Response to “Progressive Decrease in the Potential Usefulness of Meningococcal Serogroup B Vaccine (4CMenB, Bexsero®) in Gipuzkoa, Northern Spain” (Pérez-Trallero, Esnal and Marimón, PloS ONE, 26 December 2014).

Dear PLoS,

We are compelled to comment on the recent article “Progressive Decrease in the Potential Usefulness of Meningococcal Serogroup B Vaccine (4CMenB, Bexsero®) in Gipuzkoa, Northern Spain” (Pérez-Trallero, Esnal and Marimón, PLoS ONE, 26 December 2014), and highlight the many concerns we have on the content of this paper and on the inadequacy of the methodology applied to predict 4CMenB strain coverage.

As a preface and in the interest of full disclosure, all signatories to this commentary are employees or associates of Novartis group companies, the manufacturer of 4CMenB, with the exception of JAV, who is the head of the Spanish Reference Laboratory for Meningococci, and he is not an employee of or associated with Novartis group companies.

In this PloS ONE article, Pérez-Trallero and colleagues analyzed 82 meningococcal strains isolated in 2008–13 in a small region of Spain (Gipuzkoa Province) in terms of their 4CMenB vaccine antigen genetic profiles. Conclusions as to predicted vaccine strain coverage, and consequently a recommendation against vaccination in this region, rest heavily on the assumption that the presence or absence of certain alleles of genes encoding the fHbp, NHBA, NadA and PorA P1.4 antigens of 4CMenB is a valid surrogate of vaccine strain coverage. Specifically, the authors erroneously assume that only strains harboring genes encoding the vaccine homologous fHbp B1, NHBA, NadA and PorA antigens would be covered. On the contrary, we have argued that such an exercise is unwarranted, not in the least because it fails to fully take into account the significant contribution made by antigen cross-reactivity, density and accessibility on the bacterial cell surface, in engendering a strain susceptible to killing by vaccine immune sera.

Due to logistical and ethical considerations, determinations of strain coverage by meningococcal serogroup B (MenB) vaccines cannot routinely be assessed by testing all vaccinated subjects against all, or even a representative panel of endemic MenB strains in the serum bactericidal assay (SBA). To overcome these limitations, we developed the Meningococcal Antigen Typing System (MATS) to measure both the extent of antigen cross-reactivity and antigen expression in a modified ELISA sandwich format (Donnelly et al., 2010, PNAS, 107:19490-19495). Furthermore, by relating the MATS relative potency of a panel of global MenB isolates to their sensitivity to killing by pooled, 4CMenB immune sera, MATS thresholds were defined that delineate strains covered by vaccine-elicited immune sera from those not covered. The platform has been transferred to several reference laboratories worldwide, including Spain, and has been thoroughly qualified in inter-laboratory standardization studies (Plikaytis et al., 2012, Clin Vaccine Immunol, 19:1609–17). Indeed, the MATS platform was used in the initial assessment of potential MenB strain coverage in Spain based on a study of 300 strains collected in 2008–10 (69%, 95% CI: 48–85%; Vogel et al., 2013, Lancet Infect Dis. 13:416-425). By examining the presence of the genes encoding 4CMenB vaccine antigens, Pérez-Trallero and colleagues reported that 69.7% of the regional MenB strains isolated in 2008–9 would be covered, while the percentage decreased to 42.1% when strains collected in 2012–13 were considered. In the absence of confirmatory studies (e.g., investigations based on pooled sera SBA, or MATS), the possibility that such evaluations may be an under-, or indeed over-estimations of true coverage remains a distinct possibility.

Although the MATS platform remains the preferred predictive tool to measure and monitor 4CMenB strain coverage, recent data suggest that it may be a conservative predictor of actual strain coverage. When the ability of pooled infant and adolescent 4CMenB immune sera to kill a panel of 40 MenB strains collected from England and Wales in 2007–08 was tested, we observed that MATS predictions and SBA results were significantly associated (p-value = 0.022: Frosi et al., 2013, Vaccine. 31:4968-497). Importantly, MATS predicted coverage of 70% (95% CI, 55–85%) and was largely confirmed by 88% killing in pooled SBA (95% CI, 72–95%). In this study, a significant number of strains that were not predicted to be covered by MATS were in fact killed by immune sera. Indeed, MATS had a 96% positive predictive value, a 33% negative predictive value, and an overall accuracy of 78%, indicating that MATS was reasonably accurate and that any errors tended to be in the direction of under- rather than overestimated strain coverage. Similar results were observed in a recent study in which 7 out of 9 Spanish strains characterized as MATS-negative were nonetheless killed by pooled 4CMenB immune sera (Abad et al., 2015, Clin Vaccine Immunol E-pub, 28 January 2015). Importantly, in both of these studies we observed killing of strains that did not harbor any vaccine homologous antigens.
The conservative nature of MATS-based coverage predictions may be due in part to the fact that the MATS platform does not account for additive or synergistic effects by antibodies that target multiple vaccine antigens that may be sparsely distributed on the meningococcal cell surface (and might therefore be expected to score as negative in MATS). Indeed, the ability of anti-NHBA antibodies to synergize with other antibodies against sparsely distributed antigens such as fHbp and/or non-PorA components of OMV has been documented (Giuliani et al., 2010, Vaccine, 28:5023-30; Vu et al., 2011, Vaccine. 29:1968-73).

In the studies described above, pooled sera analyses were judged to be informative of responses at the individual subject level. The validity of this judgment was recently confirmed by Budroni and colleagues (Budroni et al. IPNC 2014 poster; manuscript submitted). This group generated SBA titers from individual and pooled sera from 30 infant 4CMenB vaccinees collected at different timepoints and tested against 11 antigenically diverse MenB strains. They observed that i) pooled SBA titers are linearly correlated with the arithmetic mean of the titers from the individuals composing the pool (r = 0.97, p-value < 10-3), and ii) the pooled SBA titer accurately predicts the proportion of subjects that respond with protective SBA titers (the “seroresponse rate”). Applying the relationship between the seroresponse rate vs the pooled hSBA titer established in this study to the same 40 strains used in the previous study by Frosi and colleagues, the authors observed that the seroresponse rate against MenB strains deemed covered by MATS is 77% (IQR: 66–100%). Importantly, for those strains deemed not covered by MATS, 39% (IQR: 19–46%) of subjects showed a protective SBA response.

These observations collectively support the use of MATS as a predictive tool for measuring vaccine strain coverage, even if it may result in conservative estimates.

Contrary to the methodology used by Pérez-Trallero and colleagues and the conclusions they derived from it, we argue that a more informed assessment of 4CMenB strain coverage at any time period would necessitate the use of the qualified MATS platform. The ability of MATS to provide (potentially conservative) estimates of strain coverage has been validated by several studies discussed above. Furthermore, as the authors rightly point out, meningococcal epidemiology such as that observed in Gipuzkoa Province, Spain can indeed evolve over time. Any effective MenB vaccination campaign must therefore include continuous and diligent surveillance, a critical exercise given the emergence of MenB strains devoid of certain vaccine antigens (such as fHbp, Lucidarme et al., 2011, Clin Vaccine Immunol, 18:1002–14). We would suggest that the MATS predictive tool is presently our most effective and unique means of achieving that goal.

Signed,
Philip E. Boucher1, Mariagrazia Pizza2, Julio A. Vazquez3

1PRA Health Sciences (on assignment to Novartis Vaccines and Diagnostics, Inc.)
2Novartis Vaccines and Diagnostics Srl, Siena, Italy
3Spanish Reference Laboratory for Meningococci. Institute of Health Carlos III. Majadahonda. Spain