An oseltamivir-resistant H5N1 turned up on farms with no drug pressure
A bird-flu virus from British Columbia poultry carried the H275Y mutation that blunts Tamiflu and peramivir. In the lab it stayed vulnerable to zanamivir and baloxavir — and was less virulent in mice.
The antiviral drugs that would be used against a human H5N1 case are few, and the most stockpiled of them is oseltamivir — Tamiflu. So a report that an H5N1 virus already carrying oseltamivir resistance turned up on poultry farms, without anyone having used the drug on those birds, is worth reading carefully [s1]. It is not evidence of an imminent human threat. It is a reminder that resistance can appear on its own, and that surveillance is what catches it.
What was found
Researchers characterised a highly pathogenic avian influenza A(H5N1) virus of clade 2.3.4.4b, genotype D1.1, isolated from farms in British Columbia during the fall 2024 outbreak [s1]. They compared a wild-type isolate (labelled BC-H5N1-WT) with a variant carrying the H275Y substitution in the neuraminidase protein (BC-H5N1-H275Y) [s1]. H275Y is the classic marker of resistance to the neuraminidase-inhibitor class of flu antivirals.
In laboratory susceptibility testing, the wild-type virus was susceptible to all four antivirals tested [s1]. The H275Y variant was resistant to oseltamivir (Tamiflu) and peramivir — two drugs in the same class — but remained susceptible to zanamivir (an inhaled neuraminidase inhibitor) and to baloxavir acid, the active form of baloxavir, which works by a different mechanism entirely [s1]. That split is the practically important part: a virus that shrugs off Tamiflu is not necessarily a virus that shrugs off every antiviral.
The trade-off: resistance came at a cost
Resistance mutations often carry a fitness penalty, and this one did. The wild-type virus replicated to significantly higher titres than the resistant variant at every time point tested in cell culture, and produced larger viral plaques — both signs of a fitter virus [s1]. In mice, the wild-type was more virulent, with a median lethal dose (LD50) of 1.78 × 10³ plaque-forming units, compared with 8.71 × 10⁴ for the H275Y variant [s1]. A higher LD50 means it took roughly fifty times more of the resistant virus to reach the same lethal effect — so the resistant variant was substantially weaker in this model. The wild-type also reached higher titres in the lungs and other organs [s1].
This is the reassuring half of the story, but it should not be over-read. Fitness costs can be offset by additional, compensatory mutations over time, which is exactly how drug-resistant seasonal flu strains have occasionally spread in the past. The authors' point is not that this particular variant is dangerous, but that it emerged in the absence of any antiviral selection pressure on the farm — meaning surveillance cannot assume resistance will only show up where drugs are being used [s1].
How this fits the human picture
The human side of clade 2.3.4.4b has been reassuring so far on drugs. A separate analysis found that since October 2024, 55 human A(H5N1) cases were reported in the United States, and sequencing of 46 viruses across genotypes B3.13, D1.1 and D1.3 found that, with two exceptions, the viruses were assessed as susceptible to the available antivirals [s2]. Of those two, one D1.1 virus carried an M2-S31N substitution conferring resistance to the older adamantane class, and one B3.13 virus had a PA-I38M substitution that reduced baloxavir susceptibility about 17-fold in cell culture [s2]. In other words, reduced susceptibility has been rare in human cases, and no single mutation has knocked out every treatment option at once [s2].
What it means
For readers, the takeaways are narrow and specific. First, this is a poultry-surveillance and laboratory study — it does not describe illness in people, and it does not change the current, low assessed risk of H5N1 to the general public. Second, oseltamivir resistance in an avian H5N1 virus is not the same as oseltamivir resistance spreading in humans; the resistant variant here was measurably less fit [s1]. Third, and most usefully, it underlines why the antiviral stockpile should not rest on one drug: because H275Y left zanamivir and baloxavir effective in the lab, having more than one class available is what preserves a treatment option if resistance to any single class appears [s1].
None of this warrants alarm, and it certainly does not warrant self-medicating with flu antivirals, which are prescription drugs whose benefit depends on the specific virus and timing. What it warrants is confidence in the boring, unglamorous work that produced it — routine sequencing of outbreak viruses, and laboratory testing of what those viruses are and are not susceptible to [s1][s2]. That is the system working as intended: spotting a resistant variant early, while it is still confined to birds and still carries a fitness cost.
Sources
- [s1] Characterization of oseltamivir-resistant A(H5N1) clade 2.3.4.4b, genotype D1.1 variants identified in poultry farms of British Columbia, Canada. Emerging Microbes & Infections, 8 Jul 2026. https://doi.org/10.1080/22221751.2026.2686474
- [s2] Antiviral susceptibility of clade 2.3.4.4b highly pathogenic avian influenza A(H5N1) viruses from human cases. Emerging Microbes & Infections, 23 Dec 2025. https://doi.org/10.1080/22221751.2025.2601372
Sources
- Characterization of oseltamivir-resistant A(H5N1) clade 2.3.4.4b, genotype D1.1 variants identified in poultry farms of British Columbia, Canada — Emerging Microbes & Infections , July 8, 2026
- Antiviral susceptibility of clade 2.3.4.4b highly pathogenic avian influenza A(H5N1) viruses from human cases — Emerging Microbes & Infections , December 23, 2025
More on
An H5N1 case in Vietnam was found only in spinal fluid — and nowhere else
An 8-year-old boy had brain inflammation with no respiratory symptoms. Every respiratory, blood and rectal sample tested negative; the virus was in his cerebrospinal fluid alone.
Tamiflu and Paxlovid: when antivirals actually help, and by how much
The two most-prescribed respiratory antivirals sit at opposite ends of the evidence. One shortens flu by less than a day on average; the other cut serious outcomes sharply — but only in the group at real risk.
How worried should people be about H5N1 bird flu? What the risk assessment says
The honest position is two things at once: the current risk to the general public is low, and the reason it is watched so closely is that its worst-case potential is high. The gap between the two is the whole story.
Seven of Cambodia's eleven H5N1 cases this year were reported in June alone
WHO has logged 11 laboratory-confirmed human A(H5N1) infections in Cambodia between 1 January and 1 July, including six deaths. Every case had handled or culled sick poultry.