Biomarkers Accurately Predict Cytokine Release Syndrome (CRS) after Chimeric Antigen Receptor (CAR) T Cell Therapy for Acute Lymphoblastic Leukemia (ALL)

CAR T cells with anti-CD19 specificity have demonstrated considerable promise against highly refractory hematologic malignancies. Dramatic responses with complete remission rates as high as 90% have been reported in patients (pts) with relapsed/refractory ALL treated with CTL019 (Maude et al., NEJM 2014). Marked in vivo CAR T cell proliferation (100 to 100,000x) leads to improved efficacy but can be associated with adverse events, including cytokine release syndrome (CRS). To better understand manifestations of CRS, we studied clinical, laboratory, and biomarker data of 39 children and 12 adults with relapsed/refractory ALL treated with anti-CD19 CAR T cells.

T cells were lentivirally transduced with a CAR composed of anti-CD19 single chain variable fragment/4-1BB/CD3 (Porter, NEJM 2011). 43 cytokines, chemokines, and soluble receptors (collectively termed cytokines hereafter) were serially measured, using Luminex bead array. Other biomarkers were tested in a CLIA/CAP certified lab.

48 of 51 pts developed grade 1-5 CRS (CRS1-5) (see Table). Most pts developed mild (grade 1-2) to moderate (grade 3) CRS (34/51). 14 pts developed severe (grade 4-5) CRS (12 grade 4, 2 adults with grade 5). 21 pts were treated with the IL-6 inhibitor tocilizumab, and most had rapid marked clinical improvement in CRS evidenced by quick resolution of fever and weaning of vasoactive medications.

We found peak levels of 24 cytokines, including IFNg, IL6, IL8, sIL2Ra, sgp130, sIL6R, MCP1, MIP1a, and GM-CSF during the first month after CTL019 infusion were highly associated with CRS4-5 compared to CRS0-3, significant by the Holm-Bonferroni adjusted p-value. Analyzing cytokines from the first 3 days after infusion, sent before patients developed severe CRS, only 2 cytokines, sgp130 and IFNg, were strongly associated with later development of severe CRS (p<0.001) and significant by Holm-Bonferroni. With a 3 variable regression model, found by forward selection, we accurately predicted which pts developed severe CRS using IFNg, sgp130, and IL1Ra (PPV 75%, NPV 94%, sensitivity 86%, specificity 89%, AUC=0.93). For the pediatric cohort, the modeling was even more accurate; a combination of IFNg, IL13, and MIP1a had PPV 92%, NPV 100%, sensitivity 100%, and specificity 96% (AUC=0.98). In the pediatric cohort only, a bone marrow aspirate was collected immediately prior to infusion. We found disease burden was associated with CRS severity but did not improve the predictive accuracy of the models over the cytokines alone. A combination of sgp130, IFNg and disease burden yielded PPV 77%, NPV 96%, sensitivity 91%, and specificity 88% (AUC 0.95). We are validating our models in an additional cohort and will present those data.

The 1-month peak of several clinical laboratory tests were strongly associated with severe CRS, including CRP, ferritin, LDH, AST, and BUN; however, they were not predictive of severe CRS. Some of these, including CRP, had a good NPV for early prediction but none had a good PPV. We hypothesized and demonstrated that children with severe CRS develop clinical and laboratory manifestations similar to macrophage activation syndrome (MAS)/hemophagocytic syndrome (HLH), including hyperferritinemia (>10,000ng/ml), splenomegaly, and hypofibrinogenemia. Of the tested cytokines, 18 have been previously studied in children with HLH. We found a near identical pattern of cytokines differentially elevated in HLH also elevated in pts with CRS4-5 compared with CRS0-3. IL6, sIL6R, and sgp130 were markedly elevated in pts with CRS4-5; this IL6 cytokine pattern, along with the pronounced response to tocilizumab, establishes that IL6 trans-signaling is clinically relevant.

These data represent the largest and most comprehensive profiling of the clinical and laboratory manifestations of CAR T cell related CRS and provide novel insights into CRS biology. They represent the first data that can accurately predict which pts treated with CAR T cells have a high probability of becoming critically ill. These data have direct therapeutic relevance and may guide future cytokine directed therapy.

The first 4 authors contributed equally.