Capstone Project Topic Selection and Approval

e632 www.thelancet.com/hiv Vol 6 September 2019

Viewpoint

The disconnect between individual-level and population-level HIV prevention benefits of antiretroviral treatment Stefan Baral, Amrita Rao, Patrick Sullivan, Nancy Phaswana-Mafuya, Daouda Diouf, Greg Millett, Helgar Musyoki, Elvin Geng, Sharmistha Mishra

In 2019, the HIV pandemic is growing and soon over 40 million people will be living with HIV. Effective population- based approaches to decrease HIV incidence are as relevant as ever given modest reductions observed over the past decade. Treatment as prevention is often heralded as the path to improve HIV outcomes and to reduce HIV incidence. Although treatment of an individual does eliminate onward transmission to serodifferent partners (unde tectable=untransmittable or U=U), population-level observational and experimental data have not shown a similar effect with scale-up of treatment on reducing HIV incidence. This disconnect might be the result of little attention given to heterogeneities of HIV acquisition and transmission risks that exist in people at risk for and living with HIV, even in the most broadly generalised epidemics. Available data suggest that HIV treatment is treatment, HIV prevention is prevention, and specificity of HIV treatment approaches towards people at highest risk of onward transmission drives the intersection between the two. All people living with HIV deserve HIV treatment, but both more accurately estimating and optimising the potential HIV prevention effects of universal treatment approaches necessitates understanding who is being supported with treatment rather than a focus on treatment targets such as 90-90-90 or 95-95-95.

Introduction In 2019, we are at a pivotal time in the global HIV response in that many people believe that the HIV pandemic is over given the advances in HIV treatment.1 Yet the HIV pandemic continues to grow as defined by numbers of people living with HIV. Specifically, given the encouraging decreases in overall mortality among people living with HIV, in the context of universal treat ment as prevention, approximately 930 000 more people annually (1·7 million new infections minus 770 000 deaths of people living with HIV) require anti retroviral therapy (ART) and many more would need to change ART regimens. At the current rate of new infections, over 40 million people will be living with HIV by 2025.2 The global optimism about the HIV pandemic has not been matched by decreases in new HIV infections. New infections have declined by less than 2% per year since 2005, which means that between 1·8 and 2·5 million people acquired HIV in 2017.2,3 To date, just over 60% of the 37·9 million people living with HIV are on ART; of those 37·9 million, just over half (20·1 million) are estimated to have achieved viral sup­ pression.2 Taken together, these data suggest that an estimated 18 million people living with HIV require ART or improved ART regimens given increasing resistance and concerns about drug quality.2,4

Treatment as prevention for HIV was con ceptualised in response to both observational data5–7 and individually randomised controlled trials8–10 that showed risk of HIV transmission to one’s direct sexual partners were closely linked to the viral load among people living with HIV. If people with HIV had undetectable viral loads, then risk of HIV transmission is zero—referred to as undetectable=untransmittable (U=U).11 These trials were followed by observational studies reinforcing U=U, with no linked transmission events among heterosexual and same­sex male serodifferent couples who reported condomless sex in the absence of HIV pre­exposure prophylaxis.5–7,12

Given the consistent data supporting U=U, treatment as prevention at the population level was expected to reduce HIV incidence substantially through reductions in onward transmissions at the population level. Although the data on U=U at the individual level are clear, whether treatment has decreased HIV incidence at the population level in proportion to increases in coverage of effective treatment is far less clear. However, a linear or dose­ response effect between treatment and incidence re­ ductions would require all people living with HIV to have similar risks of onward transmission if they were not virally suppressed. Data on the heterogeneities that exist in every setting in onward HIV transmission risks in the context of different sexual networks challenge the dose­ response effects of treatment as prevention. These data are further contextualised by rich literature in disparities and inequities research showing that, often secondary to structural determinants, neither treatment nor viral suppression is equal among populations living with HIV. Three large­scale cluster randomised controlled trials did not show incidence declines attributable to universal access to ART. These experimental data complement observational data showing sustained HIV incidence despite increased HIV treatment across municipalities, regions, and nationally in countries across income levels.

Here, we synthesise data supporting U=U for serodifferent couples, treatment as prevention at the population level, and potential reasons for the discon­ nect in observed effect between these two intervention strategies.

Unequivocal data supporting benefits of U=U for individuals and serodifferent couples A series of randomised controlled trials showed that early ART initiation can have immediate and clinically meaningful individual­level benefits, including reductions in morbidity and mortality among people living with HIV and reductions in the rate of linked partner transmissions

Lancet HIV 2019; 6: e632–38

Published Online July 19, 2019

http://dx.doi.org/10.1016/ S2352-3018(19)30226-7

Center for Public Health and Human Rights, Department of

Epidemiology, Johns Hopkins School of Public Health,

Baltimore, MD, USA (S Baral MD, A Rao ScM);

Department of Epidemiology, Laney Graduate School, Rollins School of Public Health, Emory

University, Atlanta, GA, USA (P Sullivan PhD); Research and

Innovation Office, North West University, Potchefstroom,

South Africa (N Phaswana-Mafuya PhD);

Enda Santé, Dakar, Senegal (D Diouf MS); amfAR,

the Foundation for AIDS Research, Washington, DC, USA

(G Millett MPH); National AIDS and Sexually Transmitted

Infection Control Programme, Ministry of Health, Nairobi,

Kenya (H Musyoki MPH); Department of Medicine,

University of California, San Francisco, CA, USA

(E Geng MD); and St Michael’s Hospital, Li Ka Shing

Knowledge Institute, and Department of Medicine,

Division of Infectious Disease, University of Toronto, Toronto,

Canada (S Mishra PhD)

Correspondence to: Dr Stefan Baral, Center for Public

Health and Human Rights, Department of Epidemiology,

Johns Hopkins School of Public Health, Baltimore, MD 21205,

USA [email protected]

For more on treatment as prevention see

http://www.cfenet.ubc.ca/

www.thelancet.com/hiv Vol 6 September 2019 e633

Viewpoint

(figure 1). The TEMPRANO ANRS 12136 trial,9 done at nine care centres in Abidjan, Côte d’Ivoire, between 2008 and 2012, found that among 2056 patients, early ART initiation compared with deferred initiation reduced the risk of death or severe HIV­related illness by nearly 50% (adjusted hazard ratio [HR] 0·56, 95% CI 0·41–0·76). Early ART initiation (ie, ART started immediately after randomisation) reduced HIV­related illness even among people with pre­ART CD4 counts of 500 cells per µL or more (0·56, 0·33–0·94).9 The START trial8 showed similar results. In this trial,8 4685 patients were recruited between 2009 and 2013 from 35 countries and followed up for a mean of 3·0 years. Adults living with HIV with a CD4 count of more than 500 cells per µL were randomly assigned to start ART immediately or to defer initiation until CD4 count decreased to 350 cells per µL, development of AIDS, or development of another condition that required the use of ART.8 Patients were followed up to ascertain the primary endpoint of any serious AIDS­related event, serious event not related to AIDS, or all­cause mortality.8 Early initiation again showed individual­level benefits over starting therapy after CD4 count had declined (eg, adjusted HR for morbidity or mortality 0·43, 95% CI 0·30–0·62).8

Between 2007 and 2010, the HPTN 052 trial10 randomly assigned 1763 index participants living with HIV from multiple low­income, middle­income, and high­income countries to early or deferred initiation to evaluate differences in genetically linked new HIV infections in previously HIV­negative partners.10 Compared with the delayed ART group, early ART was associated with a reduction in linked partner infection (HR 0·07, 95% CI 0·02–0·22).10 Of the 72 partner infections in which viral­ linkage was possible, 46 were linked and 26 were unlinked. Unlinked transmissions came from outside of the sero­ different partnership being evaluated, providing early insights into effectiveness challenges of this approach.10 Of these 26 unlinked partner infections, they were distributed relatively equally in both groups with 14 in the early­ART group and 12 in the delayed­ART group, foreshadowing limitations in HIV acquisition that happens outside primary partnerships. The HPTN 052 trial10 showed that 30% of partner infections could not be prevented by treatment as prevention. Only two­thirds of couples enrolled into the study were available for final analysis of effect, further reinforcing implementation challenges of a universal treatment approach.

To complement these three experimental studies8–10 focused on leveraging HIV treatment to prevent trans­ mission within serodifferent partnerships and preventing HIV­related morbidity and mortality, three large obser­ vational studies evaluated HIV trans mission within serodifferent couples during periods of condomless sex and suppressed viral load of infected partners. Two studies5,7 have followed both heterosexual and same­ sex male HIV serodiff erent couples to track linked HIV transmissions. In the Oppo sites Attract observational

cohort study,5 serodifferent gay male couples were recruited from Australia, Brazil, and Thailand. Between 2012 and 2016, 343 couples had at least one follow­up visit, with a total of 16 800 condomless anal intercourse acts and 258 (75%) HIV­positive partners having viral loads below 200 copies per mL. No linked transmissions were observed.5 In the prospective, observational PARTNER study,7 1166 sero different het erosexual and male homo sexual couples who reported condomless sexual activity between 2010 and 2014 were enrolled.7 Inclusion criteria included that the partner living with HIV was to be virally suppressed for the couple to be eligible.7 Male homosexual couples reported approxi­ mately 22 000 condomless sex acts and heterosexual couples reported about 36 000.7 Similarly, no phylo­ genetically linked new infections were observed.7 Results of the second phase of the PARTNER study were recently published.6 No linked trans missions were found between homosexual couples for nearly 77 000 condomless anal intercourse acts, in which the partner living with HIV was virally suppressed.6 Importantly, these prevention benefits will only be sustained in the context of programmes that address long­term treatment needs for people living with HIV. Recent data from the USA suggest that achieving sustained viral suppression, especially among the most marginalised communities living with HIV, might be a challenge yet to be overcome.18 Taken together, the observational data combined with the efficacy data from the HPTN 052 trial10 do reinforce the veracity of U=U and the efficacy for treatment to prevent HIV transmission in the context of serodifferent HIV partnerships.

Figure 1: Experimental studies evaluating HIV treatment outcomes at the population level and individual level Details of studies and comparisons are given in the main text. Blue lines represent studies with population-level outcomes and red lines are studies with individual-level outcomes. aHR=adjusted hazard ratio. aRR=risk ratio. IRR=incidence rate ratio. ART=antiretroviral therapy.

ANRS 12249 TasP13–15

Reduction in population-level incidence (aHR)

SEARCH16

Reduction in 3-year cumulative population-level incidence (aRR)

HPTN 071 (PopART)17

Reduction in population-level incidence (IRR)

TEMPRANO9

Reduction in HIV morbidity

or mortality (aHR)

START8

Reduction in HIV morbidity

or mortality (aHR)

HPTN 05210

Reduction in linked-partner infection (HR)

0

0·2

0·4

0·6

0·8

1·0

1·2

1·4

1·6

M ea

su re

o f a

ss oc

ia ti

on fo

r t ria

ls

1·01 0·95 0·93

0·56

0·43

0·07

e634 www.thelancet.com/hiv Vol 6 September 2019

Viewpoint

Treatment as prevention as a population-based approach As of 2019, three fully powered cluster randomised trials have measured the effect of universal testing and treatment on population­level reductions in HIV incidence (figure 1). The ANRS 12249 TasP study13–15 was a cluster randomised trial designed to evaluate the effect of early ART, irrespective of CD4 count, on HIV incidence in specific clusters in northern KwaZulu­Natal, South Africa. Communities were randomly assigned either to immediate offer of ART or to standard of care in South Africa at the time (ART initiation at CD4 count ≤350 cells per µL or WHO stage 3 or 4 until December, 2014, then ≤500 cells per µL from January, 2015), and no differences were observed in population­level HIV incidence by group (adjusted HR 1·01, 95% CI 0·87–1·17).14,15 In the SEARCH study,16 32 communities in Kenya and Uganda

were randomly assigned to receive universal ART with a multidisease model, which included patient­centred interventions related to hypertension, diabetes, tubercu­ losis, and HIV. The inter vention reduced annual tuberculosis incidence and improved population HIV viral suppression.16 Although HIV incidence declined across all observed communities, no difference was observed in 3­year cumulative HIV incidence between groups (adjusted risk ratio 0·95, 95% CI 0·77–1·17).16 Conclusions for non­significance associated with treat­ ment included that new infections were coming from outside the community, outbreaks of acute infection, and the small subset of the population living with HIV who were unsuppressed.16 Results from HPTN 071,17 also known as PopART, were presented at the Conference on Retroviruses and Opportunistic Infections 2019. The study17 was designed as a community­randomised trial among 21 urban com munities across Zambia and South Africa. Communities were randomly assigned to three groups between 2013 and 2018: group A (full PopART intervention, which was a combination prevention and treatment intervention, and included immediate ART for all individuals with HIV irrespective of CD4 count), group B (PopART prevention intervention with ART per local guidelines), and group C (standard of care).17

Incidence reductions were observed between the prevention only (group B) and standard of care (adjusted incidence rate ratio 0·70, 95% CI 0·55–0·88).17 However, similar to the SEARCH study and ANRS 12249, HPTN 071 showed no reduction in incidence when comparing the intervention group that included universal treatment (group A) with standard of care (0·93, 0·74–1·18).17 Low rates of linkage to care were seen across the study groups, adding to the evidence that treatment interventions are challenging to implement perfectly in trial settings; challenges that would be amplified in real­world programme settings with more restricted per person intervention budgets than those in trials.

In 2016, the HIV Prevention Trials Network released a statement in response to null results released by the ANRS 12249 investigators at the International AIDS Conference in Durban in 2016.19 Specifically, HPTN 07117 was to differ from ANRS 12249 in three important ways: first, HPTN 071 would take place in urban communities in which the hypothesis was that treatment as prevention would be more effective than in the rural setting of northern KwaZulu­Natal; second, HPTN 071 would be able to assess the full effect of the combination HIV­ prevention package; and third, HPTN 071 would have a longer follow­up period in which to assess the intervention’s effect on incidence.19 That these differences did not change the outcome is now clear with the release of the HPTN 071 study17 results. The intervention did not have differential effect in urban communities compared with rural communities, and the longer follow­up period and larger sample size than in ANRS 12249 did not change HIV incidence outcomes. HPTN 07117 and

Figure 2: Numbers of new HIV infections and HIV treatment coverage in Botswana, Rwanda, and Ethiopia 2010–17 ART=antiretroviral therapy.

2010 2011 2012 2013 2014 2015 2016 2017 0 0

5000 20

10 000 40

15 000

60 20 000

8025 000

30 000 100

N um

be r o

f n ew

in fe

ct io

ns A RT coverage (%

)

0 0

5000 20

10 000 40

15 000

60 20 000

8025 000

30 000 100

N um

be r o

f n ew

in fe

ct io

ns A RT coverage (%

)

0 0

5000 20

10 000 40

15 000

60 20 000

8025 000

30 000

Ethiopia

Rwanda

Botswana 100

N um

be r o

f n ew

in fe

ct io

ns A RT coverage (%

)

Year

New HIV infections Coverage of people receiving ART

www.thelancet.com/hiv Vol 6 September 2019 e635

Viewpoint

SEARCH16 also provided insight into the potential for HIV prevention strategies to reduce HIV incidence. Group B of HPTN 07117 was the most effective, and overall incidence was reduced in SEARCH16 but not more so with universal access to ART. These data complement results of the Ya Tsie trial20 in Botswana released in 2018. Ya Tsie20 was a pair­matched community randomised trial of nearly 9000 individuals from communities covering about 10% of the population, and reported a significant reduction in HIV incidence of at least 30% (incidence ratio 0·65, 95% CI 0·46–0·90) associated with the delivery of a combi nation HIV prevention and treatment programme. The intervention included home­based and mobile testing and linkage­to­care support, with treatment guidelines changing in both groups of the trial over time, towards a treat all approach.20 Taken together, these studies reinforce the fundamental usefulness of HIV prevention in reducing population­level HIV incidence.

Observational data from a range of settings provide consistent conclusions across epidemic and income settings. Estimates by UNAIDS for Botswana, Ethiopia, and Rwanda show a marked increase in HIV treatment coverage since 2010, but a plateau in new infections (figure 2). Botswana, for example, has made substantial improvements in treatment coverage, going from 45% of those living with HIV on treatment in 2010 to just over 80% as of 2017. During the same period, however, Botswana has seen a plateau, or even a small increase in annual new infections, from 12 000 to 14 000 (ie, 16% total increase or about 2·7% annual increase). Although these data focus on countries from across southern and eastern Africa where HIV epidemics are most broadly generalised, the same disconnect between treatment coverage and numbers of new infections and HIV incidence have been observed in high­income settings.21–24 In cities including Vancouver, Sydney, and London, and across the USA and the UK, the incidence rates have been sustained, especially among gay men and other men who have sex with men in the context of universal treatment policies.25–27 Where observed, encouraging declines in HIV incidence have been attributed to advances in HIV prevention including pre­exposure prophylaxis and novel testing approaches to address the most marginalised populations. Similar to the experimental data, these observational data highlight the underlying necessity of HIV prevention in decreasing population­level HIV incidence.

Importance of understanding heterogeneity of risks Early modelling results projected the comparative benefits of scaling up universal and early ART for all people living with HIV.28,29 However, relying on foundational work of epidemic theory might be informative for estimating the potential effect of large­ scale HIV treatment pro grammes moving forward.30,31 Core­group theory posits that pre vention gaps among the

relatively few who are most at risk of acquisition and transmission can sustain an epidemic.32,33 Implications of this epidemic theory are two fold when considering treatment as prevention. First, viral suppression within a serodifferent partnership can avert an infection in a direct partner, but also leads to indirect benefits to the partners’ partners (and any onward serial partners). These indirect benefits stem from the way prevention among a relatively few can protect many along a potential transmission chain—especially if those few are at the highest risk of acquisition and transmission.34,35 Intersecting individual and structural determinants underlie heterogeneity in the risks of acquisition and transmission, including through biology (eg, greater mucosal tearing for anal vs vaginal intercourse or differing cervical surface area exposure for younger vs older women) or structural determinants driving disparities in intervention uptake.36–40 This hetero­ geneity can create pockets of residual trans mission that might break a priori predictions of intervention effect. Acquisition risks (susceptibility) and onward transmission risks are intertwined but they are not synonymous, nor are they static: onward transmission risk, in particular, is dynamic over the sexual life course of individuals.41 Thus, for an onward prevention benefit associated with HIV treatment, the people living with HIV receiving treatment must still be at risk for onward HIV trans mission but can no longer transmit because they are virally suppressed. The intervention­specific corollary to duration of time experiencing high onward transmission risks is the person time of viraemia before viral suppression, often stemming from structural barriers to engagement in HIV testing and ART initiation.42,43

Historically, most HIV transmission models of universal test and treat in high­prevalence epidemics, such as that in South Africa, included some heterogeneity in risk between a few groups (usually high, medium, and low).28,44 However, these early models of high prevalence settings done before experimental treatment studies rarely included a focus on key populations because they are a smaller population and were assumed to be less relevant in generalised epidemic settings.28,44,45 Thus, heterogeneity has traditionally been collapsed within the number of risk strata incorporated into these models. Heterogeneity has been condensed further via assumptions about equal reach and access of interventions among populations with different transmission risks. For example, pre­trial modelling of the HPTN 071 study18 simulated heterosexual HIV transmission and anticipated over 60% reduction in HIV incidence in group A (home­based voluntary testing and counselling with universal ART) relative to group C (control group).44 The model included three levels of heterogeneity drawing on the available data at the time from demographic health surveys: low risk (on average one partner every 10 years), medium risk (on average one partner per year), and high risk (more than one partner per year). High­risk sexual prac tices were calibrated to overall HIV prevalence in the total population,

e636 www.thelancet.com/hiv Vol 6 September 2019

Viewpoint

leading to 1·8–2·0 partners per year among a high­risk group that comprised 18–23% of males and females. Thus, the high­risk group subsumed the subset of women engaged in sex work (whose number of partners in South Africa range between four and 19 per day) and who comprised 0·5–2·0% of the adult female population in South Africa.36,46,47 Importantly, the intervention was also assumed to reach each risk group equally. Although the pre­trial modelling did not include key populations, the sensitivity analyses gave prescient insight hinting at the importance of transmissions via sex with a subset of a population that did not receive the intervention. In this case, the anticipated effect of group A and group B was reduced when the proportion of sex acts with individuals outside the community went from 0% to 10%, and especially when the number of partners increased in the community as a whole after the intervention started. Given the empirical data of sexual practices, engagement in HIV prevention and treatment, and incidences of HIV infections collected in the PopART study,17 modelling after the trial of the HIV epidemic in the PopART study sites and its interventions might provide crucial insights into the role that heterogeneity has in explaining the absence of population­level effectiveness attributable to universal testing and treat ment alone.

Conclusion The tools to end the HIV pandemic have existed for several years. But in 2019, the HIV pandemic is not over and, indeed, it is still growing and will likely do so for many years. The population­level treatment as prevention trials were well designed, well executed, and answered key questions regarding the population­level prevention benefits of universal HIV treatment. Their findings should not be discounted, but rather, they should be integrated into our understanding of the underlying HIV transmission dynamics powering the HIV pandemic. Successful application of tools to end the HIV pandemic necessitates a thorough understanding of HIV acquisition and onward transmission risks and effective implemen­ tation to support sustained viral suppression among people living with HIV and to prevent HIV acquisition among those people at risk of infection. The latter is crucial as treatment alone, as shown by both experimental and observational data, remains necessary but insufficient without primary prevention.

The expectation of one to one reduction in onward HIV transmission is only applicable (or restricted) to fixed serodifferent partnerships that are mutually monogamous over time. Thus, U=U is additive across such partnerships over time. If considerable heterogeneity exists in onward transmission risks (eg, 30% of onward transmissions stem from the unmet prevention needs of 2% of the population) then equal distribution of treatment might actually reinforce disparities and continue to underserve those people at highest risk of onward HIV transmission. The assumption challenges the usefulness of targets such

as 90­90­90 and 95­95­95, which were designed with a focus on coverage of all reproductive­age results, regardless of risk, in specific geographical areas. However, results from the previously mentioned trials are consistent with core epidemic theory in that, from an HIV prevention perspective, knowledge of for whom we are providing treatment is more relevant than for how many people.48 A model leveraging data from southern Africa supported this perspective by showing that if those people most likely to be left behind are also the same people who are most at risk of onward transmission, projections could underestimate the potential effect of achieving and surpassing UNAIDS treatment goals.49 Implementation strategies of HIV treatment focused on addressing those most marginalised will have key differences compared with programmes and resources focused only on treatment numbers. The HIV community often considers the implemen tation strategies for HIV treatment as treatment and HIV treatment as prevention to be the same. However, the disconnect between population­level improvements in treatment coverage and viral sup pression and HIV incidence suggests the need for a separate set of considerations for treatment as prevention. In this frame, we leverage the concept of patient­centred medicine to suggest that maximising the prevention effect of HIV treatment programmes to support HIV prevention outcomes necessitates understanding the individuals that we are trying to support in treatment in contrast to treating them as a general population.

The three cluster randomised controlled trials evaluating treatment as prevention that we described in this Viewpoint are unlikely to be surpassed in size or comprehensiveness. Integrating these experimental data into a large body of observational data from around the world suggest that HIV treatment is treatment, HIV prevention is prevention, and specificity of HIV prevention and treatment approaches towards those at highest risk of onward HIV transmission drives the intersection between the two. All people living with HIV deserve HIV treatment, but both accurately estimating and optimising the potential HIV prevention effects of universal treatment approaches necessitates under standing who is being supported with treatment rather than just how many. Contributors SB developed the concept for this Viewpoint and led the writing of the manuscript. SB and AR led the response to peer­review. AR managed the review process to inform the Viewpoint and drafting specific sections. SM led the modelling section with EG providing key inputs to the implementation components, with GM on the policy implications, and with PS, DD, NP­M, and HM providing overview and drafting of key sections.

Declaration of interests PS reports grants and personal fees from the US National Institutes of Health, grants and personal fees from the US Centers for Disease Control and Prevention, grants from Gilead Sciences, Board Membership at the MAC AIDS Fund, and personal fees from Elsevier outside the submitted work. All other authors declare no competing interests.

www.thelancet.com/hiv Vol 6 September 2019 e637

Viewpoint

References 1 Boseley S, Devlin H. End to AIDS in sight as huge study finds drugs

to stop HIV transmission. May 2, 2019. https://www.theguardian. com/society/2019/may/02/end­to­aids­in­sight­as­huge­study­finds­ drugs­stop­hiv­transmission (accessed July 17, 2019).

2 UNAIDS. Global AIDS update 2019—communities at the centre. July 16, 2019. https://www.unaids.org/en/resources/ documents/2019/2019­global­AIDS­update (accessed July 17, 2019).

3 GBD 2015 HIV Collaborators. Estimates of global, regional, and national incidence, prevalence, and mortality of HIV, 1980–2015: the Global Burden of Disease Study 2015. Lancet HIV 2016; 3: e361–87.

4 Eban K. Bottle of lies: the inside story of the generic drug boom. New York, NY, USA: Harper Collins, 2019.

5 Bavinton BR, Pinto AN, Phanuphak N, et al. Viral suppression and HIV transmission in serodiscordant male couples: an international, prospective, observational, cohort study. Lancet HIV 2018; 5: e438–47.

6 Rodger AJ, Cambiano V, Bruun T, et al. Risk of HIV transmission through condomless sex in serodifferent gay couples with the HIV­positive partner taking suppressive antiretroviral therapy (PARTNER): final results of a multicentre, prospective, observational study. Lancet 2019; 393: 2428–38.

7 Rodger AJ, Cambiano V, Bruun T, et al. Sexual activity without condoms and risk of HIV transmission in serodifferent couples when the HIV­positive partner is using suppressive antiretroviral therapy. JAMA 2016; 316: 171–81.

8 Group ISS, Lundgren JD, Babiker AG, et al. Initiation of antiretroviral therapy in early asymptomatic HIV infection. N Engl J Med 2015; 373: 795–807.

9 Group TAS, Danel C, Moh R, et al. A trial of early antiretrovirals and isoniazid preventive therapy in Africa. N Engl J Med 2015; 373: 808–22.

10 Cohen MS, Chen YQ, McCauley M, et al. Antiretroviral therapy for the prevention of HIV­1 transmission. N Engl J Med 2016; 375: 830–39.

11 Eisinger RW, Dieffenbach CW, Fauci AS. HIV viral load and transmissibility of HIV infection: undetectable equals untransmittable. JAMA 2019; 321: 451–52.

12 Rodger AJ, Cambiano V, Bruun T, et al. Risk of HIV transmission through condomless sex in MSM couples with suppressive ART: the PARTNER2 study extended results in gay men. 22nd International AIDS Conference (AIDS 2018); Amsterdam, Netherlands; July 23–27, 2018.

13 Iwuji CC, Orne­Gliemann J, Tanser F, et al. Evaluation of the impact of immediate versus WHO recommendations­guided antiretroviral therapy initiation on HIV incidence: the ANRS 12249 TasP (Treatment as Prevention) trial in Hlabisa sub­district, KwaZulu­Natal, South Africa: study protocol for a cluster randomised controlled trial. Trials 2013; 14: 230.

14 Iwuji C, Orne­Gilemann J, Balestre E. The impact of universal test and treat on HIV incidence in a rural South African population: ANRS 12249 TasP trial, 2012–2016. 21st International AIDS Conference (AIDS 2016); Durban, South Africa; July 18–22, 2016.

15 Iwuji CC, Orne­Gliemann J, Larmarange J, et al. Universal test and treat and the HIV epidemic in rural South Africa: a phase 4, open­label, community cluster randomised trial. Lancet HIV 2018; 5: e116–25.

16 Havlir D, Charlebois E, Balzer L, et al. SEARCH Community cluster randomized study of HIV “test and treat” using multi­disease approach and streamlined care in rural Uganda and Kenya. 22nd International AIDS Conference (AIDS 2018); Amsterdam, Netherlands; July 23–27, 2018.

17 Hayes RJ, Donnel DJ, Floyd S, et al. Impact of universal testing and treatment in Zambia and South Africa: HPTN071 (PopART). Conference on Retroviruses and Opportunistic Infections (CROI); Seattle, WA, USA; March 4–7, 2019.

18 Crepaz N, Dong X, Wang X, Hernandez AL, Hall HI. Racial and ethnic disparities in sustained viral suppression and transmission risk potential among persons receiving HIV care—United States, 2014. MMWR Morb Mortal Wkly Rep 2018; 67: 113–18.

19 HIV Prevention Trials Network. HPTN responds to results from ANRS 12249 TasP study presented at AIDS 2016. July 16, 2016. https://www.hptn.org/news­and­events/press­releases/hptn­ responds­to­results­from­anrs­12249­tasp­study­presented­at (accessed May 3, 2019).

20 Makhema MJ, Wirth K, Pretorius Holme M, et al. Impact of prevention and treatment interventions on population HIV incidence: primary results of the community­randomized Ya Tsie Botswana prevention project. 22nd International AIDS Conference (AIDS 2018) Amsterdam, Netherlands; July 23–27, 2018.

21 Sullivan PS, Hamouda O, Delpech V, et al. Reemergence of the HIV epidemic among men who have sex with men in North America, Western Europe, and Australia, 1996–2005. Ann Epidemiol 2009; 19: 423–31.

22 Chapin­Bardales J, Schmidt AJ, Guy RJ, et al. Trends in human immunodeficiency virus diagnoses among men who have sex with men in North America, Western Europe, and Australia, 2000–2014. Ann Epidemiol 2018; 28: 874–80.

23 Sullivan PS. The impact of pre­exposure prophylaxis with FTC/TDF on HIV diagnoses, 2012–2016. 22nd International AIDS Conference (AIDS 2018); 2018; Amsterdam, Netherlands; July 23–27, 2018.

24 Delpech V. HIV elimination in gay, bisexual and other men who have sex with men. 22nd International AIDS Conference (AIDS 2018); Amsterdam, Netherlands; July 23–27, 2018.

25 Kirwan PD, Chau C, Brown AE, Gill ON, Delpech VC. HIV in the UK—2016 report. December, 2016. https://assets.publishing. service.gov.uk/government/uploads/system/uploads/attachment_ data/file/602942/HIV_in_the_UK_report.pdf (accessed May 20, 2019).

26 Wales NS. New HIV strategy 2016–2020: data report. https://www. health.nsw.gov.au/endinghiv/Publications/nsw­hiv­ strategy­2016­2020.PDF (accessed May 20, 2019).

27 BC Centre for Disease Control. HIV in British Columbia: annual surveillance report 2016. June 1, 2018. http://www.bccdc.ca/ resource­gallery/Documents/Statistics and Research/Statistics and Reports/STI/HIV_Annual_Report_2016.pdf (accessed May 20, 2019).

28 Eaton JW, Johnson LF, Salomon JA, et al. HIV treatment as prevention: systematic comparison of mathematical models of the potential impact of antiretroviral therapy on HIV incidence in South Africa. PLoS Med 2012; 9: e1001245.

29 Granich RM, Gilks CF, Dye C, De Cock KM, Williams BG. Universal voluntary HIV testing with immediate antiretroviral therapy as a strategy for elimination of HIV transmission: a mathematical model. Lancet 2009; 373: 48–57.

30 Volz EM, Ndembi N, Nowak R, et al. Phylodynamic analysis to inform prevention efforts in mixed HIV epidemics. Virus Evol 2017; 3: vex014.

31 Anderson SJ, Cherutich P, Kilonzo N, et al. Maximising the effect of combination HIV prevention through prioritisation of the people and places in greatest need: a modelling study. Lancet 2014; 384: 249–56.

32 Wasserheit JN, Aral SO. The dynamic topology of sexually transmitted disease epidemics: implications for prevention strategies. J Infect Dis 1996; 174 (suppl 2): S201–13.

33 Boily MC, Anderson RM. Sexual contact patterns between men and women and the spread of hiv­1 in urban centers in africa. IMA J Math Appl Med Biol 1991; 8: 221–47.

34 Anderson R, May R. Infectious diseases of humans: dynamics and control. Oxford, UK: Oxford University Press, 1991.

35 Garnett GP. Role of herd immunity in determining the effect of vaccines against sexually transmitted disease. J Infect Dis 2005; 191 (suppl 1): S97–106.

36 Schwartz S, Lambert A, Phaswana­Mafuya N, et al. Engagement in the HIV care cascade and barriers to antiretroviral therapy uptake among female sex workers in Port Elizabeth, South Africa: findings from a respondent­driven sampling study. Sex Transm Infect 2017; 93: 290–96.

37 Mountain E, Mishra S, Vickerman P, Pickles M, Gilks C, Boily MC. Antiretroviral therapy uptake, attrition, adherence and outcomes among HIV­infected female sex workers: a systematic review and meta­analysis. PLoS One 2014; 9: e105645.

38 Risher K, Mayer KH, Beyrer C. HIV treatment cascade in MSM, people who inject drugs, and sex workers. Curr Opin HIV AIDS 2015; 10: 420–29.

39 Shannon K, Crago AL, Baral SD, et al. The global response and unmet actions for HIV and sex workers. Lancet 2018; 392: 698–710.

e638 www.thelancet.com/hiv Vol 6 September 2019

Viewpoint

40 Baral S, Sifakis F, Cleghorn F, Beyrer C. Elevated risk for HIV infection among men who have sex with men in low­ and middle­income countries 2000–2006: a systematic review. PLoS Med 2007; 4: e339.

41 Watts C, Zimmerman C, Foss AM, Hossain M, Cox A, Vickerman P. Remodelling core group theory: the role of sustaining populations in HIV transmission. Sex Transm Infect 2010; 86 (suppl 3): iii85–92.

42 Sullivan PS, Jones JS, Baral SD. The global north: HIV epidemiology in high­income countries. Curr Opin HIV AIDS 2014; 9: 199–205.

43 Powers KA, Miller WC. Critical review: building on the HIV cascade: a complementary “HIV states and transitions” framework for describing HIV diagnosis, care, and treatment at the population level. J Acquir Immune Defic Syndr 2015; 69: 341–47.

44 Cori A, Ayles H, Beyers N, et al. HPTN 071 (PopART): a cluster­randomized trial of the population impact of an HIV combination prevention intervention including universal testing and treatment: mathematical model. PLoS One 2014; 9: e84511.

45 Mishra S, Boily MC, Schwartz S, et al. Data and methods to characterize the role of sex work and to inform sex work programs in generalized HIV epidemics: evidence to challenge assumptions. Ann Epidemiol 2016; 26: 557–69.

46 Coetzee J, Jewkes R, Gray GE. Cross­sectional study of female sex workers in Soweto, South Africa: factors associated with HIV infection. PLoS One 2017; 12: e0184775.

47 UCSF, Institute AH, WHRI. South African Health Monitoring Study (SAHMS), final report: the integrated biological and behavioural survey among female sex workers, South Africa 2013–2014. San Francsico, CA, USA: UCSF, 2015.

48 Hontelez JAC, Lurie MN, Bärnighausen T, et al. Elimination of HIV in South Africa through expanded access to antiretroviral therapy: a model comparison study. PLoS Med 2013; 10: e1001534.

49 Knight J, Ma H, Baral S, et al. Who is left in 10­10­10? Importance of reaching key populations with the HIV cascade. CROI 2019; Seattle, WA, USA; March 4–7, 2019.

© 2019 Elsevier Ltd. All rights reserved.