On Monday February 3rd it was announced that HVTN 702, a large-scale HIV vaccine efficacy trial taking place in South Africa, had found no evidence of protection against HIV acquisition. The information emerged from a pre-planned review of interim results conducted by the trial’s Data Safety Monitoring Board (DSMB), which revealed roughly equal numbers of HIV infections among vaccine and placebo (dummy vaccine) recipients. There were no indications that the vaccine regimen had caused harm. Following the DSMB’s recommendation, immunizations have been stopped, and plans are being made to unblind participants (inform them whether they received active vaccine or placebo) and follow them for an additional 12 months to monitor for safety.
The research was sponsored by the HIV Vaccine Trials Network (HVTN) and had hoped to build on marginal evidence of efficacy observed in RV144, a vaccine trial that took place in Thailand with results reported in 2009.
Since the initial announcement, two webinars accessible to people in the USA have provided additional details on the HVTN 702 results and plans for follow up—the first was sponsored by the HVTN and Black AIDS Institute (BAI) and held on February 5th, the second was sponsored by AVAC and Advocacy for the Prevention of HIV and AIDS (APHA) and held on February 19th (a recording of this webinar is available).
The HVTN 702 Results
HVTN 702 recruited 5,407 sexually active, HIV-negative men and women aged between 18 and 35 years. On the HVTN/BAI webinar, principal investigator Glenda Gray explained that the DSMB’s interim analysis was based on 5,383 participants – 2,694 in the vaccine arm and 2,689 in the placebo arm. The median follow up time was 18 months. The DSMB analysis found that there had been 129 infections among vaccine recipients and 123 among those who received the placebo, indicating no effect on the risk of acquiring HIV in this population. Importantly, there was no possibility of the vaccine showing a significant effect on HIV acquisition if the trial had been allowed to continue until the original planned completion date, hence the DSMB recommendation to cease further immunizations.
Gray noted that 77% of the total planned follow up between study entry and month 24 had been completed, and 62.1% of participants had reached month 18.5, two weeks after the final vaccination. Data on the incidence of HIV infection measured every three months showed that it was almost identical between vaccine and placebo arms at every timepoint.
Gray also presented the incidence rates based on participant sex at birth and again the numbers were closely matched between vaccine and placebo. There were more women than men in the trial, so this analysis was based on 1886 female vaccine recipients, 1886 female placebo recipients, 808 male vaccine recipients and 803 male placebo recipients. At month 24, 114 HIV infections had occurred in females in the vaccine arm, 107 in females in the placebo arm, 14 in males in the vaccine arm and 14 in males in the placebo arm. HIV incidence was approximately four-fold higher in females, consistent with what is seen at the population level in South Africa.
Gray made clear the disappointment felt by the HVTN 702 co-chairs (Gray, Linda-Gail Bekker, Fatima Laher and Mookho Malahleha) and everyone else involved in the trial, but also pointed to the DSMB’s commendations for the how the research was conducted and the high regard shown for the participants. The trial answered a critical question, albeit not in the way that had been hoped, and there are over 15,000 samples available that will facilitate efforts to understand why the vaccine regimen did not work.
Discussing the possible implications of the results, Gray highlighted that one key difference with the RV144 trial in Thailand is the far greater rate of exposure to HIV among women—approximately 14 times higher in the South African versus Thai trial populations. Additionally, the failure of the specific vaccine regimen used in RV144 does not mean that other approaches under study are also doomed—Gray cited several ongoing trials, such as those of an HIV vaccine designed by Janssen, the PrepVacc study which is evaluating both pre-exposure prophylaxis (PrEP) and vaccination, the AMP protocols testing the efficacy of broadly neutralizing antibodies, and research into long-acting antiretrovirals for PrEP.
Community Feedback and Next Steps
On the AVAC/APHA webinar, Linda-Gail Bekker provided additional information on the follow up plans for HVTN 702. Participants are being invited to consent to being followed for a further 12 months to evaluate safety, and at their next visit will be informed as to whether they were in the vaccine or placebo arm of the trial. Bekker also explained that a journal article describing the results is in preparation, and the most likely forum for more detailed presentations on a range of issues related to the trial will be the R4P conference which takes place in Cape Town from October 11-15, 2020.
Ntando Yola, co-founder of APHA, introduced several community representatives involved with the HVTN 702 sites in South Africa who discussed their perspectives on the trial and feedback from participants. As with those responsible for conducting the trial, disappointment was the overarching feeling but there was also recognition that the negative result was still very important. There have been some expressions of concern by participants about the slight numerical difference in the number of HIV infections between the vaccine and placebo arms of the trial (six more in the vaccine group) and communications have focused on explaining why this is not a significant difference or evidence that the vaccine was harmful.
Questions have also arisen about the possibility of vaccine-induced seropositivity (VISP), provision of PrEP and the approach taken to participants who seroconverted. Linda-Gail Bekker addressed these issues, noting that to date very little VISP has been observed in HVTN 702. PrEP was made available to participants (and continues to be available), but uptake has been disappointingly low, mirroring the situation in South Africa outside of the trial. Bekker hopes that imminent PrEP promotion and demand creation initiatives from the Department of Health will help remedy this problem. Individuals who seroconverted during the trial were immediately referred for HIV treatment and are being followed.
The Vaccine Regimen in HVTN 702 and Other Efficacy Trials
On the topic of the specific HIV vaccine regimen used in HVTN 702, Mary Marovich, Director of the Vaccine Research Program at the Division of AIDS (National Institute of Allergy and Infectious Diseases, US National Institutes of Health) presented information on the types of scientific studies likely to be conducted to try to understand why it did not work. Marovich also addressed how the regimen is different from other HIV vaccines currently in efficacy trials.
HVTN 702 tested a prime-boost regimen similar to—but not exactly the same as—that used in the Thai RV144 trial. A harmless canarypox virus vector (ALVAC) delivering selected HIV antigens (parts of the virus capable of inducing an immune response) was administered at months 0, 1, 3, 6, 12 and 18. The antigens were derived from HIV clade C, the predominant virus variant circulating in South Africa. The boost part of the regimen comprised two differing forms of the HIV gp120 envelope protein, both also derived from HIV clade C, delivered with the adjuvant MF59. The boost was administered at months 3, 6, 12 and 18.
In RV144, the HIV antigens were derived from clades B and E (also known as CRF01_AE) and the immunizations were less frequent: ALVAC was given at months 0, 1, 3 and 6 and the gp120 protein boost at months 3 and 6. The gp120 protein boost was delivered with a different adjuvant, Alum.
A preparatory study (HVTN 100) conducted in South Africa demonstrated that the vaccines used in HVTN 702 induced similar—in some cases superior—immune responses compared to those observed in RV144 participants. These results were published in Lancet HIV in 2018 with additional information on the effects of the 12-month booster presented earlier this week in PLoS Medicine.
Marovich explained that samples from HVTN 702 will be studied to confirm that the vaccines induced robust immune responses, and to evaluate whether these responses were capable of recognizing the HIV strains that participants acquired during the trial. A particular focus will be placed on looking at the types of immune responses that appeared to correlate with reduced HIV risk in RV144 (sometimes described as “correlates of protection”).
The two other large ongoing HIV vaccine efficacy trials are HVTN 705 (Imbokodo) and HVTN 706 (Mosaico). There are superficial similarities with HVTN 702 in that they both involve a prime-boost HIV vaccine regimen, but Marovich cited several key differences. The HIV antigens are, for the most part, not derived from a single virus clade but rather represent specially tailored mosaics made up of elements from multiple different clades—the goal is to develop vaccines capable of working globally, regardless of the locally prevalent HIV clade.
The regimen also comprises a different viral vector, adenovirus serotype 26 (Ad26) and an HIV envelope protein boost containing gp140 rather than gp120. The adjuvant for the boost is Alum, as was used in RV144 but not HVTN 702. Additionally, in HVTN 705, the HIV envelope protein boost is from clade C HIV, whereas HVTN 706 is employing a “bivalent” boost containing both clade C and mosaic HIV gp140 proteins.
Another important distinction is that the rationale for HVTN 705 and 706 does not primarily derive from a prior human HIV vaccine trial, but rather promising results in animal models involving exposure of macaques to SHIVs (hybrids of simian and human immunodeficiency viruses). However, researchers have reported that certain immune profiles induced by the vaccines mirror those associated with reduced risk of HIV infection in RV144.
Revisiting RV144
Arguably, a question that looms over the HVTN 702 results that was not discussed on either webinar is: was the RV144 result real, or a fluke? This has been a matter of some controversy in the scientific community, primarily because the statistical significance of the RV144 result was extremely borderline—the reported reduction in risk of HIV acquisition was 31.2%, with a wide 95% confidence interval of 1.1 to 52.1. If the lower bound of a confidence interval crosses 1, then the result is not statistically significant, emphasizing the fragility of the result. Statisticians have estimated that there is an approximately 22% possibility that the finding could have occurred by chance.
Evidence advanced in support of the protection being real include an analysis of HIV incidence after the first year of RV144, which suggested that efficacy may have been as high as 60% at this timepoint, before vaccine-induced immune responses waned. But this analysis was not planned in advance (described as “post hoc”) and so may be unreliable.
A range of potential immune correlates of protection against HIV have also been reported for RV144. In some cases, there is evidence that similar immune responses offered protection in animal models. Again, however, these results do not represent definitive proof that the vaccine regimen in RV144 was protective—for example, it can be unclear if an individual’s immune response to a vaccine reflects some aspect of their immune system that renders them less susceptible to HIV independent of vaccination. Vaccine research veteran Stanley Plotkin, in a commentary co-authored with Georgia Tomaras, emphasizes that:
“To date, correlates of HIV-1 risk involving multiple immune responses have been identified for the RV144 HIV-1 vaccine efficacy trial. However, these are not yet accepted to be CoP [correlates of protection] since they have not been confirmed in another vaccine efficacy trial.”
There have been lively debates in the scientific literature, with commentaries by Ron Desrosiers and Per Johan Klasse and John P. Moore raising questions about the veracity of the RV144 analyses. These commentaries have in turn have been criticized and at least partly rebutted in a paper by Susan Zolla-Pazner and Peter B. Gilbert, researchers heavily involved in generating the data related to potential correlates of protection.
Importantly, there is very broad agreement—albeit not universal—that none of this uncertainty regarding RV144 undermines the rationale for conducting HVTN 702. The dismaying data from HVTN 702 confirming the high rates of HIV infection in South Africa, particularly in women, underscores that it was vital for researchers to follow up on the possibility of vaccine efficacy reported in RV144. But as the HIV vaccine field works to understand the results and learn from them, revisiting RV144 should not be off the table.
Additional Reading
- HVTN Fact Sheet: Understanding why vaccinations in the HVTN 702 Uhambo study were stopped
- HVTN 702 Q&A
- Why we’re still optimistic even in disappointment about an HIV vaccine - Nandisile Luthuli and Ntando Yola, Daily Maverick, February 6, 2020
Recent Comments