Hepatitis B is one of the most widely distributed pathogens with over 300 million people chronically infected. In the World Health Organization (WHO) Africa region, an estimated 82 million people are living with chronic hepatitis B, representing the highest global prevalence. Hepatitis B causes two main health problems: extensive liver damage (cirrhosis) and liver cancer (hepatocellular carcinoma, HCC). HCC has very poor outcomes, with a multicentre study showing a median survival of just 2 months at diagnosis, and median patient age of 46 years in sub-Saharan Africa.

Hepatitis B can be effectively treated with antiviral therapy. Tenofovir, a nucleoside analogue, is widely available for 2.40 USD per month (Clinton Health Access Initiative 2022), and efficiently suppresses viral replication, improves quality of life, reverses advanced liver fibrosis, and reduces the risk of HCC. The main challenge in clinical practice is to decide which patients need treatment. Most patients living with hepatitis B will not require treatment at initial assessment and over-treatment could be detrimental to stretched health system or represent costs that cannot be borne by patients. Treatment is needed for patients with evidence of advanced liver fibrosis or those at risk of developing liver disease, represented by raised liver enzymes and elevated hepatitis B virus level in the blood.

Our study aimed to answer the question: can we diagnose advanced liver fibrosis in the WHO Africa region, using routinely available and low-cost blood tests?

Several methods to diagnose liver fibrosis are available. First, ultrasound to visualise distortion of the liver appearance or signs of portal hypertension such as abdominal fluid (ascites) or an enlarged spleen (splenomegaly). These are late signs, insensitive for diagnosing early cirrhosis. Waiting to start treatment at such an advanced phase will increase the risk of liver-related death. Second, a liver biopsy can be used to obtain a sample of liver tissue to examine for histological evidence of liver fibrosis and inflammation. Limitations include variability between pathologists, sampling error in the context of insufficient sampling or inhomogeneous fibrosis, and the risk of complications. In many centres in Africa limited access to pathologists and costs restrict access to this procedure.  Third, liver stiffness measurement by transient elastography is a non-invasive technique to estimate liver fibrosis that has largely replaced liver biopsy in high-income countries. Increased liver stiffness effectively prognosticates risk of decompensation, HCC and liver-related death. The main limitations are high capital and maintenance costs in resource-limited settings and false-positive results in situations such as heart failure or acute hepatitis. Finally, blood tests can be used. Several have been developed and validated in European and Asian studies where combinations of platelets, liver enzymes (ALT, AST, GGT) and other combinations of biomarkers and patient age are used to estimate probability of advanced liver fibrosis and cirrhosis.

In 2015, the WHO released the first guidelines for treating chronic hepatitis B. These recommended diagnosing and treating patients with cirrhosis diagnosed using an APRI (AST to platelet ratio index) score of >2.0, a very high bar. Studies from The Gambia and Ethiopia have since indicated that this threshold of 2.0 significantly underdiagnosed cirrhosis in African hepatitis B patients.

Starting as a discussion on the sidelines of the Conference on Liver Disease in Africa, several clinician researchers from across Africa decided to combine our data from studies of hepatitis B patients in community and hospital settings. Hepatitis B in Africa Collaborative Research Network (HEPSANET) was born (www.hepsanet.org). We assessed the performance of the biomarkers APRI, GPR (GGT to platelet ratio), FIB-4 and ALT alone, using liver stiffness as a reference standard for significant fibrosis (>7.9kPa) and cirrhosis (>12.2kPa). We performed a systematic review to identify any further published evidence and contacted colleagues working in this area. Finally, all eligible cohorts contributed data to this individual patient data meta-analysis representing over 3500 patients from 12 centres in Africa. We used a Bayesian model taking account of differences between patient populations and centres.

Figure: Sensitivity and specificity for the diagnosis of liver stiffness >12.2kPa (cirrhosis) among patients with hepatitis B using the APRI score

We confirmed earlier findings that showed a poor sensitivity using the WHO recommended APRI threshold of 2.0; only 16.5% of patients with cirrhosis (liver stiffness >12.2kPa) had an APRI score above this threshold and would receive treatment by applying the WHO 2015 guidelines. We developed new APRI thresholds with improved sensitivity: at a “rule-in” threshold of 0.65 the sensitivity was improved to 56% with high specificity; at a “rule-out” threshold of 0.36 sensitivity was 81% but with a lower specificity (Figure).

Using a low-cost biomarker such as APRI is imperfect, but in many settings it is the only available and affordable test to diagnose liver fibrosis and cirrhosis. Setting the correct APRI threshold must balance the risk of under-diagnosis, resulting in missed opportunities to prevent liver-related mortality, and over-diagnosis that could stretch health systems. Thus, the debate about the correct APRI threshold for treatment must include patient advocates, economists, policymakers and clinicians. The HEPSANET group is advocating for better care for millions of patients living with chronic hepatitis B in Africa. Our findings are timely since WHO guidelines for hepatitis B are currently undergoing revision. A lower APRI threshold is needed and the revised guidelines should take account of our new findings. Patients and clinicians in Africa need simple, accessible guidelines that offer improved access to treatment now.