The study included patients aged ≥18 years with a confirmed diagnosis of active MM in any treatment line and who had subsequent access to our ambulatory care services (26). Patients initially received at least two consecutive doses (i.e. D1 and D2) of BNT162b2 or mRNA1273 vaccines, three or four weeks apart respectively, according to international recommendations. After EMA approval of the booster dose (D3), the initial study design was amended to include evaluation of third dose immune efficacy. Patients with a diagnosis of SMM, or prior SARS-CoV-2 infection, or with seropositive test prior to vaccination were excluded from the analysis.

Sixty-three health-care workers receiving a full immunization schedule of mRNA vaccine served as controls.

To exclude prior SARS-Cov-2 infection, serum neutralizing IgG antibodies to the receptor binding domain (RBD) of the S1 subunit of SARS-CoV-2 spike (S) protein (anti S-RBD IgG) and anti-nucleocapsid (N) antibodies titers were evaluated within 10 days before the first dose. Anti S-RBD IgG levels were subsequently assessed by employing the same platform to longitudinally monitor the long-term kinetics of humoral response to vaccination at definite timepoints, which included the first (T1), third (T3), sixth (T6), ninth (T9), and twelfth (T12) month following the second vaccine dose (D2) and 1 month after the third dose (T1D3), when introduced.

Cellular immunogenicity was assessed simultaneously with the humoral response at T3 and T12.

Any discontinuation or modification of anti-MM treatment regimens were planned according to the recommendations of the International Myeloma Society (IMS) (5).

MM participants filled in a survey reporting any adverse event and side effect following each vaccine dose, graded according to the Common Terminology Criteria for Adverse Events ver. 5.0. Breakthrough infections were graded according to the WHO severity criteria for COVID-19 (WHO/2019-nCoV/Clinical/2022.2).

The study was approved by the Ethics Committee of the National Institute for Infectious Diseases Lazzaro Spallanzani in Rome and by the local Ethics Committee and was conducted in accordance with the International Conference on Harmonisation Guidelines on Good Clinical Practice and the principles of the Declaration of Helsinki. All participants signed an informed consent form prior to their inclusion in the study.

Total levels of anti-SARS-CoV-2 spike antigen antibodies (predominantly IgG) (S-RBD) were measured using the FDA approved Elecsys^® Anti-SARS-CoV-2 ECLIA assay Roche Diagnostics EG on the cobas e 801 analyzer (27). Manufacturer recommendations for evaluation of results were implemented by including 3 different cut-off values to accurately stratify patients into the following subgroups: negative (<0.8 U/mL); inconclusive (0.8 to <5 U/mL); and positive (≥5 U/mL).

Results for anti-N antibodies were expressed as “present” or “absent”, on the basis of a cut-off index (COI) ≥1.0 and <1.0, respectively.

SARS-CoV-2 CMI response was assessed by the IGRA (Interferon-Gamma Release Assays) QuantiFERON Human IFN-gamma SARS-CoV-2, Qiagen^® (QFN SARS) assay, an in vitro diagnostic test designed for the qualitative detection of interferon-γ (IFN-γ), produced by CD4+ and CD8+ T cells, in response to stimulation by a SARS-CoV-2 peptide cocktail in heparinized whole blood. Details about the performance and interpretation of the IGRA assay are provided in the supplementary materials and Table 1S .

Univariate analysis was performed to describe the variables, with continuous variables reported as median and inter-quantile range (IRQ) and qualitative variables reported as absolute frequency and percentages. Chi-2 test was used for assessing the presence of significant association between qualitative variables; when necessary, univariate odds ratio were computed as well, to quantify the size and direction of the statistical relation and relative p-values. The non-parametrical Kruskal Wallis test was applied to assess median differences between groups, with a significance level of 0.05. Multivariate analysis was performed by GLM linear models for anti S-RBD IgG and by logistical models for CMI response; estimated coefficients and odds ratios were presented by forest plots, with confidence intervals (CIs) at 95%.

The anti S-RBD IgG cut-off value was identified by means of ROC curve, with the aim to naturally dichotomize a continuous biomarker into high and low levels.

All data analyses were performed by R-studio (1.4.3).

One hundred and three MM patients, who received at least two doses of SARS-CoV-2 mRNA vaccines between April and June 2021, were enrolled into the study. Due to the presence of anti-N antibodies at the baseline evaluation, one patient was excluded from the analysis. Among the remaining 102 SARS CoV-2-naïve MM patients, BNT162b2 and mRNA1273 vaccines were given to 44 (43.1%) and 58 (56.9%) patients. Of these, 94 patients were subsequently treated with the third dose (BNT162b2 in 75.5% and mRNA1273 in 24.5% of them). The main characteristics, disease status and prior treatments at study entry of the overall patient population are summarized in Table 1 . The median age was 66 (range 47-83) years, and the median number of previous lines of therapy was 1 (range 1-11); most of the patients were actively receiving anti-MM therapy at the time of vaccination, while 7 (6.9%) patients were treatment-free. Forty-five (44.1%) patients had relapsed/refractory MM (RRMM) while the remaining were undergoing first-line therapy. Among 73 (71.6%) transplant-eligible patients, 63 patients had actually received autologous stem-cell transplantation (ASCT) within a median of 32 (16-68) months before vaccination. Responses to the last anti-MM therapy before vaccination in the overall patient population were as follows: very good partial response (VGPR) or higher in 71 (69.6%), partial response (PR) in 17 (16.7%), stable disease (SD) in 5 (4.9%) and progressive disease (PD) in 9 (8.8%) patients. Immunoparesis, as defined by ≥1 uninvolved immunoglobulin (Ig) class concentration below the lower limit, was observed in 83 (88.3%) patients, of whom 64 (68.1%) had a reduction of both Ig classes; lymphopenia (≤1.0 lymphocytes x 10⁹/L) and neutropenia (≤1.0 granulocytes x 10⁹/L) were observed in 3.9% and 21.6% of patients, respectively. The schedule of on-going anti-MM treatments before vaccination was modified in 36 patients, and mainly consisted in transient dexamethasone and/or daratumumab holding. Thirty-three patients who were receiving lenalidomide maintenance continued treatment without any discontinuation. Sixty-three health-care workers (median age: 63 [range 50-67] years) who received mRNA vaccination during the same period represented the control cohort of the study, regarding humoral response only.

At a median of 30 (IQR 28-32) days following D2, a seropositive response, as previously defined, was found in 90/102 (88.2%) MM patients vs a 100% immunogenicity rate in healthy controls (p<0.0001). Median levels of anti S-RBD IgG in these subgroups were 481 (range 0.4-2500) U/mL vs 835 (range 118-2500) U/mL, respectively (p=0.0031). At a median of 3 (IQR 2-4) months after vaccination (T3), serological response was evaluated in 100 out of 102 patients, the remaining 2 having died due to MM. The rates of seroconversion in both study cohorts were almost superimposable to those observed at T1 (90% for MM patients); comparisons between MM patients and healthy controls significantly favoured these latter in terms of higher seroconversion rate and median antibody levels ( Table 2 ). At a median of 5 (IQR 4-7) months following D2 (T6) the seroconversion rate remained high in both MM patients (n=92/99, 93%) and controls (n=63, 100%). Notably, two patients (without evidence of COVID-19 infection and treated immediately after D2 with ASCT and CAR T-cell therapy, respectively) showed a late seroconversion, whereas none of the seropositive patients became seronegative. However, the median titers of antibodies were significantly lower than at T1, with a reduction of about 50% in MM patients (240 vs 481 U/mL, respectively, p=0.0391), and a 35% decline in the control group (539 vs 835 U/mL, p=0.0026).

D3 vaccines were administered at a median of 6 (IQR 5-7) months after D2. The kinetics of humoral response after D3 is summarized in Table 2 . At a median of 5 (IQR 4-6) months after the booster dose (corresponding to T12), the seroconversion rate in the MM subgroup was as high as 99%, a value superimposable to the 100% seen in healthy controls (p=0.1282). In addition, the median anti S-RBD IgG titer reached the maximum measurable level (2500 U/mL) after D3 in both groups, resulting in at least a 10-fold increase in MM patients and without a significant drop at T12 (p=0.9999).

Univariate analyses of variables potentially influencing humoral response at each of the prespecified time-points ( Table 3 ) revealed a relationship between antibody titers and having received the first two doses of mRNA-1273 vaccine, attainment of at least complete response (CR), and being on treatment with lenalidomide maintenance at the time of vaccination. In addition, patients who underwent ASCT showed a median anti S-RBD IgG level significantly higher than the others. Conversely, immunoparesis with involvement of both Ig classes, and prior anti-MM therapies including PIs (proteasome inhibitors) or anti-CD38 monoclonal antibodies (MoAbs), were associated with an impaired humoral response at all time-points ( Table 3 ). Notably, patients receiving PIs or anti-CD38 MoAbs-containing therapies displayed a further marked reduction of the median antibody titer (597 U/mL and 1567 U/mL, respectively) at T12 compared with the other patients (2500 U/mL) (p=0.0061 and p=0.0172 for each comparison, respectively), as well as patients with immunoparesis involving both Ig classes (2012 vs 2500 U/mL for the others, p=0.0290).

A multivariate analysis performed at T3 (e.g., before the booster dose) with a R²-adjusted of 0.4398, confirmed ≥CR (375.62, 95% CI 75.38-675.86, p=0.0149), receiving lenalidomide maintenance (552.82, 95% CI 78.36-1027.28, p=0.0230) and to have obtained a cellular response to the vaccine (526.94, 95% CI 211.39-842.49, p= 0.0014) as independent predictors of higher antibody titers ( Table 4 ). Being in ≥CR (OR: 3.69, 95% CI 1.33-10.78, p=0.0138) and receiving lenalidomide maintenance (OR: 6.56, 95% CI 1.89-31.11, p=0.0067) were also related to the achievement of the highest median anti S-RBD IgG level (2500 U/mL) at 1 month after the booster dose (D3) ( Supplemental Table 2S ). Receiving lenalidomide maintenance (631.10, 95% CI 218.10-1044.03, p=0.0032) and having obtained a cellular response (514.10, 95% CI 116.89-911.23, p= 0.0118) significantly predicted for increased humoral response at T12 (overall R²-adjusted of 0.3381) ( Table 4 ).

Ninety-nine and 94 patients were assessable for CMI response at T3 and T12, respectively. In 5 patients in each of these two subgroups results of IGRA assay turned out to be not evaluable due to lymphopenia (≤0.5 x 10⁹/L), ultimately leading to available CMI response data in 94 patients at T3 and 89 patients at T12. Thirty-two out of 94 patients (36.2%) developed positive IGRA results at T3. Interestingly, all these patients were seropositive and had a significantly higher median anti S-RBD IgG level than those lacking cellular immunity (1264 U/mL [range 23.6-2500] vs 147 U/mL [0.4-2500], p<0.0001). Nine (9.6%) patients were double-negative for both humoral and cellular response.

At a median of 5 (IQR 4-6) months after the third dose (corresponding to T12), 42 (47.2%) patients displayed a positive IGRA response, and their median antibody titer was significantly higher compared with the subgroup showing a persistently negative cellular response (2500 U/mL [range 9.54-2500] vs 782 U/mL [range 0.4-2500], p=0.0001. Only 1 patient (1.1%) remained double negative at T12.

In univariate analysis, the following factors resulted positively related to the development of cellular response at both T3 and T12: response ≥CR, ongoing treatment with lenalidomide maintenance and prior ASCT ( Table 5 ). In addition, having received the vaccine while on or soon after first-line therapy was associated with IGRA reactivity at T3. On the contrary, immunoparesis at all time-points and anti-MM therapies with PIs or anti-CD38 MoAbs at T12 were negative predictors of cellular response.

Multivariate analysis identified receiving lenalidomide maintenance (OR: 3.18, 95% CI 1.01-0.3, p=0.0484) as the only independent predictor of positive cellular response at T3 ( Table 6 ). At T12, being in response ≥CR (OR: 4.68, 95% CI 1.40-18.44, p=0.0172) was an independent factor favourable contributing to T-cell response, while regimens containing PIs (OR: 0.12, 95% CI 0.02-0.57, p=0.0130) or anti-CD38 MoAbs (OR: 0.15, CI 95% 0.02-0.72, p=0.0249) were associated with a diminished cellular response ( Table 6 ).

At last, ROC analysis revealed an anti S-RBD IgG level ≥346 U/mL as being the optimal cut-point associated with the development of a cellular response, with reasonable test sensitivity (94.1%), specificity (75.4%) and respective Youden index (0.695) ( Supplemental Figure 1S ). More in depth, the probability of a positive cellular response was 20 times higher in patients who achieved this cut-off (OR 20.6, 95% CI 6.89-78.01, p<0.0001).

Twenty-one MM patients experienced at least one adverse event (AE) related to the first two vaccine doses. All AEs reported were grade 1–2, transient, and mainly consisting in fatigue, mild temperature increase and pain in the injection site. No AE of grade 2, or higher, were reported following D3.

Documented SARS-CoV-2 infection occurred in 11 MM patients after D2 (n=2) or D3 (n=9). Among these, 2 patients were asymptomatic, 8 developed mild COVID-19 and 1 patient presented a severe illness requiring hospitalization. This patient had a prior history of advanced RRMM in progression at the time of infection and failed humoral and cellular immunogenicity after full vaccination. Six patients received antiviral medications (n=5) and monoclonal antibodies (n=1) to treat COVID-19. None of the infected patients died for COVID-19.

The main characteristics of patients who develop breakthrough infection are reported in Supplemental Table 3S .