Simple Liver Blood Test Flags Heart Risk in People With Diabetes, Vietnamese Study Finds

A routine blood calculation already sitting in the medical charts of millions of people with type 2 diabetes may carry a warning that doctors rarely read: the risk of a heart attack, stroke, or heart-failure hospitalization within the next two years. In a prospective cohort study conducted at a tertiary hospital in Ho Chi Minh City, Vietnamese researchers found that higher scores on the fibrosis-4 index, or FIB-4, were associated with a substantially increased hazard of major adverse cardiovascular events among adults with type 2 diabetes who had never previously suffered a conventional cardiovascular event. The findings, published in Frontiers in Gastroenterology, add a Southeast Asian data point to a growing international literature suggesting that a liver-oriented marker may double as a cardiovascular red flag.

The FIB-4 index is deceptively simple. It combines four routinely measured variables—age, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and platelet count—into a single number originally designed to estimate the likelihood of advanced liver scarring without a biopsy. Because it requires no imaging, no specialist equipment, and essentially no additional cost, FIB-4 has become a workhorse of liver-fibrosis risk stratification in people with diabetes and metabolic liver disease. What the new study suggests, however, is that the same arithmetic may encode information about the heart and blood vessels that extends well beyond the liver.

The research team, led by Quoc Bao Vu of Pham Ngoc Thach University of Medicine together with colleagues at Nhan Dan Gia Dinh Hospital, recruited adults with type 2 diabetes consecutively from hospital outpatient clinics between June 2023 and January 2025. Of 452 patients screened, 301 consented to participate and 284 formed the baseline cohort; 280 ultimately contributed follow-up data. Each participant underwent standardized clinical, laboratory, and liver assessments at enrollment, including transient elastography with FibroScan, a vibration-based imaging technique that quantifies hepatic steatosis through the controlled attenuation parameter and liver scarring through liver stiffness measurement in kilopascals. The researchers then tracked participants for up to 24 months, contacting them roughly every three months and verifying every suspected cardiovascular event against source medical documentation reviewed by a physician in the relevant specialty.

The liver disease burden in the cohort was striking. Using the diagnostic framework of the Asian Pacific Association for the Study of the Liver, the researchers classified 72.5 percent of participants as having metabolic dysfunction-associated fatty liver disease, or MAFLD—a prevalence consistent with global meta-analyses suggesting that roughly two-thirds of adults with type 2 diabetes harbor fatty liver disease. Nearly one in four participants, 23.9 percent, met the study’s prespecified definition of significant liver fibrosis, indicated by a liver stiffness measurement of at least 7.0 kilopascals, a threshold validated against biopsy data in fatty liver disease. The participants had a mean age of 59.1 years, a median known diabetes duration of six years, and a heavy load of cardiometabolic comorbidity: 72.2 percent had hypertension, 92.3 percent had dyslipidemia, and more than half had LDL cholesterol above 2.6 millimoles per liter.

Over 544.92 person-years of follow-up, 11 participants experienced an expanded major adverse cardiovascular event—the study’s principal endpoint—corresponding to an incidence rate of 20.19 per 1,000 person-years. The expanded definition encompassed cardiovascular death, non-fatal myocardial infarction, non-fatal ischemic or hemorrhagic stroke, hospitalization for heart failure, transient ischemic attack, hospitalization for unstable angina, peripheral arterial events, and coronary or peripheral revascularization. A narrower conventional MACE endpoint, limited to cardiovascular death, non-fatal heart attack, and non-fatal stroke, occurred in five participants, an incidence of 9.18 per 1,000 person-years. Because the event counts were so small, the team turned to Firth-penalized Cox regression, a statistical technique that reduces the small-sample bias that can distort conventional survival models when events are rare.

The central result was consistent across every analytical variation the researchers attempted. In the main model adjusted for age and MAFLD, each increase of one interquartile range in FIB-4—0.696 units, roughly the gap between the 25th and 75th percentiles of the cohort—was associated with a 56 percent higher hazard of expanded MACE (adjusted hazard ratio 1.56, 95 percent confidence interval 1.09–1.99; p = 0.022). When age was deliberately omitted as a separate covariate, the estimate strengthened slightly to 1.68. Conventional Cox regression produced a similar figure. Within the MAFLD subgroup alone, the association persisted at a hazard ratio of 1.63. For the stricter conventional MACE endpoint, the FIB-4 association was even more pronounced, at 1.81 per interquartile-range increase, though only five events underpinned that estimate.

The age question demanded careful handling, because age is embedded directly in the FIB-4 formula. If the apparent cardiovascular signal were merely age in disguise, adjusting for age should have erased it. It did not. The variance inflation factor for age and FIB-4 was a modest 1.20, penalized likelihood-ratio tests confirmed that each variable contributed incremental information beyond the other, and FIB-4 remained significant even when age was dropped from the model entirely. Still, the authors are careful about interpretation: the analyses make it unlikely that the association is a pure statistical artifact of double-counting age, but they cannot establish FIB-4 as an age-independent cardiovascular biomarker, because age remains biologically woven into the index.

Perhaps the most intriguing negative result concerned direct liver measurements. Neither liver stiffness as a continuous variable nor the dichotomous threshold of significant fibrosis at 7.0 kilopascals showed a statistically significant association with cardiovascular events, in contrast to the Rio de Janeiro Cohort Study, which reported that greater liver stiffness—particularly above 9.6 kilopascals—predicted cardiovascular outcomes in patients with type 2 diabetes and fatty liver disease. The discrepancy may reflect the lower, earlier-stage fibrosis cutoff used in the Vietnamese study, the short follow-up, or simply the paucity of events. It also hints at something conceptually important: FIB-4 may be capturing systemic cardiovascular risk information—through platelet count, through AST that reflects not only liver but also skeletal muscle and frailty—that a purely mechanical liver stiffness measurement does not.

MAFLD itself told a murkier story. Ten of the eleven expanded events occurred among participants with MAFLD, and the adjusted point estimate suggested roughly a 3.5-fold higher hazard, but the confidence interval stretched from 0.79 to 33.09—an honest reflection of the fact that only one event occurred in the non-MAFLD comparison group. Kaplan-Meier curves visually favored worse event-free survival with MAFLD, but the log-rank test did not reach significance. The authors conclude that the study can neither establish nor exclude an independent cardiovascular effect of MAFLD, a caveat that matters given the high prevalence of the condition in this population and the ongoing international debate over how metabolic fatty liver disease should be classified and risk-stratified.

The researchers supplemented their primary models with Bayesian model averaging, an exploratory technique that weighs hundreds of competing predictor subsets rather than committing to a single regression. Across 99 retained models in the FIB-4 model space, FIB-4 achieved a posterior inclusion probability of 71.5 percent, while liver-stiffness-defined fibrosis managed only 19.9 percent in its own model space—suggestive, though not definitive, evidence that FIB-4 carries the more stable cardiovascular signal. Proportional-hazards diagnostics flagged possible non-proportionality for FIB-4, and exploratory time-varying analyses hinted that the association might strengthen after the first year of follow-up, though the formal piecewise tests were not statistically significant.

What should clinicians take away? The authors position FIB-4 not as a standalone cardiovascular prediction tool but as a cheap, already-computed flag that might prompt closer review of established cardiovascular risk factors in patients with type 2 diabetes. The categorical low, intermediate, and high FIB-4 thresholds used in liver practice produced an ordered gradient of event rates—2.6, 5.1, and 20.0 percent expanded MACE across the three categories—but with only ten participants in the high-risk stratum, these analyses remain exploratory and no cardiovascular FIB-4 threshold can yet be defined. The study’s limitations are candidly acknowledged: a single center, eleven expanded and five conventional events, sensitivity of estimates to the deletion of individual events, possible selection bias from the 151 screened patients who declined participation, and residual confounding from the full spectrum of cardiovascular determinants that the small event count made impossible to adjust for simultaneously.

Even so, the study fills a genuine gap. Prospective Vietnamese data linking liver markers to observed cardiovascular events have only begun to emerge, and this is, to the authors’ knowledge, the first Vietnamese cohort to combine baseline transient elastography with prospectively recorded cardiovascular outcomes in adults with type 2 diabetes. The consistency of the FIB-4 signal with large international cohorts—including a study of more than 44,000 individuals with obesity or diabetes and recent prospective evidence linking FIB-4 to cardiac mortality in coronary heart disease—lends biological plausibility to the finding. Whether FIB-4 ultimately earns a place in cardiovascular risk prediction alongside cholesterol, blood pressure, and glycemic control will depend on larger multicenter cohorts with enough events to test the hypothesis this Vietnamese study has now put squarely on the table.

Subject of Research: Association between the fibrosis-4 index and cardiovascular events in Vietnamese adults with type 2 diabetes

Article Title: Higher fibrosis-4 index is associated with 2-year cardiovascular events in Vietnamese adults with type 2 diabetes mellitus: a prospective cohort study

Article References: Vu, Q. B., Vo, H. M. C., Tran, X. L., Nguyen, T. V., & Tran, T. K. T. (2026). Higher fibrosis-4 index is associated with 2-year cardiovascular events in Vietnamese adults with type 2 diabetes mellitus: a prospective cohort study. Frontiers in Gastroenterology, 5, Article 1973311. https://doi.org/10.3389/fgstr.2026.1973311

Image Credits: AI Generated

DOI: 10.3389/fgstr.2026.1973311

Keywords: FIB-4, type 2 diabetes, cardiovascular events, MAFLD, liver fibrosis, transient elastography, FibroScan, MACE, Vietnam, prospective cohort study, hepatic steatosis, cardiovascular risk