Why do HIV-1 subtypes segregate among persons with different risk behaviors in South Africa and Thailand?

In the past few years, a great deal has been learned about the genetic variability of HIV [1–3]; however, we are only just beginning to understand, in the laboratory and among human populations, the functional significance of this remarkable genetic diversity. As epidemiologic investigations seek to characterize phenotypic properties of genotypic HIV-1 subtypes, it is appropriate to establish as a null hypothesis that ‘genotypic HIV-1 subtypes are phenotypically similar’. The challenge to investigators is to reject this null hypothesis by clearly demonstrating phenotypic differences among genotypic HIV-1 subtypes. Within the past year, considerable attention has been focused on the observation that HIV-1 envelope (env) subtype B predominates in industrialized Western countries, where most transmission is among homosexual men and injecting drug users (IDU), whereas other env subtypes (e.g., A, C, D and E) predominate in Africa and Asia, where the great majority of transmission is heterosexual. It has been suggested that this situation may be due to functional differences among HIV-1 env subtypes [4]. Studies implying that env subtype E may be more efficiently transmitted heterosexually than subtype B [5,6] have prompted speculation and alarm that the introduction of non-B subtypes in the West could result in more extensive heterosexual transmission [7] [the Joint United Nations Programme in HIV/AIDS (UNAIDS) organized a meeting in Berlin during March 1996 to consider the ‘Implications of HIV variability for transmission: scientific and policy issues’ for which a report from the expert group participating in this meeting is forthcoming]. In the laboratory, Soto-Ramirez et al. [5,8] reported that HIV-1 env subtype E and C strains from Thai and Indian heterosexuals, respectively, infected Langerhans' cells (found in the epithelium of the vagina, cervix, and penile foreskin) more efficiently than env subtype B strains from US homosexual men. They proposed that Langerhans' cell tropism of subtypes E and C resulted in more efficient vaginal transmission, which helped explain the explosive heterosexual subtype E and C epidemics seen in Thailand and India [9–12]. Conversely, they inferred that subtype B strains, which predominate among homosexual men and IDU in industrialized Western countries, were transmitted less efficiently vaginally, while retaining the potential for efficient transmission via blood, including transmission resulting from microscopic abrasions presumed to occur during receptive anal intercourse [5]. The unique epidemiologic situation in Thailand provided a setting in which to compare heterosexual transmission of env subtypes B and E. Extensive HIV-1 subtype B transmission began in 1988 among Bangkok IDU [13,14]. This was followed in 1989 by a larger epidemic of heterosexually transmitted subtype E, largely related to the common male practice of patronizing female commercial sex workers [15–17]. By 1993, approximately 90% of an estimated 500 000–750 000 cumulative HIV infections were caused by heterosexually transmitted subtype E [18]. In Bangkok, a case-control study of HIV-1-infected Thai men (index cases) and their female sex partners suggested that subtype E was more sexually transmissible than subtype B [6]. Analyses were designed to control for a variety of possible confounding factors, including stage of HIV-1 disease, sexual practices, other sexually transmitted diseases (STD), and condom use; however, 86 (96.6%) out of 89 men with subtype E had acquired HIV-1 sexually, whereas 76 (98.7%) out of 77 men with subtype B were IDU and were probably infected parenterally. As discussed by the authors, it may not have been possible in the analyses to fully control for the confounding effect of this correlation between viral subtype and risk behavior of the male index cases [6]. In this issue, van Harmelen et al. [19], following an earlier communication [20], describe the segregation of HIV-1 env subtypes B and C among homosexual men and heterosexual men and women, respectively, in Cape Town, South Africa. Out of 26 homosexual men, 25 (96.2%) were infected with subtype B, whereas 26 (81.3%) out of 32 persons thought to have been infected via heterosexual contact had subtype C. Subtype D was found in one homosexual and one heterosexual, and subtype E was found in one heterosexual. What contributed to this apparent segregation of HIV-1 subtypes by risk behavior? Limited epidemiologic data were reported on these persons; 21 (65.6%) out of 32 heterosexuals were black, whereas none of the 26 homosexual or bisexual men were black. Several homosexual men reported sexual contact during travel in Western countries, whereas some heterosexual men reported sexual contact in nearby southern African countries. Rather than being a result of virologic factors related to transmission mode, the epidemiologic segregation of subtypes B and C may have resulted from demographic, economic, and social network factors (other than sexual preferences) that separated the group exposed to subtype B from the group exposed to subtype C. Similarly, investigators in The Gambia, West Africa have found subtype B infection to be strongly associated with history of travel to North America or Europe; such travel was infrequent among men and women infected with non-B subtypes [21]. The data from Thailand and South Africa indicate that, due to limited mixing of population subgroups, largely separate HIV-1 epidemics caused by different genotypic subtypes may co-exist in a given geographic area. Such ecologic separation of HIV-1 subtypes would be even more likely for geographically diverse regions. In the global epidemic, the relative importance of ecologic as opposed to virologic factors, and the extent of confounding of the two, is unclear. Whether HIV-1 subtypes differ in transmission efficiency by exposure mode and whether this is the cause of segregation of viral subtypes among persons with different risk behaviors remains to be demonstrated conclusively. Although it has been proposed that subtypes E and C are better adapted to penile-vaginal transmission than subtype B, subtypes E and C, in addition to subtype B, appear to be transmitted efficiently via blood. Recent studies of IDU in Thailand have documented an increase in the prevalence and incidence of subtype E, especially among younger IDU [22–24], and limited available data suggest that most HIV-1 subtype E transmission among young injecting heroin users in Bangkok is due to needle-sharing rather than sexual behavior [23]. In Vietnam, subtype E has been identified as the predominant subtype among both IDU and persons with heterosexual risk behavior [25,26]. Subtype C has been identified in IDU in Yunnan, China [27], perhaps related to the epidemic of subtype C in nearby India, which, in turn, has been attributed to connections between India and South Africa [10,11]. Reports that HIV-1 subtypes other that B have been identified in the United States have received considerable attention [28–31], in part because of concerns that these subtypes may pose increased heterosexual transmission risk. However, recent reports from Belgium and the United Kingdom indicate that HIV-1 subtypes other that B have been present in European populations for a number of years without evidence of extensive heterosexual transmission [32,33]. Furthermore, in studies of heterosexual couples, very similar per-sex-act HIV-1 transmission probabilities have been estimated for subtype B in Europe and North America (0.001) as for subtype E in northern Thailand (0.002) [34,35]. Numerous other factors, especially sexual mixing patterns (including commercial sex) and the prevalence of other STD, contribute greatly to epidemics of sexual HIV-1 transmission, and must be accounted for when examining HIV-1 subtype transmission differences [35,36]. As we consider whether various HIV-1 subtypes have different transmission efficiencies, we should recall how phenotypic differences between HIV-1 and HIV-2 were detected [37]. Careful epidemiologic investigations determined that in areas of West Africa where the two viruses existed, HIV-1 prevalence increased while HIV-2 prevalence remained stable or even decreased [38], and heterosexual spread of HIV-2 was measured to be slower than HIV-1 [39], suggesting type-specific differences in viral infectivity. Infectivity differences may be inherent to the virus or related to differences in viral load; lower quantitative HIV-2 viral levels at any or all stages of infection may result in decreased transmission efficiency [38]. New assays should allow for the accurate measurement of viral levels in plasma, semen, and genital secretions among persons infected with various HIV-1 subtypes. Use of these assays in well-designed studies will complement epidemiologic estimates of transmission efficiency for various HIV-1 subtypes. Epidemiologic settings such as Thailand and South Africa may offer opportunities for comparisons of different transmission efficiencies between two subtypes (B and E, or B and C) that have limited confounding. In Bangkok, where subtypes B and E are found among IDU (95% are men), it may be possible to compare sexual transmission efficiency to female partners for these two subtypes without such confounding factors as mode or stage of infection of the index case, STD rates, or differences in the frequency of sexual exposure. Unlike the emergence of HIV-1 variants that can not be readily detected by current HIV antibody tests, which therefore pose a potential public health risk [40], possible phenotypic transmission differences of HIV-1 subtypes should have little impact on current prevention measures aimed at HIV transmission [7]. However, it is essential that well-designed laboratory and epidemiologic investigations are conducted to give us further insight into the transmission of this pathogen for the development of improved prevention modalities including vaccines that are effective for all strains of HIV. Acknowledgement We would like to thank D. Hu and B. Edlin for valuable discussions and helpful suggestions.