Participants: Consecutive chronic hepatitis C patients who underwent liver transplantation between January 1, 1995 and January 1,2005 in the Mayo Clinic in Rochester, MN. Patients entered the study at 4 months after liver transplantation. The study protocol was approved by the Institutional Review Board of the Mayo Clinic and was carried out in accordance with institutional guidelines. All participating patients gave informed consent.
Data assembly: Data were obtained on patient demographics (gender, ethnicity, age at transplantation), anthropomorphics (height, weight), donor demographics (donor age) and transplant procedure (cold and warm ischemic time). The body mass index (BMI) was calculated as weight divided by height squared. Patients underwent dual energy x-ray absorptiometry to assess body composition before transplantation. By employing two different x-ray energy sources, dual energy x-ray absorptiometry allows discrimination of two substances, in this case fat and lean tissue, and enables quantitation of the percentage body fat (17). Treatment of hepatitis C (treatment duration and response to treatment) was documented.
Analytical procedures: Biochemical data (glucose, insulin, cholesterol, triglycerides, high density lipoprotein [HDL], leptin, adiponectin, TNFα, IL-6, creatinine, bilirubin, sodium, aspartate aminotransferases (AST), alanine aminotransferase [ALT]), hematological data (hemoglobin, platelet count, international normalized ratio [INR]) and virological data (genotype, viral load), were measured in the certified Mayo Clinic laboratories.
The evaluation of HOMA-IR, cytokines and adipokines was at 4 months posttransplant, because at this time point it is still early enough for patients not having developed significant graft fibrosis and it is late enough for the largest acute effects of the transplantation procedure on homeostasis to have resolved. In addition, we also evaluated the effect of pretransplant diabetes mellitus on fibrosis progression.
Statistics: Baseline characteristics were compared using Mann–Whitney and chi-square tests.
Independent samples t-tests were used to compare mean fibrosis stage at year 1, 3 and 5. In order to avoid bias, missing values were substituted by the previous value in the analysis presented in Figure 1.
Figure 1. Mean fibrosis stage at 1, 3 and 5 years after liver transplantation, for patients with normal insulin sensitivity and for insulin-resistant patients.
Download figure to PowerPoint
The Kaplan–Meier method was used to estimate the effect of insulin resistance on occurrence of fibrosis over time. In this analysis, patients were classified according to the presence of insulin resistance (no insulin resistance vs. insulin-resistant, i.e. HOMA-IR >2.5 or treated for diabetes mellitus) at 4 months after transplantation.
Univariate and multivariate Cox regression analyses were applied to assess risk factors for the development of steatosis in the allograft. In addition, univariate and multivariate Cox regression analyses were applied to assess risk factors for the development of advanced fibrosis i.e. Knodell fibrosis stage 3 or 4. All of the following factors were evaluated in univariate Cox regression analysis: Age, gender, donor age, BMI, body fat percentage, insulin, glucose, HOMA-IR, IGF1, leptin, adiponectin, TNFα, IL-6, cholesterol, triglycerides, HDL cholesterol, albumin, bilirubin, INR, AST, ALT, creatinine, warm ischemia time, cold ischemia time, tacrolimus level, cyclosporine level, prednisone dose, HCV RNA, genotype, rejection.
Subsequently, a multivariate model was built, using variables that were significantly associated with progression to advanced fibrosis in univariate analysis, with a p-value <0,15. The variable ‘steatosis’ was added to the multivariate model because of the association with fibrosis progression and insulin resistance in previous studies. Steatosis was modeled as a time-dependent covariate to represent the ability of patients to change their steatosis score over time. Forward and backward stepwise analyses were used to determine the multivariate model with the best fit.
The model including the variables insulin resistance, donor age, AST, steatosis and acute cellular rejection, provided the best fit to the data. Importantly, insulin resistance was statistically significantly associated with development of advanced fibrosis in all multivariate models that were evaluated.
Since previous studies have shown that peginterferon treatment may influence fibrosis progression, we repeated the multivariate Cox regression analysis with and without peginterferon treatment as a time-dependent covariate. Including peginterferon treatment in the analysis did not change the results.
The results are reported as hazard ratios with 95% confidence intervals. The reported hazard ratios are the relative increases in hazard associated with increases of 10 years for the covariate donor age, 10 U/L for aspartate aminotransferases and one stage for steatosis. Multiple logistic regression analysis was used to assess which covariates were associated with insulin resistance. In this analysis, insulin resistance was defined as a dichotomous variable (yes/no): both patients with HOMA-IR >2.5 and patients receiving treatment for diabetes mellitus were defined as insulin-resistant. First, all of the following factors were evaluated in univariate logistic regression analysis: Age, gender, donor age, BMI, body fat percentage, IGF1, Leptin, Adiponectin, TNFα, IL-6, cholesterol, triglycerides, HDL-cholesterol, albumin, bilirubin, INR, AST, ALT, creatinine, warm ischemia time, cold ischemia time, tacrolimus level, cyclosporine level, prednisone dose, HCV RNA, genotype, rejection. Subsequently, a multivariate model was built, using variables that were significantly associated with insulin resistance in univariate analysis, with a p-value <0.15. The reported odds ratios are the relative increases in odds associated with increases of 1 g/dL for albumin, 10 ng/mL for leptin, 10 mg for mean prednisone and 1 kg/m2 for BMI. The results are reported as odds ratios with 95% confidence intervals.