Association between Post-CART Terminal Complement Complex Levels and Immune Effector Cell-Associated Neurotoxicity Syndrome (ICANS)

Background:

The cornerstone of ICANS pathogenesis is endothelial dysfunction leading to blood-brain barrier disruption driven by inflammatory cytokines. Endothelial dysfunction is often associated with complement activation as in post-transplant thrombotic microangiopathy, but the association between complement and ICANS has not yet been investigated. Herein, we describe the association between terminal complement complex (TCC, sC5b-9) levels and ICANS after CART.

Methods:

We retrospectively included 50 patients (pts) with B-cell non-Hodgkin lymphoma treated with axi-cel or brexu-cel at The Ohio State University who had had sC5b-9 levels available pre-lymphodepletion (pre-LD, between D-7 - D-4), pre-CART (between D-3 - D0) and post-CART infusion (between D5 - D8). sC5b-9 levels and other biomarkers were measured once in each of these timepoints. Then, we performed an exploratory study examining the association between TCC levels and clinically significant ICANS, defined as peak grade ≥2 (G≥2). Endothelial activation stress index (EASIX) and modified EASIX (m-EASIX) were calculated based in the original publications. Circulating CART levels were measured by flow cytometry.

Results:

The median age of our study population was 63 years (IQR: 53 – 65). Among them, 39 (78%) pts had large B-cell lymphoma and 7 (14%) had follicular lymphoma treated with axi-cel (46, 92%), and 4 (8%) had mantle cell lymphoma treated with brexu-cel. All pts received standard LD with fludarabine and cyclophosphamide, and 32 (64%) received prophylactic steroids. Following CART, 19 (38%) pts experienced ICANS, being peak G1 in 5 (10%), G2 in 8 (16%), G3 in 3 (6%), G4 in 3 (6%) pts, and no G5 cases.

First, we compared sC5b-9 levels pre-LD, pre-CART and post-CART between pts that had ICANS peak G≥2 (N = 14, 28%) vs. G0-1 (N = 36, 72%) and we found that the post-CART sC5b-9 levels were significantly higher among pts who had ICANS G≥2 [median (IQR) 274 (191 – 320) vs. 186 (137 - 254) ng/mL, p = 0.024]. However, there were no significant differences in sC5b-9 levels pre-LD and pre-CART between pts who had ICANS G≥2 vs. G0-1, median (IQR) 203 (161 – 256) vs. 196 (127 – 241) ng/mL, p = 0.18, and 159 (124 – 195) vs. 144 (122 - 192) ng/mL, p = 0.73, respectively.

Next, in a matched paired analysis we found that sC5b-9 levels significantly increased from pre-CART to post-CART infusion in both pts with ICANS G≥2 and G0-1. The median (IQR) increase was +59 (+43 - +96) ng/mL, p = 0.001, and +40 (-20 - +86) ng/mL, p = 0.013, in pts with ICANS G≥2 and G0-1, respectively. Interestingly, in a matched paired analysis from pre-LD to pre-CART, only pts with ICANS G≥2 had a significant difference in the sC5b-9 levels (median: -28, IQR: -43 - -6 ng/mL, p = 0.002) vs. (median: -1.5, IQR: -60 - +22 ng/mL, p = 0.23) in ICANS G0-1.

Moreover, there was no significant differences in the post-CART sC5b-9 levels between pts who received prophylactic steroids or not. However, the sC5b-9 levels were numeric higher with a trend of statistical significance in who developed ICANS G≥2 with prophylactic steroids (N = 9) compared to who developed ICANS G≥2 without receiving prophylactic steroids (N = 5), median (IQR) 320 (203 - 393) vs. 191 (155 – 274) ng/mL, respectively, p = 0.076.

Then, we examined the correlation between post-CART sC5b-9 levels with other biomarkers: pre/post-CART ferritin, post-CART fibrinogen, pre-LD EASIX, D+3 and D+7 m-EASIX, D+7 CART count and D+28 platelet count. Among them we only found significant correlation between post-CART sC5b-9 levels with D+7 CART count (r = 0.40, p = 0.027) and D+7 m-EASIX (r = 0.57, p = 0.0001). Also, we compared the level of the above biomarkers between pts with ICANS G≥2 and G0-1, but there were no significant differences in pre/post-CART ferritin, post-CART fibrinogen, pre-LD EASIX, D+3 and D+7 m-EASIX between pts with ICANS G≥2 vs. G0-1.

Lastly, in a multivariable logistic regression modeling including post-CART sC5b-9 and D+7 CART levels, both were statistically significant associated with ICANS G≥2. The incidence rate ratio (95% CI) per 100 units increase were 1.30 (1.03 – 1.64), p = 0.027, and 1.14 (1.02 – 1.28), p = 0.024, for sC5b-9 levels and D+7 CART cells/mm³, respectively.

Conclusions:

Our findings support a possible association between complement activation after CART and clinically significant ICANS warranting further investigation of the role of complement activation in the ICANS pathogenesis, and as both biomarker and therapeutic target.