Low counts count after CAR-T for ALL

In this issue of Blood, Nair et al¹ assessed the predictive ability of the CAR-HEMATOTOX (CAR-HT) tool for early immune effector cell–associated hematotoxicity (ICAHT) after chimeric antigen receptor (CAR) T-cell therapy for pediatric B-cell acute lymphoblastic leukemia (B-ALL). After uncovering limitations of CAR-HT in this population, the authors developed and validated a more refined model for B-ALL, termed ALL-Hematotoxicity (ALL-HT; see figure).

Across products and diseases, ICAHT is the most common CAR T-cell–related adverse event.² Because cytopenias, especially neutropenia and thrombocytopenia, can be prolonged and lead to significant complications (infection, bleeding, transfusion-dependency, and prolonged hospitalization), ICAHT has emerged as a critical area of study.^2,3 The incidence, severity, duration, and outcome of ICAHT are now well-defined in cohorts of adult patients with lymphoma or multiple myeloma.⁴ The CAR-HT score was established to predict development of ICAHT based on preinfusion factors and is validated in large B-cell lymphoma, mantle cell lymphoma, and multiple myeloma.⁵ However, ICAHT is not well described in B-ALL and CAR-HT is not validated in B-ALL. Notably, B-ALL is a disease with almost universal bone marrow (BM) involvement and with severe cytopenias observed frequently in the conventional therapy setting, factors distinct from lymphoma or multiple myeloma. Considering this, the nature and duration of cytopenias after therapy for B-ALL differ from other diseases. Thus, Nair et al set out to address gaps in understanding and predicting cytopenias after CAR T-cell (CAR-T) therapy for B-ALL, with a focus on early (prior to day +30) ICAHT.

In a cohort of 156 children and young adults treated on phase I studies of CD19- and/or CD22-targeted CAR-T therapy for B-ALL at the National Cancer Institute (NCI), the authors reported high rates of severe neutropenia and thrombocytopenia during the first 30 days after CAR infusion, with 83 (53%) meeting criteria for early grade ≥3 ICAHT.⁴ Interestingly, patterns of count decreases and recovery differed by ICAHT grade and by baseline BM disease burden. In applying the original CAR-HT tool to this cohort, the score correlated moderately well with early ICAHT (area under the curve = 0.74, P = .001). However, 87% of the patients fell into the high-risk group, despite only 49% experiencing severe prolonged neutropenia, highlighting that CAR-HT is a poor tool for distinguishing high- vs low-risk patients in B-ALL.

The authors then sought to refine CAR-HT for B-ALL, using a systematic and comprehensive approach. The NCI cohort comprised the training cohort. First, the association of each variable in CAR-HT with cumulative days of severe neutropenia was assessed individually, although evaluation of ferritin was limited by a high proportion (58%) of missing values. Next, the association of baseline BM disease burden with neutropenia was assessed, which was unsurprisingly very strong. Subsequently, the relevant CAR-HT variables were merged with disease burden (categorized as <5% blasts, 5% to 25% blasts, or ≥25% blasts) to form ALL-HT. From there, the authors optimized sensitivity and specificity to identify optimal cutoffs for the high- and low-risk groups. After optimization, ALL-HT clearly outperformed CAR-HT. Seventy-three (47%) patients were classified into the high-risk group, 79% of whom experienced prolonged, severe neutropenia. The high-risk group had a substantially longer period of severe neutropenia than the low-risk group (26 days vs 4 days, P < .0001). Finally, the score was validated in 2 unique data sets of patients with B-ALL---a pediatric and adolescent/young adult cohort at Seattle Children’s Hospital (n = 84) and a combined pediatric and adult cohort at Memorial Sloan Kettering (n = 106).

In addition to predicting early ICAHT, ALL-HT was associated with several key clinical outcomes: In the NCI cohort, high-risk patients were found to have lower deep remission rates, shorter event-free survival (EFS), and shorter overall survival (OS) than low-risk patients. The OS finding held in the validation cohorts, but the EFS finding only remained significant in patients at Memorial Sloan Kettering. The establishment of a predictive model for EFS and OS based on preinfusion factors would be clinically useful; however, it is important to recognize that BM disease burden is a significant component of the ALL-HT score and is also the factor most closely associated with EFS/OS independent of the ALL-HT score.⁶ Therefore, it is not entirely clear whether ALL-HT improves prognostication more than disease burden alone.

Although these findings are important, the study leaves several gaps to be addressed in future research. Early ICAHT is eloquently described, but late ICAHT is not addressed because most patients in the training cohort went to hematopoietic cell transplant or other therapy shortly after day +30. In patients treated with long-persisting CARs for B-ALL who do not proceed to alternative therapy post-CAR, our preliminary data from the CAR-multicenter analysis (CAR-MA) consortium demonstrates that late cytopenias occur frequently.⁷ Over 25% of the CAR-MA cohort had persistent grade ≥3 cytopenias at month 3, and 17% had persistent severe cytopenias at month 6. These late cytopenias may be more clinically relevant than early ICAHT, which can be managed with antimicrobial prophylaxis and transfusion support while patients are already hospitalized and/or receiving frequent follow-up. Thus, future studies should assess the predictive value of ALL-HT for late cytopenias. Factors associated with early cytopenias may not be the same ones predictive of late cytopenias. In addition, ALL-HT was developed exclusively in patients treated with investigational CARs. A strength of the study is its applicability to multiple CAR constructs, but the validation data sets also contained predominantly investigational CARs only available at single centers. Most children treated with CAR-T therapy for B-ALL receive tisagenlecleucel,⁸ which was only represented in a small proportion of one of the external data sets. It will be important for future studies to validate ALL-HT in patients treated with tisagenlecleucel and other US Food and Drug Administration (FDA)-approved CARs.

In conclusion, Nair et al demonstrate that early ICAHT in B-ALL is not the same as in lymphoma or multiple myeloma. The ALL-HT score that was developed with rigorous methodology refines the widely used CAR-HT score, makes it applicable for B-ALL and is easy to use. Future studies focused on later cytopenias and including large numbers of patients treated with FDA-approved CAR products will be critical to more comprehensively elucidating the scope and implications of hematotoxicity in B-ALL.

Conflict-of-interest disclosure: M.A.P. has participated on advisory boards for Novartis, Bluebird, Gentibio, Cargo, Pfizer, Autolus, and Garuda and received study support from Adaptive and Miltenyi. R.M.M. declares no competing financial interests.