Management guidelines for paediatric patients receiving chimeric antigen receptor T cell therapy

In 2017, an autologous chimeric antigen receptor (CAR) T cell therapy indicated for children and young adults with relapsed and/or refractory CD19+ acute lymphoblastic leukaemia became the first gene therapy to be approved in the USA. This innovative form of cellular immunotherapy has been associated with remarkable response rates but is also associated with unique and often severe toxicities, which can lead to rapid cardiorespiratory and/or neurological deterioration. Multidisciplinary medical vigilance and the requisite health-care infrastructure are imperative to ensuring optimal patient outcomes, especially as these therapies transition from research protocols to standard care. Herein, authors representing the Pediatric Acute Lung Injury and Sepsis Investigators (PALISI) Network Hematopoietic Stem Cell Transplantation (HSCT) Subgroup and the MD Anderson Cancer Center CAR T Cell Therapy-Associated Toxicity (CARTOX) Program have collaborated to provide comprehensive consensus guidelines on the care of children receiving CAR T cell therapy.

Leukaemias are the most common childhood cancers, and acute lymphoblastic leukaemia (ALL) is the most common form of childhood leukaemia 1 . In the century following its first description, leukaemia remained a near uniformly fatal disease. In 1948, astute observations among children with leukaemia, made by Sidney Farber 2 , led to therapeutic breakthroughs -ushering in an era of hope. Since then, paediatric oncology cooperative group trials and pioneering work by E. Donnall Thomas on haematopoietic stem cell transplantation (HSCT) among patients with leukaemia have led to dramatic improve ments in the cure rates of childhood ALL 3,4 . Despite improvements in overall remission rates (ORRs) for children with ALL, until now, the cure rates for chil dren with relapsed and/or refractory disease remain disappointingly low 5,6 . In the 1970s, E. Donnall Thomas considered a 15% salvage rate after disease relapse to be a promising result among a patient population with previously even more dismal outcomes 7 . Since that time, durable remission rates have increased, with other dramatic improvements in supportive care, such as antimicrobial prophylaxis and therapy, infection control, HLA tissue typing, immunosuppression, trans fusion support, and paediatric critical care management, resulting in improved patient survival.
In 2017, the FDA approved the first chimeric antigen receptor (CAR) T cell therapy, tisagenlecleucel, which has been associated with ORRs of almost 90% among children and young adults with B cell precursor ALL that is treatment refractory or in second or later relapse 8,9 . Notably, this novel genetically engineered, CD19targeted, autologous T cell immunotherapy was approved specifically for a paediatric and young adult indication. While CAR T cell immunotherapy is an exciting paradigm shift in anticancer therapy, this treatment modality is associated with unique toxicities, which can lead to very rapid and life threatening cardi orespiratory and/or neurological deterioration. Thus, this novel therapy requires the medical vigilance of a diverse multidisciplinary team -and the associated comprehensive clinical infrastructure -to ensure optimal outcomes. CAR T cells are generated through genetic modifica tion of the patient's own (autologous) T cells or those of an allogeneic donor. The isolated cells are activated and genetically modified via viral transduction or nonviral gene transfer 10,11 . Specifically, the modified CAR T cells express an engineered chimeric cell surface receptor comprising an extracellular antigen recognition domain, which is usually an antibody single chain variable frag ment (scFv), linked to at least one intracellular signalling domain -usually the CD3ζ chain of the T cell receptor plus one or more domains derived from co stimulatory receptors (such as CD28 or 41BB ligand receptor (41BB; also known as TNFRSF9)). The extracellular portion of the CAR enables recognition of a specific antigen (such as CD19), and the signalling domains stimulate T cell proliferation, cytolysis, and cytokine secretion to enable elimination of the target cell 12,13 . Autologous or allogeneic cells genetically engineered to express CARs targeting certain molecules commonly presented on the surface of cancer cells have been asso ciated with durable remissions among patients for whom no alternative therapies were effective [14][15][16][17][18][19][20][21][22][23][24][25][26][27] .
Cytokine release syndrome (CRS) and CAR T cell related encephalopathy syndrome (CRES) are well described unique toxicities associated with CAR T cells and some other immunotherapies [28][29][30][31][32][33] ; however, the pathophysiological mechanisms of both CRS and CRES remain poorly understood. Results from animal studies reported in May 2018 implicate recipient monocyte derived and/or macrophage derived IL1, IL6, and nitric oxide (and not CAR T cell derived cytokines) as the key determinants of the severity of CRS and CRES 34,35 . Indeed, CRS is a systemic inflammatory response caused by the CAR T cells and involving other immune cells that is typically characterized by fever, hypoxia, tachycardia, hypotension, and multi organ dysfunction 36 . CRES can occur concurrently with CRS, following its resolution, or without associated CRS and is characterized by encephalopathy, delirium, seizures, and, rarely, cerebral oedema 37 . Almost half of all patients who receive tisagenlecleucel require intensive monitor ing and critical care support, predominantly owing to these toxicities 9,38 . CRS and CRES are generally reversible but can be fatal. Paediatric specific management guide lines, comprehensive training of interdisciplinary staff, effective communication, and an appropriately phased infrastructure to ensure that adequate resources are available should facilitate the early diagnosis and appro priate management of paediatric patients who develop CRS and/or CRES after receiving CAR T cell therapy as a standard of care or according to a research protocoland thereby achieve optimal outcomes. Herein, we provide consensus guidelines for the use of CAR T cells in paediatric patients with cancer.

Guideline formulation
A panel of experts from the HSCT Subgroup of the Pediatric Acute Lung Injury and Sepsis Investigators (PALISI) Network, the CAR T Cell Therapy Associated Toxicity (CARTOX) Program at The University of Texas MD Anderson Cancer Center (Houston, TX, USA), and several other institutions with extensive interest and experience in CAR T cell therapy was con vened to develop these guidelines. The PALISI Network includes clinical researchers from >90 paediatric intensive care units across North America and beyond. The PALISI HSCT Subgroup is dedicated to improv ing the health and survival of children who require critical care during and following HSCT and cellular therapy through collaborative clinical and translational research. The guideline panel comprised a multidisci plinary and inter professional team, including physi cians with expertise in HSCT for paediatric and adult patients, translational immunotherapy researchers, paediatric intensivists, neurologists, paediatric nurses, advanced practice providers, pharmacists, clinical nurse specialists and nursing administrators, and health care administrators from a diverse group of medical centres. Using a modified Delphi method, the panel aimed to provide comprehensive guidelines for the safe administration of various CAR T cell therapies (standard of care and/or research protocol administra tion), the early recognition of CRS and/or CRES, and the management of these toxicities 39 . Levels of evidence and strength of recommendations were based on the classification schemes described by Shekelle and col leagues 40 (Supplementary Table 1). A summary of our key recommendations and the associated level of supporting evidence are provided in Table 1.

Patient selection and evaluation
Currently, CAR T cell therapies are used mainly among patients with relapsed and/or refractory haematologi cal cancers or other high risk malignancies. As new genetically modified cellular therapies are developed, differences in a range of variables related to the source cell type (T cells, natural killer (NK) cells, natural killer T cells, and cytokine induced killer cells, among others), the engineered product (for example, the co stimulatory domains and gene transfer technologies used), the man ufacturing process and reagents, the primary disease, and host factors might influence the individual tox icity profiles. Thus, providers of these treatments are encouraged to adhere to product information labels and guidance from risk evaluation and mitigation strategy (REMS) programmes as the products are approved by the FDA (Table 1).
Patient selection should be based upon the FDA approved indications and eligibility criteria used in pivotal studies but could potentially be tailored on the basis of emerging information relating to each new product (such as label updates). Patients should gen erally have an acceptable performance status according to the thresholds defined in treatment protocols and/or institutional guidelines, which might vary for different indications and products and depending on whether any deficiencies are secondary to disease specific manifes tations and thus likely to improve with primary disease response. Patients should be evaluated for uncontrolled infections, active graft versushost disease (GVHD), or recent donor lymphocyte infusion (DLI), with a thresh old of at least 6 weeks between DLI and CAR T infu sion 38 . Patients with uncontrolled infection and active grade II-IV acute or extensive chronic GVHD should be excluded. In patients with prior GVHD, the GVHD must have resolved and the patient should not have received systemic immunosuppression. Consideration should be given to the sites of active disease and specifically to whether immune activation, such as tumour flare 'pseu doprogression' , could compromise vital organ function (for example, of the airway or central nervous system (CNS)). Patients with active CNS pathology should be selected with caution, particularly for products associ ated with CRES [41][42][43] . When feasible, baseline evaluation by interdisciplinary team members, such as intensive care physicians and neurologists, could help to guide patient selection.
As new products are introduced, programmes should establish minimum eligibility criteria for patients to receive each agent. These criteria could potentially be adjusted over time on the basis of published experience gained in larger cohorts of patients. Patients identified as candidates for CAR T cell therapy should rapidly be referred for financial counselling to avoid delays in accessing care related to insurance pre authorizations. For convenience, we have provided an overview of our general recommendation relating to CAR T cell therapy eligibility and monitoring evaluations in the form of a checklist (box 1).
Informed consent and assent CAR T cell therapy is a potentially curative treatment but is associated with life threatening toxicities and can require long term local follow up evaluations and restrictions. Thus, detailed informed consent for ther apy should be obtained from the patient and/or their guardians; when appropriate, child assent should also be obtained 44 (Table 1). Incorporation of child life and psy chological services in assent discussions might be helpful.
In addition, capable patients should be asked to consider age appropriate advanced medical directives. Even before leukapheresis is performed, the patient and/or their guardians should be informed about the potential bene fits of CAR T cell therapy, as well as the potential toxicities and other risks associated with the procedure. Consent for CAR T cell manufacturing and therapy should encom pass leukapheresis, lymphodepletion therapy, CRS and CRES, and the potential need for bridging chemotherapy, intensive care support (including intubation and mechan ical ventilation, vasopressor and/or inotropic support, renal replacement therapy, and intracranial hypertension management after transfer to the intensive care unit), and anti IL6 therapy 38 . Patients should be aware that, even if the CAR T cell product is manufactured successfully, infusion of the product is contingent upon continued clinical eligibility. Patients should also be informed of the need to remain within 2 hours of the treatment facility for at least 4 weeks following infusion and of any other special precautions and limitations required during post infusion monitoring after treatment with specific products (for example, limitation of driving) 8 .
Leukapheresis for CAR T cell production Generation of autologous CAR T cells requires the col lection of CD3 + lymphocytes from the patient through leukapheresis. Absolute lymphocyte count (ALC) thresholds to proceed with leukapheresis can vary between different CAR T cell products; although an ALC of >100 cells/µl can be acceptable, >500 cells/µl (or an absolute CD3 + lymphocyte count of >150 cells/ µl) is generally preferred 45,46 . Patients should undergo pre collection testing to ensure that they are medically eligible to proceed with leukapheresis. Institutional or protocol specific guidelines should outline haemato logical and other clinical criteria to proceed with col lection. Paediatric patients can require a leukapheresis central venous catheter, rather than a peripheral venous cannula, for collection and should be haemodynami cally stable (that is, able to tolerate fluid shifts) and free

Level of evidence Grade
Providers are encouraged to adhere to product information labels and guidance from REMS programmes as they are approved by the FDA 8 IV D Patient selection should be based upon the indications approved by the FDA and the criteria used in pivotal studies and can be tailored on the basis of emerging information from each new product 8,9,49 IV D Consent should include descriptions of the risks and benefits associated with leukapheresis, lymphodepletion, CRS, CRES, bridging chemotherapy , intensive-care support (mechanical ventilation, dialysis, and inotropic support), and anti-IL-6 therapy 38

IIA B
When appropriate, child assent should also be obtained; age-appropriate advance directives should be considered. Incorporation of child life and psychological services in assent discussions can be helpful 44 IV D Paediatric patients can require a leukapheresis catheter for cell collection. Close monitoring for hypotension, hypocalcaemia, and catheter-related pain is imperative during paediatric leukapheresis, particularly among infants and younger children who might not verbalize symptoms 47,48 IIA B We recommend the selection of cyclophosphamide-fludarabine regimens for lymphodepletion, with exceptions considered in cases of haemorrhagic cystitis and/or resistance to a prior cyclophosphamide-based regimen 9,4664,78-81

IIA B
Given the potential for rapid clinical deterioration, if CAR T cell therapy is administered in an outpatient setting, a low threshold should be set for patient admission upon the development of a fever and/or signs or symptoms that are suggestive of CRS and/or CRES 38

IIA B
On the basis of the published experience for tisagenlecleucel in paediatric and young adult patients with CD19 + relapsed and/or refractory B cell acute lymphoblastic leukaemia, considering inpatient admission for a minimum of 3-7 days following infusion is reasonable 8,38 IIA B CRS grading should be performed as outlined in Table 2 at least once every 12 hours and more often if a change is noted and/or concerns exist 37

IIA B
Parent and/or caregiver concerns should be addressed because early signs or symptoms of CRS can be subtle and best recognized by those who know the child best 98

III C
CRS should be suspected if at least one of the following four symptoms or signs is present during the CRS-risk period within the first 2 weeks following CAR T cell infusion: fever ≥38 °C; hypotension (for patients aged 1-10 years: systolic blood pressure <(70 + (2 × age in years)) mmHg; for those aged >10 years: SBP <90 mmHg); a change from baseline and/or reduced requirements for chronic antihypertensive medications); hypoxia with an arterial oxygen saturation of <90% on room air ; or evidence of organ toxicity as determined by the most recent CTCAE grading system (version 5.0) 99 and paediatric considerations as outlined in

IIA B
CAR T cell-related HLH and/or MAS have been shown to resolve following administration of anti-IL-6 therapy and corticosteroids, although refractory cases can require further therapy , including consideration of systemic and/or intrathecal therapy on the basis of HLH-2004 management guidelines or use of the IL-1 receptor antagonist anakinra; further research is needed in this area 62,113,114

IIA C
We recommend that delirium screening using the CAPD tool 116 (or the CARTOX-10 grading system 37 for patients aged ≥12 years who have sufficient cognitive abilities) be performed at least twice per 24-hour period among admitted patients and at least daily among outpatients during the high-risk periods for CRES

IIA C
Consideration should be given to a prospective collaboration with intensive-care registries, such as VPS, which could allow accurate data entry of cell-therapy variables into the CIBMTR registry (by cell-therapy programmes) with concurrent entry of intensive-care variables into an appropriate registry by paediatric critical care teams IV D We strongly encourage consideration of QALYs for paediatric patients who might achieve long-term remission through this therapy and encourage all efforts to reduce the cost of care 136 haematopoietic stem cell transplantation immunosuppression should not be receiving immunosuppression before autologous leukapheresis • For autologous chimeric antigen receptor (CAR) T cell production, an absolute lymphocyte count of >100 cells/µl can be acceptable (>500 cells/µl is generally preferred), but this varies depending on manufacturer guidelines • early referral for financial counselling • obtain consents (and child assent when appropriate) • Infectious disease screening (for example, for hepatitis B virus (HBv) surface antigen, anti-HBv core antibodies, anti-hepatitis C virus (HCv) antibodies, and anti-HIv-1 and anti-HIv-2 antibodies, or using HIv, HBv, and HCv triple nucleic acid testing)

Before lymphodepletion
• Interval assessment with physical examination and screening for infection and organ toxicities • Primary disease evaluation, for example, through lumbar puncture, bone marrow aspiration and biopsy, and PeT-CT imaging • Pregnancy test, if indicated, and confirmation of no substantial interval changes in height and/or weight • Document baseline heart rate, blood pressure, mood, and cognitive and developmental status • establish central venous catheter or peripherally inserted central catheter access • Confirm consents (and assents, if applicable) for the cell therapy and anti-Il-6 therapy (if applicable) • Consider baseline assessment by neurologist and brain imaging (mRI or CT scan without contrast) and/or start levetiracetam for seizure prophylaxis a

Before CAR T cells infusion
• Confirm no uncontrolled infection; delay infusion if signs of uncontrolled infection are observed • Consider raising an electronic medical record flag for CAR T cell therapy • obtain baseline vital signs • ensure that oxygen, suction pump, and emergency medications (such as adrenaline) are readily available • At least two care providers with expertise in CAR T cell therapy should review the infusion order and CAR T cell product information • Pre-medications should not include routine steroid administration • Do not use a leukapheresis filter for infusion • Be aware of management guidelines for infusion-related complications • Infuse product according to manufacturer, protocol, and/or institutional guidelines • observe patient closely following infusion for infusion-related reactions • Decide on inpatient versus outpatient monitoring on the basis of the toxicity profile of the specific CAR T cell product, the patient's clinical status (assessed before and on the day of cell infusion) and psychosocial support network, and institutional outpatient infrastructure

Post-infusion monitoring b
• Daily history and physical examination • Daily complete blood count and blood product transfusion according to institutional guidelines for paediatric patients (without corticosteroid pre-medication) • Daily monitoring for disseminated intravascular coagulation (prothrombin time, partial thromboplastic time, fibrinogen, and D-dimer testing) • Daily monitoring for tumour-lysis syndrome (TlS) with basic metabolic panel, magnesium, phosphorus, uric acid, and lactate dehydrogenase measurements, and provide TlS prophylaxis if indicated • Daily profiling of serum levels of liver enzymes, albumin, and fractionated bilirubin • Daily serum C-reactive protein and ferritin monitoring for cytokine-release syndrome (CRS) and haemophagocytic lymphohistiocytosis • Infectious disease (viral, bacterial, fungal, and parasitic) prophylaxis, including for Pneumocystis jiroveci, as appropriate • Do not routinely administer corticosteroids (including as pre-medication) • Perform CRS and CAR T cell-related encephalopathy syndrome (CRES) grading every 12 hours or more frequently with clinical status change (with outpatient management, consider including caregiver) • ensure anti-Il-6 therapy is available for ordering by cell-therapy physician • organ toxicity monitoring and grading according to the Common Terminology Criteria for Adverse events version 5.0 (ref. 99 ) • Seizure prophylaxis with levetiracetam (10 mg/kg, up to a maximum of 500 mg per dose, every 12 hours for 30 days after CAR T cell infusion) a

Admission orders (for patients who are admitted owing to toxicities including CRS and/or CRES)
• Check vital signs every 4 hours (including pulse oximetry) • Strongly consider continuous cardiopulmonary monitoring and consider telemetry (monitoring for hypoxia and dysrhythmias) • Notify treating and/or attending physician of the following: -Temperature >38 °C, and order blood cultures, urinalysis and urine culture, and chest radiography, and consider use of broad-spectrum antibiotics (especially for patients who are neutropenic) -Heart rate or respiratory rate above or below age-specific normal range and/or baseline value (set range for sleeping and awake state) -Systolic blood pressure (SBP) <(70 + (2 × age in years)) mmHg for patients aged 1-10 years; SBP <90 mmHg for those aged >10 years; for infants aged <1 year, SBP above or below age-specific normal range and/or baseline value -oxygen saturation <92% on room air -Abnormal urine output according to age and weight (that is, none for 8 hours or <1 cc/kg per hour or >5 cc/kg per hour) -CRS or CReS of any grade or any change in mental status (such as irritability or tremors) -upward trends in serum creatinine levels or detriments in liver function test results a Recommended for patients treated with immune effector cell therapies known or suspected to cause CReS or first-in-human products and for patients with a predisposition to seizures. b Post-infusion monitoring should continue until the patient has completed a high-risk CRS-CReS observation period (defined on the basis of data from the pivotal study of the agent of choice and emerging experience with similar products in similar patient populations).
volume 16 | JANuARY 2019 | 49 NATuRe RevIeWS | CliniCAl OnCOlOGy symptoms [50][51][52][53][54] (Table 1). Packed red blood cells (irradi ated) and/or albumin can be used to prime the collec tion in children weighing <30 kg. The targeted cell dose for leukapheresis can vary depending on the specific product and manufacturing process. Once an adequate quantity of cells has been collected, the cells are sent to the laboratory for CAR T cell manufacturing -a pro cess that typically takes 2-4 weeks 8 . Cell therapy pro duction, storage, transportation, and shipping should occur in compliance with the most current standards as defined by the Foundation of Accreditation of Cellular Therapy (FACT) 55 .

Allogeneic CAR T cells
Limitations of autologous CAR T cell therapy include the time required after leukapheresis to manufac ture the cell product (especially among patients with advanced stage disease and a very narrow therapeutic window), the cost of manufacturing a patient specific product, and the fact that adequate leukapheresis might not always be possible among heavily pretreated patients. Allogeneic CAR cells would presumably offer an 'off theshelf ' , third party approach to CAR based cell therapy [56][57][58][59] . CAR transduced cord blood derived NK cells have been reported to have potent anticancer effects in preclinical studies 60 . This allogeneic NK cell approach to CAR cell therapy might also be associated with a reduced risk of GVHD compared with the use of allogeneic T cells. Moreover, the NK cell product used included a 'suicide' gene -inducible caspase 9 -that could be pharmacologically activated to eliminate the transduced cells, thus providing an additional safety mechanism 60 . A clinical trial to test the safety and effi cacy of this product is currently underway at the MD Anderson Cancer Center (NCT03056339). In other allogeneic CAR expressing cell therapy approaches, functional co expression of RQR8, a construct com bining epitopes from CD34 and CD20, renders the CAR cells sensitive to the monoclonal anti CD20 anti body rituximab, as a safety feature 56,61 . Safety mecha nisms designed to eliminate transduced cells have also been added to autologous CAR expressing cells, such as anti CD19 CAR T cells also expressing inactive, truncated EGFR, which enables the cells to be targeted through administration of the anti EGFR antibody cetuximab 49 . The recommendations provided herein should be applicable to both autologous and allogeneic CAR based cell therapies.
Bridging chemotherapy Some patients with high risk advanced stage malig nancies require bridging chemotherapy in the period immediately following leukapheresis. The goal of bridging chemotherapy is to maintain disease control and prevent progression (in addition to potentially decreasing tumour burden, which might reduce the risk of severe CRS 20,29,62 ) while the autologous CAR T cells are manufactured rather than to act as a primary treatment of the disease 20,62 . During this 2-4week period, patients should be monitored for tumour lysis syndrome (TLS) and should receive antimicro bial prophylaxis with routine infection precautions 63 .
Indeed, the bridging chemotherapy regimen should be selected carefully during this critical period in order to minimize the risk of toxicities, which might disqualify the patient from proceeding to lymphodepletion and/ or CAR T cell infusion 45 . Several bridging chemother apy regimens are commonly used in paediatric patients (box 2); however, insufficient data are currently avail able to recommend an optimal regimen. This choice might be influenced by certain study protocol guide lines, characteristics of the particular CAR T cell prod uct, and/or patient specific clinical variables (such as prior response to particular chemotherapeutic agents and baseline organ function).

Preparative lymphodepletion treatment
Several lymphodepletion regimens are commonly used in the treatment of paediatric patients 9,22,64 (box 3). The T cell pool is subject to homeostatic regulation; there fore, a lymphopenic environment might be favourable for adoptive T cell transfer owing to reduced compe tition for homeostatic factors. Indeed, the removal of 'cytokine sinks' increases the availability of cytokines that promote lymphocyte proliferation and survival, such as IL7 and IL15, and can be a contributing fac tor to the effectiveness of tumour specific T cells [65][66][67] . Furthermore, the depletion of immunosuppressive CD4 + CD25 + regulatory T cells [68][69][70][71][72][73][74][75] has been proposed as a key mechanism by which lymphodepletion augments adoptive T cell therapy 76,77 . Accordingly, lymphodeple tion has also been shown to improve the expansion and persistence of adoptive CAR T cells and to enhance their anticancer efficacy, resulting in increased ORRs 64,66 . Of note, the majority of paediatric patients enrolled in the trials of tisagenlecleucel completed lymphodepletion before CAR T cell infusion 9,38 . We recommend that all patients undergo lymphodepletion if possible (omis sion of lymphodepletion might need to be considered in patients with lymphopenia).
Cyclophosphamide based lymphodepletion regimens are commonly used before CAR T cell infusion, and the addition of fludarabine to lymphodepletion chemother apy has been associated with improved CAR T cell expan sion and persistence and prolongation of disease free survival in patients with ALL 64,78 . Alternative lympho depletion regimens (box 3) can be considered in patients with haemorrhagic cystitis and/or resistance to a prior cyclophosphamide based regimen 9,46,64,79-81 (Table 1).
To proceed with lymphodepletion and CAR T cell infusion, patients should not have uncontrolled infec tion because hyperinflammatory states pre infusion have been associated with increased risks of morbidity and mortality 49,82-84 . Thus, an interval assessment should be performed on the day of initiation of the lympho depletion regimen to identify any new complications. This evaluation usually includes screening for signs and symptoms of active infection and/or new organ toxicity, as well as exclusion of pregnancy if indicated (box 1). Anticipatory guidance regarding the adverse effects of the specific drugs and treatment modalities used should be reviewed. Haemodynamic and laboratory monitor ing and hydration should be tailored on the basis of the selected lymphodepletion regimen.

Cell infusion
Patients should not have evidence of uncontrolled infec tion and/or other contraindications before CAR T cell infusion. Contraindications include active or latent HBV infection, active HCV or HIV infection, severe acute or chronic extensive GVHD, and pregnancy 8,37,38,49,[82][83][84][85] . If any of these features are present, CAR T cell infusion should be delayed to avoid potentially severe immune activation and associated sequelae.
At the time of infusion, oxygen, suction, and emergency medications, including adrenaline, should be readily available (box 1). The patient and/or their caregiver should be instructed to report symptoms, such as shortness of breath, rash, chills, chest pain, and back pain. Infusion should occur through the largest patent lumen without a filter and generally without an infusion pump. Pre medication with drugs including acetaminophen and diphenhydramine should be administered 30-60 mins before CAR T cell infusion 8 in order to prevent infusion reactions related to cryopreservants, such as dimethyl sulfox ide. Corticosteroids should not be routinely used for pre medication, as these agents are lymphocytotoxic and, thus, their administration early in the treatment course (before CAR T cell expansion in vivo) could affect therapeutic outcomes 20 . With the use of CAR T cell products approved by the FDA, we recommend following the FDA approved package labelling; other wise, protocol specific guidelines should be followed. Vital signs and urine output should be monitored closely after the time of infusion.
Infusion of cellular therapy products is generally safe, although serious adverse infusion reactions can occur. Characteristic adverse reactions include nau sea, vomiting, abdominal pain, chills, fever, and, rarely, severe respiratory depression, neurotoxicity, and cardiac arrhythmias [86][87][88][89][90][91][92][93] . General management principles for infusion reactions associated with adoptive cell therapy include consideration of slowing or halting the infusion, activation of emergency precautions, and confirmation of product details for accuracy. If symptoms resolve uneventfully without medical intervention, a serious infusion related event is unlikely to have occurred. If an infusion related event is considered likely, a transfusion reaction laboratory evaluation and appropriate support ive care according to institutional guidelines should be initiated 94 . Bacterial infusion reactions occur from infu sion of contaminated products, typically with Gram negative organisms; therefore, prompt treatment with appropriate antibiotics and supportive care for fever, severe hypotension, and other unexpected signs and symptoms is important to preclude clinical deterioration and potentially death.

Inpatient and outpatient management
Early recognition of toxicities of CAR T cell therapy, particularly CRS and/or CRES, in paediatric patients requires detection of variations from baseline in char acteristics including heart rate, blood pressure, tem perature, and irritability (box 1). Both the efficacy and toxicity profiles of CAR T cell therapy might vary depending on the specific product administered and individual patient characteristics. Thus, the decision on inpatient versus outpatient management of patients treated with CAR T cell therapy should involve consid eration of the toxicity profile of the product used, the clinical status of the patient (including risk of TLS), and the ability of the institution to deliver prompt and comprehensive outpatient management, as well as the ability of the patient to access such care. Given the potential for rapid clinical deterioration, if CAR T cell therapy is administered in an outpatient setting, a low threshold should be used for patient admission upon development of fever and/or other signs or symptoms that are suggestive of CRS and/or CRES 38 (Table 1). The presence of a reliable, consistent, and well informed caregiver is essential to facilitate outpatient adminis tration of CAR T cell therapy. In instances in which patients are not admitted for CAR T cell infusion and CRS-CRES monitoring, adequate outpatient facilities are needed, with extended outpatient hours (as defined by the institutions), prompt access to emergency and critical care, and trained staff who are knowledgea ble of CAR T cell toxicity and are capable of prompt patient evaluation and management available at all times. This requires adequate outpatient space with a design appropriate to the protection of patients who are immunocompromised while they are being triaged. Moreover, rapid access to a cellular therapy physician, and pharmacy, laboratory, and transfusion medicine services should be guaranteed.
As products transition from research study proto cols to standard ofcare administration, institutional guidelines should consider whether most patients require inpatient hospitalization during the pivotal tri als and/or what infrastructure is needed for outpatient administration. For example, during the international phase II ELIANA study of tisagenlecleucel in paediat ric and young adult patients with CD19 + relapsed and/ or refractory B cell ALL 38 , 76% of patients underwent cell infusions in the inpatient setting. CRS occurred in 77% of patients, with a median time to onset of 3 days (range 1-22 days) 38 . Almost half of all patients required intensive care support, with a median stay of 7 days (range 1-34 days); intensive care support included the use of high dose vasopressors, oxygen supplementation, mechanical ventilation, and/or dialysis 38 . Neurological events occurred in 40% of patients, with the majority of cases occurring concurrently with, or soon after res olution of, CRS 38 . Two deaths occurred within 30 days of CAR T cell infusion (one patient died of cerebral haemorrhage and another died of progressive leukae mia) 38 . On the basis of the published experience with this product, considering inpatient admission for a minimum of 3-7 days following infusion is reasonable, especially as this treatment is increasingly being offered as a standard of care 8,38 (Table 1). Nevertheless, the length of inpatient hospitalization and/or need for daily out patient assessments might vary on the basis of the risks of CRS and CRES, the clinical and developmental status of the patient, and the social support systems available to the patient at home. The risk of developing CRS and/or CRES probably depends on the patient, the source of immune effector used to manufacture the CAR T cell volume 16 | JANuARY 2019 | 51 NATuRe RevIeWS | CliniCAl OnCOlOGy therapy, the specific CAR T cell product, and/or the associated lymphodepletion strategy.
Admission orders and handoff communication should include information regarding the patient's base line heart rate, blood pressure, and mood, cognition, and developmental status. In addition to the aforementioned screening (box 1), continuous cardiac -and, if feasible, telemetry -monitoring should be strongly considered, beginning on the day of CAR T cell infusion (to ascer tain baseline physiology) and continuing for a minimum number of days following infusion (based upon peak incidence of CRS and/or CRES) or until any emergent CRS resolves. Sinus tachycardia can be an early present ing sign of CRS 29,95 , and recognition of this clinical find ing requires high vigilance and awareness of the child's baseline heart rates as well as age specific normal values. To enable prompt medical intervention for these toxic ities, we recommend that patients have central venous access or a double or triple lumen peripherally inserted central catheter 37 . For patients with a history of, or pre disposition to, seizures or those with a high risk of CNS pseudoprogression (such as those with CNS disease, chloromas and/or leptomeningeal enhancements, or a prior history of seizures), a baseline neurology evalu ation, electroencephalography (EEG), and/or baseline MRI of the brain and spinal cord and/or anti seizure prophylaxis should be considered 96 .
We recommend that all patients have frequent physical examinations and laboratory monitoring with complete blood counts, comprehensive metabolic panels, coagulation testing, and serum ferritin and C reactive protein measurements during the post infusion period associated with a high risk of CRS (as defined in pivotal studies and by emerging data for each prod uct) 37,82,84 (box 1). Regular -daily, if possible -moni toring for TLS is also important, and TLS prophylaxis is recommended for patients with a high disease bur den 16,29,37,82,84,96 . In addition, infectious disease prophy laxis against viral, bacterial, and/or fungal pathogens should be prescribed as appropriate 63,82 . Patients should have adequate hydration with monitoring for acute fluid overload (daily weights and fluid intake-output recording). If needed, transfusions (irradiated blood products) should be ordered according to institutional guidelines for paediatric patients (without routine corticosteroid pre medication). The risk of bleeding can be exacerbated by hypofibrinogenaemia and/or thrombocytopenia, especially in patients receiving anti coagulation therapy (including through continuous venovenous haemofiltration) 38,82,97 . Conservative man agement of bleeding or hypofibrinogenaemia and/ or thrombocytopenia, with the use of cryoprecipitate or fresh frozen plasma, as needed, is recommended to avoid lethal haemorrhage 82 . Administration of growth factors (G CSF) should be considered for patients with neutropenic fever 8,82 . Conditional orders for fever and/ or neutropenia and suspected sepsis can enable rapid intervention, when needed, by the nursing unit, out patient triage personnel, and pharmacy. We recommend creation of a flag or banner in the electronic health system to alert care providers of patients who are CAR T cell recipients. Indeed, recipients of CAR T cell prod ucts should be made easily identifiable; in particular, the strong contraindication of these patients to steroids must be highly visible to avoid routine or accidental administration of these drugs (for example, during blood product and other pre medication orders). An 'as needed' order, which requires a real time electronic approval by an authorized care provider, is recom mended in order to ensure rapid access to the correct dose of anti IL6 therapy (for example, with the anti IL6 receptor antibody tocilizumab) when required for the treatment of CRS and/or CRES. General recommendations regarding the timing of treatment discontinuation • TKIs and hydroxyurea must be stopped ≥72 hours before chimeric antigen receptor (CAR) T cell infusion • The following drugs must be stopped ≥1 week before CAR T cell infusion: vincristine, 6-mercaptopurine, 6-thioguanine, methotrexate ≤25 mg/m 2 , cytarabine ≤100 mg/m 2 , and asparaginase (non-PeGylated) • The following drugs must be stopped ≥2 weeks before CAR T cell infusion: clofarabine, cytarabine >100 mg/m 2 , anthracyclines, cyclophosphamide, and methotrexate ≥25 mg/m 2 • PeGylated asparaginase must be stopped ≥4 weeks before CAR T cell infusion • Central nervous system prophylaxis treatment must be stopped ≥1 week before CAR T cell infusion CRS monitoring, grading, and management CRS reflects a systemic inflammatory response driven by rapid and excessive secretion of cytokines (a so called cytokine storm) that is associated with a spectrum of symptoms ranging from fever to multi organ dysfunc tion 34,35 . As mentioned previously, 77% of the paediatric and young adult population of the ELIANA trial devel oped CRS after treatment with tisagenlecleucel, with almost half experiencing severe symptoms requiring intensive care support 38 . In addition, 40% of patients developed CRES (grade 3 in 13%) 38 . Other notable adverse events included infection (in 43% of patients), cytopenias (grade ≥3 neutropenia and thrombocytope nia not resolved by day 28 in 35% and 7%, respectively), and TLS (in 4%) 38 . Patients with a high risk of develop ing severe CRS include those with early symptom onset (typically within 3 days of CAR T cell infusion), a high disease burden, and/or pre existing comorbidities 9,20,22 . Early detection of CRS and/or CRES in paediatric patients can be challenging; however, early diagnosis of CRS and its prompt management can mitigate the risks of life threatening sequelae. We recommend that patients who show signs of CRS and/or CRES be admit ted for observation. The grading and management of CRS have been largely based on criteria originally out lined by Lee and colleagues 29 . In the January 2018 issue of this journal, Neelapu and other members of the MD Anderson Cancer Center CARTOX Program published updated and more detailed guidelines for the manage ment of adult patients with CRS 37 . Herein, these recom mendations for adult patients have been modified with input from the PALISI Network HSCT Subgroup to provide paediatric specific guidelines.
CRS grading according to the criteria outlined in Table 2 should be performed at least once every 12 hours and more often if a change in the patient's clinical status or reasons for concern are noted ( Table 1). Parent and/or caregiver concerns should be thoroughly investigated because early signs or symptoms of CRS can be subtle and thus might be best recognized by those who know the child very well 98 (Table 1). For example, detection of CRS involving the gastrointestinal system often requires recognition of changes in the child's food intake and/or the frequency and consistency of bowel movements, as well as expression of nausea. We recommend that CRS grading performed primarily by physicians, advanced practice providers, and bedside nurses be reviewed by interdisciplinary team members immediately after each assessment and, when possible, should include participation of the patient and/or parent or caregiver at the bedside. Nurses should ideally perform assess ments mid shift and jointly with incoming nurses dur ing handoff at the end of their shifts. In the outpatient setting, properly trained caregivers could potentially perform CRS-CRES assessment in lieu of twice daily clinical assessments by a health care professional, but this approach has not been validated.
CRS should be suspected if at least one of the fol lowing four symptoms or signs is present during the CRS risk period after CAR T cell infusion: fever ≥38 °C; hypotension (defined as a systolic blood pressure (SBP) <(70 + (2 × age in years)) mmHg for patients aged 1-10 years or <90 mmHg for those aged >10 years, a change in SBP from baseline values, and/or a reduced requirement for chronic anti hypertensive medications); hypoxia with an arterial oxygen saturation of <90% on room air; and/or evidence of organ toxicity as determined using the most recent Common Terminology Criteria for Adverse Events version 5.0 (CTCAE v5.0) grading system 99 -bearing in mind specific considerations for paediatric patients 29,37,82 (Tables 1,2). Frequent monitor ing of complete blood count, coagulation, and chem istry profiles, including serum levels of liver enzymes, C reactive protein, ferritin, and lactate dehydrogenase, might be useful for early detection of CRS.
While the criteria above provide general definitions of hypotension in children, it is important that the base line blood pressure range of each child be defined before CAR T cell infusion so that relative hypotension from an elevated baseline is not missed. Reduced requirements for chronic anti hypertensive medications can also indi cate relative hypotension. Furthermore, because some symptoms can be caused by other concurrent conditions (for example, sinus tachycardia can have causes that range from crying to sepsis), care providers must use their clinical judgement to determine CRS attribution. Patients who develop fever with a temperature >38 °C, for example, should be assessed for infection using blood cultures and chest radiography; additional tests, such as viral PCR, respiratory viral screening, urine cultures, and CT of the chest, should be obtained as clinically indicated. Empiric antibiotic treatment should be initi ated, and filgrastim should be considered if the patient is Cyclophosphamide monotherapy • 2-4 g/m 2 (single dose) 64 • 1 g/m 2 (single dose) 9 • 300 mg/m 2 every 12 hours for 3 days 9

Clofarabine
• 30 mg/m 2 daily for 5 days 9 volume 16 | JANuARY 2019 | 53 NATuRe RevIeWS | CliniCAl OnCOlOGy neutropenic and septic. Importantly, the patient's orders should ensure triggering of such an escalation in care for a temperature >38 °C -some nursing units might be accustomed to higher temperature thresholds for inter vention. This practice might require re education of outpatient, emergency room, and triage staff that might not routinely care for patients treated with CAR T cells.
Careful vigilance for early recognition of haemody namic shock in the child is crucial. Symptoms such as malaise, lethargy, weakness, oliguria, irritability, and reduced appetite are not always self reported by younger children. Among infants, assessment of diapers will be critical to assess urine output and detect diarrhoea. We concur with recommendations that any patient requir ing a rapid increase in the dose of vasopressors or exhib iting evidence of end organ hypoperfusion should be treated intensively for grade 3 CRS, even if the vasopressor therapy required is 'low dose' according to the definition of Lee and colleagues 29,37 . For children with hypoten sion owing to CRS, an initial normal saline fluid bolus (10-20 ml/kg; maximum 1,000 ml) should be adminis tered; if no improvement is observed, anti IL6 therapy should be initiated (Table 2). After administration of anti IL6 therapy, the decision to repeat additional fluid boluses versus starting vasopressors should involve con sideration of the cardiac and fluid status of the child. For example, the administration of additional fluid boluses should be avoided in patients with underlying cardiac dysfunction and/or signs and symptoms of volume over load (such as pulmonary oedema). Additionally, early use of colloid solutions might be indicated because patients with CRS could potentially develop capillary leak and hypoalbuminaemia more rapidly than patients with sep sis, and the administration of additional fluid boluses might compromise pulmonary function by causing pul monary oedema. Care should be taken not to trigger acute fluid overload, cardiogenic shock, and/or respiratory compromise. Consideration should be given to adrenal insufficiency in decisions of the initial choice of corti costeroid intervention, if needed (Table 2); patients with vasopressor resistant hypotension attributed to adrenal insufficiency might respond to stress dose hydrocorti sone only and, thus, high doses of other lymphocytotoxic corticosteroids (dexamethasone or methylprednisolone) can be avoided. Transfer to an intensive care unit should be considered early in this process (Table 1).
Grading criteria for CRS related hypoxia have pre dominantly been based upon fraction of inspired oxygen (FiO 2 ) requirements and the need for mechanical venti lation 29,37 . In recognition of the differences in acute res piratory distress syndrome (ARDS) between adult and paediatric patients, a 2015 publication by the Pediatric Acute Lung Injury Consensus Conference (PALICC) group provided paediatric specific definitions for pae diatric ARDS (P ARDS) 100 ; these criteria have since been applied among paediatric patients who have under gone HSCT 101,102 . We recommend application of the PALICC at risk P ARDS criteria for the CRS grading of hypoxia 100 (Table 1). Accordingly, the grading of CRS related hypoxia should be based on the use of high flow oxygen and other non invasive forms of mechanical ventilation, whereby flow rates and FiO 2 are indicators of severity. For example, paediatric patients who require supplemental oxygen exceeding an FiO 2 requirement of 40% or those who are receiving non invasive mechanical ventilation should be classified as having grade 3 CRS and be managed accordingly (Table 2). CTCAE v5.0 grading of organ toxicity provides an objective assessment tool, although high vigilance is important for prompt recognition of CRS among chil dren. Sinus tachycardia (according to age dependent definitions) is often the earliest sign of CRS 103 (Table 1). Expected heart rate ranges should also incorporate the child's baseline measurements at patient admission before CAR T cell infusion 104 . Acute kidney injury in children can be graded according to CTCAE v5.0 criteria using the Pediatric Risk, Injury, Failure, Loss, End Stage Renal Disease (pRIFLE) and Kidney Disease: Improving Global Outcomes (KDIGO) definitions of oliguria and anuria 105,106 (Tables 1,2).
In one study 20 , the use of high dose steroids to treat CRS was associated with suppression of CAR T cell expansion and unfavourable patient outcomes. However, no adequately powered randomized studies investigating whether the administration of anti IL6 therapy and/or corticosteroids reduces the efficacy of CAR T cell ther apy have been reported to date. Notably, the use of anti IL6 therapy and/or corticosteroids for the management of CRS (usually following CAR T cell expansion) has not negatively affected disease free survival outcomes in larger cohorts 107,108 . Indeed, the anti IL6 receptor antibody tocilizumab is currently approved by the FDA for management of CRS 109 ; patients weighing <30 kg are treated at a dose of 12 mg/kg, and those weighing ≥30 kg are treated at a dose of 8 mg/kg 110 (Tables 1,2). Clinical responses following the administration of tocilizumab are often observed within 4 hours (ref. 33 ). Patients who do not respond to the first dose of tocilizumab might be less likely to respond to subsequent repeat admin istration of this agent; for these patients, consideration should be given to administration of corticosteroidsweighing the risks and benefits of steroid use for CRS and the uncertain risk of suppressing CAR T cell expan sion -and/or alternative agents, such as the anti IL6 monoclonal antibody siltuximab 37,38,111 . Corticosteroids can also be administered concurrently with repeat doses of tocilizumab (Table 2). When corticosteroids are used, the taper should be rapid and individualized according to the patient's response. The initial choice of corticosteroid (hydrocortisone, dexamethasone, or methylprednisolone) will vary depending on insti tutional preference, protocol specific guidelines, or product label specifications.

CAR T cell therapy-related HLH
Haemophagocytic lymphohistiocytosis (HLH) is a rare syndrome with severe clinical sequelae that result from a dysregulated, hyperinflammatory immune response 112 and can present in a primary (inherited) or a secondary form 113 . Secondary HLH is thought to occur in the con text of an underlying immunological condition and, in the setting of autoimmune and inflammatory disorders, is often referred to as macrophage activation syndrome (MAS). The diagnosis of HLH is made on the basis of the presence of mutations associated with primary HLH (such as mutations in PRF1, UNC13D, or STX11) and/or clinical and laboratory criteria, such as fever, cytopenias, hypertriglyceridaemia, hypofibrinogenaemia, elevated serum levels of ferritin and liver enzymes, haemophago cytosis, low or absent NK cell activity, and/or elevated soluble IL2 receptor levels 113 . Differentiation of primary HLH from MAS can be difficult, and distinction of sec ondary CAR T cell related HLH-MAS from CRS-CRES can be even more challenging owing to the overlapping symptoms associated with these conditions 16,29,37,112 . Future studies to define the role of genetic testing and functional analyses of NK cells might help identify patients who are at a disproportionately higher risk of this complication of CAR T cell therapy 30 . Nevertheless, diagnostic criteria for this rare toxicity have been pro vided in the guidelines for the management of adult patients treated with CAR T cell therapy published by  Neelapu et al. 37 in this journal. Children can also be diagnosed with CAR T cell related HLH-MAS if they have a peak serum ferritin level >10,000 ng/ml during the CRS risk period and develop any two of the follow ing: grade ≥3 organ toxicities involving the liver, kidney, or lung; or haemophagocytosis in the bone marrow or other organs 37 . Patients who develop CAR T cell related HLH-MAS can be treated simultaneously with anti IL6 therapy and corticosteroids ( fig. 1); responses to anti IL6 therapy alone might not be as common as in patients with CRS alone 33 . In addition, although HLH-MAS occurring after treatment with CAR T cells and other T cell engaging therapies has typically been shown to resolve following administration of anti IL6 therapy and/or corticosteroids 30,62,96 , refractory cases can require additional therapy, including consideration of systemic and/or intrathecal therapy according to the HLH2004 management guidelines 113 or use of the IL1 receptor antagonist anakinra 62,114 (Table 1). Further research is needed, however, to optimize the diagnosis and treatment of CAR T cell related HLH-MAS.

CRES
Neurological symptoms associated with CAR T cell therapy, referred to as CRES, commonly present as a toxic encephalopathy with delirium, seizures, and/or cerebral oedema 29,37,62,82,115 . The earliest signs and symp toms of CRES can be subtle among children, and, thus, a paediatric skill set is required in order to determine the child's baseline level of cognitive performance. Among adult patients, early presenting symptoms of CRES include inattention and impaired expression affecting language and handwriting 37 . Grading of CRES accord ing to CTCAE v5.0 (ref. 99 ) or previously published CRES algorithms for adult patients 29,37 is not optimal among infants and younger children. The Cornell Assessment of Pediatric Delirium (CAPD) 116 is a validated screening tool (Supplementary Table 2) for recognition of delirium among children and adolescents (from birth to 21 years old); the sensitivity and specificity of this tool are highest in patients aged <12 years. Use of CAPD with appropri ate developmental anchor points 117 enables developmen tally appropriate delirium screening by nurses and other members of the health care team at the bedside and is an important tool in the overall grading of CRES, as out lined in Table 3; a CAPD score >8 is indicative of delir ium 116 . Alternatively, neurological assessment scoring, as previously described with the CARTOX 10point assess ment scale (CARTOX10) grading system by Neelapu et al. 37 , can be used for patients aged ≥12 years who have cognitive abilities that are appropriate for these assess ments (Table 3). We recommend that delirium screening with CAPD or other neurological assessments be per formed at least twice per day among admitted patients and at least daily among outpatients during the high risk period for CRES (the 4 weeks after CAR T cell infusion) ( Table 1). The first neurological assessment from a nurs ing provider should occur at the end of their shift and be conducted concurrently with the incoming nurse. The severity of CRES can be labile; thus, assessments should be performed more frequently if a change from prior scores occurs and/or if a caregiver raises concerns.
Indeed, the trend in CAPD scores within an individual patient is important; increasing scores can be used as a marker of CRES severity (Table 3).
The onset of CRES can be biphasic, occurring con currently with CRS and/or after CRS has resolved, and the precise pathophysiology remains unclearalthough evidence implicates a combination of endothe lial activation in the CNS, elevated cytokine levels in the cerebrospinal fluid (CSF), and cerebral T cell infiltra tion 118,119 . The use of anti IL6 therapy seems to be more effective for the management of CRES that occurs con currently with CRS. Patients who develop CRES can also benefit from early corticosteroid administration 29,37,38 ( Table 3). CRES is generally reversible; however, cer ebral oedema and death have been reported 37,64,120-122 . No adequately powered randomized studies to identify patients who are at disproportionately higher risk of CRES have been reported to date. We recommend that recipients of CAR T cell therapy with CNS disease or a history of seizures receive anti seizure prophylaxis with levetiracetam (10 mg/kg, up to a maximum of 500 mg per dose) every 12 hours for 30 days following infusion (or through the CRES risk period as described during pivotal studies and with subsequent emerging data) 9,27,37 (box 1). Levetiracetam is generally well tolerated, with a minimal risk of adverse drug interactions, although dose adjustments might be necessary in the setting of renal dysfunction and are not thought to affect cytokine levels 37,123,124 . Neurology consultation should be consid ered if the patient develops grade 1 CRES (Table 3) and/ or for specialized screening for papilloedema. Patients should be closely monitored for signs and symptoms of cerebral oedema. Status epilepticus can be managed according to institutional guidelines (our recommended management approaches are provided in box 4). In general, first line anti seizure medications with unfa vourable cardiotoxicity profiles (such as lacosamide and phenytoin) should be avoided when possible. Increased intracranial pressure (CSF opening pressure ≥20 mmHg or clinical signs of increased intracranial pressure) will require intensive care management and osmotherapy (a management algorithm that can be tailored accord ing to institutional guidelines is proposed in box 5). A neurosurgery consultation should be considered, and brain scans can help guide patient management. Routine chemistry panels should be monitored more frequently (every 6-8 hours) and medications adjusted accordingly to prevent rebound cerebral oedema, renal failure, hypovolemia and/or hypotension, and electro lyte abnormalities. General grading and management guidelines for CRES are outlined in Table 3 (these can be tailored according to product specific approved label instructions and/or study protocols). CRES can occur as a later complication and, in some patients, after discharge from hospital; therefore, caregivers and/or the patient should be given appropriate antici patory guidance and appropriate education before being discharged from hospital. Moreover, patients should have a caregiver available who can observe for signs of CRES and seek prompt intervention for at least 4 weeks (or through the CRES risk period) after CAR T cell infusion.

Long-term follow-up assessment
Careful long term follow up assessment of patients receiving CAR T cell therapy is important. Management of on target, off tumour effects should be well coordi nated between treatment and referring centres if the patient returns to local providers following therapy.
For example, B cell aplasia and hypogammaglobuli naemia or agammaglobulinaemia are commonly seen in patients treated with anti CD19 CAR T cells. These adverse effects require long term replacement with intravenous immunoglobulins (IVIGs). We recom mend intervention to maintain serum immunoglobulin levels above 400 µg/l with IVIGs as well as consideration of IVIG administration to provide specific immunity during active infection, irrespective of immuno globulin levels 63,125 . B cell aplasia has been associated with progressive multifocal leukoencephalopathy (PML) 126,127 ; thus, patients should be closely monitored for neurological signs and symptoms that are suggestive of PML (neuropsychological deficits, progressive demen tia, apraxia, or visual and motor deficits) 126 until the resolution of B cell aplasia.
Data from studies of the effectiveness and safety of immunization with inactive or live vaccines in patients treated with adoptive T cell therapies have not been reported to date. We recommend careful assessment of immune reconstitution after lymphodepletion and CAR T cell infusion; the findings should guide decisions regarding antimicrobial prophylaxis and re vaccination 63,128 .
Patients treated with CAR T cell therapy are at risk of disease relapse and/or the development of secondary neoplasms. In addition, the use of a replication competent viral vector during CAR manufacturing could pose a theoretical risk to patients and/or their close contacts 129 . Long term clinical monitoring is important for detec tion of these complications and is mandated by the FDA for all such gene therapies 130,131 .
Local and national registries to capture the out comes, acute complications, and late effects of CAR T cell therapies might enable the establishment of qual ity benchmarks, facilitate retrospective research, recog nize potential delayed toxicities, and ultimately improve future care. As the Center for International Blood and Marrow Transplant Research (CIBMTR) develops a registry for patients receiving CAR T cell therapy, we recommend that consideration be given to the reporting of variables that are directly retrievable from electronic medical records to ensure accuracy and min imize the infrastructural burden required for compre hensive reporting. Given that toxicity grading systems are likely to evolve over time, entry of primary varia bles seems more useful in the long term. Furthermore, many of these patients require intensive care support, and therefore prospective collaborations with intensive care registries, such as Virtual paediatric intensive care unit (PICU) Systems (VPS) 132 , should be considered (Table 1). This approach could enable accurate data entry of cell therapy variables into the CIBMTR regis try by cell therapy programmes, with concurrent entry of data on intensive care variables into an appropriate registry by paediatric critical care teams.

CAR T cell therapy as a bridge to HSCT
In ELIANA 8 , the largest study of CAR T cell therapy involving paediatric patients performed to date, 83% of the infused patients (n = 63) achieved minimal resid ual disease (MRD)negative complete remission (CR) or CR with incomplete haematological recovery (CRi). After a median follow up duration of 4.8 months from response, the median CR and/or CRi duration was not reached (range 1.2 months to >14.1 months). The results of prediction based modelling suggest that more than half of the patients who received tisagenlecleucel on the ELIANA trial will be alive at 5 years after treat ment 133 . The actual allogeneic HSCT (allo HSCT) rate among those who achieved a CR or CRi was 12% in the ELIANA trial 8 .
In another paediatric study 46 , CD4 + and CD8 + T cells transfected with an anti CD19 CAR construct containing a 41BB co stimulatory domain using a lentiviral vector were administered to 45 children and young adults with pre B cell ALL; 93% of the patients achieved MRD negative remission by day 21. However, the estimated 12month event free survival was 50.8%, with the majority of these patients unfortunately expe riencing disease relapse 46 . The persistence of functional anti CD19 CAR T cells was assessed by measuring the duration of B cell aplasia using flow cytometry; the median duration of B cell aplasia was 3 months (95% CI 2.07-6.44) 46 . In this study 46 , 11 of 40 (28%) patients who were in CR underwent allo HSCT, and 2 of these 11 patients subsequently experienced CD19 + leukaemia relapse. In an open label, phase I, dose escalation study of anti CD19 CAR T cells (containing a CD28 co stimulatory domain and manufactured using a retroviral vector) involving children and young adults with ALL or non Hodgkin lymphoma performed by the US NIH, the CR rate was 66.7%. Following remission, 10 of 12 (83%) patients who achieved MRD negative remission underwent HSCT and remained disease free at the time of publication of the data 22 .
At this time, whether CAR T cell therapy is a defin itive treatment remains unclear. While strategies to understand antigen escape mechanisms and to increase rates of long term remission are developed 134 , allo HSCT can reasonably be considered for patients with Table 3 | CAR T cell-related encephalopathy syndrome grading and management

Signs and symptoms
For patients aged >12 years (with age-appropriate cognitive performance): • Grade 1 somnolence, confusion, encephalopathy , dysphasia, seizure (brief partial seizure without loss of consciousness), and/or tremor a • Neurological assessment score [7][8][9] according to CARTOX-10 grading system 37 For patients aged ≤12 years: • Grade 1 CNS toxicities as above and CAPD 116 score <9 For patients aged >12 years (with age-appropriate cognitive performance): • Grade 2 somnolence, confusion, encephalopathy , dysphasia, seizure (brief generalized seizure), and/ or tremor a • Neurological assessment score 3-6 For patients aged ≤12 years: • Grade 2 CNS toxicities as above and CAPD score <9 For patients aged >12 years (with age-appropriate cognitive performance): • Grade 3 somnolence, confusion, encephalopathy , dysphasia, seizure (multiple seizures despite medical interventions), tremor and incontinence or motor weakness a , and/or elevated intracranial pressure (stage 1 or 2 papilloedema b with CSF opening pressure <20 mmHg) • Neurological assessment score 0-2 For patients aged ≤12 years: • CAPD score ≥9 • Patient is critical, obtunded, and/or unable to perform CAPD • High-grade (stage [3][4][5] papilloedema haematological malignancies who have achieved remis sion following CAR T cell therapy. Alternatively, as CAR T cell product specific data matures, it might also be reasonable to consider CAR T cell therapy as a defini tive treatment. The decision to proceed with allo HSCT should be based upon the candidate meeting standard eligibility requirements, and the long term outcomes associated with the specific CAR T cell product used should be considered in the risk-benefit assessment.

Ethical considerations
Currently, CAR T cell therapy for paediatric patients is available for only those with high grade, relapsed and/ or refractory ALL. Remission rates among children with relapsed and/or refractory ALL, who previously had no curative options, have been impressive with current CAR T cell therapies 97 . Nevertheless, not all children with relapsed and/or refractory ALL are appropriate candidates for this therapy. Patients who do not have a reasonable expectation of survival between leukapher esis and CAR T cell administration or whose survival after CAR T cell therapy is expected to be limited by other comorbidities should not be considered as candi dates for this treatment. Among these groups, the risks of primary disease progression must be weighed against the risk of accelerating mortality and/or causing severe disability that could potentially be associated with CAR T cell therapy 135 .
Financial and health-system considerations We understand that value in health care is determined by patient outcomes balanced against costs. The current estimated cost of standard of care CAR T cell therapy for children with ALL is high 136,137 . Moreover, the ancillary administrative and supportive care service (including management of complications, intensive care unit stays, and frequent hospitalization) costs can be substantially higher than the CAR T cell product price tag 138 . We strongly encourage consideration of the quality adjusted life years gained for paediatric patients who can poten tially achieve long term remission as a result of this therapy and encourage all efforts to reduce the costs of care 137,139,140 (Table 1). We anticipate that advances in CAR T cell technology will improve our understanding of the pathophysiology of CRS-CRES and facilitate the dis covery of predictive biomarkers with which to identify patients requiring early intervention with available sup portive care, which will subsequently lead to improved outcomes and reductions in the cost of this care, in addi tion to biomarkers for identifying in advance patients who are unlikely to respond. As payers and health sys tems determine coverage benefits, it is important that the specific needs of all children be considered and that access is granted to CAR T cell therapy and the asso ciated supportive care (encompassing baseline assess ments, inpatient observation, and essential supportive care when necessary). Health institutions are encouraged to provide access to CAR T cell therapies; however, adequate strategic and operational planning and preparation are needed to ensure the safe delivery of such treatments. We rec ommend that programmes seek immune effector cell (IEC) accreditation by the FACT as a voluntary means of ensuring adherence to quality standards 55 (Table 1). FACT accreditation will require an established quality assurance programme at the institution, as well as edu cation and ongoing training for clinical staff. Institutions that offer CAR T cell therapy should support rigorous quality assurance, data management, clinical services, and education programmes for interdisciplinary staff involved in the care of patients who receive this treat ment. Additionally, emergency medical services, com munity hospitals, and local triage facilities will require high vigilance to recognize and promptly escalate care in the event that a patient treated with CAR T cell therapy presents to their facility in an emergency.

Nursing considerations
The availability of skilled interdisciplinary staff, includ ing nurses, is an essential requirement for safe admin istration of CAR T cell therapy to paediatric patients. Owing to the need for close medical attention among these patients, communication between coordinators, medical care providers, and nursing administration is important to ensuring that CAR T cell infusions are considered together with staffing decisions 141 Table 3 all need available nursing staff that have completed all required competencies to care for patients treated with CAR T cells. These nurses can assist in the rapid recognition of CRS-CRES and help avoid iatrogenic errors (for example, administration of steroids as a routine pre medication). Visual cues, such as flags in the patient chart or patient bracelets (similar to 'fall risk' identifiers), can help care teams to quickly recog nize recipients of CAR T cells, even during electronic medical record 'downtime procedures' . Discharge pro tocols should ensure comprehensive education of the caregiver and patient about signs and symptoms of CRS and CRES, and the patient should also be given a wallet or REMS identification card for their spe cific CAR T cell product. Most importantly, recipients of CAR T cells should be instructed to immediately alert all providers that they have received this therapy, especially if presenting to a facility outside of their original treatment centre.

Pharmacy considerations
As CAR T cell therapies transition from experimental therapies to standard ofcare treatments, institutional pharmacists must be engaged in the development of policies and protocols to optimize supportive care. Additionally, REMS programmes are likely to accompany new therapy approvals by the FDA. For instance, the REMS programme for tisagenlecleucel requires institutions to have a minimum of two doses of tocilizumab available on site for each patient at risk of CRS-CRES in order to enable immediate admin istration 143 . New CAR T cell therapies might also require an adequate stock of other agents directed at ameliorating CRS and/or CRES and other supportive therapies (such as rituximab and cetuximab for the aforementioned products with integrated suicide or safety switches) 49, 56,61 . Institutional pharmacies will need to ensure adequate training of pharmacy staff regarding the recognition and management of CRS and CRES and the development of policies and protocols to ensure adequate stock and the prompt dispensation of supportive therapies 55 .

Emergency contingency planning
Institutions involved in the administration of CAR T cell therapies are encouraged to develop programme specific emergency plans that take into account the particular needs of patients exposed to these treatments 144 . Many patients will be required to live within a specified dis tance of the treating hospital for a predefined amount of time after CAR T cell infusion. Should acute toxici ties develop, this stipulation will ensure prompt access to care, which will help to mitigate against additional complications. When possible, institutions should develop protocols to overcome barriers to care for these patients in the event of natural disasters and/or disrup tion of services. Prespecified electronic medical record downtime procedures, evacuation plans that include access to anti IL6 therapies, and the consideration of admission of outpatients at risk of CRS-CRES before an anticipated natural disaster are examples of disaster planning protocols 145,146 .

Conclusions
To achieve improvements in CAR T cell therapy and the requisite supportive care, and thereby sustain CR rates among children, we recommend that enrolment of children in trials of novel agents be facilitated at the earliest acceptable time points. In this regard, application of CAR T cell therapies for treatment of solid tumours and other haematological malignancies in children is being explored (NCT03126864, NCT01953900, NCT02311621, and NCT03056339). Moreover, consid eration of earlier or upfront use of CAR T cell therapy might spare patients the acute and long term toxicities associated with traditional chemotherapy and/or radi ation regimens. Whether CAR T cell therapy should be followed by allo HSCT or repeat CAR T cell infu sions also needs to be explored. Future studies aimed at improving the persistence of CAR T cells and thereby inducing long term remission without the need for further therapy are an important requirement. An over arching commitment to improve patient outcomes, especially among children with limited or no therapeutic options, is key.
Currently, the time required from leukapheresis to manufacture CAR expressing cells is a major limitation of CAR T cell therapy. Efforts to explore third party, off theshelf allogeneic approaches to therapy with CAR T cells or other immune effectors, such as CAR NK cells, are exciting potential alternatives 60 .

Box 5 | Proposed management algorithms for increased intracranial pressure
Algorithm A: stage 1-2 papilloedema with CSF opening pressure <20 mmHg and without evidence of cerebral oedema Acetazolamide 15 mg/kg (maximum 1,000 mg) intravenous (i.v.) followed by 8-12 mg/kg (maximum 1,000 mg) i.v. every 12 hours; monitor renal function and acid-base balance once or twice daily and adjust dose accordingly Algorithm B: management of stage 3-5 papilloedema, any evidence of cerebral oedema on imaging studies, or CSF opening pressure ≥20 mmHg use high-dose corticosteroids according to recommendations for grade 4 chimeric antigen receptor (CAR) T cell-related encephalopathy syndrome (CRES; see Table 3) along with the following measures for the management of cerebral oedema: • elevate head of bed to an angle of 30 degrees • Hyperventilation to achieve target PaCo 2 of 30-40 mmHg during the acute management of intracranial hypertension (or acute management of intracranial hypertension according to accepted institutional guidelines) • Hyperosmolar therapy with either 20% mannitol or hypertonic saline (3%) -mannitol: initial dose of 0.5-1 g/kg; maintenance dose 0.25-1 g/kg every 6 hours (check metabolic profile and serum osmolality every 6 hours, and hold mannitol if serum osmolality is ≥320 mosm/kg or osmolality gap is ≥40) -Hypertonic 3% saline: initial dose 5 ml/kg i.v. over 15 mins; maintenance dose 1 ml/kg per hour i.v. to reach a target serum sodium level of 150-155 meq/l (check electrolytes every 4 hours, and hold infusion if sodium level is >155 meq/l) • If patient has an ommaya reservoir, drain CSF to a target opening pressure of <20 mmHg • Consider neurosurgery consultation and i.v. anaesthetics for burst-suppression pattern on electroencephalography • Perform metabolic profiling every 6 hours, daily CT of the head, and adjust above medications to prevent rebound cerebral oedema, renal failure, electrolyte abnormalities, hypovolemia, and hypotension CSF, cerebrospinal fluid; meq, milliequivalents; mosm, milliosmole; PaCo 2 : partial pressure of carbon dioxide in arterial blood.
Additional long term prospective studies to under stand the pathophysiology and develop early recog nition and optimal treatment strategies for CRS and/ or CRES are needed. Such studies should involve broad collaborative efforts because these toxicities are multifaceted and necessitate interdisciplinary care. Response and adverse event rates, as well as late effects, are likely to vary depending on various host, disease, and CAR T cell characteristics; therefore, a comprehensive and robust registry is needed to guide future efforts to optimize and expand the use of CAR T cell therapy.