Prediction of Glioblastoma Multiforme Patient Time to Recurrence Using MR Image Features and Gene Expression Nicolasjilwan, M.1·Clifford, R.2·Flanders, A. E.3·Scarpace, L.4·Raghavan, P.1·Hammoud, D.5·Huang, E.6·Jaffe, C.7·Freymann, J.2·Kirby, J.2·Buetow, K.5·Huang, S.8·Holder, C.8·Gutman,D.8·Wintermark, M.1 1University of Virginia, Charlottesville, VA, 2SAIC-Frederick, Inc., Frederick, MD, 3Thomas Jefferson University Hospital, Philadelphia, PA,4Henry Ford University Hospital,Detroit,MI,5National Institute of Health, Bethesda, MD, 6National Cancer Institute, Bethesda, MD, 7Boston University School of Medicine, Boston, MA, 8Emory University Hospital, Atlanta, GA.

DISCLOSURE Nothing to disclose

PURPOSE  Utilize conventional MRI imaging features to predict time to recurrence of patients with glioblastoma multiforme (GBM) after initial diagnosis.  Linear regression models incorporating MR imaging features and tumor gene expression to predict patient time to recurrence.  Important role in selecting treatment options.

Materials & Methods  The study is part of The Cancer Genome Atlas (TCGA) MR imaging (MRI) characterization project of the National Cancer Institute.  MR images for 70 GBM patients made available through the National Biomedical Imaging Archive were reviewed independently by six neuroradiologists.

 The VASARI feature scoring system for human gliomas, developed at Thomas Jefferson University Hospital, was employed.  30 features clustered by categories. –Lesion Location –Morphology of Lesion Substance –Morphology of Lesion Margin –Alterations in Vicinity of Lesion –Extent of resection  (Cancer cell 17:98)  620 genes associated with angiogenesis used in this investigation; Verhaak et al. (Cancer cell 17:98)

 Associations between imaging features and time to recurrence were assessed using linear regression models. Time to recurrence was the outcome; imaging features were the predictors.

Well marginated Non-enhancing F4 Enhancement Quality: 1=None 2=Mild/Minimal 3=Marked/Avid F13 Definition of the non-enhancing margin 1= n/a 2= Smooth 3= Irregular Courtesy Dr Adam Flanders

Predominantly Non-enhancing F5 Proportion Enhancing: 1= n/a 2=None (0%) 3= 95% 8=All (100%) Courtesy Dr Adam Flanders

RESULTS

 Time to recurrence values obtained for 15 patients.  Median time to recurrence: 263 days.

Univariate Analysis of Association between VASARI Features and Time to Recurrence  Individually, 2 MRI features show association to time to recurrence with an unadjusted p-value <  Negative correlation with survival:  Ependymal extension (F19), (P = )  Positive correlation with survival:  Location of the tumor in the left (usually dominant) hemisphere (F2), (P = ).

Multivariate Analysis incorporating VASARI Imaging Features and The Expression of Angiogenesis-related Genes.  Optimal model constructed by addition and substraction of variables: Vasari feature 2 (location of the tumor in the left or right hemisphere) and expression of STAT1, ARHGAP24 and SSTR2 genes.  Predicted time to recurrence based on the model shows a Pearson correlation of (P = 1.42e−09) with observed times to recurrence.  In contrast, tumor localization to the left hemisphere alone shows a correlation of (P = 8.43e−03) with time to recurrence.

CONCLUSION  A subset of VASARI imaging features correlate well with patient time to recurrence.  Linear regression models incorporating multiple imaging features or a single VASARI feature (localization of the tumor to the left or right hemisphere) and tumor gene expression can be used to predict patient time to recurrence.  We are refining these models and are investigating whether including patient clinical characteristics into linear models can improve their predictive power.

AKNOWLEDGMENT University of Virginia Health System SAIC-Frederick Thomas Jefferson University Hospital Henry Ford University Hospital National Institute of Health National Cancer Institute Emory University Hospital Boston University School of Medicine