EHF after HBV active HAART initiation was frequently observed in this population.
We found that elevated ALT and higher HBV DNA at baseline are significant predictors of EHF. In addition the occurrence of EHF was positively associated with subsequent HBeAg and HBsAg loss, both of which occurred at elevated rates (33% and 8% respectively) during the 48-week study follow-up.
Elevated transaminases in the setting of HAART and viral hepatitis co-infection are commonly multifactorial and may reflect natural alterations in hepatitis disease activity, direct antitretroviral hepatotoxicty, drug hypersensitivity, co-existent hepatotoxic medication, alcohol excess and immune restoration disease. The reasons for the high rate of flare observed in our prospective clinical trial are unclear but can be hypothesized.
Our population had advanced immunodeficiency with a median pre-HAART CD4 count of only 36 cells/mm3
and high usage of potentially hepatotoxic medications including cotrimoxazole, fluconazole and isoniazid/rifampicin - however none of these factors were identified as associated with EHF in univariate analysis. Similarly, no difference was observed in baseline level of alcohol use between groups (Table ). Nucleoside analogue agents are also potentially hepatotoxic, but usually through the development of mitochondrial dysfunction and microvesicular hepatic steatosis - this was felt to be unlikely in our subjects given the time frame in which the HF events were observed and the absence of associated features such as lactic acidosis suggesting mitochondrial dysfunction. All subjects were prescribed EFV and EFV associated hepatotoxicity has been clearly described, although at a lesser rate than with nevirapine [4
]. EFV induced hepatotoxicity as a cause of HF is possible in the symptomatic case, (although subsequent EFV reintroduction without HF at a later date makes this unlikely); and the possibility of EFV-induced hepatotoxicity as a cause of death in the case with cirrhosis is also conceivable.
Overall however, timing of EHF, association with elevated ALT and HBV DNA and high rate of seroconversion are all consistent with immune restoration stimulating HBV-specific immune responses as the likely underlying process
Spontaneous flares in CHB are well recognised and reflect the dynamic nature of the balance between immunological control and HBV replication [7
]. Sometimes, but not universally, these flares will be followed by HBeAg loss, anti-HBe seroconversion and even occasionally HBsAg loss/seroconversion. In the context of HBV treatment and effective HBV DNA reduction the rate of HF increases, particularly in the early months of treatment [8
]. These flares likely reflect an underlying shift in host-viral balance with some regeneration of host T-cell responsiveness as HBV DNA decreases [10
], but are still accompanied by anti-HBe seroconversion in around only 20% of subjects by 1 year [8
In the setting of HIV coinfection the situation is enhanced by the simultaneous reduction in HIV RNA observed with effective HAART and the subsequent restoration of either an adaptive or innate response to HBV antigens. The concept of immune restoration disease (IRD) is characterised by the finding of rapid restoration in immunity in the presence of a high pathogen load resulting in abnormal and exaggerated immune responses, often mediated by high levels of interferon gamma [11
]. Besides elevated ALT, our study identified higher HBV viral load both at baseline and throughout the early weeks of therapy, as a significant predictor of EHF, a finding consistent with the theory of IRD and high antigen burden as a cause for flare in these subjects. This theory is supported by our findings of elevated CXCL10, an interferon stimulated gene, in patients with EHF within this study [14
The most striking aspect of our study however is the link between HF and the high rate of subsequent seroconversion. In a retrospective study examining the role of 3TC-containing HAART on serological change in 82 HIV-HBV coinfected subjects, Miailhes et al reported an anti-HBe seroconversion rate of 17% and an overall HBsAg loss rate of 6% during a median follow-up of 5 years [15
]. Interesting, they found more advanced HIV CDC stage at baseline to be associated with subsequent seroconversion rates. HBsAg and/or HBeAg seroconversion occurred in 29% of stage C subjects versus only 3% of patients with stage A disease, again suggesting a major role for immunorestoration in promoting seroconversion. One specific difference in our study to the majority of reports on outcomes in HIV-HBV coinfected individuals commencing HAART is that our population was Asian with predominantly Genotype C and likely vertically acquired HBV disease. Thus differences in the virus, host and/or natural history of HBV in this population may all be contributing to our findings. Given the overwhelming greater burden of HBV disease in Asian countries, further study of the mechanisms and outcomes of HBV-active HAART in this population are certainly required.
Our study is undoubtedly limited by the relatively small number of subjects to elucidate further any additional associations or conclusions. However, it is one of the few randomised clinical trials performed in this population.