A gonorrhoea vaccine hope was tested properly and failed; a weekly HIV pill held up

Gonorrhoea is running out of antibiotics. Ceftriaxone resistance is spreading, and the pipeline is thin, which is why an observation made several years ago attracted so much attention: populations vaccinated against meningococcal B appeared to have less gonorrhoea. The bacteria are closely related, and the vaccine seemed to be offering incidental protection.

That observation has now been tested in a randomised trial, and it did not survive.

The vaccine result

GoGoVax randomised 654 participants between July 2021 and May 2023, of whom 587 were included in the primary analysis, among men who have sex with men at high risk for gonorrhoea [s1].

The incidence of N. gonorrhoeae infection was 48.1 events per 100 person-years — 160 events among 296 participants — in the 4CMenB group, and 47.8 per 100 person-years, 155 events among 291 participants, in the placebo group [s1]. The incidence rate ratio was 1.01 (95% CI, 0.80 to 1.26; P = 0.97), giving a vaccine efficacy of −0.5% (95% CI, −26.2 to 19.9) [s1].

Efficacy of −0.5%, with a p-value of 0.97 [s1]. This is about as close to exactly nothing as a trial result comes.

The subgroups do not rescue it. Efficacy was 5.5% (95% CI, −56.0 to 42.8) for symptomatic infection, −6.4% (−47.5 to 23.2) for asymptomatic infection, and −20.0% (−90.2 to 23.9), −1.2% (−33.0 to 23.0) and 2.6% (−27.7 to 25.7) for urogenital, anorectal and oropharyngeal infection respectively [s1]. Every confidence interval spans zero, and the point estimates scatter either side of it, which is the signature of no effect rather than a small one.

Serious adverse events occurred in 4.7% of the 4CMenB group and 2.8% of the placebo group [s1].

Note also the incidence figures: roughly 48 infections per 100 person-years in both arms [s1]. This is a population with very high transmission, which is exactly where a vaccine effect would be easiest to detect. The trial had the statistical conditions to find a benefit and found none.

Why the earlier signal looked real

The observational studies that generated the hypothesis compared people who had received meningococcal B vaccine with people who had not. Vaccination is not randomly distributed. People who get vaccinated differ in healthcare engagement, testing frequency and sexual health behaviour, and several of those differences independently predict gonorrhoea incidence.

Testing frequency is a particular problem, since a large share of gonorrhoea is asymptomatic and only found by screening. If vaccinated people are also more engaged with sexual health services, they will be tested more — which should push measured incidence up, not down, making the original protective signal harder to explain by that route alone. But confounding does not have to run in one direction, and the trial is the arbiter.

This is a clean example of a plausible mechanism, supported by observational data, that did not replicate under randomisation. The mechanism was never implausible — the organisms share surface antigens. It simply was not true at a magnitude that matters.

The trial in the same field that worked

The second study concerns treatment rather than prevention, and reports the kind of incremental gain that actually accumulates.

Between October 2024 and April 2025, 727 participants were screened, 647 were eligible, 634 randomised and 626 treated — 314 to once-weekly islatravir-lenacapavir and 312 to standard of care [s2]. Of the 626, 211 (34%) were female, 193 (31%) Black, 119 (19%) Asian, 123 (20%) Hispanic or Latine, and 89 (14%) aged 65 years or older [s2]. At baseline 552 (88%) were on a single-tablet regimen and 478 (76%) on regimens containing integrase strand transfer inhibitors [s2].

At week 48, one participant (0.3%) of 314 on islatravir-lenacapavir and four (1.3%) of 312 on standard of care had an HIV-1 RNA viral load of 50 copies per mL or higher, a difference of −1.0% (95.002% CI, −3.0 to 1.1), meeting non-inferiority [s2].

Treatment-related adverse events occurred in 58 (18%) of 314 on islatravir-lenacapavir against one (<1%) of 312 on standard of care [s2]. That is a large difference in reported treatment-related events, and it belongs in any summary of this trial. The more serious measures were closer: grade 3 or higher adverse events in 24 (8%) against 28 (9%), serious adverse events in 22 (7%) against 27 (9%), and discontinuation for adverse events in two (1%) against one (<1%) [s2]. There were three deaths, one and two respectively, none considered treatment-related [s2].

The authors describe the regimen as having potential to be the first once-weekly complete oral single-tablet regimen for HIV-1 treatment, and note that longer-term safety data will provide further insight beyond week 48 [s2]. The trial was funded by Gilead Sciences and Merck Sharp & Dohme [s2].

What the pair illustrates

One field, two 2026 trials, opposite outcomes — and the useful contrast is in what each was asking. The vaccine trial tested a hypothesis generated from observation. The HIV trial tested a formulation against an established comparator on a hard virological endpoint.

The first kind fails more often, and that is not a criticism of running it. Gonorrhoea's resistance trajectory made 4CMenB worth the trial; knowing it does not work redirects effort toward vaccines designed for the organism.

What to watch

Whether gonorrhoea vaccine development refocuses on candidates designed against N. gonorrhoeae rather than borrowed from meningococcal B [s1]. And whether week-48 non-inferiority for weekly dosing holds with longer safety follow-up, given the 18% treatment-related adverse event rate [s2].

This article describes published trial results. It is not medical advice.

Sources

• [s1] Meningococcal B Vaccine to Prevent Neisseria gonorrhoeae Infection, New England Journal of Medicine, 2026;395(4):349–361.
• [s2] Switch to once-weekly, single-tablet islatravir-lenacapavir from daily standard of care for HIV-1, The Lancet, 2026;408(10557):821–832.