Serum adiponectin and glomerular filtration rate in patients with type 2 diabetes

High serum adiponectin has been increased in several conditions of kidney disease. Only sparse and conflicting results have been reported in patients with type 2 diabetes (T2D), a subgroup of individuals who are at high risk for renal dysfunction. The aim of this study was to fill up this gap of knowledge by investigating such association in a large sample of Italian diabetic patients. The association between serum adiponectin levels and estimated glomerular filtration rate (eGFR by Chronic Kidney Disease-Epidemiology Collaboration CKD-EPI equation) was investigated in 1,243 patients with T2D from two cross-sectional Italian studies: 878 from San Giovanni Rotondo (SGR) and 365 from Foggia (FG). Serum adiponectin was inversely associated with eGFR in SGR [β (standard error, SE) for 1 standard deviation (SD) of adiponectin = -3.26 (0.64)] and in FG [β(SE)=-5.70(1.28)] sample, as well as in the two studies combined [β(SE)=-3.99(0.59)];(p<0.0001 for all). In this combined analysis, the association was still significant after adjusting for sex, smoking habits, body mass index (BMI), waist circumference, diabetes duration, glycated hemoglobin (HbA1c), albumin creatinine ratio (ACR) and anti-hyperglycemic, anti-hypertensive and anti-dyslipidemic treatments [β (SE)= -2.19 (0.59), p = 0.0001]. A stronger association between each SD adiponectin increment and low eGFR was observed among patients with micro-/macro-albuminuria, as compared to those with normo-albuminuria [adjusted β(SE)=-4.42(1.16) ml/min/ 1.73m2 vs. -1.50 (0.67) ml/min/1.73m2, respectively; p for adiponectin-by-albuminuric status = 0.022]. For each adiponectin SD increment, the odds of having eGFR < 60 ml/min/1.73m2 increased by 41% (odds ratio, OR = 1.41; 95% confidence interval, CI 1.21-1.64) in SGR sample, 53% (OR = 1.53; 95% CI 1.21-1.94) in FG sample, and 44% (OR = 1.44; 95%CI 1.27-1.64) in the two studies considered together (p<0.0001 for all). In the combined sample, further adjustment for the above mentioned covariates did not change the observed association (OR = 1.36; 95%CI 1.16-1.60; p