Do low-protein diets retard the loss of kidney function in patients with diabetic nephropathy?

The world-wide epidemic of diabetes mellitus and the increasing numbers of patients with chronic kidney disease (CKD) and end-stage renal failure caused by diabetic nephropathy continue to stimulate the search for methods to prevent the development and progression of CKD. Therefore, the current meta-analysis by Pan et al (1) of the effects of low-protein diets (LPDs) on the progression of CKD and proteinuria is highly relevant. After searching several databases, the authors identified 8 published randomized controlled clinical trials (RCTs) of LPDs in 519 patients with diabetic nephropathy. Data from CKD patients with type 1 or type 2 diabetes mellitus were combined, an approach that the authors contend is justified because the pathophysiology of CKD in these 2 types of diabetes is similar. For inclusion in the analyses, RCTs had to be of 6 mo duration and had to provide outcome data on glomerular filtration rate (GFR) or creatinine clearance (Ccr) and albuminuria or proteinuria. Appropriate methods for weighting differences and testing for heterogeneity and publication bias appear to have been used. Crossover studies of LPDs were excluded because of possible carryover effects from one diet to the other. The authors’ analyses indicated that LPDs were not associated with a slower progression of GFR or Ccr in type 1 or type 2 diabetes mellitus, when considered together (P 0.61) or separately, although LPDs were associated with a decrease in glycated hemoglobin (HbA1c) concentrations (P 0.005). There was also a reduction in albuminuria or proteinuria with the LPDs (P 0.003); the latter data were associated with substantial heterogeneity. These findings stand in contrast to meta-analyses of the effects of LPDs in nondiabetic persons with CKD. Significant prolongation of the time until the beginning of renal replacement therapy and a slowing of the loss of GFR have been described with LPDs in these patients (2–5). However, whereas the prolongation of time until renal replacement therapy was substantial (2, 3, 5), the slowing of the loss of GFR was rather small: 0.53 mL min 1 y 1 (4). These studies in persons without diabetes (2–5) do not conflict with each other, because LPDs also decrease the rate of generation of protein metabolites and, presumably, uremic toxins. Hence, CKD patients ingesting LPDs may start maintenance dialysis therapy at lower GFRs, presumably because, for any reduction in GFR, they accumulate less uremic toxins and therefore are less symptomatic (6). Another recent meta-analysis concurred that there was no significant slowing of the loss of GFR in diabetic patients prescribed LPDs, although a trend toward such a reduction was observed (P 0.14) (7). Studies of the effects of LPDs on CKD patients may be deceptively difficult to evaluate. First, the actual protein intakes consumed in the LPDs evaluated in the meta-analysis by Pan et al may have been too high (1). The mean protein intakes of the subjects assigned to the LPDs ranged from 0.71 to 1.10 g kg 1 d . In more than half of the patients assigned to the LPDs, the average protein intakes ranged from 0.93 to 1.10 g kg 1 d . These actual protein intakes stand in contrast to the LPDs recommended by some workers in the field of 0.60 to 0.80 g kg 1 d 1 for nonnephrotic patients with CKD (6, 8, 9). The study by Pan et al also included patients with microalbuminuria, proteinuria, and the nephrotic syndrome; it is possible that the response to LPDs may vary in persons who have these different degrees of protein excretion. It is puzzling that, in the meta-analysis by Pan et al, although LPDs did not slow the loss of GFR, they were associated with less albuminuria or proteinuria and lower HbA1c concentrations, which are risk factors for more rapid progression of CKD (6). In this regard, it would have been helpful to analyze separately the LPD effects on the patients with greater degrees of proteinuria, including those with the nephrotic syndrome. This latter analysis was probably not possible, however, because the sample size of the patients with heavy proteinuria was probably too small for a valid statistical examination of the effects of LPDs on GFR loss; in addition, it may not have been possible to obtain individual outcome data from these latter studies. Moreover, the duration of treatment ranged from 9 to 48 mo and was rather short in most of the studies analyzed by Pan et al; in 4 of the 8 reports, the LPD treatment lasted 6–12 mo. In some RCTs, the divergence of the GFR in the patients prescribed the LPD and that in patients prescribed higher protein intakes occurred gradually, over many months, and significant changes in GFR may not have been initially apparent (4, 10). A further confounding factor is that, when persons with normal kidneys or CKD ingest smaller amounts of protein, there is a rather abrupt but limited reduction in GFR, which is considered to be caused by hemodynamic changes (6). This phenomenon has increased the difficulties of using GFR as the key outcome measure of the