New antibiotics for resistant bacteria: what the trials actually measured

A number keeps circulating online: a new antibiotic called zosurabalpin "cured 90.1% of patients with a resistant superbug in a phase 3 trial." Go looking for that trial and you will not find it. What does exist is quieter and more useful — four drugs, three different bacteria, three very different levels of evidence. This article separates them, because in resistance the organism and the trial phase decide everything.

This is an information article, not medical advice. It does not replace a consultation with a doctor, and nothing here is a treatment recommendation.

Why there is no single "superbug antibiotic"

Resistance is one word for dozens of separate problems: a molecule that reaches one bacterium often cannot reach another at all.

Gepotidacin and zoliflodacin were developed against Neisseria gonorrhoeae, the cause of gonorrhoea. Zosurabalpin was built around a single target in carbapenem-resistant Acinetobacter baumannii, a hospital and intensive-care problem; sulbactam-durlobactam is the drug already licensed for that same organism. So "does the new superbug antibiotic work" has no answer. "Which organism, which phase, against what comparator" does.

Can a pill replace the injection for gonorrhoea?

Two oral drugs reached that question with completed phase 3 trials and United States approvals a day apart in December 2025.

EAGLE-1 was a phase 3, open-label, non-inferiority trial: oral gepotidacin against a single intramuscular dose of ceftriaxone plus oral azithromycin. The primary endpoint was microbiological response at the test-of-cure visit for urogenital N. gonorrhoeae — clearance in culture, not how people felt. Success occurred in 187 of 202 participants (92.6%, 95% CI 88.0–95.8) versus 186 of 204 (91.2%, 95% CI 86.4–94.7); the adjusted difference was −0.1 percentage points (95% CI −5.6 to 5.5), inside the trial's −10 point non-inferiority margin. About 628 people were randomised, with the primary analysis on the culture-confirmed subset. Gastrointestinal adverse events were more common with gepotidacin, and neither arm had treatment-related serious adverse events.

Zoliflodacin faced the same injection-plus-pill comparator in an international, open-label, phase 3 non-inferiority trial, 930 people randomised in the published analysis. Urogenital microbiological cure at day 6 ± 2 was 460 of 506 (90.9%, 95% CI 88.1–93.3) versus 229 of 238 (96.2%, 95% CI 92.9–98.3). The oral drug's figure is numerically lower than the comparator's and still inside the prespecified 12-point margin; in the evaluable population, after participants with an unknown test-of-cure result were set aside, cure was 96.8% versus 100%.

The approval record confirms the sequence. Gepotidacin (Blujepa) was cleared on 11 December 2025 for uncomplicated urogenital gonorrhoea in patients 12 and older weighing at least 45 kg who have limited or no other options; zoliflodacin (Nuzolvence) on 12 December 2025 for the same infection above a stated weight threshold.

So "a pill instead of a shot" means culture clearance about a week later, in urogenital infection, under narrow approved wording.

Does gepotidacin help when the usual UTI antibiotic fails?

Gepotidacin was licensed for uncomplicated urinary tract infection earlier, on 25 March 2025, on two double-blind phase 3 trials — EAGLE-2 and EAGLE-3, with 1531 and 1605 people randomised against nitrofurantoin. The primary endpoint at days 10–13 was therapeutic success: complete symptom resolution and microbiological success together.

EAGLE-2 reached it in 162 of 320 (50.6%) versus 135 of 287 (47.0%), a difference of 4.3 percentage points (95% CI −3.6 to 12.1). EAGLE-3: 162 of 277 (58.5%) versus 115 of 264 (43.6%), a difference of 14.6 points (95% CI 6.4 to 22.8). Diarrhoea was reported in 14% and 18% of those on gepotidacin, with no fatal events.

Two details there matter more than the gap. Absolute success was only about half to three-fifths of analysed patients in both arms, because the bar was "symptoms gone and bacteria below a threshold" rather than "felt better". And the primary analysis covered only nitrofurantoin-susceptible infections — the trial never tested the situation where nitrofurantoin had already failed, which is the situation most readers have in mind.

What about zosurabalpin — and that 90.1% figure?

Zosurabalpin is genuinely new chemistry. In Nature on 3 January 2024 it was described as a macrocyclic peptide that traps lipopolysaccharide inside its transporter, a mechanism no licensed antibiotic uses. In a neutropenic mouse pneumonia model with a pan-drug-resistant isolate, the highest dose tested cut lung bacteria by more than 5 log₁₀ colony-forming units.

Everything human in the registries is phase 1 — single intravenous doses in healthy adults, and in critically ill patients with bacterial infection. Both studies are completed, and both measured drug levels and tolerability, not cure. Conference posters have since reported pharmacokinetics in intensive-care patients and laboratory inhibitory concentrations across 1575 Acinetobacter isolates. A poster is not a peer-reviewed paper, and an inhibitory concentration is not a patient outcome.

The claim of a phase 3 trial with 90.1% clinical cure was not found in PubMed, in ClinicalTrials.gov, or in the manufacturer's 2024 development materials. The reposts also borrow a trial name that belongs elsewhere: EAGLE is the gepotidacin programme. Treat the figure as unverified — on the records checked this drug has no published human efficacy result and no approval anywhere.

So what is available for Acinetobacter today?

Sulbactam-durlobactam — a β-lactam plus a β-lactamase inhibitor, the established class rather than the new one — was approved in the United States on 23 May 2023 for hospital-acquired and ventilator-associated pneumonia caused by susceptible Acinetobacter baumannii-calcoaceticus complex in adults 18 and older.

In the ATTACK phase 3 trial the primary endpoint was 28-day all-cause death in confirmed carbapenem-resistant infection: 12 of 63 (19.0%) versus colistin 20 of 62 (32.3%), a difference of −13.2 percentage points with a 95% confidence interval from −30.0 to +3.5, against a +20 point non-inferiority margin. Both arms also received imipenem/cilastatin; investigators were unblinded and outcome assessors blinded.

That interval crosses zero: in 125 patients it is compatible with a large benefit and with a small harm. The trial supports "non-inferior to colistin", not a halving of deaths — and it is why a future zosurabalpin trial would have to match an outcome like 28-day mortality rather than a mouse colony count.

What these results do not show

• Nothing here speaks to the common sinus infection, bronchitis or skin infection most prescriptions are written for.

• Both gonorrhoea trials judged cure by culture at about a week, against a ceftriaxone-plus-azithromycin comparator that is no longer the only guideline regimen everywhere. Neither measured complications, infertility, or onward transmission.

• Both were open-label. Culture is objective, but missed visits still move the result: for zoliflodacin, counting missing tests as failures widened the gap (90.9% versus 96.2%) and setting them aside narrowed it (96.8% versus 100%).

• Zoliflodacin was not shown, with adequate power, to clear throat or rectal infection — those results were similar between arms but underpowered.

• Zosurabalpin is organism-selective. Even if human trials succeed, it would be an Acinetobacter drug, not a broad gram-negative antibiotic.

What to do with this

Nothing here identifies a drug to ask for by name — it gives you a way to read the next headline.

Keep "new" and "available" apart: gepotidacin and zoliflodacin carry approvals with deliberately narrow wording — specific infections, specific ages and weights, in one case only where other options are limited — and zosurabalpin carries none.

Treat a resistant infection as a microbiology question before a drug question. What can be used follows from the culture and susceptibility result for your own isolate, and that result, not an article, is what belongs at the appointment.

And when a figure like "90.1% cure" appears, four questions settle it: which organism, which phase, what comparator, over what period was cure measured. Most reposts answer none. If an antibiotic choice is on the table for you or a family member, that conversation belongs with the clinician holding your culture result.

Sources

• Ross JDC et al. Oral gepotidacin for uncomplicated urogenital gonorrhoea (EAGLE-1): phase 3, randomised, open-label, non-inferiority. Lancet. 2025;405:1608–1620. PMID 40245902 · registry record NCT04010539

• Wagenlehner F et al. Oral gepotidacin versus nitrofurantoin in uncomplicated urinary tract infection (EAGLE-2 and EAGLE-3). Lancet. 2024;403:741–755. PMID 38342126

• Luckey A et al. Zoliflodacin versus ceftriaxone plus azithromycin in uncomplicated urogenital gonorrhoea: international, randomised, open-label, phase 3, non-inferiority. Lancet. 2026;407:147–160. PMID 41391465 · registry record NCT03959527

• Zampaloni C et al. A novel antibiotic class targeting the lipopolysaccharide transporter. Nature. 2024;625:566–571. PMID 38172634

• Zosurabalpin (RO7223280), phase 1 safety and pharmacokinetics in healthy adults. NCT04605718

• Zosurabalpin (RO7223280), phase 1 pharmacokinetics in critically ill patients with bacterial infection. NCT05614895

• Gepotidacin (Blujepa), United States approval record, NDA 218230: original approval 25 March 2025, supplemental approval 11 December 2025. Drugs@FDA

• Zoliflodacin (Nuzolvence), United States approval record, NDA 219491: approval 12 December 2025. Drugs@FDA

• Sulbactam-durlobactam (Xacduro), United States approval record, NDA 216974: approval 23 May 2023. Drugs@FDA

• Kaye KS et al. ATTACK phase 3 trial, the source of the mortality figures above: Kaye KS, Shorr AF, Wunderink RG, et al. Efficacy and safety of sulbactam-durlobactam versus colistin for the treatment of patients with serious infections caused by Acinetobacter baumannii-calcoaceticus complex. Lancet Infect Dis. 2023;23(9):1072–1084. PMID 37182534 · doi:10.1016/S1473-3099(23)00184-6 · NCT03894046 · FDA Drug Trials Snapshot: XACDURO

No entry exists above for the circulating "phase 3, 90.1% cure" claim: no such record was found in PubMed, ClinicalTrials.gov, or the manufacturer's materials.