Visual Abstract
Abstract
Background and objectives Residual kidney function is important to the health and wellbeing of patients with ESKD. We tested whether the kidney clearances of proximal tubular secretory solutes are associated with burden of uremic and heart failure symptoms among patients on peritoneal dialysis with residual kidney function.
Design, setting, participants, & measurements We enrolled 29 patients on incident peritoneal dialysis with residual urine output >250 ml daily. We used targeted liquid chromatography-mass spectrometry to quantify plasma, 24-hour urine, and peritoneal dialysate concentrations of ten tubular secretory solutes. We calculated the kidney and peritoneal dialysis clearances of each secretory solute, creatinine, and urea, and we estimated a composite kidney and peritoneal secretion score. We assessed for uremic symptoms using the Dialysis Symptom Index and heart failure–related symptoms using the Kansas City Cardiomyopathy Questionnaire. We used linear regression to determine associations of composite secretory solute clearances and GFRurea+Cr with Dialysis Symptom Index symptom score and Kansas City Cardiomyopathy Questionnaire summary score.
Results Mean residual kidney clearances of creatinine and urea were 8±5 and 9±6 ml/min per 1.73 m2, respectively, and mean GFRurea+Cr was 8±5 ml/min per 1.73 m2. The residual kidney clearances of most secretory solutes were considerably higher than creatinine and urea clearance, and also, they were higher than their respective peritoneal dialysis clearances. After adjustments for age and sex, each SD higher composite kidney secretion score was associated with an 11-point lower Dialysis Symptom Index score (95% confidence interval, −20 to −1; P=0.03) and a 12-point higher Kansas City Cardiomyopathy Questionnaire score (95% confidence interval, 0.5- to 23-point higher score; P=0.04). Composite peritoneal dialysis secretion score was not associated with either symptom assessment.
Conclusions Residual kidney clearances of secretory solutes are higher than peritoneal dialysis clearances. Kidney clearances of secretory solutes are associated with patient-reported uremic and heart failure–related symptoms.
- proximal tubular solute clearance
- residual kidney function
- uremic symptoms
- creatinine
- renal dialysis
- Linear Models
- urea
- Symptom Assessment
- peritoneal dialysis
- Kidney Failure
- Chronic
- Peritoneum
- Dialysis Solutions
- heart failure
- Mass Spectrometry
- Chromatography
- Liquid
- Patient Reported Outcome Measures
- Cardiomyopathies
Introduction
Among patients receiving maintenance dialysis, the contribution of residual kidney function (RKF) toward health and wellbeing is well established. Observational studies consistently demonstrate associations of RKF with improved survival and better quality of life (1–6). The physiologic importance of RKF is highlighted by recognized limitations of currently available dialysis therapies, which cannot remove protein-bound solutes or maintain minute-to-minute electrolyte homeostasis. Consequently, estimation of RKF remains an important component of dialysis clinical care.
Current assessment of RKF is on the basis of the GFR, which can be measured by the average of creatinine and urea clearance in ESKD (7,8). However, GFR incompletely describes the multifaceted biologic functions of the kidneys. The secretion of organic solutes by the proximal tubules is a vital intrinsic kidney function that may be particularly important in ESKD because many secreted solutes are highly protein bound, preventing adequate clearance by glomerular filtration or by currently available dialysis treatments. Many solutes that are eliminated predominantly by tubular secretion are uremic toxins that accumulate and are suspected to contribute to adverse clinical outcomes (9–11). Moreover, proximal tubular secretion is an active process requiring cellular energy and coordination of numerous transporters, suggesting potential contrasts with glomerular filtration (12,13). Despite its potential importance, tubular secretory clearance is rarely estimated in ESKD.
We used a novel targeted mass spectrometric assay to measure the kidney and peritoneal clearances of tubular secretory solutes in an incident cohort of patients on peritoneal dialysis (PD). We compared residual kidney and PD clearances of secretory solutes, creatinine, and urea, and we tested whether residual clearances of these solutes were associated with the severity of uremic and heart failure symptoms.
Materials and Methods
Patient Recruitment and Data Collection
Between June 2017 and October 2018, we recruited 29 patients on incident PD from outpatient units affiliated with the Northwest Kidney Centers (Seattle, WA). Eligibility criteria included PD initiation within the previous 1 year and self-reported residual urine output >250 ml daily. Major exclusions were a history of kidney transplantation, previous hemodialysis (HD), age <18 years old, non-English speaking, and inability to provide informed consent. We further excluded persons who regularly used medications that could interfere with tubular secretion: probenecid, cimetidine, ranitidine, mycophenolate, olmesartan, indomethacin, trimethoprim-sulfamethoxazole, and procainamide. The University of Washington’s institutional review board approved the study, and all participants provided informed consent. During the visit, each patient brought in clinical collections of 24-hour urine and dialysate drainage. Clinical staff collected a blood sample, an aliquot from the 24-hour urine specimen, and an aliquot containing equal proportions (1/1000 of drainage volume) of dialysate effluent from each exchange over the preceding 24 hours. We centrifuged all blood samples immediately after collection and stored all specimens at −80°C until assay.
Measurement of Tubular Secretory Solutes
Through literature review, we initially selected a panel of secretory solutes on the basis of specificity for known proximal tubular transporters, an increase in circulating concentrations following transporter knockout in experimental models, a high reported protein binding percentage, and reported kidney clearances that substantially exceed GFR (9,10,14–18). We developed a targeted liquid chromatography-tandem mass spectrometry assay for these solutes in plasma and urine using labeled internal standards as previously described (19,20). Briefly, we extracted plasma solutes from bound proteins using protein precipitation with acidified acetonitrile and solid-phase extraction through phospholipid removal. We extracted urine solutes using solid-phase extraction using either a mixed cation exchange extraction plate or a hydrophilic-lipophilic binding extraction plate. We reconstituted dried extracts in acetonitrile/formic acid and injected solution onto a Restek PFPP chromatographic column. The column eluate was subsequently introduced into a triple quadrupole tandem mass spectrometer (Sciex 6500). To maximize the accuracy of the calibration and minimize laboratory imprecision, we used standard addition of purified analytes that were quantified using nuclear magnetic resonance to determine the concentration of the analytes in the calibrator. We performed further validation experiments to maximize precision. Inter- and intra-assay coefficients of variation for individual solutes in plasma and urine range from 3.4% to 14.7% (Supplemental Table 1).
We were unable to determine the lower limits of detection for three solutes due to isobaric interference (adipic acid, suberic acid, and succinic acid), and therefore, we excluded these solutes from further analyses. We further excluded pantothenic acid from consideration on the basis of previous studies demonstrating glomerular filtration and tubular reabsorption of this compound (21,22), and we excluded 3-hydroxy hippurate due to implausibly high kidney clearance. Subsequent protein binding studies performed in our laboratory revealed that three additional solutes (isovalerylglycine, tiglylglycine, and xanthosine) had lower binding percentages than previously reported (Supplemental Table 1). However, we retained these solutes in this study because their kidney clearances greatly exceeded GFR, suggesting tubular secretion as the major kidney pathway of elimination.
Symptom Assessment
We administered the Dialysis Symptom Index (DSI) (23) to evaluate uremic symptoms and a modified version of the Kansas City Cardiomyopathy Questionnaire (KCCQ) (24) to further elicit specific symptoms of dyspnea, fatigue, and lower-extremity swelling. The DSI queries regarding the presence and severity of 30 physical or emotional symptoms during the past week. Participants rated each symptom using a five-point severity scale. The overall DSI symptom score ranges from 0 to 150, with higher scores indicating worse symptom burden. The DSI was systemically developed and validated among patients on long-term dialysis and demonstrates good test-retest reliability and content validity in HD. The questionnaire assesses numerous symptoms that are both highly prevalent and clinically relevant in ESKD on the basis of literature review, dedicated focus groups consisting of patients and providers, and experts’ assessments (23). However, we currently know very little regarding which specific symptoms are causally related to the retention of protein-bound solutes. Therefore, we assessed the association of secretory solute clearance with overall DSI severity.
The KCCQ is a 23-item questionnaire that evaluates the severity of dyspnea, fatigue, and lower-extremity swelling and the effect of these symptoms on physical functioning and quality of life. The KCCQ was developed and validated in heart failure populations (24), and it has been modified for use in CKD (25,26). However, this instrument has not been validated in ESKD. We transformed raw KCCQ scores into a summary score ranging from 0 to 100, with higher scores indicating better overall function (24). Prior studies have suggested that a KCCQ summary score <75 indicates clinically significant symptoms (24,27,28). The KCCQ has been well validated and carries prognostic significance in patients with established heart failure and in patients with CKD but no history of heart failure (25,27,29). One participant did not complete the questionnaires and was excluded from subsequent analyses involving symptom assessment.
Measurement of Other Study Data
Study participants provided self-reported demographics and medical conditions via study questionnaires. We used the electronic medical record to ascertain active medications, modality of PD, and the clinically suspected cause of ESKD. Clinical staff measured height, weight, 24-hour urine volume, and 24-hour dialysate volume during the adequacy visit. We measured serum and urine creatinine concentrations using the modified Jaffe method and urine urea concentration using enzymatic conductivity on the Beckman Coulter DxC 600.
Calculations
We calculated the kidney and peritoneal clearances of urea, creatinine, and each secretory solute as the volume of blood that is cleared of the solute per minute (milliliters per minute):
where [x] represents the solute concentration in urine, plasma, or dialysate and V represents the urine or dialysate volume per minute over the collection period. All clearances represent total urinary clearances, which include the bound and unbound fractions. We measured GFRurea+Cr by calculating the average of urea and creatinine clearances and standardizing to 1.73 m2 body surface area (milliliters per minute per 1.73 m2) according to clinical guidelines established by the National Kidney Foundation (30). We also standardized all clearance values to the same units.
Statistical Analyses
We expressed kidney and peritoneal solute clearances as means and 95% confidence intervals (95% CI). We computed a composite score for kidney and peritoneal secretory clearance, which represents a scaled average of the ten individual secretory clearances. We first standardized log-transformed secretory clearance of each individual solute to a 0–100 scale as described below:
We then determined the average of the ten standardized secretory clearances to create a composite secretion score. We used scatter plots, Pearson correlation coefficients, and linear regression to determine associations of secretory solute clearances and GFRurea+Cr with DSI and KCCQ scores. Given the small sample size, we were parsimonious in the selection of covariates: age, sex, and GFRurea+Cr. P values are presented without correction for multiple comparisons. A nominal P value of 0.05 was taken as evidence of statistical significance for regression analyses. Analyses were conducted using STATA 14 (StataCorp, College Station, TX).
Results
Description of the Study Population
Among the 29 study participants, the mean duration of PD was 4.0±2.9 months. The mean measured residual GFRurea+Cr was 8±5 ml/min per 1.73 m2 (Table 1). Mean age of the study cohort was 55±12 years old; 69% of participants were men, and 52% were white. Hypertension (96%) and diabetes (48%) were the most prevalent comorbidities. Detailed dialysis prescriptions are provided in Supplemental Table 2. Metabolic parameters of hemoglobin, bicarbonate, parathyroid hormone, and phosphorus levels were within target ranges for ESKD. The mean 24-hour ultrafiltration volume was 652±840 ml (Supplemental Table 2). Higher kidney clearances of secretory solutes, reflected by the composite secretion score, were correlated with higher residual urine volume (Rho=0.66; P=0.001) and with lower 24-hour ultrafiltration volume (Rho=−0.62; P<0.001).
Participant characteristics
Comparison of Residual Kidney and PD Solute Clearances
The mean residual kidney clearances of creatinine and urea were 8±5 and 9±6 ml/min per 1.73 m2, respectively (Table 2). The kidney clearances of most secretory solutes were considerably higher than that of creatinine or urea, with the highest values observed for hippurate (mean of 95±72 ml/min per 1.73 m2) and dimethyluric acid (mean of 57±67 ml/min per 1.73 m2). In contrast, the kidney clearance of p-cresol sulfate (mean of 1±0.8 ml/min per 1.73 m2) was lower than that of creatinine and urea. With the exception of p-cresol sulfate, the kidney clearance of all other solutes exceeded GFR, with the ratio of residual secretory clearance to glomerular filtration ranging from 1.03 for indoxyl sulfate to 13.7 for hippurate. The residual kidney clearance of each secretory solute was substantially higher than its respective PD clearance, ranging up to 23-fold higher for kynurenic acid clearance. The peritoneal clearance of most secretory solutes was lower than the average of peritoneal urea and creatinine clearances. Correlations between kidney secretory clearances and GFRurea+Cr ranged from 0.38 for dimethyluric acid to 0.91 for isovalerylglycine (Supplemental Table 3). Secretory solute concentrations in urine were substantially higher compared with concentrations in dialysate and plasma (Supplemental Table 4). Composite kidney solute clearance was inversely correlated with composite peritoneal solute clearance (Supplemental Figure 1).
Residual kidney and peritoneal dialysis solute clearance (n=29)
Associations of Residual Kidney and Peritoneal Solute Clearances with DSI Symptom Score
The median overall DSI symptom severity score was 25 (interquartile range [IQR], 10–44), and the median number of symptoms was 9 (IQR, 4–14). The most prevalent symptoms were muscle cramps (61%), leg swelling (59%), fatigue (57%), and difficulty with sexual arousal (57%) (Supplemental Table 5). Higher residual GFRurea+Cr and higher composite kidney solute clearance were correlated with lower DSI scores. This correlation tended to be modestly higher for composite kidney solute clearance compared with GFRurea+Cr (Figure 1, Table 3). Among the individual solutes, the residual kidney clearances of tiglylglycine, kynurenic acid, and isovalerylglycine were most strongly correlated with DSI scores. After adjustments for age and sex, each SD higher composite kidney secretion score was associated with an estimated 11-point lower DSI score (95% CI, 20- to 1-point lower; P=0.03) (Table 5). This association was similar in magnitude after adjustment for residual GFRurea+Cr. Conversely, composite peritoneal solute clearance was not associated with uremic symptom severity (Figure 1, Tables 4 and 5). In examining individual symptoms, higher kidney secretory clearance and GFR were most strongly associated with decreased appetite, leg swelling, and insomnia (Supplemental Table 6).
Kidney tubular soute clearance correlates with severity of uremic and heart failure symptoms. Scatter plots with fitted least squares regression lines are shown. Rho and P values generated from Pearson regression are shown. GFRurea+Cr is expressed in milliliters per minute per 1.73 m2. Composite secretion score is standardized on a scale of 1–100.
Correlations of kidney solute clearances with symptoms (n=28)
Correlations of peritoneal solute clearances with symptoms (n=28)
Association of kidney solute clearances with symptoms (n=28)
Associations of Residual Secretory Clearances with KCCQ Summary Score
The median overall KCCQ score was 80 (IQR, 51–91), and 43% of participants had overall KCCQ scores <75. Higher residual GFRurea+Cr and higher residual kidney solute clearances tended to track with higher (better) KCCQ scores (Figure 1, Table 3); however, correlations were weaker than those observed for DSI scores. After adjustment for age and sex, each SD higher composite kidney secretion score was associated with an estimated 12-point higher KCCQ score (95% CI, 0.5- to 23-point higher score; P=0.04). However, this association was weakened by adjustment for residual GFRurea+Cr. Higher peritoneal solute clearance was correlated with lower KCCQ scores (Figure 1, Table 4); however, this association was not statistically significant (Table 5). Correlations of combined kidney and PD solute clearances were comparable with correlations with kidney solute clearances alone given the relatively small contribution of PD clearance toward total clearance (Supplemental Table 7).
Discussion
In a primary cohort of patients recently initiating PD, we observed substantially higher residual kidney clearances of tubular secretory solutes compared with their respective peritoneal dialytic clearances. The residual kidney clearances of most secretory solutes were also considerably higher than the clearances of creatinine and urea. Higher kidney secretory clearances but not PD clearances were correlated with a lower burden of uremic and heart failure symptoms. Taken together, our findings suggest that the residual kidney clearance of tubular secretory solutes is an important component of RKF that may hold clinical significance.
Consistent with our findings, two prior studies of PD have shown the superiority of residual kidney clearance compared with dialytic clearance in eliminating p-cresol sulfate, a small protein-bound molecule that is primarily cleared by proximal tubular secretory transport. These studies found a four-times higher residual kidney clearance of p-cresol sulfate compared with its peritoneal clearance (31,32). Similarly, among patients on HD with RKF, the fraction of p-cresol sulfate, indoxyl sulfate, and hippurate that is removed by residual kidney clearance was two to three times higher than that of urea. Higher RKF, measured by 24-hour kidney creatinine and urea clearance, was also associated with lower plasma levels of the three protein-bound solutes (33).
The relatively low PD clearance of secretory solutes likely reflects their larger size and high degree of protein binding. The peritoneal capillary membrane is the principal determinant of solute transport into the peritoneal cavity. Small and water-soluble solutes diffuse through protein-restrictive pores, which account for most of the capillary exchange membrane. Larger protein-bound solutes are carried by convection through larger pores, which exist in substantially fewer numbers (34,35). Intriguingly, we observed relatively low dialytic clearances of solutes that demonstrated a low-protein binding percentage (isovalerylglycine, tiglylglycine, and xanthosine), suggesting that laboratory assessment of protein binding on the basis of equilibrium dialysis or centrifugation may not fully correspond with the free movement of these solutes across the peritoneal membrane in uremia. Additionally, the molecular weight of the evaluated secretory solutes is 1.4–2.5 times higher than that of creatinine and 2.6–4.7 times higher than that of urea, potentially contributing to their slower dialytic clearances.
We found that the composite residual kidney clearance but not peritoneal clearance of secretory solutes is associated with lower uremic and heart failure symptom burden. The pathogenesis underlying the clinical syndrome of uremia remains unclear because only a few small studies have directly examined relationships between candidate solutes and uremic symptomology. In a study of 30 patients on incident PD, the severity of gastrointestinal and neurologic symptoms was significantly correlated with blood levels of p-cresol sulfate but not with urea or creatinine (32). Among 141 patients on maintenance HD, circulating concentrations of phenylacetylglutamine and hippurate were most strongly associated with impaired executive function (36). Furthermore, we currently know very little regarding which specific solutes of the >150 candidate uremic solutes actually exhibit clinical toxicity (14). The association between kidney clearance of secretory solutes with heart failure–related symptoms may be partially explained by improved volume control because higher kidney clearances of secretory solutes were correlated with higher 24-hour urine volume and with lower ultrafiltration volume. Our findings are an important addition to the existing literature, suggesting that the accumulation of protein-bound solutes may contribute to symptom development in ESKD and motivate future larger metabolomics studies to identify clinical consequences of individual solutes.
Our study has several strengths. We used a targeted mass spectrometry assay that was specifically developed for the secretory solutes of interest using labeled internal standards and external calibration. We calculate solute clearances using 24-hour urine and dialysate samples, which are theoretically independent of characteristics that might influence their production. We also assessed for symptom burden using well validated questionnaires. However, we are not powered to examine associations with individual symptoms. Another limitation of our study is that DSI and KCCQ scores may not be specific for uremic toxicity because the symptoms elicited by these instruments may be caused by conditions other than uremia. The KCCQ also has not been validated in patients with ESKD. Furthermore, given the known diurnal variations of plasma concentrations of secretory solutes (20), estimation of kidney clearance from a single plasma level may be prone toward error. For instance, the 24-hour clearance of a solute may be falsely high if its plasma concentration obtained on a single occasion is lower than the true average plasma value over the 24-hour collection period. Therefore, further refinements of laboratory assessments of secretory solutes may strengthen the association between solute clearance and symptoms. Despite these limitations, our results suggest a potential mechanism underlying the pathogenesis of uremic symptomology and motivate future larger studies to explore associations between tubular secretory clearance with detailed symptom assessments.
In summary, we found residual kidney clearances of tubular secretory solutes to be substantially higher than their respective peritoneal clearances among patients initiating PD. Higher residual kidney secretory clearances were associated with fewer uremic and heart failure symptoms. The inclusion of tubular secretory clearances into the assessment of RKF could enhance assessment of RKF and uniquely inform uremic toxicity. Future strategies to improve the dialytic clearance of secretory solutes may improve patient-reported symptom burden in ESKD.
Disclosures
Dr. Hoofnagle reports receiving grants from Waters, Inc. during the conduct of the study. Dr. Kestenbaum reports personal fees from Reata Pharmaceuticals outside the submitted work. Dr. Becker, Dr. Chen, Dr. Goodling, Dr. Himmelfarb, Dr. Kundzins, Dr. Nguyen, Dr. Wang, and Dr. Zelnick have nothing to disclose.
Funding
This work was supported by National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases grants R01 DK107931 (to Dr. Kestenbaum) and R01 DK 103986 (to Dr. Kestenbaum) and an unrestricted gift from the Northwest Kidney Centers (to the Kidney Research Institute). Dr. Wang is supported by the American Society of Nephrology Ben J. Lipps Research Fellowship.
Supplemental Material
This article contains the following supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.11120919/-/DCSupplemental.
Supplemental Figure 1. Correlation between composite secretion scores (peritoneal versus kidney).
Supplemental Table 1. Detection limits, laboratory variability, and characteristics of candidate secretory solutes.
Supplemental Table 2. Peritoneal dialysis prescriptions for study participants.
Supplemental Table 3. Correlations of kidney and peritoneal solute clearances.
Supplemental Table 4. Solute concentrations in urine, dialysate, and plasma.
Supplemental Table 5. Dialysis Symptom Index symptom summary.
Supplemental Table 6. Correlations of kidney functions with Individual Dialysis Symptom Index symptoms.
Supplemental Table 7. Correlations of combined residual kidney and peritoneal solute clearance with symptoms.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
- Received September 15, 2019.
- Accepted February 7, 2020.
- Copyright © 2020 by the American Society of Nephrology
References
In this issue
Jump to section
More in this TOC Section
Original Articles
Cited By...
- No citing articles found.