The development of chimeric antigen receptor T-cell (CAR-T) therapy has revolutionized the treatment landscape for multiple myeloma (MM). Thus, CAR-T therapy is considered one of the key therapeutic options, particularly for patients with relapse/refractory multiple myeloma (R/R MM). Among the principal CAR-T therapies worldwide are idecabtagene vicleucel¹ (ide-cel) and ciltacabtagene autoleucel (1, 2) (cilta-cel), both of which have demonstrated a favorable efficacy profile. However, the management of the complications associated with CAR-T remains a significant challenge. Notably, non-relapse mortality (NRM) is higher in multiple myeloma (8.0%) compared to other hematologic malignancies such as large B-cell lymphoma (6.1%) and indolent lymphoma (5.7%) (3). Inflammation-related complications, such as cytokine release syndrome (CRS) and Hemophagocytic Lymphohistiocytosis (HLH), account for up to 25% of NRM as reported by Sidana et al. Furthermore, the immunosuppressive treatments used to manage these complications increase the risk of infections, which represent the leading cause of death in these patients (4). Unfortunately, despite the existence of various consensus definitions of CAR-T-related HLH-like, the ambiguous terminology and lack of strict diagnostic criteria hinder accurate diagnosis (5–8).

Secondary HLH related to cellular therapy has gained increasing recognition in recent years with the expanding use of CAR-T. Initially identified through the classical clinical features of HLH, more recently proposed diagnostic frameworks have introduced related definitions, such as Immune Effector Cell-Associated Hemophagocytic Lymphohistiocytosis-Like Syndrome (IEC-HS) and Macrophage Activation Syndrome-like (MAS-like). Although awareness of this manifestation is increasing, it remains insufficiently characterized in clinical trials of anti-BCMA CAR-T therapy (1, 9). Nevertheless, MAS-like constitutes a relatively frequent complication of BCMA-directed CAR-T therapy, with prior studies reporting an incidence of approximately 20% (7, 10). Despite growing concern, the pathogenesis of MAS-like remains unclear. It is characterized by a distinct cytokine profile compared to CRS, with marked elevations of interferon-γ, granzyme B, interleukin-1 receptor antagonist (IL-1RA), and interleukin-10 (11). Some patients exhibit a biphasic expansion of CAR-T cells. The first peak is predominantly composed of PD1+ T effector cells, while the second peak is characterized by T-cell factor 1 positive (TCF1+) T effector cells, which are primarily involved in memory cell responses. This second expansion phase of CAR-T cells has been associated with the development of an uncontrolled inflammatory response (12).

Controlling the inflammatory response is a key challenge in CAR-T therapy for MM. Academic approaches to CAR-T cell therapy are offering novel insights aimed at enhancing safety, minimizing neurological toxicities, and enabling point-of-care treatments where the bench-clinic-bench time is reduced. For example, cesnicabtagen autoleucel (13) (ARI0002h), an academic anti-BCMA CAR-T developed by Hospital Clínic of Barcelona, is administered in three different fractions (10%, 30%, and 60% of total cells), on non-consecutive days. This methodology has been employed since the development of the anti-CD19 CAR T-cell product ARI0001, with the aim of reducing systemic complications. In this regard, a low incidence of high-grade adverse events and only mild and transient neurological toxicity has been observed in ARI0002h, with no late neurotoxicity (13, 14).

With these results from the multicentre clinical trial CARTBCMA-HCB-01, the approval for use under hospital exemption was granted in Spain (13–15). Our group aimed to investigate the incidence, clinical manifestations, predictive factors and prognosis of MAS-like in patients treated with ARI0002h.

Anti-BCMA CAR T-cell therapy has demonstrated a significant impact on the treatment of patients with MM; however, this novel therapeutic approach is associated with a distinct spectrum of immune-mediated adverse events, such as MAS-like. Notably, this complication is likely underreported in both clinical trials and real-world studies. In the present cohort of 80 patients treated with the anti-BCMA product ARI0002h, we observed an incidence of 15%, which was correlated with poorer treatment responses and reduced survival outcomes.

The diagnosis of secondary HLH/MAS-like remains challenging due to the absence of a definitive diagnostic test and the overlap of its clinical manifestations with other adverse events, such as CRS (21). Consequently, the incidence of MAS-like is not consistently reported in clinical trials and varies depending on the diagnostic criteria employed; many reports are based on retrospective analysis. Furthermore, both the type of tumour and the specific CAR T-cell product used are critical factors influencing the development of this complication. Anti-CD22-targeted CAR T-cell therapies have been associated with a notably high incidence of HLH-like reactions, with rates reaching up to 36% in certain cohorts (22, 23). Anti-CD19-based CAR T-cell therapies, in contrast, appear to be associated with a lower incidence of HLH/MAS-like (24). For instance, the reported incidence of HLH in real-world studies of axicabtagene ciloleucel (axi-cel), based on the HLH-2004 criteria, is up to 5% in patients with diffuse large B-cell lymphoma (DLBCL) (25).

MAS-like appears to be a relatively common entity observed in patients with MM undergoing CAR-T cell therapy. The incidence rates vary from 4% in the ide-cel clinical trial (1) to approximately 20% in certain real-world cohorts of patients treated with anti-BCMA CAR T-cell products (7, 10). As a potential pathophysiological explanation, the upregulation of major histocompatibility complex (MHC)-II expression on MM cells, triggered by IFN-γ, may contribute to the dysregulated immune response within the pro-inflammatory microenvironment following CAR T-cell infusion (26). The lower incidence of MAS-like, according to UCSF criteria, observed in ARI0002h (15%) compared to the 22% reported by Kennedy et al. (7) may be related to the initial fractionated infusion strategy and the withholding of treatment in the presence of adverse events. This hypothesis is supported by the reduced incidence of severe adverse events, such as CRS, in the few products administered in a fractionated schedule, including LCAR-B38M (LEGEND-2 study) and ARI0001 (CART19-BE-01), although differences among constructs should be taken into consideration (17, 27). Differences between MAS-like cases reported with other commercial CAR-T products and ARI0002h may be attributed to the distinct diagnostic criteria used. For this study, the UCSF consensus criteria were selected among other available definitions, owing to their clarity and well-defined requirements. Based on our findings, the UCSF criteria appear to be highly sensitive but may prone to overdiagnosing MAS-like. Nevertheless, our group posits that early diagnosis—prior to the onset of full organ failure and the severe manifestations of the syndrome—may enhance management and improve patient outcomes (28).

Based on our results, a higher tumour burden, as indicated by an increased serum monoclonal component, and a higher ISS at enrolment, were significantly more frequent in patients who developed MAS-like. Indeed, an ISS of III and elevated serum monoclonal component were associated with an increased risk of developing this complication. The involvement of high tumour burden has been previously identified in other hematologic malignancies (29, 30); however, MAS-like predisposing factors associated with CAR T-cell therapy have not been well characterized in multiple myeloma. These findings support the hypothesis of an uncontrolled inflammatory response activated by high tumour antigen concentrations (31). The prevalence of extramedullary disease prior to infusion demonstrated a borderline association (p = 0.057), although its potential role in MAS-like is supported by previous reports indicating that its presence increases the risk of CRS (32). In this context, extramedullary disease may elicit a MAS-like through its tumour microenvironment, which is enriched with M2 polarized tumour-associated macrophages (TAM) that secrete IL-6 (33, 34).

According to the literature, this complication typically manifests in a delayed manner following infusion, most often after the onset of CRS (7, 10, 35). Our findings support this observation, as MAS-like syndrome developed later (median of 10 days after the first infusion to meet UCSF criteria, compared to a median of 7 days for CRS), even when the product was administered in progressively increasing doses, as in the case of ARI0002h. The increase in ferritin levels >100 mg/L/h within a 24-hour period was the most sensitive parameter analysed. This finding supports the value of ferritin measurement when CAR-T inflammatory toxicity is suspected, as it may help to identify patients at risk of developing MAS-like syndrome. LDH appeared to meet the criteria more frequently than fibrinogen, and a significant poor prognosis was observed in patients who met all three criteria. Patients who met the UCSF criteria exhibited other clinical and laboratory features commonly associated with classical MAS, such as hypertransaminasemia, hypertriglyceridemia, and cytopenias (36, 37). Additionally, most patients showed histopathological findings consistent with MAS.

In line with previous reports, ARI0002h demonstrates favourable efficacy outcomes. The median PFS of the entire cohort was 20 months, with the median OS not reached, which is in line with the outcomes of currently approved CAR-T therapies for MM (1, 9). However, patients who developed MAS-like exhibited a poorer response rate (CR 25% vs. 68%) and impaired survival (median PFS 7 months vs. 21.4 months, median OS 18 months vs. not reached). The differences observed were more pronounced in patients who met all three criteria compared to those who met only two (median PFS 5.5 months vs. 11.8 months, median OS 7.8 months vs. 21.5 months). The better outcomes in patients who met only two UCSF criteria may support the necessity of early detection of MAS-like syndrome before all manifestations are fully developed. However, subgroup analyses are severely underpowered (12 events, 4 with UCSF = 3). Therefore, results should be interpreted as hypothesis-generating only. Furthermore, the contribution of a higher tumour burden to the poorer outcomes observed in these patients must be taken into account. Together with the limited sample size and the relatively low frequency of MAS-like as an adverse event, this may explain the loss of statistical significance in the multivariate analysis.

Our genetic analysis suggests a complex interplay between inherited genetic factors and the development of MAS-like in patients undergoing CAR-T therapy. Variants in UNC13D and PRF1, previously reported to impair perforin activity (20), were not statistically significant predictors on their own. However, when combined with high disease burden and extramedullary disease, they may increase the risk of MAS-like, as suggested by our clinical correlation analysis. These findings reinforce the multifactorial nature of MAS in the setting of advanced therapies such as CAR-T. However, the reduced sample size, together with the limited incidence of MAS-like, hampers the ability to draw definitive conclusions and may increase the influence of outliers.

The treatment of MAS-like syndrome requires the control of the inflammatory loop and close monitoring. There is a wide variety of drugs that are often employed without a standard algorithm. Although controversy exists in some cases, the most frequently used drugs are antibodies or recombinant proteins targeting interleukins, such as anakinra (IL-1) or tocilizumab (IL-6), systemic steroids, ruxolitinib (a JAK inhibitor), emapalumab (an IFN-γ inhibitor), and etoposide (31, 38–42). In this study, most patients who developed MAS-like received standard CRS management, which comprises anti-IL6, anti-IL1, and steroids. This is consistent with the initial diagnosis of CRS at the beginning of treatment. Noticeably, a half of patients required two or more drugs (mainly tocilizumab and dexamethasone). The limited use of anakinra may be related to its approval for IEC-HS treatment in our country only after July 2023, by which time most patients had already received ARI0002h. Nonetheless, these observations suggest that CRS and MAS-like syndrome share several features of systemic inflammation, in which prompt therapeutic intervention may be crucial to prevent the establishment of a positive feedback loop.

Our study has several limitations. This multicentre study involved a cohort of 80 patients treated with the same anti-BCMA CAR-T therapy, in a context of scarcity of studies focused on this complication. Nevertheless, the retrospective design and the limitation to a single country may introduce potential biases. Moreover, the analytical criteria employed may be insufficient for accurately differentiating between the hyperinflammation associated with CRS and MAS-like syndrome. Other criteria with different sensitivity and specificity could be explored prospectively in the future.

In summary, the development of MAS-like is associated with poorer responses and shortened PFS and OS in patients with MM treated with ARI0002h, especially in those who meet all three criteria: elevated ferritin, decreased fibrinogen, and increased LDH. Therefore, further research is required to establish diagnostic criteria that enable early identification and timely therapeutic management of MAS-like.