Association of metabolic vulnerability index with various chronic liver diseases and mortality: a large prospective cohort study Authors: Zihan Qin, Kaiqi Yang, Zheng Liang, Le Gao, Yifei Liu, Jiafei Peng, Hongtao Wei, Shutian Zhang Published: 2026 ABSTRACT BackgroundThe association between metabolic vulnerability index (MVX), a composite biomarker integrating inflammatory and metabolic pathways, and chronic liver diseases (CLDs) is still unclear.MethodsThis prospective cohort study included 255,171 participants from the UK Biobank, with a median follow-up period of 13.71 years. Multivariable-adjusted cox proportional hazards models, restricted cubic splines (RCS) and sensitivity analyses were used to evaluate the association between MVX and CLDs.ResultsThis study showed that higher levels of MVX (Q4 vs Q1) significantly increased the risk of CLDs after adjusted by socio-demographic and health status factors (all P < 0.05): cirrhosis (HR = 1.60, 95%CI:1.38–1.86), alcoholic-related liver disease (HR = 1.34, 95%CI:1.07–1.68), autoimmune liver disease (HR = 1.35, 95%CI:1.12–1.62), liver and biliary tract cancer (HR = 1.32, 95%CI:1.05–1.66), liver disease-related death (HR = 2.11, 95%CI:1.59–2.80), and all-cause mortality (HR = 1.17, 95%CI:1.13–1.21). No association was found between MVX and viral hepatitis. RCS and threshold analysis showed a reverse L-shaped association between MVX and cirrhosis and liver-related deaths, a linearly increasing association with AILD and LBC, and a J-shaped association with ALD and all-cause mortality. Subgroups analysis and sensitivity analyses demonstrated stable correlation between MVX and CLDs and death events.ConclusionMVX is a promising indicator for independently predicting CLD events, liver-related mortality, and all-cause mortality. As a marker of metabolic-inflammatory dysregulation, MVX may enable early identification of high-risk individuals for targeted primary prevention of CLDs.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12876-026-04932-0. Text box 1. Contributions to the literatureThe link between MVX and liver disease is a new field, and research on the link between the two is urgently needed.MVX has the unique advantage of linking metabolism and inflammation in predicting various liver diseases, and may be able to predict high-risk patients more accurately.There are currently no public health policies specifically for liver disease in MVX and there is an urgent need for relevant public health policies. Text box 1. Contributions to the literatureThe link between MVX and liver disease is a new field, and research on the link between the two is urgently needed.MVX has the unique advantage of linking metabolism and inflammation in predicting various liver diseases, and may be able to predict high-risk patients more accurately.There are currently no public health policies specifically for liver disease in MVX and there is an urgent need for relevant public health policies. Introduction As the global population ages and life expectancy continues to increase, chronic liver diseases (CLDs) are becoming increasingly significant for public health [1]. CLDs have emerged as a serious public health issue, leading to a large number of deaths and cases of illness, severely reducing the quality of life for patients, and also imposing a heavy economic burden [2, 3]. These liver diseases often progress over time, leading to liver fibrosis, cirrhosis, and hepatocellular carcinoma [4], thereby causing more severe health issues and wide-ranging social and economic impacts. Although some chronic liver diseases have effective management strategies including lifestyle modification, pharmacotherapy, or abstinence, prevention remains a key public health priority [5]. In recent years, metabolic inflammatory dysregulation has garnered widespread attention as a core driver of progression in certain CLDs, such as metabolic dysfunction-associated steatotic liver disease (MASLD). The Metabolic Vulnerability Index (MVX), as a composite biomarker that integrates inflammatory and metabolic states, provides a novel perspective for studying this mechanism [6], filling the gap between metabolic and inflammatory assessments and providing a new tool for systemic physiological evaluation. MVX is calculated from the Inflammatory Vulnerability Index (IVX) and the Metabolic Malnutrition Index (MMX). IVX is based on the levels of glycated albumin (GlycA) and small high-density lipoprotein particles (sHDL), reflecting the systemic inflammatory state, while MMX links branched-chain amino acids (such as leucine, isoleucine, etc.) with citric acid metabolism, quantifying the body's nutritional metabolic reserves [6, 7]. Compared to single metabolic markers, MVX, as a composite index, can more comprehensively reflect the body's metabolic and inflammatory states, overcoming the limitations of single markers in capturing a broad metabolic background. Previous research demonstrated that physiological or pathological changes reflected by MVX often occur in the preclinical stage of the disease, giving it a predictive advantage over some traditional risk factors like [8] allowing it to indicate potential risks earlier. Moreover, previous hospital-based studies have demonstrated that MVX significantly impacts patient survival rates, with its predictive efficacy even surpassing traditional clinical indicators such as age and cardiac function classification [6, 9]. This suggests that MVX may serve as a general indicator for assessing overall vulnerability in the body. Additionally, MVX is also associated with reactive oxygen species (ROS) [10] further emphasizing its importance in reflecting the body's health status. Given that MVX covers inflammatory and metabolic dimensions that closely align with the core pathophysiological mechanisms of CLDs such as insulin resistance and chronic inflammation [11], this provides important justification for MVX's role in predicting the onset of CLDs. While the impact of MVX on survival and some other diseases and its predictive efficacy have been confirmed [6, 9], the specific associations and risk implications of MVX in CLDs still require further exploration. Based on the above background, this study used large prospective cohort from 255,171 participants in the UK Biobank (UKB) to clarify the relationship between MVX and CLDs and CLD complications [including cirrhosis, alcoholic-related liver disease (ALD), viral hepatitis, autoimmune liver disease (AILD), liver and biliary tract cancer (LBC)] and verify its predictive value for liver-related death and all-cause mortality. Materials and methods Study participants As a prospective cohort study, the UK Biobank recruited approximately 500,000 British volunteers between April 2006 and December 2010 across 22 assessment centers in England, Wales, and Scotland. Participants completed questionnaires, underwent physical measurements, and provided biological samples (e.g., blood, urine, saliva). During this period, researchers continuously tracked participants' health data and updated the database. The study included 502,128 individuals from the UK Biobank. All participants were followed from their initial visit to the assessment center until loss to follow-up, first occurrence of liver diseases, any-cause death, or the end of the study (October 31, 2022), with the first event taking precedence. Baseline magnetic resonance metabolite data were not available (N = 228,152) and participants with a history of liver disease (N = 1,245) or incomplete covariate data (N = 17,560) were excluded. Ultimately, 255,171 participants were included in the primary analysis of the MVX and liver-related events. The selection process and the number of final cases are detailed in Fig. 1.Fig. 1Flow chart for eligible participants selection Flow chart for eligible participants selection Metabolic biomarker profiling and MVX calculation In the present study, participants who attended their first follow-up visit were randomly chosen to undergo metabolic biomarker profiling using high-throughput NMR spectroscopy at Nightingale Health. Detailed experimental procedures and quality control standards have been published in previous studies [12]. The sex-specific multi-marker score for MVX was calculated based on six metabolites, following the method described earlier [13]. This score ranges from 1 to 100, with higher values indicating greater metabolic vulnerability. The calculation equation for MVX is provided in the supplementary Methods. Outcome assessment The research results were obtained through an association analysis with National Health Service (NHS) records, covering the primary and secondary diagnoses in medical records of hospitalized patients, as well as the underlying or direct causes of death in death records. We obtained the cause and date of death for each participant through linkage to national death registries. Disease records and corresponding dates were extracted from hospital inpatient data. Using ICD-10 and expert consensus statements [14]: the primary outcome indicator is cirrhosis, with secondary outcomes including other liver diseases such as alcoholic-related liver disease, autoimmune liver disease, hepatobiliary malignancies, and liver-related mortality [15]. Table S1 details the ICD-10 codes used to define these outcome indicators. Covariates Covariates were selected based on previous epidemiological evidence and data availability at baseline [7, 16]. The covariates in this study included demographic factors (age, sex and ethnicity), Socio-economic factors [Education and Townsend deprivation index (TDI)], lifestyle and health status factors [body mass index (BMI), smoking status, alcohol intake status, diabetes mellitus (DM), hypertension (HTN) and cardiovascular diseases (CVD)] and laboratory indicators [alanine aminotransferase (ALT), aspartate aminotransferase (AST), triglycerides and cholesterol]. Age was calculated using the date of birth and the date of baseline assessment. Patient-reported sex at baseline data collection was categorized as female or male. Ethnicity was classified into two categories: White and Non-White. Education level was divided into college, other levels, and unknown. TDI was calculated immediately prior to participant joining UKB using preceding national census output areas. Smoking and alcohol intake status were self-reported and categorized as never, previous and current. Heart attacks, angina pectoris and strokes were all classified under the category of CVD. DM, HTN and CVD were categorized as yes or no. Statistical analysis Continuous and categorical data were presented as mean ± standard deviation (SD), median and interquartile range (IQR), or frequency (percentages). Cox proportional hazards regression models were employed to assess the association between MVX score and liver-related events, with hazard ratios (HR) and 95% confidence intervals (CI) calculated. In Cox regression model 1, no covariates were adjusted for; model 2 adjusted for age and sex; while model 3 further adjusted for ethnicity, educational level, TDI, BMI, smoking status, alcohol consumption, DM, HTN, CVD, ALT, AST, triglycerides and cholesterol. The MVX level was categorized into quartiles [the first quartile (Q1), the second quartile (Q2), the third quartile (Q3), and the fourth quartile (Q4)], and linear trends were tested by treating these quartiles as continuous variables. Restricted cubic spline (RCS) regression analysis was further utilized to examine the dose–response relationship between MVX score and liver-related events. Threshold effect analysis was utilized to explore this relationship and identify the inflection point. Subgroup analyses were performed based on different sociodemographic characteristics, lifestyles, and behaviors, including age (≤ 60 years and > 60 years), sex (female and male), ethnicity (white and others), BMI [underweight and normal (< 25 kg/m2), overweight (25.0–30.0 kg/m2), and obese (≥ 30.0 kg/m2)], smoking and alcohol history (never, previous, and current), DM status (yes or no), HTN status (yes or no) and CVD status (yes or no). The sensitivity analysis was conducted as follows: firstly, we excluded participants with a history of liver transplantation, excessive alcohol use, and illicit drug use disorders as defined by ICD-10 codes (Table S1); secondly, further adjustments were made for diet score, physical activities, sleep score, stroke, cancer, albumin, total bilirubin, high-density lipoprotein (HDL) and low-density lipoprotein (LDL). All results were considered statistically significant at P < 0.05 (two-tailed). All analyses were performed using R statistical software, version 4.4.1 (R Project for Statistical Computing). Results Baseline characteristics of participants Among the 255,171 UK Biobank participants included in the primary analysis, the median follow-up period was 13.71 years. The baseline characteristics of the participants are summarized in Table S2. The total study population had a mean age of 56.6 ± 8.1 years, with 54.1% being female and 95.3% identifying as White. In terms of educational attainment, 31.8% had a college education, 49.9% had other education levels, and 18.3% had unknown education status. The total population had a mean BMI of 27.4 ± 4.8 kg/m2, with 54.7% never having smoked, 34.9% being previous smokers, and 10.4% being current smokers; 92.3% reported current alcohol intake status. After grouping by MVX quartiles, participants with higher MVX scores showed distinct characteristics: they were more likely to be female, had higher BMI, and a higher proportion of current smokers (P < 0.001). In terms of disease burden, participants with higher MVX scores had a greater prevalence of metabolic and cardiovascular conditions, including higher rates of diabetes mellitus (DM), hypertension (HTN), and cardiovascular diseases (CVD) (all P < 0.001). Correlation of MVX with liver-related events Table 1 showed the association between MVX and liver cirrhosis, ALD, viral hepatitis, AILD LBC, liver-related deaths and all-cause mortality was evaluated using the Cox proportional hazards model in three models. For cirrhosis, the hazard ratio (HR) per SD increment of MVX increased from 0.99 (Model 1) to 1.08 (Model 2) and 1.20 (Model 3, all P < 0.001). Compared with the lowest quartile (Q1), the highest MVX quartile (Q4) conferred a 60% higher risk (HR = 1.60, 95%CI: 1.38–1.86, P < 0.001) with a significant linear trend. For ALD, the fully adjusted HR for Q4 versus Q1 was 1.34 (95%CI: 1.07–1.68, P = 0.010). No significant association was observed between MVX and viral hepatitis in Model 3. For AILD, each SD increase in MVX was associated with a 13% higher risk (HR = 1.13, 95%CI: 1.06–1.21, P < 0.001), and Q4 versus Q1 yielded an HR of 1.35 (95%CI: 1.12–1.62, P = 0.001). For LBC, the corresponding HR per SD was 1.11 (95%CI: 1.02–1.20, P = 0.018) and HR for Q4 versus Q1 was 1.32 (95%CI: 1.05–1.66, P = 0.015), both with significant linear trends. For liver-related mortality, the HR per SD rose from 0.99 (Model 1) to 1.31 (Model 3, P < 0.001), and Q4 versus Q1 showed a more than twofold higher risk (HR = 2.11, 95%CI: 1.59–2.80, P < 0.001). For all-cause mortality, MVX was consistently associated with higher risk across all models, with a fully adjusted HR of 1.17 (95%CI: 1.13–1.21, P < 0.001) for Q4 versus Q1.Table 1Association of MVX with incident liver-related eventsModel 1Model 2Model 3HR (95%CI)P valueHR (95%CI)P valueHR (95%CI)P valueCirrhosisMVX per SD0.99 (0.94 ~ 1.04)0.6171.08 (1.02 ~ 1.13)0.0051.20 (1.13 ~ 1.27)< 0.001 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.84 (0.73 ~ 0.96)0.0100.85 (0.74 ~ 0.97)0.0170.95 (0.83 ~ 1.10)0.499 Q30.91 (0.79 ~ 1.04)0.1471.00 (0.87 ~ 1.15)0.9811.15 (1.00 ~ 1.33)0.057 Q40.96 (0.84 ~ 1.10)0.5691.25 (1.09 ~ 1.43)0.0021.60(1.38 ~ 1.86)< 0.001 Trend.test0.8280.001< 0.001ALDMVX per SD0.82 (0.77 ~ 0.88)< 0.0010.92 (0.85 ~ 1.00)0.0371.05 (0.96 ~ 1.14)0.280 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.62 (0.51 ~ 0.76)< 0.0010.65 (0.53 ~ 0.79) < 0.0010.81 (0.66 ~ 0.99)0.042 Q30.56 (0.46 ~ 0.69)< 0.0010.68 (0.55 ~ 0.83) < 0.0010.83 (0.67 ~ 1.04)0.099 Q40.64 (0.53 ~ 0.78)< 0.0010.98 (0.80 ~ 1.19)0.8111.34 (1.07 ~ 1.68)0.010 Trend.test< 0.0010.3720.066Viral hepatitisMVX per SD1.07 (0.96 ~ 1.19)0.1971.14 (1.02 ~ 1.28)0.0191.05 (0.93 ~ 1.19)0.392 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.95 (0.71 ~ 1.29)0.7620.97 (0.72 ~ 1.32)0.8680.94 (0.69 ~ 1.28)0.680 Q31.05 (0.78 ~ 1.41)0.7581.15 (0.85 ~ 1.55)0.3621.03 (0.75 ~ 1.40)0.875 Q41.24 (0.93 ~ 1.64)0.1421.49 (1.11 ~ 2.00)0.0081.21 (0.87 ~ 1.66)0.255 Trend.test0.1060.0050.204AILDMVX per SD1.01 (0.95 ~ 1.07)0.8121.04 (0.98 ~ 1.10)0.2441.13 (1.06 ~ 1.21)< 0.001 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.98 (0.83 ~ 1.16)0.8090.97 (0.82 ~ 1.14)0.6911.05 (0.89 ~ 1.25)0.563 Q31.05 (0.89 ~ 1.24)0.5461.05 (0.89 ~ 1.24)0.5451.21 (1.02 ~ 1.44)0.027 Q41.00 (0.84 ~ 1.18)0.9751.09 (0.92 ~ 1.29)0.3441.35 (1.12 ~ 1.62)0.001 Trend.test0.8110.2340.001LBCMVX per SD0.95 (0.89 ~ 1.02)0.1701.00(0.93 ~ 1.08)0.9021.11 (1.02 ~ 1.20)0.018 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q21.00 (0.82 ~ 1.22)0.9860.99 (0.81 ~ 1.20)0.8961.09 (0.89 ~ 1.33)0.417 Q30.93 (0.76 ~ 1.13)0.4530.95 (0.78 ~ 1.17)0.6451.11 (0.90 ~ 1.38)0.318 Q40.88 (0.72 ~ 1.08)0.2311.05 (0.85 ~ 1.29)0.6651.32 (1.05 ~ 1.66)0.015 Trend.test0.1700.7940.021Liver-related mortalityMVX per SD0.99 (0.90 ~ 1.08)0.8231.14 (1.03 ~ 1.26)0.0101.31 (1.17 ~ 1.46)< 0.001 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.83 (0.64 ~ 1.08)0.1630.85 (0.65 ~ 1.10)0.2220.98 (0.75 ~ 1.29)0.901 Q30.91 (0.70 ~ 1.18)0.4771.08 (0.83 ~ 1.39)0.5831.24 (0.94 ~ 1.64)0.120 Q41.00 (0.78 ~ 1.29)0.9811.52 (1.17 ~ 1.96)0.0012.11 (1.59 ~ 2.80)< 0.001 Trend.test0.8170.001< 0.001All-cause mortalityMVXperSD0.96 (0.95 ~ 0.97)< 0.0011.02 (1.01 ~ 1.04)< 0.0011.06 (1.04 ~ 1.07)< 0.001 Q11.00(Ref)1.00(Ref)1.00(Ref) Q20.95 (0.92 ~ 0.98)0.0010.93 (0.90 ~ 0.96)< 0.0010.96 (0.93 ~ 1.00)0.034 Q30.93 (0.90 ~ 0.97)< 0.0010.97 (0.94 ~ 1.00)0.0791.03 (0.99 ~ 1.06)0.148 Q40.88 (0.85 ~ 0.91)< 0.0011.08 (1.04 ~ 1.11)< 0.0011.17 (1.13 ~ 1.21)< 0.001 Trend.test< 0.001< 0.001< 0.001Model 1: unadjusted. Model 2: adjusted for age and gender. Model 3: Adjusted for age, gender, ethnicity, education, TDI, BMI, smoking status, drinking status, DM, HTN, CVD, ALT, AST, cholesterol and triglyceridesMVX metabolic vulnerability index, HR hazard ratio, CI confidence interval, Q MVX quartile, TDI Townsend deprivation index, BMI body mass index, DM diabetes mellitus, HTN hypertension, CVD cardiovascular diseases, ALT alanine aminotransferase, AST aspartate aminotransferase, ALD alcoholic liver disease, AILD autoimmune liver disease, LBC liver and biliary tract cancer Table 1Association of MVX with incident liver-related eventsModel 1Model 2Model 3HR (95%CI)P valueHR (95%CI)P valueHR (95%CI)P valueCirrhosisMVX per SD0.99 (0.94 ~ 1.04)0.6171.08 (1.02 ~ 1.13)0.0051.20 (1.13 ~ 1.27)< 0.001 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.84 (0.73 ~ 0.96)0.0100.85 (0.74 ~ 0.97)0.0170.95 (0.83 ~ 1.10)0.499 Q30.91 (0.79 ~ 1.04)0.1471.00 (0.87 ~ 1.15)0.9811.15 (1.00 ~ 1.33)0.057 Q40.96 (0.84 ~ 1.10)0.5691.25 (1.09 ~ 1.43)0.0021.60(1.38 ~ 1.86)< 0.001 Trend.test0.8280.001< 0.001ALDMVX per SD0.82 (0.77 ~ 0.88)< 0.0010.92 (0.85 ~ 1.00)0.0371.05 (0.96 ~ 1.14)0.280 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.62 (0.51 ~ 0.76)< 0.0010.65 (0.53 ~ 0.79) < 0.0010.81 (0.66 ~ 0.99)0.042 Q30.56 (0.46 ~ 0.69)< 0.0010.68 (0.55 ~ 0.83) < 0.0010.83 (0.67 ~ 1.04)0.099 Q40.64 (0.53 ~ 0.78)< 0.0010.98 (0.80 ~ 1.19)0.8111.34 (1.07 ~ 1.68)0.010 Trend.test< 0.0010.3720.066Viral hepatitisMVX per SD1.07 (0.96 ~ 1.19)0.1971.14 (1.02 ~ 1.28)0.0191.05 (0.93 ~ 1.19)0.392 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.95 (0.71 ~ 1.29)0.7620.97 (0.72 ~ 1.32)0.8680.94 (0.69 ~ 1.28)0.680 Q31.05 (0.78 ~ 1.41)0.7581.15 (0.85 ~ 1.55)0.3621.03 (0.75 ~ 1.40)0.875 Q41.24 (0.93 ~ 1.64)0.1421.49 (1.11 ~ 2.00)0.0081.21 (0.87 ~ 1.66)0.255 Trend.test0.1060.0050.204AILDMVX per SD1.01 (0.95 ~ 1.07)0.8121.04 (0.98 ~ 1.10)0.2441.13 (1.06 ~ 1.21)< 0.001 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.98 (0.83 ~ 1.16)0.8090.97 (0.82 ~ 1.14)0.6911.05 (0.89 ~ 1.25)0.563 Q31.05 (0.89 ~ 1.24)0.5461.05 (0.89 ~ 1.24)0.5451.21 (1.02 ~ 1.44)0.027 Q41.00 (0.84 ~ 1.18)0.9751.09 (0.92 ~ 1.29)0.3441.35 (1.12 ~ 1.62)0.001 Trend.test0.8110.2340.001LBCMVX per SD0.95 (0.89 ~ 1.02)0.1701.00(0.93 ~ 1.08)0.9021.11 (1.02 ~ 1.20)0.018 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q21.00 (0.82 ~ 1.22)0.9860.99 (0.81 ~ 1.20)0.8961.09 (0.89 ~ 1.33)0.417 Q30.93 (0.76 ~ 1.13)0.4530.95 (0.78 ~ 1.17)0.6451.11 (0.90 ~ 1.38)0.318 Q40.88 (0.72 ~ 1.08)0.2311.05 (0.85 ~ 1.29)0.6651.32 (1.05 ~ 1.66)0.015 Trend.test0.1700.7940.021Liver-related mortalityMVX per SD0.99 (0.90 ~ 1.08)0.8231.14 (1.03 ~ 1.26)0.0101.31 (1.17 ~ 1.46)< 0.001 Q11.00 (Ref)1.00 (Ref)1.00 (Ref) Q20.83 (0.64 ~ 1.08)0.1630.85 (0.65 ~ 1.10)0.2220.98 (0.75 ~ 1.29)0.901 Q30.91 (0.70 ~ 1.18)0.4771.08 (0.83 ~ 1.39)0.5831.24 (0.94 ~ 1.64)0.120 Q41.00 (0.78 ~ 1.29)0.9811.52 (1.17 ~ 1.96)0.0012.11 (1.59 ~ 2.80)< 0.001 Trend.test0.8170.001< 0.001All-cause mortalityMVXperSD0.96 (0.95 ~ 0.97)< 0.0011.02 (1.01 ~ 1.04)< 0.0011.06 (1.04 ~ 1.07)< 0.001 Q11.00(Ref)1.00(Ref)1.00(Ref) Q20.95 (0.92 ~ 0.98)0.0010.93 (0.90 ~ 0.96)< 0.0010.96 (0.93 ~ 1.00)0.034 Q30.93 (0.90 ~ 0.97)< 0.0010.97 (0.94 ~ 1.00)0.0791.03 (0.99 ~ 1.06)0.148 Q40.88 (0.85 ~ 0.91)< 0.0011.08 (1.04 ~ 1.11)< 0.0011.17 (1.13 ~ 1.21)< 0.001 Trend.test< 0.001< 0.001< 0.001Model 1: unadjusted. Model 2: adjusted for age and gender. Model 3: Adjusted for age, gender, ethnicity, education, TDI, BMI, smoking status, drinking status, DM, HTN, CVD, ALT, AST, cholesterol and triglyceridesMVX metabolic vulnerability index, HR hazard ratio, CI confidence interval, Q MVX quartile, TDI Townsend deprivation index, BMI body mass index, DM diabetes mellitus, HTN hypertension, CVD cardiovascular diseases, ALT alanine aminotransferase, AST aspartate aminotransferase, ALD alcoholic liver disease, AILD autoimmune liver disease, LBC liver and biliary tract cancer Association of MVX with incident liver-related events Model 1: unadjusted. Model 2: adjusted for age and gender. Model 3: Adjusted for age, gender, ethnicity, education, TDI, BMI, smoking status, drinking status, DM, HTN, CVD, ALT, AST, cholesterol and triglycerides MVX metabolic vulnerability index, HR hazard ratio, CI confidence interval, Q MVX quartile, TDI Townsend deprivation index, BMI body mass index, DM diabetes mellitus, HTN hypertension, CVD cardiovascular diseases, ALT alanine aminotransferase, AST aspartate aminotransferase, ALD alcoholic liver disease, AILD autoimmune liver disease, LBC liver and biliary tract cancer In addition, we used RCS flexible modeling to evaluate the relationship between the MVX and liver-related events associated with it (Fig. 2). Overall, we observed a "first stable then increasing" significant reverse L-shaped relationship with cirrhosis and liver-related mortality (P for overall < 0.05, P non-linear < 0.001), a increasing linear relationship with AILD and LBC (P for overall < 0.05, P non-linear > 0.05), and a significant nonlinear J-shaped relationship between MVX and the risk of ALD and all-cause mortality (P for overall < 0.05, P non-linear < 0.001). Threshold analysis indicated clear turning points in cirrhosis, ALD, liver-related mortality and all-cause mortality analysis with significant associations (Table 2).Fig. 2Cubic spline regression for estimated hazard ratios of liver-related events incidence according to continuous levels of MVX score. A cirrhosis; B ALD; C AILD; D LBC; E liver-related mortality; F. all-cause mortality. All were adjusted for age, sex, ethnicity, education, TDI, BMI, smoking status, alcohol intake status, DM, HTN, CVD, ALT, AST, Cholesterol and Triglycerides. Only 98% of the data was displayed. MVX, metabolic vulnerability index; HR, hazard ratio; CI, confidence interval; TDI, Townsend deprivation index; BMI, body mass index; DM, diabetes mellitus; HTN, hypertension; CVD, cardiovascular diseases; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ALD, alcoholic-related liver disease; AILD, autoimmune liver disease; LBC, liver and biliary tract cancerTable 2Threshold effect analysis of the association of MVX and liver-related eventsMVX scoreAdjust HR (95% CI)aPMVX scoreAdjust HR (95% CI)aPCirrhosisALD < 51.440.89 (0.75,1.07)0.226< 56.940.65 (0.55,0.78)< 0.001 ≥ 51.441.45 (1.32,1.60)< 0.001≥ 56.941.48 (1.22,1.80)< 0.001 Likelihood ratio test< 0.001Likelihood ratio test0.029Liver-related mortalityAll-cause mortality < 42.040.82 (0.65,1.05)0.114< 57.540.95 (0.92,0.99)0.005 ≥ 42.041.93 (1.53,2.43)< 0.001≥ 57.541.19 (1.15,1.23)< 0.001 Likelihood ratio test< 0.001Likelihood ratio test-< 0.001aThe threshold effect analysis was calculated by MVX sccore per SD increment and adjusted for age, gender, ethnicity, education, TDI, BMI, smoking status,drinking status, DM, HTN, CVD, ALT, AST, cholesterol and triglyceridesMVX metabolic vulnerability index, HR hazard ratio, CI confidence interval, TDI Townsend deprivation index, BMI body mass index, DM diabetes mellitus, HTN hypertension, CVD cardiovascular diseases, ALT alanine aminotransferase, AST aspartate aminotransferase, ALD alcoholic liver disease Cubic spline regression for estimated hazard ratios of liver-related events incidence according to continuous levels of MVX score. A cirrhosis; B ALD; C AILD; D LBC; E liver-related mortality; F. all-cause mortality. All were adjusted for age, sex, ethnicity, education, TDI, BMI, smoking status, alcohol intake status, DM, HTN, CVD, ALT, AST, Cholesterol and Triglycerides. Only 98% of the data was displayed. MVX, metabolic vulnerability index; HR, hazard ratio; CI, confidence interval; TDI, Townsend deprivation index; BMI, body mass index; DM, diabetes mellitus; HTN, hypertension; CVD, cardiovascular diseases; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ALD, alcoholic-related liver disease; AILD, autoimmune liver disease; LBC, liver and biliary tract cancer Table 2Threshold effect analysis of the association of MVX and liver-related eventsMVX scoreAdjust HR (95% CI)aPMVX scoreAdjust HR (95% CI)aPCirrhosisALD < 51.440.89 (0.75,1.07)0.226< 56.940.65 (0.55,0.78)< 0.001 ≥ 51.441.45 (1.32,1.60)< 0.001≥ 56.941.48 (1.22,1.80)< 0.001 Likelihood ratio test< 0.001Likelihood ratio test0.029Liver-related mortalityAll-cause mortality < 42.040.82 (0.65,1.05)0.114< 57.540.95 (0.92,0.99)0.005 ≥ 42.041.93 (1.53,2.43)< 0.001≥ 57.541.19 (1.15,1.23)< 0.001 Likelihood ratio test< 0.001Likelihood ratio test-< 0.001aThe threshold effect analysis was calculated by MVX sccore per SD increment and adjusted for age, gender, ethnicity, education, TDI, BMI, smoking status,drinking status, DM, HTN, CVD, ALT, AST, cholesterol and triglyceridesMVX metabolic vulnerability index, HR hazard ratio, CI confidence interval, TDI Townsend deprivation index, BMI body mass index, DM diabetes mellitus, HTN hypertension, CVD cardiovascular diseases, ALT alanine aminotransferase, AST aspartate aminotransferase, ALD alcoholic liver disease Threshold effect analysis of the association of MVX and liver-related events aThe threshold effect analysis was calculated by MVX sccore per SD increment and adjusted for age, gender, ethnicity, education, TDI, BMI, smoking status,drinking status, DM, HTN, CVD, ALT, AST, cholesterol and triglycerides MVX metabolic vulnerability index, HR hazard ratio, CI confidence interval, TDI Townsend deprivation index, BMI body mass index, DM diabetes mellitus, HTN hypertension, CVD cardiovascular diseases, ALT alanine aminotransferase, AST aspartate aminotransferase, ALD alcoholic liver disease Sensitivity analyses Stratified analysis demonstrated a basically stable positive correlation between MVX and its associated liver-related events in different subgroups (Fig. 3). Additionally, after further excluding liver transplant recipients and individuals with a history of excessive alcohol and drug use (Table S3, S4), or after additional adjustment for other covariates (Table S5), the association between MVX and CLDs, liver-related deaths, and all-cause mortality remained stable, leading to similar conclusions and indicating the robustness of the study results.Fig. 3Association between MVX and liver-related events in subgroups. A cirrhosis; B AILD; C LBC; D liver-related mortality; E all-cause mortality. All were adjusted for age, sex, ethnicity, education, TDI, BMI, smoking status, alcohol intake status, DM, HTN, CVD, ALT, AST, Cholesterol and Triglycerides. MVX, metabolic vulnerability index; HR, hazard ratio; CI, confidence interval; AILD, autoimmune liver disease; LBC, liver and biliary tract cancer Association between MVX and liver-related events in subgroups. A cirrhosis; B AILD; C LBC; D liver-related mortality; E all-cause mortality. All were adjusted for age, sex, ethnicity, education, TDI, BMI, smoking status, alcohol intake status, DM, HTN, CVD, ALT, AST, Cholesterol and Triglycerides. MVX, metabolic vulnerability index; HR, hazard ratio; CI, confidence interval; AILD, autoimmune liver disease; LBC, liver and biliary tract cancer Discussion Based on our comprehensive analysis of the UK Biobank (UKB), MVX is significantly associated with a substantially increased risk of multiple chronic liver diseases. This association remains consistent across different MVX metrics and has been sustained through a series of sensitivity analyses. Research data show that the MVX index is positively correlated with cirrhosis, ALD, AILD, LBC, liver disease-related mortality, and all-cause mortality. RCS showed a inflection point exists exists between MVX and some liver-related mortality. For cirrhosis and liver-related mortality, when MVX is below the inflection point, the risk remains relatively stable or shows slight fluctuations. However, once it exceeds this threshold, the risk increases significantly with higher MVX levels, showing a reverse L-shaped association. For ALD and all-cause mortality, when MVX is below the inflection point, the risk decreases slowly as it rises, but it increases significantly after exceeding the threshold, showing a J-shaped association. Subgroup analysis and sensitive analysis showed that the MVX index maintains a relatively stable correlation with these liver-related mortality regardless of sex, age, ethnicity, BMI, smoking, drinking habits or calculated by different populations or models, indicating that MVX's impact on liver-related events is consistent across different populations and lifestyle cohorts. The core association between MVX levels and multiple CLDs may lay in its comprehensive reflection of metabolic dysfunction and chronic inflammation, which can be explained by the biological effects of MVX's key components. In the general population (excluding liver transplant recipients), the elevated core indicator of MVX — plasma branched-chain amino acids (BCAAs) is closely related to obesity and fat accumulation [17, 18]. Although the specific mechanisms are not yet fully elucidated, animal model studies have confirmed that abnormal BCAA metabolism can exacerbate metabolic disorders through pathways such as lipotoxicity, metabolic reprogramming, activation of inflammatory cascades, and mitochondrial damage, which are important driving factors for the progression of chronic liver disease [19, 20]. This suggests that BCAAs, as a key component of MVX, may be a significant mediator of the association between MVX and chronic liver disease. In addition, the inflammatory-related components of MVX (small HDL particles and GlycA) further strengthen its association with chronic liver disease. As a comprehensive biomarker of systemic inflammation, GlycA detects N-acetylated mucopolysaccharides in glycoproteins, reflecting multiple inflammatory substance levels simultaneously [21]. It significantly increases in both acute and chronic inflammation. In chronic liver disease, the sterile inflammation triggered by the innate immune system [22] aligns with the changes in GlycA, allowing MVX to accurately capture the inflammatory state of chronic liver disease through GlycA levels. Additionally, the functional abnormalities of small HDL particles in MVX are also involved in the association mechanism: hyperinsulinemia enhances CETP activity, leading to the formation of triglyceride-rich HDL, impairing the cholesterol transport capacity of small HDL particles, and consequently causing hepatic fat accumulation [23]. Existing studies have confirmed the association between small HDL particles and chronic liver disease [23], further supporting the potential pathway by which MVX mediates the occurrence of chronic liver disease. In summary, MVX may comprehensively capture the core pathological mechanisms of chronic liver disease development by integrating metabolic dysfunction reflected by BCAAs, chronic inflammation represented by GlycA, and functional abnormalities of small HDL particles, becoming the key biological basis for its association with chronic liver disease. This study revealed a J-shaped trend in MVX and ALD incidence, contrasting with the patterns observed in cirrhosis, AID, LBC, and liver-related mortality. This phenomenon may be associated with two pathological subtypes of ALD [24, 25], differing in steatosis severity: one showing significant fat accumulation while the other exhibits only minor changes. Regarding all-cause mortality, analysis also displayed a J-shaped relationship with MVX. While these results showed minor variations across studies, the general trend of increased mortality with elevated MVX aligns with other research findings [7]. Minor differences in conclusions may stem from variations in population screening criteria or baseline characteristics, but the overall upward trend underscored MVX's critical importance for overall health. There are several advantages to this study, including a prospective design, large sample size, long-term follow-up, the ability to adjust for multiple confounding factors and so forth. Additionally, since MVX can capture systemic metabolic vulnerability and oxidative/antioxidant imbalance—pathophysiological processes that typically precede significant overweight or hypertension—it can identify high-risk individuals earlier and implement personalized interventions [6]. However, this study also has some limitations. First, the observational design cannot establish causality and did not clarify the specific association between MVX components and chronic liver disease. Second, since all participants were from the UK Biobank database, calibration may be needed for a broader population. Third, although we explored the relationship between MVX changes and chronic liver disease, its classification remains in the preliminary stage. Subsequent follow-up data are needed for a more detailed analysis of metabolic trajectories and their correlation with chronic liver disease risk. Fourth, even after adjustment, residual confounding factors (such as genetic susceptibility) may still influence the results. In addition, heterogeneity and variability in ICD-10 coding practices across medical centers may lead to potential misclassification of liver diseases and their mortality. Furthermore, the clinical translation of NMR-derived MVX is limited by its relatively high analytical cost, technical requirements, and limited scalability in routine clinical practice and resource-limited settings. Finally, given the complex etiology of chronic liver disease, a limited sample size may exclude rare conditions, and the lack of pediatric liver disease data limits the applicability of the conclusions to young populations. Conclusions In summary, this study shows a positive correlation between MVX and most chronic liver disease while validating its predictive value for liver-related mortality and all-cause mortality. These findings suggest that early management of MVX may become a potential primary prevention strategy for various chronic liver diseases. This approach not only helps control the prevalence of chronic liver disease but also reduces mortality in patients with chronic liver disease. Supplementary Information Supplementary Material 1. Supplementary Material 1.