Frequency and Implications of HIV Superinfection

Abstract

HIV superinfection (HIV-SI) occurs when an HIV infected individual is infected with a new distinct HIV viral strain. HIV-SI has been reported throughout the world, and studies have reported HIV-SI incidence rates of 0% to 7.7% per year. Recent use of next-generation sequencing has improved detection of HIV-SI. Several studies have suggested that HIV-SI may occur at rates comparable to initial HIV infection. HIV-SI has been shown to be transmitted by injecting drug use, heterosexual and homosexual intercourse. Clinicians should encourage safe sexual and injecting drug use practices for HIV-infected patients since HIV-SI has detrimental effects on clinical outcomes, and may pose a concern for large scale antiretroviral treatment plans. The occurrence of HIV-SI has implications for vaccine research since it appears natural HIV infection is not fully protective against a subsequent HIV infection. Additional collaborative HIV-SI research is necessary which might benefit patient care and inform future vaccine design.

Introduction

HIV superinfection (HIV-SI) occurs when an HIV-infected individual becomes re-infected with a new phylogenetically distinct viral HIV strain. The possibility of HIV-SI was first demonstrated by the observation of co-infection of both HIV-1 and HIV-2, which are evolutionarily distinct viral species, but share approximately 42% of nucleotide homology in their envelope genes.^1–3

Additional suggestions of HIV-SI came from HIV-1 recombinant forms, which are HIV virions that contain separate genomic sections from distinct HIV-1 subtypes. HIV-1 is differentiated by genetic sequence into nine HIV-1 subtypes (A, B, C, D, F, G, H, J, and K). The subtypes are associated with different rates of disease progression, viral load, detection method assay sensitivity, and distinct geographic regions.^{4, 5} HIV-1 virions are diploid and viral strains are able to recombine when two distinct subtypes infect a single cell. If this new recombinant strain is transmitted it can become a circulating recombinant form (CRF). Approximately 10% of all HIV-1 infections involve recombinant viruses, and this high rate of CRFs provided further evidence of HIV-SI.⁶

While HIV-SI has been strongly suspected for many years, it was difficult to distinguish whether individuals were infected with two distinct HIV-1 viruses simultaneously (HIV-coinfection or dual infection), or whether there was a secondary infection after the initial infection (HIV-SI). The lack of well-documented longitudinal samples and sensitive techniques for detection of HIV-SI prevented documentation of HIV-SI until 2002.^7–9 Understanding both intra-and inter-subtype HIV-SI, is not only important for appropriate patient management, but may also provide insights into viral evolution and host immune responses to repeat HIV challenges pre- and post-superinfection. Such information could have substantial implications for HIV vaccine development, global public health efforts and patient care.

Detection of HIV Superinfection

The initial studies that identified individuals dually infected with HIV-1 and HIV-2 utilized serological assays that could easily distinguish between the two viral species.³ This approach, however, cannot distinguish between different HIV-1 subtypes or strains. The initial cases of HIV-SI were identified in injecting drug users in Thailand by performing a restriction fragment analysis on amplified viral sequences from longitudinal samples followed by confirmatory viral sequencing.⁹ At the same time, two separate HIV-SI cases were identified in two homosexual men being monitored as part of larger clinical studies after they experienced unexplained spikes in their set-point viral loads.^{7, 8} Samples from these individuals prior to and after the spike were subsequently analyzed by subtype specific PCR amplification⁷ or direct sequencing⁸ to confirm the presence of new viral populations. Other groups subsequently utilized these strategies of screening populations for spikes in viral load^10–14 or subsequent restriction fragment analysis¹⁵ followed by direct sequencing to identify HIV-SI.

After these initial studies, three diagnostic strategies emerged for screening HIV-SI in populations; heteroduplex mobility assays (HMA),^16–18 multi region hybridization assays (MHA),¹⁹ and bulk viral sequence analysis^20–25 followed by selective cloning of those samples that suggested emergence of new viral variants. MHA screening is limited by the fact that it only can identify inter-subtype superinfection; whereas HMA can detect samples with greater than 1.5% differences in genetic distance. However, HMA is susceptible to false positives due to insertions or deletions.²⁶ Examining bulk sequencing for changes in the viral population is achieved by either looking for new phylogenetic species at a later time point or quantifying the amount of degenerate bases in a given sequence. The sensitivity of this strategy relies on the likelihood of amplifying the new viral population and not just the original strain. Not surprisingly it was found that examining degenerate bases poorly detected minor variants at levels ≤20%.²⁶ Additionally, all these methods require confirmation using cloning and sequencing.^27–29

In order to avoid issues associated with less sensitive screening techniques, a variety of populations with differing risk behaviors were screened for HIV-SI by either single-viral genome amplification (SGA)^{27, 28} or cloning^{30, 31} and then compared for evidence of phylogenetically distinct species. The ability of these two approaches to accurately identify HIV-SI events relies heavily on the amount of sequences generated for each sample.^{28, 29} The sensitivity of all sequence based assays for identifying HIV-SI are also dependent on the number of viral genomic regions evaluated.²⁸ These requirements make these strategies prohibitively expensive and too labor intensive for large scale studies.^{29, 32}

Due to these limitations, we and others have recently designed and verified next-generation sequencing (NGS) assays that have the ability to detect minor variants at 1% or lower of the total viral population in a high-throughput manner.^{29, 33, 34} We subsequently used this assay to identify HIV-SI in both virally discordant couples and in a large population of African HIV seroconverters^{34, 35}(Figure1). NGS based assays allow one to combine the screening and verification steps into one, and can be done in a relatively high-throughput manner allowing for the accurate and timely assessment of large at-risk populations or on a single patient level.³⁵ In addition, it was found that NGS based assays could capture more viral diversity than an SGA based approach at 40% of the cost and with 20% less labor.²⁹ Consequently, NGS is becoming the assay of choice for HIV-SI detection.

Figure 1. Representative phylogenetic tree for identification and confirmation of HIV-SI.

Phylogenetic tree of consensus viral sequences (≥10 NGS reads) of a recent seroconversion sample (red) and a follow-up sample from two years later (blue) demonstrating an inter-subtype superinfection event. The diagnostic requirements for HIV-SI are shown with a selection of subtype reference and random sequences from individuals in Uganda shown (black).

Currently, determining the timing of the HIV-SI events relies on the sampling intervals available for the cohorts being examined. However, as more is learned about the rates of evolution and recombination after HIV-SI occurs it is theoretically possible to better estimate timing of HIV-SI using phylogenetic and evolutionary modeling analyses.

Epidemiology

HIV-SI has been reported throughout the world. Observational studies and case reports have documented HIV-SI cases in North America (both United States^{8, 23, 31, 36–38} and Canada³⁹), Europe,^{7, 10, 13, 22, 40–42} Australia,¹¹ Asia,^{9, 15} and Africa^{17, 20, 21, 27, 28, 35, 43} (Figure 2). Except for a small study in Brazil which did not find HIV-SI when using less sensitive detection methods than NGS,¹⁶ HIV-SI has been reported in numerous countries and in multiple risk populations (heterosexual, homosexual, injecting drug use). The widespread observation of HIV-SI suggests that it is a substantial problem that has been underreported.

Figure 2. Worldwide cases of documented HIV-superinfection.

Countries with HIV-SI cases reported from observational studies are represented in dark green. Countries with documented case reports are represented in light green. HIV-SI has been reported in North America (both United States,^{8, 23, 31, 36–38} and Canada³⁹), Europe,^{7, 10, 13, 22, 40–42} Australia,¹¹ Asia,^{9, 15} and Africa.^{17, 20, 21, 27, 28, 35, 43} Except for a small study in Brazil which did not find HIV-SI when using less sensitive detection methods than NGS,¹⁶ all countries in gray as reported in the medical literature have not been evaluated for HIV-SI. Text boxes indicate the mode of transmission [heterosexual (HS), female sex workers (FSW), men who have sex with men (MSM), injecting drug users (IDU)] and number of cases/number screened by country for all observational studies that detected HIV-SI.

HIV-SI has been found in individuals with varying modes of transmission. Two of the initial case reports of HIV-SI were among men who have sex with men,^{7, 8} and supported by several observational studies.^{23, 37,40} In both Thailand and the United States, it has been shown that IDUs can acquire HIV-SI.^{9, 15,31} In addition, there have been numerous reported cases of HIV-SI among female sex workers.^{17, 20, 27, 28,43} While HIV-SI was first studied in high risk populations, HIV-SI was recently demonstrated to occur at a significant level among a general heterosexual population in rural Africa.^{34, 35}

Several case reports and population-based studies have reported HIV-SI incidence rates of 0% to 7.7% per year.^{23, 27, 28, 43–45} The incidence rates can vary substantially by the type of population evaluated, the frequency of antiretroviral drug (ARV) use, the length of population follow-up and the detection methods used. These differences in study design have led to several negative studies^45–47. However, a recent study using NGS found HIV-SI incidence rates which are comparable to the rate of initial HIV-infection in the same area.³⁵ This study also highlighted the difficulty in comparing rates of HIV-SI to HIV incidence in a population given the fact that HIV-infected individuals are inherently higher-risk than their uninfected comparison group.³⁵ This and others studies performed in regions with multiple circulating subtypes have found both inter- and intra-subtype superinfection events, and there does not appear to be a predisposition for either form.^{27, 28, 34, 35, 48} It will be important for future studies that compare rates of HIV-SI to primary HIV incidence to consider individual risk practices as well as ensure that the study is powered to identify a difference.

One additional problem for the HIV field is that risk factors for HIV-SI are not well known. It is suspected that acquisition of HIV-SI shares the same risk as primary infections, such as increased number of sexual partners,⁴⁹ non-marital relationships,^{49, 50} limited condom use,^{51, 52} lack of ARV use,^{52, 53} and absence of male circumcision.^{54, 55} However, the risk factors of HIV-SI have not been clearly defined due to the limited number of cases. This is a critical area of future research to better understand the frequency and risk for HIV-SI acquisition on a population level.

It was originally hypothesized that HIV-SI is most common during the initial stage of primary infection due to the lack of immunity, and this was initially supported by numerous studies and case reports.^{9, 12, 23, 56} The close timing of the primary infection and HIV-SI, however, are likely due to the convenient sampling techniques that have been previously used to detect HIV-SI. The time to HIV-SI has been previously reviewed,⁵⁶ and it has been shown that HIV-SI can occur more than two years after primary infection.^{7, 8, 13} Determining if there is a predisposition in the timing of HIV-SI is important to our understanding of the role that the immune system has on protecting individuals from a second HIV challenge. More long-term follow-up studies are needed to address this issue.

Transmission and the global pandemic

Modelling studies have been performed to estimate the effect that varying amounts of HIV-SI could have on levels of recombination.^{57, 58} These models suggested that HIV-SI was most likely a pivotal component in the creation and maintenance of recombination rates within a community. However as stated by the authors, these models relied on many epidemiological assumptions that will need to be clarified before the accurate impact of HIV-SI on the global pandemic.

One such issue is how often HIV-SI leads to subsequent transmission to uninfected partners; and if the original, superinfecting, or a recombinant strain is transmitted (Figure 3). Although no study to date has addressed this question directly, several studies have attempted to find linked HIV-SI events by examining virally discordant HIV-infected partners.^{21, 30, 34, 37} In two such couples, the HIV-SI events were found to be linked as verified by clonal sequencing²¹ or in-depth recombination analysis.³⁷ Given the rarity of these linked cases it will be important that groups attempting to investigate HIV-SI induced transmission events combine their findings as much as possible.

Figure 3. Hypothetical HIV-SI transmission pattern.

Sequence of a typical heterosexual sexual network involving initial transmission with an original viral strain (red) followed by superinfection with a new viral strain (blue). This can result in the newly superinfected partner transmitting one or both of their two strains or a new recombinant strain (purple) to a subsequent partner. Although this demonstrates a heterosexual pattern, a similar pattern would be anticipated for MSM and IDUs.

Pathogenesis

As previously discussed, the initial reports of HIV-SI, as well as other later studies, used distinct spikes in HIV viral load as a way to identify new HIV-SI cases, and it is generally accepted that in most cases HIV-SI will cause this type of response(Figure 4).^{7, 8, 56,59} It is unclear, however, if HIV-SI results in a sustained rise in set-point viral load. Several studies have found an increased viral load in individuals who have evidence of HIV-SI or dual infection;^{10, 23, 60, 61} whereas others found no such link.⁴⁰ In two studies that examined confirmed HIV-SI events from Africa, no consistent pattern of increased viral load set-point was found even though set point viral load increased by 0.5 login 7 of the 16 cases identified.^{27, 28, 35} Taken together, these data suggest that HIV-SI may lead to an increased viral load on a population level, but that it is not a necessary phenotype of HIV-SI.

Figure 4. Representative clinical and virological aspects of HIV-SI.

Putative representative graph indicating the HIV viral load (solid black), CD4+ cell count (red), CD8+ cytotoxic T-lymphocyte response (dotted black), and anti-HIV neutralizing antibody (NAb) response (purple)of a typical HIV-SI case. The graph demonstrates a spike in viral load after HIV-SI similar to what is found during acute infection which may or may not result in a higher set-point viral load. An increase in NAb response is also shown post-SI.

Although viral load is a strong predictor of HIV disease progression, the effect of HIV-SI on overall disease progression is unclear.⁵⁶ Multiple studies have linked HIV-SI and dual infections to more rapid CD4+ cell loss.^{19, 23, 56, 62} Yet, some studies that have looked at HIV-SI and dual infections have not found a significant difference in CD4+ cell loss.⁶¹ One caveat to these studies is that the number of HIV-SI cases in these studies are relatively limited so they are often grouped together with dual infections. Gottlieb et al. described an interesting case of a subject who was superinfected with a highly pathogenic dual-tropic HIV strain (utilizing both CCR5 and CXCR4 as a co-receptor for cellular entry) which led to rapid disease progression.³⁶ In addition, others have documented cases of HIV-SI causing either long-term nonprogressors or elite controllers to subsequently progress to disease.^{41, 63} However, another study documented HIV-SI in an elite controller who regained some control of the infection post-superinfection.⁶⁴ A mathematical model that aimed to predict this effect estimated that only HIV-SI by a more fit strain would result in faster disease progression.⁶⁵ These data suggest the full extent and potency of the detrimental effects of HIV-SI remain unclear, and may depend on a variety of viral and host parameters.

Immunology

The immunological aspects of HIV-SI are inherently related to HIV vaccinology,⁵⁹ and the initial studies that described HIV-SI all highlighted the significance of their finding on this field.^7–9 These studies, and others that have observed HIV-SI in a variety of populations, provide a sobering fact for HIV vaccine design, i.e. that natural HIV infection and the host’s subsequent immune responses are not fully protective against a new HIV challenge. However, these findings also provide investigators with unique populations and novel research paths to identify 1) which components of the natural HIV immune response may be protective against HIV-SI, 2) which components are NOT protective, and 3) what happens to the immune response after a second successful HIV viral challenge.

Initially, Altfeld et al. demonstrated in a case report that HIV-SI occurred even in the presence of a broad HIV-specific cytotoxic T-lymphocyte (CTL) response.⁸ This lack of a protective effect for CTL responses has since been confirmed in several studies.^{61, 66,67} Neutralizing antibodies (NAb) have long been thought to be an essential component of any successful protective HIV vaccine. Examination of NAb response prior to HIV-SI in case-control studies has produced mixed results (Figure 4). Smith et al. observed that individuals prior to an HIV-SI event appeared to lacka NAb response.⁶⁸ This was supported by another small study in Zambian heterosexual couples that found superinfected individuals had a delayed NAb response to their autologous virus pre-SI.⁶⁹ However, a larger case-control study of Kenyan female bar workers found no significant difference in NAb strength prior to HIV-SI.⁴⁸ This same group also reported that antibody-dependent cell mediated viral inhibition was not associated with protection from HIV-SI.⁷⁰ Although the protective effect of NAb towards HIV-SI remains unclear, a number of studies have reported that HIV-SI boosts the NAb response post-superinfection.^{48, 68, 69} The extent of this boost varies between cases, but appears to increase the potency of the NAb response, as measured by the plasma titer levels needed for neutralization, as well as the breadth.⁴⁸ However, it is unknown if this level of Nab is protective of a third phylogenetically distinct strain of HIV.

Since HIV infection damages the host immune system, aspects of the anti-HIV immune response that are not found to be fully protective against HIV-SI may still protect against initial infection if recreated by a vaccine in the context of a healthy immune system. Therefore, the underlying immunological health of subjects examined for protection against HIV-SI should be considered when analyzing these studies.

One additional difficulty with studying the effects of HIV-SI on the host immune response is that cases are relatively difficult to identify in large enough numbers to make it possible to perform in-depth immunological analyses. In addition, longitudinal samples of a wide enough assortment of specimen types (PBMC, mucosal excretions, serum, etc) to fully examine the host immune system are extremely rare. Therefore, for our understanding of the relationship of HIV-SI and host immunity to continue to expand, it will be essential that physicians and researchers examining this topic work in concert using equivalent laboratory and clinical practices so that studies can be easily compared and combined.

HIV-SI and implications for clinical care

HIV-SI has implications for the clinical care of HIV-infected individuals. The long standing practice of serosorting involves HIV-infected patient partnering with other HIV infected individuals. However, transmission of HIV-SI is greater among HIV-infected individuals who do not incorporate safe sexual practices and HIV-SI can potentially lead to increased viral load and disease progression.^{23, 27, 35}

Thus, it is important to encourage safe sexual and injection practices, regardless of HIV infection status. This includes counseling HIV-infected patients on the risk of HIV-SI and encouraging monogamous relationships, condom use, and use of clean needles. In a study on men’s attitudes of HIV-SI among men who had heard of HIV-SI, 82% (135/165) believed that potential HIV-SI could be damaging to their health and 74% (122/165) practiced safer sexual practices due to concern of HIV-SI.⁷¹ The most important aspect of men in this study for improving safer sexual practices was learning about the negative health consequences of HIVSI, and if counseling is performed correctly, it could have a substantial positive impact on reducing the risk of HIV-SI. Therefore, we believe that clinicians and other healthcare providers should counsel and give HIV-infected individuals information regarding the possible detrimental effects of HIV-SI as a component of their continuing care.

ARV therapy is highly effective at perturbing and in many cases reversing the detrimental clinical effects of HIV infection. With the increased use of ARV around the world, one of the biggest concerns of HIV-SI is the transmission of anti-retroviral drug resistant strains (ARV-resistant) or ARV-susceptible strains masking HIV-resistant strains. There have been reported cases of individuals with ARV-resistant strains acquiring an ARV-susceptible strain¹² and also individuals with an HIV-susceptible strain acquiring an HIV-resistant virus.^{39, 72} Due to this negative impact on treatment, clinicians should be aware of the risk of HIV-SI and examine HIV-positive individuals who present with a substantial spike in their viral load or drop in CD4 count for emergence of a new ART-resistance strain. Standard HIV resistance testing soon after these clinical parameters will likely detect the resistance profile of this secondary strain given that the strain is most likely rapidly replicating. However, the advent of ultra-deep sequencing technologies which are becoming cheaper and easier may become more clinically available in the near future for the routine detection of HIV-SI and acquired ARV resistance with the HIV-SI event.

HIV transmission is directly related to HIV viral load and transmission is rare among individuals with levels of less than 1500 copies of HIV-1 RNA per milliliter.⁷³ Recent randomized controlled trial data also confirmed that ARVs substantially reduce HIV transmission.⁵³ There are no randomized trial data on the impact of ARVs to reduce HIV-SI. However, the majority of HIV-SI cases have occurred prior to ARV initiation or during treatment interruption. In one study of 14 high-risk HIV-seroconcordant couples (28 individuals total) that were all treated with ARV, there were no documented cases of HIV-SI.⁴⁷ With the advent of increased ARV use at earlier time points, this will also hopefully reduce the incidence of HIV-SI, but further research is needed in this area.

Future research questions and important issues

The observation and subsequent description of HIV-SI can be viewed as both a set-back and an advancement of HIV research depending on the prism through which one views the field. The magnitude of the impact that HIV-SI will have on future HIV research is largely unknown, and will rely on the ability of investigators to more fully answer some important fundamental biological questions surrounding this field.

1. What is the impact of HIV-SI on disease progression, and what factors effect this impact (ie subtype of superinfecting strain, timing of HIV-SI, magnitude of VL change, etc)?

2. Can a robust NAb response protect individuals from HIV-SI, and what level of response is needed for this protection?

3. What aspects of the immune response change after HIV-SI, and does this prevent a possible subsequent HIV-SI event?

4. Are certain individuals more susceptible to or protected from HIV-SI (e.g., circumcised vs. uncircumcised men, HLA-B57+ individuals, etc.)?

5. What are the risk factors for HIV-SI, and do these differ from primary HIV-infection?

6. Does HIV-SI increase the likelihood of transmitting to an uninfected partner, and what is the role of HIV-SI in transmission chains?

7. Does ART decrease HIV-SI and any subsequent transmission events?

One thing that is becoming clear is that due to the difficulty of identifying HIV-SI cases and the labor and cost associated with screening large populations for these events it will be essential for researchers and clinicians to work in concert to address these and other interesting scientific questions. Therefore, when researchers utilize new and more technologically advanced diagnostic and experimental assays to examine HIV-SI, it is important that the strengths and limitations of every technique be clearly stated and that the population being sampled be carefully described.

Conclusions

HIV-SI has a potentially deleterious pathogenic effect on infected individuals. Initial studies have suggested that by simply informing patients of the potential effects of HIV-SI, clinicians may possibly lower risky behaviors. It is therefore important that information on the risks and prevention of HIV-SI should be included as part of any comprehensive counseling strategy. Over the past decade HIV vaccine research has focused on better understanding HIV-infected individuals who naturally control their initial infection, and have attempted to replicate this state in uninfected individuals with the hopes that it would provide some level of protection against future HIV challenge. The significant levels of HIV-SI seen around the world suggest that this will most likely not succeed, and vaccine strategies attempting this have been largely unsuccessful. Conversely, the study of HIV-SI allows for the examination of the aspects of the naturally occurring immune response which may, or may not, be important to protect against subsequent viral challenge in a natural environment. These insights will allow for a more targeted HIV vaccine research agenda by suggesting promising targets for study, and possibly even more importantly ruling out aspects of the immune response that do not seem to be protective.

Search strategy and selection criteria

We searched PubMed for papers that were published in English between January 1, 1980 and December 1, 2012. The keyword search terms utilized included "HIV superinfection"[All Fields] OR "HIV dual infection" [All Fields] OR (("hiv"[MeSH Terms] OR "hiv"[All Fields]) AND double[All Fields] AND ("infection"[MeSH Terms] OR "infection"[All Fields] OR "communicable diseases" [MeSH Terms] OR ("communicable" [All Fields] AND "diseases" [All Fields]) OR "communicable diseases" [All Fields])) OR "HIV-1 superinfection" [All Fields] OR "HIV-1 dual infection" [All Fields] OR "HIV-1 double infection" [All Fields] OR “HIV reinfection”[All Fields] OR “HIV-1 re-infection”[All Fields]. Articles of relevance from reference lists were also incorporated. Only studies of original research or case reports of distinct HIV infection were included.