This meta-analysis demonstrated that the diagnostic value of FT was similar in the four most frequent chronic liver diseases. This meta-analysis also demonstrated that the diagnostic value of FT, as for liver biopsy, was similar between all the adjacent fibrosis stages but without a specific "gray zone" or "inaccurate zone" between intermediate stages. FT, like biopsy, has lower diagnostic value to discriminate between two adjacent stages than between two extreme stages .
The advantages of the present study are the large number of studies included, as well as the opportunity to analyze an integrated database, which included the individual characteristics of 3,282 (51%) patients out of 6,378 patients included in the published studies. This permitted to better take into-account the variability factors associated with FT diagnostic value.
Comparison of FT diagnostic values between different chronic liver diseases
One limitation of the study is that the number of studies and patients in non HCV related diseases is smaller than those in HCV. However we analyzed a total of eleven studies including 2,877 non HCV or mixed causes, and 851 non-HCV patients in the integrated data base. Another limitation was that there were few independent studies in other chronic liver diseases (1 for HBV, 1 for NAFLD and none for ALD). However two studies in HBV [4, 34] and two studies in ALD [5, 35] were mixed and three independent studies included HBV and ALD in their analyses [6, 7, 38] with same results than in non-independent studies (Table 1).
To compare as well as possible the FT diagnostic value according to liver diseases, we used the standardization of the AUROCs, and sensitivity analysis in both the meta-analysis and the integrated data base with individual data.
We applied the standardization of the observed AUROCs according to the spectrum of fibrosis stages among advanced and non advanced fibrosis. We recently demonstrated that this standardization is mandatory for any interpretation of AUROCs estimating the diagnostic value of a fibrosis marker . For instance, this method allowed an adjustment to be made in the ObAUROCs of FT according to the cause of liver disease, which had significant difference in fibrosis stage spectrum. The significant difference observed between ALD and HCV ObAUROCs disappeared after adjustment (Table 2). In HBV studies patients had lower DANA than in studies of ALD patients. After standardization, the difference between AUROCs was reduced by two (0.77 versus 0.88 before and 0.80 versus 0.86 after standardization) (Table 2).
These data are also in accordance with the similarities of advanced fibrosis stages among chronic hepatitis C and B, NAFLD and ALD. Despite differences in the dynamics of fibrosis progression  and the initial fibrosis stages, the bridging stages are very similar including cirrhosis and were estimated in the same way by the METAVIR scoring system for advanced fibrosis [40, 41]. Fibrosis stages and pathogenetic mechanisms are very similar in NAFLD and ALD . Repeated FT improved similar to fibrosis as estimated by repeated biopsies during treatment for HCV [22, 23], HBV [4, 34] and NAFLD . The components of the FT had similar modifications according to fibrosis stages for these four chronic liver diseases [1, 3, 5, 8].
The sensitivity analyses did not reveal any significant differences between AdAUROCs according to all the other characteristics analyzed (Table 3 and Table 4). Significant differences or the absence of differences between ObAUROCs could be due to confounding factors. A demonstrative illustration is the artificially higher ObAUROCs for fragmented versus non-fragmented biopsies. Because of a higher prevalence of cirrhosis in patients with fragmented biopsies, the DANA was higher than in patients with non-fragmented biopsies . This difference was no longer seen after adjustment [Table 4).
Comparison of FT diagnostic values between adjacent stages
There is still a controversy among experts concerning the FT diagnostic value for "intermediate fibrosis stages". For panel biomarkers including FT, Gebo et al. stated that "One of the major limitations may be in the lack of reliable identification and classification of the intermediate stages of fibrosis". Bissell also stated that for panels including FT "Their accuracy for intermediate fibrosis is relatively poor." . Rockey and Bissell stated that "Decision-making requires a test that differentiates minimal disease [stage 0/1 fibrosis) from intermediate fibrosis [stage 2/3). For this purpose, the current generation of non-invasive tests falls short, and liver biopsy still is needed for definitive staging" . These statements are not evidence-based.
The first error is stating that "liver biopsy is still needed for definitive staging of intermediate stages". The entire liver is certainly the gold standard but the liver biopsy is an imperfect gold standard. The present overview of the 25 studies giving biopsy length, all performed in tertiary centers, observed among 5,404 patients that the median of mean biopsy length was 18 mm. For the two larger studies including more than 500 patients (total 1,428) the median was 14 mm and in the integrated data base the mean was 17 mm out of 3,282. In our tertiary center a prospective study observed in 1,769 patients that biopsy was greater than 25 mm in only 16% (280/1769) of patients .
A liver biopsy of 15 mm has an AUROC of 0.82 between F1 and F2, being around 20% of false positives or false negatives . Therefore FT with an AUROC of 0.66 (usually described as a weak value when using a true gold standard) between F1 and F2 has a relative AUROC versus the best AUROC possible of 0.66/0.82 = 0.80, which is in the end acceptable for a non invasive test.
The second error is the confusion between intermediate stages and adjacent stages. For any estimate of liver fibrosis, the diagnostic values (AUROCs) between adjacent stages need to be assessed. There are no significant differences in the diagnostic values (AUROCs) for FT as demonstrated in this study, or for liver biopsy as demonstrated by Bedossa et al  according to intermediate stages as opposed to extreme stages, with the AUROCs for all adjacent stages being similar. The same results have been observed for all the combinations of stages in a previous study  and with this database (data not showed).
The third error is assessing the diagnostic value of a biomarker in a subpopulation of patients defined by liver biopsy such as F2/F3 vs. F0/F1. The exclusion of F4 patients defined by a 15 mm biopsy will not exclude the risk of false positives or false negatives of the remaining non-F4. It is much more important to assess the spectrum of fibrosis stage among the F0/F1 and F2/F3; if the AUROCs are not standardized according to the DANA, the ObAUROCs will be misleading . This once again underlines that assessing the AUROCs between all adjacent stages remains the best way, knowing that for the "perfect" biomarker, the best possible achievable AUROC is 0.82 for a 15 mm biopsy.
There are also different methodological approaches for the overview of fibrosis markers. Parkes et al. arbitrarily defined an "inaccurate" zone of a marker when it "cannot reliably attribute result for tests as tests perform with lower sensitivities/specificities at thresholds, where positive predictive value < 90%, negative predictive value > 95%" . There is no rationale for choosing these thresholds, but this definition could be acceptable if a true gold standard existed. This is not the case for fibrosis markers. If this definition is applied to 15 mm liver biopsies, the biopsy will be inaccurate in 40% of cases for a diagnosis between F1 and F2.
The only significant difference identified using AUROCs between adjacent stages (Table 5) was for HBV versus ALD. The obAUROC for ALD was particularly high and this should be validated in population with greater sample size.
High risk profile
"The observed high risk profile of FT in the published studies (4.1%) and in the integrated database (1.9%) were concordant with the post marketing analyses finding (2.1%) in 32,527 consecutive tests [2, 46]. In these analyzes there were 272 cases (0.8%) with a high-risk profile of false positives, for which the other components were not concordant in favor of significant fibrosis. Patients with extremely low haptoglobin, particularly when the rest of the exams were hardly modified, could have had hemolysis. A high-risk profile of false positives due to possible Gilbert syndrome was observed in 409 (1.3%) cases. In the presence of acute inflammation (i.e., sepsis or acute hemolysis), FT analysis must be postponed ."