AUTHORS
Paula Sulikowski, MD, Regina M. Myers, MD2, Yimei Li, PhD3, Kelly Diringer Getz, PhD, MPH4, Anna Sparrow5, Brynn Henry6, Kathlyn Phengchomphet7, Lucy E. Cain, MBBS FRACP8, Amanda DiNofia, MD9, Shannon L. Maude, MD10, Stephan A Grupp, MD, PhD11, Allison Barz Leahy, MD MSCE12, Caroline Diorio, MD13
AFFILIATIONS
1 Children's Hospital of Philadelphia, Philadelphia, United States, Children's Hospital of Philadelphia, Philadelphia, United States
2 Children's Hospital of Philadelphia, Philadelphia, United States
3 Children's Hospital of Philadelphia, Philadelphia, United States
4 Children's Hospital of Philadelphia, Philadelphia, United States
5 Children's Hospital of Philadelphia, Philadelphia, United States
6 Children's Hospital of Philadelphia, Philadelphia, United States
7 Children's Hospital of Philadelphia, Philadelphia, United States
8 Children's Hospital of Philadelphia, Philadelphia, United States
9 Children's Hospital of Philadelphia, Philadelphia, United States
10 Children's Hospital of Philadelphia, Philadelphia, United States
11 Children's Hospital of Philadelphia, Philadelphia, United States
12 Children's Hospital of Philadelphia, Philadelphia, United States
13 Children's Hospital of Philadelphia, Philadelphia, United States
ABSTRACT
Introduction: Cytokine release syndrome (CRS) is the most common and potentially life-threatening toxicity of CD19-directed chimeric antigen receptor T-cell (CART19) therapy in children and young adults (CAYA) with B-cell acute lymphoblastic leukemia (B-ALL). High disease burden (DB) at the time of CART19 infusion is the most established risk factor for severe CRS. However, not all patients with high DB experience severe CRS, suggesting other factors contribute to the risk. We hypothesized that baseline pro-inflammatory state and aggressive disease biology would be predictive of severe CRS. Identifying predictors may enable better risk stratification, earlier intervention and insights into biology associated with severe CRS.
Methods: We conducted a retrospective cohort study of CAYA with B-ALL and high DB treated with 4-1BB-based murine or humanized CART19 on 6 clinical trials or with commercial tisagenlecleucel at Children’s Hospital of Philadelphia. The primary endpoint was development of severe CRS defined as ASTCT Grade ≥3. We evaluated preinfusion patient, disease, treatment characteristics, laboratory values, Endothelial Activation and Stress Index (EASIX) and modified-EASIX (mEASIX) as possible predictors of severe CRS. Univariable and multivariable logistic regression models were developed to estimate odds ratios (OR) and confidence intervals (CI). The multivariable model included bone marrow blast percentage (BM blast %; selected a priori) and factors associated with severe CRS at P<0.20 in univariable analysis. Secondary outcomes included time to CRS onset, frequency of neurotoxicity, complete remission (CR), and 28-day mortality.
Results: During the study period, 504 patients were treated with CART19 for B-ALL. To determine a threshold to define high DB, we conducted a spline regression of the risk of severe CRS by DB. This model yielded an inflection point at ~25% marrow blasts. Based on the ≥25% blasts cutoff, 106/504 patients were classified as high DB and included in this analysis.
Median age at infusion was 10 years; 50% (n=53) were in first relapse, 36% (n=38) had undergone prior HCT, and 82% (n=87) were CART-naïve. Median BM blast % was 84% (IQR 50-92%). Overall, 46% (49/106) of patients developed severe CRS. 28-day mortality was 5.7% (6/106), including 2 patients who died of toxicity and 4 from disease progression. Day 28 CR rates were similar for no/mild vs severe CRS (74% vs 76%, P>0.99).
Severe CRS occurred more frequently in patients receiving their first CART therapy vs retreatment (53% vs 15%, Fisher's exact P=0.004). Conversely, severe CRS was less frequent in patients with prior HCT than those without prior HCT (29% vs 56%, P=0.009) and among patients in ≥2nd relapse compared with first relapse (33% vs 57%, P=0.053).
In multivariable analysis, first CART exposure was independently associated with severe CRS (OR 5.19, 95%CI 1.52-24.1, P=0.016) and prior HCT was associated with lower odds of severe CRS (OR 0.38, 95%CI 0.15-0.91, P=0.033). BM blast % ≥25% was not associated with severe CRS (OR 1.06 per 10% increase, 95%CI 0.89-1.25, P=0.5). Relapse number was not included in the multivariable model given collinearity with prior CART and HCT. No baseline clinical characteristics, laboratory markers, EASIX or mEASIX scores were associated with severe CRS.
Time to CRS onset (i.e., first fever) was substantially shorter in patients with severe CRS compared with no/mild CRS (median 1 vs 5 days, P<.001). Severe CRS occurred in 65% of patients with fever onset by day 4 post-infusion compared with 10% of those with later onset (RR 6.26, 95%CI 2.12-18.54, P<.001). Neurotoxicity (any grade) occurred more frequently in patients with severe CRS (83% vs 43%, P<.001).
Conclusions: Preinfusion high DB has been identified as the best predictor of severe CRS. We demonstrate that CAYA with ≥25% pre-infusion blasts have a ~50% risk of developing severe CRS, but disease burden above the 25% threshold does not further increase risk. In patients with high DB, first CART exposure was associated with a higher risk of severe CRS. Prior HCT independently identified patients at lower risk of severe CRS. These results suggest that prior exposure to CART or HCT may modify the inflammatory response to subsequent CART19 infusion. This hypothesis warrants mechanistic investigation. Early fever onset also identifies patients at increased risk for progression to severe CRS, with first fevers by day +4 conferring sixfold higher risk.
DISCLOSURES
Paula Sulikowski, MD: No relevant conflicts of interest to declare.