Preoperative estimation of residual volume for WHO grade II glioma resected with intraoperative functional mapping

Abstract

Despite the lack of class I evidence, it is widely agreed that surgery can improve the functional and vital prognosis for WHO grade II gliomas when the resection is at least subtotal radiologically, that is, leaving less than 10 cm(3) of visible residual tumor. Because these tumors frequently invade functional areas, the preoperative estimation of the probable residual volume remains challenging. This article presents a probabilistic map of postoperative residues, with the aim of predicting before the decision for surgical intervention whether the resection could be subtotal. We selected 65 patients who underwent surgery with intraoperative functional mapping between 1999 and 2004 for a WHO grade II glioma located in a sensorimotor and/or language area. For each case, the postoperative image was normalized on a standard atlas, and the residual tumor was segmented. A probabilistic map of residues was then computed. The fusion between the map and a preoperative image allowed a preoperative estimation of the expected extent of resection. The map enhances the regions where grade II glioma cannot be resected. The success rate for the preoperative classification of partial versus subtotal resection is 82%. Although both its reliability and accuracy have to be improved, this probabilistic map gives preoperatively an objective estimation of the expected extent of resection for grade II glioma resected under intraoperative functional mapping. This rationale will assist in decisions regarding surgical resection and may thus contribute to the elaboration of a therapeutic consensus for WHO grade II glioma.

Fig. 1

Overlay of a patient MRI (gray) with the MNI atlas (yellow). This typical case shows the absence of distortion induced by the left frontal lesion. The mismatch for periventricular white matter is due to the variability of the ventricular sizes. Nevertheless, the residual tumor in the head of the left caudate nucleus is correctly located on the atlas.

Fig. 2

Principle of expected residual tumor volume estimation: The black line corresponds to the segmented, preoperative lesion. In this simplified case, the white voxels on the map have a 100% probability of residual tumor, the light gray voxels, 50%, and the dark gray voxels, 0%. The estimated postoperative volume is as follows: 100% of the volume of the 9 white voxels inside the lesion + 50% of the volume of the 11 light gray voxels + 0% of the 7 dark gray voxels; that is, 14.5 mm³.

Fig. 3

Probabilistic map of residual tumor: Voxels in yellow and red correspond to the sites that are usually not resected (because of their functional role).

Fig. 4

Estimated probable residual volume versus effective segmented residual volume: Twelve patients are not correctly classified in subtotal versus partial resection (⋄, erroneous classification; three cases in the upper left box have a residual volume overestimated by the probabilistic map, whereas nine cases in the lower right box are underestimated).

Fig. 5

Sites with probability of residues greater than 70%, corresponding to specific areas. A. Primary motor area. B. Internal capsule. C. Anterior perforated substance medially and inferior occipitofrontal fasciculus laterally. D. Internal capsule and arcuate fasciculus.