Concordance assessment between a multiplexed competitive Luminex immunoassay, a multiplexed IgG Luminex immunoassay, and a pseudovirion-based neutralization assay for detection of human papillomaviruse types 16 and 18

Vaccine 32 (2014) 5880–5887

Contents lists available at ScienceDirect

Vaccine journal homepage: www.elsevier.com/locate/vaccine

Concordance assessment between a multiplexed competitive Luminex immunoassay, a multiplexed IgG Luminex immunoassay, and a pseudovirion-based neutralization assay for detection of human papillomaviruse types 16 and 18 Darron Brown a,∗ , Martin Müller b , Peter Sehr c , Michael Pawlita b , Hanna Seitz b,1 , Ivonne Rubio b,2 , Joseph Antonello d , David Radley d,3 , Christine Roberts d , Alfred Saah d a

Department of Internal Medicine, Indiana University School of Medicine, 645N. Barnhill Drive, Indianapolis, IN 46202, USA Infections and Cancer Program, German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany c EMBL-DKFZ Chemical Biology Core Facility, European Molecular Biology Laboratory, Meyerhofstraße 1, 69117 Heidelberg, Germany d Merck & Co., Inc., 1 Merck Dr, Whitehouse Station, NJ 08889, USA b

a r t i c l e

i n f o

Article history: Received 20 December 2013 Received in revised form 22 May 2014 Accepted 6 August 2014 Available online 19 August 2014 Keywords: Vaccine Antibodies HPV

a b s t r a c t There are two approved vaccines against anogenital human papillomaviruses (HPV) and a nine-valent vaccine is currently under development. Although there are several assays available to measure antibodies elicited by HPV vaccines, there is currently no global standard for HPV antibody assays. In the current study, antibody responses to HPV16 and HPV18 among young men and women vaccinated with a quadrivalent HPV6/11/16/18 (qHPV) vaccine were assessed using three assays: a competitive Luminex immunoassay (cLIA-4) which measures antibodies directed against a single neutralizing epitope, an immunoglobulin G Luminex immunoassay (IgG-9) which measures both neutralizing and non-neutralizing antibodies, and a pseudovirion-based neutralization assay (PBNA) which functionally measures the full spectra of neutralizing antibodies. To assess HPV16 and HPV18 responses, 648 and 623 serum samples, respectively, were selected from three prior clinical trials of the qHPV vaccine. For each HPV type, the functional relationship between pairs of assay methods was estimated using a linear statistical relationship model and Pearson correlation coefficients. For both HPV16 and HPV18, the agreement between the PBNA and IgG-9 (correlation coefficients of 0.95 and 0.93, respectively) was comparable to the agreement between the cLIA-4 and IgG-9 (correlation coefficients of 0.92 and 0.92, respectively). Of 478 and 399 post-dose 3 samples that tested positive in the cLIA-4, 100% and 98% also tested positive in the IgG-9 and PBNA. The proportion of cLIA-4 seronegative post-dose 3 samples that tested positive in both the IgG-9 and PBNA was 68% (19/28) for HPV16 and 58% (71/122) for HPV18. The data demonstrate the three assays are highly correlated and reflect the measurement of neutralizing antibody. This further verifies that the IgG-9 assay, which is used to assess the immune response to an investigational nine-valent vaccine, is similarly sensitive to the PBNA for the detection of HPV16 and HPV18 neutralizing antibodies. © 2014 Elsevier Ltd. All rights reserved.

Abbreviations: BPV, bovine papilloma virus; cLIA-4, HPV6, 11, 16 & 18 competitive Luminex immunoassay; cLIA-9, HPV6, 11, 16, 18, 31, 33, 45, 52 & 58 competitive Luminex immunoassay; DKFZ, Deutsches Krebsforschungszentrum Laboratory (Heidelberg, Germany); HPV, human papillomavirus; IgG-9, HPV6, 11, 16, 18, 31, 33, 45, 52 & 58 immunoglobulin G Luminex immunoassay; LLOQ, lower limit of quantitation; LSR, linear statistical relationship; mAb, monoclonal antibody; mAb-PE, phycoerythrin-labeled monoclonal antibody; mMU/mL, milliMerck unit per milliliter; MS, microsphere (used in Luminex assays); PBNA, pseudovirion-based neutralization assay; PE, phycoerythrin; PPD VBL, PPD Vaccines & Biologics Laboratory, Wayne, PA; qHPV, quadrivalent HPV vaccine; SC, serostatus cutoff; VLP, virus-like particle; VLP-MS, virus-like particles coupled to Luminex microspheres. ∗ Corresponding author. Tel.: +1 317 274 1425; fax: +1 317 274 1587. E-mail addresses: [email protected] (D. Brown), [email protected] (M. Müller), [email protected] (P. Sehr), [email protected] (M. Pawlita), [email protected] (H. Seitz), [email protected] (I. Rubio), joseph [email protected] (J. Antonello), david.radley@pfizer.com (D. Radley), christine [email protected] (C. Roberts), alfred [email protected] (A. Saah). 1 Current address: National Institutes of Health, NCI/CCR/LCO, 37 Convent Dr, Room 4112, Bethesda, MD 20892, USA. 2 Current address: Steinhofweg 100, 69123 Heidelberg, Germany. 3 Curent address: Pfizer Inc, 500 Arcola Road, Collegeville, PA 19426, USA. http://dx.doi.org/10.1016/j.vaccine.2014.08.004 0264-410X/© 2014 Elsevier Ltd. All rights reserved.

D. Brown et al. / Vaccine 32 (2014) 5880–5887

1. Introduction Approximately seven years ago, two vaccines were approved by the United States Food and Drug Administration and the European Medicines Agency for the prevention of HPV infection and related disease: a bivalent HPV16/18 vaccine (Cervarix) and a quadrivalent HPV6/11/16/18 vaccine (Gardasil, abbreviated as qHPV vaccine). In addition, an investigational nine-valent vaccine targeting HPV6/11/16/18, as well as five of the next most frequent HPV types found in cervical cancers worldwide (HPV31/33/45/52/58) [1] is currently under development (Merck, V503, NCT00543543). The vaccines are composed of virus-like-particles (VLPs), which are made by expressing the L1 major capsid protein of the respective-HPV-types in eukaryotic cells (qHPV vaccine) or a Baculovirus expression system (Cervarix). Vaccination with L1 VLPs induces a complex polyclonal response that generates a large number of antibodies directed against specific conformational and linear epitopes displayed on the VLP surface [2–4]. The mechanism of protection is believed to depend largely upon the induction of neutralizing antibodies directed against L1 surface loops of the viral capsid; however, to date, there is no established immune correlate of protection, nor an antibody threshold that correlates with protection against HPV infection or disease. In the clinical trials of the qHPV vaccine, antibodies to the L1 VLPs were measured by a competitive Luminex immunoassay (cLIA-4) [5,6]. This type-specific assay measures antibody binding to a single neutralizing epitope for each HPV-type VLP, but does not measure complete antibody binding. Thus, the cLIA measures only a subset of the total immune response to qHPV vaccination. In the pivotal phase 3 clinical trials in women, it was noted that 4 years after the primary vaccination series, the seropositivity rates for HPV18 as measured by the cLIA-4 showed about 35% of subjects had undetectable antibodies, yet vaccine efficacy though 4 years post-vaccination for HPV18-related disease was 98.4%, with only 1 case in the qHPV vaccine group [7]. This suggests that the cLIA-4 was not optimized for sensitivity to all neutralizing antibodies induced by the qHPV vaccine. A newer immunoglobulin G Luminex immunoassay (IgG-9), which also utilizes a Luminex microsphere platform, was subsequently developed and validated [8]. The IgG-9 is a more sensitive assay that measures a broader subset of the total immune response to vaccination; however, it does not distinguish between neutralizing and non-neutralizing antibody binding. In a prior study where sera from women vaccinated with the qHPV vaccine were tested in both the cLIA-4 and IgG-9, seropositivity rates at 4 years post-dose 1 using the cLIA-4 were 98.5% and 64.8% for HPV16 and HPV18, respectively, whereas for the IgG-9, seropositivity rates were 100% and 96.7%, respectively [9]. The study illustrated potential important differences in serologic assays utilized in the clinical trials of the two currently available HPV vaccines, and the potential under-representation of the complete VLP-induced protective antibody response by the cLIA-4. Both a nine-plexed version of the cLIA (cLIA-9) and the IgG-9 assay are being used to assess the immune responses of the investigational nine-valent HPV vaccine. The current ‘gold standard’ for measuring the full spectra of neutralizing anti-HPV antibodies is a manually performed pseudovirion-based neutralization assay (PBNA) [10]. This manual assay is tedious and variable, preventing its use in large-scale clinical trials. Recently however, a highly sensitive, automated, high-throughput PBNA with excellent reproducibility and little run-to-run variability was developed for HPV types 16/18/31/45/52/58 and bovine papillomavirus type 1 [11]. We conducted a concordance study of the cLIA-4, IgG-9, and PBNA to assess the relationship among the three assays for HPV16 and

5881

Table 1 Number of samples by HPV type, study protocol, and time point* . Protocol

Time point

Protocol-011†

Month 0 (baseline) Month 7 Month 24 Month 48 Month 0 (baseline) Month 7 Month 24 Month 48 Month 0 (baseline) Month 7 Month 24 Month 36

Protocol-019†

Protocol-020†

Total‡

HPV16

HPV18

37 63 50 52 50 49 57 85 33 52 48 72

24 52 69 56 40 54 69 62 30 43 65 59

648

623

*

Baseline samples were intentionally selected to fall in the vicinity of the cLIA serostatus cutoff. Non-baseline samples were taken after completion of a 3-dose vaccination regimen. It is not the case that the different bleed intervals are from the same subjects. † Protocol-011 was a safety and immunogenicity study of the qHPV vaccine in 16- to 23-year-old females when administered alone or concomitantly with a recombinant hepatitis B vaccine [12]. Protocol-019 was a safety, efficacy and immunogenicity study of the qHPV vaccine in 24- to 45-year-old females [13], and Protocol-020 was a safety, efficacy and immunogenicity study of the qHPV vaccine in 16- to 27-year-old males [14]. ‡ A total of 24 HPV16 PBNA sera were excluded from the evaluations due to testing error (non-transfer of sera due to a misplacement of the pipet tips on the robotic instrument).

HPV18, and to determine if samples testing negative in the cLIA-4, but positive in the IgG-9, have neutralizing antibodies. 2. Methods 2.1. Study objectives The primary objective was to compare the serologic response to the qHPV vaccine in the cLIA, IgG and PNBA. The secondary objective was to potentially broaden the number and type of serological assays available for anti-HPV detection. For the purpose of this study, the comparisons among assays were restricted to HPV16 and 18. 2.2. Sample selection Serum samples were selected from three prior phase 3 clinical trials of the qHPV vaccine (Protocols-011, -019, and -020), as described in Table 1 [12–14]. In each of the three trials, those randomized to the qHPV vaccine arms received intramuscular injections at day 1, months 2 and 6. Within each trial, samples were selected from among four time points: (1) day 1 (i.e. the baseline sample); (2) month 7 (i.e. one month post-dose 3); (3) month 24; and (4) either month 36 (for Protocol-020) or month 48 (for Protocols-011 and -019). Samples were intentionally selected to span a broad range of antibody response based on the anti-HPV16 and anti-HPV18 antibody concentrations as originally measured in the cLIA-4 assay [12–14]. It bears noting that the samples within each of the bleed intervals were not selected so as to be from the same subjects, as it was not the intent of this study to characterize antibody levels across time within each of the assays. Rather, the intent of this assessment was to compare assays throughout the entire range of response, and the extensive range in antibody response was accomplished through the inclusion of the different bleed intervals. It also bears noting that the baseline samples included in this evaluation were not a random selection, but rather their selection was weighted such that the proportion of samples having antibody concentrations in the region of the cLIA-4’s lower limit of quantititation (LLOQ) and serostatus cutoff (SC) was much

5882

D. Brown et al. / Vaccine 32 (2014) 5880–5887

higher than what exists in the general clinical trial population. This was done to strengthen the comparisons among assays in the neighborhood of the serostatus cutoffs, a region of primary clinical interest. Consequently, the percentage of cLIA-4 baseline positive samples in this set (approximately 50% for HPV16 and 30% for HPV18) far exceeds what would have been obtained had a random selection of baseline samples been taken (i.e. the cLIA-4 seropositivity rate in the general clinical trial population of approximately 17,000 women was 11% for HPV16 and 4% for HPV18). 2.3. cLIA-4 This assay measures antibody binding to a single neutralizing epitope for each HPV type as previously described [5,6,9]. A set of four distinct Luminex microspheres are coated with the respective VLP and incubated with the test serum and phycoerythrin (PE)-labeled monoclonal antibodies (mAb). Fluorescent signal from the PE-labeled mAbs bound to each VLP-microsphere is measured on a Luminex (or equivalent) instrument. Displacement of the PE-labeled mAb binding is an indirect measure of human serum antibody binding to the monitored neutralizing epitopes and is quantitated relative to the displacement produced by the reference standard. Titers are reported in milliMerck Units per milliliter (mMU/mL). This assay was developed by Merck and is currently performed at PPD Vaccines and Biologics Lab (Wayne, PA, USA). In this study, the original cLIA values from the respective clinical trials were utilized [12–14]. 2.4. IgG-9 A nine-valent IgG was developed utilizing yeast-derived L1 VLP of HPV types 6/11/16/18/31/33/45/52/58 coupled to a set of nine distinct fluorescent Luminex microspheres, as previously described [8,9] and is performed by PPD Vaccines and Biologics Lab. The assay is a direct binding immunoassay that measures HPV-specific antibodies to epitopes on VLPs from an individual serum sample. Antibody concentrations are determined in a multiplexed, direct binding format by measuring the amount of VLP-specific IgG bound to each VLP-microsphere following incubation with human serum. Fluorescent signal from an anti-human IgG detection antibody conjugated to PE that binds directly to serum IgG subclass 1-4 bound to each VLP-microsphere, is measured on a Luminex (or equivalent) instrument. For each HPV type, binding of the anti-human IgG-PE to serum IgG bound to the VLP-microsphere is referenced against that obtained for a human reference standard serum. 2.5. PBNA The purpose of the PBNA is to detect the presence of antibodies capable of inhibiting infection of HPV pseudovirions. The original PBNA [10] was adapted to a high-throughput setting by developing a purely add-on system in which the serial dilution of serum samples is separated from the cell-based assay, providing a high degree of flexibility [11]. This assay was developed and performed by Deutsche Krebsforschungszentrum (DKFZ) laboratories, Heidelberg, Germany. Pseudovirions were produced at DKFZ laboratories by co-transfecting the 293TT human embryonic kidney cell line with an expression plasmid encoding the HPV L1 and L2 capsid genes and another encoding luciferase from the marine copepod Gaussia princeps. Pseudovirions of L1/L2 self-assemble and package the luciferase reporter plasmid within. Pseudovirions are incubated with HeLaT K4 cells and when pseudovirions enter cells, Gaussia luciferase is expressed and secreted into the cell culture supernatant. If neutralizing antibodies are present in the test sera, infection of cells by pseudovirions and subsequent expression of luciferase reporter is inhibited. The addition of luciferase

substrate, coelenterazine, to the reaction results in luminescence when luciferase is present in the cell culture supernatant. Bovine papillomavirus (BPV) pseudovirion assays are run as a control to verify that the test serum is not toxic to the cells, which can mimic neutralization. The LLOQ for the HPV16 and HPV18 PBNA is a reciprocal dilution of 40. A sample is classified as seronegative if the PBNA titer is <50; seropositive if the PBNA titer is ≥50 and ≥2 times the BPV titer; or serostatus indeterminate if the PBNA titer is ≥50 and <2 times the BPV titer. The PBNA serostatus cutoffs were determined utilizing a panel of likely HPV-negative clinical samples comprised of 64 sera from 9 to 12 year old adolescents. 2.6. Statistics Separate analyses were performed for HPV16 and HPV18. All quantitative comparisons were performed on the log-transformed data and were limited to the subset of sera having a quantifiable result in the assays being compared. For each HPV type, the functional relationship between pairs of assay methods (i.e. cLIA-4 vs. IgG-9, PBNA vs. cLIA-4 and PNBA vs. IgG-9) was estimated separately by study protocol and time point using the linear statistical relationship (LSR) model of Tan and Iglewicz [15]. In comparing measurements between two assays, the LSR model, also referred to as an errors-in-variables model, is a regression model that recognizes that measurement error is present in both of the assays being compared. In contrast, standard regression models account for the presence of measurement error in just one of the two assays and regard the measures from the other assay as having been obtained exactly, without error. Failure to account for measurement error in both assays results in a biased (i.e. inaccurate) determination of the relationship between assays. The Pearson correlation coefficient was also calculated for each protocol and time point. Qualitative comparisons between assay methods were based on serostatus assignment using the respective SCs. For each protocol and time point, 2 × 2 cross-classification tables were generated and the agreement rate (proportion of double positive and double negative samples relative to the total number of samples) was reported. The agreement in serostatus assignment between assays was also assessed using Cohen’s Kappa coefficient. The statistical significance of the kappa estimate was determined using an exact test. Additionally, the level of imbalance in the discordant samples between assay methods was assessed for statistical significance using an exact McNemar’s test. The agreement measures and the assessment of imbalance among discordant pairs were also determined for the combined set of assay results. 3. Results 3.1. Quantitative comparisons A total of 648 and 623 samples were selected to assess HPV16 and HPV18 responses, respectively (Table 1). In the quantitative comparisons (Fig. 1), for both HPV16 and HPV18, the cLIA-4 and IgG-9 were strongly associated, and the PBNA was strongly associated with both the cLIA-4 and IgG-9. For all samples combined, the correlation coefficient computed on the log-transformed data ranged between 0.92 and 0.95 across the three assays for both HPV types. Additionally, the estimates of slope from the fitted LSR model were all near 1, indicating strong agreement between assays. For both HPV16 and HPV18, the strong agreement among the three assays was consistent across protocols (Supplementary Tables 1–3) and post-dose 3 time points (Supplementary Table 2), with the majority of correlations falling between 0.85 and 0.95. Among the post-dose 3 time points, the lowest correlations of 0.65

D. Brown et al. / Vaccine 32 (2014) 5880–5887

5883

Fig. 1. (A) Quantitative comparison of cLIA and IgG; (B) PBNA and cLIA; and (C) PBNA and IgG by time point. Samples testing below the LLOQ (indicated in red) in the respective assays were excluded from the correlation and regression calculations.

5884

D. Brown et al. / Vaccine 32 (2014) 5880–5887

Table 2 Previously established and LSR estimated serostatus cutoffs and lower limits of quantitation (LLOQ).

Serostatus cutoff HPV16 HPV18 LLOQ HPV16 HPV18

cLIA (mMU/mL)

IgG (mMU/mL)

Estab.

Estab.

cLIA LSR Equiv.

Estab.

PBNA titer cLIA LSR Equiv.

IgG LSR Equiv.

20 24

7* 10*

14 24

50† 50†

107 140

53 52

11 10

5* 10*

7 10

40† 40†

54 52

38 52

* The results of the additional testing, which were combined with the results from the initial validation study, resulted in a serostatus cutoff of 7 and 10 mMU/mL for HPV16 and HPV18, respectively, compared to 8 and 5 mMU/mL as previously reported [8,9]. After the initial validation study [8,9], the LLOQs were reassessed based on the results of a follow-up experiment designed specifically for that purpose by including a larger panel of HPV-negative and low positive samples, and multiple preparations of the VLP microspheres. The reassessed LLOQs are 5 and 10 mMU/mL for HPV16 and HPV18, respectively. † The LLOQ for the HPV16 and HPV18 PBNA is a reciprocal dilution of 40. A sample is classified as seronegative if the PBNA titer is <50; seropositive if the PBNA titer is ≥50 and ≥2 times the BPV titer; or serostatus indeterminate if the PBNA titer is ≥50 and <2 times the BPV titer.

and 0.66 were observed for comparisons involving the HPV16 cLIA on month 7 sera (Supplementary Table 4–6). The relatively weaker correlation for that subset appears to be driven by approximately 15 sera samples having low cLIA-4 concentrations in comparison to their corresponding PBNA titers and IgG-9 concentrations, as shown in Fig. 1A. The inconsistency was restricted to the cLIA-4 test results as no such difference was seen between the PBNA and IgG-9 assays for these particular sera. As expected, the association among the assays was notably weaker for the day 1 samples than for the post-dose 3 samples, with the correlation coefficients ranging from 0.10 to 0.59, due to their proximity to the assays’ LLOQ and their limited range in antibody concentrations across the set. 3.2. Qualitative comparisons Based on the fits of the LSR model, the cLIA-4 corresponding LLOQ and SC values were estimated for the IgG-9, and the cLIA-4 and IgG-9 corresponding LLOQ and SC values were determined for the PBNA (Table 2). While the LLOQ values are quite similar between IgG-9 and cLIA-4, the SC in the IgG-9 is approximately half that of the LSR equivalent for the cLIA-4 for HPV16 (7 vs. 14 mMU/mL), and approximately five-twelfths for HPV18 (10 vs. 24 mMU/mL). Correspondingly, for both HPV16 and HPV18, the seropositivity rate in the IgG-9 exceeded that of the cLIA-4 (91% vs. 86% for HPV16, and 89% vs. 70% for HPV18) for all time points combined. The observed number of samples classified as positive in the IgG-9 and negative in the cLIA-4 was significantly greater than the observed number classified as positive in the cLIA-4 and negative in the IgG-9 (40 vs. 11, p < 0.0001 for HPV16; 130 vs. 9, p < 0.0001 for HPV18) (Table 3a). For both HPV16 and HPV18, the applied PBNA LLOQ of 40 corresponded fairly closely to the equivalent LLOQ values in the cLIA-4 and IgG-9 (Table 2). Four of the 624 samples (0.6%) were classified as serostatus-indeterminate in the HPV16 PBNA, and 31/623 (5.0%) of samples were classified as serostatus-indeterminate in the HPV18 PBNA. The difference in the number of PBNA indeterminate samples between HPV16 and HPV18 is primarily because of the higher titers against HPV16 that make the HPV16 samples more likely to be above the two times the BPV SC condition for positivity. Using the PBNA SC of 50, the resulting agreement rate in HPV16 and HPV18 serostatus assignment was 93% and 83% between the PBNA and cLIA-4 (Table 3b). The HPV16 and HPV18 positivity rates in the PBNA exceeded that of the cLIA-4 (91% vs. 87% and 88% vs. 72%), and the observed number of samples classified as positive in the PBNA and negative in the cLIA-4 for both genotypes was significantly greater than the observed number classified as positive in the cLIA-4 and negative in the PBNA (p < 0.0001 for both genotypes). The resulting agreement rate in HPV16 and HPV18 serostatus assignment was 96% and 92% between the PBNA and IgG-9

(Table 3c). The HPV16 positivity rate in the PBNA was equivalent to that in the IgG-9 (91%), and there was no evidence of imbalance in the assignment of the discordant outcomes as 13 samples were classified as PBNA positive/IgG-9 negative and 14 samples were classified as PBNA negative/IgG-9 positive (p = 1.0). All but 2 of the 27 discordances were day 1 samples. The HPV18 positivity rate in the PBNA was similar to that in the IgG-9 (88% vs. 89%), and there was also no imbalance in the assignment of the discordant outcomes (p = 0.14). The difference in the cLIA-4 positivity rate compared to the IgG-9 and PBNA was investigated by comparing the IgG-9 and PBNA serostatus outcomes based on the cLIA-4 serostatus outcome (Table 4). For HPV18, excluding PBNA indeterminate samples, 102 post-dose 3 samples were seronegative in the cLIA-4. Of these, 71 were seropositive in both the IgG-9 and PBNA and 10 tested negative in both the IgG-9 and PBNA, resulting in an agreement rate between the IgG-9 and PBNA of 79% for this subset of cLIA negative samples. Good agreement between the IgG-9 and PBNA was also observed for the 61 baseline samples that tested negative in the cLIA-4 and had a determinate serostatus assignment in the PBNA, with an agreement rate of 74%. For the entire set of 163 HPV18 cLIA-4 negative/PBNA determinate samples, the positivity rate for the IgG-9 and PBNA were quite comparable, with the IgG-9 rate only slightly exceeding that for the PBNA (67% vs. 59%). Though fewer samples were seronegative in the cLIA-4 for HPV16, comparable rates of agreement were obtained between the PBNA and IgG-9 with the overall positivity rate in the IgG-9 being slightly greater than that of the PBNA (45% vs 41%) (Table 4a). For both HPV16 and HPV18, a sizeable fraction of the cLIA-4 seronegative post-dose 3 samples had positive outcomes in both the IgG-9 and PBNA (19/28 = 68% for HPV16 and 71/122 = 58% for HPV18). As shown in Table 4b, for both HPV16 and HPV18, nearly all of the post-dose 3 samples that tested positive in the cLIA-4, also tested positive in both the IgG-9 and PBNA (478/478 [100%] for HPV16, and 399/407 [98%] for HPV18). This was less so the case for the day 1 samples that were cLIA-4 seropositive, whereby 46/60 (77%) and 17/28 (61%) of samples that were cLIA-4 positive for HPV16 and HPV18 had positive outcomes in both the IgG-9 and PBNA.

4. Discussion Several assays which measure antibodies elicited by HPV L1 VLP vaccines are available. Some immunoassays are designed to measure a broad humoral response to VLP vaccination, such as the IgG-9, and others, such as the cLIA-4 and cLIA-9, are restricted to measuring a single, conformational, neutralizing mAb for each HPV

D. Brown et al. / Vaccine 32 (2014) 5880–5887

5885

Table 3 Summary of the qualitative comparison between the assays by time interval. a. IgG vs. cLIA No. of samples with IgG/cLIA result of:

HPV16

HPV18

Time (months)

No. samples tested

+/+

−/−

+/−

−/+

IgG % positive

cLIA % positive

Agreement rate %

Kappa coefficient

Kappa P value

McNemar P value

0 7 24 36 48 Overall 0 7 24 36 48 Overall

120 164 155 72 137 648 94 149 203 59 118 623

49 163 155 58 124 549 22 148 140 38 78 426

41 1 0 4 2 48 39 0 8 3 8 58

19 0 0 10 11 40 27 1 53 18 31 130

11 0 0 0 0 11 6 0 2 0 1 9

57 99 100 94 99 91 52 100 95 95 92 89

50 99 100 81 91 86 30 99 70 64 67 70

75 100 100 86 92 92 65 99 73 69 73 78

0.50 1.0 NE 0.39 0.25 0.61 0.31 0.0 0.15 0.18 0.24 0.35

<0.001 0.006 NE 0.001 0.008 <0.001 0.001 1.0 0.001 0.041 0.006 <0.001

0.201 1.0 1.0 0.002 0.001 <0.001 <0.001 1.0 <0.001 <0.001 <0.001 <0.001

b. PBNA* vs. cLIA No. of samples with PBNA/cLIA result of:

HPV16

HPV18

Time (months)

No. samples tested

+/+

−/−

+/−

−/+

I/−

I/+

PBNA % positive

cLIA % positive

Agreement rate %

Kappa coefficient

Kappa P value

McNemar P value

0 7 24 36 48 Overall 0 7 24 36 48 Overall

118 154 153 72 127 624 94 149 203 59 118 623

51 153 153 58 114 529 21 148 139 38 76 422

44 1 0 3 1 49 43 0 13 6 5 67

13 0 0 11 10 34 18 1 40 12 25 96

8 0 0 0 0 8 4 0 2 0 1 7

1 0 0 0 2 3 5 0 8 3 9 25

1 0 0 0 0 1 3 0 1 0 2 6

55 99 100 96 99 91 45 100 92 89 94 88

51 99 100 81 91 87 29 99 73 68 72 72

82 100 100 85 92 93 74 99 78 79 76 83

0.64 1.0 NE 0.31 0.15 0.66 0.47 0.0 0.30 0.40 0.20 0.48

<0.001 0.007 NE 0.006 0.088 <0.001 <0.001 1.0 <0.001 0.001 0.006 <0.001

0.383 1.0 1.0 0.001 0.002 <0.001 0.004 1.0 <0.001 0.001 <0.001 <0.001

c. PBNA* vs. IgG No. of samples with PBNA/IgG result of:

HPV16

HPV18

Time (months)

No. samples tested

+/+

−/−

+/−

−/+

I/−

I/+

PBNA % positive

IgG % positive

Agreement rate %

Kappa coefficient

Kappa P value

McNemar P value

0 7 24 36 48 Overall 0 7 24 36 48 Overall

118 154 153 72 127 624 94 149 203 59 118 623

53 153 153 68 123 550 30 149 176 50 95 500

38 1 0 3 1 43 34 0 7 3 1 45

11 0 0 1 1 13 9 0 3 0 6 18

14 0 0 0 0 14 13 0 8 3 5 29

1 0 0 0 0 1 2 0 0 0 2 4

1 0 0 0 2 3 6 0 9 3 9 27

55 99 100 96 99 91 45 100 92 89 94 88

58 99 100 94 98 91 50 100 95 95 93 89

78 100 100 99 99 96 74 100 94 95 90 92

0.56 1.0 NE 0.85 0.66 0.74 0.49 NE 0.53 0.64 0.10 0.61

<0.001 0.007 NE <0.001 0.016 <0.001 <0.001 NE <0.001 0.001 0.340 <0.001

0.69 1.0 1.0 1.0 1.0 1.0 0.524 1.0 0.227 0.25 1.0 0.144

I/− (+) denotes PBNA indeterminate, comparator assay negative or positive * A total of 24 HPV16 PBNA sera excluded from the evaluations due to testing error (non-transference of sera due to a misplacement of the pipet tips on the robotic instrument).

type. Given the use of both the cLIA-4 and the IgG-9 in the clinical trials of the HPV vaccines, it was important to determine if samples testing negative in the cLIA-4, but positive in the IgG-9, have neutralizing antibodies present. Our results demonstrate that the cLIA-4, IgG-9, and PBNA are highly correlated and the two former assays accurately measure neutralizing antibody. There is currently no global standard for HPV antibody assays. Despite the establishment of WHO international antibody units for HPV16 and HPV18, we report the cLIA-4 and IgG-9 antibody responses in mMU/mL rather than International Units (IU) per mL, as international units are not available for the other two HPV types included in the cLIA-4 nor for the other seven types included

in the cLIA-9 and IgG-9 assays which were developed for use in large Phase 3 studies of an investigational 9-valent HPV vaccine [16]. The use of mMU/ML is also consistent with the published literature as well as global regulatory documents, allowing for a direct comparison of antibody titers across studies. Notably, however, the WHO HPV16 and HPV18 antibody reference standards have been tested in their respective PBNA assays. The WHO HPV16 standard, at an antibody concentration of 10 IU/mL, neutralized specifically HPV16 in the HPV16 PBNA with a titer of 463 [11]. The HPV18 standard, at an antibody concentration of 16 IU/mL, neutralized specifically HPV18 in the HPV18 PBNA with a titer of 579 [11]. Using the LSR estimates obtained in this study, the HPV16

5886

D. Brown et al. / Vaccine 32 (2014) 5880–5887

Table 4 Summary of the qualitative comparison between the IgG and PBNA by time point for cLIA serostatus negative (a) and serostatus positive (b) samples. a. cLIA serostatus negative No. of samples with PBNA/IgG result of:

HPV16

HPV18

Time (months)

No. samples tested

+/+

−/−

+/−

−/+

I/−

I/+

PBNA % positive

IgG % positive

Agreement rate %

Kappa coefficient

Kappa P value

McNemar P value

0 7 24 36 48 Overall 0 7 24 36 48 Overall

58 1 0 14 13 86 66 1 61 21 39 188

7 0 – 10 9 26 13 1 38 12 20 84

33 1 – 3 1 38 32 0 6 3 1 42

6 0 – 1 1 8 5 0 2 0 5 12

11 0 – 0 0 11 11 0 7 3 4 25

0 0 0 0 0 0 2 0 0 0 2 4

1 0 0 0 2 3 3 0 8 3 7 21

23 0 – 79 91 41 30 100 75 67 83 59

32 0 – 71 82 45 39 100 85 83 80 67

70 100 – 93 91 77 74 100 83 83 70 77

0.25 NE – 0.81 0.62 0.53 0.43 NE 0.47 0.57 0.0 0.52

0.087 NE – 0.011 0.182 <0.001 0.001 NE 0.002 0.025 1.0 <0.001

0.332 1.0 – 1.0 1.0 0.648 0.210 1.0 0.180 0.250 1.0 0.047

b. cLIA serostatus positive No. of samples with PBNA/IgG result of:

HPV16

HPV18

Time (months)

No. samples tested

+/+

−/−

+/−

−/+

I/−

I/+

PBNA % positive

IgG % positive

Agreement rate %

Kappa coefficient

Kappa P value

McNemar P value

0 7 24 36 48 Overall 0 7 24 36 48 Overall

60 153 153 58 114 538 28 148 142 38 79 435

46 153 153 58 114 524 17 148 138 38 75 416

5 0 0 0 0 5 2 0 1 0 0 3

5 0 0 0 0 5 4 0 1 0 1 6

3 0 0 0 0 3 2 0 1 0 1 4

1 0 0 0 0 1 0 0 0 0 0 0

0 0 0 0 0 0 3 0 1 0 2 6

86 100 100 100 100 99 84 100 99 100 99 98

83 100 100 100 100 98 76 100 99 100 99 98

86 100 100 100 100 99 76 100 99 100 97 98

0.48 NE NE NE NE 0.55 0.26 NE 0.49 NE −0.01 0.36

0.002 NE NE NE NE <0.001 0.234 NE 0.028 NE 1.0 <0.001

0.727 1.0 1.0 1.0 1.0 0.727 0.688 1.0 1.0 1.0 1.0 0.754

PBNA titer of 463 corresponds to 73.0 mMU/mL in the HPV16 cLIA-4 and 56.0 mMU/mL in the HPV16 IgG-9 assay. Correspondingly, the HPV18 PBNA titer of 579 corresponds to 85.0 mMU/mL in the HPV18 cLIA-4 and 81.7 mMU/mL in the HPV16 IgG-9 assay. Using these relationships, the conversion factor between mMU/mL and IU/mL for the HPV16 reference standard was determined as 1 mMU/mL equals 0.137 IU/mL in the cLIA-4, and 0.179 IU/mL in the IgG-9 assay. Similarly, the conversion factor between mMU/mL and IU/mL for the HPV18 reference standard was determined as 1 mMU/mL equals 0.188 IU/mL in the cLIA-4, and 0.196 IU/mL in the IgG-9 assay. The cLIA-4 measures antibodies that displace a monoclonal antibody directed against a particular neutralizing epitope of a specific-HPV-type. For example, the particular neutralizing epitope and relevant mAb selected for HPV18 (18.J4), was selected for its analytical specificity, that is, its ability to distinguish HPV18 antibody response from response to other HPV types that are phylogenetically related to HPV18 (such as HPV45), over its sensitivity and the immuno-dominance of the epitope. The inability of the cLIA-4 to measure other potentially relevant epitopes may result in a less-sensitive assay in comparison to the IgG-9 and PBNA. The same antibodies used for HPV6, 11, 16 and 18 in the cLIA-4 are used in the cLIA-9. To assess this issue, we tested a total 648 samples with the three assays. For the PBNA and the cLIA-4, the agreement rate in HPV18 serostatus assignment was 83%. The HPV18 positivity rate in the PBNA exceeded that of the cLIA-4 (88% vs.72%), and the observed number of samples classified as positive in the PBNA and negative in the cLIA-4 was significantly greater than the

observed number classified as positive in the cLIA-4 and negative in the PBNA. This verifies that the cLIA-4 assay is less sensitive than the PBNA when using their respective independently assigned SCs for the detection of HPV16 or HPV18 neutralizing antibodies. For the PBNA and the IgG-9, the resulting agreement rate in HPV18 serostatus assignment was 92%. The HPV18 positivity rate in the PBNA was similar to that of the IgG-9 (88% vs. 89%), and there was no statistical evidence of imbalance in the assignment of the discordant outcomes. This verifies that the IgG-9 is similarly sensitive to the PBNA when using their respective independently assigned SCs for the detection of HPV16 or HPV18 neutralizing antibodies. The difference in positivity rate between the cLIA-4 and IgG-9 for HPV16 and HPV18 was further investigated by comparing the IgG9 and PBNA serostatus outcomes based on the cLIA-4 serostatus outcome. For HPV18, the agreement rate between the IgG-9 and PBNA for the cLIA-4 negative post-dose 3 samples was 79%. For both HPV16 and HPV18, nearly all of the post-dose 3 samples that tested positive in the cLIA-4 also tested positive in both the IgG-9 and PBNA assays. In conclusion, the rates of positivity and the agreement between the IgG-9 and the PBNA in sera that were negative in the cLIA-4 confirm the presence of neutralizing antibody in cLIA-4 negative sera for HPV18. These findings corroborate the high efficacy that was observed across all the phase 3 studies of the qHPV vaccine in young women, older women, and males whereby HPV18 seronegative subjects showed continued high protection against cervical, vulvar, vaginal, penile, and anal disease [12–14].

D. Brown et al. / Vaccine 32 (2014) 5880–5887

5887

5. Author contributions

Acknowledgements

DB set up study sites and enrolled participants into the quadrivalent vaccine clinical trials. MM, MP, and PS designed the HT-PBNA and analyzed the data. PS performed the HT-PBNA. HS and IR produced and characterized pseudovirions under supervision of MM and contributed to the assay design. JA assisted in the design and statistical analysis of the concordance study with emphasis on test sample selection (including numbers of samples within each study and bleed interval, selection based cLIA-4 titer and desired titer representation, and blinding and randomization of test samples). CR assisted in the design of the concordance study and coordinated the specimen identification and laboratory testing with DKFZ and PPD VBL. AS managed the sponsor’s operations. DR developed and instituted the data analysis plan. All authors made substantial contributions the drafting the work or revising it critically for important intellectual content, gave final approval of the submitted version, and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

We thank Heather L Sings (Merck) for assistance in the preparation of this manuscript. We are grateful to Kerstin Putzker (EMBL-DKFZ) for excellent technical assistance (EMBL-DKFZ).

Conflicts of interest statement DB has served on an Advisory Board at Merck and Co., Inc. and has lectured on the quadrivalent HPV vaccine (honoraria received from Merck and Co., Inc. are donated to charities). His laboratory has received research funding from Merck and Co., Inc. Indiana University and Merck and Co., Inc. have an agreement that pays the University, based on certain landmarks related to vaccine development. DB receives a portion of these funds as income. MM reports having received financial support to develop and perform high-throughput neutralization assays from Merck. He also reports having received royalties paid by Loyola University of Chicago in regard to sales of the Cervarix Vaccine. In addition, MM has a patent application for high-throughput pseudovirion-based neutralization assay pending, and a patent in regard to production of chimeric virus-like particles with royalties paid. PS has a patent pending relevant to this work (WO2011151335A1). MP reports having received financial support to develop and perform high-throughput neutralization assays from Merck. In addition, MP has a patent application for high-throughput pseudovirion-based neutralization assay pending. HS reports nothing to disclose. IR reports having received financial support to develop and perform high-throughput neutralization assays from Merck. In addition, IR has a patent application for high-throughput pseudovirion-based neutralization assay pending. JA, DR, CR, and AS are current or former Merck employees and hold stock/stock options. Funding This study was funded by Merck and Co.

Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.vaccine. 2014.08.004. References [1] International Agency for Research on Cancer 2013. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans: Vol 100, A Review of Human Carcinogens. Part B: Biological Agents. Lyon, France: International Agency for Research on Cancer; 2011. [2] Orozco JJ, Carter JJ, Koutsky LA, Galloway DA. Humoral immune response recognizes a complex set of epitopes on human papillomavirus type 6 l1 capsomers. J Virol 2005;79:9503–14. [3] Christensen ND, Dillner J, Eklund C, Carter JJ, Wipf GC, Reed CA, et al. Surface conformational and linear epitopes on HPV-16 and HPV-18 L1 virus-like particles as defined by monoclonal antibodies. Virology 1996;223: 174–84. [4] Carter JJ, Wipf GC, Madeleine MM, Schwartz SM, Koutsky LA, Galloway DA. Identification of human papillomavirus type 16 L1 surface loops required for neutralization by human sera. J Virol 2006;80:4664–72. [5] Dias D, Van Doren J, Schlottmann S, Kelly S, Puchalski D, Ruiz W, et al. Optimization and validation of a multiplexed luminex assay to quantify antibodies to neutralizing epitopes on human papillomavirus 6, 11, 16 and 18. Clin Diagn Lab Immunol 2005;12:959–69. [6] Opalka D, Lachman CE, MacMullen SA, Jansen KU, Smith JF, Chirmule N, et al. Simultaneous quantitation of antibodies to neutralizing epitopes on virus-like particles for human papillomavirus types 6, 11, 16 and 18 by a multiplexed luminex assay. Clin Diagn Lab Immunol 2003;10:108–15. [7] Joura EA, Kjaer SK, Wheeler CM, Sigurdsson K, Iversen OE, Hernandez-Avila M, et al. HPV antibody levels and clinical efficacy following administration of a prophylactic quadrivalent HPV vaccine. Vaccine 2008;26: 6844–51. [8] Opalka D, Matys K, Bojczuk P, Green T, Gesser R, Saah A, et al. Multiplexed serologic assay for nine anogenital human papillomavirus types. Clin Vaccine Immunol 2010;17:818–27. [9] Brown DR, Garland SM, Ferris DG, Joura E, Steben M, James M, et al. The humoral response to Gardasil over four years as defined by total IgG and competitive Luminex immunoassay. Hum Vaccin 2011;7:230–8. [10] Pastrana DV, Buck CB, Pang YY, Thompson CD, Castle PE, FitzGerald PC, et al. Reactivity of human sera in a sensitive, high-throughput pseudovirusbased papillomavirus neutralization assay for HPV16 and HPV18. Virology 2004;321:205–16. [11] Sehr P, Rubio I, Seitz H, Putzker K, Ribeiro-Muller L, Pawlita M, et al. Highthroughput pseudovirion-based neutralization assay for analysis of natural and vaccine-induced antibodies against human papillomaviruses. PLoS One 2013;8:e75677. [12] Wheeler CM, Bautista OM, Tomassini JE, Nelson M, Sattler CA, Barr E. Safety and immunogenicity of co-administered quadrivalent human papillomavirus (HPV)-6/11/16/18 L1 virus-like particle (VLP) and hepatitis B (HBV) vaccines. Vaccine 2008;26:686–96. [13] Munoz N, Manalastas R, Pitisuttihum P, Tresukosol D, Monsonego J, Ault K, et al. Safety, immunogenicity, and efficacy of quadrivalent HPV (types 6, 11, 16, 18) recombinant vaccine in adult women between 24 and 45 years of age: a randomized, double-blind trial. Lancet 2009;373: 1949–57. [14] Giuliano AR, Palefsky JM, Goldstone S, Moreira Jr ED, Penny ME, Aranda C, et al. Efficacy of quadrivalent HPV vaccine against HPV infection and disease in males. N Engl J Med 2011;364:401–11. [15] Tan CY, Iglewicz B. Measurement-methods comparisons and linear statistical relationship. In: Technometrics Vol. 41; 1999. No. 3. [16] Roberts C, Green T, Hess E, et al. Development of a human papillomavirus competitive luminex immunoassay for nine HPV types. Hum Vaccin Immunother 2014;10 [Epub ahead of print], Pubmed ID 24955587.