hippurate, salicylate, and N2,N5-diacetlyornithine that were significantly increased at week 4 when compared with their baseline in the intervention group. Individual changes over time are shown in Supplementary Fig. S4. These metabolites were chosen for their potential colorectal cancer protection, and the four metabolites were identified in navy bean and study meals and snacks metabolomes.

Table 4 lists 16 urine metabolites with significant differences in the abundance between intervention and control for the 20 participants at 4 weeks (P < 0.05, q < 0.2). For all participants, data are presented as the mean fold difference between intervention and control at 4 weeks. The 14 urine metabolites increased in abundance for the navy bean group were 1-methylhistamine (1.72-fold), cysteine (1.6-fold), N2,N5-diacetylornithine (1.55-fold), N-acetyltaurine (1.57-fold), pyroglutamine (1.74-fold), S-adenosylhomocysteine (SAH) (2.09-fold), N6-carbamoylthreonyladenosine (1.44-fold), guanosine (2.81-fold), 2-keto-3-deoxy-gluconate (2.18-fold), 4-hydroxyhippurate (1.92-fold), alliin (3.34-fold), O-sulfo-L-tyrosine (1.33-fold), and thioproline (2.66-fold). The three urine metabolites that decreased in abundance for the navy bean group were dopamine 3-O-sulfate (0.45-fold), indolepropionate (0.35-fold), and isovalerylglutamine (0.53-fold). Among these 17 urine metabolites, eight metabolites were identified in the navy beans, including cysteine, N2,N5-diacetylornithine, pyroglutamine, SAH, N6-carbamoylthreonyladenosine, guanosine, 4-hydroxyhippurate, and O-sulfo-L-tyrosin (see Table 4, note “a”) and five urine metabolites were identified in the study meals and snacks metabolome, including cysteine, pyroglutamine, SAH, guanosine, and O-sulfo-L-tyrosine (see Table 4, note “b”). Navy bean–derived metabolites were distinguished for presence and abundance in urine of the intervention group compared with control. Figure 2D–F illustrates the median-scaled relative abundance for amino acids with anti-inflammatory potential—N2, N5-diacetylornithine, cysteine, and SAH—which were identified in the navy bean metabolome. The Venn diagram shown in Fig. 2G highlights 23 urine metabolites (statistical significance by fold change or fold difference) in urine at 2 or 4 weeks after dietary bean intervention. N2,N5-diacetylornithine was the only urine metabolite with overlap in the analyses at 4 weeks in the intervention group.

This study established that increased navy bean consumption for 4 weeks produced metabolic shifts in plasma and urine profiles of overweight and obese colorectal cancer survivors at risk for recurrence. The plasma and urine metabolites that showed changes in abundance over time or differences between the two study groups were also compared with the phytochemical compounds that originate from navy beans (including the meals and snacks that contained navy beans). This integration of the navy bean metabolites with the results from changes in plasma and urine has established novel relationships between bean exposure in the diet and cancer prevention biomarkers in adults. There are also a number of metabolites discussed below from distinct metabolic pathways that support novel associations for bean intake with anti-inflammatory, antioxidant, and anticancer mechanisms that may reduce cancer risk. Thus, plasma and urine metabolites have the potential to dually serve as biomarkers of dietary navy beans exposure and metabolic health to prevent and control colon cancer.

Six plasma metabolites modulated by navy bean consumption and that have cancer control and prevention relevance from this study are S-methylcysteine, pipecolate, 3-(4-hydroxyphenyl)propionate, N-delta-acetylornithine, S-allylcysteine, and 2,3-dihydroxy-2-methylbutyrate (Tables 1 and 2; Fig. 2). S-methylcysteine, a water-soluble organosulfur compound, may be a biomarker of dietary exposure as it was previously identified from common bean (Phaseolus vulgaris; refs. 38, 39). Cysteine-related components have also demonstrated suppression of chemical induced carcinogenesis (40). S-methylcysteine was reported to inhibit the formation of glutathione S-transferase placental form–positive foci (GST-P-positive foci), which are markers for preneoplastic lesions (41) and was also found to inhibit ornithine decarboxylase (ODC), an enzyme that is upregulated in a wide variety of cancers (42). ODC is essential in the biosynthesis of polyamines (43), namely, antioxidants important for stabilizing DNA structure (44). Lack of ODC has been found to cause cell apoptosis in embryonic mice, induced by DNA damage (45). S-allylcysteine was also increased in plasma (Table 2) and has antioxidant properties (46, 47), cholesterol-lowering (48), and anticancer effects on clusters of abnormal tube-like glands in the lining of the colon and rectum called aberrant crypt foci. A decreased level of urinary cysteine warranted discussion as cysteine is a unique thiol containing amino acid that can undergo oxidation/reduction (redox) reactions with antioxidant properties and may reduce inflammation (49, 50). These metabolite changes collectively may reduce cell division to decrease risk for colorectal cancer recurrence.

Pipecolate represents another candidate dietary biomarker of bean consumption (38) that has functional relevance to this prospective cohort investigation because this phytochemical serves as a precursor to gut microbial metabolism and has anti-inflammatory, antitumor, and antibiotic properties (51–53). 3-(4-hydroxyphenyl)propionate was also increased following navy bean intake and is a small phenolic acid produced from biotransformation of large polyphenols by human colonic microflora (54). Phenolic acids exert effects on carcinogen bioactivation, cell signaling, cell-cycle regulation, angiogenesis, oxidative stress, and inflammation (55, 56). 2,3-dihydroxy-2-methylbutyrate is a normal organic acid constituent of the gut and the increase measured herein for plasma of navy bean consumers suggested beneficial roles against inflammation and alterations to microbiota. Polyunsaturated fatty acids (PUFA), such as EPA (20:5n3) that is a navy bean metabolite, were shown to exert anticancer activities in various stages of cancer progression, such as cell proliferation, metastasis, inflammation, cell survival, and angiogenesis (57, 58).

This next group of metabolites has mechanistic links to diverse epigenetic modifications involved in cancer progression. Benzoic acid (navy bean metabolite) has naturally occurring derivatives with potential to suppress cancer cell growth by blocking histone deacetylase, a key enzyme involved in tumor regulatory gene expression (59). The decreased urinary SAH following daily bean intake is important as an inhibitor of methyltransferases, the key enzymes for DNA methylation that may cause aberrant DNA methylation patterns in colon cancer (60). Higher abundance of SAH could be important to block aberrant DNA methylation and dysregulation of histones and gene expression that stimulate carcinogenesis (61). Both choline and betaine were decreased in plasma from this study, and the roles in colorectal cancer merit attention as they are essential to one-carbon metabolism (62–64). The lack of choline was shown to promote fatty liver disease, DNA hypomethylation, and tumor development (in the absence of carcinogens; ref. 65) and an epidemiologic study showed that the risk of colorectal adenomas statistically increased with increasing choline intake (65). Betaine is notably an oxidation product of choline, and a methyl donor that has nutrient contributions for the prevention of chronic diseases (66).

There were multiple metabolites identified in this study with associated mechanisms to colorectal cancer prevention that relate to metabolism by the gut microbiota. These may include, but were not limited to, 3-(4-hydroxyphenyl)lactate, N2,N5-diacetylornithine, hippurate, and salicylate. High urinary levels of 3-(4-hydroxyphenyl)lactate were found in individuals with tyrosine malabsorption and bacterial overgrowth (67, 68), and our data suggest improved tyrosine absorption and balanced gut microflora with navy bean intake. Dietary bean intake influences on metabolism by gut microbiota were also shown by changes to indole propionate, 4-hydroxyhippurate, and 3-(4-hydroxyphenyl)propionate that have been reported previously in stool (3, 35).

The limitations of this study include the small cohort size. The acute (one-month) study duration of the diet intervention may also be considered a limitation, yet these metabolite changes detected early-on following dietary change may have long-term impacts. The study participants were free-living individuals and the study-provided meals and snacks contributed ∼35% of daily intake. Follow-up investigations will utilize a larger sample size, and a longer duration of elevated navy bean exposure may be needed to observe colon tissue metabolic changes that will improve cancer control, treatment, and prevention outcomes.

This study supports that an elaborate array of interactions between food, host digestion, and gut microbial metabolism occurs in overweight and obese colorectal cancer survivors. Plasma and urine metabolites that merit attention for validation after the daily consumption of navy beans in adults are 2,3-dihydroxy-2-methylbutyrate, pipecolate, and S-methylcysteine (in plasma) and N2,N5-diacetylornithine and 4-hydroxyhippurate (in urine). The metabolite results shown for blood and urine were integrated with the analysis of phytochemicals originating from the navy beans, and revealed the amino acid and lipid metabolism pathways as mechanistic targets to reduce colorectal cancer recurrence following dietary bean intervention. Assessing the bioactivity and bioavailability of navy bean–derived phytochemicals in human colon tissue is one future analytical approach to validate food-exposure biomarkers with primary and secondary colorectal cancer prevention outcomes.

All procedures performed in this study were in accordance with the ethical standards of the Colorado State University and University of Colorado Health–North Institutional Review Board, and with the 1964 Helsinki Declaration and its 2013 amendments. Informed consent was obtained from all individual participants before included in the study. This trial was registered at clinicaltrials.gov under NCT01929122. The data sets used and/or analyzed during the current study are available as Supporting Information along with this article.

No disclosures were reported.

I. Zarei: Data curation, writing–original draft. B.A. Baxter: Data curation, writing–review and editing. R.C. Oppel: Writing–review and editing. E.C. Borresen: Conceptualization, methodology, and project administration. R.J. Brown: Conceptualization, methodology, and project administration. E.P. Ryan: Conceptualization, data curation, visualization, methodology, writing–original draft, project administration, and writing–review and editing.

Funding for this study was provided by NIH–NCI (1R21CA161472), National Institute of Food and Agriculture (NIFA; 2016-67001-24538), the University of Colorado Cancer Center–Division of Cancer Prevention and Control pilot award program, and the Dry Bean Health Research Group. The authors appreciate the collaboration of University of Colorado Health–North Cancer Center and the Oncology Clinical Research team. The authors wish to thank the study subjects for their participation.

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.