Journal of Infection (2014) 69, 481e489
www.elsevierhealth.com/journals/jinf
Long-term responses to first-line antiretroviral therapy in HIV and hepatitis B co-infection in Ghana Fred Stephen Sarfo a,b, Adetayo Kasim c, Richard Phillips a,b, Anna Maria Geretti d, David R. Chadwick e,* a
Komfo Anokye Teaching Hospital, Kumasi, Ghana Kwame Nkrumah University of Science and Technology, Kumasi, Ghana c Wolfson Research Institute for Health and Wellbeing, Durham University, Stockton-on-Tees TS17 6BH, UK d Institute of Global Health, University of Liverpool, L69 7BE, UK e The James Cook University Hospital, Middlesbrough, TS4 3BW, UK b
Accepted 12 June 2014 Available online 27 June 2014
KEYWORDS Hepatitis B; HIV; Clinical outcomes; Lamivudine; Antiretroviral therapy
Summary Objectives: To observe the long term response to first-line antiretroviral therapy (ART) in HIV and hepatitis B virus (HBV) co-infected patients in Ghana and explore predictors of poor clinical outcomes. Methods: Retrospective cohort study of hepatitis B surface antigen (HBsAg) positive and negative patients receiving predominantly NNRTI-based ART with lamivudine plus either zidovudine or stavudine for up to seven years. Cox proportional hazards and Kaplan Meier survival analyses compared clinical outcomes and identified baseline characteristics predictive of poor outcomes. A mixed effects model compared changes in CD4 counts. Results: A total of 299 HBsAg-positive and 1869 HBsAg-negative patients started ART between 2004 and 2008. Over a median 35 months of follow-up, HBsAg-positive patients were more likely to die or default care than HBsAg-negative patients, aHR 1.36 (95% CI, 1.03e1.80). HBsAg-positive patients were also more likely to develop Grade 3/4 hepatotoxicity than HBsAg-negative patients, HR 1.99 (1.16e3.40) on survival analysis. There was no significant difference in CD4 responses between HBsAg-positive and HBsAg-negative patients.
* Corresponding author. Centre for Clinical Infection, The James Cook University Hospital, Marton Road, Middlesbrough TS4 3BW, UK. Tel.: þ44 01642 854429; fax: þ44 01642 854017. E-mail address: [email protected] (D.R. Chadwick). http://dx.doi.org/10.1016/j.jinf.2014.06.012 0163-4453/ª 2014 The British Infection Association. Published by Elsevier Ltd. All rights reserved.
482
F.S. Sarfo et al. Conclusions: HBsAg-positive patients are at significantly increased risk of adverse clinical outcomes after starting ART. Further studies are warranted to evaluate whether these risks remain now that tenofovir is becoming routinely available in Ghana. ª 2014 The British Infection Association. Published by Elsevier Ltd. All rights reserved.
Introduction Approximately 17% of HIV-infected individuals in Ghana are also chronically infected with hepatitis B virus (HBV).1 In contrast to developed countries in sub-Saharan Africa most HBV infection is acquired during early childhood, so patients with HIV/HBV co-infection have usually had chronic HBV infection for many years by the time they present with (or are tested for) HIV infection. Patients with HIV/HBV co-infection have accelerated liver disease2e4 and are more likely to develop hepatotoxicity due to antiretroviral therapy (ART).5,6 The effect of HBV infection on HIV progression or response to ART is less clear and a number of studies have shown contrasting effects of HBV coinfection on HIV virological response or CD4 cell increases.7e12 Data on the longer-term effectiveness of ART and risk of ART-induced hepatotoxicity among HIV/HBV co-infected patients are lacking, particularly in high prevalence areas for HBV. The gold standard treatment for patients with HIV/HBV co-infection is ART containing tenofovir, although until recently this drug has not been widely available in most sub-Saharan African countries. First-line ART in such regions has normally comprised a non-nucleoside reverse transcriptase inhibitor (NNRTI) in combination with two nucleoside reverse transcriptase inhibitors (NRTI) including lamivudine, which is the sole NRTI with activity against HBV. Hence in many parts of Africa, where testing for HBV was not routine in HIV clinics or tenofovir was not available, co-infected patients have started such ART combinations with the inevitable development of HBV resistance to lamivudine,13,14 and some have developed complications including hepatitis flares and progression of liver disease due to loss of HBV virological control. The objectives of this retrospective study were to explore the associations between HIV/HBV co-infection and the risks for clinical events, particularly death, loss to follow up and severe hepatotoxicity, after initiation of ART containing lamivudine as the only NRTI active against HBV. We also modelled CD4 recovery on ART over the longterm in a cohort of Ghanaian HIV infected patients with and without HBV co-infection.
Patients and methods Study population and antiretroviral therapy From 2004, patients referred to the HIV clinic in KATH, Kumasi have been treated as part of the National AIDS Control Programme. Patients were referred from a large area of central and northern Ghana. Clinical data are recorded in patients’ case notes. Routine laboratory tests include full blood count, renal and liver function tests and
CD4 counts at baseline, with blood counts, liver function and CD4 counts tested every 6 months subsequently. HIV viral load was not tested routinely and testing for HIV-2 and HBV co-infection was performed only in limited circumstances. HBV testing at the clinic was not conducted as part of routine care, however was performed intermittently as part of research projects where all patients attending clinic were tested for a period of time, using either the Bioelisa HBsAg Colour (Biokit Ltd, UK), or the Murex HBsAg (Abbott Diagnostics, UK) assays on fresh or stored serum samples, and later confirmed using the automated chemilumiscent immunoassay Architect HBsAg (Abbott Diagnostics, Maidenhead, United Kingdom) with values 0.05 IU/ml. The criteria for starting ART in Ghana followed the WHO 2003 guidelines, with a change of the CD4 threshold for initiation from 200 to 350 cells/mm3 in 2008. During the study period (2004e2010) first-line ART comprised lamivudine plus either zidovudine or stavudine, plus either nevirapine or efavirenz. The choice between the use of either zidovudine or stavudine was determined by availability but zidovudine was avoided in patients with haemoglobin below 10 g/dL. The Committee on Human Research Publications and Ethics of the Kwame Nkrumah University of Science and Technology and the Komfo Anokye Teaching Hospital (KATH), Kumasi provided ethical approval for this study. Informed consent was not obtained from patients since this was an anonymised retrospective observational study.
Data and study outcomes Data were extracted from notes of patients starting ART between January 2004 and December 2010. Information on deaths and other clinical events in patients defaulting clinic appointments were obtained from hospital records in Kumasi and telephone enquiries from relatives or friends of patients. For the present analysis, loss-to-follow up, death and severe hepatotoxicity were defined as follows. Loss to follow up was defined as missing a clinic appointment by at least 3 months without further attendance at the clinic. Death was defined as confirmed death with death certification by a medical practitioner or verbal confirmation of death by a relative or friend. A previous study in Kumasi had identified around two thirds of all patients who were lost to follow up as having died,15 so a composite outcome measure including both death and lost-to-follow-up was used. Hepatotoxicity grades were based on alanine aminotransferase (ALT) levels and defined in accordance with AIDS Clinical Trials Group criteria in the following manner: grade 1, 1.25e2.5 times the upper limit of normal (x ULN); grade 2, 2.6e5.0 ULN; grade 3, 5.1e10 ULN; grade 4, >10 ULN. In order to avoid selection bias favouring the inclusion of patients with very high baseline ALT levels, severe hepatotoxicity was defined as grade 3 or 4 increases in ALT level, or an increase in ALT
Response to ART in HIV/HBV co-infection in Ghana level of greater than 100 IU/l from baseline as previously described.9 Furthermore, when a patient was noted to have had transaminase elevations, the reasons for those elevations as recorded by the attending clinician were noted. All analyses were conducted on an intention-to-treat basis and data closed for analysis on 31st December 2011.
Statistical analyses Parametric and non-parametric methods were used to compare baseline characteristics of continuous data between patients with and without hepatitis B co-infection. Wilcoxon rank sum test was used to compare the two groups since the baseline continuous data were independent between the patients. Comparisons of dichotomous data were performed using Chi-squared or Fisher’s exact test. Risk factors associated with elevated alanine transaminase concentration before initiation of ART was examined using a logistic multivariable regression analysis. For the clinical outcome measures (severe hepatotoxicity on ART and the risk of death/loss-to-follow up after initiation of ART), time to the first occurrence was calculated by subtracting the date of the event from the date of initiation of ART. Patients were censored if none of these events were observed at the time of last visit for patients that were lost to follow-up and at 31st December 2011 for the remainder. Cox proportional hazards regression was used to model the individual and simultaneous effects of the initial NNRTI used and other baseline variables on time to death/loss-to-follow up and severe hepatotoxicity. Potential determinants of the end-points were first tested for possible association with this outcome measure in an unadjusted model for possible inclusion in multivariable models. Any predictor having significant association with these outcomes at 5% level in unadjusted analysis, was included in adjusted analysis. In multivariable analysis, statistical significance was attained if p < 0.05. All p-values were two-tailed. These analyses were performed using SAS 9.3. A linear mixed effects model was used to analyse the CD4 data. The heterogeneity between patients at the first visit and during the course of the therapy was accounted for in the random effects specification for the model. Changes in CD4 count from baseline were specified as the outcome variable for the mixed model. Age, gender and WHO stage, HBV status, time on ART and the interaction between HBV status and time on ART were specified as fixed effects in the model. The mixed effect model was implemented in R using the nlme package.
Results Baseline demographics of patients Two thousand one hundred and sixty-eight (2168) out of 4039 patients (54%) starting ART between January 2004 and December 2010 had documented hepatitis B screening results. There were no significant differences in the baseline demographic, clinical and laboratory characteristics between patients with HBV serology results and those without serology results (data not shown). Two hundred
483 and ninety-nine patients were HBsAg-positive (13.8%, 95% CI of 12.4%e15.3%). Patients were followed for up to 90 months with a median duration of follow-up of 35 months. A comparison of baseline characteristics of HBsAg-positive and negative patients is shown in Table 1. At baseline, there were no significant differences between the two groups with respect to median age, WHO clinical stage, body mass index (BMI), CD4 count, haemoglobin and ART regimen initiated. Most patients started NNRTI-based ART and all received lamivudine plus either zidovudine or stavudine. A total of 20 patients started PI-based ART with either lopinavir/ritonavir or nelfinavir. Among the HBsAgpositive group, PI use was uncommon, there were more females than males, and serum aspartate aminotransferase (AST), alanine aminotransferase (ALT) and creatinine concentrations were significantly higher compared to the HBsAg-negative group.
Factors associated with elevated ALT at baseline Prior to initiation of ART, 591 out of 2003 (29.5%) patients had elevated ALT concentrations of above 40 IU/l. Risk factors associated with elevated ALT on adjusted analysis were a positive HBsAg status, with an adjusted odds ratio (95% CI) of 1.44 (1.10e1.89) (p Z 0.009), male gender, lower baseline CD4 count and body mass index, but not WHO clinical stage nor age (Table 2). Similar risk factors were associated with elevated AST (data not shown).
Incidence and risk factors for ART related grade 3/4 hepatotoxicity For patients who remained on therapy during follow up, there was a general increase in the proportion of patients with normal ALT level as shown in Fig. 1. In the entire cohort initiating ART (n Z 4039), 143 (3.5%) patients experienced severe (Grade 3/4) hepatotoxicity during 11,236.8 person-years of follow up with an incidence rate of 1.27/ 100 person-years, 95% CI of 1.07e1.49/100 person years. Among the sub-cohort with HBsAg serology data, seventyone (71) patients experienced severe hepatotoxicity during 7057.5 person-years of follow up with an incidence rate of 1.01/100 person-years, 95% CI of 0.7e1.26/100 person years. The median (range) time to the onset of first episode of severe hepatotoxicity of 12 months (2e84) was not different from that of the entire cohort. HBsAg-positivity was significantly associated with risk of developing severe hepatotoxicity by Kaplan Meier estimates with a HR (95% CI) of 1.99 (1.16e3.40) (p Z 0.01), although the increased risk appeared to be mostly after 18 months of ART (Fig. 2). The cumulative frequency of severe hepatotoxicity amongst HBsAg-positive patients during follow up for up to 90 months was 12.9% (95% CI of 5.9%e20.0%) compared to 6.7% (95% CI of 4.3%e9.0%) among HBsAg-negative patients. The only other significant risk factor identified in univariate Cox proportional hazards analysis was baseline body mass index with an HR (95% CI) of 1.21 (1.00e1.47) for each 5 kg/m2 increase in BMI. Factors such as age, gender, WHO clinical stage, CD4 counts at baseline (and its changes within first 12 months of ART), NNRTI and NRTI used were all not associated with risk of
484 Table 1
F.S. Sarfo et al. Baseline characteristics of the study population at ART initiation.
Characteristic
HBsAg positive n Z 299
HBsAg negative n Z 1869
p-value
Age, median (IQR) Gender, male:female, n(%) WHO clinical stage,a n(%) 1 2 3 4 BMI, median (IQR) CD4 count, median (IQR) Haemoglobin, median (IQR) AST, median (IQR) ALT, median (IQR) Grade of hepatotoxictyb 0 1 2 3 4 No data Creatinine, median (IQR) NRTI backbone, n(%) ZDVþ3 TC d4Tþ3 TC NNRTI/PI, n(%) Nevirapine Efavirenz Protease Inhibitorc
37 (32e44) 98 (33):201(67)
38 (32e45) 795 (43):1074 (57)
0.45 0.002
22 (7.4) 49 (16.4) 160 (53.5) 29 (9.7) 19.9 (17.6e22.8) 141 (45e234) 10.3 (8.9e12.4) 45.0 (32.0e66.0) 33.0 (25.0e49.0)
104 (5.6) 192 (10.3) 830 (44.4) 219 (11.7) 20.1 (17.8e22.9) 147 (62e229) 10.5 (9.1e12.0) 39.0 (29.0e54.0) 28.0 (20.0e42.7)
0.08
209 (69.9) 51 (17.1) 12 (4.0) 4 (1.3) 1 (0.3) 22 (7.4) 88.4 (72e106)
1419 (75.9) 252 (13.5) 49 (2.6) 4 (0.2) 2 (0.1) 143 (7.7) 84.9 (67e105)
0.01
138 (46) 161 (54)
924 (49) 945 (51)
0.29
117(39) 180 (60) 2 (1)
785(42) 1066 (57) 18 (1)
0.55
0.44 0.36 0.86
www.elsevierhealth.com/journals/jinf
Long-term responses to first-line antiretroviral therapy in HIV and hepatitis B co-infection in Ghana Fred Stephen Sarfo a,b, Adetayo Kasim c, Richard Phillips a,b, Anna Maria Geretti d, David R. Chadwick e,* a
Komfo Anokye Teaching Hospital, Kumasi, Ghana Kwame Nkrumah University of Science and Technology, Kumasi, Ghana c Wolfson Research Institute for Health and Wellbeing, Durham University, Stockton-on-Tees TS17 6BH, UK d Institute of Global Health, University of Liverpool, L69 7BE, UK e The James Cook University Hospital, Middlesbrough, TS4 3BW, UK b
Accepted 12 June 2014 Available online 27 June 2014
KEYWORDS Hepatitis B; HIV; Clinical outcomes; Lamivudine; Antiretroviral therapy
Summary Objectives: To observe the long term response to first-line antiretroviral therapy (ART) in HIV and hepatitis B virus (HBV) co-infected patients in Ghana and explore predictors of poor clinical outcomes. Methods: Retrospective cohort study of hepatitis B surface antigen (HBsAg) positive and negative patients receiving predominantly NNRTI-based ART with lamivudine plus either zidovudine or stavudine for up to seven years. Cox proportional hazards and Kaplan Meier survival analyses compared clinical outcomes and identified baseline characteristics predictive of poor outcomes. A mixed effects model compared changes in CD4 counts. Results: A total of 299 HBsAg-positive and 1869 HBsAg-negative patients started ART between 2004 and 2008. Over a median 35 months of follow-up, HBsAg-positive patients were more likely to die or default care than HBsAg-negative patients, aHR 1.36 (95% CI, 1.03e1.80). HBsAg-positive patients were also more likely to develop Grade 3/4 hepatotoxicity than HBsAg-negative patients, HR 1.99 (1.16e3.40) on survival analysis. There was no significant difference in CD4 responses between HBsAg-positive and HBsAg-negative patients.
* Corresponding author. Centre for Clinical Infection, The James Cook University Hospital, Marton Road, Middlesbrough TS4 3BW, UK. Tel.: þ44 01642 854429; fax: þ44 01642 854017. E-mail address: [email protected] (D.R. Chadwick). http://dx.doi.org/10.1016/j.jinf.2014.06.012 0163-4453/ª 2014 The British Infection Association. Published by Elsevier Ltd. All rights reserved.
482
F.S. Sarfo et al. Conclusions: HBsAg-positive patients are at significantly increased risk of adverse clinical outcomes after starting ART. Further studies are warranted to evaluate whether these risks remain now that tenofovir is becoming routinely available in Ghana. ª 2014 The British Infection Association. Published by Elsevier Ltd. All rights reserved.
Introduction Approximately 17% of HIV-infected individuals in Ghana are also chronically infected with hepatitis B virus (HBV).1 In contrast to developed countries in sub-Saharan Africa most HBV infection is acquired during early childhood, so patients with HIV/HBV co-infection have usually had chronic HBV infection for many years by the time they present with (or are tested for) HIV infection. Patients with HIV/HBV co-infection have accelerated liver disease2e4 and are more likely to develop hepatotoxicity due to antiretroviral therapy (ART).5,6 The effect of HBV infection on HIV progression or response to ART is less clear and a number of studies have shown contrasting effects of HBV coinfection on HIV virological response or CD4 cell increases.7e12 Data on the longer-term effectiveness of ART and risk of ART-induced hepatotoxicity among HIV/HBV co-infected patients are lacking, particularly in high prevalence areas for HBV. The gold standard treatment for patients with HIV/HBV co-infection is ART containing tenofovir, although until recently this drug has not been widely available in most sub-Saharan African countries. First-line ART in such regions has normally comprised a non-nucleoside reverse transcriptase inhibitor (NNRTI) in combination with two nucleoside reverse transcriptase inhibitors (NRTI) including lamivudine, which is the sole NRTI with activity against HBV. Hence in many parts of Africa, where testing for HBV was not routine in HIV clinics or tenofovir was not available, co-infected patients have started such ART combinations with the inevitable development of HBV resistance to lamivudine,13,14 and some have developed complications including hepatitis flares and progression of liver disease due to loss of HBV virological control. The objectives of this retrospective study were to explore the associations between HIV/HBV co-infection and the risks for clinical events, particularly death, loss to follow up and severe hepatotoxicity, after initiation of ART containing lamivudine as the only NRTI active against HBV. We also modelled CD4 recovery on ART over the longterm in a cohort of Ghanaian HIV infected patients with and without HBV co-infection.
Patients and methods Study population and antiretroviral therapy From 2004, patients referred to the HIV clinic in KATH, Kumasi have been treated as part of the National AIDS Control Programme. Patients were referred from a large area of central and northern Ghana. Clinical data are recorded in patients’ case notes. Routine laboratory tests include full blood count, renal and liver function tests and
CD4 counts at baseline, with blood counts, liver function and CD4 counts tested every 6 months subsequently. HIV viral load was not tested routinely and testing for HIV-2 and HBV co-infection was performed only in limited circumstances. HBV testing at the clinic was not conducted as part of routine care, however was performed intermittently as part of research projects where all patients attending clinic were tested for a period of time, using either the Bioelisa HBsAg Colour (Biokit Ltd, UK), or the Murex HBsAg (Abbott Diagnostics, UK) assays on fresh or stored serum samples, and later confirmed using the automated chemilumiscent immunoassay Architect HBsAg (Abbott Diagnostics, Maidenhead, United Kingdom) with values 0.05 IU/ml. The criteria for starting ART in Ghana followed the WHO 2003 guidelines, with a change of the CD4 threshold for initiation from 200 to 350 cells/mm3 in 2008. During the study period (2004e2010) first-line ART comprised lamivudine plus either zidovudine or stavudine, plus either nevirapine or efavirenz. The choice between the use of either zidovudine or stavudine was determined by availability but zidovudine was avoided in patients with haemoglobin below 10 g/dL. The Committee on Human Research Publications and Ethics of the Kwame Nkrumah University of Science and Technology and the Komfo Anokye Teaching Hospital (KATH), Kumasi provided ethical approval for this study. Informed consent was not obtained from patients since this was an anonymised retrospective observational study.
Data and study outcomes Data were extracted from notes of patients starting ART between January 2004 and December 2010. Information on deaths and other clinical events in patients defaulting clinic appointments were obtained from hospital records in Kumasi and telephone enquiries from relatives or friends of patients. For the present analysis, loss-to-follow up, death and severe hepatotoxicity were defined as follows. Loss to follow up was defined as missing a clinic appointment by at least 3 months without further attendance at the clinic. Death was defined as confirmed death with death certification by a medical practitioner or verbal confirmation of death by a relative or friend. A previous study in Kumasi had identified around two thirds of all patients who were lost to follow up as having died,15 so a composite outcome measure including both death and lost-to-follow-up was used. Hepatotoxicity grades were based on alanine aminotransferase (ALT) levels and defined in accordance with AIDS Clinical Trials Group criteria in the following manner: grade 1, 1.25e2.5 times the upper limit of normal (x ULN); grade 2, 2.6e5.0 ULN; grade 3, 5.1e10 ULN; grade 4, >10 ULN. In order to avoid selection bias favouring the inclusion of patients with very high baseline ALT levels, severe hepatotoxicity was defined as grade 3 or 4 increases in ALT level, or an increase in ALT
Response to ART in HIV/HBV co-infection in Ghana level of greater than 100 IU/l from baseline as previously described.9 Furthermore, when a patient was noted to have had transaminase elevations, the reasons for those elevations as recorded by the attending clinician were noted. All analyses were conducted on an intention-to-treat basis and data closed for analysis on 31st December 2011.
Statistical analyses Parametric and non-parametric methods were used to compare baseline characteristics of continuous data between patients with and without hepatitis B co-infection. Wilcoxon rank sum test was used to compare the two groups since the baseline continuous data were independent between the patients. Comparisons of dichotomous data were performed using Chi-squared or Fisher’s exact test. Risk factors associated with elevated alanine transaminase concentration before initiation of ART was examined using a logistic multivariable regression analysis. For the clinical outcome measures (severe hepatotoxicity on ART and the risk of death/loss-to-follow up after initiation of ART), time to the first occurrence was calculated by subtracting the date of the event from the date of initiation of ART. Patients were censored if none of these events were observed at the time of last visit for patients that were lost to follow-up and at 31st December 2011 for the remainder. Cox proportional hazards regression was used to model the individual and simultaneous effects of the initial NNRTI used and other baseline variables on time to death/loss-to-follow up and severe hepatotoxicity. Potential determinants of the end-points were first tested for possible association with this outcome measure in an unadjusted model for possible inclusion in multivariable models. Any predictor having significant association with these outcomes at 5% level in unadjusted analysis, was included in adjusted analysis. In multivariable analysis, statistical significance was attained if p < 0.05. All p-values were two-tailed. These analyses were performed using SAS 9.3. A linear mixed effects model was used to analyse the CD4 data. The heterogeneity between patients at the first visit and during the course of the therapy was accounted for in the random effects specification for the model. Changes in CD4 count from baseline were specified as the outcome variable for the mixed model. Age, gender and WHO stage, HBV status, time on ART and the interaction between HBV status and time on ART were specified as fixed effects in the model. The mixed effect model was implemented in R using the nlme package.
Results Baseline demographics of patients Two thousand one hundred and sixty-eight (2168) out of 4039 patients (54%) starting ART between January 2004 and December 2010 had documented hepatitis B screening results. There were no significant differences in the baseline demographic, clinical and laboratory characteristics between patients with HBV serology results and those without serology results (data not shown). Two hundred
483 and ninety-nine patients were HBsAg-positive (13.8%, 95% CI of 12.4%e15.3%). Patients were followed for up to 90 months with a median duration of follow-up of 35 months. A comparison of baseline characteristics of HBsAg-positive and negative patients is shown in Table 1. At baseline, there were no significant differences between the two groups with respect to median age, WHO clinical stage, body mass index (BMI), CD4 count, haemoglobin and ART regimen initiated. Most patients started NNRTI-based ART and all received lamivudine plus either zidovudine or stavudine. A total of 20 patients started PI-based ART with either lopinavir/ritonavir or nelfinavir. Among the HBsAgpositive group, PI use was uncommon, there were more females than males, and serum aspartate aminotransferase (AST), alanine aminotransferase (ALT) and creatinine concentrations were significantly higher compared to the HBsAg-negative group.
Factors associated with elevated ALT at baseline Prior to initiation of ART, 591 out of 2003 (29.5%) patients had elevated ALT concentrations of above 40 IU/l. Risk factors associated with elevated ALT on adjusted analysis were a positive HBsAg status, with an adjusted odds ratio (95% CI) of 1.44 (1.10e1.89) (p Z 0.009), male gender, lower baseline CD4 count and body mass index, but not WHO clinical stage nor age (Table 2). Similar risk factors were associated with elevated AST (data not shown).
Incidence and risk factors for ART related grade 3/4 hepatotoxicity For patients who remained on therapy during follow up, there was a general increase in the proportion of patients with normal ALT level as shown in Fig. 1. In the entire cohort initiating ART (n Z 4039), 143 (3.5%) patients experienced severe (Grade 3/4) hepatotoxicity during 11,236.8 person-years of follow up with an incidence rate of 1.27/ 100 person-years, 95% CI of 1.07e1.49/100 person years. Among the sub-cohort with HBsAg serology data, seventyone (71) patients experienced severe hepatotoxicity during 7057.5 person-years of follow up with an incidence rate of 1.01/100 person-years, 95% CI of 0.7e1.26/100 person years. The median (range) time to the onset of first episode of severe hepatotoxicity of 12 months (2e84) was not different from that of the entire cohort. HBsAg-positivity was significantly associated with risk of developing severe hepatotoxicity by Kaplan Meier estimates with a HR (95% CI) of 1.99 (1.16e3.40) (p Z 0.01), although the increased risk appeared to be mostly after 18 months of ART (Fig. 2). The cumulative frequency of severe hepatotoxicity amongst HBsAg-positive patients during follow up for up to 90 months was 12.9% (95% CI of 5.9%e20.0%) compared to 6.7% (95% CI of 4.3%e9.0%) among HBsAg-negative patients. The only other significant risk factor identified in univariate Cox proportional hazards analysis was baseline body mass index with an HR (95% CI) of 1.21 (1.00e1.47) for each 5 kg/m2 increase in BMI. Factors such as age, gender, WHO clinical stage, CD4 counts at baseline (and its changes within first 12 months of ART), NNRTI and NRTI used were all not associated with risk of
484 Table 1
F.S. Sarfo et al. Baseline characteristics of the study population at ART initiation.
Characteristic
HBsAg positive n Z 299
HBsAg negative n Z 1869
p-value
Age, median (IQR) Gender, male:female, n(%) WHO clinical stage,a n(%) 1 2 3 4 BMI, median (IQR) CD4 count, median (IQR) Haemoglobin, median (IQR) AST, median (IQR) ALT, median (IQR) Grade of hepatotoxictyb 0 1 2 3 4 No data Creatinine, median (IQR) NRTI backbone, n(%) ZDVþ3 TC d4Tþ3 TC NNRTI/PI, n(%) Nevirapine Efavirenz Protease Inhibitorc
37 (32e44) 98 (33):201(67)
38 (32e45) 795 (43):1074 (57)
0.45 0.002
22 (7.4) 49 (16.4) 160 (53.5) 29 (9.7) 19.9 (17.6e22.8) 141 (45e234) 10.3 (8.9e12.4) 45.0 (32.0e66.0) 33.0 (25.0e49.0)
104 (5.6) 192 (10.3) 830 (44.4) 219 (11.7) 20.1 (17.8e22.9) 147 (62e229) 10.5 (9.1e12.0) 39.0 (29.0e54.0) 28.0 (20.0e42.7)
0.08
209 (69.9) 51 (17.1) 12 (4.0) 4 (1.3) 1 (0.3) 22 (7.4) 88.4 (72e106)
1419 (75.9) 252 (13.5) 49 (2.6) 4 (0.2) 2 (0.1) 143 (7.7) 84.9 (67e105)
0.01
138 (46) 161 (54)
924 (49) 945 (51)
0.29
117(39) 180 (60) 2 (1)
785(42) 1066 (57) 18 (1)
0.55
0.44 0.36 0.86
