Infections in children following chimeric antigen receptor T‐cell therapy for B‐cell acute lymphoblastic leukemia

CD19‐directed chimeric antigen receptor T‐cell (CAR‐T) therapy is transforming care for pediatric patients with relapsed or refractory B‐cell acute lymphoblastic leukemia (B‐ALL). There are limited pediatric‐specific data concerning the infection risks associated with CD19 CAR‐T therapy and the adequacy of current antimicrobial prophylaxis guidelines for these patients.


INTRODUCTION
CD19-targeted chimeric antigen receptor T-cells (CAR-T) have changed the treatment paradigm for children, adolescents, and young adults (CAYA) with relapsed or refractory B-cell acute lymphoblastic leukemia (B-ALL). 1 With the expanding commercial availability of CAR-T-cell products, such as Tisagenlecleucel (Kymriah), an increasing number of patients are expected to benefit from this novel immunotherapy.Given its association with prolonged leukopenia, B-cell aplasia, and low immunoglobulin levels, understanding the infectious complications after CAR-T-cell infusion (CTI) is critical to informing appropriately targeted supportive care and prophylaxis strategies. 2 Patients receiving CAR-T-cell therapy, similar to those undergoing allogeneic hematopoietic stem cell transplant (HSCT), are considered to be at an increased risk of infection. 3Data from the Summary Product Characteristics (SPC) reported a Grade 3 or more infection risk of between 19% to 35% in CAYA patients. 2However, there are emerging data that suggest post-CTI infections may be more common.One CD19-CAR-T-cell study in CAYA patients (n = 39) reported up to 50% of patients experienced a total of 35 infections, with most occurring in the early post-infusion period between day 0 and day 28. 4 These realworld data, albeit limited to date, provide important insight into the key periods of risk, types of infection encountered and infection screening and prophylaxis protocols.
Prior to the implementation of CAR-T-cell therapy in 2019 at the Royal Children's Hospital (RCH), one of the largest tertiary pediatric hospitals in Australia, an infection screening and prophylaxis guideline was developed in collaboration with key stakeholders (medical and nursing representatives from infectious diseases, oncology, intensive care, and pharmacy).The guideline considered cumulative immunosuppression and epidemiologic and prior infections as risk factors.In the absence of pediatric-specific data, recommendations for bacterial, viral, Pneumocystis jirovecii pneumonia (PJP) and fungal prophylaxis were based on expert opinion, the SPC and limited adult data. 5e objective of this study was to describe the spectrum of infections experienced by CAYA patients undergoing CAR-T-cell therapy in the first 100 days and to evaluate the effectiveness of our infection screening and prophylaxis strategies.

Patients
This was a retrospective observational, single-center study of all chil-

Definitions
Microbiologically documented infection (MDI) was defined as an infection that was clinically detectable and microbiologically proven and further categorized as bacterial, viral, fungal, or parasitic. 6Bloodstream infection (BSI), a subset of MDI, was defined as a recognized pathogen cultured from one or more blood cultures 6 (including viridans group streptococci in the setting of concomitant mucosal barrier injury).To be classified as a BSI, common commensals such as coagulase-negative staphylococci, were identified from two or more blood cultures drawn on separate occasions.Clinically documented infection (CDI) was defined as a site of infection that was diagnosed, but its microbiological pathogenesis could either not be proven or was inaccessible to examination. 6 Neutropenia was defined as an absolute neutrophil count (ANC) <0.5 × 10 9 /L and lymphopenia as an absolute lymphocyte count (ALC) <1.0 × 10 9 /L.Cytokine release syndrome (CRS) was defined and graded using the American Society for Transplantation and Cellular Therapy (ASTCT) consensus guidelines. 8Neurotoxicity was assessed using the ASTCT immune-effector cell associated-neurotoxicity syndrome (ICANS) Consensus Grading for Children. 8

Antimicrobial prophylaxis and treatment
Clinical review by an infectious diseases fellow and physician was recommended for all patients prior to CAR-T-cell therapy for the purpose of developing a personalized plan for infection prophylaxis and empiric febrile neutropenia treatment (See Appendices A1 and 2 for proforma).
An infectious diseases physician also attended the multi-disciplinary CAR-T planning meeting for each patient where the rationale for this plan was discussed.
While awaiting CAR-T-cell therapy, antimicrobial prophylaxis was according to hospital guidelines 9

Statistical analysis
Descriptive statistics (median, range, percentage) were reported for key variables.We calculated the number of infections in the 90 days prior to CTI, the first 30 days post-CTI and 31-100 days post-CTI.The density of infections were described as total number of infections per 100 days and calculated pre-and post-CTI. 4Factors significantly associated with post-CTI infection identified in previous studies, including pre-CTI ALC <0.3 × 10 9 /L, prior HSCT and infection within 90 days as well as post CTI CRS or ICANS were also examined in our cohort.
The Mann-Whitney U test was used to estimate p values for continuous data, and the Fisher exact test was used for categorical data.All tests were two-tailed, and a p value <.05 was considered statistically significant.
Patients were censored if they died, were transferred back to referring centers, required repeat CTI, received salvage therapy for relapse or transitioned to HSCT, within the first 100 days post-CTI.
TA B L E 1 Demographic data and recommended post CAR-T infusion antimicrobial prophylaxis.

Patient characteristics
Demographic and patient characteristics of the 27 patients who received their first CTI during the study period are summarized in post-CTI.There were eight patients transferred back to their original treating center at a median of 32 days (IQR: 31-32 days), two who progressed to HSCT at days 44 and day 87 and one who relapsed at day 60.The overall infection density in the 90 days prior to CTI was 0.9 per 100 patient-days at risk.

CAR-T antimicrobial prophylaxis recommendations
Twenty-six patients (96%) were reviewed by an infectious disease fellow and physician prior to CTI and all prophylaxis recommendations were followed (Table 1).Twenty-four (89%) had an MRO screen, of which three (13%) identified a relevant pathogen, including an extended spectrum beta-lactamase gram-negative (n = 2) and carbapenem-resistant Enterobacteriaceae (n = 1), and informed a personalized empiric antibiotic plan for fever management.A personalized antifungal plan was also documented for all six patients with a previous IFI (Table 1).

Fever and/or infection episodes post-CTI
Day 0 to 30.Seventeen (63%) patients experienced at least one febrile episode within the first 30 days post-CTI.The median time to first fever was 5.3 days (IQR: 3.2-6.1 days) and of these, three (18%) were attributed to infection (BSI n = 2; viral respiratory infection, n = 1).
Overall, seven infection episodes were documented in five (19%) patients in the first 30 days post CTI (four patients had one infection, one patient had three infections) (Table 3).There were four bacterial infections, including three BSI.Notably, there were no new IFIs in the first 30 days post-CTI and both patients with recent IFI diagnoses did not have evidence of progressive fungal disease.
The infection density in the first 30 days following CTI was 0.9 per 100 patient-days at risk.3/22).In contrast, patients with an ICANS grade of 2 or more were significantly more likely to experience infection than those with ICANS less than 2 (3/3, 100% vs. 3/24, 13%, p = 0.007).
We also compared infection density before and after the opening of other pediatric CAR-T centres in Australia.The overall post CTI infection density in the period before July 2020 was 1.1 per 100 patient days, compared to 0.3 per 100 patient days July 2020 onwards.There was also a trend toward a higher proportion of patients with a documented infection before July 2020 compared with after this time (4/13, 31% vs. 2/14, 14%, p = 0.4); however, this did not reach statistical significance.
Infectious causes for neurotoxicity were not identified in any of these patients and steroid therapy was administered to three of the five patients for ICANS.
Overall, two patients experienced a relapse of their underlying disease within the first 100 days, including one that occurred after transfer.There were no deaths during the study period.

DISCUSSION
This study delineates the infections encountered within the first 100 We observed a lower infection density in patients undergoing CTI compared to previous studies. 4,11Pre-CTI, the infection density was 0.9 per 100 days-at-risk, which remained static in the first 30 days post-CTI and declined to 0.6 per 100 days-at-risk from days 31 to 100.
In contrast, two prior pediatric studies reported an early (days 0-30) postinfusion infection density of 2.4 4 and 2.9 per 100 days-at-risk. 12e late (days 31-100) post-CTI infection density reported in our study was, however, in keeping with these studies.While our relatively small sample size limit firm conclusions, we theorise that the comprehensive infectious diseases review and personalized antimicrobial plans, together with the multi-disciplinary planning meeting prior to each CTI, may have, in part, contributed to this observed lower rate of infection.
In keeping with previous reports of CRS, the median time to fever onset post-CTI was five days. 13Interestingly, an infection was identified at this time point in only 18% of patients.With a median duration of neutropenia post CTI of 14.9 days, it is perhaps not surprising that bacterial BSIs were the most common type of infection documented.
However, the observed proportion of patients with a BSI in the first 30 days was lower at 7% than has been previously reported (13% and 17%). 4,11In contrast, the proportion with a documented BSI after day 30 was higher in our cohort at 15% as compared to the same previous reports (5% and 9%). 4,11While the reason for this is not clear, it is important to highlight that antibacterial prophylaxis was not used in our patient cohort.
Fewer patients in our study (11%) had a documented viral infection post-CTI as compared to previous reports of up to 18%.Previous studies have identified risk factors significantly associated with post-CTI infections in CAYA with B-ALL.Patients with higher pre-infusion disease burden, exposed to multiple lines of antitumour treatment including HSCT and/or those who had a pre-CTI IgG less than 4 g/L were found to be more prone to infectious complications during the immediate post-CTI period. 14Peri-and post-CTI factors associated with infection have included intensity of bridging chemotherapy, lymphopenia, and higher-grade CRS. 14 Our study, limited by its sample size, could not adequately address host-and treatment-related factors that might influence the risk of infectious complications.Larger studies are needed to clarify if these findings are replicated in other centers.We did, however, note a trend toward a higher proportion of infections in patients undergoing treatment prior to July 2020 when our hospital was the only national CAR-T referral center.This was similarly reflected in a higher overall infection density of 1.1 per 100 patient-days compared to 0.3 per 100 patient-days after the opening of a two other CAR-T centres in Australia.While the exact reason for this is unknown, we postulate that more complex and heavily pre-treated patients were treated in the early phases of the CAR-T program.
Although this study is limited by its retrospective design, all patients had a comprehensive assessment prior to CTI using a standardized template documented in the electronic medical record.While results are also only applicable to a single centre, over half of the patients had received their primary ALL treatment at other centers.The analysis of infections beyond 30 days post-CTI was also restricted because of patient censoring at the point of transfer to original treating center, which may have led to a bias towards more complex and critically ill patients.Overcoming the limitations of this smaller dataset might be achieved through a national CAR-T-cell infection registry.This would inform supportive care protocols and standardized dren with relapsed and/or refractory B-ALL who received their first CTI at RCH from January 2019 until May 2023.The RCH is the only tertiary pediatric HSCT center in Victoria and, up to July 2020, was the national pediatric referral center for CAR-T-cell therapy.Leukapheresis, bridging therapy, and manufacture of tisagenlecleucel were performed as per standard institutional and/or manufacturer (Novartis Pharmaceuticals) protocols.Patients received a standardized lym-phodepletion conditioning protocol with fludarabine (30 mg/m 2 day 1-4) and cyclophosphamide (500 mg/m 2 day 1 and 2) followed by CTI.We reviewed the medical records for infections occurring in the 90 days prior, and up to 100 days following CTI.Patient data were collated up to 100 days post-CTI or date of loss to follow-up, relapse, repeat CTI, or conditioning for HSCT, whichever occurred first.A secure web-based application (Research Electronic Data Capture, REDCap) was used to capture all de-identified data and complied with the National Health and Medical Research Council (NHMRC) ethical code of conduct via de-identified health information and comprehensive auditing.
and included mould-active antifungal prophylaxis for patients with relapsed disease.CAR-T-cell antimicrobial prophylaxis (see Appendix 2) was commenced with lymphodepleting chemotherapy and included herpes simplex virus (HSV) and varicella zoster virus (VZV) prophylaxis with acyclovir or valaciclovir for three to 6 months and PJP prophylaxis with trimethoprim/sulfamethoxazole twice weekly for six months post CTI.Antifungal prophylaxis consisted of fluconazole daily until patients were day 30 post CTI and had a documented ANC greater than 0.5 × 10 9 /L for 7 days, with mould-active azole recommended for patients with previous invasive fungal infection (IFI), relapsed disease within 12 months of HSCT, CRS requiring tocilizumab, or ICANS requiring steroids.Fluoroquinolone antibacterial prophylaxis was not routinely used.Intravenous immunoglobulin (IVIG) supplementation was recommended for patients with two separately measured IgG levels less than 4 g/L.Empiric intravenous antibiotics were administered to patients with fever according to our institutional guidelines for febrile neutropenia and non-neutropenic fever, 10 unless a personalised empiric antibiotic treatment recommendation had been made, based upon routine multiresistant organism (MRO) screening swabs or known history of MRO infection or colonization at any time during the patients cancer treatment.The MRO screen included stool or rectal swab for extended spectrum beta-lactamase (ESBL), carbapenem resistant Enterobacteriaceae (CRE) and vancomycin resistant enterococci (VRE) and skin swabs for methicillin resistant Staphylococcus aureus (MRSA) and, if travelled overseas in last 12 months, Candida auris.
Antimicrobial prophylaxis in the 30 days preceding CTI varied.Twentysix (96%) patients received PJP prophylaxis, one received fluoroquinolone, 19 (70%) received antiviral prophylaxis and 24 (89%) received antifungal prophylaxis (11, fluconazole; 1, voriconazole; 2, posaconazole; 3, intermittent ambisome; 7, echinocandin).While routine fluoroquinolone prophylaxis for CTI is not used at our center, there was one patient who received this as secondary prophylaxis.The patient had a preceding history of multiple episodes of severe sepsis, including an episode of Bacillus cereus bacteraemia complicated by endocarditis and meningitis.They completed six weeks of treatment with ciprofloxacin and meropenem prior to CTI.Given this history, he had an individualized secondary prophylaxis plan recommended.A total of 22 infection episodes were documented in 11 (41%) patients in the 90 days before CTI (4 patients had one infection; five patients had two infections, two patients had four infections) (Table2).Bacterial infections were the most common, of which BSI accounted for nine infection episodes in six (22%) patients.Overall, six patients had an IFI diagnosed at any time before receiving CTI, of which two (7%) occurred in the 90 days prior.These two IFI's were classified as possible pulmonary IFIs based on characteristic radiological findings in the absence of positive microbiology.One patient had a complete workup with serum galactomannan and bronchoalveolar lavage (BAL) (including culture, galactomannan, and Aspergillus PCR), all of which were negative.The second patient had an incomplete work up with no BAL or serum galactomannan done.Both patients were commenced on mouldactive treatment prior to CTI.The four patients with IFI diagnosed earlier than this included three patients with a proven IFI (cutaneous mucormycosis, n = 2; pulmonary cryptococcus disease, n = 1) and one patient with two possible IFIs (possible hepatosplenic candidiasis and possible pulmonary IFI).

Day 31- 100 .a
Seven patients were transferred back to their primary treatment centers between days 31-33, leaving 20 patients who had further follow-up at RCH at least until day 44 in this period.There were seven infection episodes in three patients (two patients had two infections; one patient had 3).Five BSI occurred in three patients, four of which occurred with a corresponding ANC <1.0 × 10 9 /L.There were no viral infections in this time period, in particular, no clinical HSV or VZV reactivation.TA B L E 3 Infections by time period post CAR-T infusion.Staphylococcus haemolyticus; Enterobacter cloacae; Klebsiella pneumoniae (all n = 1).b Klebsiella pneumonia (n = 3); Pseudomonas aeruginosa and Streptococcus oralis (n = 1); Enterobacter cloacae (n = 1).c C. difficile enterocolitis.d Klebsiella pneumonia skin and soft tissue infection (n = 1).e respiratory syncytial virus; rhinovirus/enterovirus. f BK virus hemorrhagic cystitis.g Probable invasive fungal infection based on clinical and radiologic criteria and positive galactomannan.One patient was diagnosed with a probable pulmonary aspergillosis at day 63 post-CTI, whilst receiving micafungin prophylaxis.The diagnosis was made in the context of prolonged fever, fluorodeoxyglucose (FDG)-positron emission tomography (PET)-magnetic resonance imaging (MRI) demonstrating avid pulmonary and skin nodules and a positive galactomannan on bronchoalveolar lavage fluid.This patient had a complicated course with Grade 4 ICANS/neurotoxicity requiring steroids, multi-resistant BSI, a tension pneumothorax and a prolonged admission to the intensive care unit for respiratory support.As per our CAR-T guidelines that patient had initially received voriconazole prophylaxis before being changed to micafungin to avoid drug-drug interactions.The IFI responded to treatment with voriconazole.The infection density from days 31 to 100 following CTI was 0.6 per 100 patient-days at risk.Overall there were six patients that experienced one or more infections from day 0-100 post CTI.There was no significant difference in the proportion of patients that had a documented infection post CTI, compared to those without infection, with pre-CTI ALC <0.3 × 10 9 /L (1/3 vs. 5/24), HSCT (3/9 vs. 3/18) or infection within 90 days (2/11 v 4/16), as well as post CTI CRS (5/17 vs. 1/10) or any ICANS (3/5 vs. days post-CD19 CAR-T-cell therapy in a cohort of pediatric patients with relapsed or refractory B-cell ALL.During part of the study period, our hospital served as the national referral centre for pediatric CAR-T therapy with 60% of patients referred from other hospitals.Almost all TA B L E 4 Other CAR-T cell-related complications and treatments in first 100 days.) had a comprehensive infectious diseases review as part of the pre-CTI assessment, which informed a personalised prophylaxis or fever management plan in six (22%).Overall, six (22%) patients had one or more infections during the study period, including in five (19%) patients in days 0-30 and three (15%) patients in days 31-100.Bacterial BSIs were the most common type of infection encountered during both time periods and one patient had a new IFI documented.There were no infection-related deaths.
4,11 Notably, there were no viral infections documented in the 20 patients followed from days 31-100.While this may be a limitation of the retrospective analysis, these patients received close clinical follow-up as outpatients during this time period.Our center does not routinely monitor post-CTI cytomegalovirus and Epstein-Barr viraemia, which may have also contributed to the lower observed rate of viral infections.Finally, as viral respiratory illness accounted for only two infection events, it is plausible that the heightened infection control measures during the COVID-19 pandemic influenced these results.Reassuringly, breakthrough IFI in the first 100 days post-CTI was rare, and no patient with an IFI diagnosis preceding CTI had progression of fungal disease post-CTI.The patient with an IFI diagnosed post CTI had a complicated course and had appropriately received mould-active azole prophylaxis before changing to micafungin to avoid drug-drug interactions.Although this reflects a breakthrough infection, the appropriate diagnostic workup (lung imaging and BAL) facilitated early diagnosis and targeted Aspergillus species treatment resulting in complete cure.

Table 1 .
All patients had relapsed or refractory B-ALL and the median mary therapy, or after one cycle of chemotherapy for relapsed disease.Eleven patients were from RCH, while 16 patients (59%) were referred from other sites, including 14 from interstate or New Zealand.The median duration of follow-up was 100 days (interquartile range [IQR]: 38-100) with 16 patients contributing data up until 100 days TA B L E 2 Infections pre-CAR-T infusion (day −90 to day 0).