Tubular Secretory Clearance Is Associated With Whole-Body Insulin Clearance Free

Abstract

Context

The kidneys eliminate insulin via glomerular and peritubular mechanisms; consequently, the kidney contribution to insulin clearance may be underestimated by the glomerular filtration rate (GFR) alone.

Objective

To determine associations of tubular secretory clearance with whole-body insulin clearance and sensitivity in a dedicated study of glucose and insulin metabolism.

Design, Setting, and Participants

We performed an ancillary, cross-sectional study of tubular secretion in the Study of Glucose and Insulin in Renal Disease (SUGAR). Hyperinsulinemic-euglycemic clamps were performed in 57 nondiabetic persons with chronic kidney disease and 38 persons without kidney disease.

Intervention

We measured plasma and 24-hour urine concentrations of endogenous solutes primarily eliminated by tubular secretion. Kidney clearances of secretory solutes were calculated as the amount of blood fully cleared of that solute per minute.

Main Outcome Measures

Whole-body insulin clearance, insulin sensitivity.

Results

Mean whole-body insulin clearance was 924 ± 228 mL/min. After adjustment for age, sex, Black race, fat and fat-free mass, each 20% lower estimated GFR was associated with a 13 mL/min lower insulin clearance (95% confidence interval [CI], 2-24 mL/min lower). Each 20% lower clearance of isovalerylglycine and xanthosine were associated with a 16 mL/min lower (95% CI, 5-26 mL/min lower) and 19 mL/min lower (95% CI, 7-31 mL/min lower) insulin clearance, respectively. Neither estimated GFR nor secretory solute clearances were associated with insulin sensitivity after adjustment.

Conclusions

These results highlight the importance of tubular secretory pathways to insulin elimination but suggest that kidney functions in aggregate contribute only modestly to systemic insulin clearance.

Chronic kidney disease (CKD) leads to complex and clinically important disturbances in insulin homeostasis (1-4). When measured using gold-standard hyperinsulinemic-euglycemic clamp tests, whole-body insulin clearance is lower in nondiabetic persons with CKD compared to similar persons without kidney disease (5). Patients with diabetes and kidney disease typically require lower doses of insulin and have a higher incidence of hypoglycemia (6, 7). Yet, whole-body insulin clearance, measured via hyperinsulinemic clamp, is only modestly associated with the estimated glomerular filtration rate (GFR), suggesting that nonglomerular pathways of insulin disposal may play a contributory role (5).

Insulin is secreted into the portal circulation for delivery to the liver, which closely regulates the amount of insulin released into the systemic circulation. The remaining insulin then enters the posthepatic circulation to increase substrate uptake in insulin-sensitive target tissues. The healthy kidney plays a physiologic role in systemic insulin clearance via 2 distinct mechanisms: glomerular filtration followed by apical tubular reabsorption and peritubular clearance (8). Insulin that is freely filtered at the glomerulus is subsequently reabsorbed and cleared by proximal tubular epithelial cells via receptor-mediated endocytosis not involving the insulin receptor. In parallel, nonfiltered insulin is delivered to the postglomerular circulation, where it is cleared by insulin receptors located on the basolateral surface of tubular epithelial cells (peritubular clearance) (8). Experimental studies demonstrate that approximately 40% of insulin delivered to the kidneys is extracted in a single pass with a roughly similar proportion cleared by glomerular and peritubular mechanisms (8-11). In this regard, measures of kidney function that include estimates of both GFR and peritubular clearance may improve the assessment of insulin disposition in humans.

We developed a novel method to estimate kidney tubular secretory clearance based on quantification of endogenous secretory solutes in plasma and urine. We used this procedure to estimate tubular secretory clearances in 95 nondiabetic individuals with and without CKD in a dedicated hyperinsulinemic-euglycemic clamp study. We determined associations of tubular secretory solute clearance with whole-body insulin clearance and insulin sensitivity, and we compared associations with estimated GFR, the prevailing measure of kidney function.

Methods

Data source and study participants

We performed an ancillary study in the Study of Glucose and Insulin in Renal Disease (SUGAR), a cross-sectional study of glucose and insulin homeostasis conducted in nondiabetic persons with CKD and control individuals without known kidney disease (5). From 2011 to 2014, study personnel recruited 59 patients with CKD, defined by an estimated glomerular filtration rate (eGFR) <60 mL/min per 1.73 m², and 39 control individuals with an eGFR ≥60 mL/min per 1.73 m² from nephrology and primary care clinics at the University of Washington and neighboring institutions in Seattle, Washington. Exclusion criteria included age <18 years, a clinical diagnosis of diabetes, receipt of maintenance dialysis, the presence of an arteriovenous fistula, history of kidney transplantation, use of specific medications known to reduce insulin sensitivity (including corticosteroids and immunosuppressants), a fasting serum glucose level ≥126 mg/dL, and a hemoglobin level <10 g/dL. We further excluded 2 participants from this ancillary study who did not have adequate stored plasma and urine samples for secretory solute measurements and 1 participant who had an extreme insulin clearance, yielding a final analytic sample of 95 participants.

Measurements of insulin clearance and insulin sensitivity

The SUGAR measured insulin clearance and insulin sensitivity using the hyperinsulinemic-euglycemic clamp technique adapted from the method of DeFronzo and colleagues (12-14). As described previously (5), participants were admitted to the University of Washington Clinical Research Center after an overnight fast. Participants received an infusion of unlabeled 20% dextrose (11.4 g/m² over 60 seconds). Thirty minutes after commencing the dextrose infusion, an insulin infusion was initiated as a prime (160 mU/m² per minute for 5 minutes) followed by a constant rate (80 mU/m² per minute) based on an initial dosing study demonstrating suppression of endogenous glucose production and a linear dose-response relationship with submaximal glucose uptake (5). A variable infusion of unlabeled 20% dextrose was administered to maintain blood glucose (measured every 5 minutes) at approximately 90 mg/dL. Beginning 120 to 150 minutes after initiation of the insulin infusion, the dextrose infusion rate was held constant for 30 minutes, over which time 3 steady-state plasma measurements of insulin were obtained 15 minutes apart. Plasma insulin measurements were averaged to generate the steady-state insulin concentration, and whole-body insulin clearance was then calculated as the insulin infusion rate divided by the steady-state insulin concentration. The glucose disposal rate was calculated as the glucose infusion rate during the last 30 minutes of the clamp, adjusted for the drift in plasma glucose concentration. Insulin sensitivity was calculated as (glucose disposal rate * concentration of infused glucose) / (insulin concentration at steady-state − fasting insulin concentration). Plasma concentrations of insulin (two-site immune enzymometric assay; Tosoh 2000 Autoanalyzer), glucose (glucose hexokinase method; Roche Module P Chemistry Autoanalyzer; Roche, Basel, Switzerland), and dextrose concentration of the infusate were measured at the Northwest Lipid Research Laboratories (Seattle, WA).

Measurements of tubular secretory clearance

We calculated 24-hour kidney clearances of 7 endogenous solutes that are primarily eliminated via tubular secretion: cinnamoylglycine, indoxyl sulfate, isovalerylglycine, kynurenic acid, pyridoxic acid, tiglylglycine, and xanthosine. We previously selected these solutes from the published literature based on one or more of the following characteristics: affinity for basolateral organic anion transporters in the proximal tubules, a high reported degree of protein binding suggesting minimal glomerular filtration, and/ or kidney clearances that exceed that of creatinine or GFR (15-19).

In SUGAR, participants returned approximately 1 week after the hyperinsulinemic-euglycemic clamp (median 7 days; interquartile range, 7-8 days) for further testing; they completed a 24-hour urine collection on the day prior to the visit and provided plasma samples at the start of the visit. Secretory solutes were quantified in plasma using protein precipitation, solid phase extraction, and targeted liquid chromatography-tandem mass spectrometry (LC-MS/MS); and in urine using solid phase extraction and LC-MS/MS (19). Each plasma and urine sample were normalized to labeled internal standards of purified compounds added to each well. Peak area ratios were then standardized to single point calibrators (mean of 5 replicates on each study plate) to account for potential drift. We have previously determined accurate concentrations of each solute in the external calibrators (pooled human serum and urine) via standard addition of purified compounds analyzed by quantitative nuclear magnetic resonance (NMR). Laboratory coefficients of variation for the secretory solutes ranged from 3.4% to 14.5% in plasma and from 4.5% to 10.1% in urine (Supplemental Table 1) (20). We calculated the 24-hour kidney clearance of each secretory solute, in mL/min, as:

where U_x and P_x represent the urine and plasma concentrations of solute X in ng/mL, respectively, and V represents the 24-hour urine volume in mL per minute.

Covariates

We measured serum concentrations of creatinine and cystatin C at the clamp visit using a Beckman DxC Automated Chemistry Analyzer (Beckman Coulter, Inc., Brea CA) with values traceable to isotope dilution mass spectrometry and ERM-DA471/IFCC, respectively. We calculated eGFR using the combined 2012 Chronic Kidney Disease Epidemiology Collaboration formula based on serum creatinine and cystatin C concentrations (21). For the purposes of these experiments, we expressed eGFR in mL/min, rather than standardized to body surface area, to correspond with the secretory solute clearance measurements and the direct measurements of insulin clearance obtained from the clamp. Urine creatinine concentrations were measured at the Kidney Research Institute on the Beckman DxC 600 automated platform.

We adjusted measurements of body composition determined by dual-energy x-ray absorptiometry (DXA) (GE Lunar or Prodigy and iDXA; EnCore Software versions 12.3 and 14.1; GE Healthcare, Waukesha, WI). We measured urine albumin concentrations using a turbidimetric method on a Beckman DxC Automated Chemistry Analyzer (Beckman Coulter, Inc.). SUGAR participants self-reported their demographics and medical history. We quantified physical activity levels using accelerometry, expressed as counts per minute (22).

Statistical analyses

Continuous variables were described as mean ± standard deviation and were stratified by sex-specific tertiles of whole-body insulin clearance due to differences between men and women. Scatterplots were used to describe bivariate distributions of insulin clearance with eGFR and secretory solute clearances. Linear regression with robust Huber-White standard errors was used to estimate associations of eGFR and individual secretory solute clearances with whole-body insulin clearance. To provide a singular metric for summarizing the secretory clearances, we created a summary measure by converting each clearance to a common 0 to 100 scale:

where clearance_X represents the kidney clearance of solute X, min(clearance_X) represents the minimum clearance value in the distribution, and range(secretory clearance_X) represents the difference between the maximum and minimum values. We then calculated the summary secretion score as the average of the 7 standardized clearances.

Potential confounders were controlled for in a series of nested models. Model 1 adjusted for age, sex, self-identified Black race, fat mass, and fat-free mass. Model 2 added adjustment for log-transformed urinary albumin excretion and level of physical activity (accelerometry counts per minute). A Bonferroni correction for multiple comparisons was used to control the Type I error rate; for these analyses, a 2-sided P value of < 0.05/8 = 0.0063 was taken as evidence of statistical significance for the individual secretory clearances and the summary score and a value of 0.05 taken as evidence of significance for eGFR. Analyses were repeated with 24-hour urinary creatinine clearances and after standardizing these measures of kidney function and whole-body insulin clearance to body surface area. A similar analytic plan was used to evaluate associations of eGFR and individual secretory clearances with insulin sensitivity. We investigated the incremental value of each variable in the percentage of variability in insulin clearance explained, estimated using an analysis of variance (ANOVA) approach together with 500 bootstrap replicates. A small percentage (5%) of study participants were missing data on fat mass and fat-free mass; for regression analyses, these values were multiply imputed using chained equations, with the resulting estimates combined using Rubin’s rules (23). All analyses were conducted using the R computing environment 3.4.3 (R Core Team 2017, Vienna, Austria).

Results

Description of participant characteristics

Among the 95 participants in this ancillary study, the mean age was 63 ± 13 years, 47% were female, and the mean eGFR was 65.9 ± 32.7 mL/min (Table 1). The mean whole-body insulin clearances among participants with and without CKD were 871 ± 224 and 1003 ± 213 mL/min, respectively. Participants who had lower insulin clearances tended to have lower fat mass and fat-free mass, higher urinary albumin excretion, and were more likely to use antihypertensive medications compared with participants who had higher insulin clearances. Among participants in the lowest, intermediate, and highest sex-specific tertiles of insulin clearance, mean eGFR values were 54 ± 29, 67 ± 34, and 76 ± 32 mL/min, respectively.

Unadjusted associations of kidney functions with insulin clearance

Lower eGFRs were correlated with lower whole-body insulin clearance (Fig. 1 panel a; ρ = 0.34). Lower estimated secretory solute clearances were also associated with lower insulin clearance (Fig. 1; panels b-h; ρ ranged from 0.26 for kynurenic acid clearance to ρ = 0.45 for xanthosine clearance). In unadjusted analyses, lower eGFR and lower kidney clearances of indoxyl sulfate, isovalerylglycine, tiglylglycine, and xanthosine were associated with lower insulin clearance after correction for multiple comparisons (Table 2). Each 20% lower summary secretion score, which represents the average of the individual secretory clearances, was associated with a 36 mL/min lower insulin clearance (95% CI, 18-54 mL/min lower). When expressed as the variance in whole-body insulin clearance explained by each kidney measure, xanthosine clearance was associated with 18% of the variance (95% CI, 7%-31%); tiglylglycine clearance 15% (95% CI, 4%-30%); isovalerylglycine clearance 12% (95% CI, 2%-26%); and indoxyl sulfate clearance 9% (95% CI, 1%-23%). In contrast, eGFR explained 11% of the observed variation in insulin clearance (95% CI, 3%-24%).

Adjusted associations of kidney functions with insulin clearance

After adjustment for age, sex, Black race, fat mass, and fat-free mass (model 1), lower eGFR and lower kidney clearances of isovalerylglycine, tiglylglycine, and xanthosine were associated with significantly lower whole-body insulin clearance (Table 3). These estimates were not materially altered by further adjustment for urinary albumin excretion and physical activity levels (model 2). After adjustment, each 20% lower xanthosine clearance was associated with an estimated 19 mL/min lower insulin clearance (95% CI, 7-31 mL/min lower), and each 20% lower isovalerylglycine clearance was associated with an estimated 16 mL/min lower insulin clearance (95% CI, 5-26 mL/min lower).

Cumulative variability in insulin clearance

The combination of eGFR and the summary secretion score collectively explained 16% of the variation in whole-body insulin clearance (Table 4). Adding measurements of body composition (fat-free mass and fat mass) increased the total cumulative proportion of variability explained to 35% (95% CI, 17%-52%). The sequential addition of demographic characteristics (age, sex, and Black race), physical activity levels, and urinary albumin excretion further increased the proportion of variability in whole-body insulin clearance explained to 48% (95% CI, 32%-64%).

Adjusted associations of kidney functions with insulin sensitivity

The mean insulin sensitivity values among participants with and without CKD were 3.8 ± 1.6 and 5.1 ± 2.0 (mg/min)/(μU/mL), respectively. After adjustment for demographics and body composition, lower kidney clearances of isovalerylglycine and xanthosine were associated with lower insulin sensitivity at the P < 0.05 level, but not after correction for multiple comparisons (Table 5).

Discussion

We found that lower eGFR and lower kidney clearances of the tubular secretory solutes isovalerylglycine and xanthosine were associated with lower whole-body insulin clearance measured by the hyperinsulinemic-euglycemic clamp technique. The combination of eGFR and the average of the individual tubular secretory clearances explained approximately 16% of the variation in whole-body insulin clearance. We did not observe associations of eGFR or secretory solute clearances with insulin sensitivity, assessed by hyperinsulinemic-euglycemic clamp testing. These findings, obtained in a contemporary cohort of nondiabetic persons with and without CKD using direct measurements of insulin clearance, demonstrate that glomerular and peritubular pathways contribute to kidney insulin elimination in humans but suggest that kidney functions in aggregate contribute only modestly to systemic insulin clearance.

Early experiments employing human arterial and renal vein sampling found that glomerular filtration (and subsequent apical tubular reabsorption) accounted for approximately 60% of the total kidney insulin clearance, with the remaining 40% attributable to extraction from the basolateral surface of tubular epithelial cells (9, 10). A similar study of isolated, perfused rat kidney tissue further demonstrated that basolateral peritubular insulin uptake increased as glomerular filtration fell to maintain a constant rate of kidney insulin clearance (24).

Our results suggest that the kidney clearance of secretory solutes could serve as a surrogate marker of basolateral insulin elimination. These solutes—many of which are highly protein-bound and have kidney clearances that exceed eGFR—traverse the postglomerular circulation and are delivered to the basolateral surface of the proximal tubules, where they are transported into cells via specific organic anion transporters (15, 25, 26). These solutes are then actively secreted into the urine via energy dependent transporters. Although insulin is not secreted into the urine, its elimination via basolateral tubular pathways suggest potential correlation with peritubular secretory solute clearance.

We similarly observed that the combination of eGFR and the average of individual tubular secretory solute clearances explained a relatively small proportion of the variation in whole-body insulin clearance. The liver is the primary organ responsible for both regulating peripheral insulin concentration and for degrading the hormone, and it has been estimated that the liver removes up to 65% of all endogenous insulin secreted from the pancreas (27-29). Notwithstanding, the kidney is widely believed to play an important role in posthepatic insulin clearance, degrading 25% of all endogenous insulin in healthy individuals and approximately 33% of exogenously administered insulin in experimental hyperinsulinemic clamp conditions (30, 31). Our analytic approach is not the ideal method for directly quantifying kidney insulin clearance and provides only an estimate that must be interpreted cautiously. Nevertheless, our data suggest that other peripheral tissues, such as skeletal muscle and adipose tissue, may play a larger role in posthepatic insulin disposition than previously thought (32).

We did not find significant associations between tubular secretory clearance and insulin sensitivity in the current study. Indeed, insulin sensitivity is reduced in CKD, but studies examining the intrinsic contribution of decreased GFR to this outcome have yielded conflicting results (4, 5, 33, 34). As kidney tubular function also declines in CKD, it is intriguing to speculate that a loss of tubular secretory solute clearance could contribute to reduced insulin sensitivity in this population. In animal studies, p-cresol sulfate, an avidly protein-bound uremic solute, promotes insulin resistance, loss of fat mass, and the ectopic redistribution of lipids via alteration of the insulin-induced intracellular signaling cascade (35, 36). Our study may be underpowered to detect an association between secretory solute clearance and insulin sensitivity, and further investigation is warranted.

Strengths of this study include direct measurements of insulin clearance and insulin sensitivity via the hyperinsulinemic-euglycemic clamp and 24-hour urine collections to estimate secretory solute clearances. Importantly, our study did not include a gold-standard method for quantitatively assessing tubular secretory clearance or for definitively proving that a solute is cleared exclusively by tubular secretion. As such, we could only estimate tubular secretory clearance by the total kidney clearances of specific solutes that are thought to be primarily removed by secretion. Likewise, although we used novel targeted laboratory assays to quantify the secretory solutes of interest with high precision and accuracy, diurnal variation in plasma concentrations of the selected solutes over the timed urine collection period could contribute to imprecision in these estimates of secretory clearance. We similarly acknowledge that we were unable to measure GFR using exogenous filtration markers but instead estimated GFR based on serum creatinine and cystatin concentrations using the Chronic Kidney Disease Epidemiology Collaboration equation. The sample size is relatively small due to the cost and resource-intensive nature of the study procedures, and the cross-sectional study design cannot inform temporality of the observed associations between secretory solute clearances and insulin clearance.

In conclusion, we found lower kidney clearances of the secretory solutes isovalerylglycine and xanthosine to be associated with lower whole-body lower insulin clearance in a human study of glucose and insulin homeostasis. The findings imply a larger contribution of kidney function to insulin disposition when measures of secretory clearance are included.

Abbreviations

Abbreviations
 
• CI

  confidence interval

 
• CKD

  chronic kidney disease

 
• eGFR

  estimated glomerular filtration rate

 
• GFR

  glomerular filtration rate

 
• LC-MS/MS

  liquid chromatography–tandem mass spectrometry

 
• SUGAR

  The Study of Glucose and Insulin in Renal Disease

Acknowledgments

Financial Support: This study was supported by Grant R01 DK107931 (PI Kestenbaum) and by Grant R01 DK087726 (PI de Boer) from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). This work was partially supported by an unrestricted fund from the Northwest Kidney Centers.

Additional Information

Disclosure Summary: M.P.H. and S.E.K. have nothing to declare. L.R.K. receives grant support from NIDDK. K.M.U. has previously consulted for Novonordisk and received travel reimbursement for a Medtronics insulin pump training class. I.H.dB. consults with Boehringer-Ingelheim, Cyclerion Therapeutics, George Clinical, Goldfinch Bio, and Ironwood. B.R.K. has previously consulted for Reata Pharmaceuticals.

Data Availability

The datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

References