Impact of Influenza B Lineage-Level Mismatch Between Trivalent Seasonal Influenza Vaccines and Circulating Viruses, 1999–2012

Abstract

Background. Influenza B virus strains in trivalent influenza vaccines are frequently mismatched to the circulating B strains, but the population-level impact of such mismatches is unknown. We assessed the impact of vaccine mismatch on the epidemiology of influenza B during 12 recent seasonal outbreaks of influenza in Finland.

Methods. We analyzed all available nationwide data on virologically confirmed influenza infections in all age groups in Finland between 1 July 1999 and 30 June 2012, with the exclusion of the pandemic season of 2009–2010. We derived data on influenza infections and the circulation of different lineages of B viruses during each season from the Infectious Diseases Register and the National Influenza Center, National Institute for Health and Welfare, Finland.

Results. A total of 34 788 cases of influenza were recorded. Influenza A accounted for 74.0% and influenza B for 26.0% of all typed viruses. Throughout the 12 seasons, we estimated that 41.7% (3750 of 8993) of all influenza B infections were caused by viruses representing the other genetic lineage than the one in the vaccine. Altogether, opposite-lineage influenza B viruses accounted for 10.8% of all influenza infections in the population, the proportion being highest (16.8%) in children aged 10–14 years and lowest (2.6%) in persons aged ≥70 years.

Conclusions. The population-level impact of lineage-level mismatch between the vaccine and circulating strains of influenza B viruses is substantial, especially among children and adolescents. The results provide strong support for the inclusion of both influenza B lineages in seasonal influenza vaccines.

influenza B virus, vaccine mismatch, trivalent influenza vaccines, quadrivalent influenza vaccines

(See the Editorial Commentary by Glezen on pages 1525–6.)

Since the mid-1980s, 2 antigenically distinct lineages of influenza B viruses (B/Victoria-like and B/Yamagata-like) have been circulating globally, causing disease in humans [1]. The divergence of influenza B viruses into 2 lineages has posed problems for the production of seasonal influenza vaccines, which have traditionally contained the 2 circulating influenza A strains, A(H1N1) and A(H3N2), but only a single B strain. During the past decade, selecting the correct B component of the trivalent influenza vaccine has proved particularly challenging, and lineage-level mismatches between the vaccine and circulating strains of B viruses have occurred during approximately half of the seasons [2]. The level of cross-protection between the 2 B lineages is not well known, but it is assumed to be low [3–6]. As a consequence, a lineage-level mismatch between the vaccine and the predominant circulating B strains is likely to reduce substantially the clinical effectiveness of the trivalent influenza vaccine during annual outbreaks.

In recent years, several studies have elucidated the previously poorly recognized clinical importance of influenza B viruses, resulting in increased interest in the production of quadrivalent vaccines that include both lineages of influenza B viruses [2, 7–11]. Although the first quadrivalent influenza vaccines were already introduced in the United States in the autumn of 2013, such vaccines are not yet available in most other parts of the world, and even in the United States the true clinical benefit afforded by these vaccines remains uncertain [12]. We designed this study to assess the overall impact of lineage-level vaccine mismatch against influenza B viruses during 12 influenza seasons in Finland.

METHODS

Study Design

We analyzed all available nationwide data on virologically confirmed seasonal influenza infections in all age groups in Finland between 1 July 1999 and 30 June 2012. We excluded the pandemic A(H1N1) season of 2009–2010 (defined as 1 July 2009 through 30 June 2010, for the purpose of this study) from the analyses because the focus of this study was on seasonal influenza, and the numbers of viral specimens obtained during the pandemic season were disproportionately high compared with other seasons. Hence, the total study period consisted of 12 seasonal outbreaks of influenza A and B viruses.

We analyzed all data in different age groups, categorizing children into 4 groups (0–4, 5–9, 10–14, and 15–19 years) and adults into 6 groups (20–29, 30–39, 40–49, 50–59, 60–69, and ≥70 years). Because the main purpose of this study was to determine the impact of vaccine-mismatched influenza B infections, we made no attempt to estimate the relative impact of A(H1) and A(H3) infections during the study period.

Sources of Data

We extracted the numbers of patients with virologically confirmed influenza infections during each season of the study period from the Statistical Database of the Infectious Diseases Register maintained by the National Institute for Health and Welfare, Finland. This register has been operational since 1995, and it contains monthly data on all influenza virus–positive specimens (A, B, or untyped) reported by diagnostic laboratories and practicing physicians from all regions of the country. These register data provide the most comprehensive estimate of the annual incidence of influenza A and B infections in different age groups in Finland.

We collected detailed data on the circulation of different lineages of influenza B viruses during each season in Finland from the National Influenza Center, National Institute for Health and Welfare. The National Influenza Center receives clinical specimens for sequencing and detailed identification of influenza viruses from patients with influenza-like illness seen in several sentinel clinics and collaborating medical centers in different parts of the country. Data on different lineages of influenza B viruses were available in electronic files since 1999. To ensure consistency in the data to be analyzed, we included only data from 1999 onward in this analysis. This decision was also supported by the fact that lineage-level vaccine mismatch against B viruses has been a particular problem during the past decade [2]. We estimated the extent of vaccine mismatch against influenza B viruses by comparing the known B antigens contained in the vaccine and the proportions of different lineages of B viruses circulating in Finland during each season of the study. For the season of 2003–2004, when influenza B viruses were very infrequently detected in Finland and other European countries, we based our estimation of vaccine mismatch against influenza B viruses on European-level data provided by the European Influenza Surveillance Scheme [13]. We rounded the estimated proportions of matched and mismatched influenza B viruses to the nearest 5% in the calculations.

Because this study was based solely on registry data available from the National Institute for Health and Welfare and openly accessible data from the National Infectious Diseases Register in Finland, it was not subject to ethics committee approval.

Statistical Analysis

When calculating the adjusted numbers of influenza A and B cases for each season, we allocated untyped viruses into influenza A and B viruses using the observed proportions of confirmed type A and B viruses during that season. We calculated 95% confidence intervals (CIs) for the proportions by the exact (Clopper–Pearson) method with the use of StatsDirect statistical software, version 2.7.9.

RESULTS

Overall Reporting of Influenza A and B Virus Infections

During the 12 seasons of this study, a total of 34 788 virologically confirmed influenza cases were reported to the Infectious Diseases Register in Finland (Table 1). Of all influenza virus–positive patients, 15 055 (43.3%) were 0–19 years of age, 13 100 (37.7%) were 20–49 years of age, and 6633 (19.1%) were ≥50 years of age. Across all seasons, influenza A accounted for 23 714 (74.0%) and influenza B for 8317 (26.0%) of the 32 031 typed viruses; 2757 (7.9% of all) viruses remained untyped. Following allocation of these untyped viruses into influenza A and B viruses according to the observed proportions of confirmed type A and B viruses during each season, the estimated total numbers of influenza A and B infections were 25 795 and 8993, respectively.

Table 1.

Influenza A and B Infections Reported to the Infectious Diseases Register in Finland

Season	No. of Influenza Viruses				Proportions of Influenza Viruses, % (95% CI)		Adjusted No. of Influenza Viruses
	A	B	Untyped	Total	A	B	A	B
1999–2000	1742	50	0	1792	97.2 (96.3–97.9)	2.8 (2.1–3.7)	1742	50
2000–2001	994	249	365	1608	80.0 (77.6–82.2)	20.0 (17.8–22.4)	1277	331
2001–2002	1383	158	87	1628	89.7 (88.1–91.2)	10.3 (8.8–11.9)	1460	168
2002–2003	274	743	211	1228	26.9 (24.2–29.8)	73.1 (70.2–75.8)	318	910
2003–2004	2317	38	184	2539	98.4 (97.8–98.9)	1.6 (1.1–2.2)	2499	40
2004–2005	1031	126	899	2056	89.1 (87.2–90.8)	10.9 (9.2–12.8)	1784	272
2005–2006	763	418	686	1867	64.6 (61.8–67.3)	35.4 (32.7–38.1)	1228	639
2006–2007	1763	109	245	2117	94.2 (93.0–95.2)	5.8 (4.8–7.0)	1990	127
2007–2008	1832	1806	31	3669	50.4 (48.7–52.0)	49.6 (48.0–51.3)	1848	1821
2008–2009	3478	719	27	4224	82.9 (81.7–84.0)	17.1 (16.0–18.3)	3502	722
2010–2011	2241	3552	18	5811	38.7 (37.4–40.0)	61.3 (60.0–62.6)	2247	3564
2011–2012	5896	349	4	6249	94.4 (93.8–95.0)	5.6 (5.0–6.2)	5900	349
All Seasons	23 714	8317	2757	34 788	74.0 (73.6–74.5)	26.0 (25.5–26.4)	25 795	8993

Abbreviation: CI, confidence interval.

Influenza A and B Viruses in Different Seasons

Influenza B viruses predominated during 2 of the 12 seasons (2002–2003 and 2010–2011), and in one season (2007–2008), influenza A and B viruses were detected in equal proportions (Table 1). The relative proportions of influenza B viruses among all influenza viruses in different age groups during each season are presented in Figure 1. In most seasons, there was a general pattern of proportions of influenza B viruses being highest among children 5–19 years of age and lowest in the older age groups. During the 2005–2006 season when the overall prevalence of influenza B viruses in the entire population was 35.4%, influenza B viruses predominated (52.4%–66.2%) among children and adolescents 5–19 years of age, but the proportions of influenza B viruses were only 4.0%–16.0% among patients ≥50 years of age.

Figure 1.

Proportions of influenza B viruses among all influenza viruses by season and age group.

Influenza B Lineage-Level Mismatch

We assessed the extent of lineage-level mismatch between the vaccine and circulating strains of influenza B viruses for each season by using information about the genetic and/or antigenic characteristics of circulating influenza B strains (Table 2). We observed a predominantly good match (90%–100%) between the vaccine B strain and the circulating influenza B viruses in 7 of the 12 seasons. In 4 of the 12 seasons, however, the vaccine B virus represented the other genetic lineage than the predominant B viruses characterized in the population.

Table 2.

Estimated Lineage-Level Match and Mismatch Between the Vaccine and Circulating Strains of Influenza B Viruses in Finland

Season	Vaccine B Lineage	Circulating B Lineages	Lineage-Level Vaccine Match, %	Lineage-Level Vaccine Mismatch, %
1999–2000	Yamagata	Yamagata (100%)	100	0
2000–2001	Yamagata	Yamagata (100%)	100	0
2001–2002	Yamagata	Yamagata (100%)	100	0
2002–2003	Victoria	Victoria (90%), Yamagata (10%)	90	10
2003–2004	Victoria	Yamagata (60%), Victoria (40%)	40	60
2004–2005	Yamagata	Yamagata (100%)	100	0
2005–2006	Yamagata	Victoria (95%), Yamagata (5%)	5	95
2006–2007	Victoria	Yamagata (100%)	0	100
2007–2008	Victoria	Yamagata (100%)	0	100
2008–2009	Yamagata	Victoria (100%)	0	100
2010–2011	Victoria	Victoria (90%), Yamagata (10%)	90	10
2011–2012	Victoria	Victoria (100%)	100	0

Throughout the 12 seasons of the study, we estimated that a total of 3748 (41.7% [95% CI, 40.7–42.7]) of 8993 influenza B infections were caused by B viruses that did not match the antigenic lineage of the vaccine virus (Table 3). Altogether, these opposite-lineage influenza B infections accounted for 10.8% (95% CI, 10.4–11.1) of all influenza infections during the study period, with the corresponding proportions ranging from 0.0% to 49.6% during individual influenza seasons.

Table 3.

Patients With Infections Caused by Lineage-Level Mismatched Influenza B Viruses, Compared With Vaccine Strain

Season	Total No. of Patients		Lineage-Level Mismatched B Viruses		Proportion of Patients With Mismatched B Viruses Among All Influenza Patients, % (95% CI)
	Any Influenza	Influenza B	Proportion (%)	No. of Patients
1999–2000	1792	50	0	0	0.0 (.0–.2)
2000–2001	1608	331	0	0	0.0 (.0–.2)
2001–2002	1628	168	0	0	0.0 (.0–.2)
2002–2003	1228	910	10	91	7.4 (6.0–9.0)
2003–2004	2539	40	60	24	0.9 (.6–1.4)
2004–2005	2056	272	0	0	0.0 (.0–.2)
2005–2006	1867	639	95	607	32.5 (30.4–34.7)
2006–2007	2117	127	100	127	6.0 (5.0–7.1)
2007–2008	3669	1821	100	1821	49.6 (48.0–51.3)
2008–2009	4224	722	100	722	17.1 (16.0–18.3)
2010–2011	5811	3564	10	356	6.1 (5.5–6.8)
2011–2012	6249	349	0	0	0.0 (.0–.1)
All Seasons	34788	8993		3748	10.8 (10.4–11.1)

Abbreviation: CI, confidence interval.

During the entire study period, the overall proportion of influenza B infections among all influenza infections was highest (41.2% [95% CI, 39.2–43.3]) in children 10–14 years of age and lowest (7.5% [95% CI, 6.5–8.5]) in subjects ≥70 years of age (Figure 2). Similarly, the proportion of infections caused by vaccine–lineage mismatched influenza B viruses was also highest (16.8% [95% CI, 15.3–18.4]) among children 10–14 years of age and lowest (2.6% [95% CI, 2.0–3.3]) in those aged ≥70 years. Among all B viruses, the relative proportions of strains representing the genetic lineage opposite to the vaccine lineage ranged between 34.9% and 46.6% in different age groups.

Figure 2.

Relative proportions of influenza A viruses and lineage-level matched and mismatched influenza B viruses, compared with the vaccine strain in different age groups during the 12-year study period.

DISCUSSION

Using a large, representative nationwide database over 12 recent epidemic seasons, we found that approximately 40% of all influenza B viruses causing infections in the population represented the other genetic lineage than the one included in the trivalent influenza vaccines. Overall, these mismatched influenza B infections accounted for approximately 10% of all influenza illnesses in the population. The impact of the vaccine mismatch was greatest among children and adolescents, among whom up to 17% of all influenza illnesses were due to influenza B viruses of the lineage that was not contained in the vaccine.

To our knowledge, this is the first study to provide a robust, population-level estimate of the impact of the lineage-level mismatch of the B component in seasonal influenza vaccines. A particular strength of our study is that we were able to combine the numbers of type-specific influenza infections with locally derived data on the circulation of different lineages of influenza B viruses during each season. Because the relative proportions of different types and subtypes of influenza viruses vary substantially between different countries and regions during any given season, the use of influenza B strain-specific data arising from some other region than that of the population under study would likely reduce the accuracy of the estimates. Furthermore, our 12-year study period included 7 seasons with good lineage-level match and only 4 seasons with clear lineage-level mismatch of the influenza B vaccine component, which implies that the results were not biased by primarily including only those seasons when lineage-level mismatch between the vaccine and circulating strains of influenza B viruses occurred.

The overall importance of influenza B viruses is underscored by our finding that 26% of all virologically confirmed influenza illnesses in the population during the 12 seasons were caused by B viruses. The relative proportion of influenza B viruses was greatest (approximately 40%) in children 5–14 years of age, which is in agreement with previous studies in other geographic areas [14]. A logical implication of this finding is that children and adolescents might benefit most from the addition of a second influenza B strain to the seasonal influenza vaccine. For example, if one made conservative assumptions that (1) 15% of all influenza illnesses in children were caused by lineage-level mismatched B viruses, (2) the efficacy of influenza vaccine was a modest 70% against all circulating strains, and (3) the level of vaccine-induced cross-protection between the 2 B lineages was 30% [3], a total of 640 per 1000 cases of influenza could be prevented by the use of trivalent vaccine and 700 per 1000 cases by the use of quadrivalent vaccine (6% absolute reduction). The additional 60 per 1000 cases that could be prevented with a quadrivalent vaccine would represent a 17% relative reduction of influenza illnesses in children when compared with a trivalent vaccine (60 of 360 per 1000 cases not preventable with a trivalent vaccine). The reductions would be even greater if the efficacy of the vaccine were higher and/or the level of cross-protection lower than what was assumed in the above-mentioned example.

Influenza B viruses are generally underestimated because they are thought to cause less severe illnesses than A viruses. However, although A(H3N2) subtypes have been associated with the greatest numbers of deaths and hospitalizations, the corresponding figures related to influenza B viruses have exceeded those of seasonal A(H1N1) viruses circulating prior to the 2009 pandemic [15, 16]. The histological features of fatal influenza B infections are similar to those of influenza A infections [9], and comparative clinical studies have demonstrated that influenza B infections are indistinguishable from influenza A infections [17–21]. Particularly in children, the conventional concept of influenza B viruses causing a milder disease may be seriously confounded by age because children with influenza B infections are generally older than those with influenza A infections [14, 17, 18, 22]. When adjusted for age, the clinical presentation of influenza B appears to be comparable to that of influenza A [23].

Some limitations of our study require consideration. First, it is obvious that the numbers of influenza cases reported to the Infectious Diseases Register represented only a small proportion of all influenza illnesses in the country because virologic confirmation of influenza is not routinely performed, especially among outpatients. However, the narrow confidence intervals for the observed proportions of influenza A and B viruses indicate that the numbers of viruses identified were adequate for reliable calculations. Second, the frequency of sampling for viral detection may have varied between different areas and settings. Although this may have brought some imbalance to the register data, our analyses were mainly based on the relative prevalence of influenza A and B viruses during each season, and it is unlikely that any sampling-related variation could have favored either virus type and thus biased the results. Third, the sentinel clinics that provided the National Influenza Center with clinical specimens for the identification of influenza viruses may not have closely represented the population demographics. However, even if true, this could have affected the results in the direction of substantial overestimation of the impact of mismatched B viruses only during seasons with predominant vaccine mismatch and considerable influenza B activity (2005–2006, 2007–2008, and 2008–2009). During those seasons, the mismatch between the vaccine strain and circulating influenza B viruses was virtually identical in Finland and the rest of Europe [2].

Our findings provide strong support for the inclusion of both influenza B lineages in seasonal influenza vaccines. The substantial impact of lineage-level mismatched influenza B viruses is in agreement with recent estimates which have indicated that switching from trivalent to quadrivalent influenza vaccines would result not only in reductions in influenza-related illnesses, hospitalizations, and deaths, but also in substantial cost savings to society [24, 25].

Notes

Disclaimer. The funder had no role in the initiation, design, performance, analysis, writing, or interpretation of the results of this study, and the decision to submit the manuscript for publication was made solely by the authors.

Financial support. This work was supported by GlaxoSmithKline, which provided a grant for investigator-initiated epidemiologic studies of influenza to the Hospital District of Southwestern Finland (a secondary employer of T. H.).

Potential conflicts of interest. T. H. has been a consultant to AstraZeneca/MedImmune, GlaxoSmithKline, Novartis, and Sanofi Pasteur MSD, and has given lectures at academic symposia organized by AbbVie and AstraZeneca. T. Z. has been a consultant to the World Health Organization, the European Centre for Disease Prevention and Control, and the Ministry of Social Affairs and Health in Finland, and has been employed by the National Institute for Health and Welfare, which has received grants from the European Commission and the National Institutes of Health. N. I. reports no potential conflicts.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Author notes

a

Present affiliation: Research Centre for Child Psychiatry, University of Turku, Finland.