Quantitative Imaging Study of Extent of Surgical Resection and Prognosis of Malignant Astrocytomas

To the Editor: We read with great interest the article by Kowalczuk et al.⁽⁵⁾, who attempted, using quantitative imaging studies, to evaluate the prognostic value of cytoreductive surgery in patients with high-grade gliomas. During the last 30 years, several studies have addressed this question, and until now, the influence of the extent of resection on the time of recurrence and on survival remains controversial. In many cases, this evaluation is based on the operative perception by the neurosurgeon. In some studies, however, neuroimaging techniques were used to better assess the residual tumor. In their study, Kowalczuk et al. used pre- and postoperative computed tomography or magnetic resonance imaging (MRI) combined with quantitative analysis and did not find that the extent of resection represents a prognostic factor. This study, however, contains several methodological biases that may question the validity of its conclusions, as was emphasized in the comments published with the article.

In the study presented by Kowalczuk et al., it was interesting to note that although 93% of the patients had a lobar location of their tumors, only 13% of the patients had no residual tumor revealed by postoperative imaging. It would have been of interest to know whether the extent of resection in this subgroup of patients (e.g., those with macroscopic total resection) was a predictor of better survival. Macroscopic total removal of the tumor is achievable in selected patients with high-grade gliomas and may represent an important prognostic factor in that particular situation. In an era of neurosurgery during which new technological adjuvants, such as neuronavigation, are entering common practice, we think that it is of importance to clarify whether neurosurgeons should try to achieve total resection of glial tumors, even when they are malignant.

To evaluate whether patients benefit from extensive resection of malignant gliomas, it is essential to consider the imaging procedure used for the postoperative assessment of the extent of resection. In a pilot prospective study, we have compared MRI and positron emission tomography (PET) with[18F]fluorodeoxyglucose (FDG) in a group of 10 patients, with macroscopically confirmed total resection of high-grade gliomas. All patients underwent MRI(T1-weighted images with and without gadolinium diethylene triamine penta-acetic acid injection) and PET-FDG before surgery and within 4 days after surgery. It was possible to repeat these examinations 6 months later in seven patients; two patients died at 3 months and one at 6 months after tumor resection. The extent of resection, as evaluated by MRI and PET-FDG within 4 days after surgery, and the follow-up data are summarized inTable 1. Although all patients were considered to have undergone macroscopically confirmed total resection by the neurosurgeon(ODW), seven had residual tumor demonstrated by postoperative neuro-imaging. Among them, three died at 3 months (n = 2) or 6 months (n = 1) after surgery and four showed clinical and radiological signs of recurrence at 6 months after surgery. At the time of this writing, all of the patients had residual tumor visible on both magnetic resonance images and positron emission tomographic scans. On the early postoperative images, however, signs of residual tumor were visible on both magnetic resonance images and positron emission tomographic scans for only five patients; for one patient, residual tumor was visible only on MR images, and for another one, residual tumor was visible only on positron emission tomographic scans. Conversely, the three patients with no tumor residue revealed by either early MRI or PET were still free of tumor recurrence, as revealed by both imaging modalities at 6 months after surgery, and the results of their neurological examinations remained unchanged. χ² test with Yates' correction indicates a dependency between the probability of recurrence at 6 months and the presence of tumor residue revealed by either MRI or PET (P < 0.01). Thus, these preliminary data stress the importance of MRI and PET-FDG for the estimation of tumor residue after apparent total resection of malignant glioma. Moreover, they suggest that when complete resection is identified with these imaging techniques, it may be of prognostic value.

TABLE 1: Evaluation of Postoperative Residual Tumor by Positron Emission Tomography and Magnetic Resonance Imaging¹

Although the data published by Kowalczuk et al. are inconclusive, we think that further studies with more optimal design are needed to evaluate the prognostic role of the extent of resection in malignant gliomas. In this respect, PET may be a useful tool to consider. PET allows the accurate and sensitive detection of tumor areas with anaplasia ^(3,4). Moreover, PET-FDG provides information on glucose metabolism, which has a prognostic value independent of histology, at least in low-grade gliomas ⁽²⁾. Similar conclusions have been recently drawn regarding malignant gliomas ⁽¹⁾. Also, in one case of the present study, PET-FDG was more sensitive than was MRI in the early detection of residual tumor. Nevertheless, the interpretations of PET-FDG may be difficult without the support of morphological information. For example, in three of our patients, MRI was necessary to identify and to differentiate displaced cortex from tumor residue on the PET images. Thus, the concomitant use of PET and MRI seems to be useful and could probably reach a better predictive value than could be reached by using each method separately. We have already used such an approach to guide stereotactic biopsies of brain tumors ⁽⁶⁾ and found that the adjunct of PET-FDG improves their diagnostic yield ⁽⁷⁾. We now use a similar approach to integrate PET data in the planning of neuronavigation for the resection of brain tumors ⁽⁸⁾. Taken together, our data on the combined use of PET and morphological images, if confirmed in larger series, should prompt further studies of the role of cytoreductive surgery for malignant gliomas, including early treatment of residual tumor by reoperation, radiosurgery, or other new therapeutical strategies.

Olivier De Witte

Marc Levivier

Philippe Violon

Jacques Brotchi

Serge Goldman

Brussels, Belgium