Coffee Metabolites and Kidney Disease

Coffee is one of the most widely consumed beverages in the world and has been the topic of hundreds of epidemiologic studies. Although large prospective analyses are few, meta-analyses report a significant 14%–28% lower risk of incident CKD, kidney failure, and death related to CKD among self-reported coffee drinkers compared with nondrinkers (1). Coffee-drinking behavior is captured relatively well by dietary assessment tools; however, measuring coffee exposure is a challenge (2). Coffee contains hundreds of phytochemical compounds that vary in quantity depending on factors such as bean variety, bean roast, and brewing method. Likewise, individuals vary in their ability to metabolize coffee compounds, and some of these compounds are subject to gut microbial metabolism, contributing further to intraindividual variability and assessment of coffee exposure. Metabolomic analyses of blood, urine, or other biospecimens show promise in this regard, increasing opportunities for insight into coffee-associated mechanisms of action (2). Indeed, several cross-sectional and clinical metabolomic studies have identified novel metabolites in urine and blood whose levels correlate with coffee-drinking behavior (3). The hope now is to use these “biomarkers” in prospective studies of coffee and health.

In this issue of CJASN, He et al. (4) identify 20 metabolites that correlate with self-reported coffee consumption in the Atherosclerosis Risk in Communities (ARIC) and Bogalusa Heart Study cohorts; all but one (2-hydoxy-3-methylvalerate) have been reported previously. They additionally test these 20 metabolites in ARIC with incident CKD defined by one or more criteria: baseline eGFR <60 ml/min per 1.73 m² and ≥25% eGFR decline at any subsequent visit; hospitalization or death related to CKD stage 3+; or kidney failure. After a median follow-up time of 24 years, two metabolites (O-methyl catechol sulfate and 3-methyl catechol sulfate) presented with a dose-response higher incidence of CKD. These metabolites reflect polyphenol metabolism by gut microbiota and are positively correlated with intake of coffee, a major but not sole contributor of polyphenols in the diet. Their positive relationship with CKD conflicts with epidemiologic studies of coffee (1). Coffee is an especially high source of chlorogenic acid, a polyphenol with beneficial links to inflammation, glucose homeostasis, and BP control (5); these properties might reduce risk of CKD. Indeed, quinate was identified as one of the most significantly associated metabolites with coffee consumption and was not related to CKD (4). Kidney failure is characterized by retention of uremic toxins originating from host or gut metabolism (6). Perhaps an accumulation of these sulfated catechol metabolites reflects early stages of impaired kidney function. Interestingly, serum levels of both metabolites were also positively associated with current smoking, an independent risk factor for CKD (7) that is also strongly correlated with coffee consumption behavior. Whether the findings hold among nonsmokers or with better adjustment of smoking is an open question. Sekula et al. (8) used the same metabolomics platform and reported no relationship (P=0.05) between either metabolite and incident CKD (defined as incident eGFRcr <60 ml/min per 1.73 m², n=95 cases) or eGFRcr decline over a mean follow-up time of 7 years in the Cooperative Health Research in the Region of Augsburg (KORA) cohort. Although differences in study design may have contributed to null findings, it is worth noting that KORA (approximately 12%) had fewer smokers than ARIC (approximately 28%). With potential confounding by smoking, one might have expected to observe a similar relationship between serum caffeine metabolites and CKD. However, smoking increases metabolism of caffeine and, thus, may have attenuated confounding. The null relationship with caffeine metabolites suggests that caffeine may not underlie the potential protective effect of coffee on CKD, although a nonsmoker analysis would provide clearer insight.

A third metabolite, glycochenodeoxycholate (GCDC), was inversely correlated with coffee consumption; the direction suggests an endogenous response to coffee and not a metabolite of the beverage itself. GCDC levels were associated with a higher risk of incident CKD in ARIC but only at relatively higher levels (4). Sekula et al. (8) provide no support for a linear association between GCDC and CKD (P=0.05); dose-dependent associations were not tested. GCDC is a conjugated bile acid (BA) previously identified among six other BAs that are differentially elevated in sera of patients with kidney failure compared with healthy controls (9). In a recent urine metabolomics study of patients with CKD, GCDC was positively associated with kidney failure and AKI as well as overall mortality (10). These end points overlap with CKD as defined by ARIC, and thus, GCDC may associate with CKD progression rather than CKD onset per se. Regardless, the potential for coffee to reduce risk or progression of CKD by altering BA metabolism merits further study. GCDC levels were also inversely correlated with smoking status, and thus, the protection offered by coffee via this pathway, if confirmed, may be stronger with better adjustment for smoking.

Metabolomics has emerged as a powerful approach to unveiling new insight to mechanisms linking diet to disease. However, this approach does not escape the limitations of epidemiology and also warrants caution when interpreting results. Moreover, metabolite measures capture a single time point and can vary with recent diet, medications, microbiome, and genetics; thus, they may not reflect habitual behaviors. Kidney function also affects metabolomic profiles and is one of the most important confounders in metabolomics studies of CKD (11). Metabolomics should not replace but rather should complement self-reported diet data. It would have been interesting to see whether the results of He et al. (4) were attenuated when adjusting for self-reported coffee or other diet sources of polyphenols. Integrating these data types should provide a better understanding of the role coffee and other diet factors play in the development of CKD or other diseases.

Disclosures

B. Burton-Freeman reports ownership interest in Amgen; receiving honoraria from the McCormick Science Institute, the Nutrient Institute, and the Watermelon Promotion Board; patents and inventions with Polyphenolics Inc.; and serving as a scientific advisor or member of McCormick Science Institute, Nutrient Institute, Journal of Nutrition and Healthy Aging, Journal of Berry Research, Journal of Food Science, and Institute of Food Technologist. M.C. Cornelis reports serving as a volunteer scientific advisor for International Life Sciences Institute, North America.

Funding

None.