An H5N1 case in Vietnam was found only in spinal fluid — and nowhere else

Almost everything known about how H5N1 presents in people starts from the lungs. A case report published on 7 May describes one that did not [s1].

In mid-April 2025, a previously healthy 8-year-old boy from Tay Ninh province in Vietnam, on the Cambodian border, was admitted to a local hospital with 24 hours of fever, severe headache and vomiting — and no respiratory symptoms at all [s1]. A rapid dengue NS1 test was negative [s1]. He was treated as sepsis or meningoencephalitis and transferred two days later to Children's Hospital 1, a tertiary referral centre in Ho Chi Minh City [s1].

There he had a temperature of 38.1°C, neck stiffness and altered consciousness, with normal heart and respiratory rates, normal oxygen saturation on room air and normal lung auscultation [s1]. His chest radiograph showed a consolidation in the left lower lobe, but he had no respiratory complaints [s1].

How the virus was found

Cerebrospinal fluid taken on day 3 of illness looked, on paper, like bacterial meningitis: 486 cells per microlitre with 77% neutrophils, glucose of 1.1 mmol/L against a normal range of 2.8 to 4.4, a CSF-to-plasma glucose ratio of 0.16 where 0.6 or above is normal, lactate of 10.4 mmol/L and protein of 3.6 g/L [s1]. Brain MRI showed dilated lateral ventricles measuring 14 mm, against a normal ceiling of 10 mm [s1]. Routine bacterial cultures of CSF, urine, blood and endotracheal aspirate were all negative [s1].

What found the virus was a multiplex PCR panel screening more than 60 pathogens, run on that admission CSF sample. It returned influenza A at a cycle threshold of 19 [s1]. Confirmatory testing of a second CSF sample on day 6, using US CDC assays, returned influenza A at Ct 26 and A/H5 at Ct 34 [s1].

Then the part that makes this case unusual. PCR for influenza A was negative in urine, throat swabs, rectal swabs, endotracheal aspirate and blood — while serial CSF samples stayed positive through day 10 of hospitalisation [s1]. Previously reported H5N1 cases with central nervous system involvement have had viral RNA detectable in both CSF and non-CSF specimens, and CNS disease has generally followed respiratory symptoms rather than appearing without them [s1].

The authors offer the caveats themselves: low respiratory-tract viral load, transient replication in the airway, or simply late sampling — respiratory specimens were collected from illness day 6 onward — could each explain the negative results outside the CSF [s1].

The antibody evidence

The team measured IgG against the H5 haemagglutinin HA1 subunit in paired CSF and plasma samples across the illness. In CSF, mean fluorescence intensity rose from a borderline level to 2,542 and then 29,286; in plasma, from undetectable to 1,363 and then 8,723 [s1]. Antibody levels were higher in spinal fluid than in blood [s1].

That pattern is consistent with antibody being produced inside the central nervous system rather than leaking in from the circulation — which the authors read as evidence of genuine viral invasion of the CNS [s1]. Intrathecal antibody production has been described after seasonal influenza infection before [s1]. Negative-control CSF and serum samples showed no anti-H5 haemagglutinin antibody [s1].

Which virus this was

Direct sequencing of the second CSF sample recovered all eight influenza genome segments [s1]. The haemagglutinin gene placed it in clade 2.3.2.1e — previously designated clade 2.3.2.1c — of highly pathogenic H5N1 viruses that circulate endemically in Southeast Asia [s1].

More precisely, this is a novel reassortant: a virus carrying gene segments from both clade 2.3.2.1c and clade 2.3.4.4b lineages, which emerged in late 2023 and has since caused poultry outbreaks and zoonotic infections in mammals, including 14 confirmed human cases with 6 deaths in Cambodia and one fatal human case in Vietnam [s1]. All the remaining segments in this boy's virus were closely related to that reassortant genotype [s1].

The sequences carried several amino acid substitutions previously associated with mammalian adaptation. Two of them in haemagglutinin — S123P and R167K, both linked to increased binding to α2,6 receptors, the receptor type that dominates the human upper airway — were unique to this virus among those compared [s1]. The remaining mutations had been reported before [s1].

That finding deserves a hard limit around it. Two substitutions in a single isolate, described in a case report, are a flag for surveillance, not a measurement of transmissibility. Nothing in this report tests whether the virus spreads between people, and its own phylogeographic analysis was constrained by a lack of contemporary poultry sequences from Vietnam [s1].

The exposure

Once the A/H5 result came back, epidemiological investigation found that the family kept many young fighting cocks, which the boy treated as pets and handled frequently [s1]. Some had died of unknown causes about two weeks before he fell ill, and neighbours reported groups of 40 to 50 chickens dying around the same time [s1]. No poultry samples were available for testing [s1].

Both parents tested negative for influenza A by PCR on throat swabs, and negative for anti-HA1 antibodies in plasma [s1].

Course and outcome

The boy was mechanically ventilated 15 hours after arrival at the referral hospital as his coma deepened and cerebral oedema was suspected, then deteriorated further, requiring vasopressor and inotropic support [s1]. He received mannitol and hypertonic saline for raised intracranial pressure [s1]. Once the diagnosis was established, acyclovir was stopped and oseltamivir started on hospital day 3, followed by intravenous immunoglobulin [s1].

He was extubated on hospital day 8 [s1]. CSF remained abnormal until illness day 18, at which point influenza A RNA was no longer detectable [s1]. He was discharged after 19 days with full recovery [s1].

What it changes

This is a single case. It cannot establish how often H5N1 presents this way, and the authors' own framing is a clinical alert rather than an epidemiological claim: clinicians should be aware that clade 2.3.2.1e H5N1 can cause meningoencephalitis [s1].

The operational implication is narrower and sharper than the headline. A child with encephalitis, poultry contact and negative respiratory swabs would, under most testing algorithms, never be tested for H5N1 — because the specimen that carried the virus here was the only one that would have been positive. In a region where clade 2.3.2.1e is circulating in poultry, that is a detectable gap in case ascertainment, and it points in one direction: the count of human infections in Southeast Asia is a floor, not a total.

This article is informational and does not constitute medical advice.

Sources

• [s1] Nguyen PNT, Hung NT, Minh NNQ, et al., "Central Nervous System Involvement by Novel Clade 2.3.2.1e H5N1 Avian Influenza Virus in a Pediatric Patient," Open Forum Infectious Diseases, published online 7 May 2026. https://doi.org/10.1093/ofid/ofag283