Simple Blood-Based Scores Outperform Ultrasound Expectations in Spotting Fatty Liver Among Diabetes Patients

Researchers in Sri Lanka have shown that a handful of routine blood tests and a measuring tape can do much of the work of an ultrasound scanner when it comes to detecting fatty liver disease in people with type 2 diabetes. In a study published in Frontiers in Clinical Diabetes and Healthcare, a team from General Sir John Kotelawala Defence University and the Colombo South Teaching Hospital evaluated five non-invasive liver fat indices in 260 newly diagnosed diabetes patients and found that two of them, the Fatty Liver Index and the Hepatic Steatosis Index, performed with a diagnostic precision that could reshape screening practice in clinics where advanced imaging is scarce.

The condition at the center of the study is metabolic dysfunction-associated steatotic liver disease, or MASLD, the new name for what was long called non-alcoholic fatty liver disease. It is defined by the accumulation of fat in more than 5 to 10 percent of liver weight, accompanied by at least one cardiometabolic risk factor and minimal alcohol consumption. MASLD is considered the hepatic face of metabolic syndrome, and it shares its core engine, insulin resistance, with type 2 diabetes. That overlap is not trivial: co-occurrence of the two conditions exceeds 70 percent, meaning that most patients diagnosed with diabetes carry a liver quietly accumulating fat that may progress toward inflammation, fibrosis, and cirrhosis if left unchecked.

Diagnosing fatty liver has always been a logistical problem. Liver biopsy remains the gold standard, but it is invasive, costly, and carries real procedural risk, so it is reserved for ambiguous cases rather than population screening. Ultrasonography is the workhorse alternative, reading the brightness of liver parenchyma on a scan, but it requires equipment, a trained radiologist, and patient access to imaging suites. Newer tools such as transient elastography, marketed as FibroScan, quantify fat accumulation and liver stiffness with higher sensitivity and specificity than ultrasound, yet they remain expensive and confined to specialized centers in many low-resource health systems. Magnetic resonance imaging and computed tomography face similar constraints. The gap between what guidelines recommend and what clinics can deliver is precisely where the Sri Lankan team aimed its study.

The researchers recruited 260 newly diagnosed type 2 diabetes patients registered at the diabetic and endocrinology clinic of Colombo South Teaching Hospital between January 2022 and January 2024. Each participant provided fasting blood samples after a 10 to 12 hour fast, underwent anthropometric measurements including weight, height, waist circumference, and body mass index, and received an ultrasound scan performed by a consultant radiologist using a Toshiba APLIO 500 system. The presence and grade of fatty liver were determined from hepatic echogenicity. Biochemical parameters measured on a semi-automated BTS-350 analyzer included alanine aminotransferase, aspartate aminotransferase, gamma-glutamyl transferase, serum triglycerides, high-density lipoprotein cholesterol, and fasting blood sugar, while fasting serum insulin was processed at a referral laboratory on a Roche Cobas E411. From these inputs the team calculated five standardized indices: the Fatty Liver Index, the Hepatic Steatosis Index, the Visceral Adiposity Index, the Triglyceride and Glucose Index, and the NAFLD-Liver Fat Score.

The headline finding was stark. Ultrasound diagnosed fatty liver in 69.2 percent of the cohort, or 180 of 260 patients, with 23.4 percent showing grade 1 steatosis and 45.8 percent grade 2 or above. The burden was distributed evenly across sexes, with 69.9 percent of women and 67.8 percent of men affected, and concentrated in middle age, with half of all fatty liver diagnoses falling in the 41 to 55 year bracket. Notably, most patients had normal liver enzymes despite their fatty liver: 62.8 percent of those with ultrasound-confirmed steatosis had normal aspartate aminotransferase, 56.1 percent had normal alanine aminotransferase, and 63.9 percent had normal gamma-glutamyl transferase. That pattern matters clinically, because patients are often referred for imaging only after enzymes rise, a strategy that misses the majority of cases.

When the five indices were tested against the ultrasound findings, three emerged as strong correlates. The Fatty Liver Index, the Hepatic Steatosis Index, and the NAFLD-Liver Fat Score all reached statistical significance at p < 0.001, while the Visceral Adiposity Index and the Triglyceride and Glucose Index showed moderate significance at p < 0.01. Effect size analysis told a similar story: the Fatty Liver Index explained 40.7 percent of the variance in ultrasound diagnosis, the largest of any index, followed by the Hepatic Steatosis Index at 23.0 percent and the NAFLD-Liver Fat Score at 21.9 percent. Post-hoc comparisons across fatty liver grades confirmed that these three indices discriminated not just presence from absence but also severity, with the Fatty Liver Index showing the widest mean difference between patients with no fatty liver and those with grade 2 disease or above.

Receiver operating characteristic analysis, the standard method for quantifying how well a test separates true positives from false ones, delivered the study’s most striking numbers. The Fatty Liver Index achieved an area under the curve of 0.947, with a 95 percent confidence interval of 0.923 to 0.971, at an optimal cutoff of 46.12. At that threshold it reached 75 percent sensitivity and 100 percent specificity, meaning that in this cohort no patient below the cutoff had ultrasound-confirmed fatty liver. The Hepatic Steatosis Index and the NAFLD-Liver Fat Score tied at an area under the curve of 0.847, with the Hepatic Steatosis Index showing 68.3 percent sensitivity and 95.5 percent specificity at a cutoff of 37.02, and the NAFLD-Liver Fat Score showing 76.1 percent sensitivity and 85 percent specificity at a cutoff of 0.906. The Triglyceride and Glucose Index and the Visceral Adiposity Index trailed with areas under the curve of 0.707 and 0.670 respectively, placing them in fair rather than strong diagnostic territory.

Gender-stratified analysis added a further layer. Among women, the Fatty Liver Index reached an area under the curve of 0.961 and the Hepatic Steatosis Index 0.892, both exceeding their performance in the full cohort, while the NAFLD-Liver Fat Score also strengthened to 0.898. Among men, only the Fatty Liver Index and the Hepatic Steatosis Index retained strong classification performance. The authors note that this pattern aligns with prior work suggesting sex-specific differences in how visceral fat and lipid profiles translate into hepatic steatosis, and it hints that cutoff values may eventually need sex-specific calibration.

To test whether the indices could stand on their own rather than merely correlating with ultrasound, the team ran multivariable logistic regression, adjusting for age, sex, body mass index, waist circumference, fasting blood sugar, high-density lipoprotein, triglycerides, and the three liver enzymes, while excluding variables embedded in each index’s own formula to prevent multicollinearity. Variance inflation factors confirmed no collinearity problems. Two indices survived: the Fatty Liver Index, with an adjusted odds ratio of 1.144 per unit increase, and the Hepatic Steatosis Index, with an adjusted odds ratio of 1.366, both significant at p ≤ 0.001. The remaining three indices lost independent predictive ability once adjusted, with the NAFLD-Liver Fat Score falling just short of significance at p = 0.064. In practical terms, the Fatty Liver Index and the Hepatic Steatosis Index carry information about fatty liver that is not redundant with standard metabolic variables.

The implications reach well beyond Sri Lanka. Both winning indices rely on measurements that are already collected in virtually every diabetes clinic: waist circumference, body mass index, triglycerides, high-density lipoprotein cholesterol, and gamma-glutamyl transferase for the Fatty Liver Index, and body mass index, waist circumference, and the AST-to-ALT ratio for the Hepatic Steatosis Index. No new equipment, no additional patient visits, no specialist referral. The authors argue that these scores could serve as first-line screening tools in resource-limited settings, flagging vulnerable patients early for lifestyle intervention and closer follow-up while reserving ultrasound and elastography for those who screen positive. The study has limitations the team acknowledges openly: alcohol consumption was self-reported, fasting conditions could not be fully standardized, ultrasound rather than biopsy served as the reference standard, and the cohort of 260 patients from a single clinic will need validation in larger and more diverse populations. Future work, they suggest, should validate the derived cutoffs against FibroScan and expand the sample size to build more comprehensive predictive models. Still, the core message stands: in the quiet arithmetic of a routine blood panel lies a screening strategy that could catch fatty liver in the majority of diabetes patients before it ever reaches an imaging suite.

Subject of Research: Non-invasive liver fat indices for predicting MASLD in type 2 diabetes patients

Article Title: Prediction of metabolic dysfunction-associated steatotic liver disease among patients with type 2 diabetes mellitus using liver fat indices in a resource-limited clinical setting

Article References: Prediction of metabolic dysfunction-associated steatotic liver disease among patients with type 2 diabetes mellitus using liver fat indices in a resource-limited clinical setting. (n.d.). https://doi.org/10.3389/fcdhc.2026.1916533

Image Credits: AI Generated

DOI: 10.3389/fcdhc.2026.1916533

Keywords: MASLD, fatty liver, type 2 diabetes, liver fat indices, Fatty Liver Index, Hepatic Steatosis Index, ultrasonography, ROC analysis, insulin resistance, screening, resource-limited settings, Sri Lanka