Selective suppression of de novo SARS-CoV-2 vaccine antibody responses in patients with cancer on B cell-targeted therapy

Abstract

We assessed vaccine-induced antibody responses to the SARS-CoV-2 ancestral virus and Omicron variant before and after booster immunization in 57 patients with B cell malignancies. Over one-third of vaccinated patients at the pre-booster time point were seronegative, and these patients were predominantly on active cancer therapies such as anti-CD20 monoclonal antibody. While booster immunization was able to induce detectable antibodies in a small fraction of seronegative patients, the overall booster benefit was disproportionately evident in patients already seropositive and not receiving active therapy. While ancestral virus- and Omicron variant-reactive antibody levels among individual patients were largely concordant, neutralizing antibodies against Omicron tended to be reduced. Interestingly, in all patients, including those unable to generate detectable antibodies against SARS-CoV-2 spike, we observed comparable levels of EBV- and influenza-reactive antibodies, demonstrating that B cell-targeting therapies primarily impair de novo but not preexisting antibody levels. These findings support rationale for vaccination before cancer treatment.

Keywords: COVID-19; Cancer; Immunoglobulins; Oncology.

Figure 1. Percentages of patients without antibodies against the spike or RBD protein of ancestral SARS-CoV-2, Omicron variant, or both.

(A) Each bar depicts the percentage of patients deficient in detectable antibodies against the ancestral spike or Omicron spike or antibodies recognizing either target as determined by ELISA. (B) Percentages of patients deficient in detectable antibodies against ancestral RBD or Omicron RBD or antibodies recognizing either target. Fisher’s exact test was conducted to compare the number of seronegative patients with no active therapy and active therapy. Shown are comparisons between active therapy and no therapy for the same antibody evaluation and group. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Fisher’s exact test was also conducted to compare ancestral, Omicron, or both within individual groups, and the differences were not statistically significant.

Figure 2. Positive correlation between antibody levels against ancestral and Omicron spike and RBD post-booster.

(A) Dot plot shows the post-booster patient time points graphed by Omicron spike versus ancestral spike as determined by ELISA. Red dots depict patients on active therapy, blue dots depict patients not on active therapy, and black dots indicate patients without any detectable antibodies. (B) Same as A except showing Omicron RBD versus ancestral RBD. The Spearman’s rank correlation coefficient was calculated for patients on active therapy (red) or not on active therapy (blue) with exclusion of patients without detectable antibodies.

Figure 3. ELISA and viral neutralization assay depict variable antibody response to the ancestral and Omicron SARS-CoV-2 strains before and after booster vaccination in patients with hematological cancer.

(A–C) Area under the curve (AUC) for all patients (A), patients without active therapy (B), and patients with active therapy (C). (D–F) Same as A–C except showing the 50% neutralization titer (NT₅₀). For all data, each symbol represents a time point at which serum was isolated, and the line connects individual patients. The number above each graph represent arithmetic mean. ANOVA was used to compare between pre-booster ancestral, post-booster ancestral, pre-booster Omicron, and post-booster Omicron NT₅₀/AUC values of the same patient. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 4. B cell–depleting therapy selectively impairs de novo antibody generation.

(A) Antibody levels determined by ELISA against the ancestral RBD protein. Each patient is represented by a dot pre- and post-therapy. Patients without active therapy are indicated in blue, while patients on active therapy are indicated in red. Orange circles around blue dots indicate patients who are untreated. The arithmetic mean is shown by the bar. (B) Same as A except showing antibodies against the EBV GP350. (C) Same as A except showing antibodies against influenza H1N1. (D) Same as A except showing antibodies against the OC43 common cold coronavirus spike protein. (E) Dot plot of ancestral RBD versus EBV GP350 antibody binding taken from post-booster data. (F) Same as E except showing ancestral RBD versus influenza H1N1. (G) Same as E except showing ancestral RBD versus OC43 spike. For A–D, Wilcoxon’s signed-rank test was used to compare OD values between pre- and post-booster, while Mann-Whitney test was used to compare between different conditions (treated and nontreated), and the Bonferroni correction was applied. *P < 0.05, ***P < 0.001. For E–G, the Spearman’s rank correlation coefficient was calculated.