NATAP: NAFLD Low Accuracy of FIB-4 & NAFLD Fibrosis Score for Liver Fibrosis Screenng
Inbox [email protected]
AttachmentsSat, Nov 12, 4:34 AM (2 days ago)
In the present study, we tested the ability of the 2 most used noninvasive scores, FIB-4 and NFS, for screening for elevated liver stiffness as the main predictive factor of liver fibrosis progression to decompensation and liver-related mortality.
Our large cohort study reports a significant proportion of false-negative subjects with FIB-4 or NFS, and a larger number of false-positives. A screening strategy based on either of these 2 scores would both miss patients that need referral and risk overdiagnosis and futile referrals. Waist circumference showed up as a potential useful tool to identify patients from the general population with high risk of liver fibrosis.
More than one-third of patients were misclassified by FIB-4 and NFS, worse in at-risk cohorts than the general population. Specifically, using TE ≥8 kPa, a total of 2% to 4% of all subjects had false-negative values of FIB-4/NFS and would have been missed. This proportion increased to 8% to 9% of the at-risk cohorts. In patients with diabetes, hazardous alcohol consumption, or age <35 years using FIB-4, the proportion of false-negatives was even higher. This is important, because screening tools should be especially suited for fibrosis detection in at-risk patients. Despite the high number of false-negatives, NPVs for FIB-4 and NFS were above 90% in the general population cohorts. This may give the impression of a good test performance to rule out disease. However, NPV is a misleading metric in low prevalence cohorts, because even a coin toss would have a high NPV. If the prevalence is 5.6%, as in the general population cohort presented here, flipping a coin would result in a NPV of 94.4%. Accordingly, the FIB-4 NPV of 95% or NFS NPV of 96% as seen in Supplementary Table 2 is only marginally better than a coin toss. We are therefore cautious of suggesting FIB-4 or NFS as screening tools. We instead suggest to continue searching for more sensitive tests. An optimal screening program will likely consist of 3 phases: preselection based on risk factors for chronic liver disease, followed by screening with a highly sensitive test to rule in disease, but with adequate specificity to avoid over-diagnosis. Finally, this would be followed by a confirmatory test with high specificity in those patients who screened positive.
The proportion of false-positives was also high in our study, at 28% to 29%. This was especially true for participants >65 years, where FIB-4 frequently overestimated the risk of fibrosis. Overdiagnosis is potentially harmful to healthy subjects and leads to futile use of health care resources.
An interesting finding of our study is that waist circumference was strongly associated with increased liver stiffness, and outperformed FIB-4 and NFS for detection of patients with TE ≥8 kPa in the general population, but not in at-risk cohorts. Waist circumference is easy to measure, cheap, and was recently shown to be an independent predictor of all-cause mortality in patients with ALD and NAFLD.
However, its ability as a diagnostic and prognostic marker needs validation.
We are aware that our reference standard, LSM with TE, is a surrogate for liver fibrosis and risk of liver-related events. However, outcome assessments in patients with TE ≥8 kPa require a very long follow-up period. Additionally, liver biopsy is not feasible in the screening setting and is an imperfect gold standard.
Several studies support the applicability and validity of liver stiffness as a useful surrogate for liver fibrosis in low prevalence population.
With respect to TE false-positives, it is worth noting that increased LSM due to inflammation would also raise FIB-4/NFS. Moreover, exclusion of patients with ALT above 2 × ULN or BMI >35 did not change our results. Another limitation of the present study is that we could not assess the Enhanced Liver Fibrosis Test’s role in a 2-step approach, as recently proposed.
Currently there are several projects under evaluation in Europe and the United States (LiverScreen, Renown, Seal, Scarred liver project) to establish the best strategy for fibrosis screening.
Data coming from these consortia will help define the best approach for liver fibrosis detection and personalized referral pathways.
Conclusion
In conclusion, although several studies have reported a reduction in referrals after introduction of NITs, our data show a significant percentage of false-negatives and false-positives with FIB-4 and NFS in the general population setting, even higher in the at-risk population. Therefore, although FIB-4 and NFS represent a step in the right direction, we should be cautious implementing them without first searching for better first-line screening tools and referral pathways.