Injected PrEP: cabotegravir maintains its advantage over four years

Three more possible cabotegravir breakthrough infections seen
Professor Raphael Landovitz (bottom left) at CROI 2022.
Professor Raphael Landovitz (bottom left) at CROI 2022.

An update on HPTN 083, the study that established that PrEP given as an injection every two months had superior efficacy to a daily pill in gay and bisexual men and transgender women, was presented this week at the Conference on Retroviruses and Opportunistic Infections (CROI 2022).

Four and a half years of data are now available. The efficacy of cabotegravir injections relative to oral PrEP using tenofovir disoproxil fumarate plus emtricitabine (TDF/FTC) remained almost unchanged. People taking the injections had only a third as many HIV infections as people taking the pills: an overall efficacy over TDF/FTC of 66%, compared with 68% reported last year from the blinded part of the study. Adherence was lower in the unblinded phase, but to both cabotegravir and TDF/FTC.



When a clinical trial is blinded, the participants are unaware as to whether they are receiving the experimental drug or a placebo (or another drug). Double blinding refers to the participant, their doctor and researchers running the trial not knowing which treatment is received by each group until all data have been recorded. Blinding is done to reduce bias in clinical trials.


Refers to the mouth, for example a medicine taken by mouth.


A clinical trial where both the researcher and participants know who is taking the experimental treatment. 


How well something works (in a research study). See also ‘effectiveness’.


An umbrella term for people whose gender identity and/or gender expression differs from the sex they were assigned at birth.

HPTN 083’s principal investigator, Professor Raphael Landovitz of the University of California, Los Angeles, presented an update including figures from the last months of the blinded part of the study and also from an unblinded extension.

We have described the study in detail in the report linked above and in its initial results from 2020. Briefly, 4566 participants from sites in the US, Latin America, Thailand and South Africa were randomised either to have a cabotegravir injection every two months (after five weeks of oral cabotegravir) or to take daily TDF/FTC. One in eight participants was a transgender woman; two-thirds were aged under 30 when they entered the study, and half of the US participants were Black.

The placebo-controlled, blinded study lasted 3.4 years. Participants had either received active injections and dummy pills, or dummy injections and active pills. When study results were announced in May 2020, participants were then told which type of PrEP they had really been on and invited to continue taking it. The first year’s results from this unblinded phase up to June 2021 were presented at CROI.

Last year, it was reported that there had been 39 HIV infections in people taking TDF/FTC and 12 in people on cabotegravir. This equated to 68% fewer infections on injectable PrEP.

This year, Professor Landovitz reported three sets of figures:

  • They had found four more infections in the blinded part of the study, two in each arm, bringing the totals to 41 on TDF/FTC and 14 on cabotegravir. This equates to an annual HIV incidence of 1.29% on TDF/FTC and 0.44% on cabotegravir, meaning the infection rate was 66% lower on injectable PrEP than on oral PrEP in the blinded study.
  • In the first year of the open-label phase, there were 31 infections in people taking TDF/FTC and 11 on cabotegravir. The annual incidence rates, of 2.2% on TDF/FTC and 0.76% on cabotegravir, were about 70% higher than in the blinded study. However, the incidence rate ratio was unchanged: there were 67% fewer infections on cabotegravir.
  • Combining the blinded and open-label phases, there were 72 infections on TDF/FTC and 25 on cabotegravir – which again did not change the overall efficacy of 66% fewer infections on injectable PrEP compared to oral PrEP.

Adherence was lower in the open-label study. Drug-level measurements taken in people on TDF/FTC showed that the proportion of participants taking the effective level of four or more doses a week was 73% in the blinded phase but only 59% in the open-label phase. The proportion of people who had missed their cabotegravir injections and were unlikely to have sufficient drug levels rose from 9% to 19%.

This may be related to a change in the demographics of the study, although this hasn’t been formally examined. In the blinded study, 47% of participants were in the US and 32% in Latin America. In the open-label phase, 22% were in the US and 54% in Latin America. Study sites in Peru were closed at times due to the country’s particularly grave COVID-19 epidemic and this may have contributed to lower adherence.

Analysis of infections

In the previous report, we described 16 infections in people allocated to cabotegravir. Four of these were people who had acquired HIV shortly before they started taking cabotegravir. Because of this, they were not counted as infections taking place during the study.

This left the 12 people referred to above who had acquired HIV after starting cabotegravir.

For three people, this happened during the oral ‘lead-in’ phase of the study, and five were people who stopped their cabotegravir injections, including two who switched back to TDF/FTC.

But four of the 12 infections were in people who received cabotegravir injections and had adequate drug levels.

In the new analysis, there were 13 more infections in people regarded as being in the cabotegravir arm.

  • Seven were counted in the overall calculation of efficacy, but had not actually received an injection for more than six months.
  • Three, all in the open-label phase, had had generally good adherence but tested positive after missing their injections for a period of one to 3.6 months. They were treated as possible but unproven infections on cabotegravir.
  • This left three more infections that occurred in people who appeared to have good adherence to cabotegravir.

Looking at these three in more detail, one participant still in the blinded phase tested positive after a long lapse in their cabotegravir use, but their infection could be traced back to a period when they appeared to have good adherence. In the other two infections, one in the blinded and one in the unblinded phase, the two participants tested HIV positive after apparently consistent and regular cabotegravir use; one tested positive after their fifth cabotegravir injection and one, in the unblinded phase, after 15 injections and two years in the study.

So now there are a total of seven infections in people from both phases of the study with apparently sufficient cabotegravir levels. One characteristic of these cases, as noted in the previous report, is that being on PrEP had produced ambiguous and indeterminate test results that made their HIV infections hard to detect. HIV testing and drug resistance in these cases were covered in a second presentation by Dr Susan Eshleman of Johns Hopkins University, which will be reported separately.

Professor Landovitz hailed the consistent, and persistent, advantage of injected PrEP over oral PrEP as "remarkable". He urged countries other than the eight (Brazil, the US, Australia, South Africa, Zimbabwe, Malawi, Botswana and Uganda) that have either approved or are in the process of approving injectable cabotegravir for PrEP to do so.


Landovitz RL et al. Updated efficacy, safety, and case studies in HPTN 083: CAB-LA vs TDF/FTC for PrEP. Conference on Retroviruses and Opportunistic Infections, abstract 96, 2022.

View the abstract on the conference website.

Update: Following the conference presentation, this study was published in a peer-reviewed journal:

Marzinke M et al. Extended Analysis of HIV Infection in Cisgender Men and Transgender Women Who Have Sex with Men Receiving Injectable Cabotegravir for HIV Prevention: HPTN 083. Antimicrobial Agents and Chemotherapy, 67: e00053-23, April 2023.


Correction: This article was amended on 20 February 2022. An earlier version suggested that drug levels had been tested in people with breakthrough infections on cabotegravir, but this analysis has not yet been completed.