Assessing Clinical Relevance of Uremic Toxins

Standard hemodialysis works. Indeed, this procedure maintains life in hundreds of thousands of people with little or no kidney function. Moreover, severe symptoms of uremia—coma, seizures, acidosis, hyperkalemia, and pulmonary edema—are readily reversed by dialysis. However, it is also now conspicuous that life on maintenance hemodialysis is short, with about 20% of the ESKD population in the United States dying each year, yielding a 3–5 years survival rate of only 50%. Although mortality is appallingly high, patient symptom burden is great, and for many patients, these symptoms are more daunting than the dismal mortality prospects. Quality of life has remained consistently poor for patients on hemodialysis. A recent survey reported that the physical symptoms of insomnia, fatigue, and cramping, and mood symptoms of anxiety, depression, and frustration were paramount concerns among patients (1).

Such symptoms and undoubtedly much of the mortality derive, at least in part, from solutes that are retained despite standard hemodialysis targeted at urea removal. Although the toxicity of urea continues to be debated, it seems unlikely that urea is the predominant cause of morbidity in this population. Specifically, urea was shown to be nontoxic when added in large concentrations to dialysate nearly 50 years ago and enhanced urea removal afforded no detectable benefits in the Effect of Dialysis Dose and Membrane Flux in Maintenance Hemodialysis (HEMO) study (2,3). In recent years, there has been increased interest in the harms of sodium. Although many hospitalizations, considerable morbidity, and perhaps even inflammation may be attributed to extracellular volume overload and/or sodium excess, sodium seems an unlikely causative agent for all dialysis-related symptoms and adverse sequelae—the aforementioned symptoms along with itching, cognitive impairment, platelet dysfunction, and vascular calcification, to name a few.

Chemical analyses, notably liquid chromatography–tandem mass spectrometry used with both targeted and untargeted approaches, are shedding new light on the detail and complexity of the altered chemical anatomy of patients on dialysis. Surveys of small organic compounds in individuals with ESKD find upward of 270 accumulated compounds and, when proteins and peptides are considered, even larger lists will ultimately be revealed (4). Mapping important signs and symptoms to the retained chemicals remains a major, multistep task. Such mapping will almost surely first involve metabolomics approaches. Chemicals identified by such untargeted approaches must then be validated in targeted analyses, and then sources of the chemicals must be identified. Finally, clinical trials aimed at reducing the putative culprit chemical and measuring its clinical effect must be conducted. Infusion of a candidate toxin into animals or its addition in vitro to a suitable tissue could speed discovery, but have been little used. Some find the prospects for success in this venture impossibly complex. However, we note that there are roughly 100-fold more genes than the above estimates of retained solutes, and major advances have been made in mapping genes to disease states. Although genes may be more constant than solutes, many genetic analytic approaches likely apply to the study of solutes, and the task is smaller by more than an order of magnitude.

We acknowledge that the problem is not simple. First, most of the relevant signs and symptoms have not been quantitatively determined in large enough ESKD populations with simultaneous plasma samples. Our knowledge of many important clinical findings, like symptoms, is far from granular. To some degree, challenges with quantifying such clinical data have led to a focus on events that are relatively easy to quantify such as death. Although such event data should not be ignored, patients have expressed greater concern over symptoms that are often minimized, if not ignored by us as clinicians, than over their shortened lifespan. Second, the retained solutes’ levels may vary because of yet unknown causes; diet, individual variation in metabolism, and microbial production are likely prominent effectors. In an analysis comparing indoxyl sulfate levels at the beginning to those at the end in participants in the control arm of the Frequent Hemodialysis Network trial, correlation across timepoints was poor (T.W. Meyer, personal communication). Thus, prediction of cardiovascular events on the basis of initial levels of a solute may be fraught, particularly when the time to event is long. Such variability may pose less of an obstacle to linking more proximal findings like insomnia or cognitive impairment to simultaneous solute measures. Third, solute levels may covary. We may associate one solute with a symptom when another solute produced or cleared by related processes is the true cause. Accumulation of a true toxin may be exaggerated by blockade of its secretion through residual kidney function because of competition by some other relatively inert accumulated solute(s) (5). The confusion as to which is at fault could only be resolved by studies comparing the isolated effects of each. Mounting evidence suggests that gut microbes account for a sizable fraction of retained solutes, and thus, interindividual variation in the gut microbiome may account for interindividual variations in solute levels (6). Such complexities and interindividual variations make isolating the clinical effects of toxins challenging.

The bacterial byproduct trimethylamine oxide (TMAO) has been identified by Hazen and colleagues as a predictor of cardiovascular disease in people with normal kidney function as well as those with predialysis CKD (7). However, its predictive power in ESKD is controversial. Using samples from 1232 participants in the HEMO study, Shafi et al. (8) reported that cardiovascular risk rose at higher levels of TMAO in patients on hemodialysis, especially among white participants. In contrast, studying 235 individuals with ESKD, Kaysen et al. (9) found no association between TMAO and cardiovascular outcomes.

Similarly, in this issue of the Clinical Journal of the American Society of Nephrology, Stubbs et al. (10) report no association between TMAO and cardiovascular outcomes in 1243 participants in the control arm of the Evaluation of Cinacalcet Hydrochloride Therapy to Lower Cardiovascular Events (EVOLVE) trial. Both of the reports by Stubbs et al. (10) and Shafi et al. (8) were based on over 1200 patients on hemodialysis derived from two different clinical trials with well adjudicated cardiovascular end points, making reasons for these discrepancies not immediately evident.

However, Stubbs et al.’s findings do raise interesting points. First, cardiovascular risk factors differ across dialysis and nondialysis populations. Cholesterol is a well known example. Also, the levels of TMAO in all three studies were enormous compared with those in the general population: roughly 10- to >40-fold higher. As a comparator, urea levels are only three- to five-fold higher among dialysis versus nondialysis populations. Presumably any toxin will have a level, or asymptote, above which adverse events do not increase. As a simple example, a serum potassium of 12 mEq/L is unlikely to be discernably worse than one of 10 mEq/L. If the solute has reached some asymptote of maximal toxicity, say ten-fold normal, then dissecting its individual contribution to cardiovascular burden in ESKD will be challenging.

Ultimately, clinical trials examining the effect of lowering the levels of a potential toxin will be required to establish a causal association between individual uremic toxins and outcomes like cardiovascular events. We should be mindful that enhancing toxin removal via dialysis may not be the most effective or efficient solution. Indeed, more and more urea removal schemes have shown disappointing efficacy (2). Rather, understanding the sources of toxins (diet, intestinal microbiomes, and altered metabolism) may provide the means by which to reduce their production, and hence, their levels. Using animal infusion of a toxin and/or addition in vitro of a toxin to relevant target tissues to gain insight into toxin source may offer an attractive way to inform and bolster subsequent clinical trials. Prioritizing this quest with clinically meaningful outcomes, especially those identified by patients themselves, seems a best approach (1).

Disclosures

J.E.F. has received speaking honoraria from American Renal Associates, American Society of Nephrology, Dialysis Clinic, Incorporated, National Kidney Foundation, and multiple universities, as well as research funding for studies unrelated to this project from the Renal Research Institute, a subsidiary of Fresenius Kidney Care North America. J.E.F. is on the medical advisory board for NxStage Medical and has received consulting fees from Fresenius Kidney Care North America.

T.H.H. has been a consultant and Scientific Advisory Board member for Tricida. He holds stock options in Tricida. He also has received research funding and honoraria for grant reviews from the National Institutes of Health. He also received remuneration as a Deputy Editor for the Journal of the American Society of Nephrology.