An experimental oral vaccine against Shigella sonnei prevented 89% of shigellosis cases in a deliberately-infected group of American adults, according to a phase 2 trial published June 30 in The Lancet Infectious Diseases. It is the strongest efficacy signal yet reported for a live-attenuated Shigella vaccine, in a field that has produced no licensed product in more than a century of attempts.
The headline number, however, sits on top of a trial that did not go as designed. Safety problems during the study forced the investigators to halve the vaccine dose partway through, drop one study arm, and narrow who could enrol. The pooled efficacy figure combines two different doses — and every case of shigellosis among vaccinated participants occurred at the dose that was abandoned.
What the trial found
The vaccine, WRSs2, is a live but weakened S. sonnei strain carrying deletions of the senA, senB and virG(icsA) genes, given by mouth. Between October 2022 and January 2024, 108 healthy adults aged 18 to 49 with low preexisting immunity to S. sonnei were enrolled at Emory University and Cincinnati Children’s Hospital Medical Center. Seventy-three went on to be challenged: 28 days after their final dose, they were admitted to an inpatient unit and swallowed roughly 1,500 live S. sonnei, then received ciprofloxacin five days later.
Among the 34 who had received two doses of vaccine, three developed shigellosis, as adjudicated by a blinded review committee. Among the 26 placebo recipients, 21 did — an attack rate of 81%. That gives a vaccine efficacy of 89%, with a 95% confidence interval of 71 to 96.
Severe disease was reduced further: one case among the 34 vaccinated versus 18 among the 26 on placebo, an efficacy of 96% (95% CI 78–99). That analysis was post-hoc and used a programmatic rather than adjudicated case definition, so it carries less weight than the primary result.
For context on how unusual this is: an earlier conjugate vaccine, S. sonnei-rEPA, reached 74% efficacy in a field trial of young adults but offered no protection in the youngest children, an Israeli conjugate managed 28% overall, and the previous live-attenuated candidate WRSs1 reached 40% in a Thai challenge model. A S. flexneri 2a bioconjugate, Flexyn2a, showed about 30% efficacy in a US challenge study.
The dose reduction was a safety response, not a design choice
The trial began as three arms: two doses of 10⁶ colony-forming units, one dose of 10⁶ CFU, or placebo. After 69 participants had been dosed, an independent Data and Safety Monitoring Board was convened because participants were developing grade 3 gastrointestinal reactions to the vaccine. Six of the 108 enrolled participants had grade 3 post-vaccination adverse events over the course of the trial, triggering two separate safety reviews.
The first review, in March 2023, followed severe nausea and diarrhoea in two participants — both with obesity, both taking weight-loss medication. The protocol was amended: the dose was cut in half to 5×10⁵ CFU, enrolment in the single-dose arm stopped, randomisation shifted to 2:1, and anyone with a BMI above 40 or taking weight-loss drugs was excluded. A second pause in January 2024 followed four more grade 3 events, two of them linked to intercurrent COVID or norovirus infection. No further changes were made. There were no vaccine-related serious adverse events and no deaths, and every affected participant tolerated their second dose.
The consequence for the efficacy result is significant. The primary comparison pools both dose levels, but the breakdown differs sharply: all three breakthrough cases occurred in the 16 participants who received two doses at 10⁶ CFU. Among the 18 who received the reduced 5×10⁵ CFU dose, there were none. The 13 participants who received a single 10⁶ CFU dose also had none. Those group-specific figures are exploratory, underpowered, and not the basis of any formal claim. They are the reason the authors recommend the lower dose for further development.
They also, in this writer’s assessment, make the pooled 89% a weaker summary of the trial than it appears. The figure averages across one dose that produced three failures in 16 participants and another that produced none in 18 — on numbers far too small to establish whether that difference is real or noise. Either the doses performed differently, in which case pooling them describes a regimen nobody plans to use, or they did not, in which case the safety amendment that split them was the more consequential event of the trial. The paper reports all of this plainly; it does not draw this conclusion. The conclusion is ours.
The immunology adds a further wrinkle: antibody titres peaked around two weeks after the first dose and did not rise again after the second, which the authors read as no clear benefit from a second dose in this model.
Why Shigella is a hard target
Shigellosis spreads through faecal contamination of food, water and surfaces, and through sexual contact, producing inflammatory diarrhoea, fever and abdominal cramps. Its burden is concentrated in low- and middle-income countries, where children aged one to five account for most cases and most deaths.
Species distribution complicates deployment. S. sonnei, which WRSs2 targets, predominates in high-income countries; S. flexneri predominates where mortality is highest. The authors note that a multivalent candidate covering S. sonnei alongside S. flexneri 2a, 3a and 6 would be needed for global use.
Resistance is the sharper argument. The US Centers for Disease Control and Prevention estimates about 450,000 US infections annually, roughly 242,000 of them antimicrobial-resistant. More striking is the trend: CDC surveillance published in 2026 found that extensively drug-resistant isolates — resistant to ampicillin, azithromycin, ceftriaxone, ciprofloxacin and trimethoprim-sulfamethoxazole — rose from none during 2011–2015 to 8.5% of isolates in 2023, with no FDA-approved oral antibiotic available to treat them. The XDR share was roughly twice as high among S. flexneri isolates as across surveillance overall, which is the species a S. sonnei vaccine does not cover.
Efficacy is not effectiveness
The trial’s authors are candid about how far the result can be read. Live oral Shigella candidates have a record of failing outside high-income settings — the live-attenuated SC602 showed protection in American adults but poor shedding and weak immunogenicity when tested in Bangladesh. Differences in gut microbiota, nutrition, prior exposure, co-infection and environmental enteropathy all plausibly contribute. The controlled infection model, the authors write, cannot capture that heterogeneity, and the vaccine has not been tested in children or in any endemic population.
Two further limitations deserve mention. The 28-day gap between vaccination and challenge means nothing can be said about how long protection lasts. And one placebo recipient tested positive for the vaccine strain two weeks after their second placebo dose — the authors could not determine whether this reflected a laboratory artefact or genuine person-to-person transmission of a live vaccine organism, and flag it as a caution for interpreting the shedding data.
Against those caveats sits a practical advantage. An oral vaccine requires no trained injector, no sterile equipment and less cold chain infrastructure than a parenteral one, which the authors argue makes it more deployable in exactly the resource-limited settings where the disease burden is concentrated — if it works there.
What comes next
The authors recommend the 5×10⁵ CFU dose for further evaluation and call for trials in endemic populations, including young children, alongside work on durability, cross-protection against other serotypes and co-administration with routine paediatric vaccines. Dose and schedule will likely need further optimisation in children.
What the trial establishes is a proof of principle: gut immunity to Shigella can be induced orally, and the protection can be large. What it does not establish is that the same will hold in a Bangladeshi toddler with a different microbiome, repeated natural exposure and no inpatient ward. That gap is the next several years of work.