This question has been debated for decades since studies emerged showing the superiority of renin–angiotensin system (RAS)-blocking agents over other antihypertensive therapies in terms of reducing albuminuria in people with diabetes. However, few of these trials were powered to demonstrate an effect on hard clinical endpoints such as end-stage renal disease (ESRD) or mortality. Part of the problem is the slow rate of glomerular filtration rate (GFR) decline in people with developing or early diabetic nephropathy (DN). Newly diagnosed people with type 2 diabetes in the UKPDS (UK Prospective Diabetes Study) took almost 30 years on average to develop ESRD (Bilous, 2008). Even people with established DN have an average decline in GFR of only 2.4–4.0 mL/min/year. In the IDNT (Irbesartan Diabetic Nephropathy Trial; Evans et al, 2012), RAS blockade with irbesartan cut the rate of GFR decline by approximately 38% (1.42 mL/min/year), making the logistics of performing ESRD outcome trials impractical.
In the paper summarised alongside, Schievink and colleagues tried to overcome this problem by modelling patient-level data from four trials in people with different severities of DN to see whether there was an optimum time for intervention with RAS blockade. For participants aged 60 years with early DN at baseline, RAS blockade led to a 20% increase in the median time to ESRD, a delay which increased to 30–33% in those with more advanced DN. In 45-year-olds, the effect was even greater for those with early and intermediate DN, with a 28% and 41% delay, respectively, and similar for those with advanced nephropathy (29% delay).
What can we make of these findings? The strength of the analysis is its use of patient-level data. These are always much more powerful, in that they allow individual covariates to be entered into the model. As such, the analysis adds to the consensus of support for RAS blockade in people with DN. However, as with all meta-analyses and post hoc studies, there is inherent bias from the entry criteria of the four selected studies (see the full article for details). All of the subjects were hypertensive with varying levels of glycaemic control (mean HbA1c ranged from 37 to 69 mmol/mol [5.5–8.5%]). Mean follow-up ranged from 2.0 to 3.7 years, which is short in the context of the natural history of DN. In one of the studies, the frequency of retinopathy was only 40%; therefore, it is possible that many of the participants had non-diabetic renal disease.
It is impossible to determine clinical costs and benefits from this study. On the face of it, delaying the onset of ESRD by around 4 years sounds impressive, but we have no idea of the numbers needed to treat or the possible risks (e.g. acute kidney injury) of RAS blockade for ≥20 years. Furthermore, the confidence intervals of the estimates lack precision in participants at low risk of ESRD, as the authors acknowledge.
An additional finding of no delay in ESRD in participants who showed a ≤30% reduction in albuminuria after 6 months’ RAS blockade needs further investigation, as it has important clinical implications in terms of prognosis. It would be interesting to know other clinical concomitants in these participants: was their blood pressure higher or glycaemia worse, for example?
Should we review our current guidance for the initiation of RAS blockade in type 2 diabetes as a result of this paper? All of the participants in the four studies had hypertension and/or albuminuria, so RAS blockers are recommended as first-line therapy anyway. This post hoc analysis cannot support RAS blockade in people with normal blood pressure. What it does show, however, is that, in those for whom such therapy is indicated, the sooner therapy is started the better; furthermore, even those at late stages of DN will benefit.
To read the article summaries, please download the PDF