Early Postoperative Magnetic Resonance Imaging after Resection of Malignant Glioma: Objective Evaluation of Residual Tumor and Its Influence on Regrowth and Prognosis

To the Editor: The article by Albert et al. (¹) tries to establish, in a prospective manner, a relationship between the extent of resection and the survival/progression-free interval in patients with glioblastoma multiforme. As implied by the title, the “objective” measure of the extent of resection is the early postoperative contrast-enhanced magnetic resonance (MR) image. The major conclusion of the authors is that patients with a partial resection of a malignant glioma have a 6.6 times higher risk of death as compared with patients who undergo a total tumor resection. Therefore, in their multivariate analysis, they conclude that extent of resection is the most important predictive value in the treatment of malignant gliomas and advise radical surgery whenever possible. To our perception, these conclusions are the result of considerable bias pertaining to both the analysis of the MR images as well as the clinical data.

The article creates the impression that a malignant glioma can be visualized with all of its different parts on a contrast-enhanced MR image. This remains questionable (⁴), as there are malignant gliomas with an almost intact blood-brain barrier or only slight contrast uptake where even hypodense areas represent solid tumor. What can be seen on a gadolinium-enhanced MR scan is the area with increased capillary permeability and/or increased vascular volume, which does not reflect the size of the tumor (²). The authors acknowledge this on Page 50 when they report on one of the two patients from whom they got histological confirmation of their suspected MR diagnosis. In this case, the tumor did not match the post-contrast MR image at all. To substantiate their neuroradiological criterium of solid tumor matching the enhancing area on computed tomographic or MR scan, they cited the articles by Burger and Kelly (³, ⁵). These articles deal with nontreated tumors and therefore completely rule out treatment artefects or treatment effects. Therefore, their results cannot be easily transposed to the study presented by Albert et al. Thus, the study, because of the tumor physiology itself, is biased as to the “objective” assessment of extent of resection, as the only 11 (!) patients assumed to have total resection may simply reflect the portion of patients in whom the tumor had no major disturbance in the blood-brain barrier, i.e., low capillary permeability at the resection margin.

The authors may argue that they have used a control group for the aforementioned variables. The control group consisted of eight patients with low-grade gliomas, which are known to exhibit a completely different physiological behavior, especially on the capillary level and as to their neovascularisation. Thus, an evaluate control group for malignant gliomas does not exist. Furthermore, we found it irritating that although the study should be a prospective one, the significance of the different MR findings were established retrospectively, as the authors stated on Page 46. Therefore, it remains unclear why a primary enhancement should be benign and not an indicator for tumor tissue if it appears 4 days after surgery and disappears 6 months after radiotherapy. Why then is a slight linear enhancement in the first 3 days not a benign one? And why should a disappearing enhancement 6 months after radiotherapy not reflect the effect of this treatment modality on the remaining tumor?

In addition to these basic objections, the article is seriously biased by a couple of statistical problems. The two groups (complete resection, n = 11, 7 deaths; and incomplete resection, n = 42, 40 deaths) have a very different sample size and are highly selected. Patients with incomplete resections were older and had a worse Karnofsky-Score. They had more frequent recurrences (19 versus 0%) and received radiotherapy less frequently (69 versus 91%). All these factors had prognostic validity in other studies. The inhomogenity with regard to these important covariates in both treatment groups is a serious obstacle for proper analysis. Patients with recurrent tumors, incomplete radiotherapy, and anaplastic astrocytomas should have been excluded. Because of the imbalances and incomparability of the treatment groups, it is not surprising that the result of this multivariate analysis could not identify the powerful influence of age identified by other large studies. The enormous increase of the risk potential for patients with incomplete tumor resection appears questionable and overestimated by the selection bias in the study of Albert et al.

Interestingly, the macroscopic boundary as subjectively defined by the surgeon turned out to be invalid by the criteria of MR follow-up. This could mean that either there is no such thing or the MR follow-up study is looking at an artefact. As the “objective” evaluation of the MR scans was performed according to rather subjective criteria, as stated above, the beauty of total resection of malignant gliomas, even using MR imaging, still lies in the eyes of the beholder. Thus, although the study may generate further interest in the vascular physiology of malignant gliomas, the data do not substantiate the concept of cytoreduction by gross total resection and, in the interest of our patients, should be interpreted more carefully.