Co-circulation of the two influenza B lineages during 13 consecutive influenza surveillance seasons in Italy, 2004–2017

Background Since 1985, two antigenically distinct lineages of influenza B viruses (Victoria-like and Yamagata-like) have circulated globally. Trivalent seasonal influenza vaccines contain two circulating influenza A strains but a single B strain and thus provide limited immunity against circulating B strains of the lineage not included in the vaccine. In this study, we describe the characteristics of influenza B viruses that caused respiratory illness in the population in Italy over 13 consecutive seasons of virological surveillance, and the match between the predominant influenza B lineage and the vaccine B lineage, in each season. Methods From 2004 to 2017, 26,886 laboratory-confirmed influenza cases were registered in Italy, of which 18.7% were type B. Among them, the lineage of 2465 strains (49%) was retrieved or characterized in this study by a real-time RT-PCR assay and/or sequencing of the hemagglutinin (HA) gene. Results Co-circulation of both B lineages was observed each season, although in different proportions every year. Overall, viruses of B/Victoria and B/Yamagata lineages caused 53.3 and 46.7% of influenza B infections, respectively. A higher proportion of infections with both lineages was detected in children, and there was a declining frequency of B/Victoria detections with age. A mismatch between the vaccine and the predominant influenza B lineage occurred in eight out of thirteen influenza seasons under study. Considering the seasons when B accounted for > 20% of all laboratory-confirmed influenza cases, a mismatch was observed in four out of six seasons. Phylogenetic analysis of the HA1 domain confirmed the co-circulation of both lineages and revealed a mixed circulation of distinct evolutionary viral variants, with different levels of match to the vaccine strains. Conclusions This study contributes to the understanding of the circulation of influenza B viruses in Italy. We found a continuous co-circulation of both B lineages in the period 2004–2017, and determined that children were particularly vulnerable to Victoria-lineage influenza B virus infections. An influenza B lineage mismatch with the trivalent vaccine occurred in about two-thirds of cases.


Background
Influenza A(H1N1), A(H3N2) and influenza B viruses are responsible for a significant disease burden during seasonal epidemics in humans [1]. Worldwide, these annual epidemics are estimated to cause about 3-5 million cases of severe illness and 290,000-650,000 deaths [2]. In the past, influenza B viruses were believed to cause milder illness than influenza A, but several studies have recently demonstrated that infections with influenza A and B are clinically indistinguishable and can cause severe complications in both children and adults [1, [3][4][5][6][7]. Originally, influenza B viruses represented a homogenous group but, since the late 1980s, they have evolved into two antigenically and genetically distinct lineages, defined by the reference strains B/ Yamagata/16/88 (Yamagata lineage) and B/Victoria/2/87 (Victoria lineage) [8]. In the 1990s, Yamagata-like viruses became predominant worldwide, whereas the Victoria lineage was mainly restricted to East Asia. However, during the 2000/2001 and 2001/2002 seasons, the Victoria-lineage viruses re-emerged in North America and Europe and spread globally [9,10]. Since then, the two B lineages have been co-circulating worldwide, with variability in terms of geographic distribution and genomic evolution [10][11][12].
Vaccination is the most effective way to prevent influenza virus infection and its complications. However, due to the constant evolution of influenza viruses, seasonal vaccines are regularly reformulated through continuous global monitoring of the influenza viruses circulating in humans, carried out by the National Influenza Centers (NICs) and World Health Organization Collaborating Centers (WHO-CC) within the WHO Global Influenza Surveillance and Response System (GISRS). Currently, the majority of influenza vaccines used worldwide are trivalent formulations containing one representative A(H1N1) virus, one A(H3N2) virus and only one B strain (B/Victoria or B/Yamagata). The concurrent circulation of the two B lineages over a lengthy period and the limited or absent cross-reactive protection between them have proven particularly challenging in terms of effectiveness of seasonal trivalent influenza vaccines (TIVs), as demonstrated by the frequent lineage-level mismatches occurring in the past between the predominant circulating B viruses and the WHO-recommended B vaccine strain [5,[13][14][15][16]. For this reason, quadrivalent influenza vaccines, containing viruses of both B lineages, have recently been introduced in order to reduce the risk of occurrence of B lineage vaccine mismatches [17,18].
In this retrospective analysis, we provide an update on the circulation of influenza B viruses over a 13-year period (2004-2017) of surveillance in Italy, building upon our previous reports and those at regional level [19][20][21][22][23]. In particular, we document the molecular characteristics of the hemagglutinin (HA) gene in the circulating B strains over the 13 consecutive epidemic seasons and the match between the dominant circulating B-lineage and the lineage included in the TIVs in each season.

Influenza surveillance in Italy and retrospective data collection
Influenza surveillance in Italy is based on a sentinel network of physicians (InfluNet) who report the number of patients with an influenza-like illness (ILI) on a weekly basis and collect respiratory specimens from November to April (i.e. from week 46 to week 17 of the following year), for virological analyses. Each year, a variable proportion of clinical specimens from non-sentinel sources and hospitalized patients are also collected. Virological surveillance is carried out by the NIC at the National Institute of Health (Istituto Superiore di Sanità), in collaboration with the regional influenza laboratory network (InfluNet). Preliminary analyses are performed on clinical samples at regional level and a representative subset of influenza virus-positive samples and/or virus isolates is sent to the NIC and subsequently shared with the WHO-CC for further antigenic and genetic analyses.
For this study, we analyzed all available virological data on laboratory-confirmed influenza B clinical samples collected within the framework of the InfluNet surveillance system, from 2004/2005 to the 2016/2017 season. To increase the number of specimens available for testing, seven regional laboratories of the InfluNet network (i.e. University laboratories of Milan, Trieste, Parma, Perugia, Sassari, Florence and Bari) were involved in the study and collaborated in the characterization analyses, with particular regard to those samples collected during the eight consecutive post-pandemic influenza seasons (i.e. from 2009 to 2017). In accordance with applicable laws and regulations, no clearance from an Ethics Committee is required in Italy for the retrospective analysis of anonymized data collected within the routine influenza surveillance scheme.

Laboratory testing and characterization analyses
Molecular characterization of the available clinical specimens positive for influenza B virus and/or virus isolates obtained in MDCK cells was performed following RNA extraction, using the QIAamp Viral RNA extraction kit (Qiagen, Hilden, Germany). A real-time RT-PCR assay, recommended by the WHO, was used to discriminate between influenza B lineages [24].
A subset of influenza B viruses (n = 491) underwent genetic analysis using conventional Sanger sequencing for the HA1 viral gene. Briefly, viral RNA was transcribed to cDNA using random primers and the HA gene was amplified with segment-specific primers for influenza B [25]. The PCR products were purified using a PCR purification kit (Qiagen, Hilden, Germany) and sequenced using an ABI 3500 genetic analyzer (Applied Biosystems, USA). Sequence alignment was performed using the ClustalW method implemented in the BioEdit program (v. 7.2.5). The phylogenetic trees were constructed using the Neighbor-Joining method and the Kimura 2-parameter model, using MEGA software (v. 6.0). The HA sequences obtained in this study and used in the phylogenetic analysis were entered in the EpiFlu database of the Global Initiative on Sharing All Influenza Data (GISAID) (see Additional file 1: Table S1).

Data analysis
For each season, we reported the number of samples from ILI patients that were tested, the percentage of influenza positivity, and the number and percentage of influenza-positive cases by viral type (A or B). Data on the circulation of the two B lineages in the country were obtained directly from this study or retrieved from the seasonal influenza database. A lineage-level mismatch between the WHO-recommended B component of the TIVs for use in the northern hemisphere and the circulating B viruses was defined as the percentage of influenza B viruses belonging to a lineage different from that included in the vaccine, for each season in Italy. A complete mismatch was defined for the season when > 60% of circulating B viruses belonged to the lineage not included in the TIV for that season.
For the age distribution and clinical presentation of patients with confirmed B/Victoria and B/Yamagata infections, median values and percentages were compared using Mann-Whitney U tests and Chi-square tests, respectively. The crude and adjusted relative risk (RR) were estimated using the univariable and multivariable log-binomial regression model. Analyses were performed using STATA software v. 11.2 (Stata Corporation, College Station, Texas, US).

Circulating patterns of influenza A and B viruses in Italy
During the 13-year study period, a co-circulation of both influenza A and B viruses was observed in Italy each year, although in variable proportions and with a general predominance of type A over type B viruses (Fig. 1 (Table 1). Approximately 59% (n = 2962/5028) of all laboratory-confirmed influenza B cases were from sentinel sources (community-based infections), whereas the remaining 35 and 6% were from hospitalized patients or other non-sentinel sources, respectively (data not shown).
In this study, the lineage was characterized or retrieved for 2465 (49%) out of 5028 laboratory-confirmed influenza B infections (Table 2). Of these, B/Victoria and B/ Yamagata accounted for 53.3 and 46.7%, respectively. In general, the co-circulation of the two lineages was observed in all seasons considered in Italy, although in different proportions and with alternating patterns, typically every two to three years. Nevertheless, an abrupt annual switch in the predominant lineage occurred in the most recent seasons included in the study, namely  Table 2).
Comparison between the B lineage included in TIVs and circulating B lineages in Italy in the seasons showed that more than 50% (n = 1247/2465) of influenza B viruses belonged to the lineage not included in the seasonal TIV (Table 2) Table 2).
Finally, laboratory-confirmed influenza B cases for which age information for the patient was available were stratified into five different age groups (Table 3). On the whole, the groups most affected by influenza B were children aged ≤14 years (64.5%) whereas the lowest proportion was observed among subjects ≥64 years of age (7.7%). When comparing the two lineages, patients infected by B/Victoria viruses were significantly younger than patients infected by B/Yamagata viruses (median age: 7 years vs. 13 years; p < 0.001), with increased detection rates also observed in adults (see Additional file 2: Fig. S1). In particular, the probability of infection with B/Victoria viruses was twice as high in pediatric age groups (0-4 years and 5-14 years) than in the elderly (RR = 2.13 and RR = 2.27, respectively), and slightly higher in hospitalized patients than in communityreported cases (RR = 1.24) ( Table 3).  Figure 3 shows the phylogenetic relationship of HA1 nucleotide sequences of 123 randomly selected influenza B viruses (see also Additional file 1: Table S1) [26]. Among these, the 120-loop appeared to be the most frequently mutated region in field strains (Table 4) [27]. In particular, three substitutions were detected more frequently within the 120-loop: N116K (shared by 88.6% of Yam-3 viruses), I117V and N129D (shared by 80.7 and 87.6% of Vic-1 viruses, respectively). The I146V substitution was found in the 150-loop region of 13.8% of Vic-1 viruses, and mostly identified in those strains circulating in the 2012/2013 and 2014/2015 seasons. Within the 190-helix, most Vic-1 (97.9%), Yam-2 (100%) and Yam-3 (99.6%) viruses acquired a potential glycosylation site with D197N/ D196N substitutions (based on Brisbane/60/2008-or Massachusetts/02/2012-and Wisconsin/01/2010-vaccine strain numbering) [26]. Moreover, the S202 N substitution was identified in 8.9% of Yam-3 viruses, and particularly in those circulating from 2012 to 2015. Finally, additional amino acid substitutions were found within the four antigenic epitopes and only detected in a few circulating field strains (Table 4).

Discussion
Influenza B viruses represent a significant cause of respiratory infections in humans that generally tend to be overlooked because of the dominance of influenza A [1, [3][4][5]28]. In this study, we provide a comprehensive picture of the circulation of the two influenza B lineages in Italy [5,30]. Overall, our data further highlight the unpredictability of influenza B circulation, which makes it challenging to predict which B lineage will predominate in the upcoming influenza season. This study did not permit any vaccine effectiveness estimates related to the occurrence of B mismatches because of the limited availability of data on vaccination status. However, the potential clinical impact of a B vaccine mismatch is mainly dependent on the proportion of circulating influenza B viruses, as well as pre-seasonal immunity in the population. Even though our analysis is based on observational data only, we can assume that the overall impact was minor in four out of eight seasons with a complete B mismatch, given the very low proportion of circulating influenza B viruses. For the other seasons, a much higher impact can be expected, as  [12,[31][32][33]. Thus, an alternation of two antigenically distinct B/Yamagata clades might have reduced the vaccine's ability to protect against these viruses, although they belong to the same lineage. Further studies are needed to evaluate the extent of cross-reactivity between influenza B viruses from distinct clades, and even from different lineages, and thus establish the real impact on vaccine effectiveness.
Phylogenetic analysis of influenza B viruses circulating in Italy over the entire study period confirmed the cocirculation of both lineages during each season, and revealed a mixed circulation of distinct evolutionary viral variants with different levels of match to the vaccine strains. In particular, a gradual drift was observed in both lineages and further clades and subclades were identified in each lineage. However  [34].
With regard to influenza B virus circulation among the population in Italy, most influenza B infections (> 60%) were found in children aged ≤14 years, with the highest proportion (nearly 44%) among school-age children (5-14 years), in line with previous studies globally [5,15,29,[35][36][37]. In particular, our study confirmed a higher proportion of B/Victoria virus infections in children than those caused by B/Yamagata viruses [38][39][40][41][42]. A similar distribution of B/Victoria and B/Yamagata strains was observed in the community and in hospitals, although there was a slightly higher probability of B/ Victoria infections in hospitalized patients. Indeed, previous studies have reported a greater proportion of B/ Yamagata virus infections in hospital inpatients across two Italian regions [21]. The different time period and number of patients with these characteristics may account for any difference observed in the outcomes for the two studies. Furthermore, the potential association between B lineage and age of infected patients and the increased virulence of the B/Victoria lineage suggested by others are still controversial and require further investigation [39,40,[43][44][45].
There are some limitations in this study. Only laboratoryconfirmed influenza cases were included in the study, and this could underestimate the real proportion of circulating viruses and the true incidence of influenza infections in the population. In addition, there were increased notifications received by hospitals in later years, as a result of a much higher demand for testing from the 2009 H1N1 influenza pandemic onwards. Lastly, we were only able to provide B lineage information on almost half of all laboratoryconfirmed influenza B notifications. Nevertheless, we can assume that these results would largely reflect the circulation of the two influenza B lineages in Italy.

Conclusions
This study provides information on the viral evolution and incidence of influenza B virus infections in Italy over a 13-year period. Continuous co-circulation of both B lineages highlights the complexity of antigenic variation in influenza B viruses and the differences in the epidemiological profile of the target population. In addition, the frequent occurrence of B-lineage mismatch with the TIV, as observed during the 2004-2017 seasons, likely contributed to the adoption of quadrivalent vaccines for influenza vaccination programs in the country.
Additional file 1: Table S1. GISAID accession numbers (AN) for hemagglutinin (HA) of influenza B viruses generated in this study (*) and used for the phylogenetic analysis, along with other Italian sequences retrieved from GISAID or NCBI database and reference and WHO vaccine strains (in bold).
Additional file 2: Figure S1. Violin plot comparing the median values (white dots), interquartile range (thick blue bar in the center) and distributions of age between influenza B/Victoria-and B/Yamagatalineage cases.