German CARs keeping pace

In this issue of Blood, Bethge et al¹ present findings from the analysis of 356 patients treated with axicabtagene ciloleucel (axi-cel) or tisagenlecleucel (tisa-cel) chimeric antigen receptor (CAR) T-cell therapy for large B-cell lymphoma (LBCL) in a non-trial setting in 21 centers throughout Germany with a median follow-up of 11 months. While corroborating the excellent results of these therapies in a real-world setting with an overall response rate (ORR) of 65% and 12-month overall survival (OS) of 52%, they also add important insights into this growing standard-of-care literature. Their analysis is the first to report response to, rather than use of, bridging therapy as a predictor of outcome. Additionally, they report an adjusted 12-month nonrelapse mortality (NRM) of 5.5% in all patients, most occurring late (67%) and due to infection (62%). While their analysis invites a comparison between axi-cel and tisa-cel, differences between the patients being selected for, and the sites offering, each product result in biases that should preclude such a comparison.

This German study confirms similar ORR and OS outcome data from prior real-world studies (see table). Compared with pivotal studies of axi-cel² and tisa-cel,³ the authors report a lower progression-free survival (PFS), which may reflect practices specific to the delivery of these therapies in Germany. Some of this heterogeneity in logistics, patient selection, and management could explain the higher rate of high-grade cytokine release syndrome (CRS) seen with tisa-cel (13%) and lower rate of high-grade immune effector cell associated neurologic syndrome (ICANS) seen with axi-cel reported here compared with previous real-world series (see table). Differences in toxicity incidence may reflect clinical heterogeneity of real-world subjects as well as differences in patient selection and the diagnosis/management of these toxicities between centers. Several centers offered only a single product, and differences between each center regarding eligibility criteria, referral patterns, and toxicity mitigation strategies could account for differences seen between products. Remarkably, despite these potential differences, efficacy outcomes for axi-cel and tisa-cel remain consistent with numerous other studies (see table), further confirming the transformative impact of these therapies for patients.

This study is the first of its kind to examine response to bridging and its association with outcome after axi-cel and tisa-cel. Prior studies have shown use of bridging therapy to negatively impact both OS and NRM, the latter largely through infections. However, in these series, the predominant bridging therapy was myelosuppressive chemotherapy in an already chemorefractory patient population given an absence of more effective options.⁴ Bethge et al incorporate newer bridging options and report nonresponse, rather than use, as a predictor of negative outcome. With the increasing use of non-chemotherapy–based bridging options with activity in chemorefractory lymphoma and less associated myelosuppression (ie, polatuzumab and/or radiation), the negative associations between bridging therapy and survival may no longer be relevant. These findings, that response to rather than use of bridging is what matters, indicate that previous associations of bridging therapy with adverse CAR-T treatment outcomes may be skewed by ineffective and immunosuppressive bridging regimens. Future prospective studies incorporating modern bridging modalities are needed to further investigate this hypothesis.

Defining late infectious toxicity after CAR T-cell treatment and its impact on survival is an additional unique analysis offered by this study. While B-cell aplasia and prolonged lymphopenia and cytopenias are increasingly appreciated after CD19 CAR T-cell therapy, their impact on infectious risk and survival has been poorly defined. In this study, nearly two-thirds of the 5.5% NRM occurs late, after day 28, and most of those deaths are due to infection. Risk factors include prolonged neutropenia and high-grade ICANS, which raises important questions regarding the acute and long-term management and monitoring of these patients. High-grade ICANS may have been associated with increased steroid dosing, setting patients up for late infections; if true, more judicious use of steroids for high-grade ICANS would be warranted. Additionally, immunosurveillance may need to be more prescriptive, and infection prophylaxis and monitoring may need to be extended, when applicable, to address these risks.

Finally, this study notes some key differences between the two products with respect to efficacy and toxicity. In particular, ORR, complete response rate, and 12-month PFS were superior for axi-cel when compared with tisa-cel (74% vs 53%; 42% vs 32%; 35% vs 24%, respectively). On multivariable analysis, treatment with tisa-cel was associated with an inferior PFS. Tisa-cel, on the other hand, was associated with a lower incidence of CRS overall but not high-grade CRS, and ICANS overall as well as high-grade ICANS, and 12-month NRM (3.5% vs 10.4%). Given the lack of difference seen in OS, one could argue that the improved efficacy seen with axi-cel could be deprioritized over the improved late toxicity profile seen with tisa-cel. We strongly caution against such a comparison and interpretation, however. First, in light of new effective but not definitive therapies for cases of multiply relapsed LBCL, 11 months may not be long enough to assess an OS difference among patients with relapsing LBCL. Second, and as noted by the authors, there are differences in the populations treated with axi-cel and tisa-cel based on patient characteristics that have previously been shown to be correlates of poor response to therapy as well as presumed differences in clinical practice ranging from patient selection to toxicity management and surveillance among sites that treat with only a single product, leading to biases that cannot be controlled for or defined.^4,5

In summary, Bethge et al present a well-designed, large, multicenter, retrospective study of real-world use of axi-cel and tisa-cel in LBCL in Germany. Their data add novel insights to the field, particularly with regards to response to bridging therapy as a predictor of outcome and characterization of late infectious complications. These data highlight the need for prospective studies of novel bridging regimens as well as investigation into methods to mitigate late infectious mortality. We caution against drawing conclusions regarding the differences between axi-cel and tisa-cel outcomes given inherent biases outlined above; however, the data here once again recapitulate the previously described clinical activity of these products in LBCL.