editorial Steven J. Rosansky 1 and Richard J. Glassock 2 1 Dorn Research Institute, WJBD

Size: px

Start display at page:

Download "editorial Steven J. Rosansky 1 and Richard J. Glassock 2 1 Dorn Research Institute, WJBD"

Isaac Ford
6 years ago
Views:

1 International Society of Nephrology editorial Is a decline in estimated GFR an appropriate surrogate end point for renoprotection drug trials? Kidney International (2014) 85, doi: /ki Steven J. Rosansky 1 and Richard J. Glassock 2 1 Dorn Research Institute, WJBD VA Hospital, Columbia, South Carolina, USA and 2 Geffen School of Medicine, University of California, Los Angeles, Los Angeles, California, USA Correspondence: Steven J. Rosansky, Dorn Research Institute, WJBD VA Hospital, 526 N. Trenholm Road, Columbia, South Carolina 29206, USA. sjrcra@yahoo.com Introduction Recently, the US Food and Drug Administration (FDA) and the National Kidney Foundation jointly convened a meeting of experts to examine alternative surrogate end points for randomized clinical drug trials to slow development of renal failure (renoprotection trials). 1 A 30% decline and a 40% decline in glomerular filtration rate (GFR) from baseline values were suggested as new surrogate end points with the hope of finding new drugs or drug combinations to slow progression more efficiently than is possible using the current FDA-accepted surrogate end point of doubling of serum creatinine concentration from baseline levels (corresponding to a 50% decline in GFR), or the hard end point of death or the development of treated end-stage renal disease (ESRD). The latter hard end point is usually determined by the starting time of regular dialysis treatments or receipt of a kidney transplant (renal replacement therapy, RRT). 2 The widespread secular trend toward earlier start of RRT confounds uniform application of this end point. 3 To avoid the conundrum of variability in indications and opinions concerning when to begin RRT for ESRD, we suggest use of an absolute and confirmed estimated GFR (egfr) of 15 ml/min/1.73 m 2 or less as a replacement for the hard end point of RRT initiation. The application of all-cause mortality, composite (death or RRT) hard end points, lowered proteinuria (or albuminuria) as a new surrogate end point, and renal-disease-specific mechanisms of renal progression will not be included in this Editorial. 4 6 Instead, we will focus on the use of changes in renal function over time, referred to as renal function trajectory (RFT), as new choices for surrogate end points. We examine the use of fixed (absolute) versus fractional (percent) changes in RFT and slope-based RFT as surrogate end points. Recent reports on the high frequency of nonlinear RFT patterns, and long periods of stable or slow RFT, complicate the use of surrogate end points based on RFT. Acceptance of any new surrogate end points will rest on reproducible data showing that they accurately predict hard end points and that they are involved in the causal pathway for the patient-centered outcomes, such as death or need for RRT. 2 Clinical trial end points Renal replacement therapy as a hard end point. A drawback of using time of RRT initiation as a hard end point is that egfr levels at the time of RRT initiation have risen steadily from 1995 to Patients may initiate dialysis at an egfr that ranges from 3 to more than 20 ml/min/1.73 m 2, making the RRT end point difficult to interpret. Arrival at an min/1.73 m 2 (the level of renal function used in many international guidelines as a possible time to begin RRT) might be a preferable hard end point. 3 Alterations in endogenous creatinine generation (due to diet and/or concomitant muscle wasting) must be considered if the egfr values are determined from creatinine-based formulas (such as the Modification of Diet in Renal Disease (MDRD) and CKD-EPI equations). 7,8 Short-term variations in egfr are common in advanced kidney disease, as reversible functional alterations ensue (due to drugs, infections, volume depletion). Thus, confirmation of this hard end point by serial measures as well as a confirmatory measured GFR (mgfr) should be part of this new hard end point determination. Use of 30% and 40% relative declines in egfr as new surrogate end points. Up to 40% of patients with chronic kidney disease (CKD) may have a nonlinear negative RFT These patterns include extended periods of stable renal function, probably more common in the elderly (20 40% of the CKD population); a positive RFT (5 15% of the CKD population); and variable RFT patterns (5 15% of the CKD population) These latter variable RFT patterns may include periods of stability, fast or slow negative RFT, positive RFT, and one or more episodes of acute kidney injury superimposed on RFT patterns. 11 The common existence of nonlinear Kidney International (2014)

2 negative RFT patterns in a significant percentage of patients with CKD creates the possibility that any surrogate end point based on RFT alone may not reliably predict hard end points. Data showing correlations of these RFT patterns with outcomes are limited. Assuming a normal (age-related) RFT based on an egfr or mgfr of approximately 1 ml/ min/1.73 m 2 per year, the literature supports <3 ml/ min/1.73 m 2 per year, 3 5 ml/min/1.73 m 2 per year, and >5 ml/min/1.73 m 2 per year as slow, fast, and very fast RFT, respectively. 12 The examples cited here and in Table 1 assume a stable linear negative egfr decline during and after a hypothetical 3-year study of a new alleged renoprotective drug. These examples may not be applicable to patients with variable RFT patterns. For the same percent decline in egfr, the absolute decline per year varies inversely with starting egfr (Table 1). Thus, to make comparisons consistent, percent change per year may be preferable to absolute egfr change per year. For this hypothetical 3-year study, using baseline egfr levels of 45, 60, or 70 ml/min/1.73 m 2 (Table 1), it would require a population with a very fast RFT (5 7 ml/min/1.73 m 2 ) to reach a 30% or 40% decline from baseline egfr level, but a slower RFT than required for a 50% decline ( ml/min/1.73 m 2 per year). Nevertheless, the time to reach the new suggested hard end point of an min/1.73 m 2 or less is 5 7 years for a 30% or 40% decline versus years for the 50% decline from baseline egfr. The requirement to study patients with a very fast RFT and the long interval (which may have varying RFT patterns) required to reach the hard end point may make 3-year studies unlikely to produce meaningful results. A 30%, 40%, or even 50% change in egfr from a starting egfr of 30 ml/min/1.73 m 2 would require an absolute change per year of 3.3, 4, and 5 ml/min/1.73 m 2, respectively. The time to arrive at an min/1.73 m 2 would be 4.5, 3.8, and 3 years, respectively (Table 1). Thus, from a starting egfr of 30 ml/min/1.73 m 2, any of the three surrogate end points in a 3-year study may be acceptable if the subject has a stable linear negative RFT. Change in RFT slope as a new surrogate end point. A common phenomenon, seen in many prior studies, is the short-term (first 3 months) decline in GFR after dietary protein restriction, tight blood pressure control, or use of angiotensinconverting enzyme inhibitors and/or angiotensin receptor blockers. 2 This short-term, presumably glomerular perfusion-related decline can be uncovered with the use of RFT slope change as a surrogate end point (Figure 1). The acceleration of a negative RFT (Figure 1, comparing slope A and slope B) may be followed by a slowing of RFT relative to baseline RFT (Figure 1, comparing slope A and slope C). This novel surrogate end point requires a run-in period to assess slope A. Slopebased RFT can also uncover a short-term increase in GFR (such as that seen with amlodipine treatment) without slowing loss of renal function in the longer term or even accelerating the rate of decline in GFR. In this case slope B in Figure 1 may be positive but slope C may be steeper than slope A. 2 egfr (ml/min/1.73 m 2 ) A B ACE inhibitor/arb Low BP Low protein C Time (months) Figure 1 Variable renal function trajectory pattern, including three different rates of decline in renal function (slopes). Slope A: Pretreatment (run-in period), a rapid linear negative slope of 13.3 ml/min/1.73 m 2 per year. Slope B: First 3 months after a new study drug or one of the following putative interventions: angiotensinconverting enzyme (ACE) inhibitor therapy, angiotensin receptor blocker (ARB) therapy, initial decrease of blood pressure (BP) to target level, and initial GFR decline secondary to low dietary protein intake. This presumed hemodynamic-related decrease in egfr, if continued, would translate into an egfr loss of 40 ml/min/1.73 m 2 per year. Slope C: Approximately 3 months after study intervention in B, the linear negative RFT decreases to 6.7 ml/ min/1.73 m 2. At this rate of decline it significantly delays the time to arrival at an egfr of 20 ml/min/1.73 m 2 from 1.5 to 3 years, comparing slope A with slope C. The dotted line represents the slope of the placebo group, where slope A continues. egfr, estimated glomerular filtration rate. 724 Kidney International (2014) 85

3 Table 1 Comparison of annual absolute versus annual percentage change of egfr a Decrease of egfr by 30% from baseline Decrease of egfr by 40% from baseline Decrease of egfr by 50% from baseline min/1.73 m 2 d egfr with 40% min/1.73 m 2 d egfr with 50% min/1.73 m 2 d Baseline egfr b egfr with 30% a Three-year study using hypothetical surrogate end points of 30%, 40% and 50% decline of egfr from baseline. Based on the assumption of constancy of endogenous creatinine production and secretion during the years of observation. b Estimated GFR using the Modification of Diet in Renal Disease (MDRD) four-variable equation or the CKD-EPI equation. c Percent decrease measured from baseline egfr over 3 years of follow-up. d This assumes constant absolute change per year, linear negative slope. Caveats to consider in applying RFT to surrogate end points for renoprotection trials Issues related to RFT patterns. Pharmaceutical studies of less than 5 10 years of follow-up may not be adequate, since the natural history of renal progression in many instances includes long intervals of stable or even improving renal function that may invalidate the surrogate end points Slowing of a negative RFT during a 3-year study (Table 1) has not yet been clinically demonstrated to reliably predict an min/1.73 m 2 or RRT. A rapid negative RFT, a positive RFT, and variable RFT patterns have all been associated with worse survival in numerous observational studies How a new drug may affect these patterns must be considered. For example, the increase in egfr with short-term bardoxolone therapy might have been associated with an increase in morbidity and mortality, but a causal inference is not possible from the data available. 15 Thus, postmarketing follow-up of the effect of new drugs on renal function as well as survival is essential. Subjects to include in trials and matching of RFTrelated factors. An example of the importance of appropriate patient selection for renal progression studies is the reported failure of AST-120 (Kremezin) to slow renal progression in a CKD stage 4 population. 16 A secondary analysis of results from this study suggested potential benefit in overtly proteinuric patients who had a faster decline in renal function. 17 Generally, patients with severe albuminuria (macroalbuminuria, >300 mg per 24 hours) have a two to three times faster negative RFT, versus nonproteinuric patients. 12 Thus, one way to enrich the study population for faster RFT and renal events and thereby increase the power of the study is to have an entry criterion of macroalbuminuria. A second way is to include a run-in interval during which baseline RFT could be determined, and to accept for study participation only subjects with a fast linear negative RFT (Figure 1, slope A). The relationship between primary etiology of CKD and RFT patterns has not been well researched. Patients with polycystic kidney disease may have less RFT variability, while patients with cardiovascular disease may be more likely to have nonlinear RFT. 18 Although diabetics are assumed to have faster RFT declines, similar RFTs are seen in diabetics and nondiabetics at comparable levels of albuminuria. 19 Patients with egfr >60 ml/min/1.73 m 2 may have greater RFT variability and may need mgfr to assess renal function changes over time. 20,21 Also, surrogate end points at higher egfr levels may not relate to hard end points. Treatment and control groups must have matching factors that affect RFT. These include comparability of blood pressure control, dietary protein and salt intake, constancy of dosage of angiotensin inhibition therapy, and metabolic control (blood sugar, phosphorous, metabolic acidosis). 22,23 Issues related to measurement of renal function parameters and their change with time. The gold standard for measuring renal function (GFR) is urinary clearance of an exogenous filtration marker such as inulin or radioactive iothalamate. 24 The kinetics of plasma disappearance of filtration markers, such as iohexol or cold iothalamate (which avoids radiation exposure), can also be used. Application of gold-standard filtration markers for mgfr may give disparate results compared with egfr. 25,26 Serial mgfr assessments are probably required to assess RFT in early forms of kidney disease (egfr >60 ml/min/1.73 m 2 ). 12,20 Even in the proposed study population with egfr <30 ml/min/1.73 m 2, higher protein intake, nephrotic syndrome, and study-drug effects on tubular secretion of creatinine, muscle mass, and extrarenal elimination of creatinine must be considered. 24,27,28 Most biomarkers used for estimation of GFR have both GFR-related and non-gfr-related components. 24 Kidney International (2014)

4 Nonetheless, in serial-measure studies mgfr may not be superior to egfr, since the withinpatient variability of an mgfr based on isotopically labeled iothalamate may be higher than that of a creatinine-based egfr; but the coefficient of variation of any GFR measure may be a significant issue. 29 Time requirements, expense, and patient convenience make egfr a more attractive methodology in therapeutic intervention studies incorporating frequent measurement. Furthermore, when both measured and estimated GFR are reported, change in egfr over time often reflects change in mgfr. 30 End point results of egfr may be similar to iothalamate mgfr results in some clinical entities. 29 Nevertheless, in order to use egfr determined from plasma biomarkers (such as creatinine or cystatin C) for renal progression trials, constancy of methodology, quality control of serum creatinine determinations, and attention to the confounding issues discussed above need to be assured. Issues related to slow intrinsic renal progression rates, informative censoring (death), and competitive risk analysis. The likelihood of death before development of a surrogate or hard end point is high in elderly subjects, subjects with early CKD (egfr >60 ml/min/1.73 m 2 ), and subjects with a slow RFT (<3 ml/ min/1.73 m 2 per year). 12,22 On the other hand, lower egfr, higher albuminuria, and a more rapid rate of loss of renal function (>5 ml/ min/1.73 m 2 ) make ESRD-related end points more likely, but they are also associated with higher all-cause as well as cardiovascular mortality When there are competing events, the absolute risk and comparison of mortality versus clinical end points in placebo and treatment groups are best determined by competing risk analysis. 31 Most studies examining progression to ESRD and RRT use standard survival analysis methods that treat death or dropout before ESRD treatment as a censored event. 2 Such censoring assumes the same rates of progression in patients that died or dropped out compared with those who remain in the study. Such assumptions may not be valid. The importance of considering competitive risk of death is seen with studies involving flecainide and erythropoiesis-stimulating agents. Data from these studies show that a drug may have a beneficial effect on a surrogate end point but increase the risk of death. 2 With this in mind it may be wise for the FDA to require long-term postmarketing follow-up of patients enrolled in surrogate end point studies. Concluding remarks No one doubts the need for effective and safe interventions that slow renal disease progression. Because of marked variations of egfr at the initiation of RRT over time, the hard end point of start of RRT should probably be replaced with attainment of an min/1.73 m 2. Percent change in egfr per year may be a better unit of measure of RFT than absolute egfr change per year. New putative surrogate end points of a 30% or 40% versus a 50% decline from baseline in egfr may permit studies of shorter duration and could be considered in studies of patients with baseline egfr <30 ml/min/1.73 m 2, with a linear loss of egfr of >5 ml/min/1.73 m 2 per year confirmed by a run-in period. With this criterion, all three surrogate end points may predict attainment of an min/1.73 m 2. Use of slope-based surrogate RFT end points can define variable RFT patterns (different slopes over time) and may be the only way to examine the effect of interventions that produce a temporary (usually functional, physiological, and reversible) acceleration of a negative RFT, with a longer-term favorable slowing of RFT. The change in egfr over time in elderly patients, patients with higher baseline egfr, and non-proteinuric patients will probably not yield useful results because of large potential difference between mgfr and egfr, a slower RFT, and a higher likelihood of a nonlinear RFT. The 40% putative frequency of nonlinear RFT patterns makes all RFT-based surrogate end point studies potentially not predictive of hard end points. Thus, postmarketing follow-up studies are required to examine post-study RFT patterns and hard outcomes (mortality and RRT). Importantly, all studies that use egfr determined from endogenous plasma biomarkers (for instance, creatinine, cystatin C) must acknowledge that non-gfr-related and GFR-related factors are operative in the estimating equations that may change over time. Whenever possible, the results with egfr should be confirmed by incorporation of mgfr into the study design. DISCLOSURE The authors declared no competing interests. REFERENCES 1. Earlier endpoints proposed for chronic kidney disease trials. Nephrol News Issues 5 December Stevens LA, Greene T, Levey AS. Surrogate end points for clinical trials of kidney disease progression. Clin J Am Soc Nephrol 2006; 1: Rosansky SJ, Clark WF, Eggers P et al. Initiation of dialysis at higher GFRs: is the apparent rising tide of early dialysis harmful or helpful? Kidney Int 2009; 76: Cravedi P, Ruggenenti P, Remuzzi G. Proteinuria should be used as a surrogate in CKD. Nat Rev Nephrol 2012; 8: Kidney International (2014) 85

5 5. Remuzzi G, Benigni A, Remuzzi A. Mechanisms of progression and regression of renal lesions of chronic nephropathies and diabetes. J Clin Invest 2006; 116: Formentini I, Bobadilla M, Haefliger C et al. Current drug development challenges in chronic kidney disease (CKD) identification of individualized determinants of renal progression and premature cardiovascular disease (CVD). Nephrol Dial Transplant 2012; 27(Suppl 3): iii81 iii Beddhu S, Samore MH, Roberts MS et al. Creatinine production, nutrition, and glomerular filtration rate estimation. J Am Soc Nephrol 2003; 14: Grootendorst DC, Michels MW, Richardson JD et al. The MDRD formula does not reflect GFR in ESRD patients. Nephrol Dial Transplant 2011; 26: Li L, Astor BC, Lewis J et al. Longitudinal progression trajectory of GFR among patients with CKD. Am J Kidney Dis 2012; 59: Hu B, Gadegbeku C, Lipkowitz MS et al. Kidney function can improve in patients with hypertensive CKD. J Am Soc Nephrol 2012; 23: Perkins RM, Bacaloiu ID, Kirchner HL et al. GFR decline and mortality risk among patients with chronic kidney disease. Clin J Am Soc Nephrol 2011; 6: Chapter 2: Definition, identification, and prediction of CKD progression. Kidney Disease: Improving Global Outcomes. Kidney Int Suppl 2013; 3: Turin TC, Hemmelgarn BR. Change in kidney function over time and risk for adverse outcomes: is an increasing estimated GFR harmful? Clin J Am Soc Nephrol 2011; 6: Turin TC, Coresh J, Tonelli M et al. Change in the estimated glomerular filtration rate over time and risk of all-cause mortality. Kidney Int 2013; 83: Pergola PE, Raskin P, Toto RD et al. Bardoxolone methyl and kidney function in CKD with type 2 diabetes. N Engl J Med 2011; 365: Schulman G, Agarwal R, Acharya M et al. A multicenter, randomized, double-blind, placebo-controlled, doseranging study of AST-120 (Kremezin) in patients with moderate to severe CKD. Am J Kidney Dis 2006; 47: Schulman G, Berl T, Beck G et al. EPPIC (Evaluating Prevention of Progression In Chronic Kidney Disease): results from 2 phase III, randomized, placebo-controlled, double-blind trials of AST-120 in adults with CKD. MIU ASN Abstract_EPPIC_FINAL_091212, Kidney Week www. asn-online.org. 18. Ambrogi V, Thilly N, Boini S et al. Patterns and predictors of kidney function decline in the last year prior to dialysis. Nephron Clin Pract 2009; 111: c95 c Lorenzo V, Saracho R, Zamora J et al. Similar renal decline in diabetic and non-diabetic patients with comparable levels of albuminuria. Nephrol Dial Transplant 2010; 25: Gaspari F, Ruggenenti P, Porrini E et al. The GFR and GFR decline cannot be accurately estimated in type 2 diabetics. Kidney Int 2013; 84: Ruggenenti P, Gaspari F, Cannata A et al. Measuring and estimating GFR and treatment effect in ADPKD patients: results and implications of a longitudinal cohort study. GFR- ADPKD Study Group. PLoS One [online] 2012; 7: e Rosansky SJ. Renal function trajectory is more important than chronic kidney disease stage for managing patients with chronic kidney disease. Am J Nephrol 2012; 36: Chapter 3: Management of progression and complications of CKD. Kidney Disease: Improving Global Outcomes. Kidney Int Suppl 2013; 3: Rule AD, Glassock RJ. GFR estimating equations: getting closer to the truth? Clin J Am Soc Nephrol 2013; 8: Hsu C-Y, Propert K, Xie D et al. Measured GFR does not outperform estimated GFR in predicting CKD-related complications. J Am Soc Nephrol 2011; 22: Lemley KV, Boothroyd DB, Blouch KL et al. Modeling GFR trajectories in diabetic nephropathy. Am J Physiol Renal Physiol 2005; 289: F863 F Juraschek SP, Appel LJ, Anderson CA et al. Effect of a highprotein diet on kidney function in healthy adults: results from the OmniHeart trial. Am J Kidney Dis 2013; 61: Hofstra JM, Willems JL, Wetzels JF. Estimated glomerular filtration rate in the nephrotic syndrome. Nephrol Dial Transplant 2011; 26: Padala S, Tighiouart H, Inker LA et al. Accuracy of a GFR estimating equation over time in people with a wide range of kidney function. Am J Kidney Dis 2012; 60: Lewis J, Greene T, Appel L et al. A comparison of iothalamate-gfr and serum creatinine-based outcomes: acceleration in the rate of GFR decline in the African- American Study of Kidney Disease and Hypertension. J Am Soc Nephrol 2004; 15: Grams ME, Coresh J, Segev DL et al. Vascular disease, ESRD, and death: interpreting competing risk analyses. Clin J Am Soc Nephrol 2012; 7: Kidney International (2014)

ALLHAT RENAL DISEASE OUTCOMES IN HYPERTENSIVE PATIENTS STRATIFIED INTO 4 GROUPS BY BASELINE GLOMERULAR FILTRATION RATE (GFR)

ALLHAT RENAL DISEASE OUTCOMES IN HYPERTENSIVE PATIENTS STRATIFIED INTO 4 GROUPS BY BASELINE GLOMERULAR FILTRATION RATE (GFR) 1 RENAL DISEASE OUTCOMES IN HYPERTENSIVE PATIENTS STRATIFIED INTO 4 GROUPS BY BASELINE GLOMERULAR FILTRATION RATE (GFR) 6 / 5 / 1006-1 2 Introduction Hypertension is the second most common cause of end-stage