Iron is essential for many vital functions, however, there are no regulated means by which excess iron can be disposed of in humans. Therefore, whenever systemic iron exceeds its needs and storage capabilities are saturated, toxicity due to iron overload may arise (1). The liver plays a core role in systemic iron regulation. It is a major storage site for iron, taking up iron, and releasing iron back into the circulation (2). Additionally, the liver acts as a regulator in iron absorption and recycling for the organ to synthesize proteins to modulate iron homeostasis (3). Therefore, hepatic injury and dysfunction can disturb iron hemostasis. On the other hand, the presumption that iron-catalyzed oxidative injury plays a key role in the pathology of various forms of liver disease permeates the scientific and clinical literature (3). Iron is implicated in the pathogenesis of several human liver diseases, as hepatocellular and/or mesenchymal iron deposition was found in them. The type of liver siderosis (parenchymal, mesenchymal, or mixed) and its distribution throughout the lobule and the liver are useful means for suggesting its etiology (4). Accumulated observational studies have reported the association between iron overload and liver dysfunction (5–12). Nevertheless, confusion and conflict exist as the pathogenesis of various liver diseases is complicated. Moreover, for observational studies, it is difficult to distinguish causal and spurious associations due to problems of confounding and reverse causation.

Plasma concentration of liver enzymes (i.e., alanine aminotransferase [ALT], aspartate aminotransferase [AST], alkaline phosphatase [ALP], and gamma-glutamyl transferase [GGT], direct bilirubin [DBIL], and total bilirubin [TBIL]) are routinely measured clinical markers that represent different dimensions of liver function (13). Observational and laboratory studies suggested that systemic-iron regulation was associated with plasma concentrations of these enzymes (7, 10, 14). Iron overload is supposed to correlate with liver dysfunction, however, it is currently uncertain whether iron disorders contribute to, or result from liver dysfunction.

Genes are randomly allocated at conception, so that genetic effects on the exposure cannot be affected by classical confounding factors or reverse causation, as in the situation where the phenotype level is influenced by the presence of the disease (15). Mendelian Randomization (MR), where genetic variants that are strongly associated with a risk factor of interest are used to test its causal effect on an outcome, can help to distinguish causal effects from associations due to confounding or reverse causality (16). By studying the effect of iron status related to randomly allocated alleles, such an MR approach has previously been used in targeted analyses to investigate the effect of iron status on the risk of Parkinson's disease, coronary artery disease, stroke, and asthma (15, 17–19). However, so far, the causal effects of systemic iron status on liver function and related liver diseases are unclear.

In this work, we have extracted the largest available and updated datasets from UK Biobank to interrogate the potential effect of iron status (referred to serum iron, ferritin, transferrin, and transferrin saturation in a pattern consistent with an effect on systemic iron status), on liver function, proxied by multiple biomarkers (ALT, AST, ALP, GGT, DBIL, and TBIL), as well as on related liver diseases (including nonalcoholic fatty liver disease (NAFLD), alcoholic liver disease (ALD), viral hepatitis, liver fibrosis and/or cirrhosis and liver malignant neoplasm). Furthermore, we have also investigated whether systemic iron levels have a sex-specific effect on liver diseases since the morbidity of them are gender-dependent. Replication analyses in the largest liver disease genome-wide association study (GWAS) and reverse MR analysis were performed to comprehensively evaluate the iron-liver disease associations.

Although multiple observational studies have reported the association between iron overload and liver dysfunction (7, 11, 14, 33), it is still unclear whether this association reflects causation. Our findings from the MR analyses showed strong evidence that genetic predisposition to higher systemic iron status was related to a higher risk of liver injury, which was reflected by increased blood concentration of ALT, AST, ALP, DBIL, and TBIL. In further analyses on liver diseases, our primary and replication results all showed that genetic liability to higher iron status was associated with increased risks of NAFLD, ALD, and liver fibrosis/cirrhosis. Moreover, our analyses stratified by sex indicated that genetically predicted iron status was associated with a higher risk of NAFLD and liver fibrosis/cirrhosis in men. Although serum iron wasn't found to be significantly associated with liver malignant neoplasm, other iron markers indicated the detrimental effect of iron on the progression of this severe disease. No strong evidence was found for a causal effect of genetically predicted iron status on viral hepatitis. In the reverse MR analyses, no significant genetic effects were observed for NAFLD progression on iron status.

Hereditary hemochromatosis (HH) is an iron overload disease and is the most common genetic condition in people of European descent (34). Observational evidence suggested that patients with HH were associated with a 4~11-fold risk of liver disease, including hepatocellular carcinoma, hepatitis C and nonalcoholic steatohepatitis (35). Other clinical studies also reported that HH-related iron overload can result in life-threatening clinical complications, most notably of which were severe liver diseases such as cirrhosis or hepatocellular carcinoma (HCC) (36, 37). Arguably, however, it remains confusing whether the risk of liver disease in HH homozygotes is attributable to disease penetrance, or from the independent hepatotoxic effect of iron. Epidemiological studies suffer from confounding and reverse causation, which is intrinsic to their observational nature so they can hardly provide conclusive evidence on the causality of an observed association. Thus, independent MR studies and large prospective trials using more accurate methods of diagnosing liver disease are needed to reach a conclusion on the crucial role of iron.

So far, this is the first MR study to examine the causal effect of predisposition to iron status on liver function. Previous observational studies observed an elevation of serum transaminases in most hemochromatosis patients and mouse models of iron-induced liver injury (38–40). Serum ferritin levels were also reported to associate with elevated liver function enzymes (7, 14). However, the intrinsic causality between iron status and liver function warrants to be clarified. Our results of increased systemic iron status cause elevated circulating liver injury biomarkers, including ALT, AST, ALP, DBIL, and TBIL, provide strong genetic evidence that excess iron could be a trigger for liver dysfunction.

The deleterious effect of higher iron levels on NAFLD and ALD found in our study may get support from previous observational and experimental studies. It is not uncommon that epidemiological studies point toward an association between high ferritin levels and the presence of NAFLD and steatohepatitis (7, 11, 14, 41, 42). As an indicator of systemic iron status, elevated serum ferritin concentration is an independent predictor of advanced hepatic fibrosis in patients with NAFLD (43). Accumulating evidence suggested that hepatic mitochondrial dysfunction might contribute to NAFLD development and severity, as occurs in experimental iron overload (44). Moreover, hepatic iron excess might contribute to the impairment of glucose homeostasis by influencing insulin signaling and metabolic control, reinforcing the idea of a possible role of iron in favoring NAFLD progression (45, 46). Sets of data suggested a correlation between systemic iron burden and the extent of liver damage in ALD patients (47, 48). The underlying pathogenic mechanisms of iron-promoted alcoholic steatohepatitis may be ascribed to the ability that alcohol and iron synergistically caused oxidative stress, stellate cell activation, and hepatic fibrogenesis, an exaggerated effect on liver disease progression (49). On the other hand, excess iron accumulation in Kupffer cells (KC) has emerged as a central event in hepatic toxicity in ALD (50), suggesting the dominant role of iron in the hepatic toxicity of ALD.

Studies have shown that the risk of cirrhosis is increased in iron-overloaded patients (51). Laboratory studies also reported that iron-generated oxyradicals and lipid peroxidation results in damage to hepatocellular organelles, which is thought to contribute to the development of hepatic fibrogenesis (52). It is highly necessary to use genetic variants to assess the genetic effect of iron exposure on liver cirrhosis. Our study supports the genetic association of increased iron levels with the risk of developing liver fibrosis/cirrhosis. The underlining mechanism of iron-triggered liver fibrosis/ cirrhosis progression could be explained by a newly identified, iron-dependent, and non-apoptotic cell death named ferroptosis (53). Previous studies reported that excess iron-induced liver ferroptosis played a key role in hepatic dysfunction and liver damage in different liver fibrosis/cirrhosis mouse models (54, 55), indicating the crucial effects of iron triggered ferroptosis in liver fibrosis/cirrhosis. Although epidemiologic studies have yielded mixed results for liver malignant neoplasm, there is an iron–cancer relationship among patients with HH, who are prone to develop liver cancer (33). However, we still emphasize that large prospective studies in selected groups of participants are needed to evaluate the independent pathogenic role of hepatic iron deposits in liver fibrosis/cirrhosis and malignant neoplasm.

In this study, we observed a sex-dependent casual association between iron status and NAFLD risk in men, similar sex-dimorphic result was also found for liver fibrosis/cirrhosis. These findings were corroborated by previous population-based studies which reported sex-specific prevalence of liver disease in HH patients, with the male gender increasing the estimated effect size for liver disease association (34). The prevalence of cirrhosis in the United States was reported to independently associated with the male sex (56). Males are also shown more susceptible to non-alcoholic fatty liver disease, non-alcoholic steatohepatitis, and liver fibrosis than females in humans or animals (57, 58). Previous observational studies suggested a significantly higher propensity of NAFLD among men with higher iron levels. The HH (H63D) mutation was reported to be significantly associated with male NAFLD patients, but not in female patients with NAFLD, suggesting that female patients with the H63D mutation might be protected from the development of NAFLD by iron loss through menstruation or pregnancy (59). In accordance with this, participants with higher dietary iron intake were subject to a higher prevalence of NAFLD in a dose-response relationship manner, moreover, after stratifying by gender, such association only remained in the male population (60). The observed sexual dimorphic associations between iron status with NAFLD and liver fibrosis/ cirrhosis progression may be explained by multiple pathways. It was established that iron accumulation occurs later in women, as serum ferritin levels in adult men were around 120 μg/L, while the values remained low in women at around 30 μg/L until menopause when ferritin levels increase to around 80 μg/L. This partially explained the stronger effect size of iron on liver disease among men. In addition, estradiol and its derivatives had a strong antioxidant capacity to suppress the generation of iron-induced reactive oxygen species and lipid peroxidation in the liver (61), which may prevent hepatocytes from oxidative damage, inflammation, and cell death.

Since randomized trials are very difficult to perform in the case of iron and liver disease, as investigating the long-term effect would require not only a very long follow-up but also a huge sample size, MR offers a cost-effective approach to derive well-powered causal effect estimates. However, the potential limitations of MR and this study deserve comment. In our study, genetic variants of iron status were used as IVs to explore the association between iron on liver function and liver disease based on 3 core MR assumptions: (1) The association between the genetic variants and the exposure is reliable, suggesting that the genetic IVs for the exposure should be selected at a genome-wide significance level (P < 5 × 10⁻⁸). In our MR study, we selected 3 SNPs strongly associated with all four iron status biomarkers (P < 5 × 10⁻⁸) that are broadly used in other MR studies (15, 17–19). The instrument strength was high for all of them, as shown by their F-statistic values. (2) The genetic variants should not be associated with any possible known confounders. (3) The genetic variants should affect the outcome only through the exposure. The second and third assumptions are known collectively as independence from pleiotropy. Pleiotropy is a potential source of bias specific to MR studies. In sensitivity analysis, we checked for unknown pleiotropy using the weighted median and MR-Egger methods and found little indication of pleiotropy. Investigation for SNP-associated secondary phenotypes using online databases also found no evidence of traditional liver disease risk factors, consistent with the assumption of no alternate causal pathways. Moreover, we also note that a Wald-type estimator for MR analysis of binary outcomes has been shown to induce bias, although typically of small magnitude (within 10%) (62). In addition, when interpreting our results, some concerns should be taken into consideration. Firstly, due to the low incidence of severe liver disease, the proportion of disease cases from UK Biobank was relatively small, which might reduce the effect size of the iron-disease estimate and result in statistically insufficient for MR analysis. However, our persistent results in primary and replication analyses showed a causal effect of iron overload on the enhanced risk of liver injury, providing evidence to reach the conclusion that increased iron is detrimental to liver function. Secondly, although our findings on the sex-specific analysis were supported by previous observational studies, it should be careful when interpreting the results as the statistical power was limited, suggesting that the insignificant associations might be due to a lack of power. Moreover, sex-specific IVs for iron status are not available. Therefore, further larger GWAS studies that explore the sex effects of iron are needed. Thirdly, due to not enough IVs being available for other liver diseases in current GWAS studies, the reverse MR analysis was conducted only for NAFLD patients. Moreover, a detailed fibrosis assessment was not available as the disease identification in the database has been generated from the ICD-10 codes. Further GWAS studies with a larger number and proportion of liver disease cases, and detailed information for disease characteristics, are warranted. Finally, to minimize bias from population stratification, we confined the study population to individuals of European ancestries, our findings may be limited when generalized to other populations with different ethnicities.

In conclusion, our study shows the genetic effects of increased iron levels on a higher risk of liver injury, and the development of NAFLD, ALD, and liver fibrosis/cirrhosis. In addition, sex-specific analyses found the hepatoxic role of iron in NAFLD and liver fibrosis/cirrhosis progression among men. These findings provide genetic evidence that disrupted iron metabolism may be a trigger in the pathogenesis of liver injury. Independent GWAS and large prospective studies are warranted to further validate our findings in other cohorts and ethnicities.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding authors.

Ethical review and approval was not required for this human study as it was conducted using summary data obtained from a public database.

LJ and KW: responsible for the writing of the original draft. LJ, KW, FY, YY, and PZ: reviewing and editing the manuscript and for acquisition, analysis, or interpretation of data. LJ, KW, and FY: conceptualization of the study, statistical analysis, and accessed the database and raw data. LJ: supervision. All authors contributed to the article and approved the submitted version.

This study was supported by research grants from the National Natural Science Foundation of China (32000820 to LJ) and the China Postdoctoral Science Foundation (2022M712763 to KW).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.