Adjusted NIT Thresholds Needed for MASLD Fibrosis Detection in Clinical Practice

Yasaman Vali, MD, PhD

Credit: Amsterdam UMC

New research is shedding light on the diagnostic performance of select noninvasive tests (NITs) for detecting advanced fibrosis in metabolic dysfunction-associated steatotic liver disease (MASLD), highlighting the need for adjusted thresholds to adjust for different patient characteristics.¹

Leveraging patient data from the Liver Investigation: Testing Marker Utility in Steatohepatitis (LITMUS) meta cohort of the European NAFLD Registry, the study found that although the performance of Fibrosis-4 Index (FIB-4), enhanced liver fibrosis (ELF), and vibration-controlled transient elastography liver stiffness measurement (VCTE LSM) did not vary substantially by age, sex, BMI and presence of type 2 diabetes, different thresholds were needed to achieve the same sensitivity, specificity, and predictive values across clinical subgroups.¹

Liver biopsy is the current reference standard for detecting metabolic dysfunction-associated steatohepatitis (MASH) and staging liver fibrosis in patients with MASLD, which is estimated to affect more than 100 million US adults. However, the invasive, resource-intensive, and risky nature of this procedure has led to the development and clinical use of NITs.^1,2

“Despite their mention in guidelines, variations in the reported performance of these NITs raise concerns about their reliability and potential impact on accurate diagnosis and effective disease management,” Yasaman Vali, MD, PhD, a postdoctoral researcher at Amsterdam UMC, and colleagues wrote.¹

To assess variations in the performance of 3 NITs for detecting advanced fibrosis in MASLD based on patient characteristics, investigators examined data from a sample of adult LITMUS meta-cohort participants with liver biopsy and serum samples. Those with excessive alcohol consumption (> 20–30 g/day), other chronic liver diseases such as viral hepatitis B or C, and incomplete data for analysis were excluded.¹

Investigators evaluated the diagnostic performance of the 3 following NITs:

• ELF, with scores based on hyaluronic acid, tissue inhibitor of matrix metalloproteinase-1, and aminoterminal propeptide of procollagen type III, measured in the Central Laboratory using the Siemens Advia Centaur device.
• FIB-4 index, with scores calculated using the available clinical laboratory data based on the following formula: age × AST (IU/L)/platelet count (×109/L) × √ALT (IU/L).
• Liver stiffness, measured locally by VCTE (FibroScan)

Investigators assessed the diagnostic performance of the three NITs in detecting advanced fibrosis (F ≥ 3) across different subgroups based on sex; BMI; age; diabetes status; AST; and ALT.¹

Of 966 meta-cohort participants, 586 had data for all relevant clinical variables and all 3 NIT results available. Among these patients, the mean age was 52 years and 64% were male.¹

Investigators noted the 3 tests performed differently when evaluated in the 586 included participants: ELF had an area under the receiver operating characteristics curve (AUC) of 0.79 (95% CI, 0.75–0.84); FIB-4 an AUC of 0.74 (95% CI, 0.69–0.78), and VCTE LSM had an AUC of 0.83 (95% CI, 0.79–0.86).¹

However, the diagnostic performance of the 3 tests for detecting F ≥ 3 was not significantly different between the respective subgroups (P >.05). Investigators pointed out that both male and female AUCs were around 0.80 for ELF, 0.73 for FIB-4 and 0.83 for VCTE LSM. Despite slight differences in the AUC, the performance of all 3 tests was comparable in different age and BMI subgroups. Additionally, no significant difference was observed in the performance of the tests in subgroups of patients with and without diabetes and in those with different levels of liver enzymes.¹

Investigators noted different thresholds were required to achieve the same level of accuracy with each test. Results showed slightly increased thresholds of ELF and FIB-4 were required to achieve a sensitivity of 90% in females (8.72 and 0.95 vs 8.59 and 0.83, respectively), while for VCTE LSM, a higher threshold was observed in males (7.65 kPa vs 6.75 kPa). To reach a 90% specificity the threshold of VCTE LSM had to be higher in males (VCTE LSM: 14.05 kPa vs. 13.95 kPa). The calculated ELF threshold was higher in females (10.38 vs. 9.95).¹

Investigators also observed higher thresholds of ELF and FIB-4 in higher age ranges while pre-specifying the sensitivity level at 90%. Pre-defining specificity at 90% resulted in higher thresholds of all 3 tests in higher age ranges.¹

When pre-specifying a 90% sensitivity, investigators observed greater thresholds for all 3 tests in patients with BMI ≥ 30 compared with those with BMI 25–30 and BMI < 25 kg/m2. For 90% specificity, the same trend was observed for ELF and VCTE LSM.¹

In patients with type 2 diabetes, investigators observed higher thresholds for reaching 90% sensitivity (ELF: 8.76) and 90% specificity (VCTE LSM: 16.40 kPa) compared to non-type 2 diabetes patients (ELF: 7.03, VCTE LSM: 12.95 kPa).¹

“This highlights the fact that the accuracy of NITs can vary in specific demographic groups and can be optimised by using adjusted thresholds,” investigators concluded.¹ “Clinicians should keep this in mind when using NITs in clinical practice, where different positivity thresholds are required, depending on basic patient characteristics.”

References

1. ^{Vali Y, van Dijk AM, Lee J, et al. Precision in Liver Diagnosis: Varied Accuracy Across Subgroups and the Need for Variable Thresholds in Diagnosis of MASLD. Liver International. https://doi.org/10.1111/liv.16240}
2. ^{American College of Gastroenterology. Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD). Accessed January 24, 2025. https://gi.org/topics/steatotic-liver-disease-masld/}