In 2024, liver cancer was ranked 6th most common cancer and 3rd leading cause of cancer death worldwide, with annually close to 900,000 new cases and approximately 800,000 deaths worldwide.¹ In Vietnam, primary liver cancer is the leading cancer in both incidence and mortality, accounting for 15.4% and 22.1% of total cancer, respectively.² Hepatocellular carcinoma (HCC) is the most common histological subtype, accounting for 90–95% of all primary liver cancer cases in Vietnam.³ Established major risk factors for HCC are chronic infection with hepatitis B (HBV) and hepatitis C virus (HCV), excessive alcohol use, and to a lesser extent, dietary exposure to aflatoxin in certain regions.³ Given the diminishing role of HBV (due to universal vaccination program) and HCV (available curative therapy), metabolic dysfunction-associated steatotic liver disease (or MASLD),⁴ a nomenclature of non-alcoholic fatty liver disease (or NAFLD), has emerged as the most important risk factor for HCC.

MASLD/NAFLD (both terms being interchangeable) comprises a spectrum of liver diseases from simple steatosis, non-alcoholic steatohepatitis (NASH) which progresses to fibrosis and compensated cirrhosis. Compensated cirrhosis would further develop to HCC and decompensated cirrhosis that requires liver transplantation or leads to death (hereafter referred as “end-stage liver diseases”).⁵ The prevalence of MASLD is about 25% of adult population worldwide⁶ and close to 34% in Asia during 2012–2017 period.⁷ MASLD significantly contributes to the rising incidence of liver cancer in the U.S and worldwide. It is estimated that the cumulative incidence of HCC is around 2.4–12.8% among patients with NASH with advanced fibrosis or compensated cirrhosis over 3–7 years.⁸ MASLD is the second, and will soon take over viral hepatitis as the most common indication for HCC/liver transplantation in the U.S. and worldwide.⁹ It is therefore an urgent and unmet need to identify predictors of MASLD.

MASLD is found to be associated with metabolic syndrome (i.e., obesity, insulin resistance, type 2 diabetes mellitus-T2DM), and cardiovascular complications (i.e., dyslipidemia and hypertension).^10,11 Also, while biopsy is considered a gold standard for MASLD diagnosis, its major drawback is the invasives. Different non-invasive modalities have been developed to compensate the biopsy, including diagnostic imaging or clinical-based fibrosis scores. Specifically, liver stiffness measurement (LSM) by vibration-controlled transient elastography (VCTE) has recently applied and validated, serving as a modality to reflect the degree of liver fibrosis and predict HCC, portal hypertension, and varices.¹² In addition, five laboratory tests, considered non-invasive biomarkers have been established markers for fibrosis diagnosis, including 1) The aspartate aminotransferase (AST)-to-platelet ratio index (APRI)¹³ 2) The AST/ALT ratio (alanine aminotransferase) ratio; 3) the (fibrosis 4-index) FIB-4¹⁴; 4) the NAFLD fibrosis score (NFS)¹⁵; and 5) the BARD score¹⁶ where BMI, AST/ALT ratio and diabetes are included in this score. In a recent meta-analysis to determine the diagnostic accuracy of APRI (35 studies), FIB-4 (37 studies), BARD score (30 studies), NSF (41 studies), VCTE using FibroScan (19 studies), SWE (4 studies), MRE (5 studies), Xiao et al.¹⁷ reported that among non-invasive biomarkers, NFS and FIB-4 offer optimal diagnostic accuracy. Recently, two scores, called the Agile 3+-serving -serving for the diagnosis of advanced fibrosis and the Agile 4-serving for the diagnosis of cirrhosis for the diagnosis of advanced fibrosis and the Agile 4-serving for the diagnosis of cirrhosis, were proposed by combining LSM and several clinical variables, including platelet count, AST, diabetes, age, and sex).^18,19 While such non-invasive modalities have been examined in MASLD patients of Caucasian population, little efforts have been made in Asian population and none in Vietnamese population.

To fill this gap of knowledge, we determined the potential predictors of MALSD in Vietnamese population in a case-control study of 100 MASLD patients and 119 healthy controls.

Patients with MASLD were recruited at the Department of Diagnostic Imaging, Department of Surgical Oncology and Department of General Examination of the Viet-Duc University Hospital November 1, 2020, through December 31, 2024. Physical exam was performed and clinical and laboratory tests were conducted to confirm the inclusion/exclusion criteria. Inclusion criteria⁴: 1) Age 40–79 years and 2) non-alcoholic consumption; and 3) steatosis confirmed using the international criteria²⁰ (i.e., CAP score ≥ 237 (steatosis) or kPa ≥ 5.5 (fibrosis); and 3) at least one out of five cardiometabolic criteria (i.e., obesity/overweight, diabetes, hypertension or hyperlipidemia) as follow: [1] BMI > 23 kg/m² or waist-circumference > 94 cm (Male) or 80cm (Female); or [2] Diabetes diagnosis (i.e., Fasting serum glucose ≥5.6 mml./L (100mg/dL) OR 2-hour post-load glucose levels ≥ 7.8mmol/L (≥140mg/dL) OR HbA1c≥5.7% (39mmol/L) OR Type 2 diabetes OR Treatment for type 2 diabetes); or [3] hypertension or CVD diagnosis (i.e., Blood pressure ≥135/85mmHg OR Specific antihypertensive drug treatment); or [4] Plasma triglycerides ≥1.70 mmol/L (150mg/dL) OR Lipid lowering treatment; or [5] Plasma HDL-cholesterol ≤ 1.0mmol/L (40mg/dL) (Male) and ≤ 1.3mmol/L (50mg/dL) (Female) OR Lipid lowering treatment. Exclusion criteria were (1) alcoholic liver disease or consumption of alcohol > 14 drinks/week for women or > 21 drinks/week for men in the past 12 months; (2) other chronic liver diseases including autoimmune hepatitis, biliary cirrhosis, hemochromatosis and Wilson’s disease; (3) positive serological test for HBsAg, HBV DNA, antibodies to HCV or HCV RNA; (4) the presence of HCC), (5) history of liver transplantation and/or decompensated cirrhosis; (6) cancer under active treatment; (7) uncontrolled psychiatric syndrome; (8) serious co-morbidities (e.g., unstable cardiovascular disease and dialysis); and (9) use of antibiotics for ≥7 days within the past 60 days.

Healthy control subjects were recruited from the same three departments at the Viet-Duc University Hospital using the following criteria: individual who were 40–79 years of age, BMI 20–<35 kg/m², and no history of MASLD, diabetes, or any conditions listed in the Inclusion/Exclusion criteria above and those who agreed to participate in the study.

A structured questionnaire was used by a trained interviewer to collect exposure information from the study participants at baseline. The structured questionnaire included the following components: 1) sociodemographic factors; 2) lifetime tobacco and alcohol use; 3) body height and weight; 4) dietary habits; 5) medical history; 6) family history of cancer and 7) occupational exposures. Medical information from the medical records were extracted by the trained medical extractors.

The liver stiffness measurement was performed using FibroScan model 430 (EchoSens, Paris, France), following instructions and training provided by manufacturer²¹. Briefly, measurements was performed on the right lobe of the liver through intercostal spaces with the patient lying in dorsal decubitus with the right arm in maximal abduction. The tip of the probe transducer was covered with coupling gel and placed on the skin, between the ribs at the level of the right lobe of the liver. The operator, assisted by ultrasound time-motion and A-mode images provided by the system, located a portion of the liver that is at least 6cm thick and free of large vascular structures. Once the area of measurement is located, the operator pressed the probe button to begin an acquisition. The measurement depth will be between 25 and 45mm. Ten successful acquisitions were performed on each patient. The median value was represented the liver elastic modulus. Evaluation was only be performed on 10 successful acquisitions in which the reliability was considered as at least 60% success rate. The success rate was calculated as the number of successful measurements divided by the total number of measurements. The liver stiffness was expressed in kiloPascal (kPa).

We included 100 MASLD patients and 119 healthy controls in the current study. MALSD patients were significantly older than healthy control. The mean (SD) age at enrolment for MALSD cases and controls were 50.0 (15.5) years old and 41.7 (11.7) years old, respectively. The sociodemographic characteristics and selected clinical factors of study participants are presented in Table 1. Compared with healthy controls, MASLD patients were more likely to be male, had higher BMI, more likely to have diabetes and other comorbidity conditions, had lower level of AST but higher level of ALT and GGT. MALSD patients also had higher levels of triglyceride and glucose as well s higher score of FIB-4, kPA, CAP, APRI, NFS (all P’s < 0.05). There were no statistically significant differences between MASLD patients and controls with respect to platelet, INR, albumin, cholesterol and BARD score (all P’s > 0.05).

Age, BMI, diabetes status, INR, AST, GGT, kPA, CAP score, total bilirubin and APRI were predictors of the MASLD risk in the current study. Accordingly, the ORs and 95% CIs for such factors were: 1.93 (1.00-3.72) (BMI > 23 kg/m² compared with ≤ 23 kg/m²); 4.46 (1.16–17.12) (diabetes compared with no diabetes); 2.18 (1.15–4.14) (INR > 1.0 compared with ≤ 1.0); 4.18 (1.54–11.33) (AST≥40U/L compared with < 40U/L); 5.20 (2.39–11.33) (GGT≥55738U/L compared with < 55738U/L); 11.14 (1.29–96.05) (kPA > 8.2 compared with ≤ 8.2); 30.27 (10.07–90.64) (CAP score > 257.1 compared with ≤ 257.1); 1.04 (1.02–1.07) (per age incremental); 1.02 (1.00-1.05) (per unit incremental of total bilirubin); and 49.48 (3.77-649.34 (per unit incremental of APRI score) (Table 2).

In the study of 100 MASLD patients and 119 healthy controls in the Vietnamese population, we identified that while age, BMI, diabetes status, INR, AST, GGT, kPA, CAP score, total bilirubin and APRI were predictors of the MASLD risk, other important factors, such as cardiometabolic condition, platelet, ALT, albumin, cholesterol or triglyceride as well as other non-invasive fibrosis score (i.e., AST/ALT score, NFS, BARD) were not.

To our knowledge, this might be the first study to determine the predictors of MALSD in Vietnamese population. In a recent study of 101 patients with NASH in Ho Chi Minh City, Vietnam, Tran et al.²⁴ reported that Fibroscan, APRI an NFS were more accurate than FIB-4 in the diagnostic of patients with F3 or higher. Unfortunately, because there was no control group, it is not possible to evaluate the potential predictors of MASLD in such study population.

While age appeared to be an important determinant of MASLD in our study, it is equally interesting to note that male has significant higher prevalence of MALSD in this population, an observation that is consistent with prior findings,²⁵ supporting the hypothesis of endocrine role in the MASLD development (Review in Bertolotti et al.²⁶). Accordingly, in males, the prevalence of MASLD tends to increase from younger to middle-aged and to decline at the age of 50 or 60, an “inverted U-shaped” association phenomenon.²⁶ Age, on the other hand, is an important risk factor for fibrosis and poor outcomes.²⁷

We found that BMI an important predictor or determinant of MALSD in the current study. Noted that the mean and SD of BMI was 23.4 (3.43) in the MASLD patients and 22.1 (2.39) in the control subjects, which can be considered a “lean MASLD population”. A recent study showed that the pooled prevalence of MASLD in lean individuals is about 12% in East Asia, 10% in South Asia, 15.5% in Europe whereas the highest prevalence in Mexico (37.0%) and Italy (26.1%).²⁸ It is equally important to point out that about 14% MASLD patients are lean.²⁹ A recent meta-analysis, conducted by Ye and colleagues, showed that the pooled incidence of MASLD in lean individuals was 23 per 1,000 person-years.²⁸ In a study from Hong Kong, using proton-magnetic resonance spectroscopy to measure changes in liver fat fraction among 565 individuals without MASLD at baseline, Wong et al.³⁰ found that the annual incidence of MASLD of 4.3% and an increase in waist circumference, not BMI and plasma triglyceride independent factors that were associated with increased risk of MASLD in lean individuals. In our analysis, triglyceride was one of predictor of MASLD in the univariable regression model, however, the association of triglyceride-MASLD was attenuated in the multivariable regression model.

Similarly, in a recent meta-analysis of 53 studies on 65,029 subjects, including 38,084 lean MASLD and 249,544 healthy subjects, Young et al.²⁹ reported that compared to healthy subjects without MASLD, lean MASLD individuals had increased odds for central obesity

central obesity (OR = 2.39, 95% CI: 1.75–3.25), hypertension (OR = 2.13, 95% CI: 1.70–2.68), type 2 diabetes mellitus (OR = 3.78, 95% CI: 2.54–5.63), low HDL (OR = 3.09, 95% CI: 1.59–5.99), and metabolic syndrome (OR = 5.85, 95% CI: 4.01–8.63). They also found that lean MALSD patients had higher odds for impaired fasting glucose levels and insulin resistance in a subset of four studies ((OR = 3.06, 95% CI: 2.82–3.32 and OR = 3.99, 95% CI: 2.40–6.61, respectively). Genetically, odds for presence of PNPLA3 genetic polymorphism of lean MASLD patients, compared with healthy individuals, were almost 3-fold (OR = 2.69, 95% CI: 1.34–5.38). PNPLA3 is a multifunctional enzyme implicated in the regulation of lipids and retinyl ester activity,³¹ and its variant has been strongly linked to the progression of hepatic fibrosis and an increased risk of HCC.^32–35 They also found an equally important point that MASLD patients with classic phenotype, compared with lean MALSD patients, had a significantly higher odds of abdominal obesity (OR = 12.10, 95% CI: 8.44–17.35), hypertension (OR = 1.82 (1.53–2.18), type 2 diabetes mellitus (OR = 1.71, 95% CI: 1.34–2.18), impaired fasting glucose (OR = 1.31, 95% CI: 1.18–1.45), low HDL (OR = 1.23, 95% CI: 1.11–1.36), and metabolic syndrome (OR = 3.01, 95% CI: 2.17–4.17).

Another important finding from the current study is that while both kPA and CAP score, critical liver stiffness measurement (LSM) using FibroScan, independent predictors of MALSD, FIB-4 was also found in univariable analysis, and its association was diminished in the multivariable analysis. In a study of 1,040 Indian MASLD patients, De et al.³⁶ found that there was no difference in controlled attenuation parameter, FIB-4 score, LSM, FAST score or the proportion of patients in whom advanced fibrosis was ruled-out or ruled-in using FIB-4 or LSM among lean and non-lean MASLD patients. In another study among 911 participants recruited in community in Hong Kong, Wei et al.³⁷ reported that compared with obese MASLD patients, non-obese MASLD patients had similar IHTG content but lower cytokeratin-18 fragments and LSM. Another meta-analysis of histologic data showed that compared with non-lean MASLD patients, MASLD patients had lower NAS score, NASH and fibrosis scores,³⁸ which is in line with our finding in which we identified NAS score not an independent predictor of MASLD risk in the multivariable regression model.

Pathologically, despite the fact that MASLD patients have lower prevalence of insulin resistance than overweight and obese patients, they have higher prevalence of insulin resistance than healthy individuals without MASLD,²⁹ making insulin resistance a major player in the pathogenesis of lean MASLD. An important to note that not all MASLD patients have insulin resistance and multifaceted factors play role in the pathogenesis of MASLD.³⁹ Such individuals are lean but metabolically obese. Approximately 20% of individuals with normal body weight have insulin resistance, hypertension or dyslipidemia with the lower prevalence in Caucasians but higher in Asians.⁴⁰ This unique phenotype are described as “metabolically obese normal weight” or MONW, which is associated with significantly higher risk of incident type 2 diabetes mellitus, cardiovascular events and overall mortality, which turns out to be similar outcomes in overweight and obese individuals.⁴¹

Our study has several limitations. First, because this is a hospital-based case-control study design, selection bias has potentially occurred. Next, residual confounding might also occur even though we used a comprehensive set of covariates in the multivariable regression models. Finally, our results could not be generalizable because the current study was conducted in northern Vietnam.

The present study, however, also has several strengths. To our knowledge, this might be the first study in Vietnamese population that determine the potential predictors of MASLD in a sizable sample. Second, the use of comprehensive set of covariates, including clinical and socioeconomic factors, helped minimize the potential effects of confounding factors.

In summary, in the first study of Vietnamese population, we showed that age, BMI, diabetes status, INR, AST, GGT, kPA, CAP score, total bilirubin and APRI predictors MASLD risk. Findings from our current study lay solid foundation for the construction of risk stratification of MALSD in Vietnamese population. Further studies are thus warranted to replicate our findings that help investigate the underlying mechanisms and construction of risk stratification of MASLD in Asian populations.