Hyperfiltration

Hyperfiltration, traditionally defined as a supraphysiologic elevation in GFR, has been implicated in kidney disease pathogenesis since 1931. Then, Jolliffe and Smith (1) demonstrated that dogs maintained on a pure beef diet had substantial increases in their creatinine clearances, consistent with hyperfiltration. In the 1970s, Parving et al. (2) investigated mechanisms underlying hyperfiltration in diabetic individuals . The following decade, Brenner et al. (3) famously described the mechanisms of hyperfiltration causing kidney injury. Chronic vasodilation in the kidney due to diabetes, protein intake, or a reduction in kidney mass was postulated to cause glomerular hyperfiltration and cellular injury. Ultimately, sustained increases in glomerular pressure led to increased filtration of individual nephrons, resulting in the initiation and progression of direct cellular injury and glomerulosclerosis.

These foundational experiments underlie our understanding of hyperfiltration in the pathogenesis of kidney disease. However, decision making surrounding hyperfiltration is more challenging in the clinical setting of managing persons with type 1 diabetes. Traditional teaching defines hyperfiltration as a feared complication of early diabetic kidney disease, but it is unclear whether clinicians should attempt to identify patients who hyperfilter and treat them more aggressively. This clinical conundrum is exacerbated by the question of whether pathologic hyperfiltration can be distinguished clinically from healthy supranormal GFR or from measurement errors and random fluctuations (4). For hyperfiltration detection and management to be an integral part of clinical practice, epidemiologic evidence should link high measures of GFR with increased risk of progressive diabetic kidney disease.

In this issue of CJASN, Molitch et al. (5) address the critical question of whether high measures of GFR in a range presumed to reflect hyperfiltration indicate elevated risk for adverse kidney outcomes in the setting of type 1 diabetes. The investigators evaluated participants with type 1 diabetes from the Diabetes Control and Complications Trial (DCCT)/Epidemiology of Diabetes Interventions and Complications (EDIC) study. Baseline kidney function was assessed on study entry using ¹²⁵I-iothalamate clearance for gold standard GFR measurement (mGFR) rather than relying on GFR estimates from creatinine or cystatin C. Hyperfiltration was categorized as mGFR≥140 ml/min per 1.73 m², and alternate definitions were also explored. The primary outcome was defined as incidence of creatinine-based eGFR (eGFRcr) <60 ml/min per 1.73 m², and the secondary outcome was macroalbuminuria, an albumin excretion rate >300 mg/24 h. Given the baseline mean mGFR of 126 ml/min per 1.73 m², these represent clinically meaningful end points of diabetic kidney disease.

Of 446 participants studied in the cohort, 106 (24%) met criteria for having hyperfiltration at baseline. Over a median 28 years of follow-up, there were 53 (12%) eGFRcr<60 ml/min per 1.73 m² events. The primary finding of this study was that hyperfiltration was not associated with risk of developing eGFR<60 ml/min per 1.73 m² in the multivariable adjusted Cox proportional hazards model; in fact, the direction of the association suggested that the persons with hyperfiltration had lower risk of incident eGFRcr<60 ml/min per 1.73 m² (hazard ratio, 0.77; 95% confidence interval, 0.38 to 1.54). If high mGFR indicates greater glomerular pressure, then hyperfiltration should have been associated with higher risk of macroalbuminuria, but no such association was observed (hazard ratio,1.03; 95% confidence interval, 0.54 to 1.96). Results were consistent after several sensitivity analyses, including categorizing hyperfiltration with lower (mGFR≥130 ml/min per 1.73 m²) or higher (mGFR≥150 ml/min per 1.73 m²) thresholds.

The DCCT/EDIC study cohort offered an unprecedented opportunity to evaluate rigorously the risks associated with high GFR measures that are presumed to represent hyperfiltration. Most impressively, the study had exceptionally long follow-up for a monitored cohort, and therefore, there was adequate time for participants to develop diabetic kidney disease with the aforementioned end points. Additionally, at baseline, participants had very recently diagnosed type 1 diabetes, with an average disease duration of only 4 years; this minimized confounding, because the participants were all very early in the disease course. Gold standard mGFR allowed for an optimal categorization of hyperfiltration at baseline, because hyperfiltration definitions on the basis of eGFRcr could have been unacceptably influenced by creatinine production rather than actual GFR. For all of the above reasons, this study had a near-optimal design to evaluate the consequences of high mGFR.

The primary limitation of this study was that there was a lack of precision around their estimates of association between presumed hyperfiltration and diabetic kidney disease. Despite a large cohort size, the relatively small number of end points left the study with power to detect only moderate or large associations of high GFR with either increased or decreased CKD risk. However, after almost three decades of follow-up, if high GFR measures were truly harmful, then the point estimate for the association with incident eGFR <60 ml/min per 1.73 m² should have been in the direction of increased rather than decreased risk. Additionally, although ¹²⁵I-iothalamate clearance is the gold standard, it has measurement variability, and repeated measures might have ensured a more accurate assessment of hyperfiltration (4).

From the clinical perspective, this study is convincing that high GFR measures or estimates in individual patients are unlikely to be useful as a risk factor for diabetic kidney disease in persons with type 1 diabetes. If these elevated GFR measures were truly representative of pathologic hyperfiltration, then this study should have been able to link the high GFR measures with diabetic kidney disease. As readers, we might wish that this study could definitively prove the lack of association between high GFR measures and risk of CKD or ESKD, but achieving greater precision or more advanced CKD end points would have required a study many times as large or with decades longer follow-up. Given the challenge of recruiting this unique cohort, performing ¹²⁵I-iothalamate clearance measurements, and maintaining rigorous follow-up, this study may be the most definitive on this topic that we will ever have. Longer follow-up of the DCCT/EDIC study may be possible in the future, but it is unlikely that high GFR measures would affect CKD risk in the fourth decade if no signal had been detected over the prior 28 years. A key question going forward will be how these results apply to patients with type 2 diabetes, because they represent the majority of patients with diabetes who progress to ESKD. Patients with type 2 diabetes develop CKD via multiple mechanisms, because they have more comorbid hypertension and renovascular disease, and therefore, high GFR measures could theoretically be more likely to indicate pathologic hyperfiltration.

To reconcile the results of this study with the prior investigations of hyperfiltration, we must make the critical distinction between “whole-kidney” and “single-nephron” hyperfiltration (6). The results from Molitch et al. (5) do not negate the extensive investigations at the nephron level. Classically, in remnant kidney models, a reduction in kidney mass leads to adaptive glomerular hyperfiltration, causing increased glomerular capillary pressure, glomerular injury, nephron loss, and CKD. Because we lack clinical methods to measure single-nephron GFR in humans, clinical teaching has equated high observed GFR measures or estimates with hyperfiltration at the level of the individual nephron. The findings of Molitch et al. (5) can be interpreted as a rejection of the link between “higher than expected GFR” and pathologic hyperfiltration without challenging the pathophysiologic importance of glomerular hypertension. Although the participants in this study with mGFR≥140 ml/min per 1.73 m² were at the higher range of the cohort, their glomerular intracapillary pressures and single-nephron GFR levels were unknown.

The importance of intraglomerular pressure as a contributor to progressive diabetic kidney disease has been validated by the success of therapies that modulate intraglomerular pressure, including sodium-glucose cotransporter-2 inhibitors and renin-angiotensin-aldosterone system inhibitors, on reducing the risk of progressive diabetic kidney disease (7). To evaluate the potential risks of high intraglomerular pressure or monitor the effectiveness of these therapies, future technologies may be required that quantify filtration rate at the individual nephron level or discern functional nephron density (8,9).

Until we have these advanced technologies available in clinical practice, the findings of Molitch et al. (5) are definitive that the presence of elevated GFR measures or estimates in persons with type 1 diabetes should not be used for prognostication or clinical decision making. There are many other robust risk factors and disease markers that can assist clinicians in tracking disease progression. High GFR measurements or estimates might simply represent above-average filtration capacity. By removing concern and vigilance about the potential negative effects of high GFR measures, this study may aide clinicians and patients by avoiding unnecessary worry and confusion. Conversely, normal GFR estimates in persons with early diabetes should not reassure clinicians that there is no underlying pathology. The evolving literature on hyperfiltration began in the physiology laboratory and eventually, disseminated to become part of clinical practice. After this study by Molitch et al. (5), the conclusion can be made that current clinical measurements of hyperfiltration are inadequate for use in clinical practice, and high GFR estimates are not a clinically important risk factor in patients with type 1 diabetes of long-term risk of kidney disease.

Disclosures

Dr. Shlipak is a Scientific Advisor for TAI Diagnostics. Dr. Tummalapalli has nothing to disclose.