Long-term effect of chronic hepatitis B on mortality in HIV-infected persons in a differential HBV transmission setting

Globally in 2020, 820,000 individuals died from chronic hepatitis B virus infection (HBV), compared to 680,000 people dying from HIV disease [1], and approximately 2.6 million HIV-infected individuals are co-infected with HBV [2]. Although HIV and HBV can share similar transmission routes [3], the contribution of transmission routes to infection burden may differ by geographical region [2]. For example, studies in East Africa have shown that scarification and blood transfusions are important risk factors for HBV transmission [4] and that HBV is often acquired at a younger age (before 5 years) compared to other regions in Africa. In contrast, HIV transmission in East Africa is driven by heterosexual sex in adulthood [2]. Low levels of HIV transmission among younger age groups can be explained by the recent introduction of antivirals in Africa [5] for prevention of mother-to-child transmission. The pathophysiology of HBV and the duration of the infection in East Africa, is an important factor to suspect a different risk of mortality that is assumed to be different of that in developed countries where, advanced health infrastructure for vaccination, screening and treatment make HBV acquisition in adulthood through sex and injecting drug use [6].

Several studies have found reduced survival [7‐10] of HBV and HIV co-infected individuals, while other large studies [11‐13] found no effect. These discrepancies, along with the potential lack of generalizability of previous studies for the eastern African population, raise reasonable question on the effect of HIV–HBV co-infection on mortality risk, warranting local-level investigation for more targeted policy guidance.

We conducted this study in Rwanda, an East African country where identification of people with HBV–HIV co-infection began in 2016 after a long period of uncertainty about co-infection and its impact on population mortality. Building on the previous work by Umutesi et al. (2021), we focus on HIV-infected people, and take into account the introduction of universal, early and immediate antiretroviral therapy (ART) [14] that raised ART coverage to 97.5% in Rwanda [15], and the prioritization of TDF as a first-line treatment [16]. There is little evidence of the effect of HBV on mortality in HIV-infected individuals, particularly in large cohorts with routine HIV care, routine free HIV/HBV testing, high antiretroviral treatment coverage and in the context of a differential HBV transmission route [2]. Using a large HIV-infected individual cohort in sub Saharan Africa, we examined the association between chronic HBV infection and HIV mortality among HIV-infected individuals.

Our results show that chronic HBV co-infection is not associated with time to mortality in HIV-infected individuals in Rwanda. We consider the following are reasons for minimal effect of HBV–HIV co-infection on increased mortality. First, by the end of 2007, about 10 years before participants in the current study could be tested for HBV, 80% of eligible HIV-positive Rwandans were on ART [26] a primary reason for a possible reduction in overall HIV-related mortality in Rwanda. We further note that 17,947/18,459 (97.2%) of our study population were on highly active antiretroviral therapy (HAART), and of these 13,339 (72.3%) were on a TDF-based regimen. In fact, TDF has been used in first-line ART in Rwanda since 2009 and various research studies have shown that it reduces HBV replication [27], thereby improving HIV outcomes, including mortality. Second, 100% of our population was on Lamividine (also known as 3TC, Epivir), another drug that reduces HBV replication [28]. Third, we consider our study population to be individuals who have succeeded or survived a past where the risk of infection and mortality was higher due to poor health infrastructure and insufficient knowledge of HIV and HBV in the population. In fact, African populations have carried a huge proportion of HBV infection for decades [29] and it is therefore indisputable that chronic HBV is a major risk factor for mortality in untreated individuals [30], and anti-HBV drugs have recently been introduced in African countries [31]. Consultation with traditional healers in the past is another reason that may increase the progression of the disease and the many deaths of HBV chronically infected individuals. Fourth, we suspect differences in mortality can be observed over longer periods of follow-up beyond the follow-up time after HBV diagnosis allowed by our study. Finally, our population displayed good clinical control of HIV infection as measured by CD4 cell counts, HIV RNA load, adherence to antiretroviral therapy, comorbidities including TB and WHO-HIV clinical stage. Nevertheless, a meta-analysis of studies conducted before the introduction of HAART found an increase in mortality among HBV–HIV co-infected individuals compared to mono-infected individuals [7]. However, all studies included in the review were from developed countries in Europe and the United States, and no studies were conducted in Africa. In contrast, a cohort study of HIV-infected persons enrolled between 1984 and 2003 in Greece found no impact of HBV co-infection on all-cause mortality. In another recent study conducted by Thornton [10], an increase in all-cause mortality following HBV and HIV co-infection was found in England. Higher mortality among co-infected individuals in developed countries may not translate to higher mortality in developing countries since the demographic distribution of co-infected individuals and access to prevention and treatment services between the two contexts is importantly different. For example, co-infected individuals from developed countries are more likely to be intravenous drug users and men who have sex with men, with potential differences in adherence to treatment contributing to higher mortality [10].

Although the prevalence of HBV/HIV co-infection is estimated to be over 28.4% in some African countries [32], a number of mortality studies have been conducted on specific populations. For example, a clinical trial conducted in Côte d’Ivoire by Kouamé in 2021 focused on people with low CD4 cell counts and found a higher impact of HIV and HBV co-infection on mortality [9]. However, in countries or regions where HAART is universal, this risk could be reduced, as in the case of an HIV clinic in Nigeria in 2014, where the study occurred before the implementation of TDF and HAART was a protective factor against mortality risk [12]. In the same vein, in countries where HAART was introduced in 1990s, such as in Europe [33], the effect of chronic HBV on liver-related mortality has been demonstrated but without any impact on progression to AIDS, immunological or virological response to HAART [34]. These scenarios could apply to Rwanda where, in 2020, 97.5% of HIV-infected persons were on HAART [15].

Independently to HBV co-infection status, our study showed that other factors increased the risk of mortality in HIV-infected persons, including being male. However, this finding is not consistently described in the literature where some have described a higher risk of mortality in women [35] while others a higher mortality risk among men [36, 37]. The increased case mortality risk in HIV-infected men may be related to a later diagnosis [38], later clinic attendance or lack of adherence to ART, as has been shown to be common among men in sub-Saharan Africa [39]. Age over 55 years versus 15–24 years and WHO-HIV stages 3 and 4 were predictive of mortality, which may be related to long-term HIV infection with more severe signs and symptoms or other comorbidities that, in turn, tend to increase with age [40]. The province of residence was also associated with the time to mortality, where residing in southern, northern and western province as compared to capital Kigali were highly associated with time to mortality. Previous studies have shown a higher risk of infection in provinces bordering neighboring countries [41] and a higher case fatality rate is likely. Other reasons could be related to health services centralized in Kigali and the increased likelihood of consulting traditional healers in certain peripheral regions of the county. Another predictor of mortality was alcohol consumption and its increased effect on liver disease was demonstrated [42, 43]. Finally, one co-variate, time since starting treatment, showed that taking ART for at least 7 years was associated with a lower risk of mortality, suggesting the effectiveness of ART [44].

Limitations of our study include a relatively short follow-up period that does not allow detection of a difference in mortality between the co-infected groups, because the development of cirrhosis and hepatocellular carcinoma (HCC) is too slow in patients on TDF [45]. Also the lack of quantification of the risk of HCCand liver-related mortality. As the data could not allow us to follow the study participants from the age of onset of HBV infection and that the immortal time bias is possible, we recommend further analysis to interpret the hazard ratio by age of acquisition of infection. We believe that prolonged duration between infection and diagnosis, an artefact of lack of prior health infrastructure, meaning a substantial proportion of this population died prior to the time diagnosis was offered and could not be included in our study. It is therefore unlikely that our results show a lethal effect of hepatitis B in the study population. We recognize the limitation of not being able to assess the actual survival time of those lost to follow-up, and lack of information on the completeness of EMR mortality data and other supporting sources, which may lead to reporting bias and thus conservative interpretation of the results of this study. Finally, considering that HBV viral load activity at the start of ART may also influence the difference in mortality [46, 47], we acknowledge that the lack of HBV DNA testing may limit our results.