Critically Ill Patients Treated for Chimeric Antigen Receptor Related Toxicity: A Multicenter Study

Abstract

Objective:

To report the epidemiology of patients, treatments and outcomes of adult patients admitted to the ICU after cytokine release syndrome (CRS) or immune effector cell-associated neurotoxicity syndrome (ICANS).

Design:

Retrospective cohort study.

Setting:

11 centers across the US part of the ICU-initiative.

Patients:

Adult patients treated with CAR T-cell therapy who required ICU admission between November 2017-May 2019.

Interventions:

Demographics, toxicities, specific interventions and outcomes were collected.

Results:

105 patients treated with axicabtagene ciloucel required ICU admission for CRS or ICANS during the study period. At the time of ICU admission the majority of patients had Grade 3–4 toxicities (66.7%); 15.2% had Grade 3–4 CRS and 64% Grade 3–4 ICANS. During ICU stay, CRS was observed in 77.1% patients and ICANS in 84.8% of patients; 61.9% patients experienced both toxicities. 79% of patients developed ≥Grade 3 toxicities during ICU stay, however need for vasopressors (18.1%), mechanical ventilation (10.5%) and dialysis (2.9%) was uncommon. ICE score<3 (69.7%), seizures (20.2%), status epilepticus (5.7%), motor deficits (12.4%) and cerebral edema (7.9%) were more prevalent. ICU mortality was 8.6% with only 3 deaths related to CRS or ICANS. Median overall survival time was 10.4 months (95%CI:6.64,NA). Toxicity grade or organ support had no impact on overall survival; higher cumulative corticosteroid doses were associated to decreased overall and progression free survival.

Conclusion:

This is the first study to describe a multicenter cohort of patients requiring ICU admission with CRS or ICANS after CAR T-cell therapy. Despite severe toxicities, organ support and in-hospital mortality was low in this patient population.

INTRODUCTION:

Chimeric antigen receptor (CAR) cell therapy has shown promising results in the treatment of refractory malignancies such as diffuse large B-cell lymphoma (DLBCL), acute lymphoblastic leukemia, multiple myeloma and mantle cell lymphoma^1–11. Two of the most common toxicities observed after CAR T-cell infusion are cytokine release syndrome (CRS) and immune effector cell associated neurotoxicity syndrome (ICANS)^12–15. Both syndromes are associated with an inflammatory response caused by the activation of CARs after binding tumor antigens. Signs of CRS consist of fever, hypotension and/or hypoxia; Grade 3 and 4 toxicities require vasopressors, oxygen supplementation higher than nasal cannula and multi-organ failure can occur¹⁴. Mild presentations of ICANS consist of mild aphasia, apraxia and inattention¹⁴. Patients with Grade 3 and 4 ICANS develop seizures, cerebral edema, obtundation, motor deficits and Immune Effector Cell-Associated Encephalopathy (ICE)/CAR T-cell-therapy-associated TOXicity (CARTOX) score<3¹⁴.

The incidence and severity of CRS and ICANS varies between cell products; earlier studies showed that toxicities occur in 40–93% of patients with 13–40% exhibiting Grade 3–4 toxicity^2,7,9. A recent survey showed that while all patients with Grade 3 and 4 toxicities are admitted to the ICU, up to 70% of centers will admit patients with Grade 1 and 2 toxicities for close monitoring¹⁶. Current practices in the support of critically ill CAR patients are based on extrapolation of data from the general ICU population, and can vary significantly across centers and CAR products^15,17–19. It is unclear how these practices could affect outcomes in these patients.

The objective of this study is to describe the characteristics, treatments and outcomes of adult patients admitted to the ICU for CRS and ICANS. A better understanding of these epidemiological data is necessary to identify areas for future research that aim to improve outcomes in this population.

METHODS:

Study design

This was a retrospective cohort study of adult patients requiring ICU admission for CAR related toxicities at 9 national centers of the CAR–ICU initiative between November 2017-May 2019(PA19–0365). Local Institutional Review Boards (IRB) approved the protocol and waived informed consent. De-identified data were examined independently by CG and ARB and any discrepancies were reviewed with each center. Patients admitted to the ICU for CRS or ICANS after treatment with CAR therapy were identified by each center using internal registries. Eight centers included only patients treated with axicabtagene ciloleucel (axi-cel) or tisagenlecleucel, and 1 center included patients treated with their own non-commercial product (Supplement). Patients were excluded if younger than 18 years of age or admitted to the pediatric ICU or had ICU admissions not related to CRS or ICANS. For patients with multiple ICU admissions, each admission was recorded separately and only the last admission for CRS or ICANS was included in the final analysis(Figure 1). A Redcap^20,21 data collection instrument included: a) demographics and malignancy related data; b) CAR products, the toxicities observed and their specific treatments; c) details of ICU support, Sequential Organ Failure Assessment [SOFA] scores and outcomes. The study timeframe preceded the standardized grading consensus for CRS and ICANS by the American Society for Transplantation and Cellular Therapy (ASTCT), therefore different grading criteria were utilized across institutions^14,15,18. To counterbalance this variability, we measured specific organ toxicities by recording both organ support modalities and using common terminology criteria for adverse events (CTCAE, V4.0)²². Organ support modalities included vasopressors, mechanical ventilation (MV), high flow oxygen therapy (HFOT), non-invasive ventilation (NIV) and continuous renal replacement therapy (CRRT). To account for variability in ICANS grading, specific features (seizures, status epilepticus, ICE/CARTOX score, motor deficits and cerebral edema) were documented. With this clinical information specific for ICANS and CRS, standardization of toxicity grade was made using ASTCT criteria (refer to Supplement for ASTCT criteria)¹⁴.

Figure 1.

Inclusion criteria for CAR T-cell patients treated for the period of November 2017-May 2019 in all institutions.

Statistical Analysis:

Summary statistics including mean, standard deviation, median, and range for continuous variables and frequency counts and percentages for categorical variables. Fisher exact test or χ² test were used to evaluate the association between two categorical variables. Wilcoxon rank sum test was used to evaluate the difference in continuous variables between groups. We calculated ICU and hospital length of stay using (LOS) as the time interval from admission to discharge or death. LOS was reported with median and range. The Kaplan-Meier method was used to estimate the rate and median overall survival (OS) and progression free survival (PFS). Overall survival time was calculated from CAR T-cell infusion date to death or last follow-up date. Patients were followed up for at least 1 year after infusion and median follow-up time for the censored observations (alive patients) was reported. Progression free survival was calculated from time of CAR T-cell infusion to disease progression or death, whichever happened first. Considering the small amount of events occurring after 1 year after infusion, patients with the last follow up or event occurring after one year post infusion, were censored at one year. Statistical software SAS (version 9.4, Cary, North Carolina) and S-Plus 8.2 (TIBCO Software Inc., Palo Alto, CA) were used for the analyses. A p value <0.05 was considered significant.

RESULTS:

General characteristics:

During the study period, 345 patients received CAR T-cell therapy (axi-cel, tisagenlecleucel and experimental autologous anti-CD19 CAR T-cells) of which 120 (34.7%) were admitted to the ICU. Patients with ICU admissions not related to CRS or ICANS were not part of the final analysis; their clinical course is described in the Supplement. Nine patients treated with products other than axi-cel were not part of the analysis to avoid generalizations of our findings to all CAR products, leaving 105 patients for the final analysis(Figure 1). Most patients were male (65.7%), white (71.4%), had a comorbidity index≤4 (50.5%) and average age was 57±15 years (Supplement). 83.8% had received 3 or more lines of chemotherapy prior to CAR infusion. The main reason for ICU admission was ICANS in 44.8% of patients, CRS in 40.9% and both CRS and ICANS in 14.3%. Grade 3–4 toxicities were present on admission in 66.7% of patients and were experienced by 76.2% at some point during their ICU stay. Most patients (61.9%) experienced CRS and ICANS concomitantly during their ICU stay. Median time from CAR infusion to ICU admission was 5 days (0–74). Median SOFA score on admission was 4 (1–21) and maximum SOFA score during ICU stay was 5 (1–23). Twenty two patients (21%) had documented infections and 86.4% of these had Grade 3–4 toxicities (Supplement). Median ICU LOS was 4 days (1–22) and was shorter for Grade 1–2 toxicities than for those with Grade 3–4 toxicities(Table 1).

Table 1.

Characteristics and outcomes of axicabtagene ciloleucel patients admitted to the ICU

ICU Characteristics	All patients n=105; n (%)	CRS^^ n=81; n (%)	ICANS^^ n=89; n (%)
Main reason for ICU admission  CRS  ICANS   Both	43 (40.9%) 47 (44.8%) 15 (14.3%)	-----	-----
Toxicities during ICU stay  CRS  ICANS  Both	16 (15.2%) 24 (22.9%) 65 (61.9%)	-------	-------
Grade 3–4 toxicity at the time of ICU admission	70 (66.7%)	19 (23.5%)	57 (64%)
Grade 3–5 toxicity during ICU stay*	80 (76.2%)	28 (34.6%)	67 (75.3%)
Infusion to ICU admission, days (median, range)	5 (0–74)	5 (0–40)	6 (2–74)
SOFA score ICU admission (median, range)	4 (1–21)	4 (1–21)	4 (1–14)
Maximum SOFA score (median, range)	5 (1–23)	6 (1–23)	6 (1–19)
Any organ support	24 (22.9%)
Vasopressors  1 vasopressor  ≥2 vasopressors	19 (18.1%)	19 (23.5%) 16 (84.2%) 3 (15.8%)	------
Days on vasopressors (median, range)	1 (1–4)	1 (1–4)	------
Respiratory support  Mechanical ventilation  HFOT  NIV	11 (10.5%) 9 (8.6%) 2 (1.9%)	4 (4.9%) 9 (11.1%) 2 (2.5%)	7 (7.9%) ------ ------
Renal replacement therapy	3 (2.9%)	-----	------
Outcomes
ICU length of stay, days (median, range)  Grade 1–2*  Grade 3*  Grade 4*  Grade 5*	4 (1–22) 2 (1–14) 4 (1–22) 5.5 (1–21) 6 (6–12)	4 (1–22)	4 (1–22)
Hospital length of stay, days (median, range)	24 (5–180)	23 (5–80)	23 (5–180)
ICU mortality	9 (8.6%)	1 (1.2%)#	2 (2.2%)#
Hospital mortality  CRS/ICANS related  POD related  Non-POD related   Sepsis   Other	16 (15.2%) 3 (18.7%) 7 (43.8%) 6 (37.5%) 5 (83.3%) 1 (16.7%)	-----	------

Definition of abbreviations: BAL=bronchioalveolar lavage; CRS = cytokine release syndrome; ICANS = immune effector cell associated neurotoxicity syndrome; CAR = chimeric antigen receptor; SOFA = sequential organ failure assessment; HFOT = high flow oxygen therapy; NIV=non-invasive ventilation; ICU = intensive care unit; POD =progression of disease.

^{^^}CRS and ICANS during ICU stay

^{^}Co-morbidities include: myocardial infarction, congestive heart failure, cerebrovascular disease, dementia, chronic obstructive pulmonary disease, connective tissue disease, liver disease, diabetes, hemiplegia, renal disease, malignancy, AIDS.

^*Maximum grade for any toxicity during ICU stay as per ASTCT grading consensus. Includes hypoxemia requiring HFOT, NIV, mechanical ventilation, hypotension requiring vasopressors, seizures (any type including status epilepticus), CARTOX/ICE score<3, cerebral edema and obtundation. Grade 5= death.

^**Bacteremia included: gram positive cocci (enterococcus, staphylococcus, candida, bacterioides); BAL and tracheal aspirates: stenotrophomonas maltophilia, staphylococcus (sensitive or MRSA), ESBL E.coli, cytomegalovirus, aspergillus/mold; stool: clostridium difficile and salmonella; Cytomegalovirus reactivation in serum or BAL PCR.

^***Distribution of positive cultures according to toxicity grading as per ASTCT grading consensus: Grade 1–4.5%, Grade 2–9.1%, Grade 3–54.5%, Grade 4–27.3%, Grade 5–4.5%.

^#refers to ICANS or CRS being the direct cause of death

Details of CRS and ICANS during ICU stay:

81 patients developed CRS during ICU admission. Grade 3–4 CRS was present in 23.5% at the time of admission and in 34.6% throughout their ICU stay. Median time from infusion to peak CRS in the ICU was 5 days (0–42). Forty-four (54.3%) patients were admitted with hypotension, only 23.5% required vasopressors and most (84.2%) required only one (Table 1). The main vasopressor used was norepinephrine (94.7%), followed by phenylephrine (21.1%), epinephrine (10.5%) and vasopressin (10.5%). Hypoxia related to CRS occurred in 29.6% of patients and support included HFOT (ASTCT Grade 3), NIV and MV (ASTCT Grade 4) in 11.1%, 2.5% and 4.9% of patients respectively (Table 1). Other organ toxicities associated with CRS were observed in 76.5% of patients, most commonly symptomatic arrhythmias (25.9%) of which 71.4% of these were life-threatening or required urgent intervention, hepatic toxicity (29.6%) and acute renal failure (9.8%). Additional details can be found in the Supplement.

Eighty-nine patients (84.7%) developed ICANS during their ICU stay. Grade 3–4 toxicity was present in 64% at the time of admission and in 75.3% during the ICU stay. Median time from infusion to peak ICANS grade in the ICU was 6 days (2–74). 76.4% of patients had an ICE score of 0–2 (ASTCT Grade3) or were obtunded (ASTCT Grade4). Seven patients with ICANS required MV for airway protection (Table 1). Seizures (ASTCT Grade 3–4) were observed in 20.2% of patients and 6 patients developed status epilepticus (ASTCT Grade4). Cerebral edema was present in 7 patients and management focused mainly on corticosteroids and anti-cytokine therapy. Additional details can be found in the Supplement.

Treatments:

Variability was observed in treatment approaches for CRS and ICANS. The most common treatments used were tocilizumab (82.9%) and corticosteroids (88.6%). Patients received a median of 2 doses of tocilizumab(1–7), prescribed first for CRS alone (60.9%) or concomitant CRS and ICANS (27.6%). The first dose of corticosteroids was administered for isolated ICANS in 66.7% of patients, isolated CRS in 12.9%, and concomitant CRS and ICANS in 20.4%. Median cumulative doses of corticosteroids during the hospital stay was 1,173mg of methylprednisolone equivalent (50–20,056mg). Siltuximab use was rare and prescribed after failed treatment with at least ≥2doses of tocilizumab. Use of anakinra was rare (8.6%) and for patients with Grade 3–4 toxicities (Supplement).

Outcomes:

ICU (8.6%) and hospital (15.2%) mortality were low. The majority of deaths were related to either progression of disease (n=7;43.7%) or sepsis (n=5;31.3%)(Table 1). Mortality due to CRS (1 patient;0.95%) and ICANS (2 patients;1.9%) was extremely rare. The patient whose death was associated with CRS experienced peak toxicity on day 4 post-infusion. He developed refractory arrhythmias, shock, renal failure and concomitant Grade 4 ICANS. Of the two patients who died of ICANS, one had concomitant Grade 3 CRS and both toxicities peaked at day 5 post-CAR infusion. The second patient experienced biphasic ICANS with cerebral edema developing in the second phase (day 18 post-infusion). In univariate analysis patients who died in the ICU were more likely to have hepatotoxicity, diagnosis of HLH, cerebral edema, vasopressor use, higher maximum SOFA score, longer time to peak CRS, higher corticosteroid doses(Table 2). Also in univariate analysis, patients who died during their hospital stay were more likely to have other organ toxicity associated with CRS, diagnosis of HLH, seizures, cerebral edema, longer time to peak CRS, higher ICU SOFA scores and higher corticosteroid doses(Table 2). A multi-variable analysis was not performed due to the low incidence of mortality in the study sample.

Table 2.

Characteristics associated with ICU and hospital mortality in univariate analysis.

	ICU Survival (n=96)	ICU death (n=9)	P Value	Hospital Survival (n=89)	Hospital death (n=16)	P Value
Gender (male)	65 (68%)	4 (44%)	0.3	59 (66%)	10 (63%)	0.8
Age, years (mean±stdev)	57±15	58±17	0.8	57±15	57±18	1.0
Comorbidity Index (≤4)*	48 (50%)	5 (56%)	1.00	44 (49%)	9 (56%)	0.6
Lines of chemotherapy prior to CAR (median, range)	4 (1–11)	3 (2–5)	0.05	4 (1–11)	3.5 (2–6)	0.1
Infusion to maximum CRS days (median, range)	5 (0–42)	8 (4–30)	0.02	4 (0–16)	6.5 (3–42)	0.003
Infusion to maximum ICANS, days (median, range)	6 (2–74)	8 (5–32)	0.2	6 (2–74)	6 (3–40)	0.8
SOFA score ICU admission (median, range)	4 (1–14)	7 (2–21)	0.5	4 (1–14)	7 (2–21)	0.01
Maximum SOFA score (median, range)	5 (1–14)	11 (6–23)	<0.001	5 (1–14)	10.5 (6–23)	<0.001
Vasopressors	14 (15%)	5 (56%)	0.009	13 (15%)	6 (38%)	0.03
Mechanical Ventilation	6 (6%)	5 (56%)	<0.001	5 (6%)	6 (38%)	0.002
Organ toxicity associated to CRS (yes)	55 (57%)	7 (78%)	0.2	48 (54%)	14 (88%)	0.03
Renal failure	7 (7%)	1 (11%)	0.6	3 (3%)	5 (31%)	0.004
Hepatotoxicity	19 (20%)	5 (56%)	0.03	15 (17%)	9 (56%)	0.008
HLH	2 (2%)	2 (22%)	0.04	0 (0%)	4 (25%)	<0.001
Seizures	15 (16%)	3 (33%)	0.05	12 (13%)	6 (38%)	0.008
Cerebral edema	4 (4%)	3 (33%)	0.003	3 (3%)	4 (25%)	0.004
ICE/CARTOX≤2	57 (59%)	5 (56%)	0.6	51 (57%)	11 (69%)	0.06
ICU LOS≥5 days	39(41%)	6 (67%)	---	34 (38%)	11 (69%)	----
Tocilizumab (yes)	78 (81%)	9 (100%)	0.4	72 (81%)	15 (94%)	0.30
Siltuximab (yes)	6 (6%)	3 (33%)	0.03	4 (4%)	5 (31%)	0.004
Anakinra (yes)	5 (5%)	4 (44%)	0.003	2 (2%)	7 (44%)	<0.001
Corticosteroids (yes)	84 (88%)	9 (100%)	0.6	77 (87%)	16 (100%)	0.21
Cumulative dose of corticosteroids,mg (median, range)	1,054 (50–20,056)	4,715 (1,173–10,670)	0.001	901 (50–20,056)	4805 (1,173–10,670)	<0.001
Maximum dose of dexamethasone, mg/day(median, range)	40 (0–100)	80 (0–80)	0.008	40 (0–100)	50 (0–80)	<0.001
Maximum dose of methylprednisolone, mg/day (median, range)	160 (0–1,000)	1,000 (1,000–1,000)	0.008	80 (0–1000)	1,000 (100–1,000)	0.003

Definition of abbreviations: CRS = cytokine release syndrome; ICANS = immune effector cell associated neurotoxicity syndrome; CAR = chimeric antigen receptor; SOFA = sequential organ failure assessment; ICE= ; CARTOX= HLH= Hemophagocytic lymphohistiocytosis; ICU = intensive care unit; LOS =length of stay.

^*Co-morbidities include: myocardial infarction, congestive heart failure, cerebrovascular disease, dementia, chronic obstructive pulmonary disease, connective tissue disease, liver disease, diabetes, hemiplegia, renal disease, malignancy, AIDS.

Most patients discharged from the hospital after ICU stay were discharged home (74.3%). Median overall survival of these patients was 10.4 months (95%CI 6.64,NA) with a median follow up of 11.7 months (0.9–26.1)(Figure 2). Requiring organ support or developing toxicities Grade≥3 during ICU stay had no impact on long term survival (Supplement). Cumulative corticosteroid doses ≥1173mg were associated with decreased overall and progression free survival when compared to patients receiving lower doses (Figure 3).

Figure 2.

Overall survival of patients admitted to the ICU with CRS and ICANS after CAR T-cell therapy.

Figure 3.

Overall survival, progression free survival and cumulative corticosteroid dose.

DISCUSSION:

This is the first US multicenter study describing the severity of toxicities, interventions and outcomes of adult CAR patients treated with axi-cel who required ICU admission for CRS and ICANS. During the study period, 34% of patients were admitted to the ICU, a similar rate to that described in existing literature^{7,8,10,23–25}. Most patients developed concomitant CRS and ICANS with Grade 3 and 4 toxicities, but had low requirements for organ support in the ICU. Patients with ICANS had a high prevalence of severe toxicities such as seizures, motor deficits, obtundation, cerebral edema and low ICE/CARTOX score during ICU stay. Despite this, ICU and hospital mortality in this patient population was low and mortality directly related to CRS and ICANS in this study was extremely rare (2.8%).

Although ICU admission criteria for CAR patients are available, in our study we report real life practices in experienced cancer centers^14,15,17. We observed that only 22.9% of our patients met traditional ICU admission criteria for organ support such as vasopressors, MV or CRRT. Despite this, 3 of every 4 patients met criteria for ICU monitoring and management of Grade 3 and 4 toxicities, particularly ICANS. It is possible that ICU admission of patients with Grade≥3 toxicities was beneficial, independent of the need for organ support, and may have contributed to the low overall mortality from CAR-toxicities. Concurrently, 10% of patients deteriorated during their ICU stay and progressed to Grade ≥3 toxicities. In these cases, ICU admission allows for early detection of deterioration and accelerated interventions that could improve outcomes. CAR T-cell patients are known to acutely decompensate and early support by the ICU team might be beneficial. Examining which patients with low grade toxicity derive benefit from ICU admission will be informative in guiding resource utilization. It is important to note that our data reflect practices shortly after FDA approval of axi-cel and it is possible that practices have evolved over time and these data may not reflect current ICU utilization with complete accuracy. Moreover, recent products have lower incidence of toxicities^1,9,11, and with increasing experience, oncologists have become more comfortable managing and monitoring some of these patients on the wards rather than the ICU. Measuring the impact of ICU admission for patients not requiring organ support on outcomes is required, especially in the face of scarce ICU resources during the COVID-19 pandemic.

Additional research to identify predictors of ICU and hospital mortality is needed to inform accurate prognostication in the CAR T-cell population. Due to the low hospital mortality rate in our study, performing a multivariable analysis to identify independent predictors of mortality was not feasible, limiting our ability to generate conclusions that could change practices. Prior research has identified age, number of lines of previous chemotherapy, renal failure and SOFA score on admission as predictors of mortality in critically ill cancer patients^26–29. Rather, we observed that longer times from cell infusion to peak CRS, cumulative corticosteroid use, cerebral edema and use of rescue agents (e.g: siltuximab and anakinra) were associated with both ICU and hospital mortality. While need for vasopressors, mechanical ventilation and maximum SOFA score during ICU stay where associated to increased mortality, in-hospital deaths were mostly as a consequence of progression of disease or infectious complications. A close discussion with the oncologist to determine if the cause of multi-organ failure is sepsis or progression of disease, rather than CRS or ICANS, might help prognosticate short term outcomes. However, these findings should be validated in larger studies, as ours was limited by a low mortality rate that precluded adjusted analyses.

In our study, increasing cumulative doses of corticosteroids were associated with decreased odds of survival, a finding which has been previously reported ³⁰. Additionally, patients treated with higher cumulative doses of corticosteroids had lower progression free survival. These findings may suggest that corticosteroids can have an impact on remission responses after CAR-T cell treatment and therefore survival. Prescription of steroids was also subject to prescriber bias and variability of treatment protocols across institutions. Therefore, higher corticosteroid doses may have been prescribed in sicker patients and serve as a surrogate marker of more severe illness or higher-grade toxicities. Identifying the effects of corticosteroid dose on clinical outcomes is an area of increasing interest, particularly given reported data of infectious complications, risk of suppressing CAR-T cell expansion and shorter progression- free survival with their use^30–33. Further assessment is needed to determine if interventions such as corticosteroid-sparing therapies and earlier introduction of corticosteroids (Grade 2 instead of Grade 3), could potentially reduce overall corticosteroid exposure needs further assessment^31,34,35. Evaluating if these interventions have an impact on long-term survival, disease progression and infections should be a focus of future research.

ICU and hospital mortality in our cohort of axi-cell-treated patients appears low in comparison to other studies of critically ill cancer patients^26,29,36,37. Despite 76% of patients experiencing advanced grade 3 and 4 toxicities, death due to CRS and ICANS was extremely rare and occurred in only 3 patients. High grade CRS and ICANS has been associated with high CAR proliferation, which is essential to therapy response. This suggests that severity of toxicity or organ failure associated with CRS or ICANS may not be an ideal parameter to guide conversations of goals of care. Practices such as readdressing goals of care after a 3–5-day ICU trial need to be revisited in the CAR T-cell population³⁸. Over 40% of patients in our study had an ICU LOS ≥5days, re-evaluating goals of care and de-escalating care so quickly after admission would have led to limitation of care in as many as half of patients who survived in our cohort. Furthermore, those patients who survived had a median survival of 10.4 months, superior outcomes to those of other critically ill cancer patients. Therefore, our data support a longer trial of critical care in CAR patients if malignancy progression or severe sepsis are not the cause of persistent organ failure.

We recently published a survey describing common practices in the ICU when caring for critically ill CAR patients across our centers¹⁶. In the survey, centers reported variability in indications for use of corticosteroids and other anti-cytokine therapies, which we similarly observed in the present study¹⁶. The impact of variability in treatments on mortality, symptom duration, inflammatory profile and long-term remission is yet to be determined^30,31,39. Complications from CRS such as the need for vasopressors, cardiomyopathy and non-cardiogenic edema requiring respiratory support were described as common in our earlier survey¹⁶, but were not highly prevalent in our cohort of patients. The self-reporting nature of the survey could explain some of these differences; however, it is important to note that the majority of our centers are using multiple CAR products, and our current data report only patients treated with axi-cel. Therefore, one should be cautious in extrapolating our findings to all CAR patients. As CAR therapy continues to evolve, reports of ICU management related to different products will be of importance for the critical care community to help allocate resources and educate staff.

There are some limitations to our study. First, the CAR-ICU collaborative are experienced and highly specialized tertiary care centers in the US, therefore our results may not be fully generalizable to all centers who administer CAR treatment. Second, our data evaluated one of the many CAR products, therefore caution should be taken in applying our findings to patients treated with other CAR cell therapy. In addition, our data reflect the immediate period after FDA approval of axi-cel and practices may have evolved with increasing experience. Lastly, the retrospective nature could have contributed to measurement bias related to definition of toxicity. Despite these limitations, this is the first study to look into the specifics of the critically ill CAR T-cell patient. Our data will help other intensivists and institutions learn about this specific patient population and better understand how to support these patients and develop their CAR T-cell programs.

CONCLUSION:

In this multicenter cohort study of patients requiring ICU admission with CRS and ICANS, the need for organ support in this patient population was low despite advanced Grade 3–4 toxicities being frequent. Additional study of outcomes associated with ICU interventions is needed to optimize care and inform prognostication for critically ill patients with CAR-related toxicities.

Abbreviations list:

ASTCT

American Society for Transplantation and Cellular Therapy

axi-cel

axicabtagene ciloucel

CAR

chimeric antigen receptor

CCI

Charlson comorbidity index

CRS

cytokine release syndrome

CTCAE

common terminology criteria for adverse events

ICANS

immune cell associated neurotoxicity syndrome

ICU

intensive care unit

HFOT

high flow oxygen therapy

LOS

length of stay

MV

mechanical ventilation

NIV

non-invasive ventilation

PRES

posterior reversible encephalopathy syndrome

SOFA

sequential organ failure assessment