We recruited patients via an informed consent process. Patients were required to be >18 years of age and have a cancer diagnosis either on active treatment or requiring active surveillance. Patients were also required to have received two doses of the mRNA COVID-19 vaccine or one dose of the adenoviral vaccine prior to enrollment. After drawing baseline labs that included spike antibody, a sample for T-cell assay, and a biobank sample, patients received a third mRNA vaccine (initially BNT162b2 per protocol, which was later amended to allow for third mRNA-1273 vaccine after the Food and Drug Administration [FDA] authorized ‘booster’ doses in fall of 2021). Patients who had received Ad26.CoV2.S vaccine received a BNT162b2 vaccine. The patients then returned for follow-up 4 weeks and 4–6 months after their third dose and their labs were repeated (Figure 1—figure supplement 1).

We have previously reported preliminary findings of a 56% seroconversion rate after third dose of vaccine patients with cancer who did not have a detectable immune response after two doses (Shapiro et al., 2022). For patients who did not seroconvert after three doses or had low antibody response (<1000 AU/mL as determined by our in-house assay, Abbott), we hypothesized whether a ‘mix and match’ strategy with fourth dose of COVID-19 vaccine would induce seroconversion/improved boosting of the humoral antibody responses. To study this, we designed a protocol wherein patients who had received three prior doses of mRNA vaccines and had undetectable anti-S antibody or had an anti-S antibody level of <1000 AU/mL measured at least 14 days after third dose would be randomized to an mRNA vs. adenoviral fourth vaccine dose. Responses would be then assessed at 4 weeks after the fourth dose through measurement of anti-S antibody results. We also measured complete blood counts (CBC), quantitative immunoglobulin levels (IgG, IgA, and IgM), lymphocyte subsets, T-cell responses, and neutralization activity at baseline and 4 weeks for each of these patients. Following the implementation of this protocol, the Centers for Disease Control (CDC) published a statement that advised that the mRNA vaccines should be preferentially administered over the adenoviral vaccines given concern over rare side effects such as thrombocytopenia and thrombosis syndrome. Following this advisory, we amended our protocol to allow recruitment in a cohort that would receive a fourth dose of the BNT162b2 vaccine to comply with CDC guidelines (Figure 1—figure supplement 1).

The AdviseDx SARS-CoV-2 IgG II assay was used for the assessment of anti-S IgG antibody. AdviseDx is an automated, two-step chemiluminescent immunoassay performed on the Abbott i1000SR instrument. The assay is designed to detect IgG antibodies directed against the receptor binding domain (RBD) of the S1 subunit of the spike protein of SARS-CoV-2. The RBD is a portion of the S1 subunit of the viral spike protein and has high affinity for the angiotensin converting enzyme 2 (ACE2) receptor on the cellular membrane (Pillay, 2020; Yang et al., 2020) The procedure, in brief, is as follows. Patient serum containing IgG antibodies directed against the RBD is bound to microparticles coated with SARS-CoV-2 antigen. The mixture is then washed of unbound IgG and anti-human IgG, acridinium-labeled, secondary antibody is added and incubated. Following another wash, sodium hydroxide is added and the acridinium undergoes an oxidative reaction, which releases light energy which is detected by the instrument and expressed as relative light units (RLU). There is a direct relationship between the amount of anti-spike IgG antibody and the RLU detected by the system optics. The RLU values are fit to a logistic curve which was used to calibrate the instrument and expresses results as a concentration in AU/mL (arbitrary units/milliliter) (conversion for spike antibody titers from AU/mL to BAU/mL: based on the results from the first WHO International Standard study, which demonstrated a strong correlation with the current standardization of the SARS-CoV-2 IgG II Quant assays, the mathematical relationship of the Abbott AU/mL unit to WHO unit [binding antibody unit per mL (BAU/mL)] would follow the equation: BAU/mL = 0.142*AU/mL). This assay recently has shown high sensitivity (100%) and positive percent agreement with other platforms including a surrogate neutralization assay (Bradley et al., 2021) and also demonstrated high specificity both in the post-SARS-CoV-2 infection and post-vaccination settings. The cutoff value for this assay is 50 AU/mL with <50 AU/mL values reported as negative and the maximum value is 50,000 AU/mL.

The SARS-CoV-2 Surrogate Virus Neutralization Test Kit was used to measure antibodies that inhibit the interaction between viral RBD and ACE2 receptor. This test kit uses purified human ACE2 (hACE2) protein-coated enzyme-linked immunosorbent assay (ELISA) plates and HRP-conjugated RBD to monitor the presence of circulating antibodies in samples, including peripheral/capillary blood, serum, and plasma, which block the interaction of RBD-HRP with ACE2 with excellent correlation with the gold standard live virus plaque reduction neutralization test.

The kit contains two key components: RBD-HRP and hACE2. The protein-protein interaction between RBD-HRP and hACE2 is disrupted by neutralizing antibodies against SARS-CoV-2 RBD, if present in a sample. After mixing the sample dilutions with the RBD-HRP solution, components are allowed to bind to the RBD. The neutralization antibody complexed to RBD-HRP remains in the supernatant and is removed during washing, The yellow color of the hACE2-coated wells is determined by the RBD HRP binding to the hACE2-coated wells after incubation with TMB, followed by a stop solution. After the addition of the stop solution, a light-yellow color results from blocking agents interacting with RBDs and inhibiting hACE2 interactions.

Microneutralization assays were performed in a biosafety level 3 facility at the Icahn School of Medicine at Mount Sinai (ISMMS) as previously described (Carreño et al., 2022). Briefly, Vero E6-TMPRSS2 cells were seeded in 96-well cell culture plates at 20,000/well in complete Dulbecco’s Modified Eagle Medium (cDMEM). The following day, heat-inactivated serum samples were serially diluted (threefold) starting at a 1:10 dilution in 1× MEM (10× minimal essential medium [Gibco], 2 mM L-glutamine, 0.1% sodium bicarbonate [Gibco], 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid [HEPES; Gibco], 100 U/mL penicillin, 100 µg/mL streptomycin [Gibco], and 0.2% bovine serum albumin [MP Biomedicals]) supplemented with 10% fetal bovine serum (FBS). The virus diluted at 10,000 tissue culture infectious dose 50% (TCID₅₀) per mL of 1× MEM was added to the serum dilutions and incubated for 1 hr at room temperature (RT). After removal of cDMEM from Vero E6 cells, 120 μL/well of the virus-serum mix were added to the cells and plates were incubated at 37°C for 1 hr. Mix was removed and 100 μL/well of each corresponding serum dilutions were added in a mirror fashion to the cell plates. Additional 100 μL/well of 1× MEM 1% FBS (Corning) were added to the cells. Plates were incubated for 48 hr at 37°C and fixed with a 10% paraformaldehyde solution (PFA, Polysciences) for 24 hr at 4°C.

For staining, plates were washed with 200 μL of PBS. Cells were permeabilized with 150 μL/well PBS containing 0.1% Triton X-100 for 15 min at RT. Plates were washed 3× with PBST and blocked with 3% milk (American Bio) in PBST for 1 hr at RT. Blocking solution was removed and 100 μL/well of the biotinylated mAb 1C7C7 anti-SARS nucleoprotein antibody (generated at the Center for Therapeutic Antibody Development at the ISMMS) were added at 2 μg/mL for 1 hr at RT. Plates were then washed 3× with PBST and the secondary antibody goat anti-mouse IgG-HRP (Rockland Immunochemicals) was added at 1:3000 in blocking solution for 1 hr at RT. Plates were washed 3× with PBST, and SIGMAFAST OPD (o-phenylenediamine, Sigma-Aldrich) was added for 10 min at RT. The reaction was stopped with 50 μL/well 3 M hydrochloric acid to the mixture. Optical density (OD₄₉₀) was measured on an automated plater reader (Sinergy 4, BioTek). The inhibitory dilution 50% were calculated as previously described (Amanat et al., 2020).

We previously reported outcomes for 88 patients enrolled into this study (Shapiro et al., 2022). Here, we present our final results for the complete cohort of 106 patients that were enrolled into this study for assessment of the primary endpoint of response at 4 weeks as well as 47 patients who completed 4–6 month follow-up. The baseline characteristics of this cohort are summarized in Table 1. The median age was 68 years (63.25–76.5 years). Fifty-five percent (58/106) of patients were female and 45% (48/106) were male. Our cohort was ethnically diverse and included 34% (36/106) Caucasian, 31% (33/106) African-American, 25% (27/106) Hispanic, and 8% (9/106) Asian patients. Majority of patients had received mRNA vaccines at baseline. Sixty-eight percent (72/106) received BNT162b2, 26% (28/106) received mRNA-1273, and 6% (6/106) had received Ad26.CoV2.S. Seventy-four percent of patients (78/106) received a booster BNT162b2 vaccine and 26% (28/106) patients received booster mRNA-1273 vaccine. The majority of the patients, 62% (66/106), had a hematologic malignancy and 38% (40/106) had a solid tumor diagnosis. Further breakdown of cancer type and cancer status is summarized in Table 1. The majority of patients, 75% (80/106), were being actively treated at the time of receiving the third dose of the vaccine.

Thirty-three percent of the patients (35/106) were seronegative after two doses. At 4 weeks following the receipt of the booster vaccine, 57% (20/35) of these patients seroconverted and had a detectable antibody response as demonstrated by anti-S antibody testing, meeting the primary endpoint of our study. The median titer at baseline (after primary vaccination) for the entire cohort was 212.1 AU/mL (IQR 50–2873 AU/mL) and the median titer at 4 weeks (after third dose of the vaccine) for the entire cohort was 9997 AU/mL (IQR 880.7–47,063 AU/mL) (Figure 1A). The median rise in anti-S titer for patients with hematologic malignancies was 2167 AU/mL (IQR 0–10,131 AU/mL) versus 31,010 AU/mL (IQR 9531–44,464 AU/mL) in patients with solid malignancies (p<0.001). Within the hematologic malignancies, patients with lymphoid cancers had a lower rise in median anti-S titers (1169 AU/mL, IQR 0–8661 AU/mL) compared to those with myeloid malignancies; median anti-S titer 9424 AU/mL (IQR 4381–20,444 AU/mL) (p<0.001) (Figure 1B).

We further investigated the association of specific anti-cancer therapies with the booster effect. Patients on Bruton’s tyrosine kinase inhibitor (BTKi) therapy (n=12) had a median rise in anti-S antibody of 0 AU/mL (IQR 0–3393 AU/mL) compared to a median rise of 9355 AU/mL (IQR 877.3–34410 AU/mL) in anti-S antibody for patients not on BTKi (p<0.05) (Figure 1C). Patients on anti-CD20 antibody therapy (n=25) also had a median rise in anti-S antibody level of 0 AU/mL (IQR 0–910.5 AU/mL) compared to a median rise of 12,735 AU/mL (IQR 2842–38,863 AU/mL) in patients that did not receive anti-CD20 antibody therapy (p<0.05). (Figure 1D). Nine patients had a history of SARS-CoV-2 infection and in this cohort the rise in anti-S titers was higher (median 19,350 AU/mL, IQR 9286–32,151 AU/mL), compared to those who did not have prior SARS-CoV-2 infection with a median anti-S titer rise, 6706 AU/mL (IQR 444.1–33,831 AU/mL) (Figure 1E). We also observed that the rise in anti-S titer at 4 weeks was higher for patients who received an mRNA-1273 booster compared to BNT162b2 booster; median 31,451 AU/mL vs. 5534 AU/mL, respectively (Figure 1F). This observation was not, however, statistically significant. Lastly, we also investigated the association of age with spike antibody response at 4 weeks. The median spike antibody titer for patients <65 years of age was 27,451 AU/mL and the median patients with age ≥65 years was 6152 AU/mL. This result was significant at p value 0.03438. These results are also summarized in Table 2.

We also studied T-cell immune responses through a SARS-CoV-2 IGRA. At baseline (i.e. after primary vaccination), 88 patients had evaluable T-cell results and a positive T-cell response against SARS-CoV-2 was seen in 74% (65/88) patients. Of these 65 patients, 21 patients were seronegative for anti-S antibody at baseline. At 4 weeks (after third dose), 89 patients had evaluable T-cell results and a positive result was seen in 85% (76/89) patients. Of the 15 patients with negative anti-S antibody at 4 weeks, 11 had a positive T-cell response. Fourteen patients who had a negative T-cell assay response at baseline had a positive T-cell response at 4 weeks. Anti-S titer showed a positive correlation with T-cell response at baseline and at 4 weeks for this cohort (p<0.001) (Figure 1G). These results are summarized in Table 2.

Eighteen patients were enrolled into the fourth dose study. Median age for this cohort was 69.5 years (IQR 65.5–73.8). Thirty-nine percent (7/18) were seronegative at baseline (after three doses, pre-fourth dose) and 61% patients (11/18) were sero-low (anti-S ab <1000 AU/mL). All patients had hematologic malignancies in this cohort and the breakdown of diagnoses is provided in Table 3. Eighty-three percent of the patients (15/18) received BNT162b2 fourth booster shots and 17% (3/18) patients received Ad26.CoV2.S as their fourth booster vaccine. In addition, we also measured CBC, lymphocyte subsets, immunoglobulin G, A, and M (quantitative Ig) levels at baseline (pre-fourth dose) and 4 weeks (post-fourth dose) (Table 4). The median time between second and third vaccination was 167 days (5.5 months) and that between third and fourth vaccination was 155 days (5.1 months).

A patient was classified as a responder if they (1) had positive anti-S antibody at 4 weeks if seronegative at baseline (after three doses) or (2) if they achieved a titer of >1000 AU/mL at 4 weeks if they were sero-low at baseline (after three doses). As such, we observed a 67% response rate (12/18) in patients for the fourth dose cohort. Two of seven seronegative patients seroconverted to positive anti-S antibody at 4 weeks with a seroconversion rate of 29% in this cohort. All sero-low patients (11/11) responded with an IgG level >1000 after the fourth dose (Figure 2A). For the whole cohort, the median anti-S antibody at baseline (after three doses) was 131.1 AU/mL (<50–432.9 AU/mL) and at 4 weeks (after fourth dose) was 1700 AU/mL (IQR 64.3–18,627 AU/mL). The two patients that seroconverted after fourth dose both had a diagnosis of Waldenstrom’s macroglobulinemia. Both patients had received anti-CD20 antibody and chemotherapy as part of their treatment. One patient was actively on a BTK inhibitor and the second patient was off active treatment at the time of study participation. We further investigated association of baseline laboratory values, such as CBC, lymphocyte subsets, and quantitative Ig levels and observed that patients in the responder group had higher baseline IgM (60.5 mg/dL) compared with the non-responder group (median 17 mg/dL, p<0.001) (Figure 2B). Additionally, we also observed that the median CD19+ cell count was significantly lower in the non-responder group versus the responder group (1 vs. 113, p=0.04). No patients were on intravenous immunoglobulin at the time of study participation. These results are summarized in Table 5.

T-cell activity was assessed at baseline (pre-fourth dose) and at 4 weeks (post-fourth dose) using the SARS-CoV-2 IGRA. At baseline, 14 patients had evaluable T-cell responses and a positive response was noted in 79% patients (11/14). Of these, three patients had negative anti-S antibody at baseline. At 4 weeks after the fourth dose, a positive T-cell response was seen in 17/18 (94%) patients. These results are summarized in Table 5.

We also assessed neutralization activity at baseline (pre-fourth dose) and at 4 weeks (post-fourth dose) against WT and Omicron (B.1.1.529, BA.1). The results are summarized in Table 5. Overall, neutralization activity was seen in 67% patient samples at baseline and in 72% patient samples at 4 weeks. Strikingly, neutralization activity against Omicron was absent in all patient samples at baseline, however became detectable in 33% (6/18) patients at 4 weeks after the fourth dose. The titer of anti-S antibody correlated with neutralization activity at baseline and at 4 weeks against the WT virus (p<0.001) (Figure 2C). We also observed correlation between the titer of anti-S antibody with neutralization activity at 4 weeks for the Omicron variant (Figure 2D).

The observation for baseline IgM correlating with response to the fourth dose of the COVID-19 vaccine led us to perform an exploratory analysis to assess if IgG and IgA levels would also correlate with the response. Given that our fourth dose cohort was small, we performed this exploratory analysis by combining the baseline immunoglobulin levels for the baseline seronegative cohort for the third dose study (n=35) and baseline immunoglobulin levels for the fourth dose study (n=18). In this exploratory analysis, we observed that the median levels for all immunoglobulin subtypes were lower in patients who either did not seroconvert after the third dose or did not respond to the fourth dose (IgA 49 mg/dL vs. 116.5 mg/dL, IgM 16.6 mg/dL vs. 48.3 mg/dL, IgG 488 mg/dL vs. 759.5 mg/dL with p values of 0.05, 0.002, and 0.006, respectively [Kruskal Wallis test]).