Letter: FibroTest for staging fibrosis in non‐alcoholic fatty liver disease

Abstract

We read with interest the review of non-invasive methods for the diagnosis of non-alcoholic fatty liver in liver disease (NAFLD) by Festi et al.1 We agree with several of their conclusions, but not on their ranking of biomarkers’ performance, stated in the summary, results and discussion. Several methodological errors, lead to misleading conclusions concerning the biomarkers’ comparisons. First, the authors used standard area under the receiver operating-characteristics curve (AUROC) to indirectly compare the performance between biomarkers, without taking into account the spectrum effect. If the spectrum effect is not taken into account, the AUROC for bridging fibrosis (or cirrhosis) can vary significantly according to stages’ prevalence from 0.67 to 0.98 for the same biomarker.2 Second, this review is not exhaustive. Several large studies of FibroTest have not been included. In a meta-analysis of three studies (two independent of the FibroTest inventor group) in 494 obese patients, FibroTest had a highly significant performance for fibrosis staging, with the appropriate method taking into account the spectrum effect: Obuchowski measure = 0.85 (95% CI: 0.83–0.87; P < 0.0001).3, 4 Another independent study, only partially published in the cited meta-analysis (reference 41), assessed the performance of FibroTest in 242 NAFLD patients, and found similar AUROCs to those of Hepascore and FIB4 for advanced fibrosis or cirrhosis.5 Third, several fibrosis tests, such as Fibrometer, FIB4 and NAFLD fibrosis score, are at higher risk of false positive/negative as they included ALT or AST in their components. The presence of these sensitive biomarkers of the necrotico-inflammatory histological activity in a composite fibrosis biomarker systematically increases the risk of false positive prediction before treatment and the risk of overestimating the impact of treatment on fibrosis after treatment. It took more than 20 years to separate in biopsy scoring systems (such as METAVIR) the estimate of fibrosis (stage) vs. necrosis/inflammation (activity). It is hazardous to ‘regress’ in the non-invasive biomarkers discovery by mixing fibrosis markers and inflammation/necrosis markers. Fourth, biomarkers reaching more than 90% of AUROC, such as Fibrometer, are suspect of false positive results. It is mathematically impossible to obtain such performance using a 25 mm biopsy due to its own variability.6 Fifth, the concept of grey zone used by the authors for their algorithm is another artefact due to the spectrum effect and the higher variability in biopsy staging between intermediate stages.7 Due to these methodological errors, the following authors’ statement is unfair: ‘the most accurate estimating methods are FibroMeter, FIB-4 and NAFLD fibrosis score and transient elastography’. Using evidence based published data, FibroTest does not have lower performance than these biomarkers. Even more, FibroTest has the advantages of being less corrupted by necro-inflammatory activity than biomarkers including ALT, validated by more appropriate methods and with higher applicability than transient elastography.8, 9 Declaration of personal interests: Thierry Poynard is the inventor of FibroTest with a capital interest in BioPredictive, the company marketing the FibroTest. Patents belong to the public Organization Assistance Publique Hôpitaux de Paris. Mona Munteanu is a full employee of BioPredictive. Declaration of funding interests: None.