A Biological-Systems-Based Analyses Using Proteomic and Metabolic Network Inference Reveals Mechanistic Insights into Hepatic Lipid Accumulation: An IMI-DIRECT Study

bayesian network
causal inference
masld
mendelian randomization
metabolic disease
metabolomics
multi-omics
proteomics
  • Objective: This multi-omics study used Bayesian network analysis and Mendelian Randomization (MR) on the IMI-DIRECT cohort to determine the causal network linking glucose/insulin dynamics, fat distribution (MRI), and plasma proteins to MASLD (liver fat accumulation).
  • Key Causal Driver: High Basal Insulin Secretion Rate (BasalISR) was identified as the primary causal driver of liver fat accumulation in both the non-diabetes and Type 2 Diabetes cohorts, suggesting it is a modifiable therapeutic target.
  • Mechanistic Insights: The study revealed a self-reinforcing bidirectional association between Visceral Adipose Tissue (VAT) and liver fat. It also identified sex-specific proteomic drivers of liver fat, with GUSB being more predictive in females and LEP (Leptin) in males.
Published

23 January 2026

PubMed: 40502600 DOI: 10.1101/2025.06.02.25328773 Overview generated by: Gemini 2.5 Flash, 28/11/2025

Study Goal and Context

Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD), formerly known as NAFLD, affects about a third of the global adult population, with prevalence rising to about 67% in people with Type 2 Diabetes (T2D). The underlying metabolic and proteomic features driving the association between MASLD and T2D are poorly understood.

This study aimed to delineate the organ-specific and systemic drivers of MASLD by applying integrative causal inference across clinical, imaging, proteomic, and metabolic domains using data from the IMI-DIRECT prospective cohort study.

Methods: Multi-Omics Causal Network Inference

The study analyzed data from the IMI-DIRECT cohort consisting of 331 adults with new-onset T2D and 964 adults without diabetes. Participants were comprehensively phenotyped, including:

  • Metabolic Measures: Glucose and insulin dynamics from frequently-sampled metabolic challenge tests, including the Basal Insulin Secretion Rate (BasalISR) and insulin clearance (Clinsb).
  • Imaging Measures: MRI-derived fat content in the liver and abdomen (Visceral Adipose Tissue - VAT).
  • Proteomics: Plasma proteins were quantified using Olink Proximity Extension Assays (446 proteins initially analyzed).

The core analytical approach involved:

  1. Bayesian Network Analysis: Used to quantify potential causal pathways and interactions, generating Directed Acyclic Graphs (DAGs) to suggest cause-and-effect relationships between clinical and protein features.
  2. Mendelian Randomization (MR): Employed as a complementary technique for associations where the Bayesian network could not determine the causal direction with high probability (suggesting bidirectional links).

Key Findings: Basal Insulin as the Causal Driver

1. Basal Insulin Hypersecretion Drives Liver Fat Accumulation

High basal insulin secretion rate (BasalISR) was identified as the primary causal driver of liver fat accumulation in both the non-diabetes and T2D cohorts.

  • In the non-diabetes network, BasalISR was the parental node for both liver fat and Visceral Adipose Tissue (VAT), suggesting it drives fat accumulation in both ectopic (liver) and visceral areas.
  • The effect of BasalISR on the liver is partially mediated through VAT accumulation.

2. Bidirectional and Consequence Pathways

  • VAT-Liver Fat Loop: Excess visceral adipose tissue (VAT) was found to be bidirectionally associated with liver fat, indicating a self-reinforcing metabolic loop.
  • Insulin Clearance: Basal insulin clearance (Clinsb) was identified as a consequence (downstream effect) of liver fat accumulation. This worsening of Clinsb was more pronounced before the onset of T2D.

3. Proteomic Drivers and Sex-Specific Differences

Of the 446 analyzed proteins, 34 were identified as key components of the metabolic networks (27 in the non-diabetes network, 18 in the T2D network, and 11 common).

  • Directly Associated Proteins: Key proteins directly associated with liver fat included GUSB, ALDH1A1, LPL, IGFBP1/2, CTSD, HMOX1, FGF21, AGRP, and ACE2.
  • Insulin-IGFBP Axis: The network showed that basal hyperinsulinemia has a direct inverse effect on IGFBP-1 (Insulin-like Growth Factor Binding Protein-1), a suppressive feedback loop consistent with IGFBP-1 being produced by the liver.
  • Sex-Stratified Drivers: GUSB was the most predictive of liver fat in females, while LEP (Leptin) was most predictive in males, highlighting a sex-specific proteomic architecture of hepatic steatosis.

Conclusions and Recommendations

The study concludes that basal insulin hypersecretion is a modifiable, causal driver of MASLD, especially prior to glycemic decompensation.

The findings demonstrate a complex, multifactorial, sex- and disease-stage-specific proteo-metabolic architecture of hepatic steatosis. Proteins such as GUSB, ALDH1A1, LPL, and IGFBPs warrant further investigation as potential biomarkers or therapeutic targets for MASLD prevention and treatment.