Modeling interventions to assess HIV epidemic impact in Africa

Mid-2011 marks the 30^th anniversary of the first reports of the AIDS.^1–3 Many scientists who engaged the field of HIV/AIDS in the 1980s did so because their professional disciplines were copacetic with the challenges faced in HIV discovery, as with T-cell immunologists, retrovirologists,⁴ oncologists, behavioral scientists, infectious disease clinicians, and epidemiologists. Others were moved to study HIV disease because of their clinical and/or personal experiences with the disease.⁵ Many of us who cared for dying patients in the pre-antiretroviral era still felt empowered by the epidemiological insights of the day, that we might be able to prevent infection through behavior change and scientific discovery.⁶

Behavioral and structural changes based on the “ABC” principles of delayed sexual debut among, the young (“abstinence”), reduced numbers of sexual partners (“be faithful”), and use of barriers to viral infection (“condoms”) have reduced viral transmission in such nations as Uganda and Thailand.^7,8 Distribution and use of clean syringes and needles have reduced markedly the spread of HIV among persons who inject drugs in North America, Europe, and elsewhere.⁹ Three stellar randomized clinical trials in South Africa, Kenya, and Uganda have shown that adult men undergoing medical circumcision are half as likely to acquire HIV infection as are uncircumcised men.^10–12 We know that exclusive breastfeeding by an HIV-infected mother will reduce risk of transmission to an infant compared to mixed feeding (both breast and replacement “bottle” feeding), though complete replacement feeding is safest of all if clean water and adequate infant formula can be assured.¹³ The blood supply is no longer a menace in most of the world, thanks to HIV screening of donations and blood products like factor VIII.¹⁴ And there is more.

The herculean efforts of our research colleagues in the field of HIV drug discovery and development that have saved so many lives since the advent of combination antiretroviral therapy (cART) have borne fruit in reducing HIV transmission. The first “ART for Prevention” success was in the prevention of mother to child transmission of HIV (PMTCT). In 1994, the AIDS Clinical Trials Group 076 protocol confirmed that zidovudine monotherapy in the pregnant woman and her infant reduced MTCT by two-thirds, albeit with costly oral and intravenous dosing regimens.¹⁵ In 1999, the another NIH-sponsored network (the antecedent to today's HIV Prevention Trials Network [HPTN]) reported the HIVNET 012 study with a simpler single-dose oral monotherapy with nevirapine given to the women pre-partum and the infant post-partum; its impact was almost as good as the ACTG 076 trial, cutting transmission by half but using a cheaper, easier regimen suitable for resource limited settings with underdeveloped health care infrastructures.^16,17 And subsequent trials have used both drugs or cART to drive transmission down to <2–5% of exposed infants), reducing viral resistance in the process.^18–20 Treatment of the breastfeeding mother or ART prophylaxis in her infant both work to drop breastfeeding transmission rates markedly.²¹ This progress in efficacy, however, is belied by the challenges the global health community faces in fully rolling out PMTCT programs. Years of quality improvement research, health human and systems infrastructure improvement, and community education and mobilization will be needed to overcome issues ranging from community stigma to systems underperformance in order to succeed in full coverage of the mothers and infants who need the testing and treatment for prevention.^22–26

In 2010 and 2011, investigators presented long-awaited results of studies conceived up to a decade earlier when the correlation of viral load and transmission risk was noted in observational studies.^27–38 A “chemical condom” was developed, taking advantage of features of tenofovir (long half-life and good tissue concentrations) to develop and test a microbicide, also known as topical pre-exposure prophylaxis (PrEP). South African investigators confirmed in a rigorous randomized and placebo-controlled trial (Centre for AIDS Prevention Research in South Africa [CAPRISA] protocol 004) that topical tenofovir 1% gel worked well in women using the product before and after coitus, reducing risk of HIV acquisition by 39%.³⁹ Furthermore, suggestions in post-hoc analyses that women were even more protected when they were highly adherent to the topical tenofovir gel use, linking the essential behavioral work needed for any biomedical intervention (combination prevention).

Also in 2010, the iPrEx investigators from North and South America reported that among men who have sex with men, oral PrEP (tenofovir disoproxil fumarate [TDF, 300 mg] and emtricitabine [FTC, 200 mg]) worked to prevent infection 44% of the time.⁴⁰ As with the CAPRISA 004 results, iPrEx post-hoc analyses suggested strongly that higher protection was seen in persons with higher adherence to the daily oral regimen. The CDC has issued public health guidance on use of PrEP among MSM on this basis of these findings.⁴¹ An analogous PrEP trial in women was stopped early; FEM-PrEP did not confirm oral TDF/FTC PrEP efficacy in women.⁴² This finding generated the question as to whether the two-log higher genital tract tissue doses seen with topical gel in women might work for prevention of sexual transmission, while the lower cervicovaginal levels from oral dosing may not.

Most recently, sexual partners who are infected with HIV who go on cART at higher CD4+ cell counts than would qualify them for cART in their home countries (that generally follow WHO guidelines) are 96% less likely to transmit infection to their partners than persons who do not take cART at this earlier timeframe (the HPTN 052 protocol).⁴³ Thus the circle is complete; antiretroviral drugs can help the infected person delay disease progression and can also provide the benefits to sexual partners and newborn infants alike of markedly reduced infection. And ARVs can even be used for prophylaxis uninfected persons who are especially vulnerable to HIV acquisition.

The study by Williams et al in the current issue of JAIDS⁴⁴ continues the useful investigative tradition of mathematical modeling the putative impact of an innovative prevention methodology, in this case the CAPRISA 004 topical tenofovir trial results. Use of a microbicide at a “high” level of 90% of all sexual encounters over a 20 year span in South Africa would reduce transmissions markedly, preventing 2 million infections. Use of the products at a “low” level of 25% of sexual encounters would still reduce transmissions markedly, preventing half a million infections over the 20 year span modeled. The investigators begin their model in 2011, when they would have done well to start in 2015 when it is more likely the product will actually be licensed and in use. The modeling is presented clearly and the mathematics are sound insofar as one can conclude from the summary methods. The optimistic conclusions are inspiring and suggest the urgency of confirmatory trials to ensure the validity of the single positive CAPRISA 004 trial. We likely have a new tool to combat HIV that can save lives; scientists, pharmaceutical manufacturers, and government policy makers alike must work tirelessly to make tenofovir gel microbicides available expeditiously.

There is a caveat that one must acknowledge, however. The difficulty with transmission models is not so much that the challenges in the mathematic methods are insurmountable, but rather that the parameter estimations are too often indefensible. Let's take the high end coverage for a moment and ask whether it is plausible, under any foreseeable scenario, that 90% of South African women (or any other women) would succeed in using topical tenofovir gel for 90% of their sexual encounters for the next 20 years. Whether adherence data are extrapolated from analogous preventive medicine behaviors such as adherence to male or female condom use, hormonal contraception, or daily medications for tuberculosis, one would not posit 90% adherence to gel use before and after every coital event over 20 years to be plausible.^45–53 Similarly fanciful assumptions in other models lead to such conclusions that if every HIV-infected person in the world were diagnosed and then took ART daily with 100% adherence for the rest of their lives, the epidemic would go away, or if seronegative sexually active persons took PrEP, the incidence would plummet.^54–56 Models of HIV control strategies need more plausible assumptions of uptake and adherence to guide real-world policies; in turn, unrealistic assumptions will drive fanciful modeling assumptions that may well mislead policymakers.

As for the lower coverage rate, even 25% use of a product for each and every sexual encounter for the modeled 20 year period will be challenging given adherence levels reported in prior microbicide studies. A substantial increase in preventive resources would be needed to achieve even the “low” coverage estimate over 20 years for dual use of tenofovir 1% gel for 25% of coital episodes. And time will tell the extent to which the global donor community and the government of South Africa will increase investment in HIV prevention in that comparatively prosperous nation, compared to its poorer southern African neighbors struggling with similar epidemic profiles.

Modeling single interventions with challenging population coverage rates for product use with the millions of coital episodes per month is still a useful exercise. Even more helpful are those models that tackle combination prevention approaches including identifying persons with HIV early in their disease, effectively linking diagnosed persons to care, placing infected persons on cART as best we can, and maximizing adherence to the cART regimen for many years to come (ART for Prevention). At the same time, we must roll out ARV PrEP to the vulnerable HIV uninfected persons who are willing and able to access and take regularly the oral (for men) or topical (for women) ARVs that can reduce their risk. “ABC” interventions and community education and mobilization are vital adjuncts to any biomedical intervention. Medical male circumcision programs must be expanded for willing adult men and for male infants whose parents are supportive. Imagine if a panoply of prevention tools were made available in communities at the same time, even with imperfect coverage.^57–67 Modeling the levels and combinations of coverage that will be plausibly achieved can be done using realistic assumptions given varying epidemic circumstances and community receptivity to the components of prevention offered. This is meaningful and reflects the real world, building on the insights of such unitary intervention models such as that of Williams et al.⁴⁴