Understanding the treatment benefit of hyperimmune anti-influenza intravenous immunoglobulin (Flu-IVIG) for severe human influenza

Abstract

BACKGROUNDAntibody-based therapies for respiratory viruses are of increasing importance. The INSIGHT 006 trial administered anti-influenza hyperimmune intravenous immunoglobulin (Flu-IVIG) to patients hospitalized with influenza. Flu-IVIG treatment improved outcomes in patients with influenza B but showed no benefit for influenza A.METHODSTo probe potential mechanisms of Flu-IVIG utility, sera collected from patients hospitalized with influenza A or B viruses (IAV or IBV) were analyzed for antibody isotype/subclass and Fcγ receptor (FcγR) binding by ELISA, bead-based multiplex, and NK cell activation assays.RESULTSInfluenza-specific FcγR-binding antibodies were elevated in Flu-IVIG-infused IBV- and IAV-infected patients. In IBV-infected participants (n = 62), increased IgG3 and FcγR binding were associated with more favorable outcomes. Flu-IVIG therapy also improved the odds of a more favorable outcome in patients with low levels of anti-IBV Fc-functional antibody. Higher FcγR-binding antibody was associated with less favorable outcomes in IAV-infected patients (n = 50), and Flu-IVIG worsened the odds of a favorable outcome in participants with low levels of anti-IAV Fc-functional antibody.CONCLUSIONThese detailed serological analyses provide insights into antibody features and mechanisms required for a successful humoral response against influenza, suggesting that IBV-specific, but not IAV-specific, antibodies with Fc-mediated functions may assist in improving influenza outcome. This work will inform development of improved influenza immunotherapies.TRIAL REGISTRATIONClinicalTrials.gov NCT02287467.FUNDINGFunding for this research was provided by subcontract 13XS134 under Leidos Biomedical Research Prime Contract HHSN261200800001E and HHSN261201500003I, NCI/NIAID.

Keywords: Adaptive immunity; Immunology; Infectious disease; Influenza; NK cells.

Figure 1. Flow diagram depicting the subset of patients from the INSIGHT 006 Flu-IVIG clinical trial who underwent detailed serological analyses.

Of the 308 participants included in the primary analysis, 84 (27%) had influenza B and 224 (73%) had influenza A. Of the 84 participants with influenza B, 64 patients (21% of total participants) were infected with a B/Yamagata lineage influenza virus. Of the 224 participants with influenza A, 73 patients (24% of total participants) were infected with an A/H1N1 influenza virus. Based on sample availability, serological analyses were performed with 62 sera samples from B/Yamagata-infected patients (20% of total participants) and 50 sera samples from A/H1N1-infected patients (16% of total participants). In the analyzed B/Yamagata-infected participants, 29 were infused with placebo and 33 infused with Flu-IVIG. In the analyzed A/H1N1-infected participants, 26 received placebo and 24 received Flu-IVIG.

Figure 2. Treatment with Flu-IVIG improves the odds of a more favorable clinical outcome in patients hospitalized with severe B/Yamagata influenza but not A/H1N1 influenza.

Mean ordinal outcomes (A and B) and ordinal outcome distributions (C and D) on day 3 (d3), d5, and d7 after infusion are shown for the Flu-IVIG– and placebo-infused patients infected with severe B/Yamagata influenza (Flu-IVIG n = 33 and placebo n = 29; A and C) and A/H1N1 influenza (Flu-IVIG n = 24 and placebo n = 26; B and D). The Flu-IVIG/placebo odds ratios (ORs) with 95% confidence intervals for the B/Yamagata- and A/H1N1-infected participants were calculated on d3, d5, and d7 after infusion (E). A proportional odds model with adjustment for the patient’s baseline clinical status was used to compare the Flu-IVIG– and placebo-treated groups. *P <0.05; ***P <0.001.

Figure 3. Hemagglutination inhibition (HAI) titers 1 day after infusion with influenza-specific hyperimmune immunoglobulin (Flu-IVIG) in patients hospitalized with severe influenza.

Preinfusion (d0; open symbols) and postinfusion (d1; closed symbols) median HAI titers (with interquartile range) against the B/Phuket/3073/2013 virus (A) and the A/California/07/09(H1N1) virus (B) are shown for the placebo- or Flu-IVIG–infused B/Yamagata-infected (Flu-IVIG n = 33 and placebo n = 29) and A/H1N1-infected (Flu-IVIG n = 24 and placebo n = 26) patients, respectively. Analysis of covariance, with the preinfusion or d0 level as a covariate, was used to compare differences between the placebo and Flu-IVIG treatment groups on d1 after infusion. The P value represents the difference between treatment groups for log₂(d1 postinfusion titer), controlling for preinfusion (d0) titer. The dashed line represents an HAI titer of 40, which is considered an important protective threshold for influenza and is generally defined as 50% protective against influenza virus infection. ***P < 0.001.

Figure 4. FcγR-binding antibody and antibody-dependent NK cell activation 1 day after infusion with influenza-specific hyperimmune immunoglobulin (Flu-IVIG) in patients hospitalized with severe influenza.

The FcγRIIIa dimer ELISA (A and B), the FcγRIIIa and FcγRIIa dimer bead-based multiplex (C–F) and the plate-bound NK-92 cell activation assay (G and H) were used to analyze serum Fc functional antibody levels. Preinfusion (d0; open symbols) and postinfusion (d1; closed symbols) median (with interquartile range) FcγR dimer binding and NK cell activation (CD107a⁺) against the B/Phuket/3073/2013 HA (left figure panels) and the A/California/07/09(H1N1) HA (right figure panels) were measured for the placebo- or Flu-IVIG–treated B/Yamagata-infected (Flu-IVIG n = 33 and placebo n = 29) and A/H1N1-infected (Flu-IVIG n = 24 and placebo n = 26) patients, respectively. Analysis of covariance, with the preinfusion or d0 level as a covariate, was used to compare differences between the placebo and Flu-IVIG treatment groups on d1 after infusion. The P value represents the difference between treatment groups for log₂(d1 postinfusion titer, MFI, or NK cell activation), controlling for preinfusion (d0) level. ***P < 0.001.

Figure 5. Association between clinical outcome and HAI or FcγRIIIa-binding antibody titer in B/Yamagata- and A/H1N1-infected patients.

(A) Based on postinfusion (d1) serum titers, patients with severe B/Yamagata (n = 62; blue) and A/H1N1 (n = 50; red) influenza were divided into high (≥40) and low (<40) HAI titer groups irrespective of randomization to Flu-IVIG or placebo. (B) Patients with severe B/Yamagata (n = 62; blue) and A/H1N1 (n = 50; red) influenza were also divided into high (>160) and low (≤160) FcγRIIIa-binding antibody titer groups regardless of treatment group. The association between HAI or FcγRIIIa-binding antibody titer and d5 postinfusion ordinal outcomes was investigated using univariate and multivariate proportional odds regression models, with the multivariate model adjusting for baseline antibody titer, treatment group (Flu-IVIG/placebo), and risk score on d7. Odds ratios (ORs) are shown with 95% confidence intervals, with ORs greater than 1 indicating that the high-antibody-titer group has better odds of being in a more favorable clinical outcome category and ORs less than 1 favoring the low-antibody-binding group. *P <0.05.

Figure 6. Associations between influenza antibody features and clinical outcomes on day 5 after infusion in patients hospitalized with severe B/Yamagata and A/H1N1 influenza.

The association between antibody features and day 5 (d5) postinfusion ordinal outcomes were investigated using univariate and multivariate proportional odds regression models, with the multivariate model adjusting for baseline antibody level, treatment group (Flu-IVIG/placebo), and risk score on d7. Heatmaps show summary odds ratios (ORs) for patients hospitalized with B/Yamagata (n = 62; left side of heatmap) and A/H1N1 (n = 50; right side of heatmap) influenza at preinfusion (A and C) and d1 postinfusion (B and D) time points generated using univariate (A and B) and multivariate (C and D) proportional odds regression models. ORs greater than 1 indicate that patients with higher antibody levels have improved odds of being in a better outcome category on d5 after infusion and ORs less than 1 indicating that patients with lower antibody levels have improved odds of being in a better outcome category. *P <0.05, **P <0.01.

Figure 7. Treatment effect of Flu-IVIG in B/Yamagata-infected patients with below- and above-median baseline FcγRIIIa-binding antibody titers.

Placebo- (n = 29) and Flu-IVIG–treated (n = 33) B/Yamagata-infected patients were grouped by below-median (≤160) or above-median (>160) baseline (or preinfusion) FcγRIIIa-binding antibody titer. Mean ordinal outcomes (A) and ordinal outcome distributions (B) on day 3 (d3), d5, and d7 after infusion are shown for the Flu-IVIG and placebo groups with below- (≤160) or above-median (>160) baseline FcγRIIIa-binding antibody titers. (C) The Flu-IVIG/placebo odds ratios (ORs) with 95% confidence intervals are shown for the below- and above-median FcγRIIIa-binding antibody titer subgroups. A proportional odds regression model with adjustment for the patient’s baseline clinical status was used to compare the Flu-IVIG– and placebo-treated subgroups. ORs greater than 1 indicate that patients infused with Flu-IVIG have better odds of being in a more favorable clinical outcome category, whereas ORs less than one favor placebo infusion. **P <0.01.

Figure 8. Treatment effect of Flu-IVIG in A/H1N1-infected patients with below- and above-median baseline FcγRIIIa-binding antibody titers.

Placebo- (n = 26) and Flu-IVIG–treated (n = 24) A/H1N1-infected patients were grouped into below-median (≤80) or above-median (>80) baseline (or preinfusion) FcγRIIIa-binding antibody titer. Mean ordinal outcomes (A) and ordinal outcome distributions (B) on day 3 (d3), d5, and d7 after infusion are shown for the Flu-IVIG– and placebo-infused patients with below-median (≤80) or above-median (>80) baseline FcγRIIIa-binding antibody titers. (C) The Flu-IVIG/placebo odds ratios (ORs) with 95% confidence intervals are shown for the below- and above-median FcγRIIIa-binding antibody titer subgroups. A proportional odds model with adjustment for the patient’s baseline clinical status was used to compare the Flu-IVIG– and placebo-treated subgroups. ORs greater than 1 indicate that patients infused with Flu-IVIG have better odds of being in a more favorable clinical outcome category, whereas ORs less than 1 favor placebo infusion. *P <0.05.