Background

Chimeric antigen receptor T-cell (CAR-T) therapy is an innovative approach for managing refractory haematological malignancies. Autologous cytotoxic T lymphocytes are genetically modified to specifically recognise a tumour antigen, thereby causing tumour lysis [1, 2]. CAR-T therapy, first approved by the Food and Drug Administration in 2017 and by the European Medicines Agency in 2018, has provided extended survival and complete remission in patients with relapsed or refractory diffuse large B-cell lymphoma [3, 4] or acute lymphoblastic leukaemia (ALL) [5]. With newer indications such as multiple myeloma (MM) [6, 7], earlier treatment in lymphoma [8], and ongoing clinical trials in patients with solid tumours [9, 10], the number of patients given CAR-T therapy is predicted to increase steadily.

However, the inflammatory response generated by activated CAR-T cells can lead to potentially life-threatening complications, namely, cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS). CRS is the most frequent, develo** in up to 93% patients [3,4,5], and is defined as a febrile capillary leak syndrome, with hypotension, hypoxia, and organ failures in severe cases [19,20,21]. All our patients with ICANS also had CRS, in accordance with current pathophysiological knowledge [32, 33], although in the international CARTTAS cohort, 7/238 patients had ICANS without CRS [19]. The treatment of CRS relies on the IL-6 receptor antagonist tocilizumab and on corticosteroids as a second-line treatment, whereas corticosteroids are the recommended first-line treatment for ICANS [30, 31]. All ICU studies found similar rates of tocilizumab and corticosteroids use, but with major differences across centres regarding the drugs chosen, doses, timing of administration and discontinuation, and additional treatments [19,20,21]. Although guidelines have been issued, current recommendations rely on expert opinion and observational data, as no randomised controlled trials are available. Concern has been expressed regarding potential deleterious effects of high-dose corticosteroids on the efficacy of CAR-T therapy [20, 34], as the overwhelming majority of deaths after CAR-T infusion are due to disease progression or relapse [19, 34, 35]. Further studies are therefore needed to define the optimal management of CRS and ICANS.

HLH developed in 17.9% of our patients, compared to 3.8%–5% in previous reports [19, 20]. This discrepancy may be related to differences in the underlying malignancies, type of CAR-T used, and diagnostic criteria for HLH. This syndrome is challenging to distinguish from CRS and sepsis in clinical practice [30]. A recent study emphasised the need for further research to better recognize, define, and treat HLH in CAR-T recipients [36].

As ICU admission usually occurs early after CAR-T therapy, during the neutropenic phase, sepsis is a major concern and the main differential diagnosis of CRS in daily practice. The CARTTAS study found bacterial infection to be independently associated with a twofold higher mortality rate [19]. Microbiologically documented sepsis has been reported to occur in 16% to 30% of patients, in kee** with the 27% proportion in our cohort [19,20,21]. Three of our patients had fungal infections. The optimal prophylactic or pre-emptive strategy for infection after CAR-T therapy has yet to be determined but as of now, urgent broad-spectrum antibiotics in all febrile patients with neutropenia remain essential [37, 38].

Finally, although severe CRS and ICANS may require life-supporting interventions, recovery usually occurs within a week. ICU admission after CAR-T therapy was consistently associated with greater than 90% survival at ICU discharge [19,20,21]. Our experience was similar, with an overall mortality rate of 4.8%. Nonetheless, the subset of patients who required ICU readmission had a higher mortality. Importantly, ICU survivors had similar one-year outcomes to those of patients not admitted to the ICU.

Study implications

Our findings of excellent in-ICU and one-year outcomes, despite a substantial rate of high-grade toxicities following CAR-T therapy, support early unrestricted ICU admission without undue concern about a possible negative impact of critical care on CAR-T efficacy and the haematological prognosis. Moreover, our results imply that because sepsis is common and indistinguishable from CRS in clinical practice, sepsis can never be ruled out during the neutropenic phase and should be thoroughly investigated, and broad-spectrum antibiotics administered without delay. Finally, despite rapid advances in pathophysiological understanding, the main immunomodulatory treatments and specific therapies used to control CAR-T toxicities remain largely empirical. The prospective collection of data in nationwide registries would increase the amount of available data, thereby providing a strong basis for further research.

Strengths and limitations

Our study has several limitations. First, the difficulty of distinguishing CRS from sepsis carries a risk of adjudication bias, particularly given the retrospective design. Similarly, confounding factors such as uraemia, high fever, and use of beta-lactams or neurotropic agents may lead to ICANS-like symptoms. However, in the absence of a reference-standard diagnostic strategy, we strictly applied ASTCT grading recommendations for CRS and ICANS and reported microbiologically documented infections, refraining from classifying patients based on subjective criteria. Second, we included patients over a nearly 4-year period, during which both critical-care and haematology teams gained experience in managing CAR-T therapy recipients. The criteria for transferring these patients to the ICU may therefore have changed over the recruitment period. Our results may not apply to other ICUs with different admission policies. However, our study adds relevant and comprehensive data from two experienced ICUs that were among the first to treat CAR-T therapy recipients in France. Third, the study design prevented us from evaluating how the specific treatments used in the ICU may have affected patient outcomes. However, current recommendations rely chiefly on low-level observational evidence. Fourth, more than 70% of patients were treated with axicabtagen ciloleucel, and our findings may not apply to patients treated with other CAR products. Finally, the very low mortality precluded a multivariable analysis designed to identify independent predictors of death.

Conclusion

In conclusion, our study confirms that intensive care is an integral part of the management of patients given CAR-T therapy. Both specific toxicities (CRS, ICANS, and HLH) and sepsis may require intensive care. The short-term outcomes are excellent, and critical care is not associated with worse one-year haematological outcomes. Studies are needed to investigate the interplay between CAR-T efficacy, toxicity, and the impact of immunomodulating treatments. Moreover, as the current standard of care remains largely empirical, interventional studies are now needed to guide clinical practice.