Serological evidence of H9N2 avian influenza virus exposure among poultry workers from Fars province of Iran

Most emerging diseases are of zoonotic origin, with wild and domestic animals acting as natural reservoirs [1]. Globalization and intensive animal farming have led to an increased spread of zoonotic infections [2]. Influenza type A viruses include several distinct subtypes based on the antigenic properties of the two major surface glycoproteins, the hemagglutinin (HA) and the neuraminidase (NA). To date, 18 subtypes of HA (H1-H18) and 11 subtypes of NA (N1-N11) have been described [3]. A number of influenza A subtypes have successfully crossed the species barrier and have established in the mammals and human population, causing yearly seasonal epidemics or they have sporadically been directly transmitted from poultry to humans causing zoonotic infections [4, 5]. The influenza A viruses of the H9N2 subtype are classified as low pathogenic avian influenza (LPAI) viruses. They cause infections both in wild birds and in the poultry population worldwide, including several countries in Asia, Europe, North Africa and North America [6, 7]. A significant proportion of recent H9N2 avian influenza (AI) isolates contains the L226Q (H3 numbering) amino acid substitution in their hemagglutinins (HAs) showing preferential binding to analogs of receptors with sialic acid linked to galactose by α2,6 linkage (SAα2,6Gal), a phenotypic portrait which is characteristic of human influenza viruses. Thus, these AI viruses might possess one of the key elements for infection in humans [8‐10]. Indeed, H9N2 viruses were isolated for the first time from humans in Hong Kong in 1999 and further human infections were reported in 2003 [11, 12]. These studies have shown that avian H9N2 viruses isolated from chickens are closely related to the H9N2 viruses responsible for human infection [13]. One human case of H9N2 AI was reported in Bangladesh [14] and the World Health Organization (WHO) in 2015 has reported new cases in Egypt and Bangladesh [15, 16]. In 1998, domestic pigs from Hong Kong were confirmed as being infected with H9N2 influenza, and infections have been reported also in recent years in swine along with other mammals [17, 18]. Furthermore, H9N2 viruses can contribute with gene segments during reassortment events leading to the generation of novel avian influenza virus that can infect humans (e.g. recent Chinese H7N9 and H10N8 viruses) [19, 20]. Recent transmission studies have demonstrated that some natural isolates of H9N2 viruses can acquire the ability to transmit efficiently between ferrets via respiratory droplets. In addition, it has been reported that serial passages of an H9N2 virus through guinea pigs can result in the introduction of amino acid substitutions, which increases contact transmission efficiency in this mammalian model [21, 22].

The wide circulation of H9N2 viruses throughout Eurasia, along with their ability to cause direct infections in mammals and humans, raises public health concerns on their potential role as candidates for the next influenza pandemic [23]. H9N2 human infection is generally asymptomatic or responsible for mild clinical signs. This may explain the scarcity of evidence accounting for the circulation and transmission of this virus subtype [24]. Nonetheless, human sera positive for H9 subtype were identified in China, India, Iran, Thailand, Cambodia, Romania, Egypt and Pakistan [25‐34].

In Iran, the H9N2 subtype was identified for the first time in 1998 and is still circulating in the poultry population. In the affected farms the mortality rate ranges between 20 and 60 %, although this may also be attributable to co-infections with other pathogens, such as IBV or Mycoplasma gallisepticum [35]. In spite of the implemented national control measures, which include the mass vaccination of poultry, the virus has rapidly spread and can be considered endemic in the Iranian poultry [36].

Exposure to H9N2 AI viruses in Iranian poultry workers was previously revealed by means of HI test, using serum titre ≥20 as positive cut-off [28, 29]. In previous reported studies, the H9N2 AI seroprevalence, assessed by means of HI test, in Iranian poultry workers ranged from 1.6 to 15.7 % (Median 9.5). Amongst the Middle-Eastern and Southern Asia countries, the highest seroprevalence was observed in Pakistan (47.8 %) by means of HI test [34] and 7.5 % by means of MN test in Egypt [33]. In the present study, two different serological tests were used and compared to screen 200 individuals from the Fars province, Iran to better understand the risk of infection with the H9N2 virus in the poultry sector. The two different diagnostic tests were applied to assess whether (i) the human exposure to this virus subtype can be confirmed and whether (ii) the poultry operators in the Iranian endemic areas (the Fars province, in this study) are at risk of exposure to the H9N2 infection.

The HA gene phylogenetic analysis of 88 Iranian H9N2 strains collected between 1998-2014 shows that they belong to two different groups of G1 lineage, named sub-lineage A and sub-lineage B for the purpose of this study (Fig. 1). Sub-lineage A refers to the H9N2 isolates collected between 1998 and 2007, which apparently are no longer circulating among the Iranian poultry, while B includes more recent isolates collected between 2003 and 2014. The amino acid sequence analyses of the Iranian H9N2 viruses showed that sub-lineage A consists of 47 % (14/30) of the strains with amino acid Q at position 226 (H3 numbering) and 53 % (16/30) with L226 substitution. Interestingly, sub-lineage B encloses mostly (95 %) (55/58) the strains with amino acid L226. Based on these results, two Iranian H9N2 isolates representatives of sub-lineages A (A/chicken/Iran/12VIR/9630/1998) and B (A/chicken/Iran/10VIR/854-5/2008) were selected as antigens for the serological study (Fig. 1). In particular, A/chicken/Iran/12VIR/9630/1998 was characterized by the amino acid Q226 while the A/chicken/Iran/10VIR/854-5/2008 virus had amino acid L at this position. The HA amino acid sequence distance between the two selected strains was 8.1 %.

The percentage of seropositive individuals against different influenza viruses obtained with the HI and MN tests along with the p-value of the chi-square test related to the association between exposure and response to each virus tested, per method are reported in Table 1. Detailed results of serological investigations obtained for the exposed and unexposed groups against different influenza viruses are described in the supplementary material (Additional file 1: Table S2). Serological results showed that the prevalence of antibodies against A/chicken/Iran/12VIR/9630/1998 H9N2 (Q 226) was of 2 % in the exposed group by the HI and MN tests, while no positive results were identified in the unexposed group.

An association was observed between professional exposure to poultry species and the presence of antibodies to the A/chicken/Iran/10VIR/854-5/2008 H9N2 (L 226) virus, as revealed by the HI and MN test (p < 0.05). The percentage of positivity was of 12 % for the exposed group and 2 % for the unexposed group in the HI test, while it respectively amounted to 17 % and 3 % for the exposed and unexposed groups in the MN test. Antibody prevalence to H1N1pdm 2009 was of 28 % in the exposed versus 53 % in the unexposed groups by the HI test, and 25 % versus 33 %, respectively, by the MN test. For H3N2, the prevalence was 36 % in the exposed group compared to 44 % in the unexposed group with the HI test, while it was 28 % and 24 % in the exposed and unexposed groups with the MN test. The titer distribution percentage against different viruses in different tests (Fig. 2) showed that with the increase of the titre range from ≥40 to ≥80 and from ≥80 to ≥160, the percentage of antibody positivity drastically decreased. Cross MN and HI tests on homologues and different chicken H9N2 hyper immune sera used as positive controls showed the absence of cross reactivity with H1N1pdm 2009 and H3N2 seasonal viruses; the same results were observable in human immune sera for H1N1pdm 2009 and H3N2, which showed the absence of a cross reactivity with the two tested H9N2 viruses.

The probability to be positive against A/chicken/Iran/10VIR/854-5/2008 H9N2, H3N2 and H1N1pdm 2009 viruses in different tests was calculated. The generalized linear model for binary responses applied on the H9N2 viruses indicated that the association between professional exposure or absence of professional exposure and H9 infection was significant (p = 0.0018). In particular, the exposed and unexposed groups showed a significantly different positivity both in the HI test (9 % vs. 1 %) and in the MN test (13 % vs. 2 %). For the H3N2 virus, the professional exposure was not significantly associated to the presence of antibodies (p = 0.22). This means that in both groups the probability of testing positive is almost similar. Differently, the unexposed group showed a significantly higher number of individuals (p = 0.001) seropositive to H1N1pdm 2009. If we consider that the H1N1pdm 2009 subtype does not circulate in avian species, it can be assumed that the serological data obtained for this subtype do not depend on the exposure to poultry species; other factors may have been at the origin of the distinct immunoreactivity observed against the H1N1pdm 2009 subtype in the two groups under study.

Results demonstrated that exposure to avian H9N2 viruses had occurred in humans in Iran (Fars province). In particular, the analysis showed a significantly higher prevalence of neutralizing antibodies against the A/chicken/Iran/10VIR/854-5 H9N2 virus in poultry workers than in the professionally unexposed group. These results are in accordance with a previously reported study from Iran by Alizadeh et al. [28] confirming that poultry workers are at risk of infection from zoonotic avian influenza virus. For the first time in this study, two distinct H9N2 LPAI viruses were used as antigens for serological investigations in humans. Interestingly, neutralizing activity against just one of these two antigens was revealed in several sera, which has resulted in significant different prevalence. This is consistent with the amino acid diversity of the two H9N2 antigens used in this study. In fact, the HA amino acid sequences of A/chicken/Iran/12VIR/9630/1998 (Q-226) and A/chicken/Iran/10VIR/854-5/2008 (L-226) differed for 43 amino acids, seven of which were located in antigenic sites (S158N, S160L, S175D, E190A, S193N, N208D, Q226L (H3 numbering)) previously reported by other groups [37‐40]. These amino acid substitutions may account for the different serological reactivity against these antigens. However, the difference in prevalence observed in this study between two tested H9N2 viruses might have different explanations.

The receptor binding properties of A/chicken/Iran/12VIR/9630/1998 H9N2 (Q-226)-like viruses affect their ability to infect humans. As a consequence, the infections of this virus in the exposed population are limited. On the contrary, the A/chicken/Iran/10VIR/854-5 H9N2 (L-226) – like viruses, have increased ability to infect humans resulting in higher prevalence in the exposed population. However, it should be considered that the molecular analysis of Iranian H9N2 viruses demonstrated that those harbouring the Q-226 mutations apparently are no longer in circulation. Furthermore, one previous study investigating the serologic response in humans exposed to avian viruses concluded that the incidence of seroconversion is low and that the antibody response after mild/asymptomatic infections is short-lived [41]. Therefore, the reduction of the circulation of Q-226 viruses in the poultry in recent times and/or the low antibody response after infection might be responsible for the decreased prevalence of antibodies against these AI viruses.

Consistently with the results discussed so far, the unexposed group counting 100 healthy subjects showed a very low percentage of positivity for H9N2 by the HI and MN tests, while the distribution of positive titres against H1N1pdm 2009 and seasonal H3N2 influenzas were similar in the exposed and in the unexposed groups.

According to Stephenson et al. [42] the cross reacting antibodies against H2 might explain the antibody reactivity against avian H9N2, especially among people born before 1968. Although we did not test the sera for this antigen, to evaluate this hypothesis, the age of each individual who had tested H9 positive was verified. Only one sample was from a person born before 1968, while all the others were collected from younger subjects with an age between 22-46 years, who could not be positive for H2N2.

The comparison between the serological results obtained using the HI and MN tests against H9N2 showed that the MN test detected more positive samples within the same group, however, the triple test results proved that the HI titer was reasonably more consistent than the MN test. In addition, different researches have reported that intra-laboratory reproducibility by MN is lower than the one resulting from the HI test [43], a fact that may explain the moderate discrepancy observed between the two serological assays applied for the detection of antibodies against H1N1pdm 2009 and H3N2 seasonal influenza. Both in the HI and MN tests, the percentage of positivity for H9N2 decreased with the increase of the titre range; the obtained results were justified on the grounds that humans have a poor immune response to infections with avian influenza viruses. For instance, studies have shown that the individuals with mild or asymptotic infections had a lower antibody titer compared to individuals who had seriously developed the illness [44]. H9N2 AI viruses have some of the specific molecular markers for the infection of mammals; however, they are still unable to completely adapt to mammals and to cause serious disease in humans, which may explain the low titers observed. As well, the time elapsed between the infection and sampling dates can also account for the decrease of the antibody titer.

In conclusion, this study has highlighted the potential of avian to human transmission of H9N2 AIVs, and indicated that poultry workers are at risk of infection. Avian transmission of H9N2 viruses in humans can increase the probability of human adaption, while genetic reassortment with other human seasonal viruses are means of generating an influenza virus with epidemic or pandemic potential. Hence, integrated medical-veterinary surveillance and research activities are essential in order to identify the emergence of new influenza viruses, assess the clinical significance of seropositivity and understand more on the mechanisms that favour the virus to cross the species barrier. In the poultry sector, surveillance and control programmes should be implemented to reduce the prevalence of H9N2 AIV in poultry population and minimize the risk of exposure in poultry operators.