Is the selenium paradox in NAFLD a result of reverse causation and nutritional confounding?

To the Editors, We read with great interest the comprehensive analysis by Wang et al1 employing advanced mixture models to examine the relationship between blood metals and liver outcomes. While the application of weighted quantile sum (WQS) and Bayesian kernel machine regression (BKMR) methods is commendable, the identified “selenium paradox” – where selenium appears strongly “steatogenic” yet “anti-fibrotic” – requires careful interpretation regarding its pathophysiology and clinical safety. Recent NHANES-based cross-sectional studies (2021–2025) consistently report an identical paradox: higher serum selenium positively associates with nonalcoholic fatty liver disease (NAFLD) prevalence yet negatively associates with fibrosis risk (Table 1). Table 1 - Summary of recent NHANES-based studies reporting the paradox between selenium levels, NAFLD, and liver fibrosis. Author (year) Design/population Association with NAFLD/steatosis Association with fibrosis/cirrhosis/mortality Limitations Wang et al (2025)1 Cross-sectional N = 7588 Positive association (OR = 7.40) Negative association (fibrosis OR = 0.22; cirrhosis OR = 0.007). Inflection point: 183.49 μg/L Cross-sectional design limits causal inference. Reliance on VCTE cutoffs. Potential confounding by obesity on LSM Shan et al (2024)2 Cross-sectional N = 5893 Positive association (Per quartile increase) Negative association for advanced fibrosis (male OR = 0.62; female OR = 0.48) Cross-sectional design. Potential confounding by nutritional status. • Male: OR = 1.18 • Female: OR = 1.29 Guo et al (2024)3 Cross-sectional N = 6715 Positive correlation (with CAP) Negative correlation (with LSM) Reliance on VCTE cutoffs. Potential confounding by obesity on LSM. Wang et al (2021)4 Cross-sectional N = 3831 Positive above 130 μg/L (flat below this threshold) Not assessed Threshold (130 μg/L) is significantly lower than current study. Dong et al (2024)5 Prospective (27.4 y follow-up) N = 2274 Not assessed (NAFLD cohort) U-shaped mortality association. Only longitudinal study found. Optimal intake: 122–126 μg/day Outcome was mortality, not fibrosis progression. Reja et al (2020)6 Retrospective cohort Not reported Negative association (28% lower mortality with higher selenium) Focus on mortality rather than fibrosis progression. Spaur et al (2022)7 Cross-sectional (US adults) Positive association (steatosis grade) Negative association (fibrosis stage) Cross-sectional CAP, controlled attenuation parameter; LSM, liver stiffness measurement; NAFLD, nonalcoholic fatty liver disease; VCTE, vibration-controlled transient elastography. First, the reported positive association between selenium and NAFLD (OR: 7.40, 95% CI: 1.63–33.53)1 may be driven by nutritional confounding rather than a causal role in steatogenesis. Selenium is a micronutrient abundant in protein-rich foods, grains, and nuts – key components of hypercaloric diets. Since NAFLD typically develops in a state of metabolic surplus, higher selenium levels likely reflect total caloric and protein intake. Although the authors adjusted for BMI and diabetes, the absence of data on total energy intake leaves substantial residual confounding. This helps explain why higher selenium correlates with NAFLD (a disease of overnutrition), while lower levels correlate with advanced cirrhosis (a catabolic state). Second, the striking inverse association between selenium and cirrhosis (OR: 0.007, P = 0.003)1 likely represents reverse causation. The liver is the primary organ responsible for synthesizing Selenoprotein P (SELENOP), which accounts for the majority of circulating selenium. In advanced liver disease, hepatocyte synthetic function is progressively impaired, naturally leading to reduced SELENOP production and lower serum selenium levels8. This is strongly supported by a comprehensive meta-analysis of 50 studies demonstrating a disease-stage-dependent pattern: NAFLD patients showed no significant selenium deficiency (SMD = 1.06, 95% CI: −1.78 to 3.89), whereas patients with cirrhosis (SMD = −2.06) and liver cancer (SMD = −2.71) exhibited progressively lower levels compared to controls (all P < 0.001)9. Thus, the selenium depletion observed in the current study is more likely a consequence of advancing liver disease rather than a causative factor. Finally, the proposed inflection point of 183.49 μg/L warrants caution as it notably diverges from prior evidence. Wang et al (2021) previously identified a much lower threshold of 130 μg/L, above which selenium positively associated with NAFLD4. Furthermore, the only available prospective study (Dong et al) found a U-shaped relationship with mortality, identifying a much lower optimal intake range of 122–126 μg/day5. Current cross-sectional data showing an “L-shaped” benefit should not imply safety at higher levels1,2. Pushing patients toward the 183.49 μg/L threshold – especially those with high baseline status – risks selenium toxicity and insulin resistance10. In conclusion, while Wang et al have significantly advanced the discussion on trace elements in liver disease, resolving the tension between selenium’s “steatogenic” signal and its “anti-fibrotic” statistical association is essential. However, given the lack of randomized controlled trials demonstrating that selenium prevents fibrosis in humans and the inconsistency in safety thresholds, these findings remain hypothesis-generating until prospective trials confirm clinical safety. Ethical approval Not applicable. Consent Not applicable. Sources of funding None. Author contributions T.M.H.: Conceptualization, literature review, drafting of the manuscript, and critical revision. N.G.: Conceptualization, literature review, drafting of the manuscript, correspondence, and critical revision. T.O.: Methodological input, interpretation of the literature, and critical revision. All authors approved the final manuscript. Conflicts of interest disclosure The authors declare that they have no conflicts of interest. Research registration unique identifying number (UIN) Not applicable. Guarantor Naveen Gautam. Provenance and peer review Not commissioned, externally peer reviewed.