Introduction

Cytokine release syndrome (CRS) and immune effector cells associated neurotoxicity syndrome (ICANS) are common CAR T cells toxicities. Various grading systems that attribute different grades of severity to symptoms are currently used, preventing comparisons of different products and possibly leading to management implications.

Methods

We included 53 patients (pts) with B acute lymphoblastic leukemia (B-ALL) treated with 1928z CAR T cells (NCT01044069) and 49 pts with diffuse large B cell lymphoma (DLBCL) who received axicabtagene ciloleucel (axi-cel, n=36) or tisagenlecleucel (tisa-cel, n=13) after FDA approval. CRS and ICANS were retrospectively graded according to the new ASTCT consensus grading system, for comparison with other available systems: Lee, Penn, Memorial Sloan Kettering Cancer Center, CARTOX and CTCAEv5.0 for CRS; CTCAEv4.03 and CARTOX for ICANS. We then used ASTCT grades to predict management according to current guidelines for DLBCL pts (axi-cel and tisa-cel insert packages, CARTOX and NCCN guidelines) and compare it to the actual treatment received at our center.

Results

According to the ASTCT grading, B-ALL pts had 87% CRS (28% grade ≥3), while DLBCL pts had 86% CRS (14% grade ≥3) with axi-cel, and 54% (no grade ≥3) with tisa-cel. B-ALL pts had 55% ICANS (45% grade ≥3), while DLBCL pts had 55% ICANS (33% grade ≥3) with axi-cel, and 15% with tisa-cel (no grade ≥3) (fig. 1). When comparing grading systems, agreement on CRS and ICANS diagnosis was found in 99% and 91% cases, respectively. However, when evaluating toxicities grade by grade, only 27% pts had the same grade in each system for CRS, and 55% for ICANS (fig. 2).

When predicting management for DLBCL pts, we found some relevant differences across current guidelines (fig. 3). At our center 58% and 13% of patients with CRS received tocilizumab (toci) and steroids, respectively, similarly to what predicted according to axi-cel's, CARTOX and NCCN guidelines, but differently from tisa-cel's label (10% and 5%). For ICANS, while tisa-cel's label does not provide any recommendations, other guidelines-based predictions were mostly overlapping. Indications for treatment vary across current guidelines, which were developed on single products and different grading systems, thus, should not be universally applied. This discrepancy becomes particularly relevant for cases with discordant CRS/ICANS grades. As such, we gave toci to 66% pts upgraded to grade 3 by Penn (grade 2 by ASTCT), while they would receive both toci and steroids according to axi-cel/CARTOX/NCCN guidelines, but would not be treated according to tisa-cel's.

Conclusions

Different grading systems provide inconsistent CRS/ICANS scores. To avoid discrepancies in assessing and managing toxicities of different products, a unified grading should be used and paired management guidelines with product-specific indications should be developed.

中文翻译：

---

比较汽车T细胞毒性分级系统：Astct分级系统的应用及其对管理的启示

介绍

细胞因子释放综合征（CRS）和免疫效应细胞相关的神经毒性综合征（ICANS）是常见的CAR T细胞毒性。当前使用了将等级的严重程度归因于症状的各种分级系统，从而阻止了不同产品的比较，并可能导致管理上的隐患。

方法

我们纳入了53例接受1928z CAR T细胞（NCT01044069）治疗的B急性淋巴细胞性白血病（B-ALL）的患者和49例接受阿昔卡巴坦基因细胞纯化的弥漫性大B细胞淋巴瘤（DLBCL）患者（axi-cel，n = 36 ）或tisagenlecleucel（tisa-cel，n = 13），经FDA批准。根据新的ASTCT共识评分系统对CRS和ICANS进行了回顾性评分，以便与其他可用系统进行比较：Lee，Penn，Sloan Kettering纪念癌症中心，CRS的CARTOX和CTCAEv5.0；CTCAEv4.03和用于ICANS的CARTOX。然后，我们根据当前DLBCL pts指南（axi-cel和tisa-cel插入包装，CARTOX和NCCN指南）使用ASTCT等级来预测管理，并将其与我们中心接受的实际治疗进行比较。

结果

根据ASTCT分级，B-ALL评分的CRS为87％（28％≥3），而DLBCL评分的86％的CRS（14％≥3）和axi-cel为54％（无评分≥3）。和蒂莎 B-ALL pts具有55％的ICANS（45％≥3），而DLBCL pts具有55％的ICANS（33％≥3），含axi-cel和15％的tisa-cel（无≥3）（图1）。比较评分系统时，分别在99％和91％的病例中对CRS和ICANS的诊断达成了共识。但是，在逐级评估毒性时，对于CRS，每个系统中只有27％的患者具有相同的等级，对于ICANS，只有55％的患者具有相同的等级（图2）。

在预测DLBCL pt的管理时，我们发现当前指南之间存在一些相关差异（图3）。在我们中心，分别有58％和13％的CRS患者接受了tocilizumab（toci）和类固醇治疗，与根据axi-cel's，CARTOX和NCCN指南预测的结果相似，但与tisa-cel的标签不同（分别为10％和5％） ）。对于ICANS，虽然tisa-cel的标签未提供任何建议，但其他基于指南的预测大多重叠。在当前指南中，治疗指征有所不同，这些指南是针对单一产品和不同分级系统开发的，因此不应被普遍采用。对于CRS / ICANS等级不一致的情况，这种差异尤为重要。因此，我们将Penn提升至3级（ASTCT为2级）的TOCI降至66％

结论

不同的评分系统提供的CRS / ICANS评分不一致。为了避免在评估和管理不同产品的毒性方面出现差异，应使用统一的等级，并应将管理指南与产品特定的适应症配对。