Rituximab-treated patients have a poor response to influenza vaccination

Abstract

The efficacy of influenza vaccination in patients treated with rituximab is a clinically important question. Rheumatology clinics are populated with patients receiving rituximab for a broad array of disorders. Although several studies have explored the efficacy of other vaccines in rituximab-treated populations, results have been conflicting. We wished to define influenza vaccine efficacy in a rituximab-treated cohort. We examined 17 evaluable subjects treated with rituximab for rheumatologic conditions. T cell subsets, B cells subsets, T cell function, and B cell function were evaluated at specific time points along with hemagglutinination inhibition titers after receiving the standard inactivated influenza vaccine. T cell subset counts were significantly different than controls but did not change with rituximab. B cells depleted in all patients but were in various stages of recovery at the time of vaccination. Influenza vaccine responsiveness was poor overall, with only 16 % of subjects having a four-fold increase in titer. Pre-existing titers were retained throughout the study, however. The ability to respond to the influenza vaccine appeared to be related to the degree of B cell recovery at the time of vaccination. This study emphasizes that antibody responses to vaccine are impaired in subjects treated with rituximab and supports the concept that B cell recovery influences influenza vaccine responsiveness.

Figure 1. B cell depletion kinetics

Total B cells (CD19+) were measured at baseline, 4 weeks after rituximab, the day of vaccine administration (7-9 months after rituximab administration), 2 months after vaccination and 6 months after vaccination. B cells depleted well in all patients but recovery was variable.

Figure 2. B cell subsets

B cell subsets were defined using flow cytometry. The mean and standard deviation are shown and the baseline evaluations for the controls are shown at the right of each graph. Baseline levels were compared between patients and controls. The IgM memory and switched memory subsets were different from controls at baseline (p<0.001 for both). On the day of vaccination, memory subset counts were significantly lower in patients than controls with p<0.001.

Figure 3. T cell subsets

T cell subsets were defined using flow cytometry. The mean and standard deviation are shown at each time point and the baseline evaluations for the controls are shown on the far right of each graph. Baseline levels were compared between patients and controls and the asterisks indicate results statistically different from controls. The p values for CD4 naïve and effector subsets were p=0.05 and p<0.01 respectively. For CD8 naïve and effector subsets, the p values were p=0.01 and p<0.01 respectively.

Figure 4. Antibody production

Antibody responses to the vaccine were measured using the hemagglutination inhibition assay matched for each year of the vaccine (top panel). Mean titers and standard deviation are shown in the graph. Post vaccine responses in the patients were significantly lower than controls with p<0.001 for all three serotypes. In addition, the difference between the pre- and post-vaccine titers in the patients were non-significant. The existing HAI titers in the patients varied little over the course of the study. The second panel uses a sum of the individual HAI serotypes in the graph. Each line represents an individual study patient. We also assessed global IgG production and anti-influenza responses using a B cell ELISPOT assay. Although baseline levels are lower in patients than controls, there is detectable antibody production in this assay. Nevertheless, all time points exhibited significantly less total IgG and influenza-specific responses than controls.

Figure 5. Characteristics of the responder group

Within the cohort, we compared B cell subsets on the day of vaccination between responders and nonresponders. The mean and standard error are shown. The non-responders are shown in dark grey but the values are so low, they appear superimposed on the X axis. The naïve B cell and IgM memory percentages were lower in nonresponders than responders with p<0.01 and p<0.05 respectively. The absolute counts were lower in non-responders than responders for IgM memory and switched memory B cell subsets with p<0.05 for IgM memory and P<0.01 for switched memory B cells. Although the responders had variable B cell counts, the non-responders uniformly had <1 B cell/mm³. Asterisks indicate significant differences.