Four of the six oral presentations in the "Treatment is prevention: the proof is here" session reported on the results from HPTN 052.1 This study had been unblinded four years earlier than planned due to a review by the data and safety monitoring board (DSMB), with all patients now being offered active treatment.
In summary, HIV-positive people on treatment had a 96% reduction in sexual transmission to their HIV-negative partners if they started ARVs with a CD4 count of 350-550 cells/mm3 compared to waiting until it dropped below 250 cells/mm3. As with all prevention studies, condoms, testing and intensive counselling was included throughout the study.
The main study results were presented by Myron Cohen from University of North Carolina.2
HPTN 052 screened over 10,000 couples in order to randomise 1763 HIV-positive people with CD4 counts 350-550 to either immediate or delayed HIV treatment (CD4 confirmed <250 or an AIDS-defining illness). Screening failure was mostly due to CD4 or other criteria in the positive partner, but 300 couples were already both HIV-positive. This was an international study predominantly recruiting in Africa (Botswana, Kenya, South Africa and Zimbabwe, n=954), Asia (India and Thailand, n=531) and Latin America (Brazil, n=276). Men and women were equally distributed as the positive partners. Median baseline CD4 count was 436 cells/mm3 (IQR 365-522) and viral load was approximately 25,000 copies/mL (IQR 6,000-80,000) respectively.
This was generally a low risk population with only 6-8% reporting recent unprotected sex and only 16% aged 18-25 years (~60% were 25-40 years and ~20% >40 years).
The primary transmission endpoint was the prevention of virologically linked transmissions with a primary clinical endpoint of WHO Stage 4 events (including pulmonary TB, severe bacterial infections and death).
Transmission events (n=39) occurred significantly less frequently in the immediate (n=4) compared to the deferred (n=35) treatment arms, p<0.0001. Of these, only 28/39 were linked transmissions (within the couple) with 1 case in the immediate arm vs 27 cases in the deferred arm, p=0.001 (see below for details). Eleven transmissions were either unlinked or undetermined. This translated to incidence rates of linked transmission of 0.1 (95%CI 0.00-0.04) vs 1.7 (95%CI 1.1-2.5) per 100 person years respectively over a median follow-up of 1.7 years.
The single transmission in the immediate treatment arm was detected at the first follow-up visit. However viral diversification analysis estimated that transmission occurred prior to the positive partner initiating treatment (baseline 87,000 viral load) or certainly prior to viral suppression to <400 copies/mL which was recorded at day 28.
Other transmission risk factors were similar between arms, including rates of STIs (low at <5% in both index and partner at baseline and during the study), sexual activity (approximately 70%) and condom use (>90% by all throughout).
Viral suppression (<400 copies/mL) was maintained by >90% of participants in the immediate arm. There was a slow increase in this percentage over time in the deferred arm as people started treatment (from <10% over the first year, 20% by month 24 and increasing to 50% at month 45, though with much fewer patients). The median viral load closest to the time of transmission in the deferred arm was considerable at 80,000 copies/mL but had a wide range from 600 to 630,000 copies/mL.
In multivariate analysis, treatment was the strongest protective effect [HR=0.04, 95%CI 0.01-0.28] compared to condom use [HR=0.33; 95%CI 0.12-0.91]. Factors associated with increased transmission included baseline viral load [per log increment: HR 2.84, 95%CI 1.51-5.41] and baseline CD4 count [per 100 count increment: HR 1.24 95%CI 1.00-1.54].
|Table 1: Key Demographics and Results From HPTN 052|
|Immediate (n=886)||Deferred (n=877)|
|Baseline CD4 med (IQR)||442
|Baseline viral load med (IQR)||4.4
|Age (index partner)||33||32|
|Any unprotected sex||6%||8%|
|Linked transmissions (n)||1||27||HR 0.04 (95%CI 0.01-0.28)|
|WHO Stage IV events, pulmonary tuberculosis, severe bacterial infection or death (n=pts)||40
(2.4 per 100 PY)
(4.0 per 100PY)
|HR 0.59, 95%CI: (0.40, 0.88), p=0.01|
|Extrapulmonary tuberculosis (n)||3||17||p<0.002|
|Deaths||10||13||HR 0.77, 95% CI: (0.34, 1.76), p>0.5|
The second presentation by Susan Eshelman from Johns Hopkins University School of Medicine focused on the analysis of linked transmission.3 This included a helpful introduction to the three types of phylogenetic analyses used: phylogenetic analysis of HIV pol sequences using population sequencing, and statistical analysis of genetic distances from pol sequence pairs for the clearest cases (n=26), and phylogenetic analysis of env sequences obtained by deep sequencing for more complex cases (n=12). Together these provided a high level of reliability for indentifying whether the source of new infections was the HIV-positive partner or whether this was from another partner.
Transmissions in previous serodifferent couple studies have been from outside the main relationship in 25-50% cases.
The deep sequencing ("ultradeep pyrosequencing") supported linked two further cases and confirmed non-linkage for seven others (4 in the immediate and 3 in the deferred arm). Three cases remained unidentified (all in the deferred arm). Transmission linkage was not associated with index partner gender or CD4 count, geographical regions or time on study but this was strongly associated with study group and number of sexual partners in the three months prior to new seroconversions.
Results on the clinical outcomes for the HIV-positive participants in HPTN 052 were presented by Mina Hosseinipour from the UNC Project, Malawi.4
Results comparing the two groups were presented as ITT analyses and included the approximate 20% (184/877) people randomised to the deferred arm who started treatment during follow-up.
Over two years, median CD4 counts increased from 442 to 662 cells/mm3 in the immediate group compared to reducing from 428 to 390 cells/mm3 in the deferred arm. These differences are blunted as the deferred arm includes the response for the 20% people who started treatment. Viral suppression was achieved and maintained <400 copies/mL by >90% of the immediate arm. Less than 5% of patients on immediate treatment experienced virological failure during follow-up with most (60%) of these switching to a second-line combination.
The decision to start treatment in the deferred arm was driven by CD4 declines in 75% of cases. This occurred at a median count of 225 cells/mm3 (IQR 199-247), with 25% over people not starting until their CD4 count was less than 200. Treatment in both arms was predominantly AZT/3TC/efavirenz (70%) with ~10% using AZT/3TC/atazanavir, and ~10% using tenofovir/3TC/efavirenz. CD4 responses in the deferred arm were similar to absolute increase in the immediate treatment arm but remained significantly lower at all timepoints, reflecting the lower counts when starting treatment. Although there are fewer patients with longer duration of follow-up in the deferred arm, other studies have reported that baseline CD4 correlates with CD4 response after treatment.
The analysis by geographical region reported that about 80% of both the linked and unlinked transmission events occurred in African sites, likely a reflection of the higher background population prevalence rates in those countries, although the researchers highlighted higher rates of unprotected sex in the last week (by 9% vs 4% of African vs non-African) and higher sexual activity (>3 acts). However, baseline CD4 count, viral load and adjusted time to initiation, median adherence (99%) and treatment responses were similar between African and Asian sites.
Further details on clinical outcomes were presented by Beatriz Grinsztejn from the Oswaldo Cruz Foundation, Rio de Janeiro.5
Primary clinical events occurred at least once in 105 participants over 3304 person-years (PY) of follow-up; 40 in the immediate arm (2.4/100PY) and 65 in the delayed arm (4.0/100PY), hazard ratio (HR) 0.59, 95% CI: (0.40, 0.88), p=0.01. Seventeen people experienced more than one event. Time to event was significantly shorter in the deferred arm (HR 0.6, 95%CI 0.4, 0.9, p=0.01)
CD4 counts were significantly higher in the immediate arm vs deferred arms for all clinical endpoints (TB 518 vs 316; bacterial infection (mainly pneumonia) 551 vs 337 and death 476 vs 372 cells/mm3 respectively).
The between-arm difference was driven by extrapulmonary tuberculosis with 3 cases in the immediate versus 17 cases in deferred arms (p<0.002). These were peripheral lymph nodes (2 vs 4), abdominal (0 vs 8), pleural (1 vs 3), skeletal (0 vs 1) and meningeal (0 vs 1). Isoniazid prophylaxis was only being used by 4% of patients in each arm at baseline.
Of the 23 deaths observed, there was no difference between arms: 10 in the immediate arm and 13 in the delayed arm [HR 0.77, 95% CI: (0.34, 1.76), NS p>0.5]. Causes of death were similar, but with 3 vs 3 suicides; 0 vs 2 accidents; and 3 vs 6 unknown).
Adverse events potentially related to ART were reported in 24% of subjects in the immediate arm and 5% in the delayed arm, but severe or life-threatening events occurred equally in 14% of each group and grade 4 lab events were also similar (in <1-2% of participants).
Since the DSMB recommendation in April 2011, all participants in the deferred arm have been offered HAART based on the strength of the study findings. This study continues to monitor all participants and results will add to clinical data from use of earlier vs later treatment in people with CD4 counts >350 cells/mm3.
These results add to research that not only correlates viral load with risk of sexual transmission but specifically demonstrates a protective impact with treatment. The two cases of transmission in the early treatment arm (a second was discussed during the presentation) were both detected at the beginning of the study prior to the positive person reaching suppressed viraemia <400 copies/mL.
The fewer clinical endpoints from earlier treatment for the HIV-positive partners in this study are important but were driven by extrapulmonary TB. This clinical difference has significance for people in geographical regions where this study was run, but this aspect of the results was unexpected and has yet to be explained. A more generalisable benefit to people in Western countries is probably the reduced CD4 response in the deferred arm and this needs to be supported by longer follow-up. The ongoing START study will report on whether clinical benefits result from earlier treatment in Western countries.
It would be interesting to model the potential number of transmissions that have already been prevented over the last ten years from the seven million people globally on HAART. Given the financial constraints of access to treatment the additional impact on prevention should be included in future cost: benefit analysis.
The results from HPTN 052 clearly support offering an option for treatment to HIV-positive people who have HIV negative partners. This has been included in UK (BHIVA) guidelines for many years.
When access to treatment is limited with a waiting list using CD4 upper cut-offs to access treatment, those with the most severe medical should clearly be prioritised. However, the majority of the nine million people currently identified by UNAIDS and WHO analyses as requiring but not yet able to access treatment are likely to be undiagnosed. Broadening the CD4 criteria for access to treatment as prevention at higher CD4 counts is unlikely to directly deny access to treatment for more advanced patients.
It was unfortunate that a WHO guideline due to be launched at the IAS meeting, that included the recommendation for treatment people with CD4 counts higher than 350 and who have HIV-negative partners, based on the HPTN 052 study was withdrawn at the last minute.6
Although printed for a launch at the conference there is concern that while the scientific evidence is clear -- and this should be the focus for clinical guidelines -- practical issues on implementation have stalled their release perhaps under pressure from prominent WHO funders. It is difficult to understand how such a useful document that included broad community consultation and approval to the stage of print would have been retracted at such a late stage. WHO say this is due to a need to make "small modifications" and "to review their modeling data they used to inform investment structures". The timeline for these changes are 2-3 months.
This plausibility for intervention from outside the extensive WHO guidelines writing and advisory panels is supported by an article in Science magazine that names the Gates Foundation specifically related to their interest in the latest PrEP results also being included.7
Unless stated otherwise, all references are the Programme and Abstracts of the 16th IAS Conference on HIV Pathogenesis, Treatment and Prevention, 17-20 July 2011, Rome.
Links to external websites are current at time of posting but not maintained.