MRI for Pediatric Optic Pathway Glioma Treatment Response

Children’s Hospital of Philadelphia

Robert Avery
AVERYR@EMAIL.CHOP.EDU

Children’s National Hospital (Washington, D.C.)

Marius Linguraru
MLingura@childrensnational.org )
Grant Number: UG3 CA236536

Low-grade glioma is the most common brain tumor in children and often involves one or more structures of the anterior visual pathway (i.e., optic nerves, chiasm and tracts). Nearly 20% of children with neurofibromatosis type 1 (NF1) will develop a low-grade glioma of the anterior visual pathway, which are called optic pathway gliomas (OPGs). NF1-OPGs are not amenable to surgical resection and can cause permanent vision loss ranging from a mild decline in visual acuity to complete blindness. Children with NF1-OPGs typically experience vision loss between 1 and 8 years of age and are monitored with brain magnetic resonance imaging (MRI) to assess disease progression. However, traditional two-dimensional (2D) measures of tumor size are not appropriate to assess change over time and how NF1-OPGs are responding to treatment.

This work addresses the lack of robust and standardized quantitative imaging (QI) tools and methods needed for NF1-OPG clinical trials. The team will develop and validate a novel three-dimensional (3D) MRI-based QI application for automated and comprehensive quantification of these unique pediatric tumors. They will use machine learning algorithms to accommodate MRI sequences from different manufacturers and protocols. Working hypothesis is that the novel QI application will accurately assess treatment response in clinical trials. In this project, team will validate their QI software and machine learning methods to make accurate and automated measures of tumor volume and shape using data from a phase 3 clinical trial of NF1-OPGs. From these measures, they will create methods to assess response to therapy that will enable physicians to make informed and objective treatment decisions.

The specific aims are: 1) Develop a comprehensive QI application to perform accurate automated quantification of NF1-OPGs; 2) Determine and predict treatment response using team’s 3D QI measures of tumor volume; and 3) Validate the 3D QI measures using visual acuity outcomes.

Upon study completion, team’s QI application could transform clinical care for NF1-OPG by identifying the earliest time to determine a favorable versus unfavorable treatment response. The QI application’s ability to accurately measure treatment response, along with harmonizing data across MRI manufacturers and protocols, will standardize imaging assessments essential to NF1-OPG clinical trials.

Representative publications related to the QIN project:

Tor-Diez C, Porras AR, Packer RJ, Avery RA, Linguraru MG. Unsupervised MRI Homogenization: Application to Pediatric Anterior Visual Pathway Segmentation. Mach Learn Med Imaging. 2020 Oct;12436:180-188. doi: 10.1007/978-3-030-59861-7_19. Epub 2020 Sep 29. PMID: 34327515; PMCID: PMC8317430.
Mansoor A, Linguraru MG. Communal Domain Learning for Registration in Drifted Image Spaces. Mach Learn Med Imaging. 2019;11861:479-488. doi: 10.1007/978-3-030-32692-0_55.
Avery RA, Mansoor A, Idrees R, Trimboli-Heidler C, Ishikawa H, Packer RJ, Linguraru MG. Optic pathway glioma volume predicts retinal axon degeneration in neurofibromatosis type 1. Neurology. 2016 Dec 6;87(23):2403-2407. doi: 10.1212/WNL.0000000000003402. Epub 2016 Nov 4. PMID: 27815398; PMCID: PMC5177678.
Avery RA, Mansoor A, Idrees R, Biggs E, Alsharid MA, Packer RJ, Linguraru MG. Quantitative MRI criteria for optic pathway enlargement in neurofibromatosis type 1. Neurology. 2016 Jun 14;86(24):2264-70. doi: 10.1212/WNL.0000000000002771. Epub 2016 May 11. PMID: 27170570; PMCID: PMC4909554.
Mansoor A, Cerrolaza JJ, Idrees R, Biggs E, Alsharid MA, Avery RA, Linguraru MG. Deep Learning Guided Partitioned Shape Model for Anterior Visual Pathway Segmentation. IEEE Trans Med Imaging. 2016 Aug;35(8):1856-65. doi: 10.1109/TMI.2016.2535222. Epub 2016 Feb 26. PMID: 26930677.