Abstract
Diabetic nephropathy (DN) is the leading cause of ESRD worldwide. Reduced bioavailability or uncoupling of nitric oxide in the kidney, leading to decreased intracellular levels of the nitric oxide pathway effector molecule cyclic guanosine monophosphate (cGMP), has been implicated in the progression of DN. Preclinical studies suggest that elevating the cGMP intracellular pool through inhibition of the cGMP-hydrolyzing enzyme phosphodiesterase type 5 (PDE5) might exert renoprotective effects in DN. To test this hypothesis, the novel, highly specific, and long-acting PDE5 inhibitor, PF-00489791, was assessed in a multinational, multicenter, randomized, double-blind, placebo-controlled, parallel group trial of subjects with type 2 diabetes mellitus and overt nephropathy receiving angiotensin converting enzyme inhibitor or angiotensin receptor blocker background therapy. In total, 256 subjects with an eGFR between 25 and 60 ml/min per 1.73 m2 and macroalbuminuria defined by a urinary albumin-to-creatinine ratio >300 mg/g, were randomly assigned 3:1, respectively, to receive PF-00489791 (20 mg) or placebo orally, once daily for 12 weeks. Using the predefined primary assessment of efficacy (Bayesian analysis with informative prior), we observed a significant reduction in urinary albumin-to-creatinine ratio of 15.7% (ratio 0.843; 95% credible interval 0.73 to 0.98) in response to the 12-week treatment with PF-00489791 compared with placebo. PF-00489791 was safe and generally well tolerated in this patient population. Most common adverse events were mild in severity and included headache and upper gastrointestinal events. In conclusion, the safety and efficacy profile of PDE5 inhibitor PF-00489791 supports further investigation as a novel therapy to improve renal outcomes in DN.
The global prevalence of diabetes mellitus (DM) is increasing, with more than 400 million people projected to be affected by 2030.1 Diabetic nephropathy (DN) is a potentially life-threatening complication of DM that affects approximately one-third of all diabetic individuals and is the leading cause of ESRD in the Western world.2 Risk for progression of DN to ESRD is highest in diabetic patients with macroalbuminuria.3 Although standard of care therapy with angiotensin-converting enzyme inhibitors (ACEi) or angiotensin receptor blocker (ARB) reduces albuminuria and may slow the progression to ESRD, there remains substantial residual risk of ESRD that correlates with the residual level of albuminuria.4 The need for new, safe, and effective approaches to slow progression of DN has been emphasized by the disappointing outcomes of recent, randomized controlled clinical trials to assess the effectiveness of more complete inhibition of the renin angiotensin aldosterone system, which were halted because of safety concerns. Combining ACEi or ARB with the direct renin inhibitor, aliskiren (The Aliskiren Trial in Type 2 Diabetes Using Cardiorenal Endpoints or ALTITUDE trial5), was associated with increased serious adverse events (AEs) of renal concern, hyperkalemic events, and nonfatal strokes. Combining ACEi with ARB (Nephropathy in Diabetes trial or NEPHRON-D trial6) induced acute loss of renal function and severe hyperkalemia. In addition, the recent Bardoxolone Methyl Evaluation in Patients With Chronic Kidney Disease and Type 2 Diabetes or BEACON Phase III randomized controlled trial to evaluate the renoprotective effect of the NRF2 agonist, bardoxolone methyl, was halted because of increased risk of heart failure, nonfatal myocardial infarction (MI) and nonfatal stroke.7 Also, a study with an endothelin receptor antagonist (A Study of Cardiovascular Events iN Diabetes; Avosentan ASCEND Study) had to be halted because of safety concerns, mainly fluid overload and heart failure.8
Despite these setbacks, substantial progress has been made toward identifying putative pathophysiological pathways involved in the development of DN. Reduced bioavailability or uncoupling of nitric oxide (NO) leading to decreased intracellular levels of the NO pathway effector molecule, cyclic guanosine monophosphate (cGMP), in the kidney has been proposed as a key contributor to mechanisms driving progression of DN.9–13 In particular, altered NO-dependent intrarenal hemodynamics demonstrated in experimental in vivo studies of the cortical and medullary renal vasculature in diabetes,13 and uncoupling of NO by scavenging of reactive oxygen species14 have been implicated in the pathogenesis of DN. Preclinical in vivo studies demonstrate that restoring the NO pathway by elevating the intracellular pool of cGMP, through inhibition of the cGMP-hydrolyzing enzyme phosphodiesterase type 5 (PDE5), is renoprotective, with reductions in albuminuria, inflammation, fibrosis, and improvement in creatinine clearance.15–17 Specifically, PDE5 inhibition-mediated elevation of cGMP in podocytes attenuates podocyte damage in diabetic rats, resulting in decreased albuminuria,18 likely through correction of podocyte dysmotility as suggested by in vitro data.19 The first clinical study to suggest translation of the PDE5-related preclinical findings was conducted by Grover-Páez et al. who demonstrated that administration of a PDE5 inhibitor once daily for 30 days to microalbuminuric T2DM patients significantly reduced albuminuria.20 To further assess the renoprotective effect of chronic PDE5 inhibition in DN, the present study evaluated the safety and efficacy of the novel, selective, and long-acting PDE5 inhibitor, PF-00489791, in subjects with T2DM and overt DN receiving ACEi or ARB background therapy. In this multinational, multicenter, randomized, double-blind, placebo-controlled, parallel group trial, PF-00489791, administered at 20 mg daily for 12 weeks, was found to be efficacious in significantly reducing albuminuria as well as being safe and generally well-tolerated, thereby supporting further investigation as a novel strategy to improve renal outcome in DN.
Results
Figure 1 describes the disposition of participants in the trial. Of the 955 individuals screened, 256 were eligible for randomization and received the study drug. Two (3%) and 29 (15%) subjects from the placebo and PF-00489791 treatment groups, respectively, were discontinued from the study for reasons summarized in Figure 1. A total of 226 (88%) subjects completed the study, with 62 (97%) on placebo and 164 (85%) on PF-00489791. There were no major differences in demographic and baseline characteristics observed between subjects treated with PF-00489791 or placebo (Table 1). Male and female subjects, aged 34–84 years, with body mass index ranging from 17.2 to 45.3 kg/m2 were included in the study; the majority of subjects were male (195/256; 76%). All subjects had T2DM at baseline, and median durations since first diagnosis were 12.2 and 14.1 years for placebo and treatment group, respectively.
Summary of the disposition of study participants.
Demographic and baseline characteristics (randomized subjects)
The primary analysis was Bayesian and showed a significant reduction of 15.7% in urinary albumin-to-creatinine ratio in the PF-00489791 group at week 12 compared with placebo (ratio =0.843; 95% credible interval [95% CI], 0.73 to 0.98). This analysis of UACR at week 12 revealed a 15.4% UACR reduction from baseline in the PF-00489791-treated patients compared with a 0.4% UACR increase from baseline for placebo subjects. A Bayesian statistical summary of the change from baseline in UACR at week 12 in the PF-00489791 treatment group in comparison to placebo is provided in Table 2. A descriptive summary of UACR (geometric mean) over time may be seen in Table 3.
Primary analysis: change in UACR at week 12 - posterior distributions of treatment geometric means and comparison to placebo (full analysis set)
Statistical analysis of UACR (mg/mmol) at week 12 - MMRM (full analysis set)
A predefined frequentist sensitivity analysis to the primary UACR efficacy endpoint was conducted using observed data only and from all postdose time points (mixed model repeated measures [MMRM], unstructured covariance matrix) from the full analysis set. Three extreme outliers from 3 patients at week 12 (1 placebo, 2 active) were excluded from the MMRM analysis. This was in accordance with our prespecified rule that any values >3.5 or <-3.5 standardized residual would be considered as outliers. Consistent with the results of the primary Bayesian analysis, a significant reduction in UACR of 21.7% (P=0.003) in the PF-00489791 treatment group compared with placebo was observed at week 12. However, note that the week 12 estimated treatment effect in the primary Bayesian analysis was more conservative than in the supporting MMRM analysis. The MMRM analysis also demonstrated that a statistically significant UACR reduction with PF-00489791 compared with placebo was present as early as week 3, with further separation between the PF-00489791 and placebo groups observed through week 12 (-12.4% at week 3; P=0.04 and -16.9% week 6; P=0.01) (Figure 2); however, the reduction was reversible and no longer statistically significant at 4 weeks after end of treatment (-11.8% at week 16; P=0.15). Additional MMRM analyses were performed with the inclusion of outliers, and results were similar to MMRM analyses without outliers.
Least Square Mean of Urinary albumin-to-creatinine ratio from the MMRM analysis: time-dependent reduction in UACR (21.7% at week 12) in response to treatment with PF-00489791 (▲) compared with placebo (●). Least squares means and 95% confidence intervals are presented.
Change of eGFR was a key secondary endpoint used to assess the possibility of an acute impact of PF-00489791 on renal function. The MMRM analysis showed that for all time points, there was no statistically significant difference of eGFR from baseline observed between treatment groups. Absolute eGFR values over time are presented in Figure 3. Mean eGFR data over time (baseline to follow up) are displayed in Table 4.
Plot of eGFR observed mean (±SD) over time for PF-00489791 (▲) and placebo (●).
Descriptive summary of UACR, eGFR and supine SBP (full analysis set)
There was a statistically significant decrease (P<0.001) in mean systolic and diastolic BP (5.4 and 3.7 mmHg, respectively) at week 0, 4 hours after initial treatment with PF-00489791. After this initial effect, there were no statistically significant differences observed between treatment groups for weeks 3, 6, 12, and follow-up (Figure 4). In order to exclude the possibility that UACR reductions were related to lower BP in some individuals, a causal path analysis was performed which demonstrated that >99% of the effect of PF-00489791 on UACR could be attributed to a direct effect of treatment, independent of systemic BP response. Mean values for supine systolic BP (SBP) over time (baseline to follow-up) may be seen in Table 4.
Plot of supine systolic BP observed mean (±SD) over time for PF-00489791 (▲) and Placebo (●).
Clinical laboratory results did not indicate any cause for safety concerns and there were no potential cases of drug-induced liver injury. Analysis of change from baseline in glycosylated hemoglobin (HbA1c) at week 12 revealed a statistically significant mean decrease of 0.3% in the PF-00489791 group compared with a mean increase of 0.1% in the placebo group.
The proportion of subjects experiencing at least one treatment emergent adverse event (TEAE) was balanced between the PF-00489791 and placebo groups; a total of 94 all causality TEAEs were reported in 36 (56%) placebo subjects, compared with a total of 294 reported in 105 (55%) PF-00489791-treated subjects. As indicated in Table 5, overall, the most common treatment-related AEs reported were headache (4.7% versus 7.8%), diarrhea (3.6% versus 0%) and dyspepsia (3.6% versus 1.6%) in subjects on PF-00489791 or placebo, respectively. In addition, a higher percentage of subjects taking PF-00489791 had TEAEs of nasopharyngitis (9 [4.7%] versus 1[1.6%]) and vomiting (7 [3.6%] versus 2 [3.1%]) compared with those receiving placebo (Table 5). Most of these AEs were mild in severity.
Incidence of TEAEs (all causalities and treatment related) in ≥5% of subjects overall
Fifteen subjects (8%) permanently discontinued from study medication due to AEs which were considered related to PF-00489791 by the blinded investigators; headache (moderate or severe intensity) was the most commonly reported TEAE that led to permanent discontinuation (4 subjects), although the incidence of headache was not higher in the PF-00489791 treatment group when compared with the placebo group. In the PF-00489791 treatment group, 13 serious AEs were reported, of which two were considered treatment-related by the investigator (chest discomfort and cerebrovascular accident, CVA). Two randomized subjects experienced a MI; one PF-00489791-treated subject during the follow-up period (off drug for 5 weeks) and one placebo subject during week 7. Two subjects in the PF-00489791 treatment group experienced a CVA; one subject on day 55 (deemed related to study drug) and another subject on day 87, 3 days into the follow-up phase (deemed nonrelated to study drug). Three subjects died. One subject in the placebo group died because of hypotension during the study; two subjects receiving PF-00489791 died 8 months after study completion, one because of cardiorespiratory arrest (deemed nonrelated to study drug), the other because of MI (deemed nonrelated to study drug).
Discussion
The present study demonstrates that 12 weeks of daily oral treatment with 20 mg of the selective PDE5 inhibitor, PF-00489791, reduces UACR by 16% (95% CI, 2% to 27%) compared with placebo in diabetic subjects with stage 3 or 4 CKD and overt albuminuria, with values returning to baseline 4 weeks after treatment cessation. To be eligible for inclusion in the present study, subjects had to have been treated for at least 3 months with stable doses of current standard of care (ACEi or ARB) plus antihypertensives and antidiabetes medications and despite this treatment, demonstrate persistent macroalbuminuria (mean UACR at baseline 1609–1724 mg/g) as well as adequate BP control. Therefore, the reduction seen in this treatment-resistant albuminuria is particularly interesting, and may not have reached a nadir at week 12. Additional studies are warranted to evaluate the effect on albuminuria of longer treatment duration and additional treatment doses in this patient population. In addition, it should be noted that unlike observations in subjects treated with ACEi or ARB, there was no short-term fall in eGFR resulting from PF-00489791 treatment and there were no observed differences in the slope of eGFR in subjects treated with PF-00489791 versus placebo. Studies of longer duration are needed to evaluate the impact of PF-00489791 on slowing the rate of decline in eGFR. The observed reduction in albuminuria in the present study extends those made in a pilot study of 40 male subjects with T2DM conducted with the PDE5 inhibitor sildenafil.20 In that study, once daily treatment with sildenafil (50 mg orally) for 30 days resulted in reductions in albuminuria of approximately 50%, whereas subjects treated with placebo demonstrated a small increase in albuminuria over the same time period. Although the large reductions in albuminuria observed in this pilot study were impressive, it should be noted that these subjects were only microalbuminuric and were not receiving standard of care medication with ACEi or ARBs during the study. In addition, the study did not clearly demonstrate that the reduction in albuminuria upon treatment with sildenafil was unrelated to changes in systemic BP.
In contrast, in the current study, systolic and diastolic BPs were determined at each study visit through the 12 weeks of randomized treatment. Data indicated that there were no significant changes in either systolic or diastolic BP over time in subjects treated with PF-00489791 versus placebo, with the exception of a mean acute decrease in BP after the first dose of PF-00489791. As previously noted, there were no differences in eGFR between the PF-00489791 and placebo groups throughout the study duration. Taken together, these observations suggest that the albuminuria-lowering effect of PF-00489791 was not primarily driven by systemic hemodynamic changes. However, given the rapid effect on albuminuria, one must consider that changes in intraglomerular hemodynamic parameters may be a major contributor to the beneficial effect of PF-00489791 on albuminuria, as demonstrated by the study of Tapia et al.21 in which PDE5 inhibition prevented single-nephron hyperfiltration and intraglomerular hypertension without reducing whole-kidney GFR in rats with renal ablation. In addition, the study of Pflueger et al.13 conducted in streptozotocin-induced diabetic rats clearly demonstrated reduced responsiveness of cortical and vasa recta capillary blood flow to modulation of NO generation compared with nondiabetic rats, possibly through diminished cGMP-coupled intrarenal vasodilation. This further suggests that PDE5 inhibition-mediated elevation of cGMP in diabetes may result in renoprotection through favorable modulation of intraglomerular hemodynamics.
Consistent with data reported in the pilot study conducted by Grover-Páez et al.20 with sildenafil, a lowering of HbA1c upon treatment with PF-00489791 was also observed in the present study. Although the precise mechanism of HbA1c reduction upon inhibition of PDE5 is unclear, preclinical and clinical studies suggest that PDE5 inhibition may favorably impact glucose metabolism in T2DM by mediating enhanced glucose uptake in skeletal muscle22,23 improving β-cell function and decreasing insulin resistance.24–26 Such mechanisms may eventually contribute to the improvement in microvascular disease and subsequent reduction of albuminuria in studies of longer treatment duration with PF-0048971. In addition, a large set of preclinical studies conducted in animal models of diabetic and nondiabetic kidney disease have demonstrated the beneficial effect of PDE5 inhibition on pathophysiological pathways implicated in albuminuria, glomerulosclerosis, and tubulointerstitial injury. The data most relevant to the present study population come from the streptozotocin-induced diabetic rat,15,18 the Long-Evans noninsulin dependent diabetic rat,16 and the alloxan-induced diabetic rabbit,17 models of DN in which chronic inhibition of PDE5 reduced podocyte damage contributing to lowered albuminuria and reduced inflammation, oxidative stress, and fibrosis in the kidney in turn accompanied by improved renal structure and function. These effects have been suggested to be primarily mediated by cGMP-mediated protein kinase-G activation, secondary to phosphodiesterase type 5 inhibitor (PDE5i)-induced elevation of the NO-derived cGMP pool.19,27,28
PF-00489791 was safe and generally well-tolerated when administered at a dose of 20 mg daily for 12 weeks in T2DM subjects with overt DN. The most commonly observed AEs in the present study were consistent with those reported in subjects with mild-to-moderate hypertension treated with PF-00489791 for 28 days,29 including headaches and upper gastrointestinal events. There was an imbalance in discontinuations because of (primarily) mild or moderate TEAEs which may reflect a mild intolerance to this medication, in line with other PDE5is. However, the overall completion rate in the PF-00489791 group was 85%.
The well-characterized increased risk for cardiovascular events in subjects with diabetes and reduced kidney function was apparent in this study, with two serious AEs of MI and two of CVA reported during the randomized treatment period or follow-up phase. The current study was not powered to detect any potential benefit of PF-00489791 on cardiovascular events but evidence for a cardioprotective effect of PDE5 inhibition in the setting of small clinical studies30 offers the exciting prospect that PF-00489719 may exert concomitant renal and cardiovascular benefits in subjects with overt DN.
In the current study, a major limitation was the lack of monitoring sodium intake or sodium excretion. In a recent publication31 data from a meta-analysis of controlled trials measuring the differences in urinary albumin excretion or UACRs demonstrated that sodium intake reduction meaningfully reduces albumin excretion. In future studies, efforts to fully understand the effect of PDE5i in lowering albuminuria in patients with DN would require strict control of sodium balance.
In conclusion, PF-00489791 is the first long-acting PDE5i to demonstrate significant reduction of albuminuria when added to standard of care therapy in subjects with T2DM and overt DN. This proof of pharmacological activity together with the encouraging safety profile associated with PF-00489791 treatment in this fragile patient population and its potential to positively impact on key pathophysiological processes at play in diabetic kidney disease provide the basis for further investigation of PF-00489791 effects on clinical outcomes in overt nephropathy.
Concise Methods
Study Design
This randomized, double-blind, placebo-controlled, parallel-group study was conducted in 125 centers across 15 countries. Of 955 subjects screened, 256 were randomly assigned in a 3:1 ratio to receive either two 10 mg PF-00489791 tablets or two matching placebo tablets to be taken orally once daily. Masked treatments were assigned according to computerized randomization list; subjects were to take the masked therapy for 12 weeks with 4 weeks of safety follow-up.
Outcomes
The primary endpoint of the study was UACR at week 12, with secondary endpoints including UACR at weeks 3 and 6 and follow-up. UACR was calculated as the geometric mean of UACR from three first-morning void specimens collected prior to each study visit. Secondary endpoints included urine protein/creatinine ratio, systolic, diastolic and mean BP, serum creatinine and eGFR at weeks 3, 6, 12 and follow-up.
Eligibility Criteria and Recruitment
At screening, eligible participants had T2DM with CKD, based on an eGFRMDRD of 25–59 ml/min per 1.73 m2, with a history (≥3 months) of persistent, overt macroalbuminuria, and were males or females of nonchildbearing potential aged >18 years. Because of changes between screening and baseline, some subjects in the study had eGFR outside of the range used for eligibility. Macroalbuminuria was defined as a mean UACR≥300 mg/g (≥33.9 mg/mmol) determined from three consecutive first-morning void urine samples and with a UACR≥300 mg/g in at least two of three of these samples. Subjects were to have a stable background therapy of an ACEi or an ARB for ≥3 months before screening that was to be maintained for the duration of the study. The resting BP was to be ≤150/100 mmHg. Subjects were asked to sign and date an institutional review board or ethics committee approved informed consent. Subjects were ineligible for the study if the CKD resulted from type 1 DM or nondiabetic CKD, or if subjects had poorly controlled DM defined as HbA1c>9%. Prohibited medications included combination ACEi and ARB therapy, renin inhibitor therapy or aliskiren-containing combination therapy, aldosterone antagonist therapy, other PDE5i therapy, nitrates or NO donors, α-adrenoceptor blockers, or moderate to strong inducers or inhibitors of the cytochrome P4503A4.
Procedures
The eGFR was calculated using the four-variable formula developed by the modification of diet in renal disease study group.
AEs reporting included the severity of the events, and the investigator’s opinion of the relationship (related or nonrelated) to the study treatment. TEAEs were those AEs that occurred after randomization, once subjects began taking the study drug. Investigators determined study drug relatedness (causality) while remaining blinded to the treatment group by evaluating whether a reasonable possibility existed that the study drug caused or contributed to an AE; evidence in support of this assessment was provided if available. Laboratory assessments included: hematology, nonfasting chemistry, including liver and pancreatic enzymes as well as HbA1c and urinary chemistry.
Triplicate measurements of supine and standing BP and pulse rate, and assessment of orthostatic hypotension were conducted at all visits. For orthostatic BP measurements, after a minimum of 5 minutes in a supine position, the subjects were asked to stand upright. After 2 minutes in the standing position, BP and pulse rate were measured immediately. Orthostatic hypotension was defined as a decrease of ≥20 mmHg for systolic BP or ≥10 mmHg for diastolic BP, after 2 minutes of standing from a supine position and could have been symptomatic or asymptomatic and was assessed at visit 1 (baseline), predose at visit 2, and 4 hours (±15 minutes) postdose at visit 2 (week 0). Subjects were assessed for signs of edema and fluid overload at all visits.
Statistical Analyses
A two-part decision criteria for efficacy and futility was used at the end of this study which is described in the Supplemental Material.
The primary analysis was a Bayesian, outlier robust, analysis of covariance model of the UACR full analysis set data with the ratio change from baseline at week 12 as the response variable (log-scale) and treatment as a fixed effect, log baseline UACR and supine systolic BP included as covariates. Only the placebo geometric mean response was given an informative prior, as all other model parameters had effectively flat prior distributions. The placebo geometric mean response was given a log-normal prior distribution with a geometric mean of −4% and 80%; credible intervals, −13% to +7%. This prior was derived from the more than 600 placebo subjects in the Aliskiren in the Evaluation of Proteinuria in Diabetes or AVOID32 and Avosentan ASCEND8 studies, discounted to reflect potential differences between those and the study reported here. The suitability of the informative prior was assessed using a prior-predictive check that found no evidence of discordance between the data and the prior. Since this prior was forecast to be equivalent to approximately 100 placebo subjects, the targeted 200 completers were randomized on a 3:1 allocation ratio of 150 PF-00489791 subjects to 50 placebo subjects. The primary analysis was repeated for the per protocol population, using only those subjects who completed the study with no major protocol deviations. An interim analysis for futility was conducted when approximately 25% of subjects had completed the 12 week randomized treatment period. More details of the statistical design and analysis including the outlier robust approach can be found in Walley et al.33
In addition, secondary endpoints including urine protein/creatinine ratio, eGFR, and systolic and diastolic BP, were analyzed using a similar analysis of covariance as for the primary endpoint but with uninformative priors for all parameters.
The sensitivity analysis was MMRM, excluding extreme outliers, analysis of covariance model of the UACR full analysis set data with the UACR at all postdose time points as the response variable (log-scale) with treatment, time, the treatment-by-time interaction and the treatment-by-baseline interaction included as fixed effects, log baseline UACR and supine systolic BP included as covariates (Table 3). Thus, the MMRM mirrored the primary analysis except that all postdose time points were used and the analysis was frequentist, hence prior distributions were not incorporated. The results are presented in terms of adjusted geometric means.
The effect of BP change on the primary endpoint, UACR at week 12, was analyzed using a causal path analysis approach.34 That is, a system of linear equations were fitted to the data to estimate the direct treatment effect on UACR, and also the direct effect of treatment on BP and the effect of changes in BP on UACR. The two endpoints were ratio change in UACR and ratio change in systolic BP. Treatment was included as a fixed effect. Baseline log UACR and baseline log systolic BP were included as covariates in the linear model.
Disclosures
W.S., J.G., V.C., N.T., V.L., C.P.-H. and T.R., are employees of Pfizer Inc. and hold Pfizer stock. R.W. is a current employee of UCB Pharma and holds UCB Pharma stock. S.D., F.G.-P. and M.E.N. have no conflicts of interest to disclose.
Acknowledgments
The participants and clinical site staff are acknowledged. The study was funded by Pfizer Inc. Pfizer authors contributed to the design, conduct of the trial and collection, analysis and interpretation of the data. All authors contributed to the collection and analysis of the data and the preparation of the manuscript.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
This article contains supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2015050473/-/DCSupplemental.
- Copyright © 2016 by the American Society of Nephrology
References
In this issue
