Lifestyle and genetic risk of chronic liver disease in metabolically healthy and unhealthy individuals from the general population

Background & Aims It is unclear to what extent lifestyle and genetic factors affect the incidence of chronic liver disease (CLD) in the general population and if lifestyle affects CLD independently of underlying cardiometabolic perturbations and genetic predisposition. Methods We examined 27,991 men and women aged 44-73 years from the Malmö Diet and Cancer Study recruited between 1991-1996 and followed until the end of 2020 using registry linkage (median follow-up time 25.1 years; 382 incident first-time CLD events). Associations between cardiometabolic factors, polygenic risk scores (PRSs), and lifestyle factors in relation to CLD were examined using multivariable Cox proportional hazards regression models. Results The incidence of CLD increased with number of cardiometabolic risk factors (the hazard ratio per each additional cardiometabolic risk factor was 1.33; 95% CI 1.21-1.45; p = 5.1 x 10-10). Two novel PRSs for metabolic dysfunction-associated steatotic liver disease and a PRS for cirrhosis were associated with higher risk of CLD but provided marginal predictive utility on top of other risk factors and compared to the PNPLA3 rs738409 genetic variant. An unhealthy lifestyle (high alcohol intake, current smoking, physical inactivity and unhealthy diet) markedly increased the risk of CLD (hazard ratio 3.97, 95% CI 2.59-6.10). Observed associations between examined lifestyle factors and CLD were largely independent of cardiometabolic perturbations and polygenic risk. Conclusions We confirmed the importance of cardiometabolic dysfunction in relation to risk of CLD in the general population. Lifestyle risk factors were shown to be independently associated with CLD and added predictive information on top of cardiometabolic risk factors. Information on the polygenic risk of liver disease does not currently improve the prediction of CLD in the general population. Impact and implications: This large population-based prospective study suggests largely independent roles of cardiometabolic, lifestyle, and genetic risk factors in the development of chronic liver disease. Findings strengthen the evidence base for a beneficial effect of modification of high-risk lifestyle behaviors in the primary prevention of chronic liver disease in the general population.


Supplementary Data Description
Anthropometric, cardio-metabolic and blood measurements At MDCS baseline examinations trained nurses measured height (m) and weight (kg) and body mass index (BMI) was calculated as weight divided by height squared (kg/m 2 ).Waist circumference (cm) was measured midway between the lowest rib margin and iliac crest.
Blood pressure (mmHg) was measured using a mercury-column phygomanometer after 5 min of supine rest.Hypertension was defined as blood pressure >130/85 mmHg and/or use of anti-hypertensive medication(s).Prevalent diabetes mellitus at baseline was based on selfreported history of diabetes, diabetes diagnosis in national/local registries, current use of diabetes medications or fasting whole blood glucose of at least 6.1 mmol/l (corresponding to plasma glucose ≥7.0 mmol/L) at baseline examination.
All fasting blood samples were donated after an overnight fast and stored at −80°C.Fasting glucose, fasting insulin, high-density lipoprotein (HDL, mmol/l), and triglycerides (mmol/l) were measured at the Department of Clinical Chemistry, Skåne University Hospital in Malmö, which is attached to a national standardization system.Low-density lipoprotein (LDL) was estimated using Friedewald's formula.Fasting glucose at baseline was measured in whole blood by a hexokinase-glucose-6-phosphate dehydrogenase method (1).A constant factor of 1•11 was used to convert concentration in whole blood to the equivalent concentration in plasma (2).Homeostatic Model Assessment -Insulin Resistance (HOMA-IR) was calculated according to Matthews et al. (3) by using the formula: (fasting insulin × fasting glucose)/22.5,where insulin is expressed as mIU/l and glucose as mmol/l (1).C-reactive protein (CRP) concentration using the high-sensitive C-reactive protein (hsCRP) test, was performed using the Tina-quant® CRP latex assay (Roche Diagnostics, Basel, Switzerland) on an ADVIA® 1650 Chemistry System (Bayer healthcare, NY, USA).

Lifestyle variables
Age and sex were extracted from the participants' Swedish personal identification number.
Educational level was categorized based on years and level of education completed i.e., less than 9 years or completed elementary school, middle school, high school or at least one year of studies at advanced level after high school but without degree, or university degree.
Smoking status was categorized as never, former or current (including irregular) based on self-reported use in the baseline questionnaire.Alcohol consumption was estimated based on information from both the baseline questionnaire and the reported intake during a 7-day food record that included detailed registration of cooked meals, medications, supplements and cold beverages (4).Non-consumers of alcohol (i.e., defined as those reporting no alcohol intake during the preceding year in the baseline questionnaire and reporting no intake during the 7-day registration) were classified as zero consumers while alcohol intake among consumers was categorized as low, moderate or high (i.e.<15, 15-30, or >30 g/day for women, and <20, 20-40, or >40 g/day for men).Level of leisure-time physical activity was assessed by participants reporting reported the number of minutes per week for seventeen different leisure-time activities and combined into a physical activity score (5).Participants were ranked from low to high leisure-time physical activity level by dividing them into sexspecific quartiles of total score.Dietary intakes were assessed using a modified diet history method combining the 7-day food record with a 196-item semi-quantitative food questionnaire.Overall eating habits, quality of reported intakes in the food record and the questionnaire and potential overlap using the two modalities were further assessed using a 45-60 minutes dietary interview with a trained nutritionist (4).Reported food intake was used to calculate total dietary fiber intake using the food and nutrient data base from the Swedish National Food Agency (4).The reproducibility and validity of the diet assessment method has been described previously (6)(7)(8).We examined three previously proposed 'healthy' dietary components (dietary fiber, fruits and vegetables, and coffee) and two 'unhealthy' components (sugar-sweetened beverages and red and processed meat) dietary components.Dietary components examined were selected based on the previously reported directionally consistent associations with both liver-related outcomes and cardiometabolic diseases (9)(10)(11)(12).Selection was further guided based on availability of data in the MDCS.Dietary intakes were energy-adjusted by calculating the relative intake in grams per 1000 kcal of estimated total energy intake.

Table S1 ICD
-codes and number of prevalent and incident first events of chronic liver disease (CLD) in the Malmö Diet and Cancer Study (N=30,446) identified in Swedish national registries including the inpatient register, hospital-based outpatient care and cause-of-death register.Only the first recorded event of the included endpoints is shown.Subjects with an incident diagnosis of chronic viral hepatitis

Table S2
(14) of genetic variants and weights used to construct the weighted polygenic risk scores (PRS) for MASLD (PRS-MASLD), cALT (PRS-cALT), and liver cirrhosis (PRS-cirrhosis).Weights are the Z-scores from the GOLDPlus European ancestry meta-analysis presented in Chen et al.Nature Genetics 2023 (DOI: 10.1038/s41588-023-01497-6)(13).Negative weights were used if the reported risk-increasing allele was different from the minor allele.bWeightsare the natural log of odds ratios for liver cirrhosis from Emdin et al.Gastroenterology 2021 (DOI: 10.1053/j.gastro.2020.12.011)(14).Negative weights were used if the reported risk-increasing allele was different from the minor allele.cWeightsare the beta coefficients for unexplained chronically elevated ALT levels as a proxy for MASLD in Vojkuvic et al.Nature Genetics 2022 (DOI: 10.1038/s41588-022-01078-z)(15).Negative weights were used if the reported risk-increasing allele was different from the minor allele.*Proxyvariantforrs138764179identifiedbyChen et al. (13) ** Proxy variant for rs1799992 identified by Emdin et al.(14)TableS3Cardiometabolic risk factors for chronic liver disease (CLD) in the Malmö Diet and Cancer Study (MDCS; n=27,991) and the sub-sample with fasting blood samples taken at baseline (n=4,549).Model 1 presents hazard ratios (HR) and 95% confidence intervals from a Cox proportional hazards regression model adjusting for age and sex.Model 2 includes adjustment for age, sex, prevalent diabetes mellitus, body mass index, hypertension and use of lipid-lowering drugs.TableS4Multiplicative interaction terms between cardiometabolic, lifestyle and genetic risk factors on risk of chronic liver disease from a Cox proportional hazards regression model with adjustment for age, sex, and educational level.Nominally significant interaction terms (p<0.05) are marked in bold font. a

Table S5
Effect of genetic risk variants (single nucleotide polymorphism; SNP) included in polygenic risk scores on risk of chronic liver disease (CLD) in the MDCS (N=26,965).Hazard ratios (HR) and 95% confidence intervals (CI) from a Cox proportional hazards regression model adjusting for age and sex.Homozygous carriers of the minor allele were few and therefore heterozygous and homozygous carriers of the minor allele were collapsed into one category.TableS6PNPLA3 rs738409 genetic risk variant and polygenic risk scores (PRSs) for metabolic dysfunction-associated steatotic liver disease (MASLD), liver cirrhosis and unexplained chronic ALT elevation (cALT) in relation to incidence of chronic liver disease (CLD), steatotic liver disease (unspecified), liver cirrhosis (all-cause) and hepatocellular carcinoma (HCC) in the Malmö Diet and Cancer Study (N=26,965) stratified by age and sex.Hazard ratios (HR) and 95% confidence intervals (CI) per risk G-allele in PNPLA3 rs738409 and per standard deviation increase in normalized (z-score) PRS-MASLD, PRS-cirrhosis and PRS-cALT. *