We sincerely appreciate the interest of Wang et al. in our study [1,2]. Type 2 diabetes mellitus (T2DM) is a critical risk factor in the development and progression of metabolic dysfunction-associated steatotic liver disease (MASLD). The co-incidence rate of T2DM and MASLD is very high across all regions globally [3]. All MASLD are candidates for hepatic fibrosis evaluation, and those with coexisting T2DM are classified as a high-risk group for which hepatic fibrosis evaluation is essential [4]. However, concerns have been raised regarding the low diagnostic performance of fibrosis-4 index (FIB-4) in assessing hepatic fibrosis in T2DM for various reasons. To address these concerns, researchers have extensively investigated the diagnostic performance of FIB-4 for advanced hepatic fibrosis in T2DM patients. Our meta-analysis data showed moderate sensitivity and specificity, with an overall diagnostic accuracy represented by the hierarchical summary receiver operating characteristic curve (HSROC) value at a low cutoff. The sensitivity, specificity, and HSROC were 0.74, 0.62, and 0.75, respectively [2].

The concerns raised by Wang et al. regarding our metaanalysis are both reasonable and appropriate [1]. First, our study included both cross-sectional and cohort studies without distinguishing between them. Generally, study design can influence the level of evidence, and therefore, in meta-analyses, studies are usually analyzed separately based on their design. In our case, the primary endpoint of our study was to evaluate the diagnostic performance of FIB-4. We did not believe that the diagnostic performance of FIB-4 would influence the level of evidence based on the study design. We used only baseline biopsy data from cohort studies. Therefore, while cross-sectional and cohort studies were analyzed together without distinction, we believe this methodological choice has a minimal likelihood of introducing bias into the study results. Second, recent data have indicated that MASLD or advanced hepatic fibrosis affects long-term clinical hard outcomes such as renal disease, cardiovascular disease, and extrahepatic malignancies. However, evidence that FIB-4’s diagnostic performance is diminished in individuals with renal or cardiovascular disease remains inconclusive. The predictive ability of FIB-4 for cardiovascular and renal disease incidence and the notion that its diagnostic performance is reduced in these conditions are distinct concepts. While many studies have shown that FIB-4 has predictive power for cardiovascular mortality, there is limited evidence to suggest that FIB-4’s diagnostic performance is diminished in patients with cardiovascular disease. While some studies in our meta-analysis included participants with impaired renal function, we believe that renal dysfunction itself does not significantly impact FIB-4’s diagnostic performance. Furthermore, our sensitivity analysis confirmed that renal function did not influence the diagnostic performance of FIB-4. Third, there is a possibility of overlapping patient cohorts in three studies conducted by the same researcher (references 13, 15, 18), which is a reasonable concern [5-7]. Such overlap could introduce bias into the results. However, among the 12 studies included in the meta-analysis (totaling 5,624 participants), the potential overlap involves three studies comprising 331 participants, accounting for approximately 5.9% of the total study population. While it is clear that overlapping participants can introduce bias, we believe the impact of this overlap on the overall findings is limited.

The most important variable influencing the diagnostic performance of FIB-4 in T2DM is age [8]. FIB-4 was designed for individuals aged 35–65 years, and it is known that FIB4 tends to underestimate fibrosis in individuals under 35 years and overestimate it in those over 65 years. The reduced diagnostic performance of FIB-4 in T2DM is largely attributable to the higher average age of T2DM populations. When researchers analyzed 517 biopsy-proven T2DM patients with MASLD, the sensitivity and specificity of FIB-4 were highest at age 45, while the accuracy dropped below 60% after age 55. However, for T2DM patients aged 35–55, the diagnostic accuracy of FIB-4 was comparable to that of non-T2DM patients (Fig. 1). In contrast, for patients over 65 years, the diagnostic accuracy of FIB-4 in T2DM with MASLD dropped sharply to below 50%. In our meta-analysis, the relatively favorable diagnostic performance of FIB-4 in T2DM with MASLD may be associated with the relatively younger mean age (56.1 years) of the 5,624 T2DM patients with MASLD included in the 12 studies. The second key variable influencing the diagnostic performance of FIB-4 is the prevalence of advanced hepatic fibrosis. Generally, the diagnostic performance of FIB-4 is higher in hospital cohorts with a higher prevalence of advanced hepatic fibrosis but lower in general population or community cohorts. Among the 12 studies in our meta-analysis, the prevalence of advanced hepatic fibrosis in the 5,624 T2DM patients with MASLD was relatively high at 34.1%.

Considering the clinical characteristics of FIB-4, its diagnostic performance for detecting advanced hepatic fibrosis in T2DM patients with MASLD is thought to be at a moderate level. In the use of FIB-4 for MASLD with T2DM, the average age of the target population plays a crucial role. Therefore, the diagnostic performance of FIB-4 in MASLD with T2DM should be applied with careful consideration of the patient’s age.