Author Of 2 Presentations
P0006 - Detecting treatment response on T1 gadolinium enhancing lesion burden in clinical trials of multiple sclerosis with deep learning (ID 998)
Abstract
Background
Ocrelizumab (OCR) is a humanized anti-CD20+ monoclonal antibody approved for the treatment of relapsing and primary progressive forms of multiple sclerosis (MS). It suppresses the development of new areas of inflammation as shown by the reduction in the number of T1 gadolinium (Gd) enhancing lesions. Deep learning (DL) based segmentation of lesions has the potential to automate these manual reads; enabling faster and more reproducible quantification.
Objectives
To evaluate a DL model’s ability to detect OCR treatment response on the number of T1 Gd enhancing lesions in clinical trials of relapsing remitting MS (RRMS).
Methods
We used images from Opera I trials (NCT01247324, n=898), to train a Unet model to segment both unenhancing and Gd-enhancing T1 lesions. T1 Gd- enhancing lesions in ground truth (GT) masks were present in ~44% of patients at baseline and ~15% across all time points. We created a dataset with an equal number of imaging volumes with and without enhancing lesions and used 70%-30% data splits for training and validation. The model was tested on images from Opera II trials (NCT01412333, n=905). To detect significant differences between treatment and control arms, we performed a negative binomial regression on the number of T1 Gd-enhancing lesions with baseline imaging and clinical covariates.
Results
The DL model achieved a mean dice coefficient (DC) of 0.72, lesion true positive rate of 0.92 (LTPR), lesion false positive rate (LFPR) of 0.06 and a volume correlation coefficient (CC) of 0.97 for Gd-enhancing lesion segmentation. For unenhancing lesion segmentation, the mean DC was 0.68, LTPR 0.78, LFPR 0.18 and volume CC of 0.97. The model had the highest false positives for lesions smaller than 10 voxels (voxel size: 1x1x3 mm3). For Gd-enhancing lesion segmentation a significant OCR treatment effect (p<0.001) in reducing the mean number of Gd-enhancing lesions at 24, 48 and 96 weeks (92%, 96%, 97% reduction from GT manual reads vs 67%, 71%, 78% from model predictions) was detected.
Conclusions
Our DL model performed Gd-enhancing lesion segmentation comparably to similar DL models in the literature and showed a high correlation to GT manual reads. Our model also had sufficiently high sensitivity to detect an OCR treatment response consistent with neuro-radiologist assessments. To our knowledge, this is the first study to report that a DL model has the sensitivity to detect treatment response on Gd-enhancing lesions.
P0613 - Ocrelizumab Reduces Regional Brain Atrophy in Relapsing Multiple Sclerosis as Assessed by Longitudinal Deformation-based Morphometry (ID 1228)
Abstract
Background
Ocrelizumab (OCR) is a humanised anti-CD20+ monoclonal antibody approved for the treatment of relapsing and primary progressive forms of multiple sclerosis (MS). It has been reported that OCR reduces brain atrophy in MS patients, however, a voxel-based approach has never been reported.
Objectives
To identify brain regions where atrophy is modified by the OCR treatment in Relapsing MS (RMS), using a fully automatic, voxel-based approach.
Methods
We applied longitudinal deformation-based morphometry (L-DBM) using the Advanced Normalization Tools (ANTs) to T1-weighted 3D MPRAGE brain MRI (voxel=1x1x3 mm3 ) from two clinical trials of OCR for RMS (Opera I: NCT01247324 and Opera II: NCT01412333). For each MRI visit, a Jacobian map was derived based on the deformation generated by registering the scan to a Single-Subject Template (SST) constructed for the same patient. Jacobian maps in patient SSTs were mapped to a population brain template for group analysis. Patients were included if MRI data were available for all visits (baseline, Weeks 24, 48, and 96) and the images were of sufficient quality to be successfully processed through the L-DBM pipeline (683 in the OCR arm, 580 in the active control arm of Interferon beta-1a). Voxelwise ANOVA was iterated over the whole brain to detect an interaction between Treatment and Time (p<0.01 with familywise error correction). Post-hoc analysis was done with mean Jacobian of identified regions, including Linear Mixed Effect (LME) analysis.
Results
Voxelwise ANOVA identified one cluster of voxels in the brain (besides the ventricles) that were contained within parts of thalamus, brainstem, and cerebellum. From the analysis of mean Jacobian of this cluster region, the percent volume reduction from baseline was significantly smaller in the treatment arm than the control arm at every follow-up visit (p<0.0001, t>8.0, relative difference>75%). LME analysis of the mean Jacobian of the cluster region confirmed that there was a strong interaction between Treatment and Time (F=18.35), for which age and sex were controlled.
Conclusions
L-DBM has identified several brain structures where atrophy was modified by the OCR treatment in RMS. The L-DBM analysis was performed at the voxel level and these regional findings are consistent with previous reports. Localization of the treatment-responsive brain structures may have an implication for future assessment of clinical outcomes in MS.