Hepatitis C infection increases the risk of the modification of first highly active antiretroviral therapy in HIV-infected patients

We assessed predictors of discontinuation or change of the first regimen of highly active antiretroviral therapy in 465 HIV-infected adults, in the first year. A total of 187 patients modified their regimen: 45 discontinuing and 142 changing because of clinical/virological failure, intolerance/toxicity or non-adherence. Predictors of modification of the regimen were hepatitis virus C seropositivity, liver cirrhosis, higher baseline viral load, sex (women had a lower risk) and calendar year (lower risk for starting in 2000). Several studies addressing the reasons for the discontinuation or change of the first regimen of highly active antiretroviral therapy (HAART) reported that up to 30–54% of patients stopped or modified their initial HAART regimen over a follow-up lasting from a median of 48 weeks to 3 years [1–9]. Side-effects and toxicity were the most commonly reported reasons [4,7–15], but only a few studies addressed the predictors of modification of the first regimen [1–3,5–7,11], and just one included hepatitis C virus (HCV) serostatus in the analysis [2]. HIV–HCV co-infection increases the risk of hepatotoxicity in patients on antiretroviral drugs [16–20], and patients with hepatitis experience a greater number of adverse events [10]. We investigated the factors associated with the modification of the initial HAART regimen within the first 12 months, in adult patients naive to antiretroviral drugs and starting therapy between November 1996 and December 2000. Our unit is the only HIV centre in our district and patients’ demographic characteristics reflect the HIV epidemic in Italy [21], where access to treatment has always been free of charge. Out of 551 patients, we excluded 86 (15.6%) for whom clinical and biochemical data were incomplete. Each regimen included three anti-HIV drugs, with two nucleoside reverse transcriptase inhibitors (NRTI) as the backbone, which was associated either with one protease inhibitor (PI) or one non-nucleoside reverse transcriptase inhibitor (NNRTI), or a third NRTI. Modification was categorized as discontinuation (any interruption of HIV regimen for at least one month), or change (any substitution of at least one drug). The reasons for modification were grouped as clinical/virological failure, intolerance/toxicity and non-adherence. Patients had at least a 60-day follow-up, unless the regimen had been modified before that time. Potential factors associated with modification were: age, sex, risk category, AIDS diagnosis, baseline CD4 cell count and plasma viraemia, hepatitis B virus infection and HCV seropositivity (EIA-3 assay; Roche Diagnostics, Monza, Italy), alanine aminotransaminase (ALT) levels, aspartate aminotransaminase levels, platelet and haemoglobin values, NNRTI or PI-based regimens, year of starting HAART, and liver cirrhosis. Transaminase levels were abnormal if 46 IU/ml or greater. Liver cirrhosis included patients with clinical/biochemical or histological evidence of hepatic disease, with at least one episode of liver decompensation before or during the first year of follow-up. Association with HAART modification was measured through risk ratios (unadjusted and adjusted risk ratios) calculated using a log binomial model. The latter were calculated using a multiple log binomial model obtained after a backward selection (i.e. we excluded at each step the variable with the higher P value from the log-likelihood ratio test; the final model was obtained when only variables with a P value less than 0.2 were included). In this analysis, continuous variables are shown as categorized. However, another analysis in which variables were untransformed provided similar results. Out of 465 patients, 371 were men (79.8%), 243 (52.2%) were past or active intravenous drug users (IVDU), and the mean age was 36.2 years (SD ± 9.1). Thirty-eight (8.2%) were non-European citizens (mostly Africans). The mean baseline CD4 cell count was 277 cells/mm3 (SD ± 230), and plasma HIV-RNA levels were 4.2 log copies/ml (SD ± 1). A total of 273 subjects (58.7%) had normal baseline ALT levels, 90 (19.3%) had AIDS, 212 (45.6%) were HCV seropositive (74.9 and 11.5% in the ‘IVDU’ and ‘other than IVDU’ categories, respectively), 24 (5.3%) were hepatitis B surface antigen positive, whereas six (1.3%) were HCV–HBV co-infected. Twenty patients (4.3%) had liver cirrhosis (80% HCV seropositive). A total of 88 (18.9%) patients started HAART in 1997, 138 (29.6%) in 1998, 133 (28.6 %) in 1999, and 106 (22.8%) in 2000. A total of 368 individuals (79.1%) received a PI-based therapy including indinavir (56.2%), nelfinavir (26.6%) and saquinavir hard gel capsules (13.6%), or ritonavir (3.5%). In this group, 13 (3.5%) received a ritonavir-boosted regimen. A total of 86 individuals (18.5%) received NNRTI-based HAART (46 and 40 on efavirenz and nevirapine, respectively), whereas 11 patients (2.3%) received three NRTI. A total of 187 patients (40.2%) modified their HAART regimen within the first year of treatment. Most of them (142, 75.9%) changed their initial treatment, whereas a minority (45, 24.1%) discontinued therapy (28 patients re-started HAART after a mean of 4.8 months, range 1–23). The reasons for modification were: clinical/virological failure (33, 17.6%), intolerance/toxicity (126, 67.3%), and non-adherence (28, 14.9%). As shown in Table 1, the predictors of modification were: sex (women had a lower risk), baseline viral load (higher risk for greater values), both HCV seropositivity and liver cirrhosis (higher risk), and calendar year (lower risk for patients starting in 2000). No statistical difference was reported according to PI-based or NNRTI-based therapy.Table 1: Association of several factors with discontinuation/change of initial regimen of highly active antiretroviral therapy.Our study confirms that a large proportion of patients modifies their regimen within the first year, mostly switching as a result of intolerance or toxicity, as reported in other studies [4,7–10]. As far as the predictors, the decreasing rates of therapy modification by calendar year may reflect the longer experience of physicians in prescribing different medications to different patients. Women were less likely to modify their initial regimen (also according to the three main reasons for modification), similar to one other report [3], but not other studies [2,5]. Some data suggest that women are more likely both to experience adverse events [10] and to change therapy if on NNRTI [22], and to have higher rates of NNRTI-related rash [23]. Sex differences may account for differences in tolerance, toxicity and adherence to HIV drugs [24]. Therapy modification was more frequent in patients with cirrhosis (in which impaired drug metabolism may increase the dose-dependent side-effects), although the risk for such patients is overestimated, as only patients with a clinically overt illness were included in this group. As far as the role of HCV seropositivity is concerned, the interpretation is more complex. Most HCV-infected patients are also viraemic [25], and even with normal ALT levels, they have some degree of histological damage [26,27]. In addition, a higher degree of fibrosis [28,29] has been reported in co-infected individuals, but why HCV co-infection should lead to the premature modification of HAART is still unclear. Such patients may be less tolerant and less adherent to treatment, leading to early virological failure, and they certainly have a higher risk of hepatic toxicity [17–19], so even the analysis of the specific reasons for modification may be misleading in such a multifaceted picture.