Table 3.

Cox proportional hazard analysis of risk for CKD≥3 in patients with T1D according to clinical characteristics and circulating markers of TNF pathway in univariate and multivariate models

Baseline CharacteristicUnivariate Model: Hazard Ratio (95% CI)Multivariate Model: Hazard Ratio (95% CI)
Clinical Predictors OnlyClinical Predictors and Markersa
Clinical predictorsb
 sex (male)0.55 (0.33–0.90)
 age1.44 (1.15–1.81)
 body mass index1.11 (0.93–1.32)
 systolic BP1.12 (0.95–1.32)
 diastolic BP1.03 (0.83–1.28)
 serum cholesterol1.39 (1.11–1.75)
 ever smoking1.18 (0.73–1.91)
 age at diabetes diagnosis 1.37 (1.10–1.71)
 Diabetes duration1.08 (0.91–1.30)
 HbA1c1.52 (1.28–1.79)1.61 (1.36–1.90)1.59 (1.33–1.90)
 AER1.59 (1.33–1.91)1.20 (0.99–1.47)1.16 (0.95–1.42)
 eGFRcystatin2.00 (1.71–2.32)1.88 (1.61–2.19)1.64 (1.37–1.95)
 treated with RASi/AHTN2.17 (1.30–3.61)
 cohort (JKS2 versus JKS1)2.03 (1.13–3.66)
Circulating markers: Q4 versus Q1–Q3c
 free TNFα2.14 (1.23–3.74)1.44 (0.80–2.59)d
 total TNFα4.89 (2.81–8.51)1.87 (1.00–3.48)d
 TNFR16.42 (3.90–10.6)2.53 (1.36–4.67)
 TNFR27.16 (4.31–11.9)3.04 (1.67–5.52)
  • RASi/AHTN, renin-angiotensin system inhibitors and/or other antihypertensive agent; JKS, Joslin Kidney Study; Q1–Q4, quartiles 1 to 4.

  • a The effect of each circulating marker was examined separately while controlling for clinical predictors. Hazard ratios for clinical predictors are from the multivariate model with TNFR2.

  • b Hazard ratios for each clinical covariate are for a one-quartile increase in their respective distribution except for categorical variables and for eGFRcystatin, for which it is a one-quartile decrease.

  • c Clinical characteristics of patients with fourth quartile versus first through third quartiles are shown in Supplemental Table 1.

  • d Effects of free and total TNFα were not significant in the multivariate model.