Pre-Infusion Neurofilament Light Chain (NfL) Levels Predict the Development of Immune Effector Cell-Associated Neurotoxicity Syndrome (ICANS) - a Multicenter Retrospective Study

Chimeric antigen receptor (CAR) T-cell therapy is often associated with neurological complications termed immune effector cell-associated neurotoxicity syndrome (ICANS). There remains a critical need to identify patients most at risk for ICANS. Yet a biomarker for the development of ICANS is lacking. This retrospective multicenter study evaluates pre-infusion levels of plasma neurofilament light chain (NfL), a marker of neurodegeneration, as a predictive biomarker the development of ICANS.

Study inclusion criteria included available pre-infusion (up to 4 weeks prior to lymphodepletion) plasma from patients treated with a CAR T cell therapy (n = 30, 36% with ICANS, ASTCT consensus ICANS grade range 1-4). Exclusion criteria included confounding diagnoses known to elevate NfL levels (e.g. dementia, recent stroke). Plasma NfL was assayed using a Simoa HD-X kit (Quanterix ^TM). Demographic (age, sex), oncologic (primary, stage, mean tumor volume (MTV), and history of central nervous system (CNS) involvement), and medical history (history of non-oncologic CNS disease or neuropathy) were obtained from the medical record. MTV was derived from total lesion burden on pre-infusion positron emission tomography (PET) scans using a 41% maximum standard uptake value (SUV) threshold. Pre-infusion (i.e. during lymphodepletion) and Post-infusion Day 1 (D1) platelet count, C-reactive protein (CRP), fibrinogen, lactate dehydrogenase (LDH), and ferritin levels were also obtained from the medical record. Group comparisons used log-rank testing, followed by receiver operating characteristic (ROC) curve classification and hierarchical clustering. Validation testing used a 10,000 fold cross-validation on 80% of the data. Finally, demographic and clinical characteristics correlated with pre-infusion biomarkers using point-biserial and Spearman (rank) correlation.

Our results demonstrated that individuals who would go on to develop ICANS had elevations in pre-infusion NfL ([87.6 v 29.4 pg/ml; Fig 1A], p = 0.00004) with excellent classification accuracy for the development of ICANS (AUC 0.96; Fig 1B), sensitivity (0.91) and specificity (0.95). NfL further correlated with ICANS development (r = 0.74, p < 0.0001; Fig 1C). Among known post-infusion risk factors, D1 ferritin had the highest classification accuracy, but was inferior to baseline NfL (p < 0.05; Fig 1B). Both baseline NfL and D1 ferritin elevations clustered with ICANS grade (Fig 1D).

Our findings show that pre-infusion plasma NfL levels are a robust early marker for the development of ICANS that exceeds known post-infusion markers. This suggests the risk of developing ICANS reflects pre-existing latent neuroaxonal injury. Predictive identification of patients at risk of developing ICANS prior to cellular infusion would permit early, preemptive or prophylactic ICANS-directed therapies, thereby improving patient outcomes.

Figure 1: Baseline (pre-infusion) NfL Levels in ICANS (A) Baseline (pre-infusion) levels of NfL in patients who develop Grade 0 ICANS, Grade 1-2 ICANS, and Grade 3+ ICANS. (B) Receiver operating characteristic curve classification of patients who developed any grade ICANS (1+) vs grade 0 for baseline NfL and post-infusion day 1 (D1) markers. (C) Correlation between pre-treatment factors and pre-infusion biomarkers. All significant relationships after correction for multiple comparisons using false discovery rate (FDR) are outlined (*). (D) Hierarchical clustering of age, baseline NfL, and D1 markers. Clusters associating with ICANS are labeled in red, while those associating with cytokine release syndrome (CRS) are in blue.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal