SP #3: Quantitative MRSI to predict early response to SAHA therapy in new GBM management

Quantitative MRSI to predict early response to SAHA therapy in new GBM management


Glioblastoma multiforme (GBM) is the most common primary brain tumor and is uniformly fatal. It has become evident that the tumor suppressor genes in GBM cells are silenced by abberant histone deacetylase (HDAC) activity. Suberoylanilide hydroxamic acid (SAHA) is an orally-active, potent inhibitor of HDAC.  This agent may not only help control tumors but also alter cerebral biochemistry to improve depressive symptoms afflicting many GBM patients. However, the lack of reliable biomarkers to predict early response severely hampers the treatment of GBM patients with HDAC inhibitors. Magnetic resonance imaging (MRI) is the standard tool for monitoring therapeutic response in GBMs.  Although useful, conventional MRI has shortcomings including difficulty at distinguishing true tumor progression from “pseudo-progression,” and it is not ideal for evaluating response to SAHA since the drug response is associated with redifferentiation rather than killing/shrinking tumors.  

In this study, we will use MR spectroscopic imaging (MRSI) to measure the metabolism of cancer cells as well as normal brain. The establishment of reliable MRSI metabolic biomarkers to assess early response would be of great value in developing new treatments, especially those such as SAHA which do not work by simply killing cells. While MRSI is not new, it has not gained widespread clinical acceptance due to poor resolution, long scan times, and difficulty integrating with other types of brain scans.


We will develop state-of-the-art MRSI technology designed to generate metabolite maps of the entire brain rapidly, and will provide streamlined analysis and registration tools to facilitate clinical use.  We will then deploy these tools to evaluate the prognostic value of MRS-detected metabolites in GBM and explore the relationship of metabolite concentrations with depression in GBM patients.


Our central hypotheses, supported by data in a preliminary cohort of GBM patients, are as follows: (1) In newly-diagnosed GBM patients, MRSI showing normalization of tumor metabolites after a short pretreatment with SAHA can predict for improved outcome with SAHA in combination with standard chemoradiation (XRT+TMZ); and (2) Altered MRS metabolites (particularly low MI and NAA) in surrounding “normal” brain will predict depressive mood in GBM patients and restoration of these metabolites by SAHA will treat this depression.  In order to test these hypotheses, we will first combining 3D-EPSI with parallel imaging (as per our original aims) and with the combined parallel imaging and compressed sensing approaches developed in the BTRC.   We will develop the procedures and tools needed to import data from our analysis program MIDAS into Velocity AI, an imaging display and registration program that detects and repositions multimodality imaging data.  We will then establish inter-site MRSI reproducibility at Emory and Johns Hopkins University (with whom BTRC acceleration methods will also be shared).   

We will next investigate individual MRS biomarkers as potential predictors of good outcomes from SAHA in combination with standard chemoradiation (XRT+TMZ) for GBM patients.  To make better use of the MRSI metabolite measures, we will develop a support vector classification approach to produce signatures representing composite measures of relevant metabolites.  We will then determine whether pretreatment molecular pathologic factors including histone acetylation status are predictive of response to SAHA. 

Finally, we will correlate MRSI-detected MI and NAA concentrations with depressive symptoms assessed by a self-report depression survey (IDS-SR), and will correlate depressive symptoms with neurocognition and quality of life (QOL), in SAHA-treated versus untreated patients with GBM. 

BTRC Resources Utilized:

TR&D #1: Although we did not propose using the combined compressed sensing and parallel imaging approaches to be developed in the BTRC in our original U01 application, these approaches could substantially shorten our current MRSI acquisition time of 26 minutes (3D whole brain at 4x4x4 mm3 resolution), thereby improving the practicality of using MRSI in routine standard-of-care exams for GBM patients. In addition, both accelerated approaches in general and radial approaches in particular could markedly reduce motion artifacts in patients who have difficulty remaining still for the duration of the scans. 

Since the original submission, our collaborator in MRSI method development, Dr. Maudsley, has visited NYU and discussed requirements for evaluating the practical limits of acceleration in sample datasets.  With continued BTRC support, our plan is to test the viability of a) highly-accelerated and incoherently undersampled Cartesian acquisition trajectories, and b) 3D radial trajectories adapted for MRSI. 

Principal Investigator: 


Philanthropic Support

We gratefully acknowledge generous support for radiology research at NYU Langone Health from:
• The Big George Foundation
• Bernard and Irene Schwartz

Go to top