As shown in Table 1, several studies have evaluated the performance of FIB-4 thresholds relative to elastography-based thresholds for clinically significant liver fibrosis; typically, 8.0 kPa, equivalent to F2 liver fibrosis by liver histology, as outlined in the EASL-EASD-EASO screening strategy. The sensitivity of FIB-4 thresholds (i.e. ≥ 1.30 or ≥ 2.00 in those ≥ 65 years) used to identify clinically significant liver fibrosis ranged from 31.7% to 72.4% in all studies^[16-22]. This corresponds to false negative rates (FNR; 1-sensitivity) of between 27.6% and 68.3%, indicating that these FIB-4 thresholds may lead to a substantial proportion of patients with clinically significant liver fibrosis being misclassified as low risk. Importantly, this pattern is observed across heterogeneous cohorts of patients, including those recruited from primary care referral pathways, diabetes clinic populations, and specialist settings, and importantly persists in populations of patients perceived to be at higher risk of liver fibrosis with T2DM. When considering only those studies specifically involving populations of patients living with T2DM^[18-20,22], the sensitivity of FIB-4 thresholds ranges from 42.6% to 67.2% [Table 1]. It is important to highlight that these findings reflect the performance of FIB-4 in patient cohorts with T2DM from a range of settings, including diabetes specialities centres^[18,20], outpatient services and liver units^[19], and primary care practices^[22]. Whilst these findings indicate that FIB-4 is likely to have poor sensitivity and thereby a high false-negative rate leading to people not being referred on for VCTE throughout these patient cohorts, further studies are needed specifically in community-based cohorts of patients living with T2DM who are not known to have liver disease.

Although FIB-4 is frequently described as having a high negative predictive value (NPV), this assumption does not appear to be supported in populations with T2DM. As shown in Table 1, NPVs vary considerably across studies, particularly in cohorts of patients with T2DM, where NPVs range from 62.4% to 87.6%. This variance is likely to have profound implications for clinical decision-making. In a general population where the NPV exceeds 90%-95%, a low FIB-4 score (< 1.30) provides clinicians with adequate reassurance to “rule out” the presence of clinically significant liver disease. However, when the NPV falls to the lower ranges observed in T2DM-specific cohorts - reaching as low as 62.4% in primary care settings - the test loses its utility as an effective gatekeeper. Clinically, an NPV of 62.4% indicates that nearly 38% of patients with T2DM who are triaged as “low risk” have clinically significant liver fibrosis. Hence, we argue that instead of functioning as a reliable clinical filter, FIB-4 likely introduces an unacceptable risk of diagnostic delay to a substantial proportion of an already high-risk population of patients. Given that a test’s NPV is inherently dependent on disease prevalence, its utility as a “rule-out” marker is limited in higher-risk populations, including those with T2DM where the prevalence of liver fibrosis is higher; where even modest reductions in sensitivity translate into substantial numbers of missed cases [Figure 1].

It is important to acknowledge the biological and methodological factors which detrimentally impact the performance of FIB-4 in patients with T2DM. While the FIB-4 score heavily weighs elevations in transaminase (AST and ALT) as markers of hepatic dysfunction, patients with T2DM and MASLD frequently present with normal liver enzyme concentrations, often even in the presence of liver fibrosis^[23]. Similarly, peripheral platelet counts are frequently preserved in early-to-intermediate fibrosis stages in patient populations with metabolic derangements, as portal hypertension typically develops at more advanced disease stages.

The FIB-4 index also suffers from methodological limitations regarding age and disease prevalence. Since age is a direct multiplier in the numerator, older patients (≥ 65 years) naturally default to higher FIB-4 scores, leading to high-false positive rates that necessitate age-adjusted cutoffs (e.g., ≥ 2.00). Conversely, in younger or middle-aged patients with T2DM, their younger age suppresses the final score, resulting in a higher false-negative classification. Additionally, the high pre-test probability and prevalence of clinically significant liver fibrosis within patient populations with T2DM degrade the negative predictive value of screening tools that possess sub-optimal clinical sensitivity. Collectively, these biological and methodological factors restrict the ability of FIB-4 to perform as an effective first-tier liver fibrosis screening tool in patients with T2DM.

To mitigate false negatives, guidelines recommend repeating FIB-4 testing every 1-3 years (depending on the presence of metabolic risk factors) in patients with a FIB-4 of < 1.30^[4-8]. When evaluating these guidelines, it is crucial to distinguish between data exploring the prognostic value and evidence of screening reclassification utility. A substantial body of longitudinal evidence robustly establishes that the prognostic value of tracking FIB-4 scores over time; changes are positively associated with fibrosis progression, increased liver stiffness, and adverse clinical outcomes in patients with MASLD^[24-28]. However, demonstrating that a rising score correlates with disease progression in an established cohort is fundamentally different from proving that serial testing functions as an effective clinical safety net within a screening pathway [Figure 1].

Most longitudinal studies to date include heterogeneous populations and have not specifically studied patients with T2DM from primary care settings who may have normal liver enzyme concentrations and are most at risk of false-negative classification because they have a low FIB-4 score. This uncertainty is reinforced by a large longitudinal study, involving ~40,000 participants with repeated FIB-4 testing (within 5 years) found that whilst repeated FIB-4 testing improved the identification of individuals at an increased risk of severe liver disease, almost half of all events occurred in those consistently in the low-risk group (i.e. in those with a FIB-4 of < 1.30)^[28]. This paradox underscores the fundamental disconnect between a test’s longitudinal prognostic tracking value and its utility as a clinical screening “safety net”. That said, it should be noted that further research is urgently needed to directly explore the reclassification utility of serial FIB-4 testing specifically within T2DM screening pathways.

In addition, interpretation of longitudinal increases in FIB-4 is complicated by the influence of age, which is both a component of FIB-4 and a strong independent predictor of adverse clinical outcomes. While statistical adjustment may partially account for the impact of age, residual confounding is difficult to exclude when evaluating change over time. Taken together, although serial FIB-4 measurements clearly provide prognostic information, their role as an effective safety net for false-negative cases, especially in patients with T2DM, in screening pathways remains unclear and warrants further attention.

The limitations highlighted above raise important concerns regarding the use of FIB-4 as a first-line test in sequential screening pathways, especially in patients with T2DM. While lowering FIB-4 thresholds may lead to improvements in sensitivity and subsequently, a reduction in false negatives, this would occur at the expense of a significant increase in the proportion of individuals referred for second-line testing. Conversely (as mentioned above), maintaining current thresholds seemingly results in high false-negative rates, which are likely exacerbated in patients with T2DM [Table 1], resulting in a large proportion of patients with clinically significant liver fibrosis being missed and not referred for further assessment [Figure 1]. Alternative strategies, such as referring all patients with T2DM directly to VCTE, would maximise diagnostic sensitivity but would introduce substantial logistical trade-offs. While cost-effective in some long-term economic models due to early intervention benefits^[15,29], a direct-to-VCTE pathway would require substantial up-front investment, extensive staff training, and would likely create immediate bottlenecks in primary and secondary care settings.

To address the performance gap of FIB-4 in patients with T2DM, several non-invasive biomarker alternatives warrant further evaluation alongside the VCTE imaging pathway discussed above. Scoring systems specifically optimised or modified for patients with T2DM or metabolic dysfunction, such as metabolic dysfunction-associated fibrosis 5 (MAF-5) score and the modified FIB-4^[30,31], appear to improve diagnostic sensitivity in patients with hepatic steatosis and T2DM. Alternatively, evidence indicates that implementing the ELF test as a gatekeeper test offers superior diagnostic accuracy for early-stage liver fibrosis (≥ F2) compared to FIB-4^[32]. Importantly, given that the ELF test relies on direct serum markers of extracellular matrix remodelling, it retains high diagnostic performance even when transaminase concentrations are normal - a common observation in patients living with T2DM. However, the widespread adoption of first-line ELF testing is largely restricted by high recurring laboratory costs and lack of universal availability in routine primary care^[14,29].