Multi-system diseases and death trajectory of metabolic dysfunction-associated fatty liver disease: findings from the UK Biobank

The present study provides a panoramic view of temporal disease and death trajectories of MAFLD in the community-dwelling population of the UK Biobank cohort study. MAFLD was proved to be related to 113 medical conditions and eight causes of death when limiting disease incidence to ≥ 1%. The network of these diseases originates from asthma, diabetes, hypothyroid conditions, and tobacco abuse, with mediation primarily through diverticular diseases of intestine, chronic ischemic heart disease, obesity, benign tumors, and inflammatory arthritis. These pathways predominantly culminated in acute myocardial infarction, disorders of fluid, electrolyte, and acid–base balance, bacterial infectious agents, infectious gastroenteritis and colitis, and functional intestinal disorders. Regarding the death trajectories of MAFLD, the main causes of death included malignant neoplasm, cardiovascular, and respiratory system deaths. The prevalent lethal medical conditions associated with MAFLD were acute renal failure, heart failure, sepsis, pneumonia, and disorders of fluid, electrolyte, and acid–base balance. These maps provide a series of key pathways that link MAFLD to a broad range of health conditions. Considering the huge disease burden caused by MAFLD worldwide, these findings suggest potential key intervention targets for inhibiting the progression of MAFLD-related health events.

In the subgroup analysis, the association of MAFLD with medical conditions and death in different subgroups of sex, alcohol consumption, body weight, and liver fibrosis was largely significant. Although males had a higher prevalence of MAFLD, females had a higher risk of developing subsequent medical conditions. These sex-related differences may be explained by estrogen levels [28]. Overweight is a very common condition in the current era; even without being diagnosed with MAFLD, the proportion of overweight remained as high as 49.9%. Interestingly, this study found that individuals with lean MAFLD (BMI < 25 kg/m²) had a higher risk of multi-system diseases than did those who were overweight/obese. However, Liu et al. demonstrated that this effect remained significant only for hepatic outcomes after fully adjusting for confounding factors [5]. Based on our results, individuals with MAFLD who consumed alcohol experienced a protective effect with regard to both intrahepatic and extrahepatic outcomes and mortality. Clinical data have not conclusively confirmed the effects of alcohol consumption on MAFLD outcomes [29,30,31]. Based on the basic medical principle of “first, do no harm,” it is premature to recommend that MAFLD patients consume moderate alcohol. Similar to findings from previous studies, individuals with MAFLD and fibrosis had a higher risk of cardiac events and nonhepatic mortality [32, 33]. In the sensitivity analysis, the relationship between MAFLD and subsequent disease conditions and causes of death was still significant after excluding individuals with pre-existing diagnosis-related diseases. These results emphasize that MAFLD is a significant health problem in individuals with different characteristics.

Consistent with a previous study’s findings, the pathophysiological mechanism linking MAFLD and cardiovascular disease could be explained by diabetes, obesity, hypertension, and dyslipidemia [34,35,36]. We found that systemic inflammatory diseases and thrombotic diseases may also be critical to the development of MAFLD in cardiovascular disease. In addition, digestive system diseases, particularly infectious gastroenteritis and colitis, functional intestinal disorders, hemorrhoids, and perianal venous thrombosis, are prominent in patients with MAFLD. Regarding the gut–liver axis, these diseases may be related to enteric dysbacteriosis caused by MAFLD [37]. In this study, MAFLD mediated several common complications, particularly infections including sepsis, mycoses, infective gastroenteritis, pneumonia, lower respiratory infections, skin and subcutaneous infections, and urinary tract infections. According to animal research and ex vivo studies, the association between multi-system infection and MAFLD may be mediated by an immunosuppressive response induced by an altered hepatic metabolic profile [38, 39]. Therefore, further clinical research is required to demonstrate these relationships and provide intervention measures.

A series of medical conditions involved key nodes in MAFLD disease trajectories. Among these, hypertension, diabetes, obesity, and ischemic heart disease are widely considered adverse outcomes [7, 40, 41]. Moreover, asthma, hypothyroid conditions, tobacco abuse, and diverticulosis also dominated the beginning of the disease treegram and carried a large number of diseases downstream (e.g., complications caused by medical treatment; respiratory disease; osteoarthritis; disorders of fluid, electrolyte, and acid–base balance secondary to asthma; and anemia, infection, inflammation, and abdominal hernia secondary to diverticulosis). Furthermore, crosslinking between asthma, tobacco abuse, anxiety, depression, functional intestinal disorders, and sleep disorders suggested that individuals with MAFLD are susceptible to psychosomatic diseases [42, 43]. Our results not only emphasize the association of MAFLD with these medical conditions but also indicate the potential key intervention targets to relieve overall health issues induced by MAFLD.

MAFLD-induced malignant tumors and related deaths have always been a concern. Similar to the findings of Hwang et al., malignant neoplasms were the main cause of death (51.5%) in patients with MAFLD [44]. Our results showed that colon cancer was the cancer most likely to occur among individuals with MAFLD. In addition, disorders of fluid, electrolyte, and acid–base balance, renal failure, and infection were the main causes of death in malignant neoplasms at the end stage. Cardiovascular death was another concern, accounting for 41.4% of deaths in this study. As expected, most individuals died from the failure of multiple organs, such as the heart, kidneys, and respiratory system [45]. Although respiratory system death has attracted less attention in previous research [46], it was the third leading cause of death and accounted for 24.8% of deaths in our study. Several pathways may need to be considered. For example, (1) gastroesophageal reflux disease, tobacco abuse, metastatic cancer, falls, or asthma leading to pneumonia, (2) chronic obstructive pulmonary disease, trauma, heart failure, or renal failure leading to sepsis, and (3) hypertension leading to atrial fibrillation to stroke (supposedly intermediate factor) leading to pneumonia.

Although the results of the subgroup and sensitivity analyses were consistent, the observed disease phenotypes of MAFLD may also be explained by common etiologies, such as comorbidities, environmental factors, and lifestyle. The PRS can be used to assess the genetic risk of individual-specific diseases or disease characteristics and is not affected by other confounding factors [47]. Therefore, we utilized gene mutations strongly associated with MAFLD to investigate whether genetic susceptibility to MAFLD is associated with disease outcomes to reveal the reliability of the disease trajectory of MAFLD. Based on the results, individuals in the highest PRS tertile had 6 times the risk of MAFLD than those in the lowest PRS tertile. Critical pathways identified in the disease and death trajectories of MAFLD also exist in those with high genetic susceptibility to MAFLD, specifically in cardiovascular events, infections, organ failure, cancer, and related deaths that are of particular concern. Notably, genetic susceptibility to MAFLD had a specific pathway from liver disease to death from digestive system diseases. Consistent with previous reports that PRS of MAFLD is strongly associated with cirrhosis and hepatocellular carcinoma [48, 49], these results strongly support the key medical conditions constituting the complex disease networks of MAFLD.

The repeatability of the UK Biobank results was partially verified by the Danish National Patient Register data, but there was heterogeneity in cardiovascular diseases, such as hypertension, coronary heart disease, angina pectoris, myocardial infarction, and heart failure. However, previous studies have validated that cardiovascular diseases and cardiovascular death are important adverse events and outcomes of MAFLD [8]. Therefore, we speculate that this heterogeneity may be caused by omissions in the diagnosis of fatty liver disease using ICD codes in the Danish National Patient Register data.

Unlike the previous definition of NAFLD, which excluded other potential causes of liver diseases such as viral hepatitis or alcohol consumption, the MAFLD definition takes into account the presence of metabolic risk factors, including obesity, diabetes, dyslipidemia, and insulin resistance, in addition to evidence of hepatic steatosis [3]. It enables healthcare providers to diagnose and manage patients based on the underlying metabolic factors driving liver disease, rather than solely focusing on the presence of hepatic steatosis [50]. By broadening the scope of liver diseases associated with metabolic dysfunction, the MAFLD definition allows for a more comprehensive assessment of patients with fatty liver disease, may improve patient risk stratification, allows for a tailored management approach, and provides a better understanding of the underlying pathogenesis of metabolic liver diseases [51]. Our research findings emphasize the view that novel clinical trial designs based on the definition of MAFLD may provide a reference for comprehensive therapeutic interventions and interdisciplinary care aiming to enhance the quality of life for individuals with MAFLD [52].

The major strength of the present study is the large sample community-based cohort that prospectively collected complete data on disease diagnosis and was validated by a large sample from an electronic medical record database. Importantly, the disease trajectory analysis was used to draw a panoramic picture of the time-dependent disease occurrence and development pattern. By using data-driven methods instead of traditional analysis methods, we aimed to overcome the limitation of verifying single disease pairings based on specific hypotheses. Thus, the disease trajectory of MAFLD provides novel information that strengthens our understanding of its pathological effects. Moreover, identifying key nodes in the MAFLD disease network can provide potential intervention targets to mitigate the overall decline in health caused by MAFLD.

We acknowledge some limitations of this study. First, although liver biopsy is the gold standard for diagnosing hepatic steatosis, it is infeasible for large-scale cohort studies; thus, FLI with relatively high sensitivity and specificity is an acceptable alternative [5]. Moreover, a lack of serum insulin data may lead to an erroneous diagnosis of MAFLD in some individuals. Second, the lack of primary healthcare data may have led to the exclusion of less severe diseases, while limiting the incidence rate to > 1% in the analysis process, leaving out rare medical conditions. Therefore, the disease panorama of MAFLD has not yet been completely elucidated. Third, although previous studies have demonstrated that the accuracy of ICD-10 coding for primary diagnoses is relatively high, insufficient accuracy of secondary diagnoses may have affected the research conclusions. In addition, due to inconsistent guidelines or standards referenced by clinical doctors, the same ICD-10 code may originate from different defined diseases. Fourth, although propensity score matching was performed, many confounding factors in the PheWAS of MAFLD were not fully adjusted. To avoid the influence of confounding factors, we conducted a disease trajectory analysis at the level of genetic susceptibility to MAFLD and obtained similar results. To the best of our knowledge, this is the first study to adopt this method. Therefore, the scientific nature of this method needs to be validated. Fifth, the disease tree diagram of MAFLD has been presented according to the referred analysis methods [14,15,16], the causal relationship between paired diseases has not been fully demonstrated, and some diseases have bidirectional relationships that cannot be fully analyzed. Therefore, in the future, a more rigorous trajectory analysis method should be developed to provide more accurate data, comprehensive disease profiles, and clearer interpretations. Sixth, the response rate of participants in this study was only 5% approximately, the median age of the participants was 59 years (which differs from the progressively decreasing age of onset of MAFLD [53]), and the population included was mainly Caucasian; all of these impact factors may potentially introduce bias and limit the generalizability of the findings. Finally, considering that the disease trajectory analysis approach is purely data-driven, we conducted a test and verified the reliability and reproducibility using the Danish National Patient Register, which is the only available resource to our knowledge. However, these electronic medical record resources do not record complete baseline information of the population; thus, MAFLD cannot be determined. Additionally, the fatty liver disease referred to by ICD-10 codes was used to conduct disease trajectory analysis. Therefore, even if high consistency results are presented, caution still needs to be maintained.