Can HIV-1 superinfection compromise antiretroviral therapy?

Antiretroviral therapy has been able to restore health to HIV-infected patients, and this success may have played a role in the increasing rates of unsafe sex among HIV-infected individuals [1]. In addition, recent studies have shown that chronically HIV-infected individuals may be superinfected with a second HIV-1 strain [2–5]. We used a small cohort of patients to test whether the occurrence of antiretroviral therapy failure might be related in some instances to superinfection with HIV-1 drug-resistant variants. To detect potential HIV-1 superinfection events, we analysed samples from the specimen repository at the Center for AIDS Research, Case Western Reserve University, Cleveland, Ohio, USA. Plasma samples from HIV-infected individuals who suddenly experienced virological failure were identified. Patients who were receiving antiretroviral therapy, who had non-progressive HIV-1 infection with low to undetectable plasma viremia (<400 copies/ml), and who unexpectedly experienced virological failure (more than one log increase in the plasma viral load in less than a month) despite recorded good adherence to treatment were selected. Ninety-four patients were originally identified, and two samples were analysed, one obtained before and one after virological failure, from six patients with the highest increase in viral load. Two additional clinical cases were included, which had a suspicion of HIV-1 superinfection after having unprotected sex with HIV-infected individuals. Plasma viral RNA was reverse transcribed and the protease-coding region, the reverse transcriptase (RT) polymerase domain, and the C2V3 gp120-coding region were polymerase chain reaction amplified and sequenced [6]. Intra and interpatient nucleotide sequence diversities were calculated and neighbor joining trees were constructed to compare the phylogenetic relationships of pol and env global sequences [6]. HIV-1 sequences from seven patients formed monophyletic clusters in each genomic region analysed. The median nucleotide distance between intrapatient pol and env sequences was 1.5 and 4.1%, respectively. For patient 3, however, the pol and env longitudinal sequences did not cluster together in the phylogenetic trees, and consequently higher nucleotide distances were obtained (i.e. 5% in pol and 15% in env) (Fig. 1a). This highly antiretroviral-experienced patient showed a sudden increase in viral load (from 12 532 to 662 889 copies/ml) in the short period between samples 3A and 3B (9 November 1998 and 7 December 1998, respectively) (Fig. 1b).Fig. 1.: Screening for HIV-1 superinfection in patients failing antiretroviral therapy. (a) Phylogenetic analysis of 16 C2V3 gp120-coding region sequences from HIV-infected individuals with sudden increase in viral load. Bootstrap resampling values (1000 sets) greater than 75% are indicated by an asterisk. Longitudinal samples from patient 3 are indicated by black boxes. The HIV-1HXB2 isolate was used as an outgroup. The distance between two sequences is obtained by summing the length of the connecting branches, using the corresponding scale (in nucleotide substitutions per nucleotide; s/nt). (b) Antiretroviral treatment history, plasma HIV-1 RNA levels (—♦—), and CD4 T-cell counts (—▪—) from patient 3. Time points of samples 3A and 3B are indicated. ABC, abacavir; ADF, adefovir; APV, amprenavir; ddI, didanosine; d4T, stavudine; LPV, lopinavir; RTV, ritonavir; SQV, saquinavir; 3TC, lamivudine; ZDV, zidovudine. (c) Phylogenetic analysis of eight longitudinal C2V3 gp120-coding region sequences from patient 3 collected in a 6-month period. (d) Multiple sequence alignments of 88 amino acids from the reverse transcriptase (RT) polymerase domain and the C2V3 gp120-coding region (Env) of eight longitudinal plasma samples from patient 3. Amino acid substitutions associated with resistance to RT inhibitors for each HIV-1 isolate are indicated in bold (as described in International AIDS Society – USA [7]).To investigate this ‘multiple infection’ by phylogenetically unlinked viruses further, six extra plasma samples were analysed. All longitudinal sequences in both genomic regions clustered together, with the exception of sample 3B (Fig. 1c). Exhaustive quality control experiments ruled out potential cross-contamination with other sequences in our laboratory (data not shown). Although sequences before and after the rapid increase in viral load showed a profile of drug-resistance mutations compatible with the antiretroviral history of the patient, sequence 3B was wild type in both protease and RT and showed significant amino acid diversity in the env gene (Fig. 1d). A detailed analysis of his clinical history revealed that patient 3 was not adherent to antiretroviral therapy during the period between the collection of samples 3A and 3B, which could explain the rapid increase in viral load (Fig. 1b). Was this the result of transient superinfection with a wild-type (more fit) HIV-1 isolate during treatment interruption or the selection of a pre-existent virus established during an original HIV-1 co-infection? Previous studies have described HIV-1 superinfections with different wild-type viruses [2,3], or by wild-type HIV-1 strains re-infecting patients harboring drug-resistant viruses after a short period of treatment interruption [4,5]. The latter could have been the case for patient 3, but without the source of the second genetically distinct HIV-1 strain, it is difficult to distinguish between consecutive or simultaneous infection. There is still the possibility, although not described to date, that patients infected with wild-type HIV-1 isolates and under successful antiretroviral therapy may be exposed to drug-resistant viruses. These mutant strains would have a significant selective advantage in this new environment, which could lead them to outcompete the original wild-type HIV-1 strain and cause antiretroviral therapy failure. We have recently demonstrated that, in vitro, the superinfecting HIV-1 strain must have a phenotypic advantage (increased fitness) over the original virus, which favors superinfection (unpublished results). However, the probability that an individual infected with wild-type HIV-1 and receiving antiretroviral therapy will be exposed to drug-resistant HIV-1 strains may be low. A recent study [8] was not able to detect HIV-1 superinfection after analysing protease and RT sequences from more than 700 patients. Conversely, high rates of transmission and prevalence of HIV-1 drug resistance mutations in antiretroviral-naive patients have been reported [9], which implies that the theoretical risk of re-infection with mutant HIV-1 strains exists. Therefore, HIV-seroconcordant partners should be informed of the theoretical risk of superinfection, especially when only one partner is receiving antiretroviral therapy.