Virome Sequencing Detects H5N1 Avian Influenza in Wastewater in Ten Cities

To the Editor:

Highly pathogenic avian influenza viruses are virulent members of the H5 and H7 subtypes of Influenza A, cause systemic disease in avian species, and incur significant cost to agricultural production because of mass culling of infected animals.¹ Since 1955, most outbreaks have been caused by three lineages, with the recent widespread 2.3.4.4b clade evolving from a lineage that infected poultry. On March 25^th 2024, H5N1 2.3.4.4b was detected in Texas dairy cattle herds concomitantly with herds in Michigan and Kansas. The first human case of 2024 was detected shortly thereafter in Texas (March 28^th, 2024) in an individual with exposure to symptomatic cattle. As of July 28^th 2024, H5N1 has been detected in 14 human infections in three U.S. states (4 cattle associated and 10 poultry associated)², 171 dairy herds across 13 states, and 94 poultry flocks across 26 states.³ Another group has reported influenza with an H5 hemagglutinin gene present in wastewater beginning in March of 2024, and has subsequently identified H5 influenza at 25 sites across 9 states (https://data.wastewaterscan.org/).

Since May of 2022, the Texas Epidemic Public Health Institute (TEPHI) has been using hybrid-capture sequencing to test weekly wastewater samples in mostly urban areas in cities throughout Texas⁴ (Table S1), detecting over 400 human and animal viruses to date, several of which (e.g. SARS-CoV-2, Influenza, and Mpox) correlate to clinical case data.⁵ Seasonal influenza serotypes H3N2 and H1N1 are routinely detected in TEPHI wastewater samples, and levels have corresponded to clinical caseloads from May 2022 through the beginning of March 2024. Until that point, serotype H5N1 had not been detected (0 out of 1,337 wastewater samples). However, in samples from March 4^th through July 15^th, H5N1 is detected in 10 of 10 cities, 22 of 23 sites, and in 100 of 399 samples (Fig. 1B, C, Fig. S1). The abundance of H5N1 sequences has not correlated with influenza-related hospitalizations, which declined in Texas during the spring (Fig. 1B).

Figure 1: Influenza A H5N1 serotype is detectable in wastewater via virome sequencing.

(A) Dereplicated reads from wastewater virome sequencing assigned to H5N1 from all samples with validated H5N1 signal (see methods) aligned to a recently collected H5N1 2.3.4.4b reference sequence (GISAID: EPI_ISL_19027114). (B) Temporal chart (September 1^st, 2023 – July 15^th, 2024) measuring Influenza A serotype and abundance averaged across all 23 Texas wastewater treatment sites (colored bars) and hospital inpatient rates due to influenza across Texas as reported by Texas Department of Health and Human services (black line). Note that inpatient rates also include those caused by Influenza B, which is not measured in wastewater in this chart. (C) Bars represent proportion of influenza A serotypes in wastewater samples from each city. Grey line shows the 3-week moving average of influenza A relative abundance in wastewater, normalized to each city’s seasonal maximum. Note that sampling of South Texas City I sites began April 8^th, 2024.

Wastewater sequences were assigned to H5N1 using competitive mapping to an influenza genome database. Sequences were then manually validated and documented by three independent genomics researchers (see methods). Within the set of validated H5N1 sequencing reads, there are alignments to all eight genome segments including the PB2, HA/hemagglutinin and NA/neuraminidase genes covering > 85% of the genome (Fig. 1A). All sequencing reads best match H5N1 genomes from birds and mammals (including the human cases) collected since 2023 and are assigned to the 2.3.4.4b clade. Importantly, a glutamic acid, instead of lysine, in position 627 of the PB2 gene suggests that this virus has not extensively adapted to humans (see Table S2 which shows all unique SNPs).

The widespread detection of Influenza A H5N1 virus in wastewater from ten U.S. cities is concerning. Although the exact origin of the signal is currently unknown, lack of clinical burden along with genomic information suggests multiple animal inputs. Wastewater monitoring should be considered as a sentinel surveillance tool that augments our detection of evolutionary adaptations of concern. To aid efforts to identify the source and promote efforts to mitigate the next flu pandemic, we suggest expanded agnostic sequencing of wastewater, livestock and their products, exposed agricultural workers (including fecal/urine samples), and migrating birds along major flyways.