HIV Superinfection: Can Patients Be Infected Twice?
By Joel N. Blankson, M.D., Ph.D.
How does one counsel two HIV-infected individuals who are in a monogamous relationship? Do they really need to still use condoms? The answer to this complicated question depends in part on whether or not an HIV-infected patient can be infected with another HIV strain. It is now known that dual HIV infection, defined as the presence of two distinct HIV strains in a patient, can occur. The best evidence for this phenomenon is the existence of recombinant viruses that contain genetic material from two different parental strains. Recombinant viruses can only be created by the simultaneous replication of the two parental strains in the same patient. There are two potential mechanisms by which dual infection can occur: coinfection, in which a host is infected by two distinct HIV strains at or around the same time, and superinfection, in which there is sequential infection of a patient by two different viral strains. Distinguishing between these possibilities has been an area of intense research. The evidence for and against these possibilities will not only affects how physicians counsel patients, it also has major implications for vaccine development.
A case report in 1995 from David Ho's laboratory demonstrated that HIV coinfection can occur [Zhu T, et al. J Virol 1995;69:1324]. An acute serocoverter with exposure to multiple sexual partners was shown to be infected with multiple distinct HIV clade B isolates. The degree of genetic difference between two of these strains strongly suggested that the patient had been infected by different partners. Superinfection was ruled out by a negative HIV Western blot at the time of presentation. Interestingly, there was evidence of recombination between two of the strains. A second report demonstrated coinfection in an infant who received transfusions from two different HIV-infected donors [Diaz RS, et al. AIDS Res Hum Retroviruses 1996;12:1291]. Recombination between the two distinct strains was also observed. This was direct evidence that coinfection could occur and could lead to recombinant viruses.
An elegant animal model of HIV infection has demonstrated that coinfection occurs more readily than superinfection. Macaques could be easily infected with two distinct strains of HIV-2 if the animals were inoculated with both viruses at the same time. However, if the animal was inoculated with one virus first, then, following a two-week window, it became progressively more difficult to infect the animals with a second virus [Otten RA, et al. J Infect Dis 1999;180:673]. The reason for this difference is unclear, but one hypothesis has been that the immune response generated to the first virus provides protection against infection by a second virus. This may seem counterintuitive at first, given that the immune response of many HIV-seropositive patients doesn't seem to confer protection against the replication of autologous virus. But the inoculating dose of a second virus in most cases will be relatively small, and thus even a partially effective immune response may be able to prevent superinfection.
Seven years after the initial reports of coinfection, three reports of superinfection were published. The first two studies involved cases where the initial virus and the superinfecting virus were from different HIV clades. Ramos and colleagues demonstrated superinfection of two injection drug users who were enrolled in prospective cohort study in Thailand [Ramos A, et al. J Virol 2002;76:7444]. Clades AE and B viruses are the predominant isolates in this region. The subjects were screened for the presence of recombinant clade AB virus at different time points. Two patients were initially found to be infected with virus from a single clade. Approximately one year after sero-conversion, only recombinant AB virus was found, suggesting that superinfection had occurred and had led to recombination.
Jost and colleagues reported a case of a patient who had been placed on HAART during primary infection [N Engl J Med 2002;347:731]. He was on a supervised treatment interruption 2 years later when he was noted to have a rise in his viral load and symptoms of fatigue and fever. Analysis of his plasma virus at this time revealed clade B, virus whereas only clade AE virus had been detected at earlier time points. The patient had had a high-risk sexual exposure three weeks earlier in a country where clade B virus is endemic. Phylogenetic analysis of the 2 viruses confirmed that they were distinct strains, leading to the conclusion that super-infection had occurred.
In a recent study of highly exposed, HIV-positive commercial sex workers from Kenya with chronic infection, one individual who was initially infected with clade A virus was found to have only recombinant clade A/C isolates 10 years after seroconversion [Fang G, et al. AIDS 2004;18:153]. This patient had experienced a febrile illness and a drop in her CD4 count a few years before this. A detailed analysis revealed no evidence of clade C virus shortly after seroconversion, suggesting that superinfection had occurred with a clade C isolate leading to the formation of the recombinant AC isolates.
Viruses from different clades have significant differences in their genetic composition. Therefore, it is not surprising that an HIV-specific immune response to the initial virus was not able to prevent superinfection by the second isolate. More concerning are two reports of superinfection by viruses from the same clade. The first involved another patient who had been treated with HAART during primary infection and who was then placed on a supervised treatment interruption [Altfield M, et al. Nature 2002;42:434]. Super-infection was detected when he experienced a sudden rise in his viral load and a significant drop in his CD4 count after having demonstrated excellent immune control of viral replication for close to a year. Phylogenetic analysis revealed that the circulating virus was genetically distinct from the isolate the patient had originally been infected with 3 years earlier. The patient had a high-risk sexual exposure 2 to 3 months prior to the rise in his viral load and had experienced symptoms consistent with the acute retroviral syndrome. Of note, he had a very broad CD8 T-cell mediated response to the first virus prior to superinfection, yet his immune system was unable to prevent infection with a second virus from the same clade. Subsequent studies, however, have shown that the assay used to analyze the CD8 T-cell responses in their study may not be a good predictor of HIV specific immunity [Addo MM, et al. J Virol 2003;77:2081].
Another noteworthy study demonstrated superinfection by a wild type clade B virus in a treatment-naïve person who was initially infected with a multidrug-resistant clade B isolate [Koelsch KK, et al. AIDS 2003;17:F11]. The wild type isolate predominated shortly after superinfection, but surprisingly, subsequent studies showed that the two viruses had similar replication capacities as determined by the virologic assay [Daar E, et al. Abstract 394, 11th CROI, San Francisco, 2004]. This illustrates the point that "viral fitness" is determined by more than just mutations in the protease and reverse transcriptase genes. In both cases of clade B superinfection, it was found that the superinfecting virus differed genetically from the initial virus in key regions or epitopes that were targeted by the immune system. This may have allowed the viruses to evade the immune response and thus to out-compete the initial virus.
It should be noted that all of these studies have failed to definitively rule out coinfection as the cause of dual infection. The inability to track the second virus could simply be due to a sampling error. A change in selective pressure could lead to preferential replication of one isolate and this could explain why the second virus was detected at a later date. The fact that some of these individuals had high risk exposures and symptoms of the acute retroviral syndrome shortly before the second virus was detected makes it more likely that superinfection had indeed occurred.
There have only been five case reports of HIV superinfection in the literature, but it is possible that superinfection occurs much more frequently and is simply not detected. Two recent large studies would argue against this possibility. Gonzales and colleagues looked at the protease and reverse transcriptase genes of HIV isolates obtained from 718 individuals [J Infect Dis. 2003;188:397]. The sequence of these genes was followed longitudinally to screen for superinfection. Major genotypic changes were seen in some subjects, but this was mostly due to the loss or gain of drug resistant mutations associated with changes in antiretroviral therapy (ART). Sequence from tat and gag, genes not affected by changes in ART, were not significantly different in these cases. Thus superinfection was highly unlikely. They found no cases of superinfection during 1072 person-years of observation.
In a similar study, Tsui and colleagues performed longitudinal analyses of the gag and env genes from isolates obtained from 37 injection drug users [Tsui R, et al. J Virol 2004;78:94]. They found no evidence of superinfection over 215 person-years during which there was continued high risk exposure. They estimated that based on the reported exposure, there should have been 3.4 episodes of superinfection. They thus concluded that existing HIV infection conferred a significant degree of protection against infection with an isolate from the same clade. Both of these studies are limited by the relatively low sensitivity of the methods used to screen for superinfection. Neither assay would routinely pick up an isolate that accounted for less than 20% of the circulating virus. Thus if a superinfecting virus did not out compete the existing virus, it would probably not be detected.
In a preliminary report, Smith and colleagues used similar techniques to look at 54 patients with early HIV infection who had deferred treatment with ART [Smith D, et al. Abstract 21, 11th CROI, San Francisco, 2004]. Curiously, three cases of super-infection were detected in 46 person-years of follow up. One patient, who was initially infected with a wild type strain, later became infected with a drug-resistant isolate. Superinfection generally was associated with an increase in viral load (mean of 1.6 log) and a decrease in CD4 count (mean 132 cells/mm3). It's unclear why there is such a discrepancy between this study and the prior two larger studies. It should be noted that people with early HIV infection have a more homogeneous population of isolates than chronically infected persons, and thus a new viral isolate might be easier to detect. Alternatively, individuals with early infection may be more susceptible to superinfection, possibly because their HIV-specific immune response is still evolving.
There are now sufficient data suggesting that superinfection can occur. Significant drops in CD4 counts and transient increases in viral loads often accompany super-infection. Furthermore a recent retrospective analysis showed a very rapid progression from seroconversion to clinical AIDS in five untreated patients with dual HIV infection (range of 1.0 to 3.4 years) [Gottlieb GS, et al. Lancet 2004; 363:619]. It should also be noted that patients on HAART with undetectable viral loads may be at increased risk of superinfection, as the levels of HIV-specific neutralizing antibody and effector CD8 T-cells have been shown to decline over time when viral replication is suppressed [Morris MK, et al. J Acquir Immune Defic Syndr 2001;28:405, Casazza JP, et al. J Virol 2001;75:6508]. There is also the danger of infection with drug-resistant viruses, which would have a selective advantage over wild type virus in patients on effective antiretroviral therapy. Based on these findings, HIV-infected people should be advised to practice safe sex with other HIV-infected partners in order to prevent HIV superinfection and all of its associated consequences. Additionally, safe sex practices are indicated to prevent transmission of other sexually transmitted infections.
Return to top | Next Article: Syphilis Rates Climb Again