Pre-existing and cross-reactive immunity to avian influenza H5N1 in humans: Implications for pandemic risk and vaccine strategies Authors: Iván Sanz-Muñoz, Carlos J. Ciria-Gil, Marta Hernandez, Cesar Santiago, Ahmed M. Elsayed, Jose M. Eiros, Luis Martinez-Sobrido, Aitor Nogales Published: 2026 ABSTRACT SummaryDue to the continuous evolution of Influenza A viruses (IAVs), novel strains with efficient human-to-human transmission may emerge and cause future pandemics. Among these, highly pathogenic avian influenza (HPAI) H5N1 remains a major concern because of its impact on wildlife, livestock, and human health. The widespread circulation of H5N1 clade 2.3.4.4b, detected in hundreds of bird species and numerous mammals worldwide, highlights important changes in viral ecology and transmission, increasing its zoonotic and pandemic potential. This review summarizes current evidence on cross-reactive and cross-protective immunity to H5N1 in humans, focusing primarily on humoral immune responses. We examine the presence of pre-existing H5N1-reactive antibodies in individuals without known exposure and discuss how previous seasonal influenza infection or vaccination may contribute to their development. Particular attention is given to antibodies targeting conserved regions of hemagglutinin (HA), especially the stalk domain, as well as neuraminidase (NA), which may provide heterosubtypic protection. We also evaluate the ability of seasonal influenza vaccines and infections to induce cross-reactive responses against H5N1 and their potential role in partial protection or immune priming. Finally, we review current and emerging H5N1 vaccination strategies, including adjuvanted and mRNA-based platforms, and identify priorities for surveillance, population immunity assessment, and the development of broadly protective influenza vaccines. Introduction Influenza A virus Influenza A viruses (IAV) are zoonotic agents of major importance to both human and animal health. Primarily originated from avian reservoirs, they exhibit a remarkable evolutionary capacity, which facilitates the emergence of variants capable of infecting mammals and, in some cases, spreading to and among humans with epidemic or pandemic consequences1, 2, 3, 4, 5, 6. Their genetic diversity, broad host range, and segmented genome nature make them a constant threat, demanding integrated surveillance and control measures within a One Health framework1. IAV belong to the family Orthomyxoviridae and are classified according to their surface glycoproteins: hemagglutinin (HA) and neuraminidase (NA) (Fig. 1). To date, 18 HA subtypes (H1–H18) and 11 NA subtypes (N1–N11) have been described, whose combinations define IAV subtypes1, 2, 3, 4, 5, 6. In addition, a new candidate HA subtype (nominally H19) of IAV has been recently reported7. HA is the most abundant surface glycoprotein on IAV. Its main functions are to bind sialylated glycan receptors on the host cell, initiating viral entry, and mediating the fusion of the viral and host cell membranes during infection. HA is also the major antigen of influenza, driving immune responses. HA consists of a globular head (responsible for receptor binding) and a stalk region (important for membrane fusion); the stalk is highly conserved and a promising target for broadly neutralizing antibodies (NAbs) and universal vaccine development8 (Fig. 2). IAV HA subtypes are divided into two main phylogenetic groups: group 1 (H1, H2, H5, H6, H8, H9, H11, H12, H13, H16, H17, H18) and group 2 (H3, H4, H7, H10, H14, H15)9, 10. These two HA groups have diverged structurally and genetically, which influence in cross-reactivity and has significant implications for the design of broadly protective influenza universal vaccines10. Neuraminidase (NA) is the second major glycoprotein on the surface of IAV (Fig. 3). Its primary function is to cleave terminal sialic acids from glycoproteins on the host cell surface and newly formed virions, facilitating release and spread of progeny viruses11. NA also helps the virus penetrate mucus barriers and avoid trapping on host surfaces, contributing to viral fitness and infectivity. NA is antigenically variable and is targeted by immune responses and antiviral drugs11, 12, 13.Fig. 1Genomic organization and virion structure of IAV. (A) Schematic representation of the eight viral RNA (vRNA) segments of IAV and the major proteins encoded by each viral segment. Main (PB1, PA, M1, and NS1) and alternative reading frames or spliced products (PB1‑F2, PA‑X, M2, and NEP) are indicated below their corresponding segments. (B) Diagram of IAV virion showing the viral lipid envelope and the viral surface glycoproteins hemagglutinin (HA) neuraminidase (NA), and the ion channel protein M2. The internal viral structural proteins including the matrix protein M1, NEP, and the vRNA associated with the nucleoprotein (NP) and the trimeric RNA polymerase complex (PB2, PB1, PA) that made the ribonucleoprotein complexes (vRNPs) are indicated. Boxes at the end of each of the vRNAs indicate the 3′ and 5′ non-coding regions (NCR). PB1/PB1-F2, PA/PA-X, M1/M2 and NS1/NEP open reading frames (ORFs) are represented as blue boxes. The figure has been created/assembled with BioRender.com.Fig. 2Structural organization of HA from influenza A/Indonesia/5/2005 (H5N1). (A) Side view of the trimeric HA ectodomain in its prefusion conformation. Each monomer is shown in a different color (blue, pink, yellow). The HA1 and HA2 subunits are displayed, with the receptor‑binding site (RBS) highlighted in space‑filling representation. The approximate position of the viral membrane is indicated in beige, showing the location of the transmembrane region. The cytoplasmic tail is also indicated. (B) Top view of the HA trimer, illustrating the arrangement of the three HA1 globular heads (blue, pink, yellow) surrounding the central stalk formed by HA2. The sialic acid analog bound at the receptor-binding site (RBS) is shown as a space‑filling model. (UniProt: A8HWY8).Fig. 3Structural organization of NA from Influenza A/Thailand/460/2003 (H5N1). (A) Side view of the tetrameric NA showing the major structural domains: the globular head, the stalk region, the transmembrane domain (beige), and the cytoplasmic tail. Each monomer of the NA tetramer is displayed in a different color. (B) Top view of the NA head domain highlighting key catalytic and framework residues (magenta spheres), including R118, D151, R152, R224, E276, R292, R371, and Y406. These residues form the highly conserved active site essential for sialic‑acid cleavage (UniProt: Q6Q793). Genomic organization and virion structure of IAV. (A) Schematic representation of the eight viral RNA (vRNA) segments of IAV and the major proteins encoded by each viral segment. Main (PB1, PA, M1, and NS1) and alternative reading frames or spliced products (PB1‑F2, PA‑X, M2, and NEP) are indicated below their corresponding segments. (B) Diagram of IAV virion showing the viral lipid envelope and the viral surface glycoproteins hemagglutinin (HA) neuraminidase (NA), and the ion channel protein M2. The internal viral structural proteins including the matrix protein M1, NEP, and the vRNA associated with the nucleoprotein (NP) and the trimeric RNA polymerase complex (PB2, PB1, PA) that made the ribonucleoprotein complexes (vRNPs) are indicated. Boxes at the end of each of the vRNAs indicate the 3′ and 5′ non-coding regions (NCR). PB1/PB1-F2, PA/PA-X, M1/M2 and NS1/NEP open reading frames (ORFs) are represented as blue boxes. The figure has been created/assembled with BioRender.com. Structural organization of HA from influenza A/Indonesia/5/2005 (H5N1). (A) Side view of the trimeric HA ectodomain in its prefusion conformation. Each monomer is shown in a different color (blue, pink, yellow). The HA1 and HA2 subunits are displayed, with the receptor‑binding site (RBS) highlighted in space‑filling representation. The approximate position of the viral membrane is indicated in beige, showing the location of the transmembrane region. The cytoplasmic tail is also indicated. (B) Top view of the HA trimer, illustrating the arrangement of the three HA1 globular heads (blue, pink, yellow) surrounding the central stalk formed by HA2. The sialic acid analog bound at the receptor-binding site (RBS) is shown as a space‑filling model. (UniProt: A8HWY8). Structural organization of NA from Influenza A/Thailand/460/2003 (H5N1). (A) Side view of the tetrameric NA showing the major structural domains: the globular head, the stalk region, the transmembrane domain (beige), and the cytoplasmic tail. Each monomer of the NA tetramer is displayed in a different color. (B) Top view of the NA head domain highlighting key catalytic and framework residues (magenta spheres), including R118, D151, R152, R224, E276, R292, R371, and Y406. These residues form the highly conserved active site essential for sialic‑acid cleavage (UniProt: Q6Q793). Influenza antigenic variation occurs through two mechanisms: Antigenic drift results from the accumulation of point mutations in HA epitopes, driving the need for periodic or seasonal vaccine updates. Antigenic shift, typically mediated by viral reassortment, resulting in influenza viruses with novel surface antigens and pandemic potential in human populations lacking pre-existing immunity3, 4, 14. IAV is enveloped in a lipid membrane, and the virion size is approximately 80–120 nm in diameter, with projections of HA and NA (Fig. 1). The viral genome comprises eight segments of negative-sense RNA15, 16. The polymerase complex is composed of the individual polymerase basic 2 (PB2), polymerase basic 1 (PB1), and polymerase acidic (PA) proteins, and is responsible for viral genome replication and gene transcription15, 16, 17. PB1 also encodes the accessory protein PB1-F2, involved in cell death and pathogenicity (Fig. 1), and the PA segment encodes PA-X, which contributes to host gene shutoff and immune evasion18, 19 (Fig. 1). The viral nucleoprotein (NP) wraps the viral RNA and, together with the polymerase complex, forms the viral ribonucleoprotein (vRNP) complexes responsible for viral genome replication and gene transcription15, 16, 17. The segment 7 (or M) encodes the structural matrix protein 1 (M1), which shapes and stabilizes the virion, and matrix protein 2 (M2), an ion channel essential for virus entry and uncoating15, 16, 17. The segment 8 (or NS) encodes the non-structural protein 1 (NS1) that regulates host immune responses and supports viral replication, and the nuclear export protein (NEP) that mediates transport of viral ribonucleoprotein complexes out of the nucleus18, 19 (Fig. 1). Avian Influenza A Virus (H5N1) Molecular determinants, ecology, and evolution of H5N1 HPAI viruses Viral entry relies on HA binding to sialic acid containing receptors, with α2,3-linked receptors (avian) and α2,6-linked receptors (human) determining host tropism8. Ecologically, wild aquatic birds serve as the natural reservoir, typically sustaining subclinical enteric infections that spread along migratory flyways3. Avian influenza viruses (AIV) are classified into Highly Pathogenic Avian Influenza (HPAI) and Low Pathogenic Avian Influenza (LPAI)2, 3. HPAI viruses, typically of the H5 and H7 subtypes, contain a multibasic cleavage site in their HA protein, enabling widespread replication in host tissues and causing severe, often fatal, disease outbreaks with high mortality rates in domestic poultry and in wild birds. In contrast, LPAI viruses commonly produce mild or no clinical signs and are generally restricted to the respiratory and gastrointestinal tracts due to their monobasic cleavage site; however, certain LPAI viruses can mutate into highly pathogenic forms, underscoring the importance of surveillance2, 20. Both HPAI and LPAI can spread rapidly in bird populations, but only HPAI is linked to substantial economic losses and large-scale bird die-offs6, 21, 22, 23. Recent epizootics have demonstrated that HPAI viruses, such as H5N1 clade 2.3.4.4b, not only affect poultry but also induce unprecedented mortality between wild avian and even some mammalian species (Fig. 4). Furthermore, HPAI viruses exert a dual impact on human health, with high lethality rates such as those associated with H5N1 and H7N9, and on the economy, through losses caused by mass culling and trade restrictions6, 21, 22, 23, 24, 25. Control strategies rely on prompt outbreak management and rigorous biosecurity measures to limit spread and further mutation.Fig. 4Timeline of key events in the emergence, evolution, and spread of HPAI H5N1 viruses (1959–2025). Chronological overview of the history of HPAI H5N1. It begins with the first recognized outbreak in poultry in 1959 and continues with the re‑emergence of the virus in 1996–1997, when the first human cases and fatalities were reported in Hong Kong. Between 2003 and 2014, HPAI H5N1 expanded globally, and sporadic human infections were documented across multiple regions. The period from 2014 to 2016 marks the appearance of H5Nx viruses, including H5N2 and H5N8, within clade 2.3.4.4. Further diversification occurred between 2018 and 2020, when clade 2.3.4.4 evolved into several sublineages, notably the widely circulating 2.3.4.4b. From 2021 to 2025, HPAI H5N1 clade 2.3.4.4b spread extensively among wild birds, poultry, and several mammalian species, and sporadic zoonotic infections continued to be reported in humans. In 2024–2025, new HPAI H5N1 genotypes emerged in Canada and the United States, including infection of dairy cattle accompanied by human cases linked to livestock exposure. Human silhouettes indicate periods with reported human infections and fatalities, while animal icons represent the predominant host species involved at each stage. The figure has been created/assembled with BioRender.com. Timeline of key events in the emergence, evolution, and spread of HPAI H5N1 viruses (1959–2025). Chronological overview of the history of HPAI H5N1. It begins with the first recognized outbreak in poultry in 1959 and continues with the re‑emergence of the virus in 1996–1997, when the first human cases and fatalities were reported in Hong Kong. Between 2003 and 2014, HPAI H5N1 expanded globally, and sporadic human infections were documented across multiple regions. The period from 2014 to 2016 marks the appearance of H5Nx viruses, including H5N2 and H5N8, within clade 2.3.4.4. Further diversification occurred between 2018 and 2020, when clade 2.3.4.4 evolved into several sublineages, notably the widely circulating 2.3.4.4b. From 2021 to 2025, HPAI H5N1 clade 2.3.4.4b spread extensively among wild birds, poultry, and several mammalian species, and sporadic zoonotic infections continued to be reported in humans. In 2024–2025, new HPAI H5N1 genotypes emerged in Canada and the United States, including infection of dairy cattle accompanied by human cases linked to livestock exposure. Human silhouettes indicate periods with reported human infections and fatalities, while animal icons represent the predominant host species involved at each stage. The figure has been created/assembled with BioRender.com. Human exposure to AIV primarily occurs through direct contact with infected birds, contaminated secretions, surfaces, and/or unprocessed animal products6, 26. Although there is currently no evidence of sustained human-to-human transmission, the acquisition of genetic adaptations that enable efficient respiratory transmission among humans is the primary pandemic threat. These mutations, particularly in HA and polymerase genes, have been the focus of extensive research on host adaptation mechanisms, with implications for predicting and mitigating pandemic risks26, 27. Mammals such as pigs may act as “mixing vessels” that facilitate reassortment, while recent outbreaks in mustelids, canids, and cattle suggest an expanding host range2, 28, 29. HPAI H5N1 viruses were first identified in birds in the 20th century and gained particular attention following the 1997 outbreaks in Hong Kong, when the first severe human cases associated with this virus were documented (Fig. 4). Since then, HPAI H5N1 viruses have shown an extraordinary evolutionary capacity, giving rise to multiple lineages and clades with distinct geographic distribution, pathogenicity, and zoonotic potential6, 30. The “goose/Guangdong” (Gs/GD) lineage has served as the foundation of successive epizootic and zoonotic events, generating clades and subclades that remain in global circulation6, 26, 27, 30 (Fig. 4). Over the past decade, clade 2.3.4.4b HPAI H5N1 viruses have become predominant in both wild and domestic birds, exhibiting remarkable genetic plasticity and ecological expansion. This clade has achieved sustained intercontinental dissemination, with detections across Eurasia, Africa, and the Americas since 2021, and has demonstrated an unprecedented ability to infect mammals (Fig. 4). Notably, in 2024, widespread outbreaks of HPAI H5N1 clade 2.3.4.4b were reported in dairy cattle in the United States, with infections associated with reduced productivity, viral presence in raw milk and secretions, and transmission to exposed humans6, 30 (Fig. 4). This event marked a milestone in the epidemiology of HPAI H5N1 virus, highlighting its capacity to colonize atypical hosts and increase zoonotic risk. Zoonotic transmission and public health impact of H5N1 H5N1 human cases have been relatively limited compared with the magnitude of its circulation, yet they are often associated with severe clinical outcomes. Zoonotic transmission occurs primarily via direct exposure to infected birds, contaminated fomites, or virus-laden aerosols in agricultural or live-animal market settings. Since 2003, around 992 confirmed human infections have been reported globally, with approximately 476 deaths, resulting in a case fatality rate close to 50% as of November 202531. In the Americas alone, between 2022 and 2025, about 75 human cases have been documented, mainly in the United States, Canada, Mexico, Chile, and Ecuador, reflecting both sporadic zoonotic spillover and enhanced surveillance efforts. Since 2022, sporadic cases have been tied primarily to direct exposure to infected birds and, more recently, since 2024, to cattle6, 30, with no evidence of sustained transmission between humans (Fig. 4). These findings have prompted intensified surveillance and risk assessments by national and international agencies. Human H5N1 infection is characterized by high viral loads, dysregulated innate immune responses, and excessive production of pro-inflammatory cytokines (“cytokine storm”). In humans, HPAI H5N1 infection typically presents as a severe respiratory illness, including viral pneumonia, acute respiratory failure and, in some cases, extrapulmonary manifestations. Pathological studies have described diffuse alveolar damage, intense inflammation, and occasionally systemic viral dissemination32. In animals, the clinical spectrum is broad, ranging from subclinical infections in wild birds to mass mortality in poultry, as well as severe disease in experimental and domestic mammals33. The co-circulation of AIV H5N1 with seasonal H1N1 and H3N2 IAV in susceptible hosts raises the possibility of reassortment events that could generate variants with increased transmissibility in humans. The recent expansion of its host range, including bovine infections, has heightened concerns over a potential adaptive shift favoring human-to-human spread. In response to this scenario, control strategies are framed within the One Health program, integrating genomic and epidemiological surveillance across birds, mammals, and humans, along with biosecurity measures, outbreak control, and protection of exposed personnel. Pre-pandemic vaccines based on H5 HA have been developed for both humans and poultry, although their effectiveness depends on antigenic matching with circulating clades34. Antivirals remain the first-line therapeutic option. While several licensed antivirals remain effective against H5N1, the widespread resistance to M2 inhibitors, together with the emergence of mutations that reduce susceptibility to NA and polymerase inhibitors, underscores the need for new therapeutic strategies. Although a range of regionally approved drugs and investigational compounds have demonstrated promising results in vitro and in animal studies, supporting clinical evidence in humans is still limited35. Pandemic risk Although we are accustomed to viewing IAV as an epidemic disease that occurs every winter in temperate countries, from time to time, IAV has been responsible for pandemics of varying intensity in humans. These influenza pandemics occur sporadically, and to date, no mechanism has been identified that causes them to behave according to a specific temporal pattern4, 36. Understanding the historical patterns of influenza pandemics is essential for contextualizing the potential pandemic risk posed by HPAI H5N1. According to the latest scientific evidence, there are approximately 3 to 4 pandemics caused by different IAV every century. In the 20th century, for example, humanity suffered three pandemics caused by IAV: the H1N1 pandemic in 1918 (Spanish flu), the H2N2 pandemic in 1957 (Asian flu), and the H3N2 pandemic in 1968 (Hong Kong flu) (Fig. 5A). Additionally, at the beginning of the 21st century, humanity suffered another flu pandemic in 2009, caused by a pandemic H1N1 (pH1N1) subtype, which continues to circulate seasonally in humans along with the H3N2 subtype responsible for the 1968 pandemic4, 36 (Fig. 5B). On the other hand, additional information is available from historical and sero-epidemiological studies of influenza pandemics that occurred in the 19th century, such as those of 1889 and 184737. There is no definitive evidence regarding the IAV subtype that caused these pandemics, but according to some evidence, they may have been caused by H3N8 (in the case of 1889) and H1N8 (in the case of 1847). However, with regard to the 1889 pandemic, some authors have recently raised the debate that this pandemic may actually have been caused by a coronavirus, although there is no clear evidence to support this hypothesis38, 39, 40.Fig. 5Historical timeline of IAV pandemics and subsequent evolution of circulating seasonal lineages. (A) The five major IAV pandemics of the 20th and 21st centuries: 1918 (H1N1), 1957 (H2N2), 1968 (H3N2), 1977 (H1N1 re‑emergence), and 2009 (pH1N1). Representative schematic virion depictions for each pandemic virus. (B) Evolution and persistence of IAV subtypes following pandemic emergence: Clade 1 includes the 1918 H1N1 and its descendants, the 1957 H2N2 that replaced H1N1, the 1968 H3N2 that replaced H2N2, and the re‑emergent H1N1 strain detected in 1977. Clade 2 corresponds to the lineage established by the 1968 H3N2 pandemic virus, which has continued to circulate globally to the present day. The 2009 pandemic introduced the pH1N1 lineage, which subsequently replaced the previously circulating H1N1 strain. Together, the figure highlights the sequential replacement or coexistence of IAV lineages over more than a century of viral evolution. The figure has been created/assembled with BioRender.com. Historical timeline of IAV pandemics and subsequent evolution of circulating seasonal lineages. (A) The five major IAV pandemics of the 20th and 21st centuries: 1918 (H1N1), 1957 (H2N2), 1968 (H3N2), 1977 (H1N1 re‑emergence), and 2009 (pH1N1). Representative schematic virion depictions for each pandemic virus. (B) Evolution and persistence of IAV subtypes following pandemic emergence: Clade 1 includes the 1918 H1N1 and its descendants, the 1957 H2N2 that replaced H1N1, the 1968 H3N2 that replaced H2N2, and the re‑emergent H1N1 strain detected in 1977. Clade 2 corresponds to the lineage established by the 1968 H3N2 pandemic virus, which has continued to circulate globally to the present day. The 2009 pandemic introduced the pH1N1 lineage, which subsequently replaced the previously circulating H1N1 strain. Together, the figure highlights the sequential replacement or coexistence of IAV lineages over more than a century of viral evolution. The figure has been created/assembled with BioRender.com. The impact of all these influenza pandemics on humans has been very uneven. From the great Spanish flu pandemic in 1918, which some authors suggesting a mortality rate of around 2–3% with a total of 15 to 50 million deaths41, 42, to the most recent one caused by pH1N1 (approximately 550,000 deaths)43 very close to the mortality caused by seasonal flu each year44, 45, 46. What this dynamic of influenza behavior shows us is that influenza pandemics are inevitable events that cannot be predicted with the currently available surveillance methods, and that we will face more influenza pandemics in the future. In fact, Mr. Christopher Chadwick, Technical Officer for Influenza Preparedness and Response at the World Health Organization (WHO), commented in 2018 that "The next flu pandemic is a matter of “when”, not “if”47, reflecting that we will face influenza pandemics in the future, but we cannot precisely identify when it will happen. The recent COVID-19 pandemic in 2020 has also reminded us of the unpredictability of such global events. The appearance of cases of atypical pneumonia at the Wholesale market in the Chinese city of Wuhan alerted health authorities in 2019 to an outbreak of a new respiratory disease, but in just one month the virus had caused cases in countless cities in China and had spread to other countries in Southeast Asia and Europe48. Although there are many human coronaviruses closely related to SARS-CoV-2, the causative agent of the COVID-19 pandemic, it was largely unknown whether these had previously emerged as a pandemic, and what was known about them did not provide much useful information that could be used to predict the next one, as happened in 2019. The fact that human coronaviruses were considered minor viruses, mostly causing common colds, was probably one of the obstacles that hindered early detection and preparation for a crisis such as the coronavirus pandemic. However, in the case of influenza, the situation is completely different. Influenza has a robust and sophisticated global surveillance system that was established in 1957, the GISRS (Global Influenza Surveillance and Response System)49. This international surveillance network consists of more than 150 national reference laboratories49, as well as seven other collaborating laboratories (WHO CCs) located in Melbourne (Australia), Beijing (China), Tokyo (Japan), London (United Kingdom), Atlanta (USA), Memphis (USA), and Novosibirsk (Russia)49. This WHO network has been monitoring epidemic and pandemic influenza for more than 75 years and has led efforts to implement and maintain virological and epidemiological surveillance measures, as well as promoting prevention through the annual updating of seasonal influenza vaccines. This surveillance system analyzes information on both seasonal (human influenza) and avian influenza viruses. The existence of such networks has made it possible, since the discovery of the H5N1 virus linked to the current outbreak in 1996, to closely monitor suspected and confirmed cases in humans worldwide, as well as the number of deaths over the last 28 years. In addition, monitoring influenza outbreaks in wild and domestic birds has also made it possible to track the disease in animals. Although H5N1 human infections show a lethality rate close to 50%32, 33, 50, 51, several studies before and after the current avian influenza outbreak show that this lethality rate is not accurate, but would be much lower because asymptomatic, paucisymptomatic and mild infections are also common with H5N1 infections. For example, a meta-analysis conducted in 2020 that compiled results from 56 studies analyzing 35,159 individuals found that the seroprevalence of antibodies against H5N1 in China was between 2.4% and 7.3%, with higher rates in the central regions of the country (those that are more rural areas)52. Another meta-analysis also conducted in 2020 included 66 studies worldwide that analyzed poultry workers, poultry cullers and persons with both poultry and human case exposures, and found that the prevalence of protective antibodies was 0.2%, 0.6% and 1.8%, respectively53. This study also determined that the seroprevalence in the general population is practically 0%. Additionally, during the avian influenza outbreaks in the United States between 2021 and 2024, several studies demonstrated that H5N1 circulated quite frequently among poultry and dairy farm workers, as well as among veterinarians who treated these animals. In fact, a study conducted in 2024 showed that 8% of farm workers that worked on farms that had experienced outbreaks, had protective antibody titers (≥1:40) and 40% reported feeling ill with flu-like symptoms during the outbreak54. On the other hand, in the case of veterinarians who specifically managed dairy farms, antibodies were detected in 2% of them when serological surveys were conducted55. Since 2021, HPAI virus H5N1 of clade 2.3.4.4b has caused countless cases in wild and domestic birds, as well as in domestic and wild mammals, throughout the world56, 57, 58, 59, 60. Specifically, since then, there have been outbreaks in mammals in 19 countries51, demonstrating a clear adaptation of the virus to animals that were not previously reported to be infected and are phylogenetically closer to humans. Although there is currently no imminent risk of an H5N1 pandemic61, there are several facts that should at least make us alert. The first is the appearance of various mammalian adaptive mutations, such as E627K and D701N in PB262, 63, 64. In addition, the number of cases of HPAI H5N1 reported in the United States since the beginning of the outbreak in March of 2024 until February 2025 is unprecedented, all but three of which are related to wild and farm animals infected with the virus65. The virus is highly active, and species barriers are being crossed with increasing frequency, making it necessary to begin exploring protective measures that could be useful in the event of a pandemic. Moreover, although the current situation in the United States seems to slowed down, cases have been arisen during 2025 in other countries of the world, including Cambodia and Bangladesh, among others31, 66, 67. Cross-protection and pre-existing immunity: the key for pandemic preparedness before specific vaccines being available IAV pandemics represent some of the most disruptive events in the human history. Unlike seasonal epidemics, which arise from antigenic drift and the circulation of previously established strains, pandemics occur when a new IAV emerges bearing distinct surface antigens, or antigenic shift, against which the population lacks pre-existing immunity and the virus transmits efficiently. This scenario results in rapid chains of transmission, high morbidity, and profound socioeconomic impact on a global scale. The emergence of such new IAV in humans often involves avian or swine hosts as sources of new genomic segments, typically through co-infections and reassortment processes26, 27. The ability of the resulting IAV to transmit efficiently among humans depends on the compatibility of its cellular tropism, replication efficiency, and capacity to evade or modulate the host immune response26, 27. In this context, pre-existing immunity in humans become critically important. Although individuals generally lack NAbs against novel surface antigens, prior exposure to closely related IAV may confer a degree of partial protection68, 69, 70. This baseline immunity arises from several mechanisms. The first one is the heterosubtypic T-cell immunity. CD8 and CD4 T lymphocytes recognize conserved internal viral epitopes. Notably, previous studies have shown that the majority (>80%) of pre-existing CD4 and CD8 T-cell responses to IAV (e.g., H1N1 and H3N2) are directed against conserved core proteins such as NP, M, NS, PA, and PB proteins, which may mediate cross-reactive protection against different IAV71.These T cells can therefore cross-react with multiple IAV subtypes or strains due to the conservation of these internal antigens. Heterosubtypic T-cell immunity plays a critical role in limiting viral replication and reducing disease severity by killing infected cells and producing antiviral cytokines, even though it does not prevent viral transmission. Experimental models have shown that memory CD8 T cells can clear virus-infected cells more rapidly upon re-exposure to heterologous IAV subtypes, thereby reducing lung pathology and mortality72, 73. CD4 T cells also contribute by helping B cell responses and may exert direct cytolytic functions, further broadening protection across different IAV strains74, 75. In addition to internal antigens, T cells can also recognize epitopes derived from viral surface glycoproteins such as HA and NA, contributing to helper and cytotoxic responses that support antibody production and enhance the clearance of infected cells76, 77. A systematic analysis of human epitope conservation between HPAI H5N1 (considering the A/Anhui/1/2005 prototype) and seasonal IAV sequences revealed that 57% of CD4 and 63% of CD8 T-cell epitopes, respectively, are either identical or conserved to a level of sequence identity that is expected to be conducive to T-cell cross-reactivity. Overall, this analysis predicted that pre-existing T-cell immune memory would, to a large extent, cross-recognize HPAI H5N1 influenza viruses if widespread human infections start to occur78. The second one is antibodies targeting conserved epitopes. Aside from the highly variable HA head domain, antibodies targeting conserved epitopes on the HA stalk domain and relatively stable regions of NA can provide cross-protection across multiple IAV subtypes. These antibodies are generally no NAbs but mediate protection via effector functions such as antibody-dependent cellular cytotoxicity (ADCC) and phagocytosis. Their presence can blunt disease severity and partially prevent infection with novel IAV strains, making them valuable targets for universal influenza vaccine design79, 80, 81, 82. The third one is mucosal immunity and trained innate memory. The mucosal immune system, especially at the respiratory tract, constitutes the first line of defense against influenza infections. Components such as secretory IgA and resident innate immune cells provide a rapid but less specific response to viral entry. In the event of an avian flu pandemic, it is likely that the initial stages could be just as chaotic as they were in the case of the recent COVID-19 pandemic. Although the situation differs in that the virus is already known and at least 20 H5N1 vaccine candidates have been developed34, the initial availability of these vaccines is expected to be limited due to constraints in manufacturing capacity, production timelines, and distribution logistics83, 84. In addition, a search in ClinicalTrials.gov indicates that there are at least 179 studies analyzing H5 vaccines at some stage (https://clinicaltrials.gov/search?cond=h5n1%20vaccine). However, until vaccine production can adapt to global demand, it will be necessary to find mechanisms that can protect the population by other means, one of them being cross-protection. Cross-protection against AIV can occur either because of previous infections with other IAV, whether seasonal or avian, or through seasonal vaccination. A very clear example of how previous infections can protect against a potential pandemic can be found precisely in the history of these same events. Several studies have documented that people previously exposed to IAV with some kind of phylogenetic and antigenic relationship to the virus causing a pandemic have suffered lower mortality than other people who were not previously exposed. For example, in the 2009 pandemic caused by the H1N1, lower mortality was observed in people over 65 years of age compared to younger people and, especially, pregnant women85, 86, 87, 88, 89. Various studies, both in humans and animals, have shown that this lower mortality rate was due to previous exposure of elderly people to H1N1 viruses closely related to the 2009 subtype90, 91, 92, 93. A study published in 2018 showed that the genetic homology of the HA of the (H1N1)pdm09 subtype with various ancestral H1N1 viruses, such as A/South Carolina/1/1918 (Spanish flu), A/Puerto Rico/8/1934, or viruses from the 1940s, such as A/Weiss/1943 and A/ForthMonmouth/1947, was greater than 70%94. Furthermore, in this same study, the authors demonstrated that, before vaccination, individuals over 65 years of age had protective antibody titers against these strains in some cases in more than 38% of individuals. However, this protective effect of previous infections has also been observed in IAV pandemics before 2009. For example, in the 1968 Hong Kong flu pandemic, caused by H3N2, the risk of death in people over 75 years of age was similar to that of seasonal epidemics, despite being a completely new virus. The authors of this study hypothesized that previous exposure to viruses related to H3N2, such as H3N8, which theoretically caused the Russian flu pandemic in 1889, may have provided some kind of cross-protection that had a positive impact on survival in 1968. Another example is the Spanish flu pandemic of 1918. The highest death toll occurred in the youngest age group, especially among adolescents. A study published in 201437 showed that elderly people who were exposed to influenza H1N1 in 1918 may have been primed by the H1N8 subtype responsible of the flu pandemic reported in 184743. This same effect, but in reverse, is observed when evaluating the response to different subtypes of seasonal IAV in current populations. For example, the elderly population has higher mortality rates when H3N2 is circulating, while they respond better when (H1N1)pdm09 is circulating95, 96. This effect, documented in multiple seasonal epidemics since 2009, stems from the primary infections that elderly people currently facing these viruses had during their childhood. All of this is due to priming and its derived effect, Original Antigenic Sin (OAS). This concept is key to explaining age-related differences in susceptibility observed during IAV pandemic outbreaks97, 98, 99. There is abundant scientific literature demonstrating that early infections (priming) leave immunological footprints on individuals such that, when naturally exposed to microorganisms, they respond more readily to strains that primed them than to more current strains (OAS)100, 101, 102. This phenomenon, which has been documented not only for influenza but also for other infectious diseases, such as SARS-CoV-2 and respiratory syncytial virus (RSV)103, 104, among others, promotes a more powerful anamnestic response to ancestral epitopes and, therefore, when people are exposed to evolved epitopes, there is a biological resistance to generating new responses. This effect takes on another dimension when, in the case of influenza, there is a jump from one subtype to another that is completely different. In the case of IAV, viruses are divided into two different groups (group 1 and group 2) according to the genetic evolution of the HA protein. Group 1 includes such us H1, H2, H5 and H9, while group 2 includes others like H3 and H78, 10, 105, 106. Genetic homology, for example, between IAV with HA of the H1 and H2 subtypes, is quite high, whereas if we compare H1 and H3, it is considerably lower. This explains why priming during the early years of influenza pandemics has such an apparent impact on mortality rates when new IAV emerge. This is particularly relevant in the case of H5N1, as this virus has a fairly high homology in the HA protein with H1N1 and (H1N1)pdm09, between 40% and 50%107, which seems to justify many of the results regarding the cross-reactivity of antibodies against HPAI H5N1. In fact, a study published in 2016 showed that people born between 1918 and 1968, who were probably first infected by H1N1 or H2N2, both belonging to HA group 1, had greater cross-protection against H5N1 than those born later108. Studies conducted on animals have observed that exposure of ferrets to the (H1N1)pdm09 virus reduces susceptibility to H5N1 when healthy animals were placed in the same cages with infected animals109. Another study conducted on mice also showed that, after prior immunization with H1N1, survival after challenge with H5N1 was much higher (around 50%) than if they were immunized with H3N2 or influenza B virus110. In the same study, the authors also demonstrated that stimulation with a trivalent seasonal vaccine and subsequent immunization with a specific H5N1 vaccine produces ten times more avian subtype-specific IgG antibodies than stimulation with the H5N1 vaccine alone. However, it is important to note that seasonal influenza vaccines confer little to no protection against HPAI H5N1, as they are formulated to target circulating human subtypes70. Cross-reactive antibodies generated by seasonal vaccination may display some in vitro activity against H5N1, but clinical protection remains insufficient, highlighting the need for strain-specific or broadly protective vaccine platforms in the face of emerging pandemic threats4, 5, 64, 70. All these data show us that cross-reactivity between different influenza antigens are common, occur frequently both in natural exposure to the virus and in vaccinations, and can be very useful in protecting populations from pandemic threats, situations in which the availability of specific vaccines, at least initially, is not guaranteed. Moreover, these findings highlight how pre-existing immunity can significantly modulate the population-level impact of a novel IAV. From a pandemic preparedness perspective, understanding cross-immunity and immune memory is essential. Surveillance of HA stalk-specific antibodies, characterization of memory T-cell repertoires, and investigation of heterosubtypic immunity provide critical insights for the development of universal influenza vaccines, which aim to induce broad and long-lasting protection against multiple IAV subtypes. At the same time, these approaches may complement traditional seasonal vaccination strategies, whose effectiveness is limited by ongoing antigenic drift111, 112. Ultimately, the interplay between pre-existing immunity, viral transmissibility, and social factors can determine the actual dynamics of a pandemic. Understanding how cross-immunity modifies susceptibility, clinical severity, and transmission is key not only for risk assessment, but also for optimizing control measures, including vaccination campaigns, or antiviral development and distribution4, 5, 64. Cross-neutralizing antibodies against hemagglutinin and neuraminidase Induction of cross-neutralizing antibodies against hemagglutinin after infection with seasonal IAV While HA from H5N1 and seasonal IAV shares some structural conservation, cross-reactivity at the level of surface glycoproteins is limited. This is largely driven by differences in glycosylation, sequence variability within immunodominant regions, and receptor-binding preferences, all of which influence antigenicity8. HA, the most abundant viral surface protein, mediates host cell entry through its receptor-binding site (RBS), located in the globular head domain. The RBS is formed by the 130-loop, 150-loop, 190-helix, and 220-loop, and contains several conserved amino acid residues involved in sialic acid binding. However, structural variability across strains, including insertions or mutations within these loops, can significantly alter receptor engagement and antigenic properties. A key determinant of host specificity is receptor binding preference: avian viruses bind α2,3-linked sialic acids, whereas human-adapted strains preferentially bind α2,6-linked sialic acids8, 26. The acquisition of α2,6 specificity is critical for efficient transmission in humans and has been associated with specific mutations in the HA RBS, such as Q226L and loss of glycosylation at residue 158 in H5N18, 26. Notably, the mutations required for this switch differ across HA subtypes, highlighting the complexity of host adaptation. In addition, the pH threshold required for HA-mediated membrane fusion varies among strains and is typically lower in human-adapted viruses than in avian strains, reflecting increased stability in the human respiratory tract26, 27, 28. Together, these structural and functional differences constrain cross-reactivity between avian H5N1 and seasonal influenza viruses, while also defining key determinants of host range and transmissibility. Antibody responses to IAV infection or vaccination are dominated by epitopes in the HA globular head domain, with most antibodies targeting regions near the RBS and neutralizing the virus by blocking attachment to host cells113, 114, 115, 116. These HA head-specific antibodies are typically strain-specific and long-lived, but their effectiveness is reduced by antigenic drift in circulating viruses [30535315]. Although a subset of antibodies can target conserved residues within the RBS, the overall response remains focused on highly variable regions in HA117, 118, 119. In contrast, antibodies directed against the HA stem region are much less abundant but recognize more conserved epitopes across influenza subtypes. Importantly, while some stem-specific antibodies can directly neutralize the virus, many rely on Fc-mediated effector functions to confer protection in vivo. Although some HA stem-reactive antibodies can recognize both group 1 and group 2 IAV120, 121, most are group-specific due to structural differences between these groups. Recent studies have identified classes of broadly cross-group stem antibodies in individuals vaccinated against H5N1 or H7N9. Notably, H5N1 vaccination tends to elicit primarily group 1-specific stem antibodies, whereas H7N9 vaccination more frequently induces antibodies that cross-react with both HA groups120, 121. As previously stated, there is scientific evidence both locally and globally of the existence of people who have antibodies against avian influenza H5N1 without any record of confirmed exposure to the virus52, 53. However, as these are seroepidemiological surveys of people in environments and professions at risk of occupational exposure (farmers, veterinarians, healthcare workers who have treated infected people, etc.), contact with the virus may have been likely, especially if the case has been documented in regions where outbreaks have occurred in wild and/or domestic animals. However, other studies have analyzed the presence of antibodies that may be of heterotypic origin and have been identified in people who have no known contact with farm or wild animals, nor do they work in high-risk environments, both in children and adults. For example, in a study conducted in 2009, researchers found NAbs against H5N1 at titers ≥1/20 in 10 out of the 236 individuals tested, especially in the elderly122. Another study published in 2017 described hemagglutinating HA antibodies at protective titers in people living in Spain who were not exposed in 2.9% of individuals against H5N1, 1.2% against H7N3, and 25.9% against H9N2107. Other studies published in 2025 show that the presence of antibodies against both H5 and H7 in unexposed individuals is low70, 123, 124, as are B cells capable of recognizing both viruses, representing only 0.13% of the sample analyzed. Most of these studies highlight that the prevalence of antibodies against avian influenza viruses, and more specifically against H5N1, is low or very low, warning of the low or non-existent protection of the population against an avian influenza pandemic due to the level of antibodies capable of combating the infection by one of these viruses. On the other hand, these studies emphasize that the antibodies that have been found are probably the result of natural infections by seasonal influenza viruses due to their genetic and antigenic homology with avian influenza viruses, specifically referring to the high homology between the H5N1 subtype of avian influenza and the H1N1 subtype of seasonal human influenza. Induction of cross-neutralizing antibodies against hemagglutinin after seasonal vaccination In addition to the information available on seasonal influenza vaccines not specifically designed to induce heterotypic responses, several vaccine designs seek to achieve the paradigm of a “universal flu vaccine”. All of these designs seek to induce responses capable of producing protective antibodies that recognize conserved epitopes in the HA, so that they can be useful against various IAV subtypes. The response to seasonal flu vaccines is based primarily on the induction of protective antibodies. Most of the seasonal vaccines currently available are based on surface protein technology or, to a lesser extent, on split viruses. Due to the way the antigens are presented, these types of vaccines induce mainly humoral immunity, whose greatest potential is the induction of NAbs that target the most immunodominant areas of the HA125, 126. The humoral response is specifically directed towards protective response against infection, while the cellular response, especially that based on CD8+ T cell response, is more directed towards preventing severe disease through the destruction of infected cells127, 128. In this regard, it is important to note that vaccines that use other types of technologies that include a larger part of the virus or the whole virus, such as virosomal vaccines, Virus-like Particles (VLPs) or the classic inactivated whole viruses and attenuated vaccines, produce both humoral and cellular responses, which may reinforce this second aspect125. Some vaccines include adjuvants to enhance the broad-spectrum humoral response and induce cellular responses126, 129. Given the importance of the humoral response, it is necessary to determine the usefulness of heterotypic responses induced by seasonal vaccines to protect against IAV not included in the vaccines. However, vaccines designed against the globular head of HA have certain limitations in serving as a universal model. The main limitation is the variability of this portion of the protein. As it is the part most in contact with the immune system, it has a high rate of variation, which is also variable depending on the antigenic epitopes (Sa, Sb, Ca1, Ca2, Cb)130. Furthermore, after vaccination there is a marked decline not only in the recognition capacity of antibodies as the antigenic drift of the virus occurs, but also in the titer itself in the blood, which reduces the effectiveness of the vaccine as time passes since its administration131. Host factors also limit this response, with age, history of previous exposure, imprinting, and repeated vaccinations being the most relevant factors132. Despite these limitations, there is currently evidence that seasonal HA-based vaccines have the capacity to induce heterotypic responses against other viruses not included in their composition. Some animal models show that vaccination with trivalent seasonal vaccines induces heterotypic responses against subtype H5N1, inducing partial protection against challenge in mice110. This response appears to be exclusively dependent on the H1N1 subtype included in these vaccines, mainly due to the greater antigenic homology of this virus with the avian subtype. This response appears to be even more potent if the vaccine is administered through the mucosa in the same type of animals133, 134, 135, or using technologies such as influenza-immunostimulating complexes (influenza-ISCOM)136. In humans, there is also evidence that seasonal influenza vaccines are able of inducing antibodies capable of recognizing H5N1. For example, a 2008 study conducted in healthcare workers vaccinated with a seasonal protein-based seasonal influenza vaccine showed that 34.2% of them increased their NAb titers by >20-fold above baseline68. Other studies conducted with different types of vaccines and different age groups have shown mixed results70, 107, some even with negative results137, but in most of them, a notable heterotypic response was observed that is worth taking into account. In any case, current evidence shows that seasonal human influenza vaccines have limited usefulness in protecting against avian influenza H5N1 but may serve as initial priming before specific doses of avian vaccines are available. On the other hand, one of the most promising strategies for achieving this goal is to induce responses directed towards the most conserved areas of the HA, which is the main objective sought by universal influenza vaccines. According to the influenza vaccine roadmap (IVR) initiative at the University of Minnesota (https://ivr.cidrap.umn.edu/universal-influenza-vaccine-technology-landscape), a total of 244 universal influenza vaccine candidates have been explored with 32 active clinical trials using six different platforms. A total of 8 of the 221 universal influenza vaccine candidates are already in phase 3 clinical trials, but none have been approved at this time. Until a few years ago, the most common universal influenza vaccine designs were based on classical technologies, many of them based on chimeric protein technologies. Among these, it is worth highlighting the design by E. Puente-Massaguer and colleagues138, which based the response to the HA stem on the induction of specific antibodies against this domain using chimeric HA proteins containing exotic globular heads and conserved stems. Specifically, the stem or HA2 region is the least variable region of this protein139, as it is largely integrated into the virus membrane itself and, being below the globular head of HA (HA1), is protected from the pressure to the immune system. However, these designs are not without limitations. The main one is that the stem region is not immunodominant and, therefore, antibody induction is poorer than against the head. Chimeric HA proteins, in fact, seek to solve this problem by preferentially directing the response towards the stem. On the other hand, the location of the stem itself makes the accessibility of the antibodies created more limited or restricted to certain moments in the viral cycle when the HA stem is highly exposed, such as during viral budding. All these limitations have led to the diversification of vaccine designs targeting the stalk domain, with the search for headless HA designs140, lacking the globular head to only exposes the stalk domain, or even coupling them to viral vectors or nanoparticles that present stalk antigens on a surface141, 142. Neuraminidase also matter in cross-protection As a sialidase, NA cleaves sialic acid linkages on host cell surfaces, facilitating the budding and egress of IAV particles from infected cells. This allows mature virions to infect new host cells, initiating additional cycles of viral replication and promoting the spread of infection. In addition to aiding viral release, NA also hydrolyzes sialic acid residues present in respiratory mucus, thereby enhancing viral migration through the respiratory tract11, 143, 144. At the structural level, NA displays a highly organized architecture that is closely linked to its enzymatic function and antigenic characteristics145, 146. Structurally, NA is a tetrameric glycoprotein composed of four identical monomers, each approximately 470 amino acids in length that fold into four distinct structural domains11. Each monomer consists of a cytoplasmic tail, a transmembrane region, a stalk domain, and a membrane-distal catalytic head domain, which is responsible for catalyzing the hydrolysis of sialic acid (Fig. 3)11, 147. The NA active site contains an inner shell of eight highly conserved residues (R118, D151, R152, R224, E276, R292, R371, and Y406; N2 numbering) that directly interact with sialic acid, with Y406 functioning as the key nucleophilic residue (Fig. 3)148. Additionally, NA contains several structural loops important for its function, including the 150-loop (residues 147–152), 270-loop (residues 267–276), 340-loop (residues 342–347), and 430-loop (residues 429–433)149. Phylogenetic analysis based on genetic and structural features of NA from different IAV subtypes has classified NA proteins into two groups150. Group 1 includes the N1, N4, N5, and N8 subtypes, while group 2 comprises N2, N3, N6, N7, and N9 subtypes. Several broadly reactive epitopes on NA have been identified. However, compared to HA, antigenic epitopes on NA are less well characterized12, 13, 151, 152, 153, 154, 155, 156. Although point mutations occur in the NA protein, the rate of antigenic drift around its active site is slower than that observed for HA14. As a result, NA is generally more conserved than HA. Importantly, resistance caused by NA mutations tends to occur within the head domain, especially in or near the active site. Using extensive panels of monoclonal antibodies (mAbs) and recombinant NA proteins, several studies have defined key antigenic regions on NA145, 157, 158. Two main functional sites targeted by neuraminidase-inhibiting (NAI) antibodies have been identified: (i) the catalytic site and its surrounding residues; and (ii) regions located outside the catalytic site, including interfaces between adjacent NA monomers. Notably, many NAI mAbs exhibit inhibitory activity in enzyme-linked lectin assays (ELLA) but not in assays using small synthetic substrates, suggesting that they do not directly bind the catalytic site159. Instead, these antibodies likely act through steric hindrance, blocking access of larger substrates to the active site160. This mechanism of inhibition has been associated with reduced viral release from infected cells and impaired viral spread145. In addition, NA-reactive antibodies can enhance virus clearance through ADCC and complement-dependent cytotoxicity (CDC)143, 151, 157, 160, 161, 162. NA has gained increasing attention for its role as a protective immunogen, and it has become one of the most promising targets for preventing or combating IAV infections145. Although the NA is expressed at lower abundance than HA on the viral surface, it has been shown to elicit a robust and protective antibody response, making NA a valuable target for broadly protective vaccines12, 144, 151, 163, 164. Previous studies in both animal models and humans have demonstrated that preexisting NA-reactive antibodies can reduce infection and mitigate disease severity143, 151, 152, 161, 165, 166, 167, 168. However, most IAV vaccine development efforts remain focused primarily on targeting HA. Despite growing evidence supporting the importance of NA as an immunogen and its role in protection, current seasonal influenza inactivated vaccines contain inconsistent, and often minimal, amounts of NA. Moreover, there are no specific regulatory requirements to ensure a standardized amount of NA per vaccine dose5, 9, 166, 169, 170. It is hypothesized, that induction of antibodies capable of inhibiting NA activity could provide broader protection against antigenically drifted IAV strains. Additionally, mucosal NA-specific immune responses have been shown to prevent viral transmission, further supporting NA's potential as a key component of next-generation influenza vaccines, including universal vaccines167. An analysis of the antibodies elicited by the FDA-approved AS03-adjuvanted inactivated H5N1 (A/Indonesia/05/2005) vaccine (ClinicalTrials.gov Identifier: NCT01578317) demonstrated high levels of NAI titers169. Moreover, the fold increase in NAI titers from day 0 to day 42 was significantly greater in recipients of the AS03-adjuvanted vaccine compared to the non-adjuvanted group, suggesting that adjuvanted NA-based vaccines can elicit stronger and more cross-reactive NAI responses. In addition, the AS03-adjuvanted H5N1 vaccine promoted expansion of the anti-NA epitope repertoire in serum antibodies, highlighting the importance of identifying new antigenic epitopes within the viral NA169. In a related study, ferrets vaccinated with trivalent inactivated influenza vaccines, with or without adjuvant were partially protected against a lethal challenge with A/Vietnam/1203/2004 (H5N1). This protection was attributed to the H1N1 component of the vaccine, with the dominant protective response targeting NA. Importantly, this protection correlated with the presence of antibodies that inhibited sialic acid cleavage168. Current seasonal inactivated influenza vaccines primarily elicit antibody responses targeting the viral HA. However, they poorly display antigenic NA epitopes and rarely induce robust NA-reactive B cells responses12. In contrast, during natural infection with seasonal H1N1 or H3N2 subtypes, the immune system produces antibodies targeting the viral NA and NA-reactive B cells12. These NA-targeting antibodies have been shown to provide broad and potent protection, including against IAV variants that are resistant to NA inhibitors such as oseltamivir. In a key study, researchers isolated and characterized a panel of NA-reactive mAbs from individuals infected with seasonal influenza viruses12. These antibodies demonstrated broad binding across diverse IAV strains, often spanning the full historical circulation of a given NA subtype in humans. Importantly, when administered either prophylactically or therapeutically, these NA-specific mAbs conferred protection in mice against lethal influenza virus challenge12. These findings highlight the potential of NA as a valuable but underutilized target for next-generation influenza vaccines and therapeutics, with the capacity to enhance both the breadth and durability of protective immunity. Consistent with these findings, NA-reactive mAbs isolated from rabbits or mice have demonstrated protective efficacy against both homologous and heterologous IAV infections in vivo146, 157, 160, 171. For example, Doyle et al. used a mAb targeting a conserved NA epitope located between amino acids 22–230 (N2 numbering), which inhibited all nine NA subtypes in vitro, including H5N1146. This epitope (ILRTQESEC) is part of the NA active site147, 172. Importantly, the mAb conferred heterosubtypic protection in mice challenged with lethal doses of mouse-adapted H1N1 and H3N2, representing group I and group II IAV, respectively146. In another study, the antigenic domains of the NA from a seasonal pre-pandemic H1N1 virus (A/Brisbane/59/2007 H1N1) were mapped using a panel of N1-specific mAbs. It was found that some cross-reactive anti-NA antibodies could protect mice against both homologous and heterologous H1N1 and A/Vietnam/1203/2004 H5N1 viruses157. Similarly, another study demonstrated that N1 antigens from an early human isolate provided partial protection against challenge with a pH1N1 strain as well as a low-pathogenicity A/Vietnam/1203/04 H5N1143. This conclusion is supported by multiple reports showing cross-protection against H5N1 induced by H1N1 exposure157, 168, 173, 174, 175 due to the presence of N1 cross-reactive mAbs157. In 2007, a study reported immunization of mice against the NA of a pre-pandemic seasonal human H1N1 virus using DNA vaccination174. When challenged with a lethal dose of A/Vietnam/1203/04 H5N1, immunized mice showed partial protection. Additionally, passive transfer of sera from these NA N1 immunized mice to naïve animals conferred similar protection against A/Vietnam/1203/04 H5N1174. The study also demonstrated that sera from some human individuals contained NA-specific antibodies capable of inhibiting the sialidase activity of NA from HPAI H5N1 strains174. Subsequent research revealed that anti-NA antibodies generated by immunization with A/California/7/2009 H1N1 in ferrets cross-reacted with the NA of H5N1 but not with seasonal H1N1 NAs173. Sequence and structural analyzes of these N1 proteins showed that the A/California/7/2009 H1N1 NA shared at least 22 more amino acids in the head domain with the avian A/Hong Kong/213/2003 H5N1 and A/Vietnam/1203/2004 H5N1 strain than with seasonal human H1N1 NAs173. In another study, mice either infected with a sublethal dose of A/California/7/2009 H1N1 or vaccinated and boosted with VLPs containing NA and matrix proteins were challenged with a lethal dose of A/Vietnam/1203/2004 H5N1. Mice previously infected with the sublethal dose of A/California/7/2009 H1N1 survived without detectable A/Vietnam/1203/2004 H5N1, and those immunized with H5N1 or pH1N1 NA VLPs were also fully protected176. Recently, Daulagala et al., evaluated the presence of cross-reactive HAI and NAI antibodies to H5N1 through previous exposure to seasonal influenza infections163. For that, serum samples from 63 healthy donors 18–73 years of age were collected in 2020. Cross-reactive HAI and NAI antibodies to a clade 2.3.4.4b H5N1 virus (A/black-faced spoonbill/Hong Kong/AFCD-HKU-22–21429-01012/2022), which showed high homology to the HA and NA proteins of clade 2.3.4.4b candidate vaccine viruses A/chicken/Ghana/AVL-76321VIR7050–39/2021 (98.8% [HA] and 97.3% [NA]) and A/American wigeon/South Carolina/AH0195145/2021 (98.6% [HA] and 97.2% [NA])163. In addition, in this work the HAI and NAI antibody responses to A/California/04/2009 (pH1N1) were also determined. The authors detected high titers of cross-reactive NAI antibodies to H5N1 virus in 96.8% (61/63) of serum samples. To assess whether exposure to A/California/04/2009 (pH1N1) contributed to the cross-reactive NAI antibodies, authors used serum samples of 50 healthy donors (17–55 years of age) collected in 2009, before pH1N1 had become widespread. Results indicated that cross-reactive NAI titers detected in 2020 were greater than those detected in 2009, suggesting that exposure to pH1N1 virus is the potential source of cross-reactive NAI antibodies to H5N1163. Later, antibody responses to the HA and NA proteins of circulating bovine H5N1 strains were evaluated in a cohort of 73 individuals with prior influenza vaccination or infection124. As expected and consistent with previous findings70, low levels of HA-specific and NAbs were detected. In contrast, NA-binding and NA-inhibiting antibody responses were comparable to those observed against human-origin NA proteins. Notably, seasonal influenza vaccination did not significantly boost antibody titers against either the HA or NA of bovine H5N1 viruses. Interestingly, infection with pH1N1, but not with H3N2, led to significant increases in NAb titers against H5N1 as well as NA-binding and NA-inhibiting antibody responses. In this line, a recent study assessed population immunity to clade 2.3.4.4b A/Texas/37/2024 H5N1 and A/American wigeon/South Carolina/22–000345-001/2021 H5N1 viruses across multiple geographic regions in the United States between 2021 and 2024177. A total of 1794 serum samples from 723 individuals aged 0.5 to 88 years were analyzed. Low levels of pre-existing NAbs and HA head-binding antibodies were observed, whereas substantial cross-reactive binding antibodies to H5N1 NA were detected. The youngest age group (0.5–3 years) had the lowest, and older adults (>70 years) had the highest levels of pre-existing NA-binding antibodies to H5N1. Interestingly, infection with pH1N1 during the 2023–2024 influenza season not only boosted NA-binding antibody responses to circulating seasonal pH1N1 but also significantly increased NA-binding antibody levels to H5N1. In contrast, seasonal influenza vaccination with inactivated vaccines during the same period induced only modest, non-significant increases in H5N1 NA-binding antibodies, primarily in adult and older adult age groups. Functional analysis of NAI antibodies revealed that pH1N1 infection during the 2023–2024 season significantly enhanced cross-reactive NAI antibody responses targeting the NA of H5N1 viruses. Additionally, seasonal vaccination in 2023–2024 elicited significant NAI antibody responses across all age groups. Multiple studies have highlighted the importance of NA as an immunogenic viral protein, with growing evidence suggesting that influenza vaccines should be optimized to enhance the NA-specific humoral immune response144, 170. Such improvements could help induce broadly cross-reactive and cross-protective NA-targeting antibodies against divergent IAV strains. However, our understanding of human antibody responses to NA remains limited, with several critical knowledge gaps that need to be addressed. Although studies in humans have demonstrated that pre-existing NA-reactive antibodies can reduce infection rates and mitigate disease severity165, 166, most seasonal inactivated influenza vaccines primarily target the viral HA, with limited focus on NA. Understanding population-level susceptibility to novel influenza viruses is crucial for pandemic preparedness. Several reports have shown that NA from human pre-pandemic or pandemic H1N1 viruses can induce cross-reactive NA antibodies against H5N1 and other IAV subtypes following infection or vaccination. Furthermore, these NA-specific humoral responses have been shown to inhibit H5N1 infection both in vitro and in vivo, highlighting the potential of NA as a key target for broad and effective influenza immunity. Vaccination strategies Unlike the COVID-19 pandemic, we are certainly better prepared for a possible pandemic caused by HPAI H5N1. On one hand, seasonal vaccines have a certain heterotypic activity that provides individuals with some degree of protection against HPAI H5N1. This protection is likely to be non-sterilizing and heterogeneous across populations, largely depending on prior exposure to influenza and immune imprinting, but it may still contribute to reducing disease severity and viral shedding to some extent12, 70, 122, 173. Indeed, such heterotypic protection has been associated with cross-reactive antibodies targeting conserved regions of HA, particularly the stalk domain, as well as NA-specific responses10, 12, 122, 141, 173. Although it is not the most ideal protection paradigm to implement in the event of a pandemic of this type, seasonal vaccination is a tool to be considered given the ease of manufacture and availability of doses, and the logistics already in place for seasonal vaccination campaigns. In a pandemic preparedness context, seasonal vaccination could therefore serve as an immediately deployable, interim strategy while strain-matched vaccines are being produced. In this regard, it would be important to increase seasonal flu vaccination coverage in people who are in contact with wild or farm animals, as coverage in these groups is currently very low. Targeting these occupational groups (e.g., poultry workers, veterinarians, farm personnel) could also reduce opportunities for viral reassortment and zoonotic transmission, representing a dual individual and population-level benefit54. Seasonal influenza vaccination is already recommended in some countries, such as in Spain and the United States, for veterinaries and people working with animals, but regretfully, vaccination coverage does not reflect this need. For that reason, appropriate efforts should be done to increase the awareness of these professionals in a One Health context. On the other hand, unlike SARS-CoV-2, which appeared suddenly and required real-time follow-up actions and vaccine design, we have been monitoring H5N1 for more than 30 years, analyzing its evolution and changes, so information obtained during all these years is vital to address its possible jump and establishment in humans. This accumulated knowledge enables pre-pandemic preparedness strategies, including antigen selection, candidate vaccine virus (CVV) development, and the identification of genetic markers associated with mammalian adaptation and transmissibility63, 178. In addition, vaccines against H5N1 are available, some approved and others in different phases of clinical trials. According to WHO data and a systematic review conducted in 2025, a total of 23 vaccines approved by different specific international agencies are available against H5N1 and H5N834, 179. These vaccines have been designed using different technologies, 18 of them based on chicken embryonated eggs and 5 using cell cultures, one of them being a live-attenuated influenza vaccine (LAIV). Most of them contain some type of adjuvant in their composition (usually aluminum, AS03, and MF59). Adjuvanted vaccine formulations are particularly relevant in this context, as they allow antigen dose-sparing and have been shown to enhance the breadth of the immune response, which may be critical when facing antigenically drifted or shifted viral strains130. Of all these vaccines, 10 are approved for use in children, 16 in healthy adults, and 9 in older adults. However, the breadth and durability of the immune responses elicited by these adjuvanted vaccines, particularly against heterologous H5 clades, remain variable and not fully characterized using standardized correlates of protection. There are also many other vaccine designs that are in various stages of clinical trials and are likely to be approved in the coming years. In fact, there are 12 vaccine designs currently in phase I/II, and another 5 designs in the pre-clinical stage34, 179. These new yet unapproved vaccines explore new technologies, such as mRNA and recombinant proteins. These platforms offer key advantages in a pandemic scenario, including faster manufacturing timelines, improved scalability, and the possibility of incorporating multiple antigens or conserved regions to enhance cross-protective immunity. In particular, mRNA-based vaccines may enable rapid updating in response to emerging H5 variants and facilitate the design of multivalent or universal-like formulations180, 181. The growing number of designs by both public and private institutions demonstrates the great interest and need to develop preventive tools against this potential pandemic threat of H5N1. Nevertheless, important challenges remain, including the identification of robust correlates of protection beyond HAI titers, the evaluation of cross-neutralizing antibody responses, and the integration of these next-generation platforms into existing vaccination strategies and regulatory frameworks. Altogether, these strategies highlight that an effective response to a potential H5N1 pandemic will likely require a combination of readily deployable vaccines, targeted immunization of high-risk populations, and the progressive integration of next-generation platforms capable of inducing broad and cross-reactive immunity83, 129. Future perspectives Influenza, like other respiratory zoonotic viruses, recognize no borders and they can occasionally cross species barriers. The prevention and control of influenza would be impossible without strong surveillance activities and continuous monitoring of circulating IAV strains. In this regard, integrating genomic surveillance with antigenic characterization and real-time data sharing across countries is essential to detect emerging viral variants with pandemic potential. In addition, expanding surveillance at the animal-human interface, including systematic monitoring in poultry, swine, and cattle farms will be critical to identify early signals of zoonotic potential spillover. However, the surveillance and monitoring of influenza viruses in different animal species are not homogeneous across the world, and not many countries have a developed surveillance system beyond humans or, at most, in wild birds and poultry. HPAI could trigger outbreaks in the human population through two main mechanisms. The first involves an intermediate host, such as pigs, which can be co-infected by both avian and human IAV. These co-infection events may result in viral reassortment and the emergence of new IAV strains, with an antigenic composition unfamiliar to the human immune system, increasing the risk of infection and transmission. The second mechanism involves direct transmission from infected avian species to mammals, as we have previously observed with direct transmission of AIV to mammalian species such as horses182, dogs183, 184, dairy cows6, 30, or directly to humans32, 53. These pathways highlight the importance of a One Health approach that integrates veterinary, environmental, and human health surveillance systems. According to the CDC and Prevention influenza risk assessment tool (https://www.cdc.gov/flu/pandemic-resources/national-strategy/risk-assessment.htm) and the WHO tool for influenza pandemic risk assessment (https://www.who.int/teams/global-influenza-programme/avian-influenza/tool-for-influenza-pandemic-risk-assessment-(tipra)), population immunity to an emerging IAV is a key parameter to assess its pandemic risk. Therefore, it is important to evaluate the presence of cross-NAbs targeting HA or NA because, as indicated before, they have been shown to correlate with protection against IAV infection. However, current risk assessment frameworks would benefit from incorporating more granular immunological data, including the breadth and functionality of cross-reactive antibodies, as well as standardized assays to measure these responses at the population level. Large-scale seroepidemiological studies and longitudinal cohorts could provide valuable insights into baseline immunity and its evolution over time, improving pandemic preparedness models. Although it is not the focus of this review, in addition to humoral responses, cellular responses targeting epitopes of highly conserved proteins between AIV and seasonal IAV could have a key role in cross-protection. For instance, it has been suggested that cross-reactive T-cell immunity in seasonal and pandemic H1N1, could explain differential disease severity during the 2009 pH1N1 outbreak. In that context, it was reported that 41% of the CD4 and 69% of the CD8 T-cell epitopes were conserved between pH1N1 and previously circulating H1N1 viruses71, 185, 186, 187. Recently, a study of human epitope conservation between HPAI H5N1 (considering the A/Anhui/1/2005 prototype) and seasonal IAV sequences revealed that 57% of CD4 and 63% of CD8 T-cell epitopes, respectively, are either identical or conserved and they could contribute to T-cell cross-reactivity78. This prediction was experimentally verified testing responses from human T cells to non-avian IAV and their HPAI H5N1 counterparts78. Therefore, cross-reactive T-cell responses might be able to limit disease severity in the case of a HPAI H5N1 pandemic. From a translational perspective, these findings support the inclusion of conserved internal proteins as vaccine targets and highlight the need to better define T-cell correlates of protection, particularly in the context of severe disease prevention rather than infection blocking. Because of the evolving nature of IAV, it has gained importance the development of influenza universal vaccines, which provide broader, long-lasting, and universal protection. Contrasting traditional vaccines against specific viral strains, influenza universal vaccines would induce robust immune responses against conserved viral components shared across diverse IAV subtypes or against diverse IAV subtypes. Current strategies include targeting the HA stalk domain, NA, and conserved internal proteins to elicit both broadly NAbs and cross-reactive T-cell responses. However, key challenges remain, such as overcoming immunodominance of variable epitopes, ensuring durability of protection, and defining regulatory pathways for vaccines that do not rely on traditional strain-matching paradigms. Advancing these approaches will be critical to move from strain-specific responses to broadly protective immunity against emerging avian influenza viruses. In this context, advancing our understanding of cross-reactive immunity, strengthening integrated surveillance systems, and accelerating the development of broadly protective vaccines should be considered key priorities to mitigate the impact of future H5N1 emergence in humans.