Is Treatment of Nephropathy in Type 1 Diabetes Efficacious but Ineffective?

Increased urinary protein excretion in patients with diabetes has long been known to predict increased mortality, and its absence is associated with near-normal life expectancy.^1,2 Recent studies confirmed and extended these findings by illustrating the progressive increase in mortality by degree of albuminuria.^3,4 The excess mortality is due primarily to ESRD⁵ and to cardiovascular disease,⁶ which share many risk factors. Differential expression of these risk factors within an individual may play an important role in determining which disease ultimately prevails.

Abundant observational data indicate that hyperglycemia and hypertension strongly predict the appearance and progression of albuminuria, and hyperlipidemia strongly predicts cardiovascular disease in those with elevated albuminuria, although its relationship with kidney disease is less clear. Together, these modifiable risk factors provided important therapeutic targets for a number of highly influential clinical trials that changed the management of diabetic kidney disease in recent years because of their demonstrated efficacy in slowing development or progression of kidney disease and in reducing the risk for cardiovascular disease in patients with diabetes and macroalbuminuria. Given the efficacy of these trials, their rapid incorporation into national guidelines, and generally widespread use by many healthcare providers, one might expect to see increased effectiveness of these therapeutic regimens in patients with type 1 diabetes and macroalbuminuria.

In this issue of JASN, two articles examine this expectation by evaluating adverse outcomes, namely pre-ESRD death and progression to ESRD in patients with type 1 diabetes and chronic kidney disease defined by macroalbuminuria. Forsblom et al.⁷ used an extension of the Cox proportional hazards analysis, a Fine and Gray analysis, to examine predictors for each of these outcomes in the FinnDiane study population while adjusting for the competing risk for the alternate outcome. By using this approach, they sought to avoid the confounding effect of the alternate outcome, which may be considerable when the incidence of the competing risk is high.

Progression to ESRD was by far the more common of the two outcomes, and the investigators found that of the modifiable risk factors, poor glycemic control increased the cumulative incidence of ESRD but was not associated with pre-ESRD mortality. Conversely, poor lipid control increased the cumulative incidence of both pre-ESRD death and progression to ESRD, suggesting that appropriate therapeutic targets for preventing or delaying each outcome may differ. Of note, despite aggressive treatment, 45% of the 592 patients who had type 1 diabetes and participated in this study progressed to ESRD or died before onset of ESRD during a median follow-up of just under 10 years. The authors expressed hope that the better treatment options and stricter therapeutic targets in recent years may have reduced the frequency of both outcomes during the study period, but they did not provide a secular trend analysis over the 10-year study period to confirm this hypothesis.

In a separate report in this issue of JASN, Rosolowsky et al.⁸ did find a secular trend in a similar cohort of patients with type 1 diabetes from the Joslin Clinic. They found a rate of ESRD that was virtually identical to that of the FinnDiane cohort and likewise observed that progression to ESRD was the predominant outcome among their patients with type 1 diabetes and macroalbuminuria. In their trend analysis, they found no significant decline in the rate of progression to ESRD over the past 20 years despite widespread adoption of renoprotective drugs during the same period accompanied by significant improvements in BP and total serum cholesterol concentration.

Although these results are disappointing, they are not universal. The Joslin Clinic investigators noted that investigators at the Steno Memorial Hospital in Denmark reported a diminishing risk for ESRD over time in a similar group of patients with type 1 diabetes.^9,10 The factors responsible for this disparity are uncertain, although the pre-ESRD mortality in the Steno cohort was much higher than in the other two cohorts.

National data from the US Renal Data System, which include patients with either type 1 or type 2 diabetes regardless of the level of urinary albumin excretion, also suggested wide variations in secular trends of diabetic ESRD by age and ethnicity.¹¹ In white patients aged 30 to 39 years, for example, the gender-adjusted incidence rate of ESRD declined by 7% between 2000 and 2008, whereas it increased by 2% in those aged 50 to 59 years. In black patients, by contrast, the incidence rate of ESRD, which is approximately fourfold higher than in white patients, increased by 65% in those aged 30 to 39 years and declined by 13% in those aged 50 to 59 years. Perhaps most troubling in the US Renal Data System report is the finding that incident ESRD as a result of diabetes has generally increased among younger minority patients, as illustrated by the remarkable increase among the younger black patients noted here, whereas it has been generally stable or declining in older populations and among white patients.¹¹ A shift toward a younger age at onset of type 2 diabetes, particularly among some minority populations, may be partly responsible for this increase, as illustrated by the Pima Indians.^12,13 Whereas the incidence of diabetic ESRD in Pima Indians aged ≥45 years declined after 1990, those who were younger than 45 years experienced no such decline. The lack of decline in the younger Pima Indians associates with a lower percentage use of renin-angiotensin system inhibitors than in older patients. Women of childbearing age were least likely to receive these inhibitors, presumably because of concerns about their use in pregnancy.¹⁴

In epidemiologic parlance, efficacy refers to the effect of a given treatment in a controlled environment where the extraneous effects of other factors are eliminated from consideration through randomization, and effectiveness refers to the effect of the same treatment in a population in which no such conditions exist. The studies by Forsblom et al.⁷ and Rosolowsky et al.⁸ raise important questions about the effectiveness of efficacious treatments in patients with type 1 diabetes and macroalbuminuria. Which factors determine whether efficacious treatments will be clinically effective? Do improvements in general clinical management often found in the resource-intense environment of a clinical trial enhance the efficacy of the study treatment beyond what can be reasonably expected in clinical practice? Do the high rates of progression to ESRD in these cohorts truly reflect a lack of effectiveness, or would the rates of ESRD have been even higher if newer treatments had not been used? What roles do age, ethnicity, type of diabetes, and stage of disease play in determining effectiveness of an intervention?

Several types of investigations will be required to address the therapeutic challenges identified by these provocative studies. Forsblom et al.⁷ proposed that we may need to refine our therapeutic targets by defining more precisely the determinants of competing risks. This approach may ultimately enhance therapeutic effectiveness through better targeting of existing therapies. Rosolowsky et al.⁸ argued that new therapies targeted toward other causal pathways are needed. A careful examination of the process of health care delivery may also identify points in translation where corrective steps can be taken to enhance therapeutic effectiveness. Certainly, there was much room for further improvement of glycemia, lipid levels, and BP control in both of these study cohorts. Ultimately, the difference between efficacy and effectiveness is determined by clinical practice, which is fundamentally about individualizing treatments and depends on many factors that are outside the control of clinicians.

• Copyright © 2011 by the American Society of Nephrology