List of communications

Prognostic and predictive significance of geometric measures from high resolution pretreatment T1+Gd MR images of glioblastoma multiforme

Pérez-Beteta J1, Martínez-González A1, Luque B1,2, Arregui E3, Calvo M3, Borrás JM3, López C3, Claramonte M3, Barcia J4, Iglesias L4, Albillo D5, Martino J6, Velasquez C6, Asenjo B7, Benavides M7, Herruzo I7, Arana E8, García A6,Sepúlveda JM17, Peralta S16,Gil-Gil MJ15, Oscar Gallego8, Pérez-Segura P4, Reynes G14, Herrero A13, las Peñas R10, Luque R9, Capellades J12, Balaña C11, Pérez-Romasanta L5, Pérez-García VM1

1: Instituto de Matemática Aplicada a la Ciencia y la Ingeniería, Universidad de Castilla-La Mancha, Spain
2: Instituto Nacional de Estadística, Málaga
3: Hospital General de Ciudad Real, Ciudad Real, Spain
4: Hospital Clínico San Carlos, Madrid, Spain
5: Hospital Universitario de Salamanca, Salamanca, Spain
6: Hospital Marqués de Valdecilla, Santander, Spain
7: Hospital Carlos Haya, Málaga, Spain.
8: Instituto Valenciano de Oncología, Spain
9: Hospital Universitario Virgen de las Nieves, Granada, Spain
10: Hospital Provincial de Castellón, Castellón, Spain
11: Hospital Universitario Germans Trias i Pujol, Badalona, Spain
12: Hospital del Mar, Barcelona, Spain
13: Hospital Miguel Servet, Zaragoza, Sapain
14: Hospital Universitario, La Fe, Valencia, Spain
15: Instituto Catalán de Oncología-IDIBELL, Hospitalet de Llobregat, Spain
16: Hospital Sant Joan de Reus, Reus, Spain
17: Hospital Universitario 12 de Octubre, Madrid, Spain

Background: The prognostic value of different geometric measures of pretreatment T1+Gd MRIs (volumes of necrosis, contrast enhancing areas, edema, different lengths, etc) of GBM patients has been controversial for years, with contradictory and/or not statistically significant results being found in the literature [1-4]. In addition, a recent paper has constructed a mathematical evolutionary model for GBM growth that supports a predictive value of some of those measures [5]. The goal of the study was to clarify the relevance of several 3D geometric measures of pretreatment T1+Gd MRIs and to validate the predictions of Ref. [5]

Patients and methods. Two sets of patients where included in the study corresponding to two retrospective clinical trials. The first one (GlioMat trial) consists of 120 patients with high-resolution pretreatment (no corticosteroids) T1+Gd primary GBMs. The second one is a subset of patients of the Genom009 trial including 40 patients with unresectable GBMs.The patients contrast-enhancing areas and inner ‘necrotic’ regions were segmented using a semi-automatic segmentation procedure. The segmented tumors were analyzed to extract: the maximal 3D diameter, the volume of necrosis, the volume of contrast enhancing tumor, the total volume, and a number of measures of the contrast enhancing rim including distance histograms, maximal, mean and median contrast enhancing width, and an averaged pseudo-spherical quantification of the rim width. Also a measure of the irregularity of the tumor surface was computed. The geometrical values obtained were correlated with measures of the response, progression free survival and overall survival.

Results: The detailed results will be presented in the talk. The maximal 3D diameter turns out to be the most relevant independent predictor of survival with a very high significance. Other geometrical measures were also predictors of survival. The subgroups of patients having only biopsy vs. complete resection displayed interesting differences. Finally the potential of these measures as predictors of response to antiangiogenic therapies will be discussed. It is proven that the number and quality of images (3D without any previous treatment), the segmentation methods, the variables defined (fully 3D) may have a key role in the powerful results of the present analysis.

[1] Nestler U et al. (2015) Anatomic features of glioblastoma and their potential impact on survival, Acta Neurochir 157:179-186

[2] Zhang Z et al. (2014) Identifying the survival subtypes of glioblastoma by quantitative volumetric analysis of MRI, J Neuro-oncol 119:207-214

[3] Gevaert O et al. (2014) Glioblastoma Multiforme: Exploratory radiogenomic analysis by using quantitative image features, Radiology 273:168-174

[4] Mazurowski M et al (2014) Computer-extracted MR image features are associated with survival in Glioblastoma patients, J. Neurooncol 120:483-488

[5] Pérez-García et al. (2011) Bright solitons in malignant gliomas, Phys Rev E 84:021921

Format: Poster

Organized by