{
  "abstract": "Background Environmental chemical exposures are emerging contributors to MASLD, yet prior studies evaluate chemicals individually, ignoring mixture effects. We applied an exposome-wide machine learning framework to identify hepatotoxic chemical fingerprints and characterise joint mixture effects on elastography-confirmed steatosis and fibrosis.Methods We analysed 15,587 adults from NHANES 2017-2023 with transient elastography (steatosis: CAP ≥268 dB/m; fibrosis: LSM ≥7.0 kPa). Fifty-one biomarkers across seven chemical classes were evaluated: PFAS, volatile organic compound metabolites, phthalates, blood metals, urinary metals, organophosphate pesticide metabolites, and flame retardants. Urinary chemicals were creatinine-adjusted and log 2-transformed. XGBoost with 5-fold cross-validation and SHAP interaction values identified chemical importance and dose-response relationships, adjusting for BMI, age, sex, race/ethnicity, diabetes, alcohol use, and lipid profile.Results Steatosis prevalence was 44.8% and fibrosis 16.0%. The full environmental + clinical model achieved AUC=0.870 for steatosis ( IDDF2026-ABS-0274 Figure 1. ROC curves for steatosis prediction) and AUC=0.819 for fibrosis (IDDF2026-ABS-0274 Figure 2. ROC curves for fibrosis prediction), with chemicals contributing independent predictive value beyond metabolic factors alone (ΔAUC +0.006 and +0.018 respectively). Cross-validated performance confirmed robustness (IDDF2026-ABS-0274 Figure 3. Cross-validated ROC curve for steatosis prediction, IDDF2026-ABS-0274 Figure 4. Cross-validated ROC curve for fibrosis prediction).Blood metals dominated chemical importance; selenium, manganese, cadmium, lead, and mercury ranked among the top five chemicals by mean |SHAP| for both outcomes. Moving the entire mixture from the 10th to the 90th population percentile predicted a +30.8 percentage point increase in steatosis probability.Selenium exhibited divergent stage-specific hepatotoxicity: positively associated with steatosis above log2 ~7.5 μg/L, yet inversely associated with fibrosis at high exposures. Pairwise analysis revealed a synergistic selenium × manganese interaction for steatosis and a mercury × selenium interaction for fibrosis. The highest fibrosis risk emerged at high selenium combined with low mercury, suggesting a seafood-mediated protective pathway. PFNA selectively predicted fibrosis but not steatosis, implicating a PFAS-specific pro-fibrotic mechanism independent of fat accumulation.Conclusions In the largest exposome-MASLD analysis to date, environmental chemical mixtures, dominated by blood metals independently associate with both steatosis and fibrosis. Selenium and mercury exhibit non-linear stage-specific effects consistent with threshold and interaction phenomena. These findings reframe MASLD as having a significant environmental component and identify blood metal co-exposure patterns as potential intervention targets.Abstract IDDF2026-ABS-0274 Figure 1Abstract IDDF2026-ABS-0274 Figure 2Abstract IDDF2026-ABS-0274 Figure 3Abstract IDDF2026-ABS-0274 Figure 4",
  "authors": [
    {
      "affiliations": [
        "New York Medical College, United States"
      ],
      "name": "Jeril Lasington"
    },
    {
      "affiliations": [
        "Rutgers University, United States"
      ],
      "name": "Lawin Steve Mathew Lasington"
    },
    {
      "affiliations": [
        "Boston University, United States"
      ],
      "name": "Swamynathan Umamaheshwaran"
    }
  ],
  "title": "IDDF2026-ABS-0274 The invisible risk factor: environmental toxicant mixtures as drivers of hepatic steatosis and fibrosis",
  "uid": "5d7d5c57-77a4-584c-b078-c262d4b5c794"
}