HIV Prevention: Rethinking the Risk of Transmission

By Myron S. Cohen, MD*

Great hope has been pinned on biological solutions to HIV prevention including vaccines, topical microbicide agents, male circumcision, and novel use of antiretroviral therapy.

The key to the successful development of these interventions is a complete understanding of the transmission of HIV. The purpose of this communication is to examine new ideas about the sexual transmission of HIV.


The routes for transmission of HIV have been well established (Table 1) (1). Heterosexual vaginal intercourse is of greatest overall importance to the epidemic. However, anal intercourse may also play an important and underestimated role, even in heterosexual couples.




 Female-to-male transmission  1:700 to 1:3,000
 Male-to-female transmission  1:200 to 1:2,000
 Male-to-male transmission  1:10 to 1:1,600
 Fellatio  0 to 6%
 Transfusion of infected blood  95:100
 Needle sharing  1:150
 Needle stick  1:200
 Needle stick/AZT PEP  1:10,000
 Without AZT treatment  1:4
 With AZT treatment  Less than 1:10
 Adapted from Royce, Sena, Cates and Cohen, NEJM 336, 1072 (1997)


Epidemiological studies designed to determine the transmission of HIV per risk Mother-to-child and needle-stick transmission have been defined with great accuracy (Table 1), because in these situations the status of the index case is known and the precise event(s) responsible for HIV acquisition risk can be defined.

Studies of the sexual transmission of HIV have included the use of three methods: cross-sectional collection of data with modeling, prospective longitudinal cohorts of high-risk subjects, and cohorts of discordant couples. The results collected are confounded by a substantial number of problems that lead to considerable inaccuracy and great heterogeneity of transmission probability (Table 2). The aggregate of these limitations has led to widely varying estimates of risk and the somewhat misguided dogma that HIV transmission is inefficient.


Complex sexual behaviors with potential concomitant exposure of several different mucosal sites; anal intercourse in heterosexual couples may be fairly common.

Sexual history provided by study subjects limited by their memory. All studies depend on the reports of study subjects about the quantity and quality of sex, signs and symptoms of sexually transmitted diseases, use of medications, etc. Sexual diaries have proven cumbersome and sometimes inaccurate

Lack of knowledge of the HIV status of the sexual partners (except in the case of discordant couples).

Unrecognized or undetected factors that can amplify transmission, especially STDs. Many STDs are asymptomatic yet might still increase the risk of HIV acquisition (2).

Long periods of follow-up between visits of people at risk, confounding accurate interpretation of risk. HIV-uninfected subjects probably suffer only very brief periods of high risk, but if they are studied infrequently a large number of low risk sexual encounters are included for consideration, reducing the calculated probability of HIV transmission

However, HIV transmission risk is greatly amplified during brief, discrete periods of time. Studies of amplified transmission may provide a more realistic view of HIV risk and the biological forces that a vaccine or other interventions must confront.

Can HIV transmission be amplified?

The probability of HIV transmission depends on the infectiousness of the index case and the susceptibility of the exposed subject (2). Considerable progress has been made regarding infectiousness, with far less progress in understanding susceptibility.

Infectiousness is determined by the concentration of HIV during exposure and viral phenotypic factors. The inoculum required for HIV transmission has been estimated from a series of studies. The largest and most compelling work was conducted in the Rakai district of Uganda, where HIV transmission was directly related to the blood HIV concentration (3). Transmission was not observed when the blood viral burden was less than 3,500 copies/ml, whereas about 50% of transmission events were observed in subjects with the greatest blood viral burden.

Neither this nor other transmission studies examined the genital tract viral burden directly, which can differ from that in the blood. Nor is it clear whether HIV is transmitted by cell-free or cell-associated virus. While the cell-free virus in genital secretions has been carefully measured (4), the number of genital tract cells infected has been far more difficult to quantitate. In addition, HIV RNA copies in genital secretions include a large but unknown number of defective viral particles which may not actually be infectious.

Another way to measure the potential effect(s) of the inoculum on HIV transmission is to study examples of amplified and reduced HIV transmission. Reduced HIV transmission has been associated with usage of antiretroviral (ARV) drugs in some (but not all) studies (reviewed in ref. 5). For example, in an early study of discordant couples significant reduction of HIV transmission among men who were taking AZT was reported (6). More recently, several population-based studies have suggested that widespread usage of ARV therapy in a community or country can lead to reduced incidence of HIV (reviewed in ref. 5).

Conditions associated with increased excretion of HIV are of considerable importance for biological prevention strategies. Viral burden in blood is greatest during the first weeks of infection (7), consistent with modeling and epidemiology studies that suggest extremely efficient transmission of HIV from subjects with acute infection (7-9). New analysis of the Rakai data demonstrates that nearly half of the HIV transmission events observed could be ascribed to the earliest time(s) of infection (8). Accordingly, describing HIV transmission risk in terms of intercourse events over extended periods of time is an oversimplification, and HIV transmission should probably be considered in phases, the most important of which may well be acute infection (Figure 1, adapted from reference 9).


Figure 1. Sexual transmission of HIVThe relative risk of transmission over the course of the disease, as a function of viral load in semen. Orange indicates HIV RNA copies per ml semen, yellow indicates the reduced viral burden expected from host defens¬es evoked by a vaccine or early use of antiviral therapy. The dashed line offers a theoretical viral burden threshold below which HIV transmission will not occur. Numbers provided at the different stages of disease represent the probability of HIV transmis¬sion/episode of heterosexual intercourse. Adapted from ref. 9. [Large image]. 

What is the role of other STDs?

Classical STDs likely play a central role in the HIV epidemic (2, 4, 10-12). Mucosal STDs (gonorrhea, Chlamydia, trichomonas) and genital ulcer pathogens (herpes simplex virus [HSV], syphilis, chancroid) affect both infectiousness and susceptibility. STDs clearly facilitate HIV transmission (10) because: i) STDs cause mucosal breaks and increase tissue inflammation and the number of receptive cells; ii) inflammation associated with STDs increase the concentration of HIV in genital secretions ; iii) STDs and HIV are frequently co-transmitted. Several recent studies (13-14) have led to detection of a substantial number of subjects with acute HIV infection in STD clinics, supporting the idea of co-transmission of a classical STD and HIV. In a detailed study of HSV and HIV in India, incident HIV and HSV were closely correlated (14). Since many STDS are rapidly and efficiently transmitted they may set the stage for concomitant acquisition of HIV from virus harbored in the mucosa or during subsequent episodes of intercourse.

What must host defenses confront during exposure to HIV?

The HIV RNA concentration in semen under different conditions has been quite well defined. Subjects in the US and Europe with acute HIV clade B infection can be expected to have 1 - 2 x 105 copies/ml semen (16), whereas subjects in Africa with HIV clade C infection appear to have substantially more HIV in this secretion during acute infection (unpublished observations from Malawi). The copy number of HIV in the female genital tract is not easily defined, in part because the sample collection procedures may be traumatic and often involve dilution with collection buffer. Menstrual cycle stage and oral contraceptive methods also influence the results.

Male patients with established HIV who acquire an inflammatory STD may excrete >1 x 105  copies HIV/ml, or 6-8 times more than at baseline (12), and substantial increases of HIV in female genital secretions have been observed in women with STDs (2, 4). HIV excretion stimulated by inflammation can on occasion overwhelm the suppressive effects of antiviral therapy (17).

Does viral phenotype of genital tract isolates affect HIV transmission?

Only a small number of investigators have compared the genotype or phenotype of HIV in secretions harvested concomitantly from genital secretions and blood (4, 18). The predominance of CCR5¬tropic (R5) HIV in primary infection suggests either a selective advantage of R5 HIV in transmission, selective viral expansion after infection, or escape from host defenses. In general, HIV recovered from blood and genital secretions demonstrates homology, whether detected by heteroduplex mobility assay, heteroduplex tracking assay, envelope sequence analysis or aniviral resistance markers. However, some differences surface in the face of inflammation (18). There are few data related to viral factors that might affect transmission efficiency, but one recent report suggests that HIV strains with unique envelope properties that confer increased susceptibility to neutralizing antibodies may be preferentially transmitted (19). In addition, uneven distribution of resistance mutations in untreated patients with early HIV infection suggests the possibility that viral fitness could affect transmission efficiency (20).

Men and women may differ in the number of unique variants recovered after HIV transmission. It appears that heterosexual men acquire a single variant, whereas heterosexual women acquire multiple variants (21, 22). Gay men may also have one or many variants, suggesting that a large surface area (i.e., the vagina and the rectum) may permit more access for simultaneous discrete transmission events (22).

What about susceptibility?

Nearly all studies related to HIV transmission have focused on the individual who has become HIV infected, and factor(s) that allowed or promoted infection (10, 11). These factors include (but are not limited to) host sexual behaviors, host anatomy (e.g., circumcision), host genetics, genital tract flora including acute or chronic STD infections, ill-defined innate or acquired immune defenses, cytokines, and drugs that affect mucosal biology (e.g., oral contraceptives). However, determining the relative importance of each of these factors has been virtually impossible. Unfortunately, animal models are not sufficiently similar to humans to provide definitive conclusions. Rather, clinical trials in humans will likely prove most instructive. For example, if an ongoing clinical trial of cervical barriers supported by the Gates foundation demonstrates that these devices prevent HIV infection, the central role of the endocervix will be clarified.


Improved understanding of HIV transmission has resulted from a large number of studies of the epidemiology and biology of HIV transmission. Sexual transmission of HIV is clearly not homogeneous. Successful interventions (whether behavioral or biological) should anticipate and be directed toward the time(s) of greatest transmission risk (Figure 1). Given our current difficulty in recognizing and treating people with acute HIV infection this is a substantial challenge, but one that must be met if we are to control the HIV epidemic (9). In addition, biological interventions should be developed to deal with the real conditions of transmission: high viral burden, the potential for viral diversity, and substantial STD-mediated inflammation.


1. Royce RA, et al., N. Engl. J. Med. 336, 1072 (1997). PubMed  
2. Galvin S and Cohen MS, Nat. Rev. Microbiol. 2, 33 (2004). PubMed  
3. Quinn TC, et al., N. Engl. J. Med. 342, 921 (2000). PubMed  
4. Coombs RW, et al., AIDS 17, 455 (2003). PubMed  
5. Cohen MS, et al., AIDS in the 21st Century. Mayer, et al., (in press, 2004).  
6. Musicco M, et al., Arch. Intern. Med. 154, 1971 (1994). PubMed  
7. Pilcher CD, et al., J. Infect. Dis. 189, 1785 (2004). PubMed  
8. Wawer MJ, et al., 10th Conference on Retroviruses and Opportunistic Infections, Boston, 40 (2003).  
9. Pilcher CD, et al., J. Clin. Invest. 113, 937 (2004). PubMed  
10. Fleming DT, et al., Sex. Transm. Infect. 75, 3 (1999). PubMed  
11. Rotingen J-A, et al., Sex. Trans. Dis. 28, 579 (2001). PubMed  
12. Cohen MS, et al., Lancet 349, 1868 (1997). PubMed  
13. Bollinger RC, et al., JAMA 278, 2085 (1997). PubMed  
14. Reynolds SJ, et al., J. Infect. Dis. 187, 1513 (2003). PubMed  
15. Pilcher CD, et al., AIDS 18, 517 (2004). PubMed  
16. Pilcher CD, et al., AIDS 15, 837 (2001). PubMed  
17. Sadiq ST, et al., AIDS 16, 219 (2002). PubMed  
18. Ping LH, et al., J. Virol. 74, 8946 (2000). PubMed  
19. Derdeyn CA, et al., Science 303, 2019 (2004). PubMed  
20. Turner D, et al., J. Acquir. Immune Defic. Syndr. 15, 1627 (2004). PubMed  
21. Long EM, et al., Nat. Med. 6, 71 (2000). PubMed  
22. Ritola K, et al., J. Virol. 78, 11208 (2004). PubMed 

*Myron S. Cohen, MD is J. Herbert Bate Professor of Medicine, Microbiology and Immunology and Public Health, Chief of the Division of Infectious Diseases, and Director of the Center for Infectious Diseases at The University of North Carolina at Chapel Hill.