Immune responses against SARS-CoV-2 variants after two and three doses of vaccine in B-cell malignancies: UK PROSECO study

Abstract

Patients with hematological malignancies are at increased risk of severe COVID-19 outcomes due to compromised immune responses, but the insights of these studies have been compromised due to intrinsic limitations in study design. Here we present the PROSECO prospective observational study ( NCT04858568 ) on 457 patients with lymphoma that received two or three COVID-19 vaccine doses. We show undetectable humoral responses following two vaccine doses in 52% of patients undergoing active anticancer treatment. Moreover, 60% of patients on anti-CD20 therapy had undetectable antibodies following full vaccination within 12 months of receiving their anticancer therapy. However, 70% of individuals with indolent B-cell lymphoma displayed improved antibody responses following booster vaccination. Notably, 63% of all patients displayed antigen-specific T-cell responses, which increased after a third dose irrespective of their cancer treatment status. Our results emphasize the urgency of careful monitoring of COVID-19-specific immune responses to guide vaccination schemes in these vulnerable populations.

Fig. 1. CONSORT diagram of study cohort.

A total of 592 patients were assessed for eligibility and subsequently 578 were recruited into the study. At the time of data cutoff, blood samples were processed from 457 participants. The vaccination schedule and numbers of participants sampled at the following time points are shown: pre-vaccination (pre-D1), 4 weeks after first dose (post-D1), 2–4 weeks after second dose (post-D2), 20–26 weeks after second dose (pre-D3) and 4–8 weeks after third dose (post-D3).

Fig. 2. Antibody responses but not cellular responses to SARS-CoV-2 vaccination are impaired by systemic therapy.

a, Participants with lymphoma vaccinated with ChAdOx1 vaccine while on systemic treatment or within 24 weeks of treatment completion (designated ‘on treatment’) had reduced antibody levels compared to participants with previous or untreated lymphoma (designated ‘no treatment’) and healthy controls (designated ‘healthy donors’). The dashed line and shaded region indicate undetectable antibody. GMC is shown; two-sided Kruskal–Wallis test, Dunn’s test for multiple comparisons, ****P < 0.0001. b, As in a but with BNT162b2 where the same phenomenon is observed. c, Reduced anti-S IgG antibody levels in HL, aggressive B-NHL (A B-NHL) and indolent B-NHL (I B-NHL) participants in the ‘on treatment’ group and in I B-NHL participants in the ‘no treatment’ group despite two vaccine doses. GMC is shown; two-sided Kruskal–Wallis test, Dunn’s test for multiple comparisons, *P = 0.0288, **P = 0.0008, ***P = 0.0004, ****P < 0.0001. d, Anti-S levels in participants with I B-NHL are similar between those who had received previous antisystemic therapy compared to treatment-naive. GMC is shown; two-sided Kruskal–Wallis test. P <0.05 was considered significant. e, Univariable logistic regression showing that low serum IgG, anti-CD20 exposure 12 months preceding the first vaccine dose; ‘on treatment’ group and older age (represented by ‘Age’ as a continuous variable) were associated with an increased OR of having undetectable anti-S IgG. Participants with HL favored detectable anti-S IgG. n = 428 participants. Error bars represent 95% CI. *P = 0.005, **P = 0.001, ***P < 0.001. f, Similar IFN-γ cellular responses in participants with lymphoma between ‘on’ and ‘no treatment’ groups. The dashed line and shaded region indicate a negative IFN-γ response. Median and 95% CI are shown; two-sided Kruskal–Wallis test, Dunn’s test for multiple comparisons, *P = 0.0132, **P = 0.0023, ****P < 0.0001. g, Reduced IFN-γ cellular responses were preserved in participants with lymphoma except for those with I B-NHL in the ‘no treatment’ group. The dashed line and shaded region indicate a negative IFN-γ response. Median and 95% CI are shown; two-sided Kruskal–Wallis test, Dunn’s test for multiple comparisons, *P = 0.0356, **P = 0.0262, ***P = 0.0093, ****P = 0.0010, *****P < 0.0001. Source data

Fig. 3. The effect of specific therapies on antibody and cellular responses.

a, Anti-S IgG levels in participants vaccinated while receiving anti-CD20 (on) or <6, 7–12, 13–24 or >24 months of completion of treatment. The dashed line and shaded region indicate undetectable antibody. GMC is shown; two-sided Kruskal–Wallis test, Dunn’s test for multiple comparisons, *P = 0.0057, **P = 0.0060. b, Preservation of cellular responses in anti-CD20-treated lymphoma participants, regardless of treatment timing. All points below the dashed line represent a negative test. Medians are shown; two-sided Kruskal–Wallis test, P < 0.05 was considered significant. c, A reduced peripheral blood B-cell count is associated with impaired anti-S IgG level in aggressive B-NHL. GMC is shown; two-sided Mann–Whitney test, ****P < 0.0001. d, As in c but for indolent B-NHL; two-sided Mann–Whitney test, ****P < 0.0001. e, Anti-S IgG levels in participants vaccinated ≤3 months of autologous stem cell transplant have reduced anti-S. GMC is shown; two-sided Kruskal–Wallis test, Dunn’s test for multiple comparisons, *P = 0.0112. Matching antigen-specific T-cell responses are also shown. f, Participants with HL have reduced anti-S when vaccinated within the 3-month period before (designated ‘before’) commencing chemotherapy. GMC is shown; two-sided Kruskal–Wallis test. Dunn’s test for multiple comparisons, *P = 0.0455. Antigen-specific T-cell responses are detectable in participants regardless of timing of chemotherapy. Source data

Fig. 4. Correlates of antigen-specific T-cell responses.

a, Lack of correlation between antigen-specific IFN-γ T-cell responses and anti-S levels. n = 191 participants with paired T-cell and anti-S levels. b, ChAdOx1 vaccination induces higher IFN-γ T-cell responses than BNT162b2. All points below the dashed line represent a negative result. Medians are shown; two-sided Mann–Whitney test, *P = 0.0005. c, Similar CD4⁺ T-cell counts were observed between ChAdOx1 and BNT162b2 recipients. Medians are shown; two-sided Mann–Whitney test, P < 0.05 was considered significant. d, Similar CD8⁺ T-cell counts were observed between ChAdOx1 and BNT162b2 recipients. Medians are shown; two-sided Mann–Whitney test, P < 0.05 was considered significant. e, ChAdOx1 vaccination was associated with a lower cellular response on multivariable logistic regression analysis. Error bars represent 95% confidence intervals, n = 191 participants, *P = 0.031. f, Bendamustine (Benda) treatment was associated with a trend toward reduced IFN-γ T-cell responses. All points below the dashed line represent a negative test. Medians are shown; two-sided Kruskal–Wallis test. P < 0.05 was considered significant. g, No difference in CD4⁺ T-cell counts between participants on or recently treated with anti-CD20, BTK inhibitor (BTKi) or bendamustine. Medians are shown; two-sided Kruskal–Wallis test. P < 0.05 was considered significant. h, No difference in CD4⁺ T cells was observed in bendamustine-exposed individuals who did not have an antigen-specific T-cell response (NR) compared to those who had a positive test (R); two-sided Kruskal–Wallis test. NR, non-responder; R, responder. P < 0.05 was considered significant. i, Pie chart showing that the majority of participants with both negative antibody and cellular responses were from the ‘on treatment’ group. j, Bar chart showing that within the double-negative ‘on treatment’ group in i, most participants had received combination therapy containing anti-CD20 rituximab. Source data

Fig. 5. Anti-S level correlates with ACE2 receptor inhibition against wild-type SARS-CoV-2.

a, Correlation between anti-S IgG and ACE2 receptor inhibition; two-sided Spearman test. b, Data from a but grouped according to ACE2 receptor inhibition titers; two-sided Kruskal–Wallis test, Dunn’s test for multiple comparisons, ****P < 0.0001. Source data

Fig. 6. Antibody and pseudoneutralization responses to third vaccine dose.

a, Antibody responses to the third vaccine dose according to therapy; concurrent or anti-CD20 therapy within 6 months in B-NHL (anti-CD20), concurrent BTK inhibitors or venetoclax in B-NHL (BTKi/VEN) and concurrent chemotherapy in HL (chemo), compared to those not on treatment (I B-NHL and A B-NHL and HL); two-sided Wilcoxon test, ****P < 0.0001. b, IFN-γ T-cell responses in healthy donors and participants with lymphoma who had a negative result (initial NR) or positive result (initial R) after two vaccine doses; two-sided Wilcoxon test, *P = 0.0361, **P = 0.0122. c, Good agreement is observed between binding of participant IgG to the wild-type SARS-CoV-2 and Omicron, Delta, Alpha, Beta and Gamma variants, undertaken in pre-D3 and post-D3 samples. The antibody concentration for each variant (value on inside of the y axis) equivalent to anti-S wild-type IgG 400 BAU ml⁻¹ is shown; two-sided Spearman correlation and simple linear regression. Source data

Extended Data Fig. 1. Correlation between anti-S IgG and ACE2 reception inhibition post third vaccine dose.

Good correlation was observed between anti-S IgG antibody levels in post-D3 samples and ACE2 reception inhibition. Spearman correlation test, two-tailed. Source data