Loss of anti‐spike antibodies following mRNA vaccination for COVID‐19 among patients with multiple myeloma

Abstract Background Multiple myeloma (MM) patients have variable responses to mRNA vaccination to COVID‐19. Little is known regarding their vaccine‐induced antibody levels over time. Methods We monitored spike IgG antibody levels over 24 weeks among a subset of 18 MM patients who showed a full response after two mRNA vaccinations. Results MM patients had a more rapid decline in antibody levels as compared to eight healthy controls, with power law half‐lives of 72 days (vs. 107 days) and exponential half‐lives of 37 days (vs. 51 days). The patients with longer SARS‐CoV‐2 antibody half‐lives were more likely to have undetectable monoclonal protein than those with shorter half‐lives, suggesting better disease control may correlate with longer duration of vaccine‐induced antibodies. Regardless, by 16 weeks post‐second dose of mRNA vaccination, the majority of patients had antibody levels below 250 binding arbitrary units per milliliter, which would be unlikely to contribute to preventing COVID‐19. Conclusions Thus, even MM patients who respond adequately to vaccination are likely to require more frequent booster doses than the general population.


| INTRODUCTION
Vaccination for COVID-19 reduces risk of severe disease and hospitalization. 1 Unfortunately, immunocompromised individuals are less likely to develop a protective response to vaccination while simultaneously experiencing more complications from COVID-19. [2][3][4] One such group is patients with multiple myeloma (MM), a plasma cell dyscrasia resulting in excess production of monoclonal antibodies. These patients show impaired humoral immunity with reduced levels of uninvolved immunoglobulins; treatments for MM-typically geared toward suppressing malignant plasma cells-have off-target effects that may further reduce B-cell function and uninvolved antibody production. As a result, MM patients are at increased risk for infection. 5 In our recently published study, only 45% of patients fully responded to mRNA vaccination for COVID-19, which was considered a level above the 6th percentile of healthy controls. 3 Thus, most patients remain susceptible to breakthrough infections despite vaccination.
Unfortunately, widespread vaccination has failed to halt transmission, 6 presenting an ongoing SARS-CoV-2 exposure risk for this patient population.
Quantitative anti-spike antibody levels can be used to monitor vaccination responses with widespread use despite still-yet-undefined cutoffs for protection. Moreover, vaccine-induced protection reduces with time, 1 resulting in increased susceptibility to infection even among individuals who initially developed a fully protective antibody response. 7 In this study, we monitored serial anti-spike IgG levels to compare antibody decay in MM patients versus healthy controls.

| Participant selection
Participants included MM patients receiving care at a single oncology clinic specializing in the care of MM patients, as well as healthy controls that were typically spouses, friends or other family members of patients. To be able to accurately evaluate half-lives of vaccineinduced antibodies, only individuals who responded adequately to vaccination were included: those whose anti-spike IgG levels increased at least 10-fold from prevaccination to 2-3 weeks following their second dose (D2W2) with minimum levels of 250 BAU/mL (binding arbitrary units per milliliter calibrated to the WHO 20/136 international standard). Individuals were excluded if they were not evaluable at all of the time points or if they elected to obtain a booster vaccination prior to the end of the study. If individuals had evidence of SARS-CoV-2 exposure following their first vaccination, as defined by having an increase in anti-spike antibody levels not attributable to vaccination, they were also not included in the study. Based on these criteria, 18 MM patients aged 61-84 (9 female, 9 male) and 8 controls aged 53-75 (3 male, 5 female) were evaluable. Individual participant information, including treatment regimen, is included in Table S1.

| Half-life calculations and statistical tests
Weeks 2 through 24 half-lives were calculated using both exponential decay and power law methods. 8 Comparisons on half-lives between groups were analyzed using unpaired t tests due to normally distributed data. Comparisons between antibody levels were done by logtransforming the antibody data (which has a lognormal distribution) and then comparing it with unpaired t tests. Z tests were used to compare the number of subjects in each response group (<50, 50-250, and >250 BAU/mL). An analysis of covariance (ANCOVA) test was done to assess whether half-life differences between groups were still significant in spite of the controls being slightly younger.
Tests were deemed statistically significant if p < .05. Analysis was done using R version 4.1.2 and GraphPad Prism 9.

| RESULTS
MM patients and healthy controls were recruited from a single clinic specializing in treatment of this B-cell malignancy. Initially, 48 patients and 11 age-matched controls were recruited with planned follow-up for 24 weeks. However, many patients and some controls independently elected to receive booster vaccinations prior to 24 weeks and were excluded from the analysis. Others had evidence of SARS-CoV-2 re-exposure during the follow-up period and were also excluded from half-life calculations to avoid this confounder ( Figure 1). This limited the study to 18 MM patients and 8 controls who were evaluable for their antibody decay following vaccination at all of the study time points.
Patients had lower antibody levels at week 2 (geomean 1161 BAU/mL) as compared to controls (1835 BAU/mL) that were not statistically significant (p = .62). The geometric mean antibody levels among patients relative to week 2 declined 3.6-, 8.6-, and 15.3-fold at 8-, 16-, and 24-weeks post vaccination, respectively. In controls, the fold reduction in antibody levels was less (2.8-, 4.9-, and 7.77 at 8, 16, and 24 weeks, respectively). Half-lives were estimated using two methods: the power law method and exponential decay. The former is better suited for modeling antibody decay following vaccination as it accounts for ongoing antibody production by plasma cells and memory B-cells, 9 while the latter is best suited for modeling antibody decreases following monoclonal antibody infusions, which lack ongoing antibody production. Using power-law, median half-lives were with no difference between ages 56-70 and age >70. 8 The median exponential decay antibody half-life in MM patients was 37 days (IQR 35-45), which was significantly less than the median 51-day (IQR 48-54) half-life in controls (p = .0004; Figure 2C and Table 1). MM patients had more variability in their antibody decay rates compared to healthy controls, with power-law half-lives ranging from 39 to 124 days in myeloma patients versus 72-149 days in healthy controls (Tables 1 and S1). Controls were significantly younger than patients by 8 years (Table 1), and the results of an ANCOVA test demonstrated that patients' half-lives were still statistically significantly lower than controls after controlling for age differences (p = .03 for power law and p = .003 for exponential decay; Table 1).
Due to shorter half-lives and reduced peak antibody levels at D2W2, patients with MM showed significantly lower antibody levels at week 24 ( Figure 2B and Table 1; p = .027), in spite of only including patients with a full response to vaccination. Previously, we had defined a full response as antibody levels above 250 BAU/mL, corresponding to above the 6th percentile of healthy controls and matching the F I G U R E 1 Participant flow diagram. Multiple myeloma (MM) patients and healthy controls were recruited at a single center specializing in plasma cell dyscrasias. Patients and controls were selected for half-life analysis if they met all blood draw timepoints, developed an adequate response to vaccination (>250 BAU/mL), and had no evidence of re-exposure to SARS-CoV-2 during the study (as evidenced by continuous decline in antibody levels at each subsequent analysis after D2W2).  (Table S2).
In contrast, just 4/13 of the remaining patients had undetectable M-protein, with 6/13 exceeding 0.81 g/dL. This difference was statistically significant by Mann-Whitney U test (p = .036). The geomean IgG of these 5 patients was 728 BAU/mL at D2W2, which was less than the other patients (1390 BAU/mL; p = .17) and lower than controls by unpaired t test (1835 BAU/mL; p = .20). Despite peak antibody levels that were less than half of the other patients, these 5 patients demonstrated higher geomean antibody levels at the end of the study (D2W24) of 118 BAU/ mL, compared to 64 BAU/mL among the remaining patients (p = .15).
These differences were not statistically significant but may merit further evaluation in a larger cohort. There was no difference in the average age of longer half-life patients compared to the rest of the patients.

| DISCUSSION
The data from this study demonstrate that the subset of MM patients who respond to mRNA vaccination for SARS-CoV-2 have more rapid antibody decay as compared to that of the controls. Nearly all MM patients had antibody levels decline to <250 BAU/mL, a level that is below fully protective, by 24 weeks. In 2020 and early 2021, >147 anti-spike BAU/mL was associated with protection from infection in healthy individuals 10 and 126 BAU/mL in a MM patient was associated with incomplete protection from severe disease. 7 Given their reduced antibody responses to vaccination and shorter half-lives, most MM patients would be expected to be susceptible to wild-type SARS-CoV-2 infection 4-6 months after vaccination. We noted that patients with the longest half-lives were more likely to have undetectable M-protein, but overall observed variable responses to COVID-19 vaccination with respect to both antibody levels and half-lives.
The study was limited by lack of blinding. Patients were aware of their anti-spike IgG levels, prompting some individuals to obtain additional vaccine doses prior to the end and thus were excluded from the study. Nine patients left the study between D2W16 and D2W24.
They had lower geomean antibody levels from both D2W2 (783 vs. 1161 BAU/mL; p = .27) and D2W16 (88 vs. 136 BAU/mL; p = .30) that were not statistically significant at a 0.05 cutoff, so perhaps their antibody levels may have influenced a decision to leave the study in order to have additional vaccination(s), but this did not impact the median half-life, which was nearly identical in both groups (Table S3).
Measuring other immunologic parameters was outside the scope of this work. In other studies, MM patients who developed adequate antibody responses to vaccination for COVID-19 also developed spike-specific T-cells at similar rates to healthy controls. 2 These T-cells have longer half-lives than anti-spike antibodies and may confer additional protective benefits independent of antibodies. 11,12 The study was also limited by small sample size. To be able to accurately measure vaccine-induced antibody half-lives, antibodies must increase after vaccination and then be allowed subsequently to decay without repeat exposure to the vaccinating antigen (which would boost antibody levels). This study took place during a 6-month