Comment
Treating penicillin-susceptible Staphylococcus aureus bacteraemia: confusing the issue with facts
Annette C Westgeest, Vance G Fowler Jr
Lancet Available online 17 June 2026
https://doi.org/10.1016/S0140-6736(26)00857-3
Staphylococcus aureus is a leading cause of bacteraemia, responsible for over a million deaths annually. Despite its ubiquity, the quality of evidence upon which clinicians make treatment decisions on bacteraemia is surprisingly poor. As a result, there is currently no global agreement on many basic clinical questions involving S aureus.1 Penicillin-susceptible S aureus (PSSA) bacteraemia is an illustrative example of how dogma can sometimes outlive evidence. Penicillin resistance due to the blaZ gene (encoding the penicillinase that causes penicillin resistance) rapidly emerged after penicillin introduction and widespread use.2 This emergence, coupled with concerns about the reliability of penicillin-susceptibility testing,3 led to guideline recommendations to preferentially use semisynthetic anti-staphylococcal penicillins instead of benzylpenicillin for PSSA bacteraemia.4 As a result, anti-staphylococcal penicillins were the preferred therapy for PSSA for years—a practice unencumbered by supporting evidence from even a single randomised trial—until recent increases in the frequency of PSSA in many parts of the globe5 renewed interest in the use of benzylpenicillin. But although interest in the question of whether penicillin could be safely and effectively used to treat PSSA bacteraemia was there, the mechanism by which to answer it was not. At least, not before the Staphylococcus aureus Network Adaptive Platform (SNAP).
In The Lancet, the SNAP Trial Group6 answer this question.7 SNAP is an adaptive platform trial designed to simultaneously answer multiple therapeutic questions for S aureusbacteraemia in much the same way as the RECOVERY platform addressed COVID-19.8 Of 8546 patients admitted to participating hospitals with S aureus bacteraemia across eight countries, SNAP investigators included 2602 adult patients, divided into mutually exclusive prespecified silos according to the antibiotic susceptibility of the S aureusisolate (PSSA; penicillin-resistant, meticillin-susceptible S aureus [MSSA]; and meticillin-resistant S aureus [MRSA]). 493 (19%) of these patients had PSSA bacteraemia, of whom 281 were randomly assigned to receive open-label intravenous benzylpenicillin, or flucloxacillin or cloxacillin. Median age was 67 years, 87 (31%) were female, and 194 (69%) were male. The primary endpoint of the study was incontrovertible: all-cause mortality at 90 days. Frequent prespecified Bayesian updates were performed on accumulating data and prespecified criteria were used to define non-inferiority, superiority, and futility.
The study was discontinued by the data and safety monitoring committee in response to a safety signal for acute kidney injury (AKI) in patients in the flucloxacillin or cloxacillin group in both this trial and the simultaneously enrolling SNAP silo that compared cefazolin and anti-staphylococcal penicillins for MSSA. Despite this premature study closure, the investigators were able to make important observations. First, 90-day mortality was 14% in the benzylpenicillin group and 22% in the flucloxacillin or cloxacillin group (adjusted odds ratio [OR] 0·67, 95% credible interval [CrI] 0·35–1·28). This difference corresponded to a 96·1% posterior probability of non-inferiority and an 88·9% posterior probability of superiority of benzylpenicillin compared with flucloxacillin or cloxacillin. Patients in the benzylpenicillin group had a lower risk of AKI than those in the flucloxacillin or cloxacillin group (11% in the benzylpenicillin group vs22% in the flucloxacillin or cloxacillin group; adjusted OR 0·50, 95% CrI 0·26–0·94). This difference corresponded to a 99·8% posterior probability of non-inferiority and a 98·4% posterior probability of superiority of benzylpenicillin compared with flucloxacillin or cloxacillin for risk of AKI. Results were consistent across intention-to-treat and per-protocol analyses. The results were also consistent when patients lost to follow-up, who were excluded from the primary analysis, were analysed as having died by 90 days.
As the first results of the SNAP trial to be published, this study adds key contributions to the field. First, it shows that benzylpenicillin is likely to be at least non-inferior to flucloxacillin or cloxacillin in terms of 90-day mortality and is probably less nephrotoxic. Second, the results support the use of a combination of automated screening and confirmatory modern phenotypic penicillin-susceptibility testing to guide treatment decisions involving benzylpenicillin.
However, for several reasons, the study's long-term impact on clinical practice is uncertain. First, PSSA is still relatively infrequent. Only 493 (19%) of 2602 adults with bacteraemia had PSSA, and the proportion of PSSA might be even lower in regions with higher rates of meticillin resistance. Second, the phenotypic susceptibility tests necessary to accurately identify PSSA are not readily available in many hospitals worldwide, and their interpretation can be complex. Third, AKI was mild in most patients and is generally reversible. Finally, the study did not answer the most relevant question of benzylpenicillin versus cefazolin for PSSA. Cefazolin was recently shown by SNAP investigators and others9, 10 to exhibit similar efficacy to anti-staphylococcal penicillins for the treatment of MSSA, as well as improved safety, and is already the preferred antibiotic for the treatment of MSSA in the USA.1
Most importantly, however, the SNAP investigators showed the power of international collaboration. By creating an unprecedented international network focused on S aureus, SNAP has now enrolled more S aureus bacteraemia patients than in all previous randomised controlled trials combined. One country was conspicuously absent from the collaboration: the USA. Once again,11 the USA failed to contribute patients to a landmark global trial in antibacterial resistance, primarily due to the high cost of conducting clinical research there. This inability to participate in high-quality international trials reiterates a growing concern over the position of the USA as a global science superpower12 and underscores the National Academy of Medicine's call to “revitalize the US biomedical research enterprise”.13
S aureus is a global threat; it will require a global response. By providing a historic example of international cooperation and innovation, SNAP serves as a model for addressing unresolved questions for common infectious conditions.
