Summary: T cells genetically modified with CD19 chimeric antigen receptors have produced impressive clinical responses in patients with refractory B-cell malignancies, but therapeutic responses are often accompanied by cytokine release syndrome (CRS), which can cause significant morbidity and mortality. Teachey and colleagues have identified predictive biomarkers for this complication that may allow testing of earlier intervention with agents such as the IL6 receptor blocker tocilizumab to evaluate whether CRS can be ameliorated without jeopardizing clinical responses. Cancer Discov; 6(6); 579–80. ©2016 AACR.
See related article by Teachey et al., p. 664.
The infusion of T cells genetically modified with chimeric antigen receptors (CAR) with specificity against CD19 after lymphodepleting chemotherapy has been one of the major success stories in the field of immuno-oncology. Several groups have reported high remission rates in heavily pretreated patients with refractory B-cell malignancies (1–5); however, significant side effects have accompanied these impressive clinical responses. Antitumor activity is dependent on T-cell activation and proliferation, which results in a systemic inflammatory response known as cytokine release syndrome (CRS). CRS can range from a mild syndrome requiring minimally invasive supportive care to a more severe systemic response necessitating intensive interventions, such as vasopressor support and mechanical ventilation (1–5). In some cases, CRS can present with a constellation of symptoms indistinguishable from hemophagocytic lymphohistiocytosis (HLH) or macrophage activation–like syndrome (MAS). Although CRS has rarely been reported after the infusion of T cells that recognize antigens through their native receptor (6), it has been consistently observed with a frequency of up to 90% in all the successful trials of CD19 CAR–modified T cells despite variation in constructs (single-chain variable fragment and costimulatory moieties), vectors, T-cell manufacturing methods, and lymphodepletion regimens. In addition, a subset of patients treated with CD19 CAR–modified T cells experienced neurologic toxicity that may be concurrent with CRS or may occur as a distinct entity after CRS has resolved (5).
As the number of patients treated with CD19 CAR therapy continues to rise, a major research focus has been to identify characteristics and biomarkers that accurately predict whether patients will develop severe CRS. Algorithms have been developed and widely adopted to define severity and guide management (7), but the ability to predict which patients are more likely to develop severe CRS would be valuable, allowing us to test the role of earlier intervention. One consistent risk factor has been disease burden, and most groups are now using lower doses of T cells in patients with more extensive disease. Although the biology of CRS is incompletely understood, measurement of serum cytokines in symptomatic patients has enabled the identification of a group of cytokines deemed responsible for the defining symptoms of CRS. The marked elevation of IL6 in patients with CRS led to the institution of successful targeted therapy for the treatment of CRS using IL6 receptor blockade (tocilizumab; refs. 7, 8). Measuring IL6, however, is a research tool not rapidly available as a Clinical Laboratory Improvement Amendments–approved test in most centers. Therefore, Davila and colleagues' observation that levels of C-reactive protein (CRP), for which there is a rapidly available test, correlated with CRS severity was a valuable contribution that has since been widely adopted (2). Davila and colleagues showed that significant differences in CRP in patients who developed severe CRS were evident as early as 2 days after infusion and correlated with serum IL6 levels (2).
In a study published in this issue, Teachey and colleagues sought to address the need for effective predictive biomarkers for CRS by using regression modeling to identify factors predicting patients more likely to develop severe CRS before they become critically ill (9). They measured the levels of cytokines and other clinical biomarkers in 51 patients, the majority of whom (39 patients) were pediatric and treated with the CD19 CAR CTL019 containing the 4-1BB costimulatory moiety. Forty-eight of the 51 patients developed CRS, which was grade 1–2 in 16, grade 3 in 14, and grade 4–5 in 14. The authors reported peak levels of 24 cytokines (including IFNγ, IL6, sgp130, and sIL6R) in the first month after infusion that were highly associated with CRS. In pediatric patients, the modeling analyses were highly accurate and, in a forward-selected logistic regression model, including IFNγ, IL13, and MIP1, both highly sensitive at 100% and specific at 96% (9).
They validated this predictive cytokine signature in an independent cohort of 12 pediatric patients, accurately predicting which patients would develop severe CRS. Of note and in contrast to previous reports (2, 5), although peak serum CRP and ferritin were higher in the majority of patients with grade 4–5 CRS, they failed to predict the development of severe CRS.
Despite the success of therapy targeting IL6 in treating CRS, the authors did not find an appreciable rise in IL6 prior to development of CRS and concluded that measurement of serum IL6 early after infusion of CD19 CAR–modified T cells is of no benefit (9). This observation is in contrast to the findings of Turtle and colleagues (5), who not only found elevated levels of IL6 (along with IFNγ and TNFα) as early as 1 day following infusion, but also noted higher IL6 levels in patients who developed higher-grade CRS and also neurotoxicity. Turtle and colleagues also reported that high CAR T-cell doses and larger tumor burden contribute to an increased risk of severe CRS and neurotoxicity (5), which could be mitigated by adoption of a risk-stratified approach where dosing was adjusted based on marrow disease burden. Their stratified approach lowered ICU admissions for both CRS and neurotoxicity (5).
Both Turtle and colleagues and Teachey and colleagues highlight the importance of developing algorithms that accurately identify patients more likely to develop severe CRS, so we may intervene early and test strategies to lower their risk. Although the predictive biomarkers identified by Teachey and colleagues were validated in a second pediatric cohort, it is not yet clear whether the results can be extrapolated to adult patients receiving the same product and to other studies with different CD19 CAR constructs and T-cell–manufacturing regimens. It seems reasonable to surmise that, just as there are differences in the kinetics of peak expansion and persistence with the different CD19 CAR-modified T-cell products currently being tested in the clinic, there may be differences in biomarkers predictive of severe CRS. Indeed, the results from Turtle and colleagues, who are also using a lentiviral construct with the 4-1BB costimulatory domain, support the contention that factors such as the CAR construct, T-cell manufacturing, conditioning regimen, and age may influence the biological features of CRS and thus its predictive markers.
One important question still to be evaluated is whether earlier intervention with tocilizumab might reduce the morbidity and mortality from CRS without jeopardizing the impressive clinical responses of CD19 CAR-modified T cells. It is unclear whether, if validated, the use of rapid, real-time serum cytokine analysis would be feasible as a clinical decision-making guide in the larger number of centers currently participating in licensing trials of CD19 CAR-modified T cells, or whether a clinical manifestation of CRS such as fever reading would be a better guide for intervention. As Teachey and colleagues suggest, prospective trials initiating early intervention based on cytokine profile models or clinical findings will need to be implemented cautiously in order to ensure early intervention strategies do not prematurely abrogate antitumor response, especially given the long half-life of tocilizumab.
Disclosure of Potential Conflicts of Interest
R.H. Rouce has received speakers bureau honoraria from the Novartis Treatment Advisory Board Landscape Meeting and is a consultant/advisory board member for the same. H.E. Heslop reports receiving commercial research support from Celgene and has ownership interest (including patents) in ViraCyte and patents with Cell Medica.
The authors are supported by Lymphoma Research Foundation grant 337548 (R.H. Rouce), the American Society of Hematology Harold Amos Foundation (R.H. Rouce), NCI SPORE P50 CA126752 (H.E. Heslop), NCI:PPG PO1CA094237 (H.E. Heslop), Leukemia and Lymphoma Society SCOR 7018-04 (H.E. Heslop), and CPRIT RP110553-C1 (H.E. Heslop).
- ©2016 American Association for Cancer Research.