Non‐neutralizing antibody responses following A(H1N1)pdm09 influenza vaccination with or without AS03 adjuvant system

Abstract Background Non‐neutralizing antibodies inducing complement‐dependent lysis (CDL) and antibody‐dependent cell‐mediated cytotoxicity (ADCC) activity may contribute to protection against influenza infection. We investigated CDL and ADCC responses in healthy adults randomized to receive either non‐adjuvanted or AS03‐adjuvanted monovalent A(H1N1)pdm09 vaccine (containing 15 µg/3.75 μg of hemagglutinin, respectively) on a 2‐dose schedule 21 days apart. Methods We conducted an exploratory analysis of a subset of 106 subjects having no prior history of A(H1N1)pdm09 infection or seasonal influenza vaccination enrolled in a previously reported study (NCT00985673). Antibody responses against the homologous A/California/7/2009 (H1N1) vaccine strain and a related A/Brisbane/59/2007 (H1N1) seasonal influenza strain were analyzed up to Day 42. Results Baseline seropositivity determined with hemagglutination inhibition (HI), CDL and ADCC antibody titers against viral strains was high; A/California/7/2009 (HI [40.4‐48.1%]; CDL [34.6‐36.0%]; ADCC [92.1‐92.3%]); A/Brisbane/59/2007 (HI [73.1‐88.9%]; CDL [38.0‐42.0%]; ADCC [86.8‐97.0%]). CDL seropositivity increased following vaccination with both adjuvanted and non‐adjuvanted formulations (A/California/7/2009 [95.9‐100%]; A/Brisbane/59/2007 [75.5‐79.6%]). At Day 21, increases in CDL and ADCC antibody geometric mean titers against both strains were observed for both formulations. After 2 doses of AS03‐adjuvanted vaccine, vaccine responses of 95.8% (≥9‐fold increase from baseline in CDL titers) and 34.3% (≥16‐fold increase from baseline in ADCC titers) were seen against A/California/7/2009; and 22.4% and 42.9%, respectively, against A/Brisbane/59/2007. Vaccine responses after 2 doses of the non‐adjuvanted vaccine were broadly similar. Conclusions Broadly comparable non‐neutralizing immune responses were observed following vaccination with non‐adjuvanted and AS03‐adjuvanted A(H1N1)pdm09 formulations; including activity against a related vaccine strain.


| INTRODUC TI ON
Following influenza virus infection, a robust immune response is observed involving the generation of both neutralizing and non-neutralizing antibodies. 1 Neutralizing antibody responses are directed toward the viral hemagglutinin (HA) glycoprotein that mediates virus attachment to host cells via sialic acid receptor binding and subsequent cell entry. The importance of neutralizing HA-antibodies in protection is well established, and influenza vaccines are developed and assessed primarily by their ability to induce hemagglutination inhibition (HI) as a surrogate for neutralizing antibodies. 2,3 However, most conventional HA-specific neutralizing antibodies target epitopes of the HA globular head that, while immunodominant, are subject to substantial antigenic drift and are typically strain-specific; hence the need for annual updating of the composition of the seasonal influenza vaccines so as to target and induce protective neutralizing antibodies to the anticipated predominant seasonal strains. 3 Non-neutralizing antibodies to influenza are also generated following infection and provide additional protection via a range of mechanisms including complement-dependent lysis (CDL) and antibody-dependent cell-mediated cytotoxicity (ADCC). [4][5][6][7][8][9][10][11] Such non-neutralizing antibodies can recognize and bind a range of viral epitopes expressed by influenza virus on the surface of infected cells with subsequent complement activation or direct cell lysis by natural killer (NK) cells, monocytes, and macrophages in conjunction with antiviral cytokine release. 6,11 A potential benefit of such non-neutralizing antibodies is their recognition of epitopes within the HA globular head and also of more highly conserved epitopes (eg, in the HA stalk domain) than those recognized by neutralizing antibodies, so offering a broader cross-reactive protection. 5,12 These non-neutralizing antibodies include those directed against internal proteins such as nucleoprotein and matrix 1 protein to which ADCC antibody responses are observed following clinical infection or influenza vaccination. 9,13 These aspects are of particular relevance to pandemic influenza and associated vaccine development, where virus genomic reassortment events result in novel strains with novel viral epitopes in their more variable antigenic domains. 10,11,14 In this respect, the role of CDL and ADCC antibodies in response to influenza infection or following vaccination is of considerable interest, with a number of reports in recent years. 7,9,13,[15][16][17][18] However, data from vaccine clinical studies are more limited.
Previously, we reported on a randomized controlled trial (NCT00985673) evaluating immunogenicity and safety of a monovalent A(H1N1)pdm09 pandemic influenza vaccine given with or without AS03 adjuvant. 19 Use of adjuvants such as AS03 is an important consideration in pandemic vaccine development as it provides an antigen sparing component to vaccine composition, which may be relevant when antigen availability to novel strains is limited.
In that study, robust HI antibody responses were observed with both the non-adjuvanted (15 µg of hemagglutinin) and AS03-adjuvanted vaccine formulations (3.75 µg of hemagglutinin); where the differences in hemagglutinin content in these different formulations represent such an antigen sparing effect. 19 To investigate the effect of these vaccine formulations on non-neutralizing antibody responses, we performed an exploratory evaluation and analysis of CDL and ADCC antibody responses, using sera collected in a sub-population of this study cohort. The aims were to characterize non-neutralizing antibody immunogenicity against the homologous A(H1N1)pdm09 vaccine strain (A/California/7/2009) and also against a related seasonal influenza strain; A/Brisbane/59/2007(H1N1) representative of a previously circulating seasonal H1N1 subtype (which in the context of the present analysis we consider this to be a heterologous vaccine strain).

| Study design and study population
This was an exploratory analysis of a sub-population of a previously reported randomized, observer-blind, controlled clinical trial (NCT00985673) conducted in the United States and Canada between October 2009 and December 2010. The study design, inclusion criteria, and primary objectives (immunogenicity and safety) have previously been published. 19 Participants were healthy adults 19 to 40 years of age, excluding subjects with any prior history of A(H1N1)pdm09 influenza vaccination or physician-confirmed A(H1N1)pdm09 infection, and those with a history of previous seasonal influenza vaccination. 19 The study was conducted in accordance with Good Clinical Practice and the Declaration of Helsinki.
All study-related documents were approved by the appropriate Institutional Review Boards of participating Centres; and written informed consent was obtained from all subjects prior to enrollment.
Anonymized individual participant data and study documents can be requested for further research from www.clini calst udyda tareq uest.

com.
The current analysis involved a subset of those participants from this parent study (corresponding to groups E and F) who were randomized to receive non-adjuvanted or AS03-adjuvanted formulations, respectively, of a A(H1N1)pdm09 pandemic influenza vaccine. 19 The A(H1N1)pdm09 pandemic influenza vaccine is a monovalent, inactivated, split-virion antigen either in non-adjuvanted form (with 15 µg of hemagglutinin) or as an AS03-adjuvanted formulation (Arepanrix, GSK, Belgium); with 3.75 µg of hemagglutinin, 19 and which contains DL-α-tocopherol and squalene in an oil-in-water emulsion. 20 Subjects in this exploratory analysis were randomly drawn from the according-to-protocol (ATP) cohort in each group, with subject K E Y W O R D S A(H1N1)pdm09 vaccine, AS03 adjuvant system, cross-reactivity, non-neutralizing antibodies selection based on available blood samples for additional testing for non-neutralizing antibody responses. Subjects received either non-adjuvanted or AS03-adjuvanted formulations on Days 0 and 21 administered in the deltoid muscle; no other vaccines were administered during this time-frame (and so for the purposes of the present analysis received only the monovalent vaccine). 19 Blood samples were collected on Day 0 (pre-vaccination) and on Days 21 and 42 (ie, 21 days after each vaccine dose). All samples were coded with a unique identification number, aliquoted, and stored at −80°C until analysis. in a similar manner as for the CDL assay, that is, the highest serum dilution at which ≥50% peak SIL of the sample was observed, with a similar cut-off value (32.0 1/DIL) used.

| Immunogenicity assessments
HI seropositivity status was assessed at baseline (Day 0) and on Day 42; seropositivity rate was defined as the percentage of subjects with HI titer equal to or above the assay cut-off value (≥10), consistent with the approach used in the parent study and other studies evaluating HI immunogenicity in response to the H1N1pdm09 vaccine. 19,21 For both CDL and ADCC assays, we evaluated seropositivity, geometric mean titer (GMT), vaccine response (VR), and mean geometric increase (MGI). Subjects were considered seropositive for CDL or ADCC antibodies if their antibody titer was equal to or above the assay cut-off (≥32.0 1/DIL). For GMT calculations, titers were log 10 transformed and then calculations performed using the antilog of the mean of the log 10 titer transformations. For CDL, titers <32.0 1/DIL were assigned a value of 10.7 1/DIL (to account for the 3-fold dilution), and for endpoint titers above the maximum assay readout value of the serum dilutions tested (>7776 1/DIL), a maximal titer of 23 328 1/DIL was assigned. For ADCC, titers <32.0 1/DIL were assigned a value of 8.0 1/DIL (to account for the 4-fold dilution); endpoint titers above the maximum assay readout (>32 768 1/DIL) were assigned a maximal titer of 131 072 1/DIL.
No standardized VR criteria exist, and to account for this uncertainty, we used two exploratory levels of response thresholds to determine VR, and applied both to assess VR at Day 21 and Day 42. For the CDL assay, VR was defined as the proportion of subjects who showed a 3-fold or 9-fold increase in the post-vaccination reciprocal titer from baseline. For those subjects with baseline titers of <32.0 1/DIL (considered seronegative at baseline), post-vaccination reciprocal titers of ≥96.0 1/DIL or ≥288.0 1/DIL were required to meet these 3-fold and 9-fold thresholds, respectively. For the ADCC assay, a similar approach was adopted (but using subsequent 4-fold or 16-fold increases from baseline in post-vaccination reciprocal titers; subjects seronegative at Day 0 required post-vaccination reciprocal titers of ≥128 1/DIL or ≥512.0 1/DIL to meet these thresholds). MGI was defined as the geometric mean fold rise in GMTs at Day 21/Day 42 relative to Day 0.

| Data analysis
Descriptive analyses were performed for all data and summarized

| RE SULTS
From the original ATP cohort from the primary study, 106 subjects were included in this exploratory analysis (accounting for approximately 50% of eligible participants); 52 receiving the non-adjuvanted F I G U R E 1 Study flow. From the original according-to-protocol cohort from the primary study, 106 subjects were included: 52 receiving the nonadjuvanted vaccine and 54 receiving the AS03-adjuvanted vaccine. ADCC, antibody-dependent cell-mediated cytotoxicity; ATP, according-to-protocol; CDL, complement-dependent lysis; HA, hemagglutinin; HI, hemagglutinin inhibition; n, number of subjects with available results for all three antibody responses (HI, CDL and ADCC); N, total number of subjects in the non-adjuvanted or AS03-adjuvanted vaccine group; TVC, total vaccinated cohort vaccine and 54 receiving the AS03-adjuvanted vaccine. Study flow and number of subjects assayed at study time-points are shown in Figure 1. Cohort demographics are presented in Table 1 Figure 2).  Table S1).

| Immunogenicity based on CDL antibody assay
Differences in VRs between the AS03-adjuvanted vaccine and the non-adjuvanted groups are presented in Table S2 and adjusted GMTs, MGI and adjusted-GMT ratio in Table S3. Between-group differences in CDL VRs and adjusted-GMT ratios against either vaccine strain were inconsistent, and with overlapping CIs in between-group comparisons.
We evaluated correlations between CDL and HI antibody titers at Day 42, in both vaccine groups and in the total exploratory analysis cohort (Table S4 and Figure S2). Overall, there was fair linear correlation between HI and CDL titers against A/California/7/2009, with correlation coefficient (r) values of 0.68-0.69 ( Figure S2).

Correlations between HI and CDL antibody titers against A/
Brisbane/59/2007 were strong on initial analysis but after removal of an outlier patient the adjusted correlation was very poor, with r between 0.14 and 0.26 (Table S4).

| Immunogenicity based on ADCC antibody assay
A high proportion of subjects in either study group were seropositive for ADCC antibodies against both vaccine strains at baseline,  Figure 3).  Table S1).
Similar to our CDL data, between-group differences in VRs and adjusted-GMT ratios for ADCC antibodies against either strain were inconsistent (Tables S2 and S3 (Table S4). In addition, correlations between CDL and ADCC titers were also poor for either strain.

| D ISCUSS I ON
We evaluated CDL and ADCC antibody responses after immunization with either non-adjuvanted or AS03-adjuvanted pandemic A(H1N1)pdm09 vaccine in a subset of participants from a previously reported clinical trial, 19  Abbreviations: ADCC, antibody-dependent cell-mediated cytotoxicity; AS03, Adjuvant System containing DL-α-tocopherol and squalene in an oil-inwater emulsion; CDL, Complement-dependent lysis; CIs, confidence intervals; GMT, geometric mean titer. a n, number of (seropositive) subjects with antibody titer ≥ 32.0 1/dilution on CDL or ADCC assay; N, number of subjects with available results. than due to previous pandemic influenza infection. This is supported by data from previous studies where healthy adults had high levels of ADCC antibodies to pandemic A(H1N1) and A(H5N1) and A(H7N9) virus strains, even though it was considered unlikely that previous clinical exposure had occurred. 9,13,15,16 In this context, although we observed lower baseline with the CDL assay, this is consistent with previous data in which CDL antibodies against pandemic strains were detected in only a fraction of subjects with high ADDC antibody titers, 7,15 although different assay sensitivities may also have influenced our results.
Immunization with either unadjuvanted or AS03-adjuvanted A(H1N1)pdm09 influenza vaccine was followed by increasing seropositivity rates and substantial increases in GMTs (for both CDL  Table S2. ADCC, antibody-dependent cell-mediated cytotoxicity; CDL, complement-dependent lysis; D, Day The generally comparable immunogenicity we observed with the AS03-adjuvanted vaccine in generating robust CDL and ADCC responses is consistent with that seen for HI responses in the primary study, 19 as well as that from other studies evaluating conventional HI immunogenicity of AS03-adjuvanted pandemic A(H1N1)pdm09 vaccines. In the latter, lower HA antigen induces comparable HI antibody responses to conventional non-adjuvanted formulations with higher HA content. 21,22 Our results are also consistent with other recent data on AS03-adjuvanted H1N1pdm09 vaccine, where robust ADCC responses were observed in subjects regardless of baseline HI seropositivity. 17,18 They are also consistent with studies using other influenzas vaccine, including a recent study demonstrating ADCC responses to quadrivalent and MF-59 adjuvanted vaccines in older adults. 23 The present study has some limitations.

| CON CLUS IONS
Cross-reactive CDL and ADCC antibodies may constitute crucial components of immune responses and provide some level of protection against existing and emerging pandemic influenza viruses. 24 Their role in providing protection against seasonal influenza is increasingly recognized, as is the need to more fully consider non-neutralizing antibody responses in vaccine development for a better characterization of the immune response of next-generation vaccine candidates. 25 Our data support these views and provide further evidence that ADCC and CDL assays should be an important consideration in vaccine development and evaluation and for the design of future, more cross-reactive vaccines (eg, universal vaccine

| TR ADEMARK S
Arepanrix is a trademark owned by or licensed to the GSK group of companies.