Increased Prevalence of Liver Fibrosis and HIV Viremia among Patients with HIV, HBV, and Tuberculosis in Botswana

People with concomitant human immunodeficiency virus (HIV) and tuberculosis (TB) have an increased risk of hepatotoxic reactions due to antiretroviral therapy (ART) and anti-TB therapy (ATT). Concomitant hepatitis B virus (HBV) in these patients may lead to poorer health outcomes. To assess liver enzyme levels and immune response in adults with HIV, HBV, and TB, data from 300 antiretroviral-naïve people living with HIV (PLWHIV) were analyzed. The prevalence of HIV/HBV (cHIV/HBV) and HIV/TB (cHIV/TB) was 28% (95% CI: 23.0–33.4) and 10% (95% CI: 6.8–14.0), respectively. HIV/HBV/TB (cHIV/HBV/TB) prevalence was 5.3% (95% CI: 3.1–8.5). There was a statistically significant difference between the groups of participants in HIV viral load (p = 0.004), hemoglobin levels (p = 0.025), and body mass index (p = 0.011). A larger proportion of cHIV/HBV/TB participants (37.5%) had an aspartate aminotransferase to platelet ratio index (APRI) score ≥0.5 (p = 0.013), a lower cutoff for significant liver fibrosis. Immunological non-responders (CD4+ T-cell count <20% gain and HIV viral load <400 copies/mL at 6 months) were observed in all groups except those with cHIV/TB. Our findings support the need to screen for infections that could cause excessive liver damage prior to ATT or ART initiation, such as HBV.


Introduction
Globally, approximately 251,000 tuberculosis (TB)-associated deaths were recorded among people living with human immune deficiency virus (PLWHIV) in 2018 with a high proportion of people with

Participant Baseline Demographics and Clinical Characteristics
At baseline, there were no statistically significant differences across the study groups in median CD4+ T-cell count, ALT, AST, APRI, or FIB-4 levels. However, there was significant statistical difference in median hemoglobin levels (p = 0.025), median body mass index (BMI) (p = 0.011), and log HIV viral load (p = 0.004) at baseline (see Table 1).

Participant Baseline Demographics and Clinical Characteristics
At baseline, there were no statistically significant differences across the study groups in median CD4+ T-cell count, ALT, AST, APRI, or FIB-4 levels. However, there was significant statistical difference in median hemoglobin levels (p = 0.025), median body mass index (BMI) (p = 0.011), and log HIV viral load (p = 0.004) at baseline (see Table 1). There was a statistically significant difference in the proportion of participants with previous exposure to HBV (HBcAb), (p = 0.001). There is evidence of previous exposure to HBV in PLWHIV and participants with cHIV/TB. However, the highest proportions were observed in participants with cHIV/HBV/TB (82.3%) and those with cHIV/HBV (69.1%), which is not surprising as some of the participants were also HBsAg positive. Five participants-4 with cHIV/HBV and 1 with cHIV/HBV/TB-were HBsAg positive but HBcAb negative (data not shown). There was no significant difference between participants with cHIV/HBV who had chronic infection versus OBI for all variables analyzed (data not shown).
These participants had APRI scores >1.0, which is a low cutoff for cirrhosis, a high cutoff being 2.0. The lowest APRI scores were among mono-infected HIV participants (BB1-BB3). Participants with cHIV/HBV had the highest APRI scores (BB8 and BB6). Participants BB6-BB10 had APRI scores of greater than 2.0 indicating significant liver cirrhosis. Participant BB6, deceased, had the highest APRI scores indicating severe liver damage (Table 3).

Immunological Response
All

Discussion
In this study conducted in a cohort of PLWHIV initiating ART in Botswana, we found a cHIV/HBV/TB prevalence of 5.3% with cHIV/HBV/TB participants having the lowest hemoglobin level and high HIV viral loads. Participants with cHIV/TB had a low BMI. Liver damage was more prevalent among participants with cHIV/HBV/TB at a low cutoff for fibrosis. In addition, immunological non-response was observed in all groups of participants at 17-20% apart from those

Discussion
In this study conducted in a cohort of PLWHIV initiating ART in Botswana, we found a cHIV/HBV/TB prevalence of 5.3% with cHIV/HBV/TB participants having the lowest hemoglobin level and high HIV viral loads. Participants with cHIV/TB had a low BMI. Liver damage was more prevalent among participants with cHIV/HBV/TB at a low cutoff for fibrosis. In addition, immunological non-response was observed in all groups of participants at 17-20% apart from those with cHIV/TB. Concomitant HIV, HBV, and TB is poorly studied in sub-Saharan Africa; however a prevalence of 7.5% has been observed in China [11]. Our study has therefore revealed a possibly overlooked health problem that should be studied more in our settings. People with cHIV/HBV/TB may require unique patient care due to the considerable burden on the liver from ATT, HBV, and ART. Our estimated prevalence of cHIV/HBV in Botswana is 28%, including both chronic infections and OBI, which is comparable to prevalence rates that have been reported in Botswana in similar populations [6,9,26]. The prevalence of cHIV/HBV in other sub-Saharan African countries was shown to range from 0% to 27%, with this large range being affected by the management of HBV in different countries [27]. A cHIV/TB prevalence of 10% is reported here, slightly lower than what was previously reported in a previous analysis of the Bomolemo cohort as we did not exclude patients with HBV [20]. Evidence of cHIV/HBV/TB in our participants can therefore be used to propose guidelines for screening PLWHIV for HBV prior to ART and ATT initiation. The five participants who were HBsAg positive but HBcAb negative are suspected to be cases of recent infections; however, this should be confirmed with immunoglobulin M (IgM) testing. Several other reasons, such as the presence of diagnostic escape mutants for the core gene and immunotolerance of core antigen have been discussed to explain these serology results [28].
Baseline participant demographics revealed that there were more females enrolled in the study, which could explain the difference in proportions in gender for PLWHIV and participants with cHIV/HBV. However, this trend was not observed for participants with cHIV/TB, which is also a global trend. In Botswana, incident TB is more prevalent in males than in females [20]. The difference in TB susceptibility between the genders has been discussed to be possibly due to behavioral, physiological, and genetic differences as reviewed by Nhamoyebonde and Leslie [29]. This gender bias is attributed to risk factors such as smoking [30] and protective female sex hormones that modify immune responses and regulate immune cell functions for a controlled mycobacterial infection [31]. Participants with cHIV/HBV/TB exhibited high HIV viral load and low hemoglobin levels. Co-infections are well known to result in worsened prognosis than mono-infections, which is no exception is this study. Concomitant HIV, HBV, and TB can bring in more complications in this case possibly resulting in late viral suppression and a high viral set point. In a previous analysis of the Bomolemo cohort, there was no association between HIV viral load and the risk of TB [20]. Therefore, the association between HIV viral load and multi-infection could be due to the high burden of HBV infection in this cohort. A larger South African study, however, showed HIV viral load as a risk factor for TB [32]. The present study observed the lowest median hemoglobin levels among participants with cHIV/HBV/TB possibly due to TB. Low hemoglobin level has been shown to be a predictor of incident TB in Bomolemo participants [20] and in South African patients during long-term ART [33]. The association of HBV and low hemoglobin levels should be explored further. Participants with cHIV/TB had the lowest BMI, which could be attributed to nutritional deficiency due to TB infection. Nutritional deficiency, resulting in low BMI, was found to be high in pulmonary TB patients [34], while an overweight BMI is shown to significantly reduce risk for TB [35].
With regards to liver enzyme levels, the majority of our study participants had normal liver enzyme levels at baseline. A high incidence of abnormal liver function tests (LFTs) was observed among participants with cHIV/HBV/TB in China [11]. In contrast, in our study, ALT and AST data were only available at baseline, and there was no statistically significant difference in the median liver enzyme levels between all groups. However, there is a possibility of advancing to more toxic grades, which could have been assessed with longitudinal follow-up of liver enzyme tests as more individuals initiate ART. Fibrosis grading in our study showed that participants with cHIV/HBV/TB had a higher proportion of patients with APRI score of greater or equal to 0.5, a low cutoff point for significant liver fibrosis. At this cutoff point, the sensitivity and specificity for the detection of fibrosis is 78% and 68%, respectively. At a cutoff of 1.5, the sensitivity and specificity for the detection of fibrosis is 36% and 92%, respectively [25]. These non-invasive tests are utilized in most resource-limited settings but may miss some cases. Local ALT and AST reference ranges are yet to be defined specifically for men and women in our setting. A recent study, using African reference ranges versus American reference ranges emphasized the need to have specific ranges for different populations to avoid overestimation or underestimation of abnormalities [36]. FibroTest and FibroScan may be preferred over APRI and FIB-4 but may not be suitable for resource-limited settings. The upper cutoff was not pursued due to the modest sample size. Other studies have revealed an association between concomitant HIV/HBV and liver fibrosis [37,38], while one has shown TB to be a risk factor for liver fibrosis [39]. In patients suspected of having cirrhosis (APRI > 1), all had high HIV viral loads and those with HBV had detectable HBV viral loads at baseline, suggesting active HBV replication resulting in liver damage. The association between HBV and HIV viral load with liver fibrosis has been described previously [40].
We also assessed immune non-response in these patients, defined as a less than 20% gain in CD4+ T-cell count at 6 months of follow-up with viral suppression at less than 400 copies/mL. The lack of immune non-responders in the HIV/TB cohort could be due to the general small sample size as well as our definition of non-responders. There is currently no consensus on the definition of INRs, defined as reviewed by Yang et al. [41]. In light of this, cHIV/HBV was previously shown to result in slow CD4+ T-cell recovery in these patients compared to PLWHIV only [9,42]. TB also results in slow CD4+ T-cell recovery even after ART initiation [43][44][45]. In one South Africa cohort of 15,646 adults, there was no difference in immune restoration between participants with cHIV/TB and those without TB after ART initiation [46]. The differences in observations from these studies could be attributed to factors such as sample size, immune reconstitution definition, and cofounding factors among others.
Our study bears the limitation of a lack of longitudinal liver enzyme tests. This limits our analysis in terms of assessing for incidence of liver damage in each group of participants, particularly with initiation of ART. A larger population size would have also allowed for other fibrosis cutoff points in the analysis.

Conclusions
To our knowledge, this is the first study in Botswana to report the prevalence of cHIV/HBV/TB prior to ART initiation. Participants with cHIV/HBV/TB had low hemoglobin levels and high HIV viral loads. Patients with undiagnosed HBV who initiate ATT before ART may have poorer viral hepatitis health outcomes such as developing chronicity, advancing to liver fibrosis, cirrhosis, and possibly hepatocellular carcinoma without any intervention. The study has revealed the need to screen for HBV in people with cHIV/TB infection, as this will lead to better patient management.