Human surveillance and phylogeny of highly pathogenic avian influenza A(H5N8) during an outbreak in poultry in South Africa, 2017.

Abstract Background In June 2017, an outbreak of the highly pathogenic avian influenza A(H5N8) was detected in commercial poultry farms in South Africa, which rapidly spread to all nine South African provinces. Objectives We conducted active surveillance for the transmission of influenza A(H5N8) to humans working with infected birds during the South African outbreak. Methods Influenza A(H5N8)‐positive veterinary specimens were used to evaluate the ability of real‐time PCR‐based assays to detect contemporary avian influenza A(H5N8) strains. Whole genome sequences were generated from these specimens by next‐generation sequencing for phylogenetic characterization and screening for mammalian‐adaptive mutations. Results Human respiratory samples from 74 individuals meeting our case definition, all tested negative for avian influenza A(H5) by real‐time PCR, but 2 (3%) were positive for human influenza A(H3N2). 54% (40/74) reported wearing personal protective equipment including overalls, boots, gloves, masks, and goggles. 94% (59/63) of veterinary specimens positive for H5N8 were detected on an influenza A(H5) assay for human diagnostics. A commercial H5N8 assay detected H5 in only 6% (3/48) and N8 in 92% (44/48). Thirteen (13/25; 52%) A(H5N8) genomes generated from veterinary specimens clustered in a single monophyletic clade. These sequences contained the NS (P42S) and PB2 (L89V) mutations noted as markers of mammalian adaptation. Conclusions Diagnostic assays were able to detect and characterize influenza A(H5N8) viruses, but poor performance is reported for a commercial assay. Absence of influenza A(H5N8) in humans with occupational exposure and no clear impression of molecular adaptation for mammalian infection suggest that this avian pathogen continues to be low‐risk human pathogen.


| INTRODUC TI ON
In June 2017, an outbreak of the highly pathogenic avian influenza (HPAI), A(H5N8) was detected on commercial poultry farms in the Mpumalanga Province of South Africa and reported to the World Organization for Animal Health (OIE). 1 This followed soon after reported outbreaks in neighboring Zimbabwe. 2 Over the months that followed, the virus rapidly spread across the country to all nine South African provinces resulting in the death and culling of millions of commercially farmed birds as well as mortalities in several species of wild birds. 1 People that are in close contact with infected birds or carcasses are regarded as being at potentially elevated risk of acquiring avian influenza (AI) as the virus may be transmitted through infectious secretions and excretions. [3][4][5][6] While thousands of people worldwide working in close contact with infected birds have been exposed to influenza A(H5N8), no human infections have been reported to date. [6][7][8] This suggests the risk of human infection is low. However, the risk in South Africa may differ when compared to other countries as a result of an HIV prevalence of 18.8% within the 15-49 year old population. 9 It should be noted that influenza A(H5N8) has been shown to infect and be mildly pathogenic in ferrets and mice. 5,10,11 Mutations facilitating adaptation of avian influenza viruses for the infection of mammalian hosts consist of individual or limited sets of point mutations in specific gene segments like PB2 (L89V, E627K, D701N) and hemagglutinin (HA) (A149V) and NS1 (P42S). [11][12][13][14][15][16][17][18][19][20][21][22][23] Thus, while influenza A(H5N8) poses limited zoonotic transmission risk, given its evolutionary history and relatively simple genetic adaptations required for potential mammalian infection, it could pose a potential pandemic risk.
The one health concept recognizes that human, animal, and environmental health is interlinked and encourages close collaboration between respective health authorities to work toward achieving health for all. 24 Close interaction and possible collaboration between human and animal health authorities during influenza surveillance plays an important role in ensuring that both sectors are aware and prepared to detect, respond, and control potential zoonotic influenza viruses. In this study, we conducted active surveillance for the transmission of influenza A(H5N8) to humans working with infected birds during the 2017 outbreak in South Africa. We also evaluated the ability of 2 real-time PCRbased assays to detect avian influenza A(H5N8) strains that circulated in birds during the outbreak period and characterized them by genome sequencing for known adaptive mutations that could augment host range and virulence.

| Human surveillance for potential zoonotic transmission
The Outbreak Response Unit (ORU) and Centre for Respiratory Diseases and Meningitis (CRDM) of the National Institute for Communicable Diseases (NICD) launched active epidemiological and laboratory investigations to screen in-contact workers and animal health personnel for influenza A(H5N8) viruses. A case under investigation was defined as a person who presented with any one or combination of symptoms including cough, fever, sore throat, runny nose, difficulty breathing, or conjunctivitis while also having a documented history of exposure (direct contact or proximity of <15 m) to potentially infected birds (alive or dead) or having worked in a poultry house with potentially infected birds, in the 10 days preceding the onset of symptoms. Active surveillance was conducted on three affected commercial poultry farms located in Mpumalanga and Gauteng Provinces. Demographic and clinical data, as well as information on personal protective equipment use and hand hygiene practices, were collected by an interviewer using a case investigation form (CIF). Oropharyngeal or combined oropharyngeal and nasal swabs were placed in Universal Transport Medium (UTM) (Copan Italia, Brescia, Italy) and were transported to the NICD within 24 hours of collection for testing.
In addition, passive surveillance included samples from animal health personnel meeting the same case definition and who were involved in outbreak response activities on multiple A(H5N8)affected farms in the Western Cape Province (WCP) and from workers from a bird park with laboratory-confirmed A(H5N8)-positive birds in Gauteng Province were submitted to the CRDM laboratory after completion of the CIF. Specimens from patients meeting the case-under-investigation definition referred from private and state pathology laboratories were also included.

| Personal protective equipment (PPE) recommendations
The NICD recommended that all people that work in close contact with poultry should wear appropriate PPE when handling potentially infected birds, carcasses, contaminated material, or when cleaning equipment and production houses in which infected poultry were kept. Recommended PPE to be worn in addition to normal overalls and gumboots included disposable overalls, gloves, protective eyewear, of molecular adaptation for mammalian infection suggest that this avian pathogen continues to be low-risk human pathogen.

K E Y W O R D S
avian influenza, H5N8, South Africa and masks capable of preventing inhalation of aerosolized virus particles. Handwashing with a disinfectant soap after handling of any potential contaminated material was also advocated. 25 The NICD also advised that persons exposed to infected poultry or their products should be followed up for 7-10 days to identify influenza symptoms, in which case samples should be collected and submitted to the NICD.

| Testing of avian samples for influenza A(H5N8)
Avian specimens were initially screened by the Western Cape Department of Agriculture for avian influenza A(H5N8). Total nucleic acid were extracted from specimens using the QIAcube automated nucleic acid extraction platform using the QIAcube HT kit (Qiagen), and AI virus infection was detected using the VetMAX™-Gold AI virus detection Kit (Life Technologies). Influenza virus H5 hemagglutinin and N8 neuraminidase subtype identities were determined by real-time PCR using previously published methods. 28,29

| Real-time PCR assays for avian influenza diagnostics in humans
Two assays were evaluated for their ability to detect A(H5N8) strains in circulation during the outbreak period: (a) The CDC influenza typing (A/B) (FluRUO-01) assay combined with assays for H1pdm09 and H3, H5, and H7 (FluRUO-09 and FluRUO-08) subtyping (International reagent resource (IRR)), and (b) the commercial FluHunter A(H5N8) kit (Genekam). Assays were conducted as described and according to manufacturer's instructions. 26,27 The FluHunter A(H5N8) kit detects both the H5 and N8 subtypes, using a single primer-probe set for each target. Congruence of results between the assays was determined by calculating Cohen's Kappa coefficient. 30

| AI A(H5N8) genome amplification and sequencing
For pandemic preparedness, we assessed if procedures routinely used for human influenza A virus characterization can be used to characterize AI A(H5N8) viruses. Briefly, nucleic acid extracts were treated with RNase-free DNase I (New England Biolabs) to enrich for RNA at 37°C for 30 minutes followed by heat inactivation after addition of 5 µL of 50 mmol/L EDTA. The eight genomic segments of influenza A viruses were simultaneously amplified using a onestep RT-PCR. 31,32 Briefly, enriched RNA served as template in PCR reaction mixture which combined 0.2 μmol/L of each of the cUni-12; cUni12G; and cUni-13 primers with SuperScript III one-step RT-PCR system with Platinum Taq high-fidelity DNA polymerase system (Thermo Fisher). 31,32 The temperature cycling profile parameters were 42°C for 60 minutes, 94°C for 2 minutes, followed by five cycles of (94°C for 30 seconds, 45°C for 30 seconds, and 68°C for 3 minutes), followed by 31 cycles of (94°C for 30 seconds, 57°C for 30 seconds, and 68°C for 3 minutes). PCR products were confirmed on a 1% agarose gel. PCR amplicons were enriched for viral templates by MspJI restriction enzyme (New England Biolabs) digestion at 37°C for 16 hours followed by incubation at 72°C for 20 minutes to inactivate the MspJI enzyme. 33 Specimens were then processed for next-generation sequencing on the Illumina MiSeq instrument.

| Phylogenetic and sequence analysis
Illumina sequencing library was prepared from amplified PCR products using Nextera XT Sample Preparation kit (Illumina). Briefly, PCR amplicons were tagmented enzymatically (55°C for 5 minutes), barcoded through PCR amplification, purified using AMPure XP beads dk/servi ces/NetNG lyc/) using default settings. HA amino acid numbering excluded the signal peptide sequence while NA amino acid numbering included the signal peptide sequence to conform to the methods of. 35

| Ethics
This study was conducted as part of the NICD outbreak response   sampled and interviewed individuals at the different establishments ( Figure S1). The majority (54%, 40/74) indicated that they wore overalls, boots, masks, gloves, and goggles. The 74 samples identified through human surveillance were collected between 7

| Human surveillance
and 36 days post-exposure and tested negative for AI A(H5) viruses. Two samples (3%) tested positive for seasonal human influenza A(H3N2) viruses (Table 1).

| Evaluation of influenza A(H5N8) diagnostic assays
Sixty-three influenza A(H5N8)-positive nucleic acid extracts and 32 HPAI-negative specimens were tested. All 63 AI A(H5N8)-positive veterinary specimens tested positive for influenza A on the CDC assay; 94% (59/63) tested positive with the CDC influenza A/H5 assay (   (Table S1). The accuracy of genome sequence data generated in our laboratory was determined by full-genome ML tree analysis; comparing with reference sequence data for 7 samples (S2017-08-0561_NICD, S2017-09-0065_NICD, S2017-08-0340_NICD, S2017-08-0274_NICD, S2017-08-0243_NICD, S2017-08-0581_NICD, and S2017-08-0336_NICD), which were provided by the OVI reference laboratory (Figure 1). All consensus sequence duplicates generated in our study and by OVI were identical.      As a result of farms being quarantined, an extended lag was observed between exposure of symptomatic individuals and sampling (7-36 days). It is therefore possible that potential infections by avian influenza may have resolved within this period. The lag time though was estimated from the first date of exposure, which is when the outbreak started on the farm. Subsequent to initial exposure, farm workers also assisted with the euthanasia and disposal of carcasses, which took place over several days. The workers therefore could potentially have been continually exposed throughout the culling operation, which may reduce the estimated lag period between exposure and sampling and limit the possibility of missing resolved infections.

| D ISCUSS I ON
In addition to active surveillance conducted at the three commercial poultry farms, surveillance guidelines were also provided to public health clinics within all affected areas. This allowed us to identify any potentially infected individuals through the public health system.  surveillance at the human-animal interfaces is important for the control or aversion of potential zoonotic and novel pandemic events.