A six-month prospective-observational study was conducted from July 1 to December 31, 2022. We screened 238 IEI patients >18 years regularly followed up in the Care Centers in Rome (Policlinico Umberto I, Sapienza University), Naples (Federico II University), and Milan (Ospedale Maggiore Policlinico, University of Milan), Italy. The population study included subjects with primary antibody deficiency diagnosed according to the ESID criteria (26). All participants were treated with immunoglobulin replacement therapy (IgRT) by intravenous (IVIG), subcutaneous (SCIG), or facilitated immunoglobulins (fSCIG). Booster doses of BNT162b2 mRNA vaccine were offered to the entire IEI cohort in September 2021, six months after completing the primary immunization cycle, and in April 2022. From June 2022, AZD7442 (300 mg: 150 mg tixagevimab/150 mg cilgavimab) was offered as PrEP for SARS-COV-2 infection to patients with a minimum weight of 40 kg who did not have a positive test result (PCR or antigen) in the previous month, and who were not vaccinated against COVID-19 in the previous two weeks. Eligible patients received an SMS and an email advising they were suitable for PrEP and inviting them to contact their IEI centre to make an appointment for AZD7442 administration. If no appointment was made within 15 days, an SMS and an email were sent again. Clinicians provided informed consent on the treatment, effectiveness, and contraindications. AZD7442 was administered at least one week after the IgRT administration.

Enrolment was allowed between July 1, 2022, and August 31, 2022. The study population was divided into two groups: 33 patients received AZD7442 as PrEP (AZD7442 group), and 170 patients did not receive AZD7442 (no-AZD7442) group (either did not read the SMS/email, or declined to receive AZD7442, or did not take steps to attend an appointment for whatever reason). Data from 35 patients excluded from enrolment as being positive for SARS-CoV-2 in the previous three months were separately analysed (recently infected group) ( Figure 1 ). For the AZD7442 group, we used the date of administration of tixagevimab/cilgavimab to identify the date of entry into the study; for the no-AZD7442 group, we used the date of the signature of the informed consent as the date of entry into the study. For the recently infected group, July 1st was considered the date of entry into the study. All participants who agreed to participate signed the written informed consent form at enrolment.

After enrolment, clinicians reminded all patients of prevention measures and on the need to contact the hospital in case of SARS-CoV-2 infection. Patients were tested for SARS-CoV-2 by PCR or by nasopharyngeal swab antigen detection every time they attended a hospital site (i.e., to receive IgRT), in case of positive household contacts irrespective of symptoms and upon onset of symptoms possibly related to COVID-19. During the study period, we interviewed the participants monthly via in-person visits or SMS/emails to verify if an infection had occurred from the last observation. All participants were allowed to continue their treatments, including IgRT as standard therapy for the underlying antibody deficiency. In the case of SARS-CoV-2 infection, specific treatments were prescribed by clinicians based on the clinical assessment and ongoing recommendation.

At enrolment, demographic and IEI-related health issues were collected to allow comparison of outcome measures, adjusting for differences between groups. Health factors included IEI diagnosis, immunoglobulins serum levels, lymphocyte count, and IEI-related comorbidities (Chronic Obstructive Pulmonary Disease (COPD) (27), malignancies, autoimmune manifestations, and granulomatous diseases), immunisation status, and number of doses of BNT162b2 mRNA vaccine administered. We defined IEI with a complicated phenotype as having at least one of the non-infective IEI-related comorbidities (25). The Ethical Committee of the Sapienza University of Rome (Prot. 0521/2020, July 13, 2020) approved the study. The study was performed following the Good Clinical Practice guidelines, the International Conference on Harmonization guidelines, and the most recent version of the Declaration of Helsinki.

The primary study outcome was SARS-CoV-2 infection, defined as any person with a positive PCR or positive antigen test result from 1st July 2022 – 31st December 2022. Secondary endpoints included: 1) severe COVID-19 disease; 2) symptomatic infection; 3) duration of the viral shedding; 4) additional SARS-CoV-2 specific treatments; 5) anti-Spike IgG serum levels.

COVID-19 symptoms severity was scored according to the World Health Organization stage. We defined severe COVID-19 disease as either COVID-19-related hospitalisation or all-cause mortality, assessed in each group for the same periods. In SARS–CoV–2 positive patients, we calculated the duration of viral shedding by recording the dates of the first positive and the first negative nasopharyngeal swab. In the AZD7442 group, we measured the anti-Spike IgG serum levels at enrolment (T0), seven days (T7), 120 days (T120), 150 days (T150), and 180 days (T180) after PrEP administration. We collected samples for anti-Spike IgG levels on the day of the IgRT administration, just before immunoglobulins administration.

We evaluated SARS-CoV-2 spike S1/S2 protein-specific IgG antibody levels by the DiaSorin Liaison SARS-CoV-2 Trimeric S IgG chemiluminescence immunoassay (CLIA) (DiaSorin S.p.A, Saluggia, VC, Italy). We reported the sensitivity and specificity performance according to the manufacturer’s instructions. We expressed the levels of anti-SARS-CoV-2 IgG antibodies using the World Health Organization International Standard (NIBSC code. 20/268) binding antibody unit (BAU/mL). The assay’s lower and upper detection limits were 4.81 BAU/mL and 2080 BAU/mL, respectively. Samples testing >2080 BAU/mL were further diluted.

We studied the B-cell responses to SARS-CoV-2 immunisation with the BNT162b mRNA vaccine in a subgroup of 78 patients randomly selected over the participants. Blood samples were obtained ten days after a booster dose administration. We isolated the peripheral blood mononuclear cells (PBMCs) by Ficoll Paque™Plus 206 (Amersham PharmaciaBiotech) density gradient centrifugation and immediately froze and stored in liquid nitrogen until use. The freezing medium contained 90% Fetal Bovine Serum (FBS) and 10% DMSO. To detect SARS-CoV-2 specific B cells, a biotinylated protein spike was individually multimerized with fluorescently labelled streptavidin as previously described (28). Briefly, to identify the Spike-specific memory B cells (MBC), we used two different streptavidin-conjugated fluorochromes, one with a very high brightness index (PE) and the other with a moderate brightness index (BUV395), in order to be able to distinguish low-affinity MBCs (only visible with a very bright fluorochrome such as PE) from high-affinity MBCs (double positive for PE and BUV395). We defined the spike-specific MBCs as low-affinity (positive for PE, S+) or high-affinity (double positive for PE and BUV395, S++). We acquired stained PBMC samples on FACs LSRFortessa (BD Bioscience). At least 2x106 cells were acquired and analysed using Flow-Jo10.8.1 (BD Bioscience). We only performed phenotype analysis of antigen-specific B cells in subjects with at least ten cells detected in the respective antigen-specific gate. The gating strategy is shown in Supplementary Figure 1A .

The primary analysis was to investigate the risk for SARS-CoV-2 infection in the AZD7442 vs. the no-AZD7442 group. We described continuous variables using median and interquartile ranges (IQR) and categorical variables using frequencies and percentages. Comparisons of continuous parameters between groups were calculated with a t-test if normally distributed (as tested by the Kruskal-Wallis’s test) and with a Mann-Whitney U test if not normally distributed; differences in frequencies between groups were calculated by using the χ2 exact test. To assess the relationship between AZD7442 administration status and outcome variables over time, we used the Kaplan-Meier product-limit estimates and based on a Log-rank of the difference between the two treatment groups at the time of the SARS-CoV-2 infection, with no adjustment for baseline covariates. The difference between the two treatment groups was calculated monthly from the enrolment until the end of the study period. In the no-AZD7442 group, we compared specific anti-Spike IgG serum levels at different time points over the study period. Values were compared by the nonparametric Kruskal-Wallis’s test, and if not significant, the Wilcoxon matched-pair signed-rank test or the two-tailed Mann-Whitney U test was used. Differences were significant when p was less than 0.05. Analyses were carried out using SPSS software, version 18 (SPSS, Chicago, IL).

Of 238 IEI candidates, 33 patients (16%) were treated with 300-mg AZD7442 (tixagevimab/cilgavimab, 150 mg/1.5 mL and 150 mg/1.5 mL) and 170 patients (84%) did not take steps to attend an appointment for whatever reason or declined the treatment. The AZD7442 administration was well tolerated, with no patient reporting major adverse events. When comparing variables considered as potential confounders ( Table 1 ), the AZD7442 group had lower serum IgA levels (p=0.006) and lower post-booster anti-Spike IgG serum levels (p=0.010, Table 2 ), suggesting a more severe immune defect in those who agreed to receive PrEP. Age, gender, complicated phenotype, total lymphocyte, lymphocyte subsets count, number and timing of previous vaccine doses were comparable in the two groups ( Table 1 ). Thirty-five patients (age 54 years, IQR: 41-66) who tested positive for SARS-CoV-2 in the three months before the enrolment were considered as controls and included in the recently infected group. Characteristics of recently infected patients and comparisons with the study groups are reported in Supplementary Table 1 .

During the study, 86% of patients were immunized with three or four doses of the BNT162b2 vaccine ( Table 1 and Supplementary Table 1 ). As previously shown (10, 28), the efficacy of SARS-CoV-2 immunisation is heterogeneous in IEI. Data on the production of anti-Spike IgG and specific MBC with low (S+) and high (S++) affinity, and Spike-specific RBD developed post-BNT162b2 booster were available for 78 participants, of which 16 were treated by PrEP. In our cohort, combining specific antibodies and specific B-cell responses, none of the patients produced a response comparable to what was observed in healthy subjects. Thirty-six percent of patients did not produce specific antibodies nor specific B cells (non-responders), 36% produced low levels of specific antibodies but did not generate specific high-affinity (S++) MBC and RBD+ B cells (poor responders), and 28% produced low levels of specific antibodies and generated low frequency of specific high-affinity MBC (weak responders) ( Table 2 ). Post-immunization anti-SpikeIgG serum levels were lower in the AZD7442 group (median 4.5, IQR: 0.5-111.0) compared to the no-AZD7442 group (median 160.3, IQR 8.7-657.3, p=0,010) ( Figure 2 ), suggesting a more severe immune-defect in those who agreed to receive PrEP. None was infected by SARS-CoV-2 in the AZD7442 group. In particular, none of the non-responders/poor-responders was infected in the AZD7442 group, whereas 27% were infected in the no-AZD7442 group (p=0.041) ( Table 2 ).

A total of 60 SARS-CoV-2 infections occurred during the study period. Six out of thirty-three (18.2%) patients who received AZD7442 were infected with SARS-CoV-2 compared to 54/170 (31.8%) patients who did not receive AZD7442 (p = 0.048). Infections occurred respectively after a median time of 167.1 days (95%CI 154.9-179.3) and 148 days (95%CI 138.9-157.33) from the enrolment. The risk of SARS-CoV-2 infection was reduced by 85% among those who received the PrEP at a median follow-up of 90 days from treatment administration (hazard risk [HR] 0.15; 95%CI 0.14-0.96; log-rank p=0.036). The comparison was not significant at the end of the study period (HR 0.6; 95%CI 0.33-0.1.84; log-rank p=0.278) ( Figure 3 ). No patient who had received prophylaxis became infected in the first 60 days, while those who had not received prophylaxis continued to become infected with a constant trend over time. To note, when considering only SARS-CoV-2 naive patients, the AZD7442 group had a lower risk for infection than the no-AZD7442 group at a median follow-up of 90 days (HR 0.20, 95%CI 0.07-0.56, log-rank p=0.049, Supplementary Figure 2 ). Time of viral shedding did not differ in the AZD7442 and no-AZD7442 groups (19 days, IQR: 10-28 vs. 13 days, IQR: 7-20, p=0.614). At univariate analysis, age was positively associated with SARS-CoV-2 infection (OR 4,841, 95%CI 0,07-9,61, p= 0.047), while a prior episode of SARS-CoV-2 infection was found to be protective (OR 0.09 95%CI 0,03-0,32, p<0.0001). Gender, lymphocyte counts, IgA serum levels, IEI comorbidities, complicated phenotype, and the number of doses of COVID-19 vaccine were not associated with infection risk ( Supplementary Table 2 ). Immunophenotype of infected patients revealed a lower frequency of MBC cells in the AZD7442 group compared to the no-AZD7442 group (median 7.5%, IQR 26.5-48.5 vs. 1.0%, IQR 4.2-16.0, p=0.040) ( Supplementary Figure 3 ), suggesting that PreEP was ineffective mainly in patients with severely impaired memory B function.

The recently infected group was 89% and 92% less likely to have a SARS-CoV-2 infection compared to the AZD7442 and the no-AZD7442 group (HR 0.11; 95%CI 0.03-0.61; log-rank p=0.011 and HR 0.08; 95%CI 0.16-0.64; log-rank p=0.001, Figure 3 ), as a single patient was infected during the study.

None of the patients was hospitalized for COVID-19, had severe SARS-CoV-2 infection, or died during the study. A mild symptomatic infection was observed in 29% of patients of the no-AZD7442 group and in 18% of the AZD7442 group (p=0.500). The risk of symptomatic infection was reduced among the AZD7442 group within the first 90 days (HR 0.18; 95%CI 0.14-1.05; log-rank p=0.049). Univariate analysis identified low lymphocyte counts as a risk factor for developing symptomatic COVID-19 ( Supplementary Table 3 ). SARS-CoV-2 infection required treatments in 33% of patients in the AZD7442 group (antivirals only) and in 56% of patients in the no-AZD7442 group (antivirals and monoclonal antibodies).

In the AZD7442 group, median anti-Spike IgG serum levels at enrolment were 18.5 BAU/mL (IQR: 3.7-77.7). Seven days after AZD7442 administration (T7), IgG median levels were 6780 BAU/mL (IQR: 6305-7075) and they remained high and at T120 (1695 BAU/mL, IQR: 1548-2080), at T150 (1580 BAU/mL, IQR 1345-1935), and at T180 (1120 BAU/mL, IQR 754-2013) ( Figure 4A ).

As IgRT might contribute to the passive transfer of anti-Spike IgG, we randomly measured anti spike IgG serum levels in plasma samples of patients who did not receive AZD7442 as therapy or as prophylaxis. To avoid bias, we excluded patients infected in the previous six months or vaccinated in the previous four weeks. Anti-Spike IgG serum levels increased over time, from a median level of 7.5 BAU/mL (IQR: 0-101.1) recorded in the trimester October-December 2021 to a median level of 1175 BAU/mL (IQR: 663-1728) recorded in the trimester October-December 2022 (p<0.0001) ( Figure 4B ), showing the progressive increase of the anti-Spike IgG concentration in the polyclonal immunoglobulin brands administered as IgRT.