Protective efficacy of live-attenuated influenza vaccine (multivalent, Ann Arbor strain): a literature review addressing interference

Abstract

Selecting the B strain for inclusion in a trivalent seasonal influenza vaccine has been difficult because two distinct influenza B lineages frequently co-circulate, prompting consideration of a quadrivalent vaccine containing two A and two B strains. Because interference among wild-type influenza viruses is a well-documented phenomenon and viral replication is required to elicit protection by the licensed live-attenuated influenza vaccine (LAIV; MedImmune, LLC, Gaithersburg, MD, USA), a potential quadrivalent formulation raises considerations of interference among the LAIV strains contained in the vaccine. We reviewed the available clinical and nonclinical literature to understand the potential impact of viral interference on immunogenicity, efficacy and shedding of LAIV strains. We have found no clinically significant evidence of viral or immune interference affecting efficacy of LAIV strains in multivalent vaccine formulations. Future clinical studies should compare the safety and immune responses of children and adults to licensed trivalent and investigational quadrivalent LAIV formulations.

Keywords::

• interference
• LAIV
• live-attenuated influenza vaccine
• quadrivalent LAIV
• trivalent LAIV

Figure 1. Vaccine virus replication in ferret nasal wash samples (study 1).

Vaccine virus present in nasal wash samples after dose 1 (day 0) and dose 2 (day 28) was detected and quantitated by inoculation of decimal dilutions of samples into Madin–Darby canine kidney cells. The amount of vaccine virus present was expressed as log₁₀ median TCID₅₀ per ml.

TCID₅₀: Median tissue culture infectious dose.

Figure 2. Vaccine virus replication in ferret nasal wash samples (study 2).

Vaccine virus present in nasal wash samples after dose 1 (day 0) and dose 2 (day 28) was detected and quantitated by inoculation of decimal dilutions of samples into Madin–Darby canine kidney cells. The amount of vaccine virus present was expressed as log₁₀ median TCID₅₀ per ml.

TCID₅₀: Median tissue culture infectious dose.

Figure 3. Although high rates of seroconversion following live-attenuated influenza vaccine are associated with high efficacy, high efficacy is observed with low rates of seroconversion.

Serum antibody was measured before and after vaccination using HAI assays, and seroconversion was defined as a fourfold or greater rise in HAI antibody titers. HAI data are for all subjects, regardless of baseline serostatus. Efficacy indicates protection of vaccine recipients against wild-type influenza strains that were matched to the live-attenuated influenza vaccine strains. Error bars represent the 95% CI.

HAI: Hemagglutination inhibition.

Viral interference

Influenza B viruses have evolved into two antigenically and genetically distinct lineages that can co-circulate and cause disease globally [1,2]. Only one B lineage is represented in trivalent influenza vaccines, and in some years, B viruses of the opposite lineage have predominated during seasonal epidemics [3,101,102]. This situation has prompted development of investigational quadrivalent influenza vaccines containing antigens representing both B lineages in addition to the two type A strains already contained in licensed trivalent vaccines. Inclusion of an additional antigen or vaccine virus in quadrivalent vaccines has raised considerations of interference and its potential for affecting clinical efficacy against all strains contained in the vaccines.

Viral interference, defined as the interference of one virus with the replication of another when both are administered simultaneously to susceptible cells, has long been recognized for animal, plant and bacterial viruses [4]. The goal of vaccination with multivalent vaccines is to induce protection against each of the individual vaccine components. Because vaccination with live viral vaccines relies on vaccine virus replication and consequent induction of protective immunity, successful vaccination requires sufficient replication of each of the individual vaccine strains. Although viral interference is not an issue for nonreplicating vaccines, it is recognized that immune interference can reduce the immune response to individual antigens of multivalent inactivated or live vaccines [5].

A variety of mechanisms could contribute to viral interference, such as competition for cellular receptors, competition for molecular substrates required for replication and induction of inhibitory host proteins such as interferon. A detailed discussion of viral interference mechanisms is beyond the scope of this article. Regardless of the mechanisms involved, the end result of viral or immune interference among components of multivalent live vaccines could be reduced immunogenicity or efficacy of one or more vaccine strains. For the live-attenuated influenza vaccine (LAIV; MedImmune, LLC, Gaithersburg, MD, USA), reduced immunogenicity of one or more vaccine strains is not necessarily indicative of clinically significant viral interference because the immune mechanisms by which the vaccine exerts its effects have not been fully elucidated and the most commonly employed measure of immunogenicity (hemagglutination inhibition [HAI] antibody response) has not been shown to be an absolute correlate of protection for LAIV [6]. Furthermore, it must be kept in mind that some influenza hemagglutinins are intrinsically less immunogenic than others owing to variations in sequence, independent of whether they are presented to the immune system as a wild-type virus, a live-attenuated vaccine or an inactivated vaccine.

To provide a consensus description of the data regarding interference among Ann Arbor LAIV strains, and whether such interference impacts vaccine efficacy, we reviewed the current literature and all available clinical evidence from studies performed with investigational (99 published reports) and licensed (27 published reports) formulations of LAIV. The scope of this article is limited to discussion of vaccine strains based on the Ann Arbor LAIV master donor viruses, A/Ann Arbor/6/60 and B/Ann Arbor/1/66.

Studies in cell culture

Description of the model

Cell culture models have been used to evaluate viral interference among LAIV strains. The strength of cell culture models is that the kinetics, magnitude and duration of vaccine virus replication can be directly compared between monovalent and multivalent formulations without the potential confounding influence of a host immune response. However, cell culture systems are subject to a variety of limitations, including requirements for culture conditions and temperatures that may not reflect the normal host environment (which is highly relevant for temperature-sensitive LAIV vaccine strains). In addition, cultured cells used in laboratory studies may not be representative of the targeted human cell types.

Cell culture interference studies with multivalent LAIV

Rhesus monkey kidney cell culture has been used to study interference among LAIV strains [7]. Monovalent (A/H1N1, A/H3N2 or B) or bivalent (A/H1N1 and A/H3N2) LAIV formulations replicated to comparable levels in Rhesus monkey kidney cell cultures. In one study, when the potency of the type A strains was high (10⁶ median tissue culture infectious doses [TCID₅₀]) relative to that of the type B strain (10⁴ TCID₅₀), reduced replication of the type B LAIV strain was observed with a trivalent LAIV formulation. However, when the potency of the type B LAIV was increased to 10⁶ TCID₅₀, type A and type B strains were observed to replicate similarly [7]. Thus, the ‘interference’ effect observed in cell culture was not an intrinsic property of LAIV strains because it was not present when the culture was inoculated with LAIV strains of equivalent potency. Similar results were obtained using a Madin-Darby canine kidney cell culture model [8].

Studies in ferrets

Description of the model

The ferret is a frequently used model of human influenza infection and immunity because ferrets are susceptible to infection by intranasal inoculation with type A and B viruses, develop signs of disease that resemble human influenza illness, have different biologic responses to virulent and attenuated influenza strains, and develop a protective immune response after intranasal infection [9]. Seronegative ferrets have been used to evaluate viral interference among LAIV strains. The strengths of the seronegative ferret model include the ability to sample the nasal mucosa frequently, which allows measurement of the magnitude and duration of replication of LAIV strains in the absence of underlying pre-existing anti-influenza immunity. In addition, the robust immune responses of seronegative ferrets to influenza allow assessment of the serum antibody response to individual strains. Finally, ferrets vaccinated with monovalent or multivalent LAIV can be challenged with wild-type influenza viruses to assess protection in the upper and lower respiratory tracts. The latter feature is a key strength of the model because it allows an assessment of whether any observed viral or immune interference actually results in reduced protection against influenza infection. One limit of the model is the current lack of commercially-available ferret-specific immunologic reagents to assess cellular immunologic responses.

Ferret interference studies with bivalent type A LAIV

An early study used ferrets as a model to assess the potential for viral interference between influenza type A LAIV strains [10]. In this study, each LAIV strain induced an antibody response in all three vaccinated ferrets when administered as a monovalent formulation. However, with a bivalent formulation of strains with equal potency (10⁷ median egg infectious doses [EID₅₀]), one ferret did not develop an antibody response to the A/H1N1 strain and two ferrets developed a lower antibody response to the A/H1N1 strain. The difference in immune response was greater when the potency of the A/H1N1 strain was tenfold lower than that of the A/H3N2 strain. Based on these results, the authors concluded that a higher dose of the A/H3N2 strain in the bivalent formulation may have interfered with the replication of a lower dose of the A/H1N1 strain. However, the number of animals studied was low (three) and vaccine virus replication was not measured.

A quite different result was obtained when these same strains were administered to a small number of human adults. Specifically, in adults, the immune response to the A/H1N1 LAIV strain was similar after administration of monovalent and bivalent formulations. However, the proportion of subjects with an immune response to the A/H3N2 LAIV strain was 63% (17 out of 27) when administered as a monovalent formulation compared with 35% (14 out of 40) when administered as a bivalent formulation (p = 0.05) [10]. These differing results highlight the value of strain-specific clinical efficacy data for multivalent formulations, which is discussed later.

Ferret studies with bivalent influenza type B LAIV

Three studies evaluated the immunogenicity of a bivalent B formulation compared with monovalent formulations [3]. The underlying assumption in these studies was that comparability in antibody titers would suggest that two B LAIV strains of different lineages could be simultaneously administered to seronegative animals without significant interference (i.e., a level of interference causing a reduced immune response). The vaccine formulations each had one B/Yamagata lineage strain and one B/Victoria lineage strain: B/Jilin/20/03 and B/Hong Kong/330/01; B/AnnArbor/01/94 and B/Beijing/243/97; or B/Jiangsu/10/03 and B/Malaysia/2506/04. For each of the three formulations studied, equivalent amounts of each of the vaccine viruses were administered, and HAI titers generated against both influenza B strains were compared with those generated by monovalent formulations. In all three studies, HAI titers induced against each virus in the bivalent formulation were similar to those induced by the monovalent formulation. These results indicated that simultaneous vaccination with two influenza B lineages had minimal effect on the immunogenicity of each virus and there was no serologic evidence of virus interference.

Ferret studies with trivalent & quadrivalent LAIV

Two studies in seronegative ferrets were performed using two different investigational quadrivalent LAIV formulations that consisted of A/H1N1, A/H3N2 and one representative strain from each influenza B lineage. Results of these studies are provided in detail because they have not been previously published. The purpose of these studies was to examine the immunogenicity of each influenza B lineage virus when both lineages were present in a quadrivalent formulation, to assess the impact of a quadrivalent formulation on the immunogenicity of the A/H1N1 and A/H3N2 strains, and to assess efficacy of the quadrivalent formulation. Both studies included two comparator trivalent LAIV formulations that contained the same A/H1N1 and A/H3N2 strains as the quadrivalent formulation and either a B/Yamagata or a B/Victoria lineage strain. In Quadrivalent Study 1, the quadrivalent vaccine consisted of a total volume of 0.2 ml containing 10^7.0±0.5 focus forming units (FFU)/dose of each of the following LAIV strains: A/New Caledonia/20/99 (H1N1), A/Wisconsin/67/05 (H3N2), B/Florida/07/2004 (B/Yamagata lineage) and B/Malaysia/2506/04 (B/Victoria lineage). This formulation was compared with two trivalent formulations that contained the same A/H1N1 and A/H3N2 strains but included either B/Florida/07/2004 or B/Malaysia/2506/04 as the comparator influenza B strain virus. As with the quadrivalent vaccine, the trivalent vaccines contained 10^7.0±0.5 FFU of each of the three vaccine virus components in a total volume of 0.2 ml. In Quadrivalent Study 2, the quadrivalent vaccine formulation consisted of 10^7.0±0.5 FFU/dose in a total volume of 0.2 ml of each of the following LAIV strains: A/South Dakota/06/07 (H1N1), A/Uruguay/716/07 (H3N2), B/Florida/04/2006 (B/Yamagata lineage), and B/Malaysia/2506/04 (B/Victoria lineage). This formulation was compared with two trivalent formulations that contained the same A/H1N1 and A/H3N2 strains but included either B/Florida/04/2006 or B/Malaysia/2506/04 as the comparator influenza B strain virus.

In both studies, comparable HAI responses were generated by the quadrivalent and trivalent vaccines for the strains contained in the vaccines (Tables 1 & 2). As expected, there was little evidence of any influenza B strain cross-lineage immunogenicity and the quadrivalent vaccine induced superior responses to the influenza B strain not contained in the trivalent formulations. The HAI titers against each of the influenza A and B strain viruses were robust and comparable when inoculated as a quadrivalent or trivalent vaccine; differences observed were small, not statistically significant and similar to those seen for the A/H1N1, A/H3N2 and B components of the two different trivalent formulations. For both studies, vaccine virus replication kinetics and titers from nasal washes were similar for the quadrivalent and trivalent formulations, which also indicated that there was minimal impact of the addition of a fourth strain, consistent with a lack of interference that might be manifested as either reduced replication titers and/or a truncated kinetic curve (Figures 1 & 2).

Efficacy of the quadrivalent LAIV in the ferret model was determined by inhibition of replication of wild-type challenge virus. Wild-type virus challenge was performed at 21 days postvaccination. Results from studies 1 and 2 were similar and indicate that wild-type virus shedding in nasal washes was inhibited in both quadrivalent and trivalent vaccine groups, as determined by the lack of virus detection from the nasal wash samples collected from individual animals in these vaccine groups. This is in contrast with placebo groups in which robust viral titers (ranging from a peak of 10^1.3–10^4.8 TCID₅₀/ml at day 1 after challenge to 10^1.4–10^3.4 TCID₅₀/ml at day 3 after challenge) were consistently detected in nasal wash samples from all animals after challenge with wild-type virus. In addition to inhibition of local nasal influenza viral replication after challenge, influenza viral replication in the lung was markedly inhibited in animals vaccinated with either trivalent or quadrivalent LAIV in studies 1 and 2 when compared with placebo. Specifically, no wild-type challenge virus was detected in lung samples from vaccinated animals, whereas a range of levels of wild-type challenge virus (10^0.8–10^3.3 EID₅₀/ml) were measured in lung samples from placebo control animals. This complete protection of vaccinated animals was observed for all four wild-type challenge viral strains used. Hence, differences in efficacy were not detected with the addition of a fourth vaccine virus, and trivalent and quadrivalent formulations were protective in ferrets.

A final ferret study evaluated the potential for interference when seasonal trivalent LAIV was concomitantly administered with the 2009 pandemic H1N1 LAIV [11]. This study also demonstrated that simultaneous administration of four LAIV strains, in this case through the addition of a second antigenically distinct H1N1 strain, did not impact the serum antibody response to the strains administered and there was no evidence of interference.

Clinical studies

Clinical studies with investigational formulations of LAIV

In the 1960s, type A and type B cold-adapted, temperature-sensitive, attenuated influenza vaccine strains were developed by John Maassab (University of Michigan, Ann Arbor, MI, USA). Clinical studies of the Ann Arbor LAIV strains in the USA began with monovalent type A vaccines, and later expanded to include bivalent type A vaccines and trivalent vaccines containing A/H1N1, A/H3N2 and B strains, as reviewed in [12]. Thus, during development of LAIV, there were opportunities to compare vaccine strain replication and immunogenicity using monovalent, bivalent and trivalent formulations to directly assess the potential for interference among LAIV vaccine strains. Only studies that compared monovalent with multivalent formulations and were prospectively designed to directly address the question of LAIV vaccine strain interference are formally included in this article [7,10,13–15]. These studies are summarized in Table 3 and discussed later.

Pediatric and adult studies, summarized in Table 3, support the overall conclusion that LAIV vaccine strains, whether given in monovalent or trivalent formulations, exhibit immunogenicity that is generally comparable [7,13]. While in some studies shedding and/or immunogenicity of a multivalent component strain was lower than that of the same strain given as a monovalent formulation, these differences were not necessarily due to interference; equally likely explanations for differences in shedding or immunogenicity could be differences in vaccine potency, the number and type of assays used and the specific study population, as discussed later. All studies were small; apparent differences could have been due to chance or unmeasured differences in the study cohorts.

Vaccine strain potency

Vaccination of young seronegative children with trivalent LAIV, in which the potency of the type B strain was 100-fold lower than that of the type A strains, resulted in less shedding of the influenza B strain compared with trivalent LAIV in which all strains were equipotent (also seen in early ferret studies), although antibody response rates were similar in both cases [7]. Dose escalation studies in young children have demonstrated that the potency of each constituent strain in multivalent LAIV should be at least 10⁶ TCID₅₀ to ensure adequate immunogenicity of all strains after a single LAIV dose [16].

Number & type of assay

The number and type of assays used to study the response to vaccination with LAIV (i.e., assays to detect virus shedding or immune responses) also influenced whether reduced immunogenicity or shedding were observed for multivalent LAIV. In a small study in which monovalent type B, bivalent type A and trivalent vaccines were administered to adults, immunogenicity of the influenza B strain appeared to be reduced after vaccination with trivalent LAIV compared with monovalent B LAIV [15]. However, shedding of the B strain appeared similar after vaccination with monovalent or trivalent LAIV. In another small study, a lower rate of shedding of the influenza B strain in trivalent LAIV compared with monovalent B LAIV was not mirrored by serologic data; the serum antibody response to the influenza B strain was actually higher after vaccination with trivalent LAIV compared with monovalent type B LAIV [7].

Variability in study populations & assays

Variability in study population demographics and size and variability in assays appear to have influenced conclusions regarding interference [7,10,14,15]. Specifically, based on antibody responses it was suggested that the H3N2 (A/Los Angeles/2/87) strain interfered with the H1N1 (A/Kawasaki/9/86) strain when these strains were administered to seronegative children as a bivalent LAIV vaccine [14]. However, no interference as measured by viral shedding or serologic responses was reported when these same strains were administered to seronegative children as components of a trivalent LAIV vaccine [7]. This lack of interference was confirmed in a study in which serologically unscreened adults were vaccinated with a trivalent vaccine containing these same LAIV strains, with no interference against any of the three vaccine strains reported [13]. These apparently conflicting data obtained using the same LAIV strains strongly suggest that the reported interference was not an intrinsic replicative property of these LAIV strains. Instead, the observed differences were more likely due to differences in the study populations, methods used to produce the investigational vaccines for the various studies, assay variability and/or the limited number of subjects tested, and consequent limited statistical power.

In addition to the studies summarized in Table 3, several other published reports have speculated that antibody responses to multivalent LAIV may reflect viral interference, even though the studies were not designed to allow an adequate assessment of interference because the studies did not include monovalent control groups. For example, three studies [17–19] reported that an A/H3N2 strain (A/Los Angeles/2/87) may have interfered with an A/H1N1 strain (A/Kawasaki/9/86), whereas two other studies [20,21] reported no interference with the Kawasaki A/H1N1 strain by the Los Angeles A/H3N2 strain or another A/H3N2 strain (A/Korea/1/82). Because none of the five studies included a monovalent control group, no valid conclusion could be drawn as to whether viral interference was responsible for the observed antibody responses. Furthermore, in two studies in which the same A/H1N1 (A/Kawasaki/9/86) and A/H3N2 (A/Los Angeles/2/87) vaccine strains were administered to children or adults in monovalent or trivalent formulations (which allowed direct assessment of interference), no interference between the same Kawasaki A/H1N1 and Los Angeles A/H3N2 LAIV strains was reported (Table 3)[7,13]. Taken together, these results illustrate that small clinical studies of the same LAIV strains in different populations using different assays have led to different conclusions regarding the potential for interference among LAIV strains. These results also highlight that inclusion of monovalent LAIV control groups can facilitate interpretation of the immunogenicity results after vaccination with multivalent LAIV and can eliminate interference as an explanation for differences in immunogenicity of individual strains in multivalent LAIV formulations.

Clinical studies with licensed LAIV formulations

The clinical relevance of viral interference would be best evaluated in randomized, placebo-controlled efficacy studies that evaluate the incidence of confirmed influenza illness and vaccine virus replication and immunogenicity end points, in influenza seasons in which all influenza strains represented in the vaccine co-circulate in the community. However, these conditions occur very rarely owing to the unpredictable nature of annual influenza epidemics.

Although postvaccination HAI response and vaccine virus shedding can be accurately and reliably measured, they do not adequately predict the clinical efficacy of LAIV [22]. This principle is illustrated in Figure 3, which provides a summary of pediatric efficacy trials of LAIV that evaluated serum HAI responses in a study cohort and clinical efficacy against circulating influenza types/subtypes. Although high rates of seroconversion after vaccination with LAIV are associated with high efficacy, high efficacy is also observed with low rates of seroconversion (Figure 3).

There are two pediatric efficacy studies that are relevant to the question of potential interference occurring in trivalent LAIV formulations in that they measured immune responses and were able to assess safety and efficacy against all three circulating influenza strains during the same influenza season. Adverse events reported during these studies were consistent with previous observations that the most common adverse events after vaccination with LAIV are runny nose, nasal congestion or fever (100–101°F oral). The first study was conducted in Asia and included 3174 children 12–35 months of age; 111 children were also evaluated for immunogenicity [23]. After vaccination in year 1 with two doses, 60, 61 and 59% of children had an HAI response to the A/H1N1, A/H3N2 and influenza B strains, respectively. During the subsequent influenza season, LAIV efficacy against A/H1N1, A/H3N2 and influenza B strains was 81 (95% CI: 69–89), 90 (95% CI: 71–98) and 44% (95% CI: 6–67), respectively. A subsequent analysis of efficacy by time demonstrated that the lower efficacy against influenza B could be attributed to progressive antigenic drift among circulating B strains. The CDC isolate data from Asia demonstrated a progressive antigenic drift for influenza B viruses during the study season [24]; at 1 to <5 months after vaccination when influenza B viruses were generally well-matched to the vaccine, efficacy was 74% (95% CI: 39–89). The second study was conducted over two influenza seasons in Northern Europe among children 6–35 months of age, who were vaccinated with two doses of LAIV in year 1 and a single dose in year 2 [25]. All three strains circulated during year 2. In spite of modest year 2 seroconversion rates of 50, 25 and 38% to A/H1N1, A/H3N2 and influenza B strains, respectively, LAIV efficacy was 90 (95% CI: 56–99) 90 (95% CI: 83–95), and 82% (95% CI: 54–94) for these strains.

Lastly, a meta-analysis of all randomized, placebo-controlled pediatric efficacy studies conducted with LAIV demonstrated similar high efficacy for all three vaccine types/subtypes. Among previously unvaccinated children, the efficacy of two doses of LAIV compared with placebo for antigenically similar strains was 77% (95% CI: 72–80). For A/H1N1, A/H3N2 and influenza B strains, efficacy was 85 (95% CI: 78–90), 76 (95% CI: 70–81) and 73% (95% CI: 63–80), respectively [26].

Conclusion

Multiple studies in cell culture, ferrets and humans have attempted to evaluate interference among LAIV strains. The studies have reported conflicting results, even when the same strains were examined. These apparently disparate results may be explained by differences in vaccine potency, the number and type of assays used, and the specific study population. Additionally, the small number of animals or individuals vaccinated in these studies may not have been adequate to describe the true effects of the vaccine. The fact that the same LAIV strains appeared to exhibit interference in some studies, but not in others, supports the conclusion that the observed interference is not an intrinsic property of the LAIV strains.

In contrast, several large, randomized, placebo-controlled clinical efficacy studies of LAIV in children have demonstrated consistently high levels of protection against all three strains contained in the trivalent vaccine, despite at times apparently low serum HAI responses. Similarly high levels of efficacy were demonstrated against all three vaccine strains in the single season in two studies [23,25]. Thus, low serum HAI responses to individual components of multivalent LAIV should not necessarily be interpreted as evidence of interference among vaccine strains, or as being predictive of low vaccine efficacy. In some small immunologic studies, some LAIV strains in multivalent formulations have induced lower serum antibody responses than when the LAIV strains were administered as monovalent formulations. While some researchers have interpreted these results as serologic evidence of interference, these differential immune responses do not appear to translate to reduced protection against strains contained in multivalent vaccines. Indeed, results of efficacy studies in children indicate that even if some level of viral interference did occur with LAIV, this has not translated into diminished efficacy because LAIV has consistently provided high levels of protection against all three components of the trivalent vaccine. If quadrivalent influenza vaccines are licensed in the future, clinical studies could assess the impact of an additional influenza B strain on vaccine effectiveness.

Expert commentary

Influenza vaccines are formulated annually to protect against strains expected to circulate globally during the upcoming influenza seasons in the Northern and Southern Hemispheres. Recommendations regarding strain composition are issued by the WHO and national regulatory authorities. All licensed influenza vaccines are currently trivalent, containing two type A strains (A/H1N1 and A/H3N2) and one type B strain.

Vaccine strains in licensed LAIV are generally comparable in immunogenicity and efficacy, whether given in monovalent or trivalent formulations. In the future, it is possible that four strains could be recommended for inclusion in influenza vaccine formulations, based on epidemiology and the inability of trivalent formulations containing a single influenza B strain to protect against the two distinct B lineages that frequently co-circulate.

Quadrivalent influenza vaccine formulations are in development and could be commercialized after regulatory approval is obtained and recommendations are made by policy bodies. For LAIV, because vaccine strains for each B lineage (B/Victoria and B/Yamagata) have been previously shown to be efficacious in the trivalent formulation [3], the primary question is whether the addition of a second B lineage LAIV strain might interfere with the replication of the other vaccine strains. This question has been evaluated in nonclinical studies in seronegative ferrets, with no evidence of significant interference. In humans, demonstration of quadrivalent vaccine efficacy in randomized controlled field studies with all four vaccine components is not feasible because it would require multiple years and because in many countries, placebo-controlled studies are no longer ethical owing to recommendations for the annual vaccination of children against influenza. As a result, consistent with regulatory agency guidance, the clinical development of a quadrivalent LAIV would involve assessment of safety and immunogenicity in children and adults, with the goal of demonstrating comparable safety and immunogenicity between a quadrivalent LAIV and two trivalent LAIV formulations, each including a B strain of a different influenza B lineage. If demonstrated, a quadrivalent vaccine would be expected to have comparable safety and effectiveness to the currently licensed trivalent vaccines with additional protection against the alternate influenza B lineage.

Five-year view

Since 1977, influenza A/H1N1, A/H3N2 and B viruses have circulated globally and have been included in all licensed trivalent influenza vaccines. Influenza B has evolved into two distinct lineages: Victoria and Yamagata lineages. The lineage responsible for annual epidemics caused by influenza B has been difficult to predict; in the USA, in five of the past ten influenza seasons, the predominant circulating influenza B lineage was different from that contained in the vaccine. As a result, the inclusion of an additional B strain in the annual influenza vaccine formulations would increase the chance of the vaccine matching the circulating epidemic strains of influenza.

Human clinical trials in adults and children of quadrivalent LAIV and inactivated vaccines containing influenza B strains representing both lineages are currently underway, and safety and immunogenicity data in comparison with trivalent vaccines will be available in the near future. Safe and immunogenic vaccines containing four seasonal influenza strains would be expected to provide enhanced protection against both influenza B lineages. Quadrivalent seasonal influenza vaccines are expected to enter the commercial market within 2–5 years.

Table 1. Ferret hemagglutination inhibition titers and seroconversion rates at day 14 after vaccination with trivalent or quadrivalent live-attenuated influenza vaccine formulations.

Vaccine	HAI mean titer (log2) ± SD^† (% seroconversion)
	H1N1	H3N2	B/Florida/7/2004	B/Malaysia
Quadrivalent 1 B/Malaysia/2506/04 B/Florida/7/2004	6.6 ± 1.3 (100)	3.5 ± 1.7 (100)	4.2 ± 0.8 (100)	3.5 ± 1.1 (100)
Trivalent B/Malaysia/2506/04	6.4 ± 0.9 (100)	3.4 ± 1.8 (100)	BLD	5.0 ± 0.7 (100)
Trivalent B/Florida/07/2004	7.8 ± 0.8 (100)	5.6 ± 0.5 (100)	4.8 ± 0.8 (100)	BLD

^†Limit of detection = 1.0 log₂ HAI.

BLD: Below limit of detection; HAI: Hemagglutination inhibition; SD: Standard deviation.

Table 2. Ferret hemagglutination inhibition titers and seroconversion rates at day 14 after vaccination with trivalent or quadrivalent live-attenuated influenza vaccine formulations.

Vaccine	HAI mean titer (log2) ± SD^† (% seroconversion)
	H1N1	H3N2	B/Florida/4/06	B/Malaysia/2505/04
Quadrivalent 2 B/Malaysia/2506/04 B/Florida/4/06	4.7 ± 1.3 (100)	2.4 ± 1.2 (0)	6.8 ± 0.6 (100)	4.3 ± 0.6 (100)
Trivalent B/Malaysia/2506/04	5.0 ± 0.8 (100)	2.1 ± 1.6 (0)	BLD	5.8 ± 0.6 (100)
Trivalent B/Florida/04/2006	5.4 ± 1.2 (100)	2.2 ± 1.4 (0)	6.5 ± 0.9 (100)	BLD

^†Limit of detection = 1.0 log₂ HAI.

BLD: Below limit of detection; HAI: Hemagglutination inhibition; SD: Standard deviation.

Table 3. Studies comparing responses to monovalent and multivalent live-attenuated influenza vaccine.

Study (year)	Strains in vaccine	Potency^† (log10 TCID50)	Study population, age	Proportion of vaccinees shedding designated vaccine strain n/N (%)				Proportion of vaccinees developing ≥fourfold rise in serum antibody n/N (%)			Evidence for interference	Ref.
				H1N1	H3N2	B		H1N1	H3N2	B
Gruber et al. (1993)	Study 1 A/H1 A/H3 B A/H1, A/H3, B	6.0 6.0 4.0 6.0, 6.0, 4.0 (Low B)	Children, 6–24 months	5/8 (63) 9/10 (90)	11/12 (92) 10/10 (100)	7/8 (88) 3/11 (27)		5/8 (63) 8/10 (80)	11/12 (92) 10/10 (100)	0/8 (0) 2/11 (18)	Shedding data indicate that A strains interfered with B strain only when potency of B strains was 100-fold lower in trivalent vaccine than potency of A strains. Antibody responses were comparable in monovalent and trivalent formulations	[7]
	Study 2 A/H1 A/H3 B A/H1, A/H3, B	6.0 6.0 6.0 6.0, 6.0, 6.0 (High B)	Children, 6–24 months	5/8 (63) 24/25 (96)	11/12 (92) 24/24 (100)	10/11 (91) 21/26 (81)		5/8 (63) 18/25 (74)	11/12 (92) 24/24 (100)	0/8 (0) 5/26 (19)
Gruber et al. (1996)	A/H1N1 A/H3N2 A/H1, A/H3	6.2 6.2 6.2, 6.2	Children, 6–18 months	7/44 (16) NR	16/45 (36) NR			30/36 (83) 10/32 (31)	31/31 (100) 32/32 (100)		Antibody responses to A/H1N1 were lower in bivalent formulation compared with monovalent formulation	[14]
Keitel et al. (1993)^‡	Study 1 B A/H1, A/H3 A/H1, A/H3, B	7.1 7.1, 7.1 7.1, 7.1, 7.1	Adults, 18–40 years	7/9 (78) 6/11 (55)	0/9 (0) 1/11 (9)	10/10 (100) 10/11 (91)		8/9 (89) 7/11 (64)	2/9 (22) 2/11 (18)	4/9 (44) 2/11 (18)	Shedding data indicate interference with the B strain in study 2, but not in study 1. Antibody responses to B strain were lower in trivalent formulations	[15]
	Study 2 B A/H1, A/H3 A/H1, A/H3, B	7.1 7.1, 7.1 7.1, 7.1, 7.1	Adults, 18–40 years	8/9 (89) 4/10 (40)	6/9 (67) 1/10 (10)	10/10 (100) 5/10 (50)		9/9 (100) 7/10 (70)	4/9 (44) 3/10 (30)	6/10 (60) 0/10 (0)
Atmar et al. (1995)	A/H1 A/H3 B A/H1, A/H3, B	7.1 7.1 7.1 7.1	Adults, 18–40 years	ND	ND	ND		35/80 (44) 36/80 (45)	22/80 (28) 15/80 (19)	21/80 (26) 24/80 (30)	Antibody responses to all strains were comparable in monovalent and trivalent formulations	[13]
Potter et al. (1983)	A/H1 A/H3 A/H1, A/H3	8.0 8.0 8.0, 8.0	Adults	ND	ND	ND		10/30 (33) 13/40 (33)	17/27 (63) 14/40 (35)		Antibody responses to A/H3N2 were lower in bivalent formulation compared with monovalent formulation	[10]

^†Potency expressed in terms of log₁₀ TCID₅₀.

^‡Two doses of vaccine were administered in each study; data presented here are for the first dose only.

A/H1N1 LAIV strains used were A/Kawasaki/9/86 [7,13–15] and A/USSR/77 [10]. A/H3N2 LAIV strains used were A/Los Angeles/2/87 [7,13–15], A/Bethesda/1/85 [13], and A/Victoria/3/75 [10]. B LAIV strains used were B/Ann Arbor/1/86 [7,15] and B/Yamagata/17/88 [13].

ND: Not done; n/N: Number shedding/number of vaccine recipients; NR: Not reported; TCID₅₀: Median tissue culture infectious dose.

Key issues

• • Interference among live virus vaccine strains is defined as partial or complete inhibition of replication of a vaccine strain due to replication of another vaccine strain. Interference could theoretically attenuate the immune response or vaccine efficacy.

• • Studies in which investigational monovalent and multivalent live-attenuated influenza vaccine (LAIV) were compared in children and adults have shown that LAIV strains are generally comparable in immunogenicity, whether given in monovalent or trivalent formulations. In some studies, shedding and/or immunogenicity of a multivalent component strain was lower than that of the same strain given as a monovalent formulation, but these differences were not necessarily due to interference. Equally probable explanations for these results were differences in vaccine potency, the number and type of assays used and specific populations studied.

• • Early small studies of LAIV immune responses reported conflicting results regarding interference, even when using the same vaccine strains. This supports the conclusion that interference is not an intrinsic property of LAIV strains.

• • Nonclinical studies in ferrets demonstrated that quadrivalent and trivalent LAIV formulations are comparable in immunogenicity, replication kinetics and magnitude of replication. LAIV efficacy was maintained after the addition of a fourth vaccine strain.

• • Randomized, placebo-controlled efficacy studies of licensed trivalent LAIV have not yielded evidence of clinically-relevant interference among vaccine strains. Two of these studies conducted in young children demonstrated high levels of efficacy against all three vaccine strains in the same influenza season.

Acknowledgements

The authors thank Brian Murphy and Christopher S Ambrose for careful review of the manuscript and helpful suggestions, and Raburn Mallory for preparingFigure 3.

Financial & competing interests disclosure

The authors are employees of MedImmune, LLC, the manufacturer of LAIV (FluMist^®). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Writing assistance was utilized in the production of this manuscript. The manuscript was formatted for submission with the assistance of Complete Healthcare Communications, Inc. (Chadds Ford, PA, USA) and funded by MedImmune.