One-year breakthrough SARS-CoV-2 infection and correlates of protection in fully vaccinated hematological patients

The long-term clinical efficacy of SARS-CoV-2 vaccines according to antibody response in immunosuppressed patients such as hematological patients has been little explored. A prospective multicenter registry-based cohort study conducted from December 2020 to July 2022 by the Spanish Transplant and Cell Therapy group, was used to analyze the relationship of antibody response over time after full vaccination (at 3–6 weeks, 3, 6 and 12 months) (2 doses) and of booster doses with breakthrough SARS-CoV-2 infection in 1551 patients with hematological disorders. At a median follow-up of 388 days after complete immunization, 266 out of 1551 (17%) developed breakthrough SARS-CoV-2 infection at median of 86 days (range 7–391) after full vaccination. The cumulative incidence was 18% [95% confidence interval (C.I.), 16–20%]. Multivariate analysis identified higher incidence in chronic lymphocytic leukemia patients (29%) and with the use of corticosteroids (24.5%), whereas female sex (15.5%) and more than 1 year after last therapy (14%) were associated with a lower incidence (p < 0.05 for all comparisons). Median antibody titers at different time points were significantly lower in breakthrough cases than in non-cases. A serological titer cut-off of 250 BAU/mL was predictive of breakthrough infection and its severity. SARS-CoV-2 infection-related mortality was encouragingly low (1.9%) in our series. Our study describes the incidence of and risk factors for COVID-19 breakthrough infections during the initial vaccination and booster doses in the 2021 to mid-2022 period. The level of antibody titers at any time after 2-dose vaccination is strongly linked with protection against both breakthrough infection and severe disease, even with the Omicron SARS-CoV-2 variant.


INTRODUCTION
Patients with hematological disorders (HD) are at high risk of severe coronavirus infectious disease 2019 (COVID-19) caused by the new coronavirus (SARS-CoV-2) [1]. The dramatic impact of COVID-19 on hematological patients was observed during the early phase of the pandemic, with mortality rates then exceeding 25% [2][3][4][5][6][7]. However, encouraging preliminary data suggest that SARS-CoV-2 vaccination has played a major role in reducing the severity of breakthrough COVID-19 in these immunocompromised patients, with mortality dropping to less than 10% during 2022 [8]. A recent retrospective case series supports the reduced mortality (12.4%) of breakthrough COVID-19 (mainly with the alfa variants of concern (VOC) after vaccination in hematological patients, but with high burden of hospitalization (66.4%) and intensive care unit (ICU) admission (21.3%) irrespective of the vaccine doses administered or serological response [9].
The first generation of SARS-CoV-2 vaccines failed to prevent community transmission of the virus, in particular the most recent Omicron VOC, but protection still remained high against severe COVID-19 [10]. In fact, it is expected that most HD patients worldwide will be infected several times with different SARS-CoV-2 VOC despite vaccination and boosters, and a significant number of them could still experience a severe course. The poor humoral response and impaired cellular immunity in HD patients contribute to a higher incidence of post-vaccination SARS-CoV-2 infection and its severity [11,12]. To date, there is a lack of studies evaluating the kinetics of breakthrough SARS-CoV-2 infection in a prospective fashion as well as its severity in vaccinated HD patients. Thus, we still do not know to what extent these patients will develop breakthrough SARS-COV-2 infection after vaccination, their potential severity and the risk factors for these breakthrough infections.
In view of these gaps, the current study analyzes the incidence, characteristics, severity and clinical and immune factors associated with breakthrough SARS-CoV-2 infections through a prospective observational registry conducted by the Spanish Hematopoietic Stem Cell Transplantation and Cell Therapy Group (GETH-TC) aimed at monitoring the humoral response to SARS-CoV-2 vaccines through 1 year after complete vaccination in a large cohort of 1551 HD patients.

PATIENTS AND METHODS Study population
In December 2020 the Infectious Complications Subcommittee (GRUCINI) of the GETH-TC in collaboration with the Spanish Society of Hematology and Hemotherapy (SEHH) launched a national prospective multicenter registry aimed at evaluating the humoral response for SARS-CoV-2 vaccination and its duration through one year after full vaccination (defined as two vaccine doses for RNA vaccines or one dose for adenoviral vector-based vaccines). Characteristics of this registry have been described in detail elsewhere [13,14]. The registry included consecutive HD patients who were vaccinated against SARS-CoV-2 between December 30, 2020 to June 30, 2021 in 21 participating Spanish centers. In June 2021 a specific form for booster doses and breakthrough SARS-CoV-2 infection was implemented in the registry. Patients were followed as initially planned and monitored for the development of SARS-CoV-2 reactive IgG antibodies (SCoV2-R-A) and breakthrough SARS-CoV-2 infection at different time points (3-6 weeks, 3, 6, and 12 months) after the complete vaccination schedule. The status of all included patients was updated on July 28, 2022. All patients included in this registry gave their signed informed consent in accordance with the declaration of Helsinki. The local Research Ethical Committee of the Hospital Clínico Universitario of Valencia approved the registry and study protocol (reference code 35.21).

Inclusion criteria, cohort selection and data check
As of July 28, 2022, the GETH-TC registry recruited 1783 patients with different hematological disorders who were fully vaccinated against SARS-CoV-2. With the aim of assessing (1) the incidence and severity of breakthrough SARS-CoV-2 infection at one year after full vaccination (2) the effect of serological response at different time points and (3) the effect of booster doses, the current study focused on patients with complete demographic and survival data. We excluded 232 patients from 3 centers, initially included with limited data (only affiliation data), that did not obtain the institutional approval for serological testing leaving 1551 hematological patients for final study analysis. At the time of this analysis none of the patients had received pre-exposure prophylaxis with monoclonal antibodies.
Out of 1551 HD patients, 1398 (90.1%), 1174 (76%), 1023 (66%) and 849 (55%) had available serological assessment at 3-6 weeks, 3 months, 6 months, and 12 months after full vaccination, respectively. Of these 1250 (89%), 1070 (91%), 900 (88%) and 820 (97%) had quantitative assessment in BAU/mL available, respectively, reflecting a high rate of missing serology data over time. To analyze whether there were relevant unknown biases which led to not having serological testing, we compared the clinical characteristics of patients and the survival of those with and without complete serological data at different time points and we did not find any significant differences (See supplementary Table S1), suggesting random reasons for missing serological data. Accordingly, we used a listwise deletion analysis to evaluate the effect of antibody titers on the risk of breakthrough infection at each pre-specified time point. A major reason for having missing serologies, from our personal experience and through our interactions with all participating physicians/centers, the most common reasons involved the patients' and investigators' exhaustion and/or the difficulties in scheduling serological tests outside the routine schedule for outpatient visits in long-term survivors, as well as the institutional refusal for serological monitoring through academic non-funded studies while the patient's death before the scheduled serological test was a logical reason for missing serological follow-up.

Technical considerations and Definitions
We assessed humoral response at the specified time points using serological ELISA or chemiluminescence immunoassay according to their availability at the microbiology services of each participating center. Antibody detection or seropositivity was defined as detectable SARS-CoV-2-reactive IgG antibodies (SCoV2-R-A) at any level above the lower limit of detection level for each test used. As recommended by the SEHH in vaccinated individuals, serological testing included the detection of IgG against both the nucleocapsid (N) and surface (S) proteins (anti-N and anti-S IgG, respectively) [15]. Antibody levels in patients with quantitative assessment were normalized according to the WHO standards and results were reported as SCoV2-R-A binding antibody units per milliliter (BAU/mL). Supplementary Table S2 summarizes the technical characteristics of the serological tests used and the normalization of antibody titers to BAU/mL according to WHO standards.
Complete vaccination (full primary immunization) schedules were defined as per marketing authorization at the time of study conduct and included two doses for full primary immunization (except COVID-19 Vaccine Janssen®). An additional dose after completion of full primary immunization was considered as a booster dose.
Pre-vaccination SARS-CoV-2 infection was defined as patients with prior history of PCR-proven COVID-19 and/or positive SARS-CoV-2 serostatus (IgG and/or IgM) before the first vaccine dose.
Breakthrough SARS-CoV-2 infection was defined as molecular (PCR test), antigenic or serological (IgG anti-N seroconversion between two consecutive serological tests) evidence of SARS-CoV-2 infection 7 days after full vaccination and until the last follow-up. Commercial PCR test used for diagnosis was provided in Supplementary Table S3. Corticosteroids use was defined as ≥15 mg per day of prednisone or equivalent doses of methylprednisolone, dexamethasone or hydrocortisone started at least 2 weeks before vaccination.
Although we did not sequence SARS-CoV-2 strains in any case, the inference of VOC on each episode was based on the likelihood of having a specific VOC according to the Spanish sequencing epidemiological data (see Supplementary Fig. S1) [16]. We inferred a specific VOC when the infection was diagnosed during the predominance (>50% of all sequenced VOC) of a specific VOC in our country. For example, Alfa-Beta VOC was considered in SARS-CoV-2 infections diagnosed from April 2021 to June 20 2021. Delta was considered from June 21 to December 26, 2021 and Omicron from December 27, 2021 to the end of the study in July 2022.

Endpoints and statistical analysis
The primary objective of the study was to assess the incidence, severity and risk factors of breakthrough SARS-CoV-2 infection in vaccinated HD patients. Secondary endpoints include the effect of SCoV2-R-A titers on the development and severity of breakthrough SARS-CoV-2 infection at each pre-specified serological time points. We also evaluated the clinical benefit of booster vaccine doses as compared to those without boosters in terms of breakthrough infection incidence. Finally, we evaluated the effect of breakthrough SARS-CoV-2 infection in all-cause mortality.
The main patient characteristics were reported by descriptive statistics on the total available information: medians and ranges were used for continuous variables, while absolute and percentage frequencies were used for categorical variables. For comparisons, Fisher exact test or Mann-Whitney's U test were used when appropriate.
Breakthrough SARS-CoV-2 infection after vaccination was estimated by the cumulative incidence method considering death without SARS-CoV-2 infection as the competing event [17,18]. Univariate and multivariate analyses of risk factors for breakthrough infection were calculated using the Fine and Gray test [19]. For multivariate analysis, only variables with parameter estimates showing a p ≤ 0.10 in the univariate analysis were finally included. We did not include in the multivariate analysis timedependent covariates (serological assessments and/or booster doses) since the circulation burden of SARS-CoV-2 in the community was not constant J.L. Piñana et al. over time [20] which may lead to a misleading assumption (i.e. the booster vaccine doses were given after the second dose and just before the increase of community SARS-CoV-2 circulation and its inclusion as a timedepend covariate could lead to a misleading significant increase in the risk of breakthrough infection). To address the clinical benefit of booster doses as compared to those who only received 2 doses in terms of breakthrough infection incidence we performed two landmark analyses, the first one selecting a control group of patients (only 2 doses) alive without breakthrough infection and whose observation period started at the median time of booster administration in the study group (median time from second to third vaccine dose). The second landmark analysis started the observation period in December 27, 2021 in both groups (booster and non-booster) including patients alive and COVID-free at the beginning of the observation period to assure that booster and non-booster patients did not differ in the rate at which they came in contact with SARS-CoV-2 [21]. A median test analysis to check the protective effect of the amount of SCoV2-R-A was carried out in patients with available quantitative SCoV2-R-A titers normalized to BAU/mL, whether they developed or not breakthrough infections. A p-value <0.05 was considered statistically significant. All p-values are two-sided. Analyses were performed using the statistical software SPSS v. 25(IBM SPSS Statistics, Armonk, New York, USA) and R software, version 4.0.
All patients were fully vaccinated, whereas 1284 (82.8%) received a first booster dose at a median of 168 days after complete vaccination schedule, and 430 (27.7%) received a second booster dose at a median of 159 days after the first booster.

Incidence, severity, and risk factors of breakthrough SARS-CoV-2 infection
We identified 266 patients (17%) with breakthrough SARS-CoV-2 infection at a median of 158 days (range 7-391) after complete vaccination. Fourteen out of 266 (5%) episodes occurred in patients with pre-vaccination COVID-19 being detected by PCR (14/129, 11%). Characteristics of patients and SARS-CoV-2 breakthrough infection are summarized in Table 2. The distribution of breakthrough cases during the study period and type of probable VOC are provided in Fig. 1. Spanish epidemiological curve of SARS-CoV-2 infection during the study period was provided in Supplementary  Fig. S2. The one-year cumulative incidence of breakthrough SARS-CoV-2 infection in the whole cohort from the start of vaccination was 18% [95% confidence interval (C.I.), 16-20] ( Fig. 2A). Seventy-six (28.5%) developed SARS-CoV-2 infection after full primary vaccination and before the booster dose, whereas 165 (62% of breakthrough infections and 12.8% of those who received the 1st booster) and 25 (9.4% of breakthrough infections and 5.8% of those who received the 2nd booster) were infected after the 1st booster (before 2nd booster) and after the 2nd booster, respectively.
Univariate and multivariate analyses of risk factors associated with the incidence of breakthrough SARS-CoV-2 infection are shown in Table 3. CLL patients (29%) and use of corticosteroids (24.5%) at the time of vaccination were associated with a higher cumulative incidence of breakthrough infection whereas female gender (15.5%), and vaccination at least 1 year after the end of HD therapy (14%) were protective in the multivariate analysis. As shown in Table 3, diagnosis of a cMPN and having had COVID-19 before vaccination showed a trend for a protective effect. The incidences of SARS-CoV-2 infection according to the aforementioned risk factors are shown in Fig. 2B-E. To analyze whether similar risk factors existed for breakthrough infection by the omicron variants, we performed a separate multivariate analysis in those infected by these VOC and found the same risk factors as in the entire cohort. We also performed a multivariate analysis of risk factor for symptomatic breakthrough SARS-CoV-2 infection and risk factors remain unchanged except for gender which was no longer significant (details of both analyses in the footnote to Table 3).

Antibody level titers and severity of breakthrough SARS-CoV-2 infection
For this purpose, we excluded patients with breakthrough SARS-CoV-2 infection before each serological testing point. Median antibody titers according to the later development of breakthrough SARS-CoV-2 infection at different time points are shown in Table 4. As expected, the magnitude of SCoV2-R-A responses was significantly lower in patients with breakthrough SARS-CoV-2 infection as compared to those without at each of the serological points analyzed (Fig. 3A-D). Results were similar when analyzing cases attributed to the Omicron VOC (data not shown). Note that median antibody titers increased over time in both groups, reflecting the effect of booster doses, especially after 6 months from full vaccination. A serological cut-off value of 250 BAU/mL was able to discriminate the risk of being infected and the disease severity ( Table 5).
Infection severity, number of vaccine doses, and type of SARS-CoV-2 VOC We analyzed potential differences in severity whether the patient had the breakthrough infection after the second, third, or fourth dose of vaccine (Table 6) and we did not find relevant differences, except for a trend to a lower rate of ICU admission after the third dose. In contrast, differences were more obvious according to the type of VOC (alfa-beta vs delta vs omicron) which was inferred from our National epidemiological sequencing data. Omicron causes milder cases of COVID-19 requiring less ICU admission than Alfa-Beta or Delta (p < 0.01 for both comparisons). There was a trend towards a lower pneumonia rate with Delta and Omicron. However, death occurred with Delta and Omicron but with very few cases (3/45 and 2/203, respectively).

Booster doses and breakthrough SARS-CoV-2 infection
In order to evaluate the potential clinical benefit of receiving a booster dose, we performed a landmark analysis comparing the incidence of breakthrough SARS-CoV-2 infection in patients with or without a booster dose. The inclusion criterion in the booster group was no history of SARS-CoV-2 infection before the 1st booster dose (n = 1182), which was day 0 for the landmark analysis. The non-booster cohort included patients who were alive

DISCUSSION
The current study provides a real picture of SARS-CoV-2 breakthrough infections in patients with HD one year after the implementation of mass vaccination and boosters. The first observation was that the epidemiology of breakthrough SARS-CoV-2 infection in HD patients mirrored that of the community. Secondly, we identified conditions associated with higher incidence of breakthrough infection, such as CLL and corticosteroid use at the time of vaccination, whereas female gender and vaccination more than one year after the last treatment were associated with a reduced incidence. Quantitative humoral responses correlated with both the risk of developing breakthrough SARS-CoV-2 infection and its severity, irrespective of when it was evaluated. Unexpectedly, we observed an encouraging low mortality rate (1.9%) of breakthrough SARS-COV-2 infection in HD patients. In addition, breakthrough SARS-CoV-2 infections did not increase all-cause mortality in this large series of HD patients suggesting a favorable evolution of the pandemic in terms of virulence. The prospective design of our registry along with the large sample size, the longitudinal serological monitoring and the long follow-up enable us to provide an accurate estimate of several critical issues, such as real-life epidemiology of SARS-CoV-2 infection in vaccinated HD patients and cumulative incidence estimates. In addition, our design limited the likely risk of bias of reporting only severe cases, which is characteristic of retrospective registry studies, providing a precise picture of clinical characteristics and severity of SARS-CoV-2 infection in this particular scenario.
Despite intensive transmission prevention measures counseling our patients and caregivers, we still observed that SARS-CoV-2 infection in hematological patients mirrored national epidemiological data [2,20], since the three-peak prevalence of breakthrough SARS-CoV-2 infection observed in our series (July-August 2021, January 2022 and April-May 2022, see Fig. 1) fully coincided with our national epidemiological data (see Fig. 1 and  Supplementary Fig. S2). Preventive transmission measures (hand washing, social distancing, wearing mask, etc) are still highly recommended for patients but are clearly far from being satisfactory. It looks suitable that such measures would be applied not only to immunosuppressed patients and their caregivers but also to those who have a close contact with these patients (friends, other relatives, health staff, etc). We already learned that the more thorough preventive transmission     measures are, the lesser the incidence of respiratory virus infections in immunocompromised patients [22]. We report an overall cumulative incidence of breakthrough infection of 18% at one year after the start of mass vaccination. Although there is no available data in other scenarios for comparing cumulative incidence estimates, it is likely that HD patients have an increased breakthrough infection incidence as was the case in patients with solid tumors [23]. In fact, there is an evident increased risk of significant breakthrough infection immediately after vaccination in HD patients as compared to the general population [11], however, the incidence is not necessarily the same for all HD. Our findings suggest that CLL disease and receiving corticosteroid therapy at the time of vaccination experienced a significantly higher infection incidence. Several prior studies revealed that among HD patients, CLL patients are characterized by a severe cellular immune dysfunction which translates into a low humoral response rate [24,25] and a severe course of the disease [26,27], although a significant reduction in mortality has been observed in the Omicron era [28]. On the other hand, corticosteroid use has been associated with a lower probability of mounting an adequate humoral response [29]. Both conditions may facilitate the virus entry and in particular its spread. Although the use of anti-CD20 monoclonal antibodies has been consistently associated with a lower amount of SARS-CoV-2-R-A after vaccination [29][30][31][32], being treated with anti-CD20 monoclonal antibodies was not associated with an increased incidence of breakthrough infection in our study. It is likely that the number of events and in particular of patients who received anti-CD20 monoclonal antibodies within 6 months before starting vaccination (n = 97, 36%) was too low to observe significant differences. The male gender has been already associated with a more severe course of the disease [33]. In accordance, our multivariate analysis revealed that males were prone to higher incidence of breakthrough infection as compared to females. Although we do not have a likely explanation for this difference, we can speculate that hormonal differences and/or differences in social behaviors between males and females may explain such differences. Finally, patients free of treatment for at least one year before vaccination experienced a lower incidence, most likely because the probabilities of having a serological response after vaccination are consistently higher [29]. In light of these findings, it seems reasonable to maximize humoral vaccine responses according to the disease status and treatment as recently suggested by the SEHH [15] and that low-evidence-based preexposure prophylaxis recommendations in vaccinated patients with long-lasting monoclonal antibodies should be restricted to patients harboring these conditions instead of its unselective use. Of note, most of emergent Omicron BA. 4 and BA. 5 subvariants showed a loss of neutralization activity to almost all monoclonal antibodies but bebtelovimab [34]. Our prior preliminary analysis with only 37 cases of breakthrough infection, showed a higher rate of breakthrough infection and higher disease severity in patients without SARS-CoV-2-RA at 3-6 weeks of full vaccination [8]. With a significantly larger number of cases, we hereby confirm this observation and extended this finding beyond 3-6 weeks after full vaccination to later times points from vaccination and irrespective of the number of booster doses given. In this sense, higher levels of both binding and neutralizing anti-SARS-CoV-2 antibodies may protect against COVID-19 and are correlated with each other [35,36]. Data in the general population also support this statement [37,38]. Thus, the quantitative assessment of anti-SARS-CoV-2 antibodies at any time after vaccination could be justified for a benefit/risk calculation before additional vaccine doses or anti-SARS-CoV-2 monoclonal antibodies are given in these immunosuppressed population.
The severity of SARS-CoV-2 infection in patients with HD appeared to decrease over time most likely due to the implementation of mass vaccination programs. Initial reports, including ours, described high rates of pneumonia (>70%), hospital admission (>50%), ICU requirement (>20%) and overall mortalities exceeding 25% in unvaccinated HD patients [2][3][4][5][6][7]. A recent study focusing on breakthrough infections in vaccinated HD patients found a reduced severity but still a significant burden of hospitalizations (66%), ICU admissions (21.3%) and SARS-CoV-2related mortality (12.4%) [9]. In contrast to this retrospective study, our data indicate that the severity is lower than previously reported in terms of hospitalizations (18%), ICU admissions (1.9%) and especially, mortality (1.9%) which was encouragingly low. These observations could be justified by several factors; first, the prospective design of the current study led to the capture of both symptomatic and asymptomatic infections, reducing the bias of registry-based retrospective studies. Second, the vast majority of cases (76%) were attributed to the Omicron VOC, which has a lower intrinsic virulence [28,39]. Third, efforts in maintaining a high titers of anti-SARS-CoV-2 antibody titers in this vulnerable population may have paid off, as higher levels of antibody titers were consistently associated with lower/less disease severity. In fact, the cut-off of 250 BAU/mL was able to predict the risk of breakthrough infection as well as the disease severity.
Finally, we did not observe any significant differences according to the number of booster doses administered, which may merely indicate that more important than the number of booster doses is the amount of anti-SARS-CoV-2 antibody titers. In fact, no patient with SCoV2-R-A > 250 BAU/mL died from SARS-CoV-2 in our cohort irrespective of the serological time points assessed and the number of booster doses. The monitoring period during which our study was conducted (from early 2021 to end July 2022) spanned between the fifth, sixth and seventh COVID-19 waves in Spain, which were sequentially caused by Alfa-Beta, Delta and Omicron VOC according to our national epidemiological sequencing data. Current data show that infections by the Omicron VOC are more frequently symptomatic but less severe in term of ICU support compared to Alfa-Beta or Delta VOC. Our findings are in line with prior studies in the general population [39,40].
The limitations of this study comprise the use of different serological tests, absence of neutralizing antibody testing, the absence of cellular immune response analyses and the lack of molecular data regarding the SARS-CoV-2 variants in patients with breakthrough infections. In addition, although a significant drop out at the 12 months after full vaccination period was observed and could be regarded as a limitation, all patients alive at the given follow-up time points were clinically assessed, and there were thus no patients who were alive and lost to follow-up during the study period. In this sense, the random appearance of serological missing data along with a still large number of evaluable cases and consistent results at each time point assessment may support a limited potential bias. The use of anti-N IgG seroconversion in defining breakthrough SARS-CoV-2 Table 5. SARS-CoV-2 infection severity according to anti-SARS-CoV-2 IgG reactive antibody cut-off before the occurrence of SARS-CoV-2 infection in evaluable patients.    infection could be regarded as an additional limitation given the possibility of passive immunization transfer in patients receiving blood products support from seropositive donors. However, only 7.2% of patients with anti-N IgG seroconversion were on transfusion and/or immunoglobulin support at the time of vaccination.

Variable
CONCLUSION CLL patients and those under corticosteroids showed a higher incidence of breakthrough SARS-CoV-2 infection one year after the start of mass vaccination. The amount of antibody titers appears useful in identifying HD patients at higher risk of breakthrough SARS-CoV-2 infection by any VOC and its severity, and titers of SCoV2-R-A above 250 BAU/mL appear to be a good cut-off value. Last but not least, we report a reassuring reduction of SARS-CoV-2 mortality in this vulnerable population.

DATA AVAILABILITY
Data available upon formal request by email to the Spanish hematopoietic transplant and cell therapy group (GETH-TC).