Many neurosurgeons take a nihilistic approach to surgical treatment of gliomas, stating the inability to achieve a cure. Where this idea comes from is somewhat nebulous to most neurosurgeons. A review of the scientific studies supporting the commonly held beliefs about gliomas shows that these ideas regarding the surgical treatment of gliomas are based on overgeneralizations of data from older studies. One should avoid the temptation to apply them to the greater concept of what gliomas are, how they behave, and what should be done, but rather we should continue to scientifically evaluate the role of surgical resection in glioma treatment.
Commonly held views about glioma treatment
Infiltrating gliomas, defined as World Health Organization grade II through IV astrocytic or oligodendoglial neoplasms, are known to all with even casual exposure to their clinical history as invariably multiply recurrent and eventually fatal, albeit at grade-specific rates, despite lesionectomy, adjuvant radiotherapy, and chemotherapy. Despite generations of effort, the outcome for these lesions has improved only marginally, and the cure rate remains dismally low. The future of glioma therapy is in the laboratory, and eventually some molecular-based therapy will be developed that will be the ultimate solution for this terrible disease, because surgery clearly is not. In many cases, aggressive surgery is ill-advised, pointless, and harmful. In these cases, biopsy and radiation alone serves as an acceptable alternative to surgery, despite the dismal prognosis with this approach.
The paragraph above summarizes how many surgeons and oncologists view treatment of gliomas. Many of these views are handed down between clincians, yet given that there are not definitive class 1 studies demonstrating most of these statements, we feel these views are worth a critical reassessment. This article summarizes much of what been taught regarding glioma treatment and, more specifically, glioma surgery. Unquestionably, a great deal of published data support this ideology, not the least of which are from the collective experience of many who have treated patients using this paradigm. The purpose of the article is not to entirely discount these ideas, but rather to critically address the scientific and clinical foundations on which these approaches are based. A review of the scientific studies supporting the commonly held beliefs about gliomas shows that these nihilistic ideas regarding gliomas are based on overgeneralizations of older historical studies that have been applied to the greater concept of what gliomas are, how they behave, and what should be done to treat them.
Nihilism in glioma surgery
High-grade gliomas, notably glioblastoma, behave extremely aggressively. Regardless of grade, they have a high rate of recurrence and progression to higher grades, such as glioblastoma, and are ultimately fatal in most patients. These tumors notoriously infiltrate normal brain, meaning that surgical resection generally involves removal of functional tissue. Hence, these tumors are difficult to remove completely, especially when neoplastic cells spread widely throughout the brain. Given these challenges, many neurosurgeons still (despite evidence to the contrary) choose to avoid this risk by instead biopsying these tumors and then radiating them.
Nihilism in glioma surgery
High-grade gliomas, notably glioblastoma, behave extremely aggressively. Regardless of grade, they have a high rate of recurrence and progression to higher grades, such as glioblastoma, and are ultimately fatal in most patients. These tumors notoriously infiltrate normal brain, meaning that surgical resection generally involves removal of functional tissue. Hence, these tumors are difficult to remove completely, especially when neoplastic cells spread widely throughout the brain. Given these challenges, many neurosurgeons still (despite evidence to the contrary) choose to avoid this risk by instead biopsying these tumors and then radiating them.
Is nihilism justified?
The Hemispherectomy Experience
A common belief is that, regardless of radiographic appearance, gliomas disseminate widely and have often microscopically infiltrated even across to the contralateral hemisphere at initial presentation. Thus, many claim that gliomas are incurable. As support, many authors refer back to early studies attempting to cure glioblastoma with hemispherectomy. The recurrence rates despite this approach are cited as the ultimate proof that surgery cannot definitively treat these tumors.
The most prominent report by Dandy in JAMA in 1928 described the approach in five patients. Two of these patients died of early postsurgical complications and one was still alive at last follow-up. The remaining two patients died of recurrent tumors 3 months and 3.5 years after surgery, respectively. The patient who died at 3 months was noted to have disease extending into the basal ganglia that was intentionally not resected. The patient who lived 3.5 years was in a coma at presentation and lived well beyond the present mean life expectancy despite no adjuvant therapy, which was not available in 1928.
The only other notable report, which was published in 1949, is of five patients who underwent surgery at the Cleveland Clinic. One of these patients died from early postoperative complications; two died of a recurrent tumor 15 months and 3 years after surgery; one died 4 years after surgery from a traumatic brain injury; and one lived at least 10 years after surgery. After this article was published, reports about hemispherectomy in glioma become fewer, and the latest report identified was in 1975, in which the patient died from a contralateral meningioma.
Given this paucity of evidence, particularly the lack of any current evidence, one cannot say what hemispherectomy would do for patients with glioma, and what it would accomplish if tried today when applied in the context of existing management paradigms. Given that most of these reports are 60 years to 80 years old and predate MRI and CT, many important facts about these patients are unknown, most notably whether these tumors were bilaterally disseminated at diagnosis, making hemispherectomy a suboptimal option. Information regarding radiographic extent of resection is also not available for these patients. 5-Aminolevulinic acid (5-ALA) certainly was not used, because operating microscopes with or without fluorescent filters were not available to visualize it. None of these patients received radiotherapy or chemotherapy. No repeat surgeries were performed for recurrence, because there was no way to detect it until the patient was near death. Many of these patients could have died of the siderosis that occurred frequently with old techniques of hemispherectomy, because no insight into this problem was available at the time, and detailed postmortems looking for this problem were not performed on many of these patients. Furthermore, the histopathologic diagnostic criteria of gliomas have changed several times since these reports. No control cohort is available with any meaningful relevance to contemporary therapy, and therefore perhaps all of these patients experience recurrence, but after many more years.
Thus, one of the key pieces of evidence dismissing glioma surgery as futile is from from decades-old case series of five or fewer patients who were managed using outdated surgical techniques, outdated diagnostic techniques, and outdated adjuvant treatment paradigms. In addition, these patients were diagnosed using outdated criteria, and had inadequate follow-up according to today’s standards of evidence in surgical studies. This fact is stunning when considering the large number of patients who are denied full aggressive surgical treatment for gliomas based partly on these outdated studies.
The Tumor Cell Dissemination Story
Grade II through IV astrocytomas and oligodendrogliomas are known as infiltrating gliomas based on their long observed tendency to spread widely throughout the brain far beyond the primary site of disease. This finding was first documented by Bailey and Cushing, 1926, and was most fastidiously documented by Scherer in a report in 1940. In this frequently cited autopsy study of 120 patients with gliomas, Scherer concludes that, except for ependymomas, “infiltrative growth must be regarded as characteristic of the enormous majority of gliomatous tumors.” He then concludes that “complete extirpation is not possible.” He notes that, even in this study of postmortem gliomas, bilateral involvement with tumor was observable in only 30% of these cases.
Certainly, several autopsy studies subsequent to Scherer have shown that these tumors can spread widely, to the point that this propensity of gliomas to migrate throughout the brain in some cases is beyond debate. The autopsy-based literature has failed, however, in a fallacy of extension. More specifically, many clinicians have concluded that because evidence shows that many gliomas demonstrate widespread microcellular dissemination at the end stage (ie, the subjects in most postmortem studies), this implies that all gliomas are widely disseminated at diagnosis, including lower-grade astrocytomas. Furthermore, it implies that these distant microcellular satellite cells are the source of the tumor that ultimately kills the patient. In this belief system, leaving the tumor behind seems perfectly reasonable, because aggressive treatment would be futile if more tumor is always extending beyond the margins of the resection.
The most prescient argument against this belief system is an observation practitioners are all familiar with, namely that most glioma recurrences, especially with low grades, occur near or within the previous resection cavity. Studies have shown that approximately 76% of glioblastoma recurrences occur within 2 cm of the surgical resection margin. This finding argues strongly that the problem with the current surgical paradigm is inadequate control of tumor at the margins, rather than the presence of distant disease, in most patients treated. Thus, the issue for most failures is inherently inadequate margins of resection, not far-flung satellite cells. This theory is further strengthened by studies showing that radiotherapy margins could be successfully reduced from whole-brain fields to fields extending just 2 cm beyond the resection bed, suggesting that infiltration is largely by cells just beyond the edge of MRI-guided tumor resection cavity that cause recurrence. The radiation oncology world figured out many years ago that the problem is usually at the margins and not the whole brain.
To further show that the issue for most of these tumors, at least initially, is local treatment failure, the authors refer to four well-executed, influential, and frequently cited studies from the CT/MRI era showing the existence of infiltrating tumor cells beyond the area of radiographic tumor involvement. Although frequently cited as evidence of the futility of aggressive glioma surgery by those who have not read them closely with a critical eye, these studies in fact show that the problem is probably one of inadequate resection margins in many cases.
Salazar and Rubin, 1976
Salazar and Rubin reported an autopsy study of 42 patients with glioblastoma who died shortly after diagnosis (meaning not altered by treatment), including 35 supratentorial tumors, 6 intratentorial tumors, and 2 spinal tumors. They concluded that 29 of 35 patients had tumor that extended outside the clinically expected boundary of the tumor. Of these 35 tumors, 9 had spread to the contralateral hemisphere, only 4 had spread to infratentorial structures, and 4 had spread from peripheral brain to deep structures not anticipated clinically. Of the 29 peripherally located glioblastomas, 6 crossed to the other hemisphere, 4 invaded the deep structures, and 2 invaded the infratentorial brain. Although they do not publish whether any of these patients were duplicated in these classes, the findings imply that, at worst, 48% of these patients with rapidly fatal glioblastomas had anatomic characteristics that would make aggressive surgical resection undesirable. The findings also fail to account for the fact that the study was performed on patients diagnosed in the 1950s and 1960s, when these tumors were generally caught much closer to end stage than is likely common in contemporary practice. Thus, even in the worst case scenario, many of these tumors are still not as widely disseminated as seems to be suggested in the nihilistic view.
Burger and colleagues, 1983
In perhaps one of the most interesting studies of this group, Burger and colleagues focused on analyses of surgical specimens and autopsies in 20 patients with untreated high-grade gliomas (5 cases), lesions in patients experiencing remission who died of other diseases (3 cases), and recurrent tumors (12 cases), correlating pathologic findings with imaging characteristics. They found that in untreated glioblastoma, although cells were present outside the area of imaging abnormality, no evidence of tumor cells could be found beyond 3 cm outside of tumor center, nor any tumor cells in the contralateral hemisphere. Similarly, lesions in remission did not show any evidence of widespread dissemination. Only at recurrence did the investigators note tumor cells spread far beyond the primary tumor site, but they also noted that the greatest concentration of tumor cells was seen locally. Together with the previous study, this suggests again that widespread dissemination is an end-stage, not primary, trait of glioblastomas in most cases.
Kelly and colleagues, 1987
By far the most influential study on this topic (cited more than 400 times) was that of Kelly and his team at New York University. In this landmark study, they performed serial stereotactic biopsies both inside and outside of the tumor volume, thus obtaining 195 biopsy specimens in 40 patients, including 8 grade IV tumors, 7 grade III tumors, 17 low-grade astrocytomas, and 8 low-grade oligodendrogliomas. By doing so, they were able to correlate MRI and CT findings with histologic specimens, and this study is widely cited as providing proof that MRI and CT underestimate the true extent of tumor involvement. More importantly, it showed that T2 changes, previously thought to be edema from the tumor, were often also tumor tissue.
This important study is frequently cited, but how many people have critically reviewed the numbers in this study is unclear, because these numbers show that regardless of grade, MRI does not miss many tumor cells. For example, of the 117 biopsies containing infiltrating tumor cells, only 18 (15%) were found in regions with a normal T1 signal. Higher sensitivity was found for T2 images, with 5 of 128 (4%) specimens with infiltrating tumor cells found in regions with a normal T2 signal. Thus, although tumor infiltration outside the area of tumor tissue does occur, critical analysis of this important data set shows that regardless of tumor grade, if the areas of abnormal T2 signal are resected, most of the infiltrating cells are removed in most patients. Furthermore, taking a slight margin is likely removing an even higher number of cells. Thus, in defining this report in a binary fashion, instead of a probabilistic one, many practitioners justified discontinuing glioma surgery.
Pallud and colleagues, 2010
Similar to the study by Kelly and colleagues, Pallud and colleagues reported on 16 patients with low-grade oligodendrogliomas studied with serial stereotactic biopsies performed using a rigorous paradigm. They increased their histologic yield through using mindbomb homolog 1 (MIB-1) labeling to identify mitotic cells that may have been missed by simple histology. The investigators show that although tumor cells are frequently found outside the MRI region, they are seldom found far from it in these low-grade tumors. More specifically, they did not find evidence that MIB-1–positive cells are located outside of a 2-cm perimeter around the tumor at a higher rate than in normal brain. Thus, the investigators concluded that low-grade oligodendrogliomas are locally infiltrative to approximately 2 cm but are probably not widely disseminated at initial diagnosis. This idea of the local 2- to 3-cm margin of tumor infiltration is similar to the observation of Burger and colleagues, who found that these tumors are frequently locally infiltrative at diagnosis.
Is an aggressive surgical philosophy warranted?
Based on the earlier discussion, the authors suggest that one can safely conclude that, although many gliomas are widely disseminated at end stage, sparse evidence suggests that they are always widely disseminated far beyond the imaging-defined tumor borders at initial diagnosis. At the very least, one can reasonably hypothesize that some patients present with localized lesions at diagnosis. Beyond just a simple lack of evidence, reason exists to also disbelieve that these gliomas are widely disseminated in initial stages. First, recurrent disease is classically seen as local recurrence at the margins, with multifocal disease and cerebrospinal fluid dissemination much less common. More importantly, cells distant from the primary tumor at diagnosis would lie outside of conventional radiotherapy fields, and thus distant foci would be expected to grow faster than more proximally located cells. The fact that, despite this, most recurrences occur in the margins of the lesionectomy cavity suggests that the wide-flung cells are not what kills patients with glioma, but rather the inadequately treated margins.
Furthermore, associating widespread invasion and dissemination with early-stage tumors is simply incompatible with current understanding of cancer biology, because these tumors usually need to acquire several mutations to gain those abilities, which takes time and possibly exposure to DNA-altering therapies, such as radiation and alkylating chemotherapy. Regardless, with any tumor, one can logically hypothesize that outward radial spread from the primary tumor mass should lead to a decreasing probability of finding a cancer cell with increasing distance from the primary site. In other words, even if the tumor had invasive properties uniformly at initial presentation, it takes time for tumor cells to reach distant sites, and thus the probability of finding a cancer cell infiltrating seemingly normal brain is likely not equal for distant sites and the margins.
In addition, a growing body of evidence shows that aggressive lesionectomy is superior to subtotal lesionectomy and biopsy. Class 1 evidence includes the randomized experience with 5-ALA–guided resections, which showed that gross total resection conferred improved survival compared with less-aggressive lesionectomy. Additionally, the landmark trial by Stupp and colleagues reporting a survival benefit with temozolomide and radiotherapy over radiotherapy alone showed a 6-month survival benefit with surgery over biopsy alone in both arms of the trial, although the study did not aim to address this comparison. Meta-analyses of class 2 and 3 data have suggested that aggressive surgery confers a survival benefit regardless of grade. Similarly, two recent well-executed retrospective volumetric studies have shown a survival benefit for patients with glioblastoma receiving aggressive complete or near-complete resections compared with less-aggressive resections. A large study of low-grade astrocytomas similarly showed a stepwise improvement in 8-year survival for patients with greater than 90% resection compared with those with lesser resections (91% vs 60%). In addition to extending survival, complete removal has been shown to improve seizure control in both retrospective studies and meta-analyses. Thus, although not a cure, aggressive lesionectomy improves survival and renders biopsy and radiation alone an archaic strategy for treating most of these patients.
Related posts:
Characteristics and Treatment of Seizures in Patients with High-Grade Glioma: A Review
Radiation Options for High-Grade Gliomas
Pathology: Commonly Monitored Glioblastoma Markers: EFGR, EGFRvIII, PTEN, and MGMT
The Role of BCNU Polymer Wafers (Gliadel) in the Treatment of Malignant Glioma
Alternative Chemotherapeutic Agents: Nitrosoureas, Cisplatin, Irinotecan
Temozolomide and Other Potential Agents for the Treatment of Glioblastoma Multiforme
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
