Introduction

Chronic liver disease (CLD) is a major and growing public health issue worldwide1. The main complications of CLD are cirrhosis and hepatocellular carcinoma (HCC), which account for 3.5% of worldwide global mortality2,3. About 25% of the world's population currently lives with metabolic dysfunction associated steatotic liver disease (MASLD), and 1.2% with alcohol-related disorders2,3,4. An increase in the prevalence of MASLD is expected owing to the increased in central obesity worldwide5,6,7,8. Fibrosis is the only determinant of liver-related mortality, particularly at the advanced fibrosis stage9,10. Unfortunately, CLD are discovered at a late stage: in 75% of cases, cirrhosis is diagnosed at a decompensated stage and HCC at an advanced stage not eligible for curative treatment11,12,13,14. Late diagnosis prevents the early implementation of measures likely to limit fibrosis progression and of a HCC screening programs15. Yet several studies have shown that HCC screening improves patient prognosis16,17,18. Early diagnosis of advanced liver fibrosis is critical to improve overall prognosis of CLD in general and HCC in particular19.

The European Association for the Study of the Liver (EASL) recommends assessing liver fibrosis in patients with risk factors for steatotic liver disease (SLD): excessive alcohol consumption, metabolic syndrome, obesity and diabetes20. The screening strategy relies on non-invasive tests (NIT), with FIB-4 test first in line. FIB-4 is a simple test with no additional cost, based on aspartate aminotransferase (AST), alanine transaminase (ALT), platelets and age21. If FIB-4 suggests fibrosis, EASL recommends liver stiffness measurement by vibration controlled transient elastography (VCTE), to be performed.

At Grenoble Alpes University Hospital (GA University Hospital) we implemented a care pathway based on the systematic calculation of FIB-4 by the central laboratory for medical departments managing patients with a high prevalence of metabolic risk factors for CLD. Only FIB-4 ≥ 2.67 in patients aged 18 to 70 were reported in the final lab reports along with an incentive to request a second-line fibrosis test by measurement of liver stiffness by VCTE. The aim of this study was to evaluate the effectiveness and the feasibility of this sequential care pathway.

Materials and methods

Study design and population

We conduct an observational, retrospective, single-center study at GA University Hospital (Fig. 1). Four pilot departments were chosen due to high prevalence of patients with metabolic risk factors for CLD: Endocrinology-Diabetology-Nutrition, Internal Medicine, Nephrology and Respiratory Physiology. Systematic calculation of FIB-4 was implemented from November 16th 2020 in the central laboratory system each time AST, ALT and platelets were prescribed in the same lab request. The FIB-4 result was returned to the prescriber if it was ≥ 2.67 in patients aged 18 to 70, with an incentive to request a specialized fibrosis testing by measurement of liver stiffness. Liver stiffness measurement was performed using VCTE technology (FibroScan device; Echosens) according to the manufacturer's recommendations. An experienced operator (≥ 500 examinations) recorded 10 valid measurements. VCTE results were expressed in kilopascals (kPa) as the median of these valid measurements.

Figure 1
figure 1

Care pathway implemented in the GA University Hospital.

Full size image

If VCTE was ≥ 8 kPa, the patient was referred to a liver specialist.

Patients with a FIB-4 available between November 16th, 2020 and November 16th, 2022 were included and followed up until May 16th, 2023.

Eligibility criteria to VCTE examination were as follows:

  • FIB-4 ≥ 2.67

  • Age ≥ 18 years and ≤ 70 years

Exclusion criteria were as follows:

  • Known liver disease including polycystic liver disease (PLD)

  • Identified cause of false positive FIB-4: non-liver related thrombocytopenia, acute event associated with elevated transaminase, cancer treated with systemic therapy, or hepatic metastases

  • Life-threatening condition

All procedures were followed in accordance with STROBE22. The study protocol complied with the 1975 Declaration of Helsinki. The study «ANTICIPATE CHUGA» meets the ethics requirements of the CNIL (Commission Nationale de l’Informatique et des Libertés) reference methodology 004 related to research on data not involving human subjects, waiving the need for informed consent (CNIL regulation reference 1818709X)23. Accordingly, the conduct of the study has been registered and authorized by Grenoble Alpes University Hospital Research Regulatory Board on March 18th 2022. Information was available to all patients undergoing blood tests in participating departments and objections to use personal data recorded and none of the patients objected participation (https://www.chu-grenoble.fr/sites/default/files/content/recherche/documents/2023-02/ANTICIPATE%20CHUGA_Affiche%20sp%C3%A9cifique%20V1.0%20du%2018.03.2022.pdf.).

Data collected

Data collected for all patients included age, sex, date of sampling, FIB-4 value and department of origin.

For patients with FIB-4 ≥ 2.67, additional data were collected when available: (1) clinical data: medical history, current treatments, anthropometric data, lifestyle; (2) biological data: standard biology, metabolic parameters, HBV and HCV tests, autoimmune status, specialized liver fibrosis blood tests; 3) imaging data: CT scan, ultrasound, MRI; 4) anathomopathological data, including liver biopsy results; and 5) VCTE results.

Metabolic syndrome was defined according to the World Health Organization (WHO)24.

Obesity was defined as Body Mass Index (BMI) > 30.

Alcohol consumption was assessed according to French Alcohol Society guidelines25.

Steatosis was considered significant when the control attenuation parameter (CAP) value was ≥ 248 dB/m on VCTE20,26. Fibrosis was considered significant when the elastometry value was ≥ 8 kPA on VCTE20.

Agile3 + and Agile4 scores were computed using MyFibroscan application27.

Agile3 + score ≥ 0.68 or a liver biopsy finding a fibrosis score ≥ F3 ruled-in advanced fibrosis27. Cirrhosis was confirmed if any of the following was met: Agile4 score ≥ 0.5727, elastometry ≥ 15 kPa20,28, liver biopsy confirming cirrhosis, or diagnosis combination of arguments for cirrhosis.

Outcomes

The primary endpoint was the number of patients with unrecognized significant liver fibrosis (defined as elastometry ≥ 8 kPa) identified through the fibrosis care pathway during the study period.

The secondary endpoints were: the number of FIB-4 calculated, the number of FIB-4 ≥ 2.67 out of all FIB-4 calculated for patients 18–70 years of age, the number of specialized fibrosis tests (VCTE) performed as part of the care pathway, the descriptive analysis of patients with significant liver fibrosis on VCTE and the assessment of factors associated with the presence of significant fibrosis on VCTE.

Statistical analysis

Data were described using frequencies and percentages for categorical variables; mean, standard deviation, range, median and interquartile range (IQR) for continuous variables.

For patients with more than one FIB-4 during the inclusion period, a primary analysis was performed with the first FIB-4 available. A secondary analysis was performed taking into account patients with a first FIB-4 < 2.67 but a follow-up FIB-4 ≥ 2.67. A Chi2 test was used for between-group comparisons for categorical variables if the headcount was sufficient (≥ 10). When frequencies were < 10, a Fisher’s exact test was used. For continuous variables, a Student's t-test was used if normality was met, and a non-parametric Wilcoxon (or Mann–Whitney) test if not.

Exploratory analysis were performed to assess the factors associated with FIB-4 ≥ 2.67 and significant fibrosis, using univariate and multivariate analyses with logistic regression models. The multivariate model aiming at exploring factors associated with significant fibrosis, included age and gender, variables for which the hypothesis about an influence was strong in the literature and/or variables for which the p-value was < 0.2 in the univariate analysis. Associations with a p-value < 0.05 were considered statistically significant. Analyses were performed under R4.3.0.

Missing data management

When information was not fully available to assess the presence of hypertension, dyslipidemia, diabetes, obstructive sleep apnea (OSA) or metabolic syndrome, we considered that the patient must not have it. For data relative to alcohol consumption, smoking status or employment status, when missing it was considered unknown and the patient was excluded from the analysis (complete case analysis), except when assessing the number of cumulative risk factors: in that case we considered patients had alcohol consumption only if it was stated in the medical file.

Results

During the study period, 73,145 visits were recorded in the pilot departments, enabling 6327 FIB-4 to be calculated, corresponding to 4019 patients of all ages: 2955 patients with a single FIB-4 calculated over the study period, and 1064 patients with several FIB-4 calculated (3372 FIB-4) (Fig. 2). FIB-4 distribution by sex and age is shown in supplementary data Table 1.

Figure 2
figure 2

Study flow chart.

Full size image
Table 1 Description of the 47 patients eligible for VCTE.
Full size table

Of the 4019 patients, 2963 patients were aged 18 to 70: 52.6% were women and 47.4% men, age 48.9 years +/− 13.6 (mean +/− standard deviation) (47.5 years +/− 13.6 for women and 50.4 years +/− 13.5 for men, p < 0.001).

Concerning the patient distribution across pilot departments involved, 47.7% of patients came from Endocrinology-Diabetology-Nutrition departments, 30.1% from Nephrology, 21.4% from Internal Medicine and 0.6% from Pneumology.

Distribution across FIB-4 categories was: 2056 patients (69.4%) with FIB-4 < 1.3, 772 patients (26.1%) with FIB-4 between 1.3 and 2.67 and 135 patients (4.6%) with FIB-4 ≥ 2.67. Among 795 patients (26.8%) aged 18 to 40, only 3 men and 2 women had their first FIB-4 ≥ 2.67.

Of the 135 patients aged 18 to 70 with FIB-4 ≥ 2.67, 88 were excluded: 46 with known liver disease, 4 with PLD, 20 for non-liver related thrombocytopenia, 13 for progressive cancer, 4 for acute cytolysis and one for short-term life-threatening condition.
Of the 46 patients with known liver disease, 6 were actually lost to follow-up and were referred back to specialized care after FIB-4 testing.

Overall, 47 patients with FIB-4 ≥ 2.67 (34.8%) were eligible for VCTE.

Description of the 47 patients eligible for VCTE

Most of these 47 patients were men (59.6%), with age of 60.9 +/− 7.3 (Table 1).

With regards to risk factors, data were missing for 3 (6.4%) patients regarding employment status, 10 (21.3%) patients regarding alcohol consumption and 1 (2.1%) patient regarding tobacco consumption (Table 1).

Of these 47 patients, 35 (74.4%) had at least one risk factor for SLD: obesity, metabolic syndrome, diabetes, OSA, current or past excessive alcohol consumption; 23 (49%) patients had 0 or 1 risk factor, and 24 (51%) patients had at least 2 associated risk factors.

A VCTE examination was requested for each of the 47 eligible patients. Only 7 (< 15%) patients were referred at the initiative of the patient's physician in the department where he was hospitalized. Other requests were made after an alert issued by liver specialists involved in the care pathway.

Overall, 40 patients underwent VCTE. The mean elastometry value was 8.9 kPa +/− 7.4, and 15 patients (37.5%) had significant fibrosis (elastometry ≥ 8 kPa) (supplementary Table 2).

Exploratory analysis of factors associated with significant fibrosis

The analysis was carried out on the 40 patients who underwent VCTE (40/47 patients eligible). Among those, data were missing for 2 (5%) patients regarding employment status, 6 (15%) patients regarding alcohol consumption and all the patients had data completed regarding tobacco consumption.

All patients with significant fibrosis based on VCTE (n = 15) had at least one risk factor for SLD; whereas a risk factor was found in only 15/25 (60%) of patients without significant fibrosis (p = 0.006) (Table 2).

Table 2 Associated factors with significant fibrosis and exploratory multivariate model (among the 40 patients who underwent VCTE examination).
Full size table

In exploratory univariate analysis, only the presence of dyslipidemia, metabolic syndrome and the number of risk factors were associated with significant fibrosis, with Odds Ratio (OR) of 7.04 (CI 95% [1.53–51.34]), 3.86 (CI 95% [1.03–15.82]) and 5.86 (CI 95% [1.5–26.86]), respectively (Table 2).

For exploratory multivariate analysis, we chose sex, age and the number of risk factors (among obesity, diabetes, obstructive sleep apnea, metabolic syndrome, current or past excessive alcohol consumption). The low number of patient and the collinearity of variables prevented the use of multiple variables. All 40 individuals were included in the exploratory multivariate models since employment status and alcohol consumption were not selected for multivariate analyses.

When risk factors were considered in two classes (0 to 1 versus 2 or more risk factors for SLD), and after adjustment for sex and age, having ≥ 2 risk factors for SLD was associated with a significant risk of fibrosis, with an ORa of 14.05 (CI 95% [2.59–129.54]) (p = 0.006). Age was also inversely associated with significant fibrosis (adjusted OR (ORa) 0.86, CI 95% [0.75––0.96]).

For patients with multiple FIB-4 calculated over the study period, having a FIB-4 consistently ≥ 2.67 was not associated with the presence of significant fibrosis on VCTE, compared to FIB-4 moving in different categories overtime.

Patients with significant fibrosis (n = 15) were referred to a liver specialist but only 8 attended the appointment, all diagnosed with compensated cirrhosis. Final cirrhosis diagnosis was histologically proven in 2 patients, based on hepatic elasticity > 15 kPa in 3 patients and on a positive Agile4 + score in 3 patients.

Of the 7 patients lost to follow-up after VCTE, 6 had an Agile3 + score suggesting advanced fibrosis.

Description of patients with multiple FIB-4 calculated over the study period

768 patients aged between 18 and 70 had multiple FIB-4 calculated (Fig. 3). Among these patients, mean FIB-4 was 1.7 +/− 6.2, and the number of FIB-4 per patient 3.1 +/− 2.5. In 86.2% of cases, FIB-4 was always < 2.67, in 3.9% it was always ≥ 2.67. It changed class from < 2.67 to ≥ 2.67 or vice versa in 9.9% of cases.

Figure 3
figure 3

Distribution of patients with single or multiple FIB-4 tests over the study period.

Full size image

Secondary analysis of patient with FIB-4 ≥ 2.67 after a first FIB-4 < 2.67

After a first FIB-4 < 2.67, 45 patients had a FIB-4 ≥ 2.67 during follow-up, and 24 were excluded: 9 for known liver disease, 7 for thrombocytopenia of non-hepatic origin, 4 for progressive cancer, 2 for cytolysis associated with an acute event, and 2 for PLD.

A total of 21 patients with a FIB-4 ≥ 2.67 after a first FIB-4 < 2.67 were eligible for VCTE, 18 underwent the procedure among whom 4 patients had significant fibrosis (supplementary Table 3). Patients (n = 4) with significant fibrosis were referred to a liver specialist, 2 patients actually attended the meeting: one had a confirmed diagnosis of cirrhosis (liver elasticity > 15 kPa) and the other was not ruled-in for advanced fibrosis or cirrhosis (Agile3 + and Agile4 scores negative, elasticity controlled at 6.4 kPa). Both patients lost to follow-up after VCTE had a positive Agile3 + score, suggesting advanced fibrosis.

Discussion

We report the first results of the care pathway implemented in a tertiary hospital for systematic liver fibrosis screening. The two-stage strategy, combining systematic FIB-4 calculation based on routine laboratory prescriptions, followed by VCTE, allowed the detection of significant asymptomatic liver fibrosis in 19 patients over a 2-year period. In addition, the pathway identified 6 patients with known liver disease who were lost to follow-up, and could be referred back to specialized care before onset of a liver event.

Of the 47 patients eligible for VCTE in the main analysis, 74.4% had at least one risk factor for SLD (obesity, metabolic syndrome, diabetes, OSA and/or current or past excessive alcohol consumption), confirming the relevance of a screening strategy triggered by risk factors. Our results confirmed the synergistic effect between risk factors previously reported29,30.

Age has a massive impact in the FIB-4 formula, with an increased risk of false positive tests with age. Thus, we chose to restrict the care pathway to patients up to 70 years of age in the first phase of the pathway assessment. Although the performance of FIB-4 seems lower in patients aged 65 and over31, EASL guidelines do not suggest an age range in the recommended screening strategy20. Similarly, recommendations for HCC screening do not specify an age beyond which screening is no longer relevant32. There is no consensus in publications reporting the results of sequential strategies using first line FIB-4-to screen for liver fibrosis: some retained only patients under age 7933, under age 7534, and others set no age limit35.

In our study, after age 70, FIB-4 was ≥ 2.67 in 24% of cases, and this proportion increased to 34% after age 80. Conversely, among patients aged 18 to 40, only 0.6% had a FIB-4 ≥ 2.67, raising the question of the relevance of screening in a population below age 40. Indeed, The LIVERSCREEN European primary care fibrosis screening project has restricted the target to patients over age 4036.

We also chose to include only patients with a FIB-4 ≥ 2.67, which is the threshold identifying patients at highest risk of advanced fibrosis. This choice does not comply with EASL guidelines which advocate a threshold of 1.3 for screening20. However, in the EASL guidelines, the first step of the screening algorithm is the presence of risk factors for chronic liver disease. In our study, the first step is the automatic calculation on FIB-4, irrespective of the presence of risk factors. It was therefore necessary to first assess the volume of patients eligible for Fibroscan to guarantee the feasibility of the pathway. Importantly, it has been reported that as much as 40% of patients in the primary care setting may have FIB-4 > 1.337. This prompted us to initiate the pathway with the higher threshold of 2.67. We believed this would be a safe way to start as available data show severe liver events mostly happen in the population with FIB-4 > 2.67 suggesting these patients are to be identified in priority38. We also considered patients with FIB-4 between 1.3 and 2.67, with a lower risk of developing severe liver events, could safely be re-assessed later if the pathway was successful. Importantly, 24 months after the launch of the pathway, the number of patients referred to VCTE could be easily evaluated on-site. We therefore decided to modify the care pathway, encouraging practitioners to request a VCTE as soon as FIB-4 ≥ 1.3. It should be noted that a threshold of 1.3 considerably increases the number of VCTE requested: in our study, if the threshold of FIB-4 ≥ 1.3 had been chosen, this would have represented 30% of patients with FIB-4 calculated between age 18 and 70. This increase of patients eligible for second-line testing expose to an increase in false-positives, estimated at 35% in the study by Kjaergaard et al.39, which could jeopardize the sustainability of the pathway, and which will need to be reassessed.

Despite the good negative and positive predictive value of FIB-4 at thresholds of 1.3 and 2.67 respectively in populations of patients with MASLD40, this test remains imperfect. In Hagstrom et al. study, 48.4% of cases of severe liver events were observed in the lower risk group of patients, with consistent FIB-4 < 1.333, indicating the need to improve NIT for screening CLD. However, in this low risk group, only 1% of the patients experienced severe liver events when it was 13% in the higher risk group of patient with FIB-4 consistently ≥ 2.67. Therefore, until better care pathways are recommended, the FIB-4 first appears as an opportunity to increase the chances to diagnose CLD before severe liver events.

Importantly, of 135 patients aged 18 to 70 with a first FIB-4 ≥ 2.67, 88 were excluded, mainly due to an obvious cause of false-positive FIB-4 or known liver disease. This highlights the need for medical assessment before referral to VCTE rather than an all-automated "reflex" pathway all from FIB-4 to specialized counselling.

Our study also showed the limits of automatic feedback based on unsolicited test. Of the 47 VCTE requests, only 15% originated from the prescribing physician, the other requests being triggered after an alert issued by liver specialists involved in the care pathway. This underlines low awareness of CLD among non-liver specialists. A survey of 178 French diabetologists showed that 59% underestimated the prevalence of CLD and only 29% knew FIB-441. Our study also pointed out a low patient awareness about CLD, as a significant number of patients lost to follow-up before and after VCTE. In an American study of over 10,000 patients with MASLD, 96% were unaware that they had liver disease42. It seems critical to raise awareness among patients at risk of advanced liver fibrosis as well as non-liver specialists about CLD.

The high rate of patients being lost to follow-up before VCTE could also be explained by issues to access to this test (need to return to the hospital, waiting time). In an attempt to overcome this difficulty, we recently implemented ELF® testing as a second-line test instead of VCTE (unpublished data), with a cut-off value of 9.8. This alternative had been used in a study by Srivastava et al., in which a primary care pathway to screen for advanced liver fibrosis was proposed using FIB-4 as the first-line test, and ELF® test in second line35. This approach reduced by a factor of four the number of "futile" referrals for specialist visits.

Our study provides new insights into the feasibility and performance of a systematized care pathway to screen for liver fibrosis in patients at risk, as few studies have been reported. Mansour et al. developed and evaluated a primary care pathway integrating a two-level liver fibrosis assessment (FIB-4 then VCTE) into the routine annual examination of all type 2 diabetic patients43, which significantly improved the identification of advanced liver disease in patients with type 2 diabetes. Another study is ongoing to evaluate a care pathway for liver fibrosis screening in routine care for type 2 diabetes44.

However, our study has limitations. We acknowledge the retrospective nature of our data collection exposes to the risk of missing data, particularly concerning certain risk factors, such as metabolic syndrome, alcohol consumption or OSA. Furthermore, the small number of patients eligible for second-line fibrosis testing makes it difficult to properly analyze the factors associated with significant fibrosis.

Concerning the number of subjects required for the multivariate model, we used Agesti's rule45 which requires a minimum of 10 cases for an independent variable in a logistic model (15/40 in our case) for this to be correct. We are well aware that this is a matter of debate, the results of our exploratory multivariate analyses should be interpreted with caution. More substantial studies are needed to confirm the factors associated with fibrosis.

Through the pathway, within 2 years, we identified 19 patients with significant fibrosis before onset of severe liver events. Retrospective collection of incident cirrhosis managed in our department before the implementation of the pathway is challenging, as different referral pathways co-exist. Therefore, we could not analyze changes in cirrhosis severity at diagnosis after the implementation of the pathway. Importantly, all patients identified with advanced fibrosis within the pathway were diagnosed before any liver event, allowing implementation of specialized follow up in order to prevent or delay liver events and start HCC screening. This is important to highlight as it has been reported diagnosis of cirrhosis is mostly done at the time of a complication11,12,13,14.

The implementation of care pathways to screen for liver fibrosis now seems a critical unmet need, especially as cost-effectiveness studies in other care systems have confirmed that such pathways reduce costs compared with an "all-referred" strategy46,47. It is to be hoped that early referral of patients with advanced fibrosis will enable the implementation of strategies to limit the occurrence of life-threatening complications, and screening for HCC, associated with early diagnosis and improved survival. Cost-effectiveness studies carried out in the French healthcare context are ongoing (Clinical trial NCT05880173) and will be useful to legitimize the dissemination of these pathways throughout the country.