Abstract
BACKGROUND: To evaluate the associations between pre-diagnostic levels of serum insulin, glucose and insulin resistance (HOMA-IR) and future risk of incident primary liver cancer (PLC) or chronic liver disease (CLD)-related mortality. METHODS: We used a nested case-control design to evaluate subjects over 22 years of follow-up. Glucose, insulin, and three markers of hepatitis B virus (HBV) and hepatitis C virus were measured in fasting baseline serum from 119 incident PLCs, 157 CLD-death cases and 512 matched controls. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using logistic regression to estimate the associations between insulin, glucose, HOMA-IR and the risk of PLC or CLD death. RESULTS: Compared with the lowest quartile of insulin, multivariable adjusted models showed that subjects in the highest quartile had elevated odds of developing PLC (OR(Q4/Q1) = 2.42, 95% CI = 1.26-4.75, P(trend) = 0.007), particularly in HBV-positive subjects (P(interaction) = 0.040), and of CLD death (OR(Q4/Q1) = 1.80, 95% CI = 1.02-3.21, P(trend) = 0.018). For glucose, in the HBV-positive group, subjects in the fourth quartile had an increased risk of PLC (OR(Q4/Q1) = 2.18, 95% CI = 1.07-4.60, P(trend) = 0.009), and of CLD mortality (OR(Q4/Q1) = 1.75, 95% CI = 0.95-3.28, P(trend) = 0.019). Subjects with the highest HOMA-IR values had a threefold risk of developing PLC (OR(Q4/Q1) = 2.94, 95% CI = 1.54-5.87, P(trend) = 0.001), and a twofold risk of CLD death (OR(Q4/Q1) = 2.20, 95% CI = 1.25-3.94, P(trend) = 0.005). CONCLUSIONS: We found that serum insulin and HOMA-IR could potentially be risk factors for PLC or CLD death.