Risk of infection in patients with hematological malignancies receiving CAR T-cell therapy: systematic review and meta-analysis

ABSTRACT Background Chimeric antigen receptor (CAR) T-cell therapy has emerged as a promising treatment option for relapsed or refractory B-cell malignancies and multiple myeloma. Underlying and treatment-related variables may contribute to the development of infectious complications. Research design and methods We conducted a systematic review and meta-analysis on the incidence of overall and severe (grade ≥3) infection in patients with hematological malignancies receiving CAR T-cells. Secondary outcomes included the specific rates of bacterial, viral and invasive fungal infection (IFI), and infection-related mortality. PubMed, Embase and Web of Science databases were searched from inception to 27 May 2022. Sensitivity analysis were performed according to the type of malignancy and study design (randomized clinical trials [RCTs] or observational studies). Results Forty-five studies (34 RCTs) comprising 3,591 patients were included. The pooled incidence rates of overall and severe infection were 33.8% (I2 = 96.31%) and 16.2% (I2 = 74.41%). The respiratory tract was the most common site of infection. Most events were bacterial or viral, whereas the occurrence of IFI was rare. The pooled attributable mortality was 1.8% (I2 = 43.44%). Conclusions Infection is a frequent adverse event in patients receiving CAR T-cell therapy. Further research should address specific risk factors in this population.


Introduction
As cancer incidence and mortality are rapidly increasing worldwide, about 1.2 million of cases of hematological malignancies are diagnosed every year (accounting for ≈7% of newly diagnosed cancers) [1,2]. Developments in oncological treatment over the past few decades, however, have led to a significant improvement in survival rates [3]. Chimeric antigen receptor (CAR) T-cell therapy constitutes an emerging adoptive immunotherapy that has achieved a significant success in hematological cancer patients and, particularly, those with relapsed or refractory (r/r) B-cell malignancies and multiple myeloma (MM) [4].
Autologous T-cells are collected from the peripheral blood by leukapheresis and engineered in vitro to express artificial receptors targeted to specific tumor antigens, such as CD19 or B-cell maturation antigen (BCMA) [5]. Following in vitro expansion and infusion, redirected CAR T-cells target and kill tumor cells with high specificity. CARs are fusion proteins that comprise the extracellular antigen-binding component of a monoclonal antibody, various hinge and transmembrane domains, and the intracellular signaling domains of one or more T-cell receptors [6]. Depending on the structure of this latter component, CAR T-cells are categorized into distinct generations with different antitumor activity [7]. Second-and third-generation CARs contain additional co-stimulatory domains that enhance T-cell activation and proliferation [6,[8][9][10]. Fourth-generation CARs comprise transgenes for cytokine release and co-stimulatory ligands that recruit innate immunity and increase the resistance of CAR T-cells to the immunosuppressive tumor microenvironment [9][10][11].
Unlike hematological malignancies, solid cancer cells do not often express one tumor-specific marker, which would markedly increase the risk of on-target off-tumor toxicity. Additional difficulties lie in the difficulty of T-cell trafficking into solid tumor tissues due to the stromal barriers, tumor microenvironment, and tumor-induced T-cell exhaustion. Therefore, the clinical development of CAR T-cell therapy is far more advanced for hematological malignancies -such as large B-cell lymphoma, acute lymphoblastic leukemia (ALL) or MM -than for solid cancers [12][13][14].
The efficacy and safety of CAR T-cell therapies in patients with hematological malignancies have been evaluated in previous meta-analyses [24][25][26][27][28][29]. These works were mainly focused on clinical response rates and typical adverse events like cytokine release syndrome (CRS) or immune effector cellassociated neurotoxicity syndrome (ICANS), which were well defined and reported in randomized clinical trials (RCTs). The occurrence of prolonged cytopenias, infections and different off-tumor effects are other well established complications associated to the use of CAR T-cells [30]. However, no previous systematic reviews or meta-analyses have primarily addressed the incidence and risk factors of infectious complications in hematological patients receiving this therapy.
The risk of infection related to CAR T-cell therapy results from the interplay of a number of factors, such as the immune dysfunction induced by the underlying disease, the cumulative effect of prior lines of therapy, and the lymphodepletion administered before CAR T-cell infusion [31,32]. B-cell aplasia and the resulting hypogammaglobulinemia (HGG) also act as contributing factors [33]. Finally, it should be taken into account the frequent requirement of the anti-interleukin (IL)-6 monoclonal antibody tocilizumab (TCZ) and high-dose corticosteroid boluses for the management of CRS and ICANS [32][33][34]. A comprehensive assessment of the disease burden posed by infection events in CAR T-cells receptors -including the typically older and heavily treated MM patients -may be useful to refine prevention strategies, including prophylaxis practices, in this specific population.
In the present systematic review and meta-analysis we aimed to offer an updated estimate of the incidence of overall and specific types of infection among patients with hematological malignancies that received CAR T-cell therapy in the setting of RCTs and observational studies. We also assessed the risk factors for the development of infectious complications and the pooled attributable mortality.

Study design
This systematic review and meta-analysis was designed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [35]. The protocol was registered in the international Prospective Register of Systematic Reviews (PROSPERO) database.

Eligibility criteria
We included RCTs and prospective and retrospective observational studies performed in adult patients (≥18 years) diagnosed with hematological malignancies (B-cell and MM) and treated with CAR T-cells that reported absolute and/or relative frequencies of infection. Both comparative and noncomparative studies were included. All the studies had to be published in full-text in English to be considered. Reviews, previous meta-analyses, clinical guidelines, case reports and series, editorials, animal studies and conference abstracts were excluded, as were studies lacking essential data (i.e. the overall number of included patients and the number of participants that developed infection), those conducted in the pediatric population, and those published in languages other than English. In case of partially overlapping publications, only the study with the highest number of patients was included in the analysis.

Search strategy
We searched PubMed (Medline), Embase and Web of Science databases by using the following combination of terms: ('chimeric antigen receptor' OR 'CAR-T' OR 'chimeric antigen receptor-T cell therapy' OR 'CAR T-cell therapy' OR 'receptor, chimeric antigen' OR 'chimeric antigen receptor-modified T cell therapy') AND ('leukemia' OR 'multiple myeloma' OR 'lymphoma' OR 'hematological malignancies') AND ('infection' OR 'infectious complications' OR 'adverse events' OR 'side effects' OR 'reactivation'). Electronic databases were searched from inception to 27 May 2022. The references of the resulting articles were reviewed to avoid missing relevant publications. In addition, the references cited in review articles and previous meta-analyses [24][25][26][27][28][29] were manually screened for any potentially related study.

Data extraction
The following data were extracted from each study by two investigators (G.T.D. and M.F.R.) independently: study characteristics (first author, year of publication, sample size, median follow-up period since CAR T-cell infusion); patient characteristics (median or mean age at study entry, gender, underlying hematological malignancy, disease status, median number of previous lines of chemotherapy, previous hematopoietic stem cell transplantation [HSCT]); treatment characteristics

Article highlights
• No previous meta-analyses analyzing the safety of CAR T-cell therapy in patients with hematological malignancies have been specifically focused on the incidence of and risk factors for infection in this at-risk population. • In the present systematic review and meta-analysis (45 studies with 3,591 patients) the pooled incidence rates for overall and severe infection were high (33.8% and 16.2%, respectively), although moderate-to-high heterogeneity was observed for all outcomes. • Upper respiratory tract infection, pneumonia and sepsis or bloodstream infection were the most commonly reported sites. The pooled estimate for infection-related mortality was low (1.8%). • Most of the included studies did not provide granular data regarding the site of infection and causative agent, or whether the patient was receiving or not prophylaxis. The occurrence of IFI, however, was rarely reported. • Prior HSCT, the number of previous lines of chemotherapy, the presence of baseline neutropenia, the severity of cytokine release syndrome and the requirement of anti-IL-6 agents and corticosteroids were identified in the few studies that assessed the risk factors for infection after CAR T-cell therapy.
(lymphodepletion regimen, CAR T-cell product and dose, use of corticosteroids and TCZ for the management of CRS and ICANS); type of antimicrobial prophylaxis administered;

Methodological quality of the included studies
Cohort studies included in the meta-analysis were evaluated through the Newcastle-Ottawa Quality Assessment Scale, which uses a pre-established star-rating system (from 0 to 9) to assess the quality of non-randomized studies [36]. Studies scoring ≥5 stars were considered to be of moderate to high quality. Discrepancies regarding study eligibility, data extraction or quality assessment were resolved by consensus.

Primary and secondary outcomes
The primary outcome of this meta-analysis was the incidence rate of overall and severe (grade ≥3) infection in patients with hematological malignancies receiving CAR T-cell therapy. Secondary outcomes included the incidence rates for specific types of infection (bacterial, viral and IFI) and the infectionrelated mortality rate. We performed a number of pre-specified sensitivity analyses stratified according to the type of study (RCTs versus observational studies) and the type of the underlying disease (B-cell malignancy [lymphoma and ALL] versus MM).

Statistical analysis
Statistical analysis and figures were performed with the OpenMeta [Analyst] tool. We performed a meta-analysis of proportions to estimate the pooled rates of study outcomes. Pooled incidence rates were calculated with the corresponding 95% confidence intervals (CIs). Heterogeneity was evaluated by the Cochran's Q test (which was considered significant at a P-value <0.05) and quantified with the I 2 statistic. It can take values from 0% (no observed heterogeneity) to 100% (complete heterogeneity), with I 2 values of <25%, 25% to 75%, and >75% interpreted as representing low, moderate and high heterogeneity levels, respectively [37]. Randomeffects model with the Mantel-Haenszel method was used for pooling results from primary studies in the presence of significant heterogeneity; otherwise, a fixed-effects model was applied.

Literature search and study selection
In total, 5,687 potentially relevant citations were retrieved. After deduplication, 3,648 studies were screened. By reviewing titles and abstracts, 2,752 studies excluded and 896 full-text articles were evaluated for eligibility. Finally, after exclusion of studies performed in pediatric or mixed populations, animal studies and articles with irrelevant results, 45 eligible studies [16,18,20,21, were included. The flowchart of the study selection process is shown in Figure 1.

Infectious complications
The incidence rates of primary (overall and severe infection) and secondary outcomes (bacterial infection, viral infection and IFI) across included studies are shown in Table 2.    [54] No antibacterial prophylaxis was given No antiviral prophylaxis was given No antifungal prophylaxis was given No anti-PCP prophylaxis was given (Continued )   (Figure 2). When the studies from Yan et al. [51] and Ramos et al. [58] were removed due to the very low occurrence of infection (4.8% and 5.0%, respectively), the pooled incidence rate was 39.8% (95% CI: 33.7-45.9) with a slight decrease of heterogeneity (I 2 = 83.73%; P-value <0.001). Details on microbiologically documented episodes of infection were provided in only 28 (62.2%) studies (Table S1 in Supplementary Material).

Discussion
Infection constitutes a relevant adverse event related to CAR-T cell therapies, with multiple factors contributing to this complication. The present systematic review and meta-analysis comprising 3,591 patients with hematological malignancies from 45 studies revealed a pooled incidence rate for overall        infection of 33.8%. The majority of the studies (75.6%) were RCTs, and all but one [42] were in phase 1 or 2. In the sensitivity analysis stratified by the type of design, the incidence of overall infection was even higher in observational studies (38.5%), suggesting that patients receiving CAR T-cell in real-life practice are more prone to develop infection than the highly selected population recruited in early phase trials. In a recent pharmacovigilance study and meta-analysis of safety data, the pooled incidence rate of infection of any grade was 27.7% (11 studies with high heterogeneity) [29]. The lower rate compared to our results may be explained by the inclusion of pediatric patients. In another meta-analysis restricted to ALL patients with consolidative HSCT following anti-CD19 CAR Tcell therapy, only three out of 11 studies reported data related to infections and yielded a pooled incidence of 39%, also with high heterogeneity [27]. Therefore, the present systematic review and meta-analysis provides the most comprehensive synthesis to date in terms of number of summarized studies, and the only one specifically devoted to assess the occurrence of infectious complications after CAR T-cell therapy.
The pooled estimate for severe infection -'on the basis of 26 studies with 1,745 patients -was nearly half that for overall infection (16.2%). All but two [75,78] of these studies were RCTs, which may explain the lower degree of heterogeneity observed (I 2 = 74.41%). In a meta-analysis comprising 15 studies on r/r ALL patients receiving CD19-specific CAR Tcells, the pooled cumulative incidence rate of infection events graded ≥3 was higher (29%) than that found in our study [28]. A direct comparison, however, should be taken with caution. Unlike our patient population, all participants in the metaanalysis by Aamir et al. had a diagnosis of ALL as underlying disease, and their age range (0 to 30 years) was also different. Since we specifically excluded pediatric or mixed studies, only three RCTs restricted to ALL patients [43,45,65] and six further studies comprising both ALL and B-cell lymphomas [46,56,60,61,73,78] were included. Finally, the differential impact of specific disease biology and previous lines of chemotherapy according to the type of B-cell malignancy (i.e. the common use of purine analogues in ALL) cannot be ruled out.
Two CAR T-cell therapies targeted against different tumorassociated antigens are currently used in clinical practice. Since BCMA is selectively induced during plasma cell differentiation [79,80], anti-BCMA CAR T-cells are being increasingly used in MM patients, with two FDA-approved products (idecabtagene vicleucel and ciltacabtagene autoleucel). Anti-CD19 CAR T-cells are preferred for B-cell malignancies, as CD19 is specifically expressed on the surface of normal and neoplastic B-cells and follicular dendritic cells [81]. In accordance with its more advanced clinical development, two thirds of the pooled studies were related to anti-CD19 CAR T-cells, in contrast to only five studies on anti-BCMA therapies. Heavily treated MM patients have a high risk of infection that result from B-cell dysfunction and associated HGG, immune defects involving Tcells and NK cells, older age and frequent comorbidities [82]. Therefore, it may be expected that infection events would be more common with anti-BCMA than with anti-CD19 CAR T-cell therapies. The pooled estimates for overall infection, however, were very similar between both patient populations (35.5% and 34.4%, respectively), whereas the incidence rate of severe infection was actually higher in patients with B-cell malignancies.
The pooled incidence rate of late infection was estimated at 36% based on data provided by 8 studies, again with high heterogeneity (I 2 = 94.21%). Of note, the definition of late infection and the follow-up period were not homogeneous across studies. For instance, Cappell et al. reported 6 episodes of infection requiring hospital admission (including one case of disseminated herpes zoster [HZ]) that were diagnosed up to 3 years after CAR T-cell infusion [44]. In another study that defined late infection as any event occurring beyond 90 days, the cumulative incidence was as high as 61%, which accounted for an incidence density of 0.55 per 100 days at risk (2.08 per patient-year). Although upper (48%) and lower respiratory infections (23%) were the most common syndromes, 20% and 5% of the episodes required hospital and intensive care unit admission, respectively [78]. Long-term Bcell depletion and HGG with delayed T-cell recovery have been well described following CD19-targeted CAR T-cell therapy, with some studies showing low serum immunoglobulin levels for up to 5 years after infusion [44].
antibodies, it has been shown that CD19-targeted CAR T-cell therapy negatively impacts the capacity to mount humoral responses following mRNA vaccination, although the amount of specific T-cell responses seems to be similar (or even higher) than healthy controls [83,84]. Less than two thirds of the analyzed studies provided data on the occurrence of IFI, which resulted in a pooled incidence of 2.0% with moderate heterogeneity (I 2 = 43.48%). Once again, the scarcity of specific information limits the risk assessment of this life-threatening complication. The development of invasive aspergillosis was anecdotally reported in seven studies (incidence rates: 0.9% to 3.8%) [39,46,48,56,65,76,78]. The timing of diagnosis and antimould prophylaxis status, however, were not given in most of them. The occurrence of PCP was overall uncommon [46,48,76], likely reflecting the widespread use of anti-Pneumocystis prophylaxis for at least 3-6 months or with the duration guided by the recovery of CD4 T-cell counts. Interestingly, in the study reporting the highest cumulative incidence of PCP (7.3%), all the three cases occurred within 3 months of completing the prespecified course of trimethoprimsulfamethoxazole prophylaxis despite the persistence of CD4 Tcell counts below 200 cells/μL [76].
Previous meta-analysis focused on efficacy and safety outcomes reported no separate rates for specific types of infection or causative agents. Only five out of 19 studies included in the meta-analysis by Xu et al. detailed the microbiological characteristics of the episodes of infection, reporting 36 cases of cytomegalovirus (CMV) infection [27]. The incidence of CMV infection in our meta-analysis largely varied from 0.3% [41] to 33.3% [45], although the type of event (i.e. asymptomatic viremia or clinical disease) is not usually provided. In addition, the cumulative incidence depends to some extent on the frequency of monitoring for CMV DNAemia, which was not homogeneous across studies.
Baseline neutropenia was identified in the study by Hill et al. as a risk factor for infection after CAR T-cell therapy, although the association lost its statistical significance after multivariate adjustment [46]. Monocytopenia and lymphopenia -in particular low CD4 T-cell levels -have been additionally explored as risk factors for viral and fungal infection [46,65,75,[85][86][87]. Due to the insufficiency of the data extracted these variables could not be examined in our meta-analysis. Various studies [46,64,65,69,71,75,76] have analyzed the development (and severity) of CRS as a potential risk factor, since it has been suggested that the associated endothelial damage would initiate or facilitate the infection process [85,88]. Nevertheless, only the study by Park et al. showed an independent association between grade ≥3 CRS and overall infection (adjusted hazard ratio [aHR]: 2.67; P-value = 0.05) and BSI (aHR: 19.97; P-value <0.001) [65]. After adjusting for clinical covariates in multivariate models, the number of prior lines of chemotherapy [46,76] and the use of corticosteroids for the management of CRS or ICANS [64] were ultimately identified as independent risk factors in the few studies that have specifically assessed this issue, most of them retrospective cohorts.
Whereas all-cause mortality rates are detailed in the majority of studies as a key safety outcome (with progression of the underlying disease as the most common cause of death), mortality attributable to adverse events was assessed in a limited number of papers only. One third of them reported no infection-related deaths. The resulting pooled rate was 1.8% with moderate heterogeneity, in line with a previous meta-analysis that included a much lower number of studies (1.3% [I 2 = 0%] based on three studies) [29].
Our systematic review and meta-analysis have some limitations. Since most of the articles were phase 1/2 RCTs, the number of patients per study was relatively low (20 trials had sample sizes <50 patients). Excess of small-sample studies can explain the wide variation observed in infection rates, which led to high heterogeneity for most of the estimates. Additional reasons of heterogeneity are the disparity in followup periods and the lack of consensus definitions for infection events. With a few exceptions -typically observational studies rather than RCTs [73,75,76,78] -most of the studies simply reported absolute and relative frequencies for overall and/or severe infection. Therefore, details on microbiologically documented episodes and sites of infection were limited. In addition, some trials only contained data for specific types (i.e. pneumonia) or specific pathogens (i.e. HZ). Definitions for early and late infection differed across studies. The timing of diagnosis or whether they occurred while the patient was under antimicrobial prophylaxis or following its discontinuation was not always discernible. Most of the studies that specifically investigated the risk factors for infection were retrospective in design [46,64,65,69,71,73,75,76]. In addition, the paucity and variability of available data across studies prevented us to perform a meta-analysis. Finally, some studies only reported the absolute number of infection events during the follow-up period, making difficult the estimation of cumulative incidence rates.
In conclusion, this systematic review and meta-analysis offer a comprehensive assessment on the incidence of infectious complications in patients receiving CAR T-cells for the treatment of different types of hematological malignancies. Infection revealed as a frequent adverse event associated with this emerging therapy. The occurrence of severe opportunistic infections such as invasive aspergillosis or PCP, however, was uncommon, and the attributable mortality was low. In addition, we have detected a need for prospective data evaluating the risk factors for this complication, as well as for a more detailed description of microbiologically documented infection events in RCTs.

Expert opinion
The advent of CAR T-cell therapy has revolutionized the approach to hematological malignancies, and major advances are to be expected in the coming years, with a growing number of approved products and indications. As revealed by the present systematic review and meta-analysis, infection remains a major complication following CAR T-cell therapy, since one in three patients will develop any-grade episode and one in six will suffer a severe event. Unfortunately, the implementation of effective risk minimization strategies may be problematic due to a number of reasons. First, the specific attribution of causality to CAR T-cell therapy in the pathogenesis of infection may be confounded by the impact of baseline immunodeficiency that usually develops among heavily pretreated patients with r/r ALL or MM, with a meaningful proportion of previous HSCT recipients. Second, in addition to the lymphodepletion regimen administered (FluCy in most of the studies), the variable occurrence of CRS and its therapeutic management -based on an anti-IL-6 agent with or without high-dose corticosteroids -represents a further source of patient heterogeneity. Third, a notable variation was also found across studies in terms of antimicrobial prophylaxis. In the absence of dedicated RCTs, prophylaxis practices following CAR T-cells have been mostly modelled upon regimens used in other at-risk groups such as neutropenic patients or allo-HSCT recipients. Nevertheless, CAR T-cell therapy entails some specific risks, such as long-term lymphopenia and HGG. This notion is supported by the high incidence rate of late infection observed in our meta-analysis (pooled estimate of 36.0%), even by considering the inconsistent definition of this outcome in the literature. Therefore, it is likely that the optimal timing for the discontinuation of prophylaxis must be established on an individualized basis and informed by the kinetics of CD4 + T-cells (i.e. PCP prophylaxis) or serum immunoglobulin levels. In addition, IVIg replacement therapy should be routinely implemented in patients with severe HGG (IgG <400 mg/dL), in line with the common practice in MM [89]. Lastly, the present research reveals that the majority of RCTs lack granularity in the reporting of infectious complications in study populations, which often was incomplete and unstructured. Although this drawback has been also noted in RCTs assessing other treatments for hematological malignancies [90], further efforts are urgently needed to standardize the report of infectious events after CAR T-cell with accepted definitions and classified by both type of pathogen and source of infection. Keeping in mind this limitation, the burden of IFI and other opportunistic infections (i.e. CMV disease) following CAR T-cell therapy seems to be relatively low. In fact, only a few studies required as per protocol the use of antiviral or mould-active antifungal prophylaxis. Thus, the main disease burden in patients receiving CAR T-cell therapy actually lies on the development of bacterial infection in form of pneumonia or bacteremia, which should prompt the assessment of the role of antibiotic prophylaxis and IVIg replacement in future RCTs.

Declaration of interest
The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or material discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or mending, or royalties.

Reviewer disclosures
Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Funding
This paper was funded by Instituto de Salud Carlos III (ISCIII), Spanish Ministry of Science and Innovation, and co-funded by the European Union -European Social Fund, 'Investing in your future.' M.F.R. holds a research contract 'Miguel Servet' (CP18/00073) from the ISCIII. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.