1.1 Performance

1.1.1 Pivotal study
In the pivotal study, FibroMeter was mainly evaluated in patients with chronic hepatitis due to HCV (2). The overall diagnostic performance for significant fibrosis (Metavir F2/3/4), as reflected by the area under the receiver operating characteristic (AUROC: an index summarizing sensitivity and specificity), was significantly superior to Fibrotest (5), Forns test (6) and APRI (7) in the test population. FibroMeter AUROC was also superior to those of Fibrospect (8) and the European liver fibrosis test (9).

1.1.2 Validation studies
The FibroMeter for fibrosis staging in HCV hepatitis has been evaluated in two independent studies (10, 11) and in one meta-analysis with individual data including the two previous studies, the pivotal study and other available cohorts (12).
We present here the results of this meta-analysis including 1056 patients (12) that can be summarized as follows. Briefly, the FibroMeter AUROC or correct classification rate (figure 1) for significant fibrosis was very significantly higher than that of other usual tests evaluated: Fibrotest, Hepascore (13), APRI and Fib-4 (14). The AUROC of these other tests were not significantly different between them. In a sensitivity analysis excluding the population of the pivotal study, these significant differences were maintained.

1.2 Performance profile

The previous overall diagnostic performance is a necessary descriptor to compare blood tests but it is insufficient. Indeed, this overall performance has two limits. First, it depends on the prevalence of the diagnostic target, i.e. significant fibrosis, and thus is not reproducible in different epidemiological settings. Second, it includes different fibrosis stages with variable meanings. Thus, two categories of fibrosis stages have to be distinguished: the median stages (F1/2/3) and the extreme stages (F0 and F4). Indeed, the performance of median stages is hampered by two drawbacks: it strongly depends on the diagnostic cut-off that determines the diagnostic target and, more importantly, it is strongly biased by the reference, i.e. liver biopsy, since F1/2/3 stages have a low interobserver reproducibility in histological staging (15). F0 and F4 are clinical important stages: F0 correspond to the lack of fibrosis which is very reassuring for the patient and F4 corresponds to proved cirrhosis. This diagnosis implies a stringent management with an active screening of esophageal varices and hepatocellular carcinoma. To evaluate that issue, Halfon et al have implemented the performance profile that describes the diagnostic accuracy in each fibrosis stage (10). In the previous meta-analysis quoted, FibroMeter had an unique profile (12). This was the only test to correctly classified 100% of patients in F0 or F4 stages regarding the diagnosis of significant fibrosis (figure 2).

In other words, no patient with F0 stage at liver biopsy was classified by FibroMeter as having significant fibrosis, and all patients with F4 stage were classified as having significant fibrosis. In addition, this correct classification was significantly superior in each stage of significant fibrosis that are F2, F3 and F4 compared to other usual blood tests.
The diagnostic performance of FibroMeter is higher than that of other usual tests in all values (figure 3).

However, its main advantage is to have narrower values at extreme stages (F0 and F4) as seen in box plots (figure 4).

This can be attributed to biomarkers like hyaluronic acid and platelets that have abnormal results mainly in stages of severe fibrosis at the difference of other markers. Finally, FibroMeter displays the highest overall diagnostic accuracy and is the only test to correctly classify all patients without fibrosis or with cirrhosis.

2.1 Reliable diagnosis

A reliable diagnosis corresponds to the intervals of blood test values where the diagnostic accuracy is considered sufficiently reliable for clinical practice. Thus, in these patients, a liver biopsy is considered as avoidable (12).

2.1.1 Traditional intervals
Previously, the intervals of reliable diagnosis were defined by the thresholds provided by 90% negative (NPV) and positive (PPV) predictive values (5, 16). Thus, with the traditional definition based on 90% predictive values, a reliable diagnosis for a patient means >= 90% chance to be F0/1 in the lowest interval and >= 90% chance to be >= F2 in the highest interval of blood test values (figure 3).

 

The indeterminate interval is the zone outside of these reliable intervals. Figure 5 depicts the proportion of patients with 90% predictive values; FibroMeter included a significantly higher proportion of patients than other usual blood tests, so FibroMeter was the most predictive test.

2.1.2 New intervals
Besides these traditional intervals of blood test values based on 90% predictive values, it is possible to define new intervals based on a >= 90% accuracy in each interval of blood test values. Thus, by including new intervals (between diagnostic cut-off at 0.5 and 90% NPV or PPV: figure 3),  it was possible to obtain the following four intervals of FibroMeter values with the respective reliable diagnoses: F0/1, F1/2, F2+/-1, F3+/-1 in 100% of the population, among whom 90% were correctly classified with FibroMeter (figure 6).

Thus, liver biopsy can be avoided in 100% of patients with this new classification. Finally, this provides a simplified fibrosis meter including 4 stages with high accuracy. It is possible to be more precise with the following refinement.

2.2 Meters

Blood tests are usually constructed following a binary diagnostic process (logistic regression), which provides the probability (i.e. the predictive value) of having significant fibrosis when the test is above a given cut-off. However, it would be interesting in clinical practice to know the probability of a Metavir fibrosis stage for a given blood test value. Figure 7 depicts the frequency of fibrosis stages as a function of blood test values in the previously quoted meta-analysis. For example, a patient with result close to 0 had a 100% chance of having F0/F1 with FibroMeter F virus. By contrast, a patient close to 1 had an 85 % chance of having cirrhosis. Thus, by expending this classification, it is possible to derive a more precise meter of fibrosis stages from probability of Metavir fibrosis stage at the expense of a small decrease in overall accuracy compared to the binary diagnosis (figure 7).

Thus, the diagnostic accuracy changes from 82% for significant fibrosis to 77% for the meter including 6 fibrosis stages (17). Finally, there are three ways to express the FibroMeter results for fibrosis staging (table 2)

: the original diagnosis of significant fibrosis, the reliable diagnosis intervals (figure 6), and the fibrosis meter (figure 7). The fibrosis meter is expected by the physician; consequently the website provides this practical information but the reliable diagnosis intervals are a more accurate descriptor.

2.3 Exportability

2.3.1 Reproducibility
We have tested the interlaboratory reproducibility of FibroMeters in several studies including a large number of different laboratories in France (18). Globally, it was excellent with an intraclass correlation coefficient at 0.991 (18). The “real life” simulation from quality controls of blood samples performed in 317 French laboratories using 25 different dosage methods indicated a maximum variation of 1.7% in FibroMeter results with a single biomarker (18).
The diagnostic cut-off of the blood test value distinguishes the patients according to the diagnostic target (significant fibrosis). It is usually fixed in the pivotal population at 0.5 to maximize the diagnostic accuracy according to statistic rules. However, the best cut-off might have a different value especially in other populations. Thus, the overall reproducibility of blood tests was evaluated through the variability of the best cut-off in a large population as a function of centers (17). FibroMeter was the test with the lowest variability (figure 8).

Finally, we have compared the reproducibility of liver biopsy, Fibroscan and blood test in real life conditions in several studies performed in France. It was possible to compare the reproducibility of these methods after expressing their results into Metavir fibrosis stages (19). The reference was a double reading of liver biopsy by experts. The kappa indexes were: liver biopsy: 0.336 (poor agreement), Fibroscan: 0.640 (good), FibroMeter: 0.874 (excellent).

2.3.2 Robustness
The pivotal studies include several markers with independent significant predictive values. The robustness evaluates the ability of these markers to keep this independent information in other settings, especially in conditions nearer to the real life conditions. In other words, the robustness reflects the diagnostic stability of the blood tests in various conditions. Different indices of robustness have been calculated (19). For example, the most practical index was the proportion of patients for whom the diagnosis of significant fibrosis consequently changed in a large cohort of HCV patients provided by primary to tertiary centers compared to the score of blood tests used in the pivotal studies: the variability of FibroMeter was significantly lesser than with other usual tests (figure 9).

In addition, some variables (age and bilirubin) had no independent value for Fibrotest and Hepascore calculation in this large cohort (17). By contrast, the performance and the number of independent variables of FibroMeter were stable. The inclusion of all 12 variables used in five blood tests in stepwise logistic regression showed that the 7 variables of FibroMeter kept their independent information in addition to sex. Finally, FibroMeter was the most robust test i.e. that having the most stable diagnostic performance in different centers.

2.4 Expert system

Briefly, in a first step, the system detects putative false positive or false negative according to rules precisely described by experts. For example, an isolated marked variation (increase) in blood urea is attributed to a renal dysfunction whereas an isolated marked decrease in prothrombin index is attributed to vitamin K deficiency. These undesirable abnormal results can also be noticed by the patient physician. In a second step, the system corrects this putative erroneous result by comparing two FibroMeter results where the value of the putative culprit variable is replaced by the normal mean and an alternative score of FibroMeter where the variable is excluded. Finally, the system validates or not the corrected result. If a corrected result is proposed (18% of cases), a personalised caution is provided with the patient results. Several ancillary functions are included in the system. The main one is very practical by providing a diagnosis of exclusion or affirmation of several diagnostic targets. They correspond respectively to 100% NPV and PPV.

3.1 Standard FibroMeter

The standard FibroMeter is that targeted for significant fibrosis provided by pivotal study. AUROCs for severe fibrosis and cirrhosis were 0.885 and 0.907 for FibroMeter, 0.837 and 0.882 for Fibrotest, 0.834 and 0.896 for Hepascore, 0.822 and 0.841 for APRI respectively (p<10-3, respectively, between blood tests). The proportions of patients in whom severe fibrosis and cirrhosis could be excluded (100% NPV) were significantly higher with FibroMeter than with other blood tests (figure 10).

By using the diagnostic thresholds proposed in meters of blood tests (17, 20), PPV for the diagnosis of cirrhosis was 37.1% for Fibrotest and 68.5% for standard FibroMeter.

3.2 Cirrhosis FibroMeter

In order to improve the diagnostic performance for cirrhosis, we implemented a specific score including the same variables as in the previous standard FibroMeter. The correct classification rate for the cirrhosis diagnosis of this specific FibroMeter for cirrhosis (93.0%) was significantly higher compared to the standard FibroMeter (90.9%, p=0.005); respective AUROC increased from 0.907 to 0.919. In addition, this specific FibroMeter provided a 100% PPV for cirrhosis diagnosis that was quite better than the maximum 88% PPV obtained with standard FibroMeter (21).

We compared several blood tests, Doppler-ultrasonography and ultrasonographic elastometry (Fibroscan®) in a large population (22). Briefly, for the diagnosis of significant fibrosis, two methods had independent diagnostic accuracy and thus were synergistic: FibroMeter at the first step and Fibroscan at the second step. For the diagnosis of cirrhosis, this was the same couple but with Fibroscan at the first step. A synchronous algorithm was constructed with the two methods. It had two main advantages: the diagnostic accuracy for significant fibrosis was significantly increased and it suppressed the discrepant results compared to both methods considered as separate diagnostic procedures.

The diagnostic accuracy of InflaMeter for significant activity (Metavir grade A2/3) was 74%. With the previously described method of probability of stages (17) applied to activity grade, it was possible to construct a meter distinguishing A0/1, A1/2, A2/3 with an increased diagnostic accuracy to 89% (unpublished data).

6.1 Limits of blood tests

The high diagnostic performance of blood test is attributed to the lack of observer and to the quality controls performed as well as to efforts to standardisation. Despite the risk of variability due to multiple devices and reagents, we have seen that blood tests offer the highest reproducibility compared to other diagnostic methods.
Liver biopsy is the reference of blood test and thus imposes its own limits. Studies on the causes of discrepancy between blood tests and liver biopsy are very difficult to perform since they are observer dependent. Several results suggest that the variability of blood test is less than the original liver biopsy from which it is derived. For example, figure 11,

derived of our first study (1), shows that the course of area of fibrosis as a function of clinical events is more plausible with FibroMeter than with liver biopsy (24). In addition, the variability of FibroMeter area of fibrosis is lesser than the measurement on liver biopsy from which it is derived (figure 12).

In other words, the manufactured copy -FibroMeter- is less variable and more reproducible than the original hand crafted manuscript -liver biopsy- in clinical practice.

6.2 Limits and advantages of FibroMeters

We have seen that FibroMeters have several advantages in terms of diagnostic performance, reproducibility and robustness. This can be attributed to the design of the tests and to the multiple step diagnostic process. Thus, FibroMeter F virus have the largest spectrum of biomarkers regarding fibrogenesis and fibrolysis (25). However, this has the inconvenience to require several laboratory automates (three as a rule) and to increase the cost.
Interpretation of FibroMeter requires the validation by a physician like any other diagnostic test. Although the result is analysed by an expert system on a professional website, the expert remains a qualified physician and the interpretation of non invasive test requires expertise in this field.
FibroMeters have not been validated in miscellaneous causes of liver diseases, in children, and in comorbidities. However, in our hands the FibroMeter virus is the most usable option.
All the diagnostic improvements brought by FibroMeter options need the use of sophisticated software that is available on a website.

1. Oberti F, Valsesia E, Pilette C, Rousselet MC, Bedossa P, Aube C, et al. Noninvasive diagnosis of hepatic fibrosis or cirrhosis. Gastroenterology 1997;113:1609-16.
2. Cales P, Oberti F, Michalak S, Hubert-Fouchard I, Rousselet MC, Konate A, et al. A novel panel of blood markers to assess the degree of liver fibrosis. Hepatology 2005;42:1373-81.
3. Intraobserver and interobserver variations in liver biopsy interpretation in patients with chronic hepatitis C. The French METAVIR Cooperative Study Group. Hepatology 1994;20:15-20.
4. Pilette C, Rousselet MC, Bedossa P, Chappard D, Oberti F, Rifflet H, et al. Histopathological evaluation of liver fibrosis: quantitative image analysis vs semi-quantitative scores. Comparison with serum markers. J Hepatol 1998;28:439-46.

5. Imbert-Bismut F, Ratziu V, Pieroni L, Charlotte F, Benhamou Y, Poynard T. Biochemical markers of liver fibrosis in patients with hepatitis C virus infection: a prospective study. Lancet 2001;357:1069-75.
6. Forns X, Ampurdanes S, Llovet JM, Aponte J, Quinto L, Martinez-Bauer E, et al. Identification of chronic hepatitis C patients without hepatic fibrosis by a simple predictive model. Hepatology 2002;36:986-92.
7. Wai CT, Greenson JK, Fontana RJ, Kalbfleisch JD, Marrero JA, Conjeevaram HS, et al. A simple noninvasive index can predict both significant fibrosis and cirrhosis in patients with chronic hepatitis C. Hepatology 2003;38:518-26.
8. Patel K, Gordon SC, Jacobson I, Hezode C, Oh E, Smith KM, et al. Evaluation of a panel of non-invasive serum markers to differentiate mild from moderate-to-advanced liver fibrosis in chronic hepatitis C patients. J Hepatol 2004;41:935-42.
9. Rosenberg WM, Voelker M, Thiel R, Becka M, Burt A, Schuppan D, et al. Serum markers detect the presence of liver fibrosis: a cohort study. Gastroenterology 2004;127:1704-13.
10. Halfon P, Bacq Y, De Muret A, Penaranda G, Bourliere M, Ouzan D, et al. Comparison of test performance profile for blood tests of liver fibrosis in chronic hepatitis C. J Hepatol 2007;46:395-402.
11. Leroy V, Hilleret MN, Sturm N, Trocme C, Renversez JC, Faure P, et al. Prospective comparison of six non-invasive scores for the diagnosis of liver fibrosis in chronic hepatitis C. J Hepatol 2007;46:775-82.
12. Calès P, de Ledinghen V, Halfon P, Bacq Y, Leroy V, Boursier J, et al. Evaluating accuracy and increasing the reliable diagnosis rate of blood tests for liver fibrosis in chronic hepatitis C. Liver International 2008;(in press).
13. Adams LA, Bulsara M, Rossi E, DeBoer B, Speers D, George J, et al. Hepascore: an accurate validated predictor of liver fibrosis in chronic hepatitis C infection. Clin Chem 2005;51:1867-1873.
14. Vallet-Pichard A, Mallet V, Nalpas B, Verkarre V, Nalpas A, Dhalluin-Venier V, et al. FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. comparison with liver biopsy and fibrotest. Hepatology 2007;46:32-6.
15. Rousselet MC, Michalak S, Dupre F, Croue A, Bedossa P, Saint-Andre JP, et al. Sources of variability in histological scoring of chronic viral hepatitis. Hepatology 2005;41:257-64.
16. Colletta C, Smirne C, Fabris C, Toniutto P, Rapetti R, Minisini R, et al. Value of two noninvasive methods to detect progression of fibrosis among HCV carriers with normal aminotransferases. Hepatology 2005;42:838-45.
17. Leroy V, Halfon P, Bacq Y, Boursier J, Rousselet M, Bourlière M, et al. Diagnostic accuracy, reproducibility and robustness of fibrosis blood tests in chronic hepatitis C: a meta-analysis with individual data. Clin Biochem 2008;(in press).
18. Cales P, Veillon P, Konate A, Mathieu E, Ternisien C, Chevailler A, et al. Reproducibility of blood tests of liver fibrosis in clinical practice. Clin Biochem 2008;41:10-8.
19. Calès P, Boursier J, Rousselet M, Michalak S, Oberti F, Gallois Y, et al. Comparison of reproducibility of histology, blood tests and Fibroscan for liver fibrosis. Hepatology 2007;46:834A.
20. Poynard T, Imbert-Bismut F, Munteanu M, Messous D, Myers RP, Thabut D, et al. Overview of the diagnostic value of biochemical markers of liver fibrosis (FibroTest, HCV FibroSure) and necrosis (ActiTest) in patients with chronic hepatitis C. Comp Hepatol 2004;3:8.
21. Boursier J, Bacq Y, Halfon P, Leroy V, de Ledinghen V, de Muret A, et al. Improved diagnostic accuracy of blood tests for severe fibrosis and cirrhosis in chronic hepatitis C. Eur J Gastroenterol Hepatol 2008;(in press).
22. Boursier J, de Ledinghen V, Sawadogo A, Lebigot J, Michalak S, Gallois Y, et al. Amélioration du diagnostic de la fibrose hépatique par la combinaison synchrone de tests non invasifs (résumé). Gastroenterol Clin Biol 2007;31:31.
23. Boursier J, Dib N, Oberti F, Cales P. [Characteristics and interpretation of blood tests for liver fibrosis]. Gastroenterol Clin Biol 2007;31:511-23.
24. Boursier J, Dib N, Oberti F, Hubert I, Rousselet M, Gallois Y, et al. Role of fibrosis degree in the clinical course of chronic liver diseases. Gastroenterology 2007;132:A804.
25. Pinzani M, Vizzutti F, Arena U, Marra F. Technology Insight: noninvasive assessment of liver fibrosis by biochemical scores and elastography. Nat Clin Pract Gastroenterol Hepatol 2008;5:95-106.
26. Leroy V, Sturm N, Hilleret M, Renversez P, Trocmé C, Faure P, et al. Diagnostic accuracy of blood tests of liver fibrosis in chronic hepatitis B: comparison with hepatitis C (abstract). Hepatology 2007;46:900-1A.
27. Cacoub P, Carrat F, Bedossa P, Lambert J, Penaranda G, Perronne C, et al. Comparison of non-invasive liver fibrosis biomarkers in HIV/HCV co-infected patients: The fibrovic study - ANRS HC02. J Hepatol 2008;48:765-73.
28. Calès P, Halfon P, Batisse D, Carrat F, Perré P, Penaranda G, et al. Tests sanguins de fibrose hépatique chez les patients co-infectés VIH et VHC. Gastroenterol Clin Biol 2008;(in press).
29. Oberti F, Anty R, Vanbiervliet G, Lacave-Oberti N, Gelsi E, Rosenthal A, et al. Meta-analysis of blood scores of liver fibrosis (Fibrometer, Hepascore, APRI) in alcoholic chronic liver diseases (abstract). Hepatology 2006;44:467A.
30. Calès P, Lainé F, Boursier J, Deugnier Y, Moal V, Oberti F, et al. Comparison of blood tests for liver fibrosis specific or not to NAFLD. J Hepatol 2008;(in press).
31. Angulo P, Hui JM, Marchesini G, Bugianesi E, George J, Farrell GC, et al. The NAFLD fibrosis score: a noninvasive system that identifies liver fibrosis in patients with NAFLD. Hepatology 2007;45:846-54.
32. Maor Y, Cales P, Bashari D, Kenet G, Lubetsky A, Luboshitz J, et al. Improving estimation of liver fibrosis using combination and newer noninvasive biomarker scoring systems in hepatitis C-infected haemophilia patients. Haemophilia 2007;13:722-9.
33. Sombié R, Lescure F, Bougouma A, Guiard-Schmid J, Poynard T, Calès P, et al. Evaluation de la fibrose hépatique (histologie, marqueurs sériques, élastométrie) chez des patients infectés par le VHB au Burkina Faso. JAHG 2007:(in press).
34. Foucher J, Chanteloup E, Vergniol J, Castera L, Le Bail B, Adhoute X, et al. Diagnosis of cirrhosis by transient elastography (FibroScan): a prospective study. Gut 2006;55:403-8.