Evaluation of Influenza Lentiviral Pseudotypes as an alternative source of antigen in the Neuraminidase Inhibition Enzyme-Linked Lectin Assay

The continuously changing epidemiology of influenza viruses needs to be yearly addressed with a fast and massive production of vaccines that prevent epidemic outbreaks and occasional pandemics worldwide. Currently licensed flu vaccines are designed to elicit antibodies against the viral hemagglutinin (HA). Formulations containing standardized amounts of neuraminidase (NA) are however gaining growing interest in the influenza vaccine field, as NA‐inhibiting (NI) antibodies have been associated with resistance to disease as well as reduced severity and duration of illness. The Neuraminidase Inhibition Enzyme-Linked Lectin Assay (NI-ELLA), which quantifies NI antibody titers based on the ability of specific antibodies to inhibit NA activity, is regarded as a reliable and high-throughput method to measure NI antibodies in human sera. To avoid the non-specific inhibition of HA-binding antibodies that may be present in serum samples, viruses containing an antigenically mismatched HA (reassortants) or Triton-X100(Tx)-treated wild-type (WT) viruses that retain NA functionality have been employed as NA source in the ELLA assay. An innovative approach that exploits NA-bearing pseudotypes (or pseudoviruses, PV) as surrogate viruses expressing the NA of interest has been recently investigated to provide a safer product to be used for the ELLA test. The first aim of this PhD work was to evaluate three different sources of NA (reassortant viruses, Tx-inactivated WT viruses, and lentiviral PV) to compare their ability in the measurement of NI-ELLA titers. In order to do that, two independent panels of sera were analyzed by ELLA assay to investigate the antibody response against influenza NA subtype N1 (A/California/07/2009 (H1N1pdm)) and N2 (A/Hong Kong/4801/2014 (H3N2)). The NI titers were measured either as the 50% endpoint or 50% inhibitory concentration (IC50) and were compared for every source of antigen. The linear regression, Spearman rank, Bland-Altman and Intraclass Correlation (ICC) analyses were employed to assess the comparability of the data obtained. Analysis of results showed that the ELLA performed well with all three sources of NA. The correlation coefficients (r2) for all comparisons were high (r2 > 0.92) when titers were reported either as endpoint or as IC50. Comparisons that included PV as antigen source demonstrated the best concordance, with 90% of N1 titers and 88% of N2 titers measured by the pseudotypes-based ELLA (PV-ELLA) showing less than 2-fold difference compared to those obtained in the assays performed with the other sources of NA. ICC results showed a high concordance between the different methods when either examining N1 or N2 titers (ICC coefficients of 0.861 and 0.817, respectively). Bland-Altman analyses revealed an overall agreement, and showed that titers measured against N2 antigens (reassortant H6N2 and H11N2 PV) were in better agreement than N1 titers measured using reassortant H6N1 and H11N1 PV antigens. For both anti-N1 and anti-N2 titers, the assays performed with the available sources of antigen show similar ability to detect NI titers ≥40, which have recently been associated with protection against influenza. The different ELLA platforms show comparable Geometric Mean Titer (GMT) values as well as similar titers range, suggesting the suitability of performing ELLA with each of the NA sources evaluated. The second aim of this project was to validate the PV-ELLA executed by using H11N1ACal/09 and H11N2HK/14 PV, following the recommendations provided by international guidelines. Both the ELLA performed with H11N2HK/14 PV and H11N1ACal/09 PV fulfill the pre-established criteria for definition of assay specificity, as the titer raised by the high positive serum sample against the homologous PV was always ≥ 4-fold higher than that of the heterologous and negative samples, which always displayed undetectable NI responses. The N1 and N2 PV-ELLA showed a strong linear trend between dilutions of the reference sample vs the respective GMT in the range of serum concentrations evaluated, with r2 coefficients always higher than 0.96 and 95% confidence intervals of the slope within the acceptability range, suggesting excellent linearity of the assay. The Geometric Mean Titers (GMTs) yielded in these linearity experiments show a maximum deviation of ± 1 titer step dilution (i.e. ≤ 2-fold difference in titers) when (i) collected within the same day or in different days, and when (ii) compared to those obtained in a pre-validation testing, suggesting precision and accuracy of the PV-ELLA. None of the variations applied to three critical steps of the assay produced a significant impact on NI titers, indicating that the ELLA performed with H11N1ACal/09 and H11N2HK/14 PV is adequately robust. In summary, the PV-ELLA performed with either N1- and N2-bearing PV has met all the acceptability criteria pre-defined for demonstration of specificity, linearity, precision, accuracy and robustness of the analytical method, confirming that the assay is suitable to measure NA antibody response in the experimental conditions evaluated. These results support the use of lentiviral pseudotypes as an attractive alternative source of NA in the ELLA assay and provide the means to standardize and validate the methodology in laboratories that do not have access to suitable reassortant viruses.