Background: To translate laboratory data to the bedside, we hypothesized that initiation of a thiazolidinedione would reduce the rate of progression of renal insufficiency in diabetic subjects.
Methods: We included subjects initiated on rosiglitazone for control of diabetes who had at least two consecutive serum creatinine values >= 1.5 mg/dl that were at least 4 weeks apart. We used slope estimates of reciprocal of creatinine vs. time (days) from linear models to derive the rate of decline of renal function before (Phase 1) and after (Phase 2) rosiglitazone initiation. The adjusted rate of decline of renal function was derived using repeated measure models weighted by inverse variances adjusting for hemoglobin A1C and mean blood pressure.
Results: There were 114 subjects (113 men, 1 woman; mean age 66.8 +/- 9.4 years). The mean duration of Phase 1 was 586.2 +/- 275.6 days and Phase 2 was 613.2 +/- 281.7 days (p = 0.47). The mean unadjusted Phase 1 slope of reciprocal creatinine vs. time was -0.00015 +/- 0.00021 and the Phase 2 slope was -0.00009 +/- 0.00021. The mean slope difference (Phase 2 - Phase 1) was 0.00005 +/- 0.00031 (p < 0.0001 for Wilcoxon signed rank test and p = 0.0023 for t-test). The adjusted difference in the mean slope was 0.00007 +/- 0.00042 (p = 0.0135).
Conclusion: There was a slower rate of decline of renal function after initiation of rosiglitazone in diabetic subjects with renal insufficiency. These findings warrant confirmation by a prospective randomized trial.