Functional connectivity MRI provides an imaging correlate for chimeric antigen receptor T-cell-associated neurotoxicity

Abstract Background Treatment of hematological malignancies with chimeric antigen receptor modified T cells (CART) is highly efficient, but often limited by an immune effector cell-associated neurotoxicity syndrome (ICANS). As conventional MRI is often unremarkable during ICANS, we aimed to examine whether resting-state functional MRI (rsfMRI) is suitable to depict and quantify brain network alterations underlying ICANS in the individual patient. Methods The dysconnectivity index (DCI) based on rsfMRI was longitudinally assessed in systemic lymphoma patients and 1 melanoma patient during ICANS and before or after clinical resolution of ICANS. Results Seven lymphoma patients and 1 melanoma patient (19–77 years; 2 female) were included. DCI was significantly increased during ICANS with normalization after recovery (P = .0039). Higher ICANS grades were significantly correlated with increased DCI scores (r = 0.7807; P = .0222). DCI increase was most prominent in the inferior frontal gyrus and the frontal operculum (ie, Broca’s area) and in the posterior parts of the superior temporal gyrus and the temporoparietal junction (ie, Wernicke’s area) of the language-dominant hemisphere, thus reflecting the major clinical symptoms of nonfluent dysphasia and dyspraxia. Conclusions RsfMRI-based DCI might be suitable to directly quantify the severity of ICANS in individual patients undergoing CAR T-transfusion. Besides ICANS, DCI seems a promising diagnostic tool to quantify functional brain network alterations during encephalopathies of different etiologies, in general.

• Resting-state functional MRI (RsfMRI)-based dysconnectivity index (DCI) retraced clinical immune effector cell-associated neurotoxicity syndrome (ICANS) course in each individual patient (ie, it was elevated in ICANS and decreased when symptoms resolved).
• DCI was significantly associated with ICANS grading.
• DCI depicted the most profound connectivity disruptions in cerebral hubs for language processing and multistage actions, resembling the core symptoms of ICANS.
Chimeric antigen receptor T cells (CAR T) represent a powerful new class of adoptive immunotherapy for the treatment of relapsed or refractory hematological malignancies and also solid cancer entities. 1However, its success is hampered by immune-mediated adverse events, including immune effector cell-associated neurotoxicity syndrome (ICANS).ICANS affects up to two-thirds of such patients, clinically often manifesting with nonfluent dysphasia and dyspraxia for multistage actions, without structural lesions on conventional MRI. 2,3The exact pathophysiology of ICANS remains puzzling and, among other aspects, the extent to which ICANS is associated with disruptions of dedicated neuronal networks is yet to be determined.Based on resting-state functional MRI (rsfMRI), we recently developed an imaging tool, called the dysconnectivity index (DCI), which allows for the reliable quantification of functional connectivity disruptions in individual patients by comparing the connectivity profile of each voxel of an individual patient to a reference profile of healthy controls. 4In a recent study in glioma patients this approach revealed network alterations in distant, nonlesional brain tissue.The DCI, as a quantitative measure of these alterations in each individual patient, was correlated with underlying tumor biology and cognitive performance. 4n this proof-of-concept study, we utilized rsfMRI to analyze whether the DCI is increased in ICANS and associated with ICANS grade.We furthermore aimed to depict cortical brain regions that might be predominantly affected by ICANS-associated connectivity alterations.

Material and Methods
The sample consisted of 8 patients (19-77 years; female: 2; 7 right handed), among whom 7 had systemic lymphomas, and 1 had metastasized melanoma.All of these patients developed ICANS after the administration of CAR T cells (see Table 1).None of the patients had antecedent or present disease involvement of the nervous system.For further clinical characterization of the sample, please refer to Table S1.ICANS was graded according to current guidelines (ASTCT). 2 3 Tesla structural and rsfMRI according to the protocol of the Brain Genomics Superstruct Project (GSP)5 was performed in each patient at 2 different time points: at baseline before CAR T-transfusion (pre-ICANS, n = 3) or after clinical recovery from ICANS (post-ICANS, n = 5) and during maximum ICANS (ICANS, n = 8).The control cohort consisted of 50 healthy subjects (range: 27-35 years: female: 56.0%), collected as part of the GSP. 5 For each subject, a correlation matrix of 32,492 cortical voxels in each hemisphere was calculated and normalized using z-score.
The DCI was calculated as previously described by Stoecklein et al. 4,6 Briefly, connections that diverge beyond a certain threshold (for this study 16 standard deviations) from the distribution of the respective connection in the reference group, are counted as "dysconnected." The individual patient's dysconnectivity count (DCC) is then summarized in each hemisphere, and normalized to the number of voxels in the respective hemisphere, resulting in the DCI.To evaluate the difference of DCI during versus

Importance of the Study
Immunotherapy with chimeric antigen receptor (CAR) T cells is one of the most innovative and promising treatments for cancer.Despite its success, its application is limited by neurotoxic side effects.There is an unmet clinical need for an objective, quantifiable, and readily available marker to assess immune effector cell-associated neurotoxicity syndrome (ICANS) in the individual patient, in order to inform underlying disease mechanisms, detect individuals at risk for developing ICANS, monitor patients with ICANS, and predict their course of recovery.As conventional structural brain MRI is mostly unremarkable in ICANS, we applied functional MRI and showed that resting-state functional MRI (rsfMRI)-based dysconnectivity index (DCI) can capture and quantify ICANS-related connectivity alterations in the individual patient.The results of our piloting case series propose that rsfMRI-based DCI is a promising imaging marker that should be systematically explored in prospective studies of patients undergoing CART therapy.

Stoecklein et al.: Functional connectivity in neurotoxicity
before or after ICANS, the Wilcoxon matched-pairs signedrank test was used.Correlation between ICANS grade and DCI was calculated using Spearman rank correlation.A significance level of P < .05 was applied.Statistical analysis was performed using SAS version 9.4,Copyright SAS Institute Inc., Cary, NC.
The study was approved by the local ethics committee of the Ludwig Maximilians University (No: 20-646; 19-817) and conducted in accordance with the Declaration of Helsinki.Written informed consent was obtained from each participant.

Results
In all 8 patients, whole-brain DCI was increased during ICANS as compared to the pre-or post-ICANS scan with an overall significantly elevated DCI during ICANS (P = .0039;see Figure 1 and Table 2).Higher ICANS grades were significantly correlated with a higher DCI on a whole brain level (r = 0.7807, P = .0222),and with the DCI of the left hemisphere (r = 0.7807, P = .0222),but not with the DCI of the right hemisphere (r = 0.6831, P = .0618).
Remarkably, patient #5 displayed quite a high DCI score despite only mild neurotoxicity (ie, °I, see Table 2), and patient #4 showed no complete normalization of connectivity during follow-up (see Table 2 and Figure 1).
Seven of 7 patients were right handed.In all 7 right-handed patients, relative DCI increase during ICANS was most pronounced in the left hemisphere (see Figure 1 and Table 2).Patient #3, being left handed by birth, showed the most pronounced DCI increase during ICANS in the right hemisphere.
The most pronounced connectivity increase during ICANS was visualized in the inferior frontal gyrus and the frontal operculum (ie, Broca's area), the posterior parts of the superior temporal gyrus, the temporoparietal junction (ie, Wernicke's area), and the supramarginal gyrus in the inferior parietal lobe (see Figures 2A, B).

Discussion
In the absence of structural abnormalities on conventional MRI, rsfMRI-based DCI was increased during ICANS, and the extent of connectivity changes was significantly correlated with the severity of ICANS as assessed by ICANS grading.Interestingly, DCI increase during ICANS was particularly pronounced in the language-dominant hemisphere in all 8 patients.The prominent connectivity alterations in critical hubs of language processing and multistage actions (ie, Broca´s and Wernicke's area, supramarginal gyrus, and temporoparietal junction) in the language-dominant hemisphere plausibly reflect the clinical core symptoms of nonfluent dysphasia and dyspraxia for multistage actions. 7Furthermore, these findings are in line with marked hypometabolism particularly in frontal cortical regions reported in single cases where [18F]FDG-PET/CT was performed during ICANS. 8n 1patient (#5), however, we found pronounced DCI despite only mild ICANS (°I), which might be due to ICANS grading by the Immune Effector Cell-associated Encephalopathy (ICE) Score, a bedside screening tool similar to the MMSE, with obvious limitations in detecting subtle cognitive impairments, particularly in younger subjects.In 1 patient (#4), the elevated DCI persisted during follow-up.This patient developed metabolic encephalopathy with seizures due to progressive systemic lymphoma.Furthermore, a relapse of ICANS cannot be definitively ruled out in this patient.

Stoecklein et al.: Functional connectivity in
Apart from the limited sample size, the lack of post-ICANS neuropsychological testing to assess recovery constitutes a limitation of this pilot study.Future prospective studies including non-ICANS and CRS comparators as well as detailed and repeated neuropsychological assessment are warranted to further establish the clinical potential of this imaging tool.
Collectively, rsfMRI-based DCI might be suitable to directly quantify the severity of ICANS in individual patients undergoing CAR T-transfusion.Besides ICANS, DCI seems a promising diagnostic tool to quantify functional brain network alterations during encephalopathies of different etiologies, in general.

Figure 1 .Figure 2 .
Figure 1.Intraindividual changes of the dysconnectivity indices during immune effector cell-associated neurotoxicity syndrome (ICANS).The x-axes show the scanning timepoints, the y-axes display the extent of dysconnectivity in each patient.(A) The dysconnectivity index (DCI) of the left hemisphere during ICANS was increased in all patients, when compared to either the baseline DCI (ie, before CAR T-transfusion, n = 3) or the follow-up after clinical recovery from ICANS (n = 5).Clinical severity of ICANS (ie, ICANS grade) as measured by the ICE score showed overall good correspondence with DCI values during maximum ICANS except for patient (patient #5), who exhibited a marked increase of DCI despite mild clinical neurotoxic symptoms (ie, ICANS °1).(B) The DCI of the right hemisphere was increased in all patients during ICANS (n = 8).After clinical recovery from ICANS, DCI was normalized with 1 exception (ie, patient #4) exhibiting continued elevation of DCI.(C) The whole-brain DCI during maximum ICANS was defined as the sum of DCIs of the left and right hemispheres.(D) Statistical analysis revealed a significant increase of DCI during ICANS as compared to DCI at baseline before CAR T-transfusion (n = 3) or after clinical recovery from ICANS (n = 5).Asterisk indicates P < .05.