Viral diversity, a growing feature of the HIV pandemic

By Sheri Fink, MD, PhD*

The International AIDS Conference in Bangkok added new layers of subtlety to the understanding of HIV genetic diversity and the pace of its evolution.

The picture that emerged was that of an increasingly complex molecular epidemic changing at an even more rapid pace than previously appreciated. At this scientifically challenging time for the vaccine research field, the data presented provided new insights—and raised new questions—about how viral diversity and genetic mutability will impact vaccine design.


Investigators have tracked HIV diversity and geographic distribution, detecting a high prevalence of co-infections in some populations and increasingly identifying unique and circulating recombinants. A consensus in Bangkok was that better tools are needed to pinpoint the causes and effects of viral diversity within communities and individuals, and to better assess the impact of newly-acknowledged phenomena such as fluctuating populations of unique recombinant forms within individuals.

Co-infection can occur either when an uninfected individual is infected simultaneously with more than one genetic form of HIV or when an individual with an established HIV infection becomes infected with a new viral strain; the latter is known as “superinfection”. In either type of dual infection, the various viruses can recombine, leading to the development of unique recombinant forms (URFs) within a single individual, some of which could theoretically be transmitted to many other individuals, becoming circulating recombinant forms (CRFs).

“We need a lot more surveillance of populations, particularly newly-infected people, and with full-length sequencing,” said Chris Beyrer of Johns Hopkins University. “It’s not enough to… do serotyping with a few segments of the genome.”

Understanding co-infection

Some of the most interesting data came from Francine McCutchan of the Henry M. Jackson Foundation and her colleagues, who have spent many years defining intersubtype HIV-1 genetic diversity on several continents in preparation for vaccine trials. In Bangkok they presented data from the HIV-1 superinfection study (HSIS) that focuses on a cohort of high-risk (initial seroprevalence 68%) female bar-workers in Mbeya, Tanzania, where subtypes A, C and D co-circulate. Investigators hypothesized that some of the women would be infected with more than one viral subtype.

To detect intersubtype recombinants, McCutchan’s group developed a multi-region hybridization assay (MHA) (AIDS 16, 2055, 2002) using real-time PCR to screen DNA samples across five genomic regions (short regions ofgagpolvpu, and env) with probes specific for clades A, C and D; this specific assay is named MHAacd, and the group has developed similar genotyping screens for use in other geographic regions like South America and Asia. These “high throughput” assays offer a way to identify intersubtype recombinants more rapidly and less expensively than with full genome sequencing, the gold standard technique, and also have the power to identify dual infections.

In the Tanzania study, peripheral blood mononuclear cells (PBMCs) collected from a random sample of 98 HIV-infected women from the high-risk cohort were tested with the MHAacd assay every three months for a period of four to eight visits. Seventy five percent of the women had an identical HIV genotype every time they were screened, with a single probe hybridizing to each genome region, suggesting infection with a single HIV subtype—38% had only subtype C infection, 16% only subtype A, 3% only subtype D, and 43% a variety of single recombinant forms. However, 25% of the women appeared to be infected with more than one HIV strain, evident from either dual probe reactivity in a given genome region or a shift in subtype over time. Those with dual infections had A+C (37%), A+D (15%), C+D (4%), or A+C+D (44%) genotypes, which were interpreted as representing infection with more than one subtype and/or recombinant strain.

Focusing on women with evidence of dual infection, the research group has used cloning and partial sequencing to more fully characterize HIV strains and to quantify the relative proportion of these strains over time. McCutchan described four cases in which different genetic forms of HIV fluctuated dramatically over time in PBMCs from women infected with more than one subtype, based on evidence from at least 20 gag clones, with different genetic forms apparent, then absent, then reappearing. “People who are co-infected are the source of many, many recombinant forms that presumably they can pass along. And they’re continuously generated in those people over time” concluded McCutchan. As for the fluctuation of certain recombinants, she said “you can postulate of course that the immune response is suppressing some strains… there can also be a kind of chance element; strains lying around in memory T cells [may re-emerge] when those cells become reactivated by encountering their antigen.”

There are also new efforts to try to better understand the characteristics that make certain genetic forms of HIV more likely to be transmitted. In a recent small study (Science, 303, 2019, 2004), Eric Hunter (now at Emory University, Atlanta) and colleagues found that individuals newly infected with HIV had viruses whose envelope neutralizing epitopes appeared to be more exposed, rendering them more neutralization sensitive than most of the viral forms present in their infecting sexual partners. The implication of this work may be that there is some particular characteristic of a transmitted virus that enables it to establish a new infection.

Possibly related to such a phenomenon, issues of comparative viral fitness were also discussed at Bangkok. Eric Arts of Case Western University and colleagues at Antwerp’s Institute of Tropical Medicine and the Cleveland Clinic (Abstract number MoOrA1012) reported infecting PBMCs with pairs of viruses and allowing them to compete. In all the researchers compared 11 group M isolates (subtypes A, B, C, D and CRFO1), 5 group O isolates, and 6 HIV-2 isolates obtained late in disease progression. These in vitro results, some of which were reported in 2003 (J. Virol. 77, 1021, 2003; AIDS 17, 780, 2003) showed that relative fitness in the competition assay—group M > HIV-2 > O—matched relative worldwide prevalence. But the researchers found that within group M the least fit subtype was C, a major pandemic strain now responsible for roughly 50% of worldwide infections. Although data from an in vitro system must be extrapolated to real world scenarios with caution, the investigators are now testing the hypothesis that “lower replicative capacity…leads to slower disease progression, longer times for possible exposures and transmission events,” and thus a higher “transmission fitness.”

Another factor—immune pressure on an individual and population level—also influences which viruses circulate. At a Bangkok symposium on HIV vaccine design, Simon Mallal of Murdoch University, Australia, argued that vaccine design should take into account not only viral diversity but also the HLA (human leukocyte antigen) diversity of a population (ThSy274); his work has shown that HLA, a genetically-determined component of each individual’s immune response, exerts selective pressure on HIV in a manner analogous to antiretrovirals. During chronic infection, people with certain alleles of the 1300 HLA types develop HIV escape mutants; for example, HLAB*5701 and 5801 are strongly associated with the Gag T242N mutation. Mallal has been using information about HLA and the epitopes that stimulate the strongest protective immune response to try to design vaccine immunogens that will avoid falling prey to virus escape mutations.


McCutchan’s presentation of the HSIS study also touched on an issue that was a buzzword at the Barcelona AIDS conference two years ago—superinfection. In one case of superinfection from the HSIS Tanzania cohort, a patient who was HIV uninfected at the beginning of the study was found to have seroconverted at three months and had evidence of infection with a second strain at nine months. “Because we had serial samples, we were able to make strain-specific nested PCR and go back to the earliest samples and show by our most sensitive technique the second strain wasn’t present either in plasma or PBMCs… Nested PCR is supposed to detect anything. The evidence [for superinfection] is very strong and very convincing.”

Still, McCutchan stresses that moving beyond case studies is crucial. “Understanding dual infection in high-risk populations is really a key theme that’s coming to the forefront. Our own work in East Africa is part of the picture, but this kind of work needs to be repeated in high-risk populations, to identify how easy it is to get infected with a second strain and what the consequences are.” McCutchan and her collaborators in Germany are studying the immunological profiles of the Tanzania cohort. “We have to be married up with strong immunology to determine what the CTL and the neutralizing antibody responses are in individuals who get co-infected versus those who do not in a population with similar risk.”

Several other presentations addressed the superinfection issue but key questions remain unanswered; how common is superinfection as opposed to simultaneous co-infection, and what implications, if any, does the phenomenon have for vaccine-induced immune responses?

Davey Smith of the University of California at San Diego offered three case descriptions of superinfection occurring an estimated five to thirteen months after initial HIV infection (TuOrB1140). After infection with a second HIV strain each patient experienced an increase in viral load and a decrease in CD4+ T-cell counts, suggesting to Smith that superinfection can “negatively impact individual clinical course.”

Todd Allen of Massachusetts General Hospital co-chaired the superinfection forum, which also featured a presentation by Tuofu Zhu of University of Washington on long-term HIV-exposed seronegative individuals who ultimately seroconverted. “They had HIV positive partners and weren’t infected by them, but by someone with a more distantly related [virus],” said Allen. “It’s pretty good suggestive evidence that there may be some component to an immune response that is protective in these exposed uninfected individuals…[The researchers] used the virus as a tool to measure whether or not there may be some truth to the idea that a component of the cellular immune response protected.”

Across the globe, an increasingly diverse pandemic

The consensus at Bangkok, with reports based on near full-length genome sequencing coming from disparate locales like west Africa, the Myanmar/China border region, Thailand, and South America, was that recombination is on a marked upswing, and not only because better tools are available to assess it. Chris Beyrer co-chaired one of the oral presentation panels: “It was very striking…viral diversity is increasing about as quickly as people are able to measure it.”

On a population level, groups from around the world reported finding new HIV recombinant forms in increasing prevalence, particularly among highly-exposed populations or in areas considered epicenters of the HIV pandemic. Some groups applied sequencing and phylogenetic analyses to preserved serum samples, finding evidence of recombinants stretching back for decades. For example, samples collected in Kinshasa in 1984 and 1986 were analyzed by Marcia Kalish and colleagues at the US Centers for Disease Control, the National Institutes for Health, and Project SIDA in Kinshasa (ThOrC1362), in a recently published study (Emerg. Infect. Dis. 10, 1227, 2004). Analyzing just 10% of the HIV genome yielded evidence for at least 37% intersubtype recombinant viruses, “likely a low estimate,” according to Kalish. “Virtually all the HIV-1 subtypes [with high intrasubtype diversity] and unclassifiable gene regions were identified.” In contrast to these central African samples, HIV samples collected in 1986 from Burkina Faso in west Africa showed much lower diversity, where almost 95% of circulating strains comprised two recombinant forms, CRF02 and CRF06, with a low intra-CRF genetic diversity, suggesting that the HIV epidemic was introduced there at a later time by a limited number of founder viruses.

Several investigators used the real-time generation of new forms to identify dynamic epidemics and to connect and relate epidemics in different geographical locations. There was a strong emphasis in Bangkok on characterizing emerging and established epidemics in the region. Here and also in other regions, molecular tools are painting an alarming portrait of border-crossing epidemics driven by injection drug use and sex trafficking—activities that many governments have been slow to address.

Sequencing viral genomes in seven countries across South America, Jean Carr of the US Military HIV Research Program and South American colleagues found a significant proportion of diverse BF recombinant viruses in an emerging epidemic among female commercial sex workers in the Southern Cone of the continent (ThOrC1365). Yutaka Takebe and colleagues from Japan’s National Institute of Infectious Diseases (ThOrC1364) studied viral strains from injection drug users in Central Myanmar and western Yunnan, China, finding these areas to be “geographical hotspots of extensive recombination.” Fine mapping of recombination breakpoints revealed that some Myanmar URFs shared precise breakpoints with CRFs circulating in Yunnan, China, suggesting a link in the epidemics and the possibility of common ancestors. “Connecting the epidemics in Burma and China is news,” says McCutchan. Takebe drew lessons for vaccine researchers. In an email, he wrote: “Our study suggests that the mixing of different lineages of HIV-1 strains in highly exposed populations and in the social networks…could quickly lead to the evolution of new forms of recombinants [which could], for instance, acquire new sets of ‘escape’ mutations in viral epitopes by shuffling different parts of the genome and complicate the development of effective HIV vaccines.”

Looking to the future

Presenters seemed to agree that increasingly there is even more viral diversity than previously appreciated. “On at least three continents there is good evidence of recombination between subtypes, generating both CRFs but also a host of novel one-off recombinants really found only in individual patients [URFs],” says Beyrer. “What’s driving [this increased diversity] and where it’s going is hard to say.”

It remains for researchers to sort out the implications of HIV diversity for vaccine design. “What may be more worrisome is not that recombination is happening and circulating recombinant forms are generated, but rather the increasing level of these novel recombinants [URFs], reaching levels of 30 or 40% of new infections, suggesting a tremendous amount of viral diversity particularly in east Africa,” says Beyrer. “It gets worrisome if you agree that what we need is a vaccine developed not on a single variant but a consensus sequence. That gets tough if there are multiple recombinants and there really isn’t a consensus sequence. One approach is to look for ancestral sequences.” This approach has been promoted by researchers such as Beatrice Hahn and Bette Korber. Other ideas are to consider using multi-epitope constructs or multivalent vaccines.

The high level of HIV diversity is a fact, and one that experts are viewing as a huge challenge to vaccine development but also in some respects as an opportunity. Techniques to study the molecular diversity of HIV, such as the McCutchan group’s multi-region hybridization assays, have been developed specifically to aid researchers preparing for vaccine trials. “This is an opening field, at its birth,” she says.

Even superinfection may offer more than negative implications for the vaccine research field. “It presents an excellent opportunity to try to address the ability of existing immune responses to cross-react against the virus,” says Allen. “A lot of the initial negativity over the superinfection studies was that people were starting to extrapolate this to vaccine design. But it wasn’t a fair parallel to draw because these patients are immunocompromised so we don’t know what accounts for the lack of protection. Our understanding of superinfection is still quite limited, but it does provide a unique opportunity to look at the ability of primed T cells to cross-protect ex vivo. How well are these T cells able to recognize a second clade B or C virus?” McCutchan echoes the idea that vaccine researchers should use dual infection to their advantage. “It’s the only in vivo system I know of to determine whether the immune responses generated by natural infection have the breadth of protection that we’re looking for in a vaccine,” she says. “It’s an opportunity. I’m going to stay tuned. I hope you will, too.”


*Sheri Fink, MD, PhD, is a freelance writer whose work has appeared in such publications as the New York Times and Discover Magazine, and the author of "War Hospital: A True Story of Surgery and Survival.”