Long-term Prognostic Value of the FibroTest in Patients With Non-alcoholic Fatty Liver Disease, Compared to Chronic Hepatitis C, B, and Alcoholic Liver Disease

Mona Munteanu; Raluca Pais; Valentina Peta; Olivier Deckmyn; Joseph Moussalli; Yen Ngo; Marika Rudler; Pascal Lebray; Frederic Charlotte; Vincent Thibault; Olivier Lucidarme; An Ngo; Françoise Imbert-Bismut; Chantal Housset; Dominique Thabut; Vlad Ratziu; Thierry Poynard


Aliment Pharmacol Ther. 2018;48(10):1117-1127. 

In This Article

Patients and Methods

Study Design

The design was an analysis of fresh serum specimen recorded in a non-interventional prospective cohort. Patients followed by the Hepatology department were from the "Groupe Hospitalier Pitié Salpêtrière cohort" of FIBROFRANCE, a program organized in 1997 to assess the burden of chronic liver diseases in France (Clinical trial French registry no.: DRCD–2013–1 and ClinicalTrials.org no.:CT01927133). STROBE statements were followed (Table S1). The protocol was approved by the institutional review board, regulatory agency and performed in accordance with principles of Good Clinical Practice. All patients provided written informed consent before entry. All authors had access to the study data and reviewed and approved the final manuscript.

Patients with a FibroTest performed before 2013, without previous liver transplantation, and without acute liver disease, were selected (Figure 1). Follow-up and treatments were scheduled according to the updated international guidelines.[15,16] In patients with baseline cirrhosis ultrasonography (US) examination was performed every 6 months, and AFP was recommended every 6 months.

Figure 1.

Flow sheet of population subsets

Non-invasive Liver Biomarkers

FibroTest (BioPredictive Paris, France;FibroSURE LabCorp Burlington, NC, USA) is a in vitro multi-analyte serum test including the serum concentrations of α2–macroglobulin, apolipoprotein A1, haptoglobin, total bilirubin, and GGT, adjusted for age and gender. Alpha–2–macroglobulin, apolipoprotein A1 and haptoglobin were measured using an automatic nephelometer BNII (Dade Behring, Marburg, Germany). The recommended pre-analytical and analytical procedures were applied.[17] The scores of these biomarkers range from 0 to 1.00, the highest scores being attributed to the most severe lesions.

To compare the FibroTest use in our cohort since 1997, to other tests which were validated in 2005 for transient elastography-probe-M (TE) and 2006 for FIB4, we performed a post hoc analysis of FIB4 performance cases with NAFLD who had simultaneous measurements of TE and FIB4.

To assess the possible impact of NASH or Steatosis on the survival, we also assessed as possible risk factors two validated updated blood tests, the NashTest–2 and the SteatoTest–2. NashTest–2 combined 11 components with different weights, including the seven components of FibroTest, plus alanine aminotransferase, aspartate aminotransferase, cholesterol and triglycerides.[13] SteatoTest–2 combined with different weights, 10 out of the 11 components of NashTest–2, without bilirubin, but with fasting glucose.[14]

Follow-up and Outcomes

Patients attended the Hepatology Department when justified by abnormal liver tests, or at least once a year in case of cirrhosis. The duration of follow-up was calculated from the baseline date, defined as the date when the serum used for their first analyses of liver biomarkers was collected, to the date of a lethal event occurred. Survivals were not censored at transplantation time. This interval was censored at the time of last follow-up.

The mortality rate and incidence of cardiovascular events were determined during follow-up. The causes of death were collected from the French national registry (CepiDc Inserm), according to the 10th International classification of Diseases. Mortality status obtained by physician, hospital or national register.

Cause of death was classified as liver-related, cardiovascular-related, cancer-related, and others, according to WHO codes (who.int/classifications/icd10/browse/2010). The codes considered as diagnosis of cardiovascular-related death were: ischemic heart diseases (I20–I25), cardiac arrest (I46), heart failure (I50), cerebrovascular diseases (I63 and I64) and cardiogenic shock (R57.0). The codes for liver-related death were: liver fibrosis and cirrhosis of liver (K74), non-alcoholic steatohepatitis (K75.8), portal hypertension (K76.6), oesophageal varices bleeding (I85.0), hepatocellular carcinoma (C22.0) and cholangiocarcinoma (C22.1). Death coded as R99 (other ill-defined and unspecified causes of mortality) was considered as unknown cause of death.

Statistical Methods

Survival analyses (events defined as death) were based on univariate (Kaplan-Meier, Logrank, time-dependent area under the ROC curve AUROC), and multivariate Cox-risk-ratio analyses, taking into account age, gender and response to antiviral treatment as covariates. The independent prognostic value of each FibroTest component was assessed for each liver disease. The main endpoint was the performance of the FibroTest for liver-related death in NAFLD, compared to results observed in CHC, the most validated population.

The association between ApoA1, haptoglobin and the non-liver causes of deaths (cardiovascular and non-liver cancer-related deaths), were assessed by AUROCs and Cox univariate and multivariate analyses, including three major prognostic factors, age, gender and A2M as marker of cirrhosis. To prevent colinearity, FibroTest was not used as a marker of cirrhosis as ApoA1 and haptoglobin were components of FibroTest.

In sensitivity analyses, we performed multivariate Cox models to assess if the FibroTest significant prognostic value at 10–year persisted after adjustment by gender, BMI (cutoff = 27 kg/m2), and T2–diabetes, in NAFLD compared to CHC. Analyses were repeated after exclusion of NAFLD patients with alcohol consumption missing, or with rare alcohol consumption but under the standard definition of consumption at risk (daily alcohol consumption ≥30 g for men and ≥20 g for women), and also after excluding patients with HIV-PCR missing or positive.

All statistical analyses were performed using NCSS–12.0 and R softwares, including timeROC library.