Prevalence of liver cirrhosis in individuals with hepatitis B virus infection in sub‐Saharan Africa: Systematic review and meta‐analysis

Abstract Background & Aims Chronic hepatitis B virus (HBV) infection accounts for 30%‐50% of cirrhosis related deaths in sub‐Saharan Africa (SSA). Since HBV‐related cirrhosis is an indication for immediate antiviral therapy and cancer surveillance, we aimed to estimate the prevalence of cirrhosis among treatment‐naïve patients with chronic HBV infection in SSA. Methods We performed a systematic review of published articles which evaluated liver fibrosis stage among treatment‐naïve HBV‐infected individuals who presented to care in SSA. Our primary outcome was the prevalence of cirrhosis in HBsAg‐positive persons, which was estimated using random‐effects meta‐analysis. Risk factors for cirrhosis were explored using subgroup‐analyses and multivariable meta‐regression. Results Of 2129 articles identified, 17 met our eligibility criteria. The studies described 22 cohorts from 13 countries, including 13 cohorts (3204 patients) with chronic HBV mono‐infection and nine cohorts (688 patients) with HIV/HBV‐coinfection. Liver fibrosis was assessed using transient elastography (10 cohorts), APRI score (11 cohorts), and Fibrotest (one cohort). The pooled prevalence of cirrhosis was 4.1% (95% confidence interval [CI] 2.6‐6.4) among studies from primary care facilities or general population, compared to 12.7% (95% CI 8.6‐18.3) in studies performed in referral or tertiary care facilities (adjusted odds ratio 0.29, 95% CI 0.15‐0.56). We found no association between cirrhosis and age, gender, fibrosis test used or HIV‐coinfection. Conclusions Depending on the setting, between 4% and 13% of HBV‐infected individuals in SSA have cirrhosis and need immediate antiviral therapy. These estimates should be considered when planning HBV treatment strategies and resource allocation.


| INTRODUC TI ON
Chronic hepatitis B virus (HBV) infection is the leading cause of liverrelated deaths worldwide and accounts for around one third of all deaths attributed to cirrhosis. In sub-Saharan Africa (SSA), where most infections occur at birth or during early childhood, 30%-50% of all cirrhosis-related deaths can be attributed to chronic HBV infection. 1 Antiviral treatment of patients with chronic HBV infection prevents the progression of cirrhosis 2 and reduces the incidence of hepatocellular carcinoma (HCC) in patients with and without HIV infection, 3,4 thereby decreasing HBV-related morbidity and mortality.
Although treatment eligibility criteria for individuals without cirrhosis vary across guidelines and their interpretation is often subject to debate, the presence of cirrhosis is a clear indication for antiviral therapy and ultrasound-based HCC screening. 5,6 Large studies of HBV-monoinfected 4,7 and HIV/HBV-coinfected individuals 3 8 In order to inform global resource allocation strategies, we performed a systematic review and meta-analysis to assess the prevalence of cirrhosis among treatment-naïve HBV-infected individuals in SSA.

| Search strategy and selection criteria
We performed a systematic search of PubMed, EMBASE, African Index Medicus (via Global Index Medicus), Cochrane Central Register of Controlled Trials (CENTRAL), Web of Science and CINAHL to identify articles, which reported the proportion of treatment-naïve HBsAg-positive individuals who presented with cirrhosis (see appendix for the detailed search strategy). We considered published papers from inception until June 25, 2020 with original data in which liver fibrosis was assessed systematically in cohorts from sub-Saharan Africa. We considered all types of liver fibrosis assessments, including liver biopsy, transient elastography and serological scores such as the AST-to-platelet ratio index (APRI), fibrosis-4 (FIB-4) and Fibrotest. Studies which selected participants based on symptoms or stage of liver disease, and studies which assessed less than 20 HBV-infected individuals were excluded. We did not apply any restrictions on time frame, age of study participants or language of publication. Title, abstract and full text screening as well as data extraction were performed independently by four investigators (DW, BS, CB and GW) and discrepancies were discussed until a consensus was reached. Risk of bias was assessed using a critical appraisal tool for prevalence studies. 9 The reporting of the study follows the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines. The protocol for this systematic review and meta-analysis was registered with the International Prospective Register of Systematic Reviews (PROSPERO, registration number CRD42020180422).

| Outcomes and definitions
The primary outcome was the proportion of treatment-naïve HBVinfected individuals with cirrhosis, whereas the prevalence of significant liver fibrosis was assessed as a secondary outcome. We used the transient elastography (TE) cut-offs reported by the authors which varied between studies. One study 10 used a transient elastography cut-off of >9.3 kPa to define significant fibrosis; as this value is above the WHO-recommended threshold for significant fibrosis in HBV infection (>7.5-8.5 kPa), 5 we re-categorized individuals with a value >9.3 kPa into the cirrhosis group. In another study, a substantially lower than the WHO-recommended threshold was used

Key points
• Liver cirrhosis is the most important driver of hepatitis B virus (HBV)-related morbidity and mortality.
• Individuals with liver cirrhosis should receive immediate antiviral treatment with tenofovir and screening for hepatocellular carcinoma (HCC).
• In this meta-analysis of 17 studies evaluating liver fibrosis stage among treatment-naïve HBV-infected individuals in sub-Saharan Africa (SSA), the pooled prevalence of liver cirrhosis was 6.7% and differed significantly between primary care facilities (4.1%) and tertiary care or referral hospitals (12.7%).
• Based on our findings, an estimated 2.5 of 60 million individuals with chronic HBV infection in SSA have liver cirrhosis and need antiviral therapy and surveillance for liver cancer.
to define significant fibrosis (≥5.9 kPa) 11 ; since the authors provided estimates based on transient elastography values ≥7.6 kPa, we used these numbers to define significant fibrosis. APRI scores above 1.5 were categorized as significant fibrosis, and above 2.0 as cirrhosis. 5 Studies, which reported data from more than one country or more than one patient population were separated into cohorts. If more than one method of fibrosis assessment was available, we prioritized transient elastography, followed by APRI and FIB-4. For each cohort, only one method of assessment was used to calculate the prevalence, and only the first fibrosis assessment (prior to any antiviral treatment) was considered when repeated measurements were available.

| Data analysis
The prevalence of cirrhosis and significant fibrosis was estimated using random intercept logistic regression and logit transformed proportions. The 95% confidence intervals (CI) were derived using the exact method, and statistical heterogeneity was assessed using the Cochrane's Q and the I 2 statistic. To explore potential causes of heterogeneity, we performed subgroup analyses according to the test used (transient elastography, APRI, FIB-4 and Fibrotest ® ), cohort category (referral/teaching hospital and primary care/general population), and whether HIV co-infection was present or not. In addition, we performed sensitivity analyses including only studies which used TE, the test with highest sensitivity and specificity for detecting liver fibrosis and cirrhosis. 5 A sensitivity analysis restricted to studies performed in HBV-monoinfected populations was performed to assess the impact of cohort categories in this subpopulation. Risk factors for cirrhosis were modelled using univariable and multivariable meta-regression. All analyses were performed using the meta 12 package for R version 4.0.

| Assessment of study quality
Most studies were performed among a sample of individuals representative of the target population of the specific study, and all studies measured the outcome in a standardized and reliable way within their study population. Whereas most studies reported a systematic screening of HIV among all participants, one study did not state whether HIV was screened for. 20 However, some studies failed to describe how patients were selected into the studies, and only a minority of studies performed random sampling of eligible patients.
In addition, the sample size was small in most studies, and the main outcome was incompletely assessed in three studies. Whereas 50% of all studies relied on transient elastography to assess liver fibrosis, the less sensitive APRI method was mainly used in studies of HIV/ HBV co-infected patients (Table S2).

Among treatment-naïve HBV-infected individuals in sub-Saharan
Africa, the prevalence of cirrhosis was 4.1% (95% CI 2.6-6.4) in studies performed at primary care facilities or as part of general population screening, compared to 12.7% (95% CI 8. 6-18.3) in studies performed at referral or teaching hospitals. Substantial heterogeneity was observed between studies, and the majority of them reported The studies included varied in terms of the methods used to assess liver fibrosis and cirrhosis. Whereas there was substantial heterogeneity among studies reporting on cirrhosis, we found no evidence that the method of cirrhosis assessment contributed significantly to the estimates. This finding was surprising, since the sensitivity of APRI to diagnose cirrhosis is considered low, leading to many cases of cirrhosis being missed. 5   Africa have cirrhosis and need immediate antiviral therapy and HCC screening. 8,31 More efforts are needed to identify HBV-infected individuals in SSA, and to systematically assess the extent of HBVassociated liver disease. Since widely available tests such as the APRI score miss a significant proportion of cases of cirrhosis, there is an urgent need for access to more accurate cirrhosis assessment tools such as transient elastography in order to identify patients for whom HBV therapy and surveillance are crucial.

ACK N OWLED G EM ENTS
We thank Doris Kopp from the Institute of Social and Preventive Medicine, University of Bern, Switzerland for her expert help with the literature searches.

BS reports support to his institution for travel grants from Gilead
Sciences. AR reports support to his institution for advisory boards and/or travel grants from Janssen-Cilag, MSD, Gilead Sciences, and Abbvie, and an unrestricted research grant from Gilead Sciences. All remuneration went to his home institution and not to AR personally, and all remuneration was provided outside the submitted work.
GW reports support to his home institution for advisory boards and/ or travel grants from MSD, Gilead Sciences and Abbvie, and an un-