Epidemiology and social: CONCISE COMMUNICATION
Heterogeneity in host HIV susceptibility as a potential contributor to recent HIV prevalence declines in Africa
Nagelkerke, Nico JDa,b; de Vlas, Sake Jc; Jha, Prabhatd; Luo, Mab,e; Plummer, Francis Ab,e; Kaul, Rupertf,g
aDepartment of Community Medicine, United Arab Emirates University, Al Ain, United Arab Emirates
bDepartment of Medical Microbiology, University of Manitoba, Winnipeg, Canada
cDepartment of Public Health, Erasmus MC, University Medical Center, Rotterdam, the Netherlands
dCenter for Global Health Research, University of Toronto, Toronto, Canada
eNational Microbiology Laboratory, Winnipeg, Canada
fDepartment of Medicine, University Health Network and the University of Toronto, Toronto, Canada
gDepartment of Medical Microbiology, University of Nairobi, Nairobi, Kenya.
Received 17 May, 2008
Revised 4 August, 2008
Accepted 13 August, 2008
Correspondence to Rupert Kaul, MD, PhD, Clinical Sciences Division, #6356 Medical Sciences Building, 1 King's College Circle, Toronto, Ontario M5S1A8, Canada. Tel: +1 416 978 8607; fax: +1 416 978 8765; e-mail: firstname.lastname@example.org
Background: HIV prevalence has recently declined in several African countries, and prior to this the risk of HIV acquisition per unprotected sex contact also declined in Kenyan sex workers. We hypothesized that heterogeneity in HIV host susceptibility might underpin both of these observations.
Methods: A compartmental mathematical model was used to explore the potential impact of heterogeneity in susceptibility to HIV infection on epidemic behavior, in the absence of other causative mechanisms.
Results: Studies indicated that a substantial heterogeneity in susceptibility to HIV infection may lead to an epidemic that peaks and then declines due to a depletion of the most susceptible individuals, even without changes in sexual behavior. This effect was most notable in high-risk groups such as female sex workers and was consistent with empirical data.
Discussion: Declines in HIV prevalence may have other causes in addition to behavior change, including heterogeneity in host HIV susceptibility. There is a need to further study this heterogeneity and its correlates, particularly as it confounds the ability to attribute HIV epidemic shifts to specific interventions, including behavior change.
HIV prevalence has recently fallen in some severely affected countries, including Uganda and Kenya [1,2], but the cause(s) of these declines is unclear. In Uganda, a causal relationship has been inferred between prevalence declines and timely government-initiated programs with documented impacts on risk behavior [2–6]. However, such programs were less prominent in Kenya, leading some to attribute the decline in both countries to an increased, nonexponential AIDS mortality associated with ‘maturing’ epidemics [6–8].
Recent data from a cohort of female sex workers (FSWs) in Nairobi, Kenya suggest that neither explanation is fully satisfactory . Here, the risk of HIV acquisition per act of unprotected sex declined by more than four-fold, over a decade before prevalence declines were apparent in the general population, an observation that cannot be explained by behavior change or nonexponential mortality. We hypothesized that this decline might relate to heterogeneity in HIV host susceptibility, which would give rise to a high HIV incidence among the most susceptible individuals during early stages of an epidemic, followed by a decrease in the susceptibility (and rate of HIV infection) in the remaining uninfected population. Such heterogeneity in HIV susceptibility has been demonstrated in several contexts, including the profound reduction in susceptibility associated with the CCR5Δ32 genetic polymorphism , as well as increased susceptibility associated with low copy numbers of the CCL3L1 gene duplication  and with persistent (life-long) infections such as herpes simplex type 2 .
We explored this hypothesis using a simple compartmental mathematical model and we also examined whether early depletion of susceptible individuals alone, in the absence of other proven causal factors, could partly account for the recent general population prevalence declines in Kenya and Uganda.
We modified an existing compartmental model [13,14] to explore the effect of heterogeneity in host susceptibility on the dynamic behavior of an HIV epidemic. Only heterosexual transmission was modeled. The model divided the population into groups of low-risk persons and high-risk persons (male clients and FSWs) with movement between high and low risk categories; these groups were further subdivided into compartments by stage of infection: HIV negative; early stage HIV; and late stage HIV (Fig. 1, Supplementary Table 1). To this model, we added heterogeneity in biological susceptibility, by arbitrarily distinguishing four levels of susceptibility comprising 10, 30, 30, and 30% of the population, respectively, with the lowest (10% of the population) assumed to be completely resistant. Risk levels were life-long, with no transition between susceptibility levels. Disease progression after infection was modeled as a short (mean 6 months) initial phase, followed by a second phase of an average 8 years. This second phase was modeled as exponential, and infectiousness in the early phase of HIV infection was set equal to the level of infectiousness during the second phase. Although both these choices were in all likelihood counterfactual, they were intentionally made in order to explore whether heterogeneity in susceptibility alone, in the absence of other causative mechanisms, could cause epidemic overshoot (the situation in which HIV incidence and prevalence during the early phase of the HIV epidemic exceed subsequent levels). In reality other factors, including differential infectiousness, epidemic phase, sexually transmitted infection rates, behavior, and others, will have also played an important role in shaping the epidemic, and hence our empirical datasets. Fractions of high-risk groups (FSWs and their clients) were set at equilibrium levels in the absence of HIV. Two response variables were selected: the per client contact risk of HIV infection in FSWs and the sex-specific adult HIV prevalence. The fraction of each susceptibility level, as well as the susceptibility levels themselves, was estimated by minimizing the sum of squared differences between the empirical risk of infection in the Nairobi sex worker cohort and the model predicted risk in this cohort (Supplementary Table 1). ModelMaker (version 3.0.3; Cherwell Scientific Publishing Limited, Oxford, UK) was used to implement and run the model, with HIV introduced into the population in 1980. For the sex worker model, epidemic curves obtained with the modified compartmental model were compared with per contact HIV incidence data that had been gathered empirically within a high-risk cohort of Kenyan FSWs, as previously described . For the adult HIV prevalence model, the epidemic curves obtained were compared with longitudinal HIV prevalence previously reported from urban antenatal sentinel surveillance sites in Kenya [15,16].
Determination of HIV incidence in a high-risk cohort
Empirical data regarding per contact HIV incidence in a high-risk cohort were obtained as previously described . Briefly, FSWs were enrolled through a dedicated clinic in Nairobi, Kenya from 1985–2005. All participants returned for a formal resurvey every 6 months. At enrolment and resurvey, a standardized questionnaire was completed and blood was drawn for HIV serology. Self-reported risk taking data included the number of casual clients per day, condom use with casual clients, the number of regular clients (‘boyfriends’), condom use with regular clients, and sexual practices (anal sex, sex during menses). HIV-1 serology was performed at each visit using a synthetic peptide enzyme immunoassay (EIA; Detect HIV, Biochem ImmunoSystems Inc., Montreal, Canada) and positive tests were confirmed by recombinant antigen EIA (Recombigen HIV-1/2 EIA; Cambridge Biotech Corporation, Galway, Ireland). Trends in the risk of HIV acquisition per unprotected sex act were analyzed prospectively within initially HIV-uninfected participants. The number of unprotected sex contacts per day was calculated as the product of the reported number of casual sex partners per day and the fraction sex acts unprotected by condoms .
The model that best fitted the observed empirical infection risk per 1000 client contacts within the Nairobi FSW cohort  estimated the susceptibilities in the second lowest susceptibility fractions at (near) 0. Thus, a total fraction of 40% of the population was at a very low risk of HIV infection. In reality, it is likely that a much smaller fraction of the population is relatively resistant to HIV infection, but this may relate to our use of four discrete susceptibility levels in the model, as opposed to a continuous distribution. The two highest susceptibility levels were estimated at 0.0257 and 0.0347 (male-to-female) and half that value for female-to-male per sexual-contact risk. Using these parameters, there was an approximately four-fold decline in the modeled risk of per contact HIV acquisition that followed peak infection rates. This closely mirrored empirical data collected from the Nairobi sex worker cohort (Fig. 2a).
In the general population, an initial prevalence overshoot was followed by a decline in both men and women (Fig. 2b). Again, this mirrored, although somewhat less so, empirical observations; in this case, longitudinal antenatal adult HIV prevalence surveys from Kenya [15,16]. It is possible that prevalence among (young) antenatal women more accurately reflects (changes in) incidence rather than (changes in) prevalence in the general population . Infectiousness in the early epidemic phase of HIV infection was defined in the model as equal to the level of infectiousness during the second phase. Defining infectiousness during early HIV infection as substantially higher than that during the second phase, which may be more realistic, yielded an overshoot pattern that was broadly similar, but that was more peaked and overly compressed in time. Defining late stage HIV as shorter and more infectious than early stage infection widened the overshoot peak somewhat and more closely approximated empirical observations.
Heterogeneity in susceptibility and/or transmissibility was recognized during the early phases of the HIV epidemic, when it was found that transmission in heterosexual partnerships was not solely dependent on the number of sex acts [18–20]. Strikingly, the risk of male-to-female HIV transmission in a discordant heterosexual partnership was as high as 10% for less than 10 unprotected contacts, but increased to only 23% after at least 2000 unprotected contacts . Although these studies could not differentiate between heterogeneity in infectiousness and in host susceptibility, it is plausible that the effect was at least partly due to the latter, given that several genetic, immune, and infectious correlates of altered susceptibility have since been identified [10–12,21–23]. Further evidence for heterogeneity in host susceptibility is provided by the observation that some individuals appear to be relatively resistant . However, it is difficult to measure this heterogeneity precisely, as individuals can only acquire HIV once, and therefore, estimation of individual risks becomes extremely difficult. In theory, heterogeneity in host susceptibility could be estimated from the distribution of time to infection in cohorts of equally exposed HIV-uninfected individuals, but such cohorts are not available. High-risk sex worker cohorts provide limited insights, though valuable ones , as most of these women had already been heavily exposed to HIV prior to enrolment in the cohort, presumably with infection of the most susceptible women. These difficulties in accurately measuring heterogeneity in host susceptibility may explain why the phenomenon has largely been overlooked.
We hypothesize that the decline in per act acquisition of HIV observed over time in high-risk Kenyan sex workers , independent of behavior change, may have been partly related to heterogeneity in susceptibility within the cohort. This decline occurred during a period when HIV prevalence in high-risk men was stable, making it unlikely that it was related to changing HIV prevalence in male clients, and was prior to significant roll out of antiretroviral therapy in the region . Altered patterns of sex worker client contact, particularly the establishment of ‘regular clients’, may impact sexually transmitted infection (STI) and HIV prevalence , but this effect will be blunted by the fact that regular clients of sex workers in the cohort report high numbers of concurrent partners (Kaul R, unpublished data). Additional factors are likely to have contributed to this decline, including a reduction in the prevalence of bacterial (transient) sexually transmitted infections, phase of the HIV epidemic, and other factors not included in our model. Indeed, such factors may partly explain why the reductions in population prevalence in our model were more gradual than the empirical data gathered from Kenya antenatal clinics (as demonstrated in Fig. 2b).
Nonetheless, heterogeneity in susceptibility may have a significant impact on our ability to model future epidemic spread, as well as to interpret surveillance data. As an example, a high number of transmission events are attributed to individuals with very early HIV infection, so that this phase of infection may contribute disproportionately to HIV transmission [26–28]. This phenomenon may relate to high HIV levels in genital secretions during this stage of infection , but heterogeneity in the susceptibility of their uninfected partners may be another contributor, as many of these uninfected partners will be exposed to HIV for the first time by their acutely infected partner. Available data cannot distinguish these two mechanisms. A study comparing the risk of seroconversion of steady partners of new seroconverters with that of new partners of people with established HIV infection should be able to clarify this.
Heterogeneity in host HIV susceptibility might also have more practical implications for prevention trials. HIV incidence within high-risk participants has been lower than expected in the context of trials testing STI prophylaxis/therapy , microbicides , and preexposure prophylaxis , often despite prior studies to determine ‘predicted’ incidence, and this has been a significant setback in some cases. Although it has been assumed that the reduced HIV incidence was a direct result of behavior modification and trial-associated prevention activities , it is also possible that early HIV acquisition by more susceptible hosts may have been followed by incidence declines within the remaining, less susceptible participants.
Although there is good evidence that heterogeneity in susceptibility exists, we have little information on how it is distributed in the population. In our model, we chose four discrete levels of susceptibility, though in reality a continuous distribution seems more realistic. In fact, four was the highest number of levels that we could fit in view of the dearth of data, though three levels yielded similar patterns (data not shown). This simple model demonstrates, at least qualitatively, that heterogeneity in HIV susceptibility may lead to epidemic overshoot. This overshoot is most pronounced in highly exposed populations, such as FSWs.
In summary, we propose that the phenomenon of heterogeneity in HIV susceptibility may have contributed to the observed declines in HIV incidence and prevalence, both in highly exposed ‘core transmitter’ cohorts such as FSWs, and in endemic countries. Heterogeneity in susceptibility is by no means the only contributor to these declines, and there is compelling evidence that at least part of the population HIV prevalence declines in many parts of Africa was caused by changes in risk behavior [2,3,33–36]. However, our model demonstrates that some of this decline may have occurred without behavioral change, confounding our ability to attribute HIV epidemic shifts to specific interventions.
Funding: Canadian Institutes of Health Research (R.K.; HOP-81735, HET-85518); the Bill and Melinda Gates Foundation (F.A.P.; BMG-77515) and CIHR Grand Challenges in Global Health (F.A.P.; #37873); National Institutes of Health (F.A.P.; R01 AI056980); Canada Research Chair Programme (salary support for RK, FAP).
Study concept: N.J.D.N., S.J.V., P.J., M.L., F.A.P., R.K.
Study design: N.J.D.N., S.J.V., R.K.
Analysis design and performance: N.J.D.N., S.J.V.
Initial manuscript: N.J.D.N., R.K.
Manuscript revisions: S.J.V., P.J., M.L., F.A.P.
There are no conflicts of interest.
1. Cheluget B, Baltazar G, Orege P, Ibrahim M, Marum LH, Stover J. Evidence for population level declines in adult HIV prevalence in Kenya. Sex Transm Infect 2006; 82(Suppl 1):i21–i26.
2. Stoneburner RL, Low-Beer D. Population-level HIV declines and behavioral risk avoidance in Uganda. Science 2004; 304:714–718.
3. Kilian AH, Gregson S, Ndyanabangi B, Walusaga K, Kipp W, Sahlmuller G, et al
. Reductions in risk behaviour provide the most consistent explanation for declining HIV-1 prevalence in Uganda. AIDS 1999; 13:391–398.
4. Kilian AH, Kipp W, Jhangri GS, Saunders LD, Ndyanabangi B, O'Connor H, et al
. Trends in HIV infection: prevention-related attitudes and behaviors among secondary school students in western Uganda. J Acquir Immune Defic Syndr 2007; 44:586–593.
5. Slutkin G, Okware S, Naamara W, Sutherland D, Flanagan D, Carael M, et al
. How Uganda reversed its HIV epidemic. AIDS Behav 2006; 10:351–360.
6. Wawer MJ, Serwadda D, Gray RH, Sewankambo NK, Li C, Nalugoda F, et al
. Trends in HIV-1 prevalence may not reflect trends in incidence in mature epidemics: data from the Rakai population-based cohort, Uganda. AIDS 1997; 11:1023–1030.
7. James JS. Uganda study found that death reduced HIV prevalence; did the public take home the wrong message? AIDS Treat News
8. Singh S, Darroch JE, Bankole A. A, B and C in Uganda: the roles of abstinence, monogamy and condom use in HIV decline. Reprod Health Matters 2004; 12:129–131.
9. Kimani J, Kaul R, Nagelkerke NJ, Luo M, MacDonald KS, Ngugi E, et al
. Reduced rates of HIV acquisition during unprotected sex by Kenyan female sex workers predating population declines in HIV prevalence. AIDS 2008; 22:131–137.
10. Samson M, Libert F, Doranz BJ, Rucker J, Liesnard C, Farber CM, et al
. Resistance to HIV-1 infection in caucasian individuals bearing mutant alleles of the CCR-5 chemokine receptor gene. Nature 1996; 382:722–725.
11. Gonzalez E, Kulkarni H, Bolivar H, Mangano A, Sanchez R, Catano G, et al
. The influence of CCL3L1 gene-containing segmental duplications on HIV-1/AIDS susceptibility. Science 2005; 307:1434–1440.
12. Freeman EE, Weiss HA, Glynn JR, Cross PL, Whitworth JA, Hayes RJ. Herpes simplex virus 2 infection increases HIV acquisition in men and women: systematic review and meta-analysis of longitudinal studies. AIDS 2006; 20:73–83.
13. Nagelkerke NJ, Jha P, de Vlas SJ, Korenromp EL, Moses S, Blanchard JF, Plummer FA. Modelling HIV/AIDS epidemics in Botswana and India: impact of interventions to prevent transmission. Bull World Health Organ 2002; 80:89–96.
14. Nagelkerke NJ, Moses S, de Vlas SJ, Bailey RC. Modelling the public health impact of male circumcision for HIV prevention in high prevalence areas in Africa. BMC Infect Dis 2007; 7:16.
17. Ghys PD, Kufa E, George MV. Measuring trends in prevalence and incidence of HIV infection in countries with generalised epidemics. Sex Transm Infect 2006; 82(Suppl 1):i52–i56.
18. Downs AM, De Vincenzi I. Probability of heterosexual transmission of HIV: relationship to the number of unprotected sexual contacts. European Study Group in Heterosexual Transmission of HIV. J Acquir Immune Defic Syndr Hum Retrovirol 1996; 11:388–395.
19. Kaplan EH. Modeling HIV infectivity: must sex acts be counted? J Acquir Immune Defic Syndr 1990; 3:55–61.
20. Peterman TA, Stoneburner RL, Allen JR, Jaffe HW, Curran JW. Risk of human immunodeficiency virus transmission from heterosexual adults with transfusion-associated infections. JAMA 1988; 259:55–58.
21. Ball TB, Ji H, Kimani J, McLaren P, Marlin C, Hill AV, Plummer FA. Polymorphisms in IRF-1 associated with resistance to HIV-1 infection in highly exposed uninfected Kenyan sex workers. AIDS 2007; 21:1091–1101.
22. Iqbal SM, Kaul R. Mucosal innate immunity as a determinant of HIV susceptibility. Am J Reprod Immunol 2008; 59:44–54.
23. Kaul R, Rowland-Jones SL, Kimani J, Fowke K, Dong T, Kiama P, et al
. New insights into HIV-1 specific cytotoxic T-lymphocyte responses in exposed, persistently seronegative Kenyan sex workers. Immunol Lett 2001; 79:3–13.
24. Fowke KR, Nagelkerke NJ, Kimani J, Simonsen JN, Anzala AO, Bwayo JJ, et al
. Resistance to HIV-1 infection among persistently seronegative prostitutes in Nairobi, Kenya. Lancet 1996; 348:1347–1351.
25. Ghani AC, Aral SO. Patterns of sex worker-client contacts and their implications for the persistence of sexually transmitted infections. J Infect Dis 2005; 191(Suppl 1):S34–S41.
26. Abu-Raddad LJ, Longini IM Jr. No HIV stage is dominant in driving the HIV epidemic in sub-Saharan Africa. AIDS 2008; 22:1055–1061.
27. Brenner BG, Roger M, Routy JP, Moisi D, Ntemgwa M, Matte C, et al
. High rates of forward transmission events after acute/early HIV-1 infection. J Infect Dis 2007; 195:951–959.
28. Lewis F, Hughes GJ, Rambaut A, Pozniak A, Leigh Brown AJ. Episodic sexual transmission of HIV revealed by molecular phylodynamics. PLoS Med 2008; 5:e50.
29. Pilcher CD, Joaki G, Hoffman IF, Martinson FE, Mapanje C, Stewart PW, et al
. Amplified transmission of HIV-1: comparison of HIV-1 concentrations in semen and blood during acute and chronic infection. AIDS 2007; 21:1723–1730.
30. Kaul R, Kimani J, Nagelkerke NJ, Fonck K, Keli F, MacDonald KS, et al
. Reduced HIV risk-taking and low HIV incidence after enrollment and risk-reduction counseling in a sexually transmitted disease prevention trial in Nairobi, Kenya. J Acquir Immune Defic Syndr 2002; 30:69–72.
31. van de Wijgert JH, Shattock RJ. Vaginal microbicides: moving ahead after an unexpected setback. AIDS 2007; 21:2369–2376.
32. Peterson L, Taylor D, Roddy R, Belai G, Phillips P, Nanda K, et al
. Tenofovir disoproxil fumarate for prevention of HIV infection in women: a phase 2, double-blind, randomized, placebo-controlled trial. PLoS Clin Trials 2007; 2:e27.
33. Green EC, Halperin DT, Nantulya V, Hogle JA. Uganda's HIV prevention success: the role of sexual behavior change and the national response. AIDS Behav 2006; 10:335–346, discussion 347–350.
34. Gregson S, Garnett GP, Nyamukapa CA, Hallett TB, Lewis JJ, Mason PR, et al
. HIV decline associated with behavior change in eastern Zimbabwe. Science 2006; 311:664–666.
35. Hallett TB, Aberle-Grasse J, Bello G, Boulos LM, Cayemittes MP, Cheluget B, et al
. Declines in HIV prevalence can be associated with changing sexual behaviour in Uganda, urban Kenya, Zimbabwe, and urban Haiti. Sex Transm Infect 2006; 82(Suppl 1):i1–i8.
36. Mahomva A, Greby S, Dube S, Mugurungi O, Hargrove J, Rosen D, et al
. HIV prevalence and trends from data in Zimbabwe, 1997–2004. Sex Transm Infect 2006; 82(Suppl 1):i42–i47.
This article has been cited 2 time(s).
Journal of the International AIDS SocietyMeasuring and modelling concurrencyJournal of the International AIDS Society
African Studies Review
Assessing AIDS Research in Africa: Twenty-Five Years Later
African Studies Review, 53(1):
epidemiology; heterogeneity in host susceptibility; HIV transmission; modeling
© 2009 Lippincott Williams & Wilkins, Inc.
What does "Remember me" mean?
By checking this box, you'll stay logged in until you logout. You'll get easier access to your articles, collections,
media, and all your other content, even if you close your browser or shut down your
To protect your most sensitive data and activities (like changing your password),
we'll ask you to re-enter your password when you access these services.
What if I'm on a computer that I share with others?
If you're using a public computer or you share this computer with others, we recommend
that you uncheck the "Remember me" box.
Highlight selected keywords in the article text.
Data is temporarily unavailable. Please try again soon.