key: cord-0881744-tytq9ca8
authors: Marasco, Vincenzo; Carniti, Cristiana; Guidetti, Anna; Farina, Lucia; Magni, Martina; Miceli, Rosalba; Calabretta, Ludovica; Verderio, Paolo; Ljevar, Silva; Serpenti, Fabio; Morelli, Daniele; Apolone, Giovanni; Ippolito, Giuseppe; Agrati, Chiara; Corradini, Paolo
title: T‐cell immune response after mRNA SARS‐CoV‐2 vaccines is frequently detected also in the absence of seroconversion in patients with lymphoid malignancies
date: 2021-10-14
journal: Br J Haematol
DOI: 10.1111/bjh.17877
sha: beb15ce886a3e293cc94310f6bef6c88736d534d
doc_id: 881744
cord_uid: tytq9ca8

Patients affected by lymphoid malignancies (LM) are frequently immune‐compromised, suffering increased mortality from COVID‐19. This prospective study evaluated serological and T‐cell responses after complete mRNA vaccination in 263 patients affected by chronic lymphocytic leukaemia, B‐ and T‐cell lymphomas and multiple myeloma. Results were compared with those of 167 healthy subjects matched for age and sex. Overall, patient seroconversion rate was 64·6%: serological response was lower in those receiving anti‐cancer treatments in the 12 months before vaccination: 55% vs 81·9% (P < 0·001). Anti‐CD20 antibody plus chemotherapy treatment was associated with the lowest seroconversion rate: 17·6% vs. 71·2% (P < 0·001). In the multivariate analysis conducted in the subgroup of patients on active treatment, independent predictors for seroconversion were: anti‐CD20 treatment (P < 0·001), aggressive B‐cell lymphoma diagnosis (P = 0·002), and immunoglobulin M levels <40 mg/dl (P = 0·030). The T‐cell response was evaluated in 99 patients and detected in 85 of them (86%). Of note, 74% of seronegative patients had a T‐cell response, but both cellular and humoral responses were absent in 13·1% of cases. Our findings raise some concerns about the protection that patients with LM, particularly those receiving anti‐CD20 antibodies, may gain from vaccination. These patients should strictly maintain all the protective measures.

Patients affected by lymphoid malignancies (LM) are at an increased risk for severe COVID-19 and have an exceedingly high mortality rate. [1] [2] [3] Recently, two mRNA-based vaccines were approved for the general population to prevent against COVID-19. However, individuals with LM were not included in clinical trials, and the immune response elicited by SARS-CoV-2 vaccines in this immune-compromised population is largely unknown.

The phase 3 trials of mRNA-1273 (Moderna) and BNT162b2 (Pfizer BioNTech), the first mRNA-based vaccines, which target the spike protein to elicit protective immunity, demonstrated an efficacy at preventing COVID-19 in healthy individuals of 94% and 95%, respectively. 4,5 Seroconversion occurred in almost all vaccinated individuals. 6, 7 These results suggested potential beneficial effects also in LM patients, although the seroconversion rate was expected to be lower than in the general population as it happens already after the infection. 8 In 2009, the emergence of H1N1 influenza led to the development of an inactivated virus-based vaccine. Some data showed decreased seroconversion in LM patients, particularly in those treated with antibodies targeting CD20 antigen, 9 whereas the T-cell mediated response was similar to that in healthy individuals. 10 Nonetheless, viral vaccines are routinely recommended in LM patients. [11] [12] [13] [14] In Italy, the indication from healthcare authorities was to use mRNA vaccines in LM patients, because of a supposed higher activity and better safety profile. The aim of this study was to evaluate the humoral and cellular response to mRNA-1273 and BNT162b2 vaccines in patients with LM.

This prospective study assessed the efficacy of two doses of either mRNA-1273 or BNT162b2 vaccines administered 28 days apart, according to the national healthcare system's indications in order to increase the vaccine's availability in the first phases of the national vaccination plan and contrast the risk of vaccine shortage. We included adult (age≥18 years) consecutive patients who were vaccinated at the Istituto Nazionale dei Tumori, Milan, Italy. According to national healthcare system indications, priority to vaccination was given to frail patients, defined by the presence of one of the following: presence of active disease; ongoing treatments or within 12 months from last therapy; active graft-versushost disease; allogeneic stem cell transplantation (allo-HSCT) or chimaeric antigen receptor-modified (CAR-) T-cell therapy within 3 to 12 months from administration of the first dose of vaccine. After the completion of vaccination of this high-priority population, we included also patients in remission who had completed their treatment more than 12 months prior to vaccination.

The control group consisted of age-and sex-matched healthcare workers (HCW), who were enrolled in the prospective study INT65/20 and, based on the local availability, received the BNT162b2 vaccine at the Istituto Nazionale dei Tumori, Milan, Italy. The trial was approved by the Institutional Review Board of Istituto Nazionale dei Tumori, Milan, Italy, and written informed consent was collected from all patients (INT112/21).

The primary end-point of the study was the seroconversion rate among LM patients after full-dose vaccination. Anti-SARS-CoV-2 S levels were monitored before the first dose, at the time of second dose administration, and two weeks later. Patients with a positive basal anti-SARS-CoV-2 S titre were excluded from this study.

Among patients who received chemotherapy with or without anti-CD20 antibody, immune modulatory drugs (IMIDs) or novel oral agents within the last 12 months, we selected a cohort of 99 patients and evaluated their SARS-CoV-2specific T-cell response two weeks after the second dose independently from their serological status after vaccination. The control group for the T-cell response consisted of 99 HCW, who received the BNT162b2 vaccine and whose T-cell response was evaluated two weeks after the second dose at the National Institute for Infectious Diseases "Lazzaro Spallanzani", Rome, Italy.

Briefly, the Roche Elecsysâ Anti-SARS-CoV-2 S (Roche S tAb, Roche Diagnostics International Ltd, Rotkreuz, Switzerland) was used to measure antibodies directed against the receptor-binding domain (RBD) of the viral spike (S) protein. T-cell responses were evaluated through measurement of in vitro T-helper cell type 1 (Th1)-associated cytokine release [interferon (IFN)-c, tumour necrosis factor (TNF)-a, interleukin (IL)-2] in plasma samples after stimulation with S peptides using an automatic enzyme-linked immunosorbent assay (ELISA; ELLA, Protein Simple, San Jose, CA). 15 Variable definitions, complete laboratory and statistical methods are provided in Data S1. In all, 140 patients (53Á2%) received therapies for the lymphoid malignancy within six months, whereas 29 patients (11%) received their last treatment seven to 12 months prior to the first dose of vaccine. Overall, 169 patients (64Á2%) received therapies for the LM within 12 months before the administration of the first dose of vaccine and were defined as being on active treatment. They were further stratified according to their last therapy: 51 patients (19Á3%) received anti-CD20 antibody plus chemotherapy, of whom 40 (15Á2%) in the last six months; 36 (13Á6%) received chemotherapy alone, of whom 33 (12Á5%) in the last six months; 26 (9Á9%) IMIDs, 21 (8%) novel oral agents (ibrutinib or venetoclax without rituximab in the last 12 months), 21 patients (8%) CAR-T cells or HSCT, of whom six (2Á2%) in the last six months, and 14 (5Á3%) other therapies. Patients treated with autologous HSCT were considered as treated with chemotherapy alone or anti-CD20 antibody plus chemotherapy if they received rituximab before chemotherapy conditioning. The remaining 94 (35Á8%) patients were on "watch and wait" or follow-up strategies, since they had not received treatment for their malignancy yet or they had been treated more than 12 months before vaccination.

Overall, 131 (49Á8%; 95% CI 43Á6%-56Á0%) patients seroconverted four weeks after the first dose and 39 [14Á8%; 95% confidence interval (CI) 11Á0%-19Á6%] two weeks after the second one, for a total of 170 (64Á6%; 95% CI 58Á5%-70Á4%), as shown in Table I . The median antibody titre at two weeks after the second dose was 175 U/ml [interquartile range (IQR) 0Á44-2 600]. We compared the seroconversion rate and the antibody titres of 167 LM patients and 167 HCW matched for age and sex (73 female, median age 56 years, range: 46-62 in both groups). In the HCW group, the seroconversion rate was 99Á4% (95% CI 96Á7%-100%) as compared to 64Á1% (95% CI 56Á3%-71Á3%) in the matched LM group (P < 0Á001). Interestingly, we observed lower antibody titres in LM patients compared to the healthy subjects (median = 1 078 U/ml, IQR 643-1 841 U/ml vs median = 207Á5 U/ml, IQR 0Á44-3 062 U/ml, P < 0Á001; Fig 1A) .

At two weeks after the second dose, the seroconversion rate was similar among patients who had received their last treatment within six months and in those who had been treated seven to 12 months prior the first dose of vaccine (53Á6%, 95% CI 45Á3%-61Á6% vs 62%, 95% CI 44%-77Á3%, P = 0Á42 respectively), suggesting a long-lasting immunosuppressive effect due to anti-lymphoma treatments. Considering the lack of statistical difference, we decided to consider only treatment in the last 12 months for univariable and multivariable analyses.

At univariable analysis by binary logistic models, the variables significantly associated with the lack of serological response included: treatment in the last 12 months, type of LM, lymphopenia (<800 cells/ll), IgM levels < 40 mg/dl, and, among patients on treatment, the administration of anti-CD20 antibody plus chemotherapy in the last 12 months (Table II) .

The rate of seroconversion and the antibody titre at two weeks after the second dose were lower for subjects on active treatment compared to those untreated or in follow-up [55% vs 81Á9% respectively, odds ratio (OR) 3Á7, 95% CI 2Á02-6Á79, P < 0Á001; median = 11 U/ml, IQR 0Á44-1 320 U/ml vs median=1 686 U/ml, IQR 83Á43-3 965 U/ml, P < 0Á0001; Fig 1B) ]. Among actively treated patients, those who had received anti-CD20 antibody plus chemotherapy had a lower seroconversion rate (17Á6%) and antibody titres (median = 0Á44 U/ml, IQR 0Á44-0Á44 U/ml) compared to patients receiving other treatments (71Á2%, OR 0Á09, 95% CI 0Á04-0Á20, P < 0Á001; and median=183 U/ml, IQR 0Á44-2 993 U/ml, P < 0Á0001, respectively; Fig 1C) . Moreover, seven patients received the anti-CD20 antibody alone: five were on maintenance therapy whereas two received anti-CD20 as their last no-maintenance therapy. In the maintenance subgroup, the seroconversion rate was 40% (two out of five), whereas none of the remaining two patients developed a humoral response. A low seroconversion rate was reported also among patients treated with chemotherapy alone (69Á4%) and among those treated with novel oral agents (52Á3%). Interestingly, we observed a higher seroconversion rate in myeloma patients treated with IMIDs (84Á6%). A seroconversion rate of 100% was observed in 10 myeloma patients treated with anti-CD38 antibody (daratumumab) in monotherapy or in combination with lenalidomide or bortezomib. Among subjects who were treated with CAR-T cells or HSCT the seroconversion rate was 57Á1%. Specifically, the seroconversion rate of the 12 allo-HSCT recipients was 83Á3% whereas among the nine patients treated with CAR-T cells, the seroconversion rate was 22%. Patients without lymphopenia had a higher seroconversion rate (73Á6%) and higher antibody levels (median = 639 U/ ml; IQR 0Á44-3 324 U/ml) compared to those with lymphopenia (42Á8%, OR 3Á73, 95% CI 1Á95-7Á16, P < 0Á001; median = 0Á44 U/ml, IQR 0Á44-253, P < 0Á0001; Fig 1D) . We also observed a lower seroconversion rate (51Á7%) and antibody titres (median = 3Á42 U/ml; IQR 0Á44-557 U/ml) among subjects with IgM levels < 40 mg/dl compared to those with IgM levels ≥ 40 mg/dl (71Á1%, OR 2Á3, 95% CI 1Á25-4Á23, P < 0Á011; median = 1 267 U/ml, IQR 0Á44-4 731, P < 0Á001; Fig 1E) .

Age, sex, disease status, absolute neutrophil count at the time of the first dose, IgG and IgA levels had no statistically significant association with seroconversion rate. Nonetheless, we observed a higher antibody titre in patients with IgG levels ≥ 600 mg/dl compared to those with IgG < 600 mg/dl (median = 437 U/ml; IQR 0Á44-3 675 U/ml; median = 14Á5 U/ml; IQR 0Á44-719Á8 U/ml, respectively, P = 0Á003; Fig 1F) and in subjects with IgA levels ≥ 80 mg/dl compared to those with IgA < 80 mg/dl (median = 456 U/ml; IQR V. Marasco et al. Fig 1H) .

In the multivariable logistic model (Table III) , the independent predictors for seroconversion were: type of treatment (P < 0Á001), type of LM (P < 0Á001), absolute lymphocytic count (ALC) < 800 cells/ml (P = 0Á001) and IgM levels < 40 mg/dl (P = 0Á004; Table III ). In the multivariate analysis conducted in the subgroup of patients on active treatment, the independent predictors of humoral response were: type of treatment (P < 0Á001), type of malignancy (P = 0Á002), and IgM levels < 40 mg/dl (P = 0Á030; Table IV) .

With a median follow-up of three months after two doses of mRNA vaccines, we did not observe any case of SARS-CoV-2 infection. 

Considering the importance of T-cell immunity in COVID-19, 16 Figure S1 and Data S1. Moreover, when compared to the control group, the ppike-specific T-cell response from LM patients showed a lower median value of IFN-c (median = 179Á5 U/ ml, IQR 28Á4-369Á6 pg/ml vs median = 309Á0 pg/ml, IQR 181Á3-662Á8 pg/ml, P < 0Á0001) and of TNF-a (me-dian=32Á71 pg/ml, IQR 6Á2-92Á5 pg/ml vs median=104Á0 pg/ ml, IQR 17Á4-199Á0 pg/ml, P < 0Á0001 ; Fig 2A) . Unexpectedly, we reported an increased median value of IL-2 among LM patients compared to healthy subjects (median=553Á6 pg/ ml, IQR 90Á2-1039Á0 pg/ml vs. median=196Á0 pg/ml, IQR 99Á9-303Á8 pg/ml, P < 0Á0001 ; Fig 2A) that would require further investigations in a larger cohort. IFN-c values were directly correlated with IL-2 and TNF-a levels (R = 0Á5723, P < 0Á0001 and R = 0Á4813, P < 0Á0001, respectively; Fig 2B, C) . Moreover, we also confirmed a positive correlation between IL-2 and TNF-a levels (R = 0Á4813, P < 0Á0001; Fig 2D) . Th1-associated cytokine release was not significantly correlated with serological responses, neither when evaluating IFN-c levels (R = 0Á071, P > 0Á05) nor IL-2 levels (R = 0Á048, P > 0Á05). T-cell immune response was detectable in 47 (98%) and 38 (74%) seropositive and seronegative patients, respectively. IFN-c levels were lower among subjects who received chemotherapy with or without anti-CD20 antibody in the last month (75%) compared to those who received chemotherapy with or without anti-CD20 antibody more than two months before vaccination (90%). Cytokine release was higher in patients on active treatment with IMIDs (94%) or with novel oral agents (ibrutinib or venetoclax; 91%). Thirteen individuals (13%) were defined as "double negative" with neither T-cell nor humoral responses after a complete vaccination schedule. Among these nonresponders, 11 received chemotherapy with or without anti-CD20 antibody in the last month, one was on treatment with lenalidomide (IMID) and one with ibrutinib. This represents a population potentially at high risk for COVID-19 infection. state and have a mortality rate ranging from 30% to 37%. 1,3,17 It is therefore essential to implement an effective vaccination strategy. For these reasons, in Italy, LM patients have been among the first categories vaccinated with mRNAbased SARS-CoV-2 vaccines. To the best of our knowledge, this is the first report comparing both humoral and Tcellular responses after mRNA full-dose vaccination in LM and healthy individuals. As reported in other studies evaluating the efficacy of seasonal vaccines among cancer patients, 10,18 we observed a seroconversion rate in LM patients which is significantly lower than in healthy individuals. Interestingly, the seroconversion rate increased from 49Á8% to 64Á6% after the second dose, emphasising the importance of receiving both doses and complying with the official guidelines regarding the timing of administration of the second dose. Our results are in line with those from other groups who reported a lower seroconversion rate in LM after only one dose. [19] [20] [21] [22] The main negative predictive factor for the seroconversion was an active treatment, however even in the watch-and-wait cohort we observed a lower response compared to the healthy individuals, supporting the notion that patients can be immunecompromised by the disease itself.

Consistent with previous reports on pneumococcal and H1N1 vaccination, 9, 23 we detected an extremely low seroconversion rate among patients who received anti-CD20 antibody plus chemotherapy in the 12 months preceding vaccination. This finding can be explained by the known prolonged half-life of rituximab which is detectable in the serum at lympholytic levels for up to six months after therapy completion and by the subsequent long-lasting B-cell depletion. 24, 25 The B-cell depletion is more durable in patients receiving rituximab maintenance, raising the problem of the more appropriate timing for vaccination in this setting.

We observed a low seroconversion rate among patients treated with novel oral agents. Sun et al. have also reported low antibody response rates after influenza vaccine in CLL patients treated with Bruton Tyrosine kinase (BTK) inhibitors, 26 due to the alteration of the B-cell receptor signalling pathway. Our results are slightly better than those recently reported by Herishanu et al., 22 probably because we considered CLL patients treated with venetoclax without the anti-CD20 antibody. Furthermore, we found a decreased seroconversion rate among patients who received CAR-T therapy, especially in those treated in the six months preceding vaccination, but this is expected considering the impaired immune reconstitution reported after these procedures. 27, 28 The high seroconversion among allo-HSCT recipients was unexpected, but it is probably due to the small number of patients evaluated and to the fact that all these patients had suspended their immunosuppressive therapy.

Interestingly, we observed that myeloma patients receiving IMIDs had an increased antibody response relative to subjects receiving other therapies. Our finding is in line with other reports highlighting the potentially immune-adjuvant role of IMIDs after pneumococcal or H1N1 vaccinations. 29 The elevated response rate reported in patients treated with daratumumab is in contrast with other recent studies showing a negative impact on humoral response. 20, 30 It is possible that small numbers and concomitant treatment with IMIDs are confounding factors; nevertheless further investigations are required to clarify this aspect.

Previous studies have reported the impact of low IgM serum levels on the incidence of non-neutropenic infections in patients treated with rituximab-based regimens. 31, 32 This particular form of hypogammaglobulinaemia probably is a manifestation of B-cell compartment perturbation secondary to rituximab-containing regimens and the lymphoid malignancy itself.

Although the precise definition of all factors responsible for protection against COVID-19 remains to be fully determined, it is known that both neutralising antibodies and antigen-specific T cells play important roles. 16,33-35 Given the rather low seroconversion rate among patients who received their last treatment in the 12 months preceding vaccination, we decided to investigate their T-cell response.

We documented the in vitro release of Th1-associated cytokines upon SARS-CoV-2 spike stimulation, suggestive of an effective T-cell-mediated immune response, in an unexpectedly high rate of patients receiving active treatment. This was confirmed in almost every patient with a serological response after two doses of vaccine, as T lymphocytes are necessary for the production of high-affinity antibodies and generation of memory B-cells. Interestingly, cytokine release was detected in 74% of seronegative patients. This finding is novel and relevant, as it highlights the potential protection against infection provided by T lymphocytes in seronegative patients. Finally, we also described a small fraction of "double-negative" patients with neither serological nor cellular responses, who mostly received chemotherapy with or without anti-CD20 in the month preceding vaccination. This subgroup is at high risk of infection and potentially lifethreatening complications.

Although our study shows a disappointingly low seroconversion rate among patients on active treatment with anti-CD20 antibody plus chemotherapy, the intriguing data on the T-cell-specific response suggest a potential benefit for vaccination despite the lack of fseroconversion.

Additionally, the preliminary data in recipients of a solid organ would imply that a third vaccine dose can be a reasonable option to improve seroconversion rates in particularly fragile settings. 36 Thus, although prospective studies are required to assess the efficacy of T-cell immunity in preventing morbidity and mortality of COVID-19, our data may be in favour of trials evaluating the administration of a third vaccine dose to patients with evidence of at least T-cell immune response after the first two doses. On the other hand, in double-negative subjects the rationale for a third dose seems weak and the prophylactic administration of second-generation neutralising monoclonal anti-spike antibodies should be tested in clinical trials.

Our study was not able to clearly identify an optimal time for vaccination for LM patients that could probably depend on the last treatment. Our serological findings support the idea that all LM patients should receive vaccination prior to therapy at disease onset or at disease relapse and should wait at least 12 months after any treatment containing anti-CD20 antibody and at least six months after chemotherapy alone. However, the analysis of the T-cell response showed more encouraging data, suggesting a possible benefit of the vaccination even early after treatment discontinuation. Thus, further investigations will be needed to define the best accurate timing for vaccination.

In conclusion, we demonstrated a lower seroconversion rate among LM patients compared to healthy individuals, in particular among patients on treatment with anti-CD20 antibody and chemotherapy and a better than expected T-cellmediated response. These results confirm the value of vaccination strategy, but indicate the need for more effective measures for LM patients, to prevent serious SARS-CoV-2 infections and reduce their exceedingly high mortality rate.

Clinical characteristics and risk factors associated with COVID-19 severity in patients with haematological malignancies in Italy: a retrospective, multicentre, cohort study

Risk factors and outcome of COVID-19 in patients with hematological malignancies

Outcomes of patients with hematologic malignancies and COVID-19: a systematic review and meta-analysis of 3377 patients

Efficacy and safety of the mRNA-1273 SARS-CoV-2 vaccine

Safety and efficacy of the BNT162b2 mRNA Covid-19 vaccine

An mRNA vaccine against SARS-CoV-2 -preliminary report

Safety and immunogenicity of the SARS-CoV-2 BNT162b1 mRNA vaccine in younger and older Chinese adults: a randomized, placebo-controlled, double-blind phase 1 study

COVID-19 elicits an impaired antibody response against SARS-CoV-2 in patients with haematological malignancies

Rituximab blocks protective serologic response to influenza A (H1N1) 2009 vaccination in lymphoma patients during or within 6 months after treatment

Long-term patterns of humoral and cellular response after vaccination against influenza A (H1N1) in patients with hematologic malignancies

Vaccines for prophylaxis of viral infections in patients with hematological malignancies

Aetiology and resistance in bacteraemias among adult and paediatric haematology and cancer patients

Vaccination of haemopoietic stem cell transplant recipients

Vaccination of patients with haematological malignancies who did not have transplantations: guidelines from the 2017 European Conference on Infections in Leukaemia

Coordinate induction of humoral and spike specific T-cell response in a cohort of italian health care workers receiving BNT162b2 mRNA vaccine

CD8+ T cells contribute to survival in patients with COVID-19 and hematologic cancer

Outcomes of patients with hematologic malignancies and COVID-19: a report from the ASH Research Collaborative Data Hub

Efficacy of the influenza vaccine in patients with malignant lymphoma

Safety and immunogenicity of one versus two doses of the COVID-19 vaccine BNT162b2 for patients with cancer: interim analysis of a prospective observational study

Fifth-week immunogenicity and safety of anti-SARS-CoV

BNT162b2 vaccine in patients with multiple myeloma and myeloproliferative malignancies on active treatment: preliminary data from a single institution

Response to first vaccination against SARS-CoV-2 in patients with multiple myeloma

Efficacy of the BNT162b2 mRNA COVID-19 vaccine in patients with chronic lymphocytic leukemia

Infectious diseases and immunological markers associated with patients with non-Hodgkin lymphoma treated with rituximab

Rituximab in B-cell hematologic malignancies: a review of 20 years of clinical experience

Rituximab exhibits a long half-life based on a population pharmacokinetic analysis in non-hodgkin's lymphoma (NHL) patients

Seasonal influenza vaccination in patients with chronic lymphocytic leukemia treated with ibrutinib

Immune reconstitution after allogeneic hematopoietic stem cell transplantation

Hematopoietic recovery and immune reconstitution after axicabtagene ciloleucel in patients with large B-cell lymphoma

Lenalidomide-induced immunomodulation in multiple myeloma: impact on vaccines and antitumor responses

The neutralizing antibody response post COVID-19 vaccination in patients with myeloma is highly dependent on the type of anti-myeloma treatment

Impaired response to influenza vaccine associated with persistent memory B cell depletion in non-hodgkin's lymphoma patients treated with rituximab-containing regimens

High incidence of nonneutropenic infections induced by rituximab plus fludarabine and associated with hypogammaglobulinemia: a frequently unrecognized and easily treatable complication

Deep immune profiling of COVID-19 patients reveals patient heterogeneity and distinct immunotypes with implications for therapeutic interventions

Approaches and challenges in SARS-CoV-2 vaccine development

T cell responses in patients with COVID-19

Three doses of an mRNA Covid-19 vaccine in solid-organ transplant recipients

The authors wish to thank the patients who made this research possible, Ilaria Lo Russo and Chiara Ghidoli for the collection of blood samples and the study coordinators. 

VM, CC, CA, and PC designed the study and wrote the manuscript; AG, LF, GA and GI helped design the study; CC, MM and DM conducted the experiments; RM made statistical analyses and wrote the paper; SL and PV made statistical analyses; LC wrote the manuscript; FS enrolled patients; all authors provided final approval of the manuscript.

The authors declare no competing financial interests.

Additional supporting information may be found online in the Supporting Information section at the end of the article.

Data S1. Supplementary methods.