Pentoxifylline for chronic kidney disease in people with diabetes: a meta-analysis of randomized trials

Pentoxifylline is an inexpensive anti-inflammatory agent tested in diabetic kidney disease for four decades; prior meta-analyses report reduced albuminuria. Whether it is secure, extends to kidney function, or reaches hard outcomes is unclear; we hypothesized the surrogate evidence might be conclusive and the hard-outcome evidence not.

We reviewed randomized trials of oral pentoxifylline in diabetic kidney disease to June 2026, without language restriction. Co-primary surrogate outcomes were albuminuria (ratio of means) and glomerular filtration rate (GFR) change (mean difference); the primary hard outcome was end-stage kidney disease or death (risk ratio). We pooled with random-effects models, assessed risk of bias with RoB 2, and report a prediction interval and trial sequential analysis with every estimate.

Thirty-four studies were included, 32 contributing data. Albuminuria fell by about a third (ratio of means 0.69, 95% CI 0.50-0.94; 22 studies contributing 23 comparisons), but heterogeneity was extreme (I2 97%), the prediction interval 0.15-3.10, and only 53% of the required information had accrued, with no boundary crossed. Over ≥ 12 months, GFR changed 4.49 mL/min/1.73 m2 more favorably with pentoxifylline than with control (0.84-8.14; 5 studies contributing 5 comparisons); in two trials at low or some-concerns risk of bias it was 2.96 (-15.34 to 21.26), with 11% of the required information accrued. For end-stage kidney disease or death, 29 events in 6 studies contributing 7 comparisons gave a risk ratio of 0.90 (0.46-1.77), 3% of that required. Gastrointestinal adverse events were three times as common (3.00, 1.94-4.64); serious ones were not (0.83, 0.56-1.22).

Pentoxifylline reduces albuminuria, but the effect is heterogeneous, depends on which trials are counted, and may not reappear in a new trial. Kidney failure and death have scarcely been tested. The priority is an adequately powered, long-term trial reporting hard endpoints; gastrointestinal intolerance should be anticipated.