Abstract
BACKGROUND & AIMS: Liver-related mortality represents a growing public health concern, disproportionately affecting younger subjects. Because there are no established tools for early detection of individuals at risk for liver-related death (LRD), we analyzed LRD predictors in the UK Biobank (UKB) data and validated the usefulness of serum insulin-like growth factor-1 (IGF-1).
METHODS: The UKB dataset encompassing 325,981 participants, a median follow-up of 13.5 years, and 846 LRDs was used as a training cohort. IGF-1 was validated in several independent cohorts of different liver disease etiologies and fibrosis stages. A Cox proportional hazard model was used to develop the gamma-glutamyl transferase (GGT)-IGF-1 score that was validated in an independent UKB cohort with 83,528 subjects and 237 LRDs.
RESULTS: Among 59 variables in the UKB training cohort, GGT and IGF-1 were identified as the LRD predictors with time-dependent area under the curve (AUROC) >80%. Phenome-wide association study demonstrated the higher liver specificity of IGF-1 compared with GGT. In validation cohorts, IGF-1 levels: (1) increased in subjects with alcohol misuse after alcohol detoxification; (2) were reduced in individuals with alcohol-related/steatotic liver disease or severe alpha-1 antitrypsin deficiency and higher fibrosis stages; and (3) were diminished in participants with more advanced liver cirrhosis and lower levels associated with higher mortality. In the UKB training and validation cohorts, the novel GGT-IGF-1 score achieved an AUROC of 0.87 for LRD and was significantly better than established risk scores (AUROC = 0.77-0.81).
CONCLUSIONS: The study highlights the usefulness of IGF-1 as a reliable predictor of LRD and identifies a novel, population-based screening tool outperforming the currently used scores.