Initial controlled trials with the combination of abacavir, lamivudine and zidovudine (ABC/3TC/ZDV) showed equivalent efficacy to the regimen of indinavir, lamivudine, and ZDV in achieving a plasma HIV-1 viral load (pVL) <400 copies/mL at 48 weeks in antiretroviral therapy-naive (ART-naive) HIV-1-infected patients.1 The virologic response of this triple-nucleoside reverse transcriptase inhibitor (NRTI) regimen was later shown to be inferior to that of efavirenz-based regimens and also to that of indinavir-based regimens in patients with a high pVL.2,3 For this reason, current guidelines for the initial treatment of HIV-1-infected patients recommend 2 NRTIs combined with a nonnucleoside reverse transcriptase inhibitor (NNRTI)-or boosted protease inhibitor/ritonavir-based (PI/r) regimen and consider the triple-NRTI regimen of ABC/3TC/ZDV as an alternative only when a preferred regimen cannot be used.4
Treatment recommendations are usually based on the results of randomized clinical trials because they are considered the “gold standard” for evaluating the efficacy of treatment regimens. We should take into account that selection bias is a well-known limitation of this kind of study, however, because of the tendency to include motivated and medication-adherent patients.5 We carried out this study to analyze the safety and effectiveness of ABC/3TC/ZDV for the initial treatment of HIV infection in patients not included in clinical trials.
This was a retrospective case series study carried out in 71 clinical centers throughout Spain. We analyzed 730 consecutive adult ART-naive HIV-infected patients who initiated ABC/3TC/ZDV between July 1997 and March 2003 and had at least 1 clinical follow-up visit with determination of CD4 cell counts and a pVL at least 16 weeks after the initiation of treatment. The study was approved by the Ethics Committee of the center of the principal investigator, Juan Berenguer, Unidad de Enfermedades Infecciosas, Hospital Gregorio Marañón, Madrid.
Baseline and Follow-Up Data and Study End Points
Data were collected by chart or database review with a standard questionnaire to extract information, including baseline data, such as age, gender, HIV risk group, prior AIDS-defining conditions (ADCs), methadone use, baseline CD4 cell count and pVL, hepatitis C virus (HCV) and hepatitis B virus (HBV) serology, and date of initiation of ABC/3TC/ZDV.
We recorded follow-up information, including responses of pVL and CD4+ counts, new ADCs, deaths, and adverse events that required interruption or change in the ART regimen, paying special attention to the development of any suspected ABC hypersensitivity reaction (HSR) according to defined clinical criteria.6 We also inquired about adherence to ART during the first 6 and 12 months and about missed clinical visits during follow-up. Analysis included information recorded up to March 2003.
Virologic failure was defined as any of the following: (1) reduction in pVL <1 log during the first 12 weeks of ART, unless it was less than the lower limit of quantification (LOQ); (2) failure to achieve a pVL <LOQ after 24 weeks of ART; and (3) rebound to 2 consecutive pVLs ≥LOQ after achieving a pVL <LOQ.7
We calculated rates of events by dividing the number of patients developing the event by the number of person-years at risk. Analysis of treatment failure was performed using 2 methods. The first method was an intention-to-treat (ITT) analysis of observed data (OD) in which treatment switches or losses to follow-up were not considered necessarily as treatment failures if there were no virologic failures. Data from patients who were lost to follow-up were censored at the last visit. The second method was a stricter ITT analysis in which patients who were lost to follow-up or who stopped or switched therapy for any reason were also considered to have failed therapy (LSS = F). We assessed risk of treatment failure by Cox proportional hazards analysis. All statistical analyses were performed with SPSS (version 10). All P values are 2-tailed. Statistical significance was defined as P < 0.05.
During the study period, a total of 771 ART-naive patients started ABC/3TC/ZDV in our institutions. Of this total, 41 (5.31%) were not included in the analysis for the following reasons. Thirty (3.89%) had fewer than 16 weeks of follow-up at censorship. Of these patients, none are known to have died and 2 are known to have developed an HSR to ABC on days 3 and 5 after initiation of ART, respectively. One of these 2 patients switched therapy to 3TC, stavudine, and efavirenz. Eleven (1.43%) patients were not included in the analysis because they were lost to follow-up after the visit in which ART was prescribed, and no information is available about whether or not they ever initiated ART.
The study group was composed of 730 patients: 558 (76%) were treated with the fixed-dose combination of ABC/3TC/ZDV (Trizivir), 93 (13%) were treated with the fixed-dose combination of 3TC and ZDV (Combivir) plus ABC, and 79 (11%) were treated with ABC/3TC/ZDV with each drug administered separately.
Several techniques were used for the quantification of pVL: HIV RNA polymerase chain reaction (PCR; 68%), branched chain DNA (b-DNA; 23%) and nucleic acid sequence-based amplification (NASBA; 9%). The LOQ of the technique was <20 copies/mL in 90 patients, <50 copies/mL in 452 patients, <80 copies/mL in 14 patients, <100 copies/mL in 7 patients, <200 copies/mL in 143 patients, <400 copies/mL in 21 patients, and unknown in 3 patients.
Demographic and baseline characteristics of the patients are summarized in Table 1. The median age was 37 years, approximately two thirds of the patients were male, and almost half had acquired HIV by injection drug use (IDU). Approximately one quarter of the patients were on methadone maintenance programs. At baseline, 20% of the patients had had prior ADCs, the median CD4 T-cell count was 255 cells/mm3, and the median pVL was 4.76 log copies/mL. Nearly half of the patients had antibodies against HCV, and 7% were hepatitis B surface antigen (HBsAg)-positive.
Adherence to ART was estimated in 648 (88.8%) patients. It was based on the physician's opinion in 334 (51%) patients, pharmacy reports in 269 (42%), a validated structured questionnaire8 in 40 (6%), and other methods in 5 (1%). During the first and second semesters of therapy, adherence was estimated to be less than 90% in 130 (20%) and 104 (16%) patients, respectively. During follow-up, 116 (16%) of 730 patients missed at least 1 clinical visit.
Frequency of Clinical Events
After a median follow-up of 50.5 weeks (interquartile ratio [IQR]: 28-78, minimum of 16 to maximum of 311), 565 (77.4%) patients continued on ABC/3TC/ZDV, whereas 165 (22.6%) had discontinued the regimen. Reasons for discontinuation included adverse events in 104 (14.2%) patients, voluntary abandonment in 27 (3.7%), virologic failure in 18 (2.5%), ART intensification in 6 (0.8%), structured treatment interruptions in 5 (0.7%), and other reasons in 5 (0.7%).
As previously mentioned, ART was interrupted or switched because of adverse events in 104 patients (14.25%, 13.57 events per 100 person-years; Table 2). In a multivariate Cox proportional hazards regression analysis, including the variables listed in Table 1, the following variables were independently associated with treatment interruptions because of adverse events: HCV coinfection (hazard ratio [HR] = 1.889, 95% confidence interval [CI]: 1.139 to 3.133), prior ADCs (HR = 1.669, 95% CI: 1.131 to 2.411), and female gender (HR = 1.644, 95% CI: 1.178 to 2.294). The most frequent adverse events were an HSR to ABC in 36 (4.93%) patients and hematologic toxicity (particularly anemia) in 30 (4.10%) patients. The median time to occurrence of an HSR to ABC was 4 weeks (IQR: 1-9 weeks), and only 5 (14%) of 36 HSRs occurred after week 16.
There were 10 deaths (1.4%, 1.18 events per 100 person-years). The causes of death were end-stage liver disease in 4 patients and carcinoma of the uterine cervix, brain hemorrhage, central nervous system (CNS) mass, pneumococcal pneumonia, suicide, and unknown cause (1 patient each).
A total of 12 patients suffered a new ADC (1.6%, 1.43 events per 100 person-years): tuberculosis in 5 patients, non-Hodgkin lymphoma in 2 patients, and AIDS dementia complex, Mycobacterium avium disseminated infection, progressive multifocal leukoencephalopathy, Cytomegalovirus retinitis, and Kaposi sarcoma (1 patient each).
Treatment failure by ITT OD analysis was confirmed in 105 (14.40%) patients, resulting in a rate of 13.53 events per 100 person-years (95% CI: 10.20 to 15.40). The frequency and rate of treatment failure according to the ITT LSS = F analysis were 165 (22.92%) and 22.92 events per 100 person-years (95% CI: 19.42 to 26.42), respectively (Table 3). The frequency of virologic failure in the ITT OD analysis was 105 (14.40%), and it was 18 (2.50%) in the ITT LSS = F analysis. The apparent contradiction is attributable to the fact that that 87 patients failed therapy in the first instance because of reasons other than virologic failure (eg, adverse event, voluntary abandonment, ART intensification, structured treatment interruptions [STI], other reasons) and eventually developed virologic failure. In the ITT LSS analysis, these are not considered virologic failures; however, they are considered virologic failures in the ITT OD analysis (see Table 3).
Treatment failure risk analysis was performed by multivariate Cox proportional hazards regression analysis (Table 4). In the ITT OD analysis, the following variables were independently associated with treatment failure: adherence <90% at some time during the first year of therapy (HR = 4.248), methadone use (HR = 2.116), baseline pVL (HR = 1.651 per log), and the presence of ADCs before initiation of ART (HR = 1.639). In the ITT LSS = F analysis, the following variables were independently associated with treatment failure: adherence <90% at some time during the first year of therapy (HR = 2.130), HCV coinfection (HR = 2.017), prior ADCs (HR = 1.771), and female gender (HR = 1.623).
We performed a separate analysis with the 542 patients whose pVL was quantified using a technique with an LOQ <50 copies/mL or <20 copies/mL. The frequency and risk factors for treatment failure in this group were no different from those of the entire cohort of 730 patients (data not shown).
Changes in CD4+ Cell Count and HIV RNA Levels
We analyzed changes in CD4+ cell count and HIV RNA levels after 48 weeks according to the ITT LSS = F analysis. The proportions of patients with increases in the CD4+ cell count from baseline higher than 50, 100, and 150 cells/μL were 66.5%, 55.8%, and 43.7%, respectively. At this time point, the proportion of patients with a pVL <LOQ was 67.3%. When we categorized patients according to baseline pVL < or ≥100,000 copies/mL, the proportions with a pVL <LOQ were 70.3% and 60.0%, respectively (P = 0.055).
Patients With Baseline CD4+ Cell Count <100 Cells/mm3
We performed a separate analysis of the 109 patients treated with ABC/3TC/ZDV with a baseline CD4+ cell count <100 cells/mm3, 72 (66.1%) of whom had prior ADCs. The median follow-up time for this subgroup of patients was 45 weeks (IQR: 30-73 weeks, minimum of 16 weeks to maximum of 152 weeks). The median baseline pVL in this group was 5.20 log copies/mL, and the median baseline CD4+ cell count was 40 cells/mm3. After a median follow-up of 45 weeks, 82 (75.2%) patients continued on the initial ART regimen. The safety and effectiveness of ABC/3TC/ZDV were as follows: interruption of ART because of adverse events in 18 (16.5%) patients, suspected abacavir HSR in 7 (6.4%), death in 4 (3.7%), and new ADCs in 4 (3.7%). Treatment failure by ITT OD analysis was found in 14 (12.8%) patients, resulting in a rate of 13.77 events per 100 person-years (95% CI: 6.60 to 21.00). The frequency and rate of treatment failure according to the ITT LSS = F analysis were 27 (24.8%) and 27.44 events per 100 person-years (95% CI: 17.1 to 37.8), respectively. Significant factors associated with treatment failure according to ITT OD by Cox analysis were adherence <90% (HR = 3.86, 95% CI: 1.33 to 11.15) and baseline pVL (HR = 3.12, 95% CI: 1.21 to 8.06 per log of pVL).
In this large case series of 730 consecutive naive HIV-infected patients who started ABC/3TC/ZDV therapy and had a median follow-up of almost 1 year, 14.25% discontinued therapy because of adverse events, the most frequent being a suspected HSR to ABC (4.93%) and hematologic toxicity (4.10%). The frequency of treatment failure by an ITT OD analysis was 14.4%. In a more rigorous approach considering losses to follow-up and interruptions or switches of ART, for whatever reason, as failures (ITT LSS = F), however, the frequency of treatment failure was 22.92%. Factors independently associated with treatment failure by ITT OD analysis were adherence, methadone use, baseline pVL, and prior ADCs, as previously identified in other clinical ART studies. An ITT LSS = F analysis showed that HCV infection and female gender were also independently associated with treatment failure, however, with the most plausible explanation being a higher frequency of treatment interruptions because of adverse events in these last 2 population groups.
Some randomized controlled trials have shown favorable virologic outcomes of ABC/3TC/ZDV when compared with PI-based regimens in ART-naive patients.1,2,9 This triple-NRTI combination has been found to be inferior to PI-based regimens in patients who begin treatment with a high pVL, however. For instance, in the CNA 3005 study, which compared the equivalence of Combivir plus ABC versus Combivir plus indinavir, patients randomized to the ABC arm with a baseline pVL >100,000 copies/mL were found to have a significantly poorer virologic response than patients in the indinavir arm.1
The triple-nucleoside regimen of ABC/3TC/ZDV has also proven to be virologically inferior to a regimen containing efavirenz and 2 or 3 NRTIs. In the AIDS Clinical Trials Group (ACTG) 5095 study, subjects were assigned to 1 of 3 regimens: Trizivir, Combivir plus efavirenz, or Trizivir plus efavirenz.3 After a median follow-up of 32 weeks, 21% of subjects in the triple-NRTI group versus 11% of those in the combined efavirenz groups had virologic failure (P < 0.001). These differences were evident regardless of whether the baseline pVL was > or <100,000 copies/mL. Changes in the CD4 cell count and the incidence of grade 3 or grade 4 adverse events did not differ significantly between the groups.
Our study has several limitations. First, it is observational and retrospective. Second, it does not have a comparator arm. Finally, the pVL assay method and LOQ varied between centers. Nevertheless, we believe that our findings related to the safety and effectiveness of ABC/3TC/ZDV in ART-naive patients complement what is already known from clinical trials. In this last experimental setting, results are sometimes difficult to generalize because the study population may be quite different from the population treated in normal life.10 In this regard, we would like to emphasize that our study included several hundred patients with frequent ADCs, profound immunosuppression, methadone use, and coinfection with hepatitis viruses who were followed up for a relatively long period.
It is well known that the risk of death or AIDS progression is increased in patients who start highly active antiretroviral therapy (HAART) with profound immunosuppression.11 Little is known about the optimal ART regimen for patients with low CD4 cell counts, however, because this population group was underrepresented in clinical trials during the HAART era.12 Because of the large size of our cohort, we had the opportunity to analyze a relatively large number of patients who started ART in a state of profound immunosuppression, characterized by a CD4+ T-cell count <100/mm3 and frequent prior ADCs. We found that the triple-NRTI regimen of ABC/3TC/ZDV was relatively well tolerated and effective in this group of patients and that poor adherence and baseline pVL were independently associated with treatment failure.
Our results may be related, in part, to the fact that we used a regimen with few drug-drug interactions and a low pill burden, particularly with the fixed-dose combination of ABC/3TC/ZDV (Trizivir), in which patients had to take only 2 pills per day divided into 2 doses. It is important to mention that in the CNA 3005 study mentioned previously, patients in both arms were obliged to take 16 pills daily (including placebos), along with hyperhydration and food restrictions. The ACTG 5095 study was also a randomized, double-blind, placebo-controlled trial in which subjects were obliged to take a total of 7 pills per day divided into 2 doses.
In summary, in this large observational study, we found that the triple-NRTI regimen of ABC/3TC/ZDV was sufficiently safe and efficacious for initial treatment of chronic HIV infection. Our findings reinforce the results of randomized clinical trials in that this combination is a reasonable option for ART-naive patients with a pVL <100,000 copies/mL in whom regimens based on PIs or NNRTIs are not advisable, even in patients with a baseline CD4+ cell count <100 cells/mm3.
The following investigators participated in this study: A. Alba, R. Alemán, C. Alonso, A. Antela, A. Asorey, P. Bachiller, F. Baguena, A. Ballesteros, C. Barros, J. Blanch, J. Cadafalch, A. Cano, M. A. Cárdenes, J. Casado, M. A. Castro, M. Cervantes, C. Cifuentes, B. Clotet, J. Cobo, M. Cordero, J. Cosin, M. Cousinou, A. Chocarro, J. De Otero, F. Dronda, J. Elizaga, A. Fernandez-Rial, J. Flores, D. Fuster, M. J. Galindo, M. C. Gálvez-Contreras, F. García, M. García-Basalla, J. García-Garcia, J. J. García-Lechuz, A. Gort, V. M. Gutiérrez, J. L. Gutierrez-Zufiaurre, M. Hayek, T. Hellín, J. J. Hernández-Burruezo, R. Hervás, F. Homar, M. Javaloyas, J. J. Jusdado, J. La Cruz, A. Lerida, M. Linares, J. Locutura, J. C. López, J. López-Aldeguer, E. Losada, F. Lozano, J. Mallolas, J. A. Maradona, O. J. Martínez-Madrid, P. Miralles, C. Miralles, J. M. Miró, J. Moltó, S. Moreno, M. A. Muniain, J. Muñoz-Sanchez, E. Negredo, A. Ocampo, L. Orbea, J. A. Oteo, B. Padilla, J. Pascua, J. Pasquau, A. Payeras, E. Pedrol, J. M. Peña, J. J. Peraire, M. Perpinyà, J. A. Pineda, H. Portillo, J. Portu, A. Prieto, E. Pujol, F. Pulido, C. Quereda, M. Ramírez-Schacke, B. Roca, J. Romeu, M. Rubio, J. Ruiz Morales, M. Sala, M. Salavert, M. A. Sambeat, M. Sánchez-Conde, J. Santos, J. Sanz, R. Serrano, G. Sirera, J. Sola, M. J. Tellez, J. M. Tricas, C. Tural, J. Usó, S. Veloso, A. Vergara, and C. Vilades. The authors thank Thomas O'Boyle for writing assistance during the preparation of the manuscript.