This survey demonstrates crowdsourcing’s ability to provide insights into complex neurosurgical debates. The neurosurgical oncologists who completed this survey varied significantly in their practice setting and focus, as well as their personal experience performing SpTR for GBM. However, consensus was still reached in this large, diverse group of neurosurgical oncologists on a number of questions about SpTR for GBMs in specific anatomical regions. This encourages further use of the “wisdom of the crowd” approach to inform decisions regarding patient care where no one treatment has been conclusively proven superior [32, 34].
These results also provide support for our previously proposed consensus definition of SpTR for GBM [16]. There is a considerable need to adopt a single definition, as the multitude of definitions currently in use has obscured the impact of SpTR on OS in GBM. The 2018 systematic review and meta-analysis by Jackson et al. including 11 studies and 810 patients reported a survival benefit with SpTR over GTR in 9 of 11 studies (SpTR median OS 12-54 months; GTR median OS 11-17.5 months) [14]. However, the authors noted significant heterogeneity between the methods used in the included studies, and secondarily elected to divide studies into three subgroups based on their definition of SpTR, including extent of T2 FLAIR resection, extended anatomical resection, and intra-operative 5-aminolevulinic acid (5-ALA) fluorescence-guided resection. Meta-analysis in the more homogenous extended anatomical resection subgroup demonstrated a statistically significant 35% lower risk of mortality among 88 GBM patients who underwent SpTR versus 95 patients who received GTR (HR = 0.65, 95% CI 0.47–0.91; p = 0.003). This study, the largest systematic review to date on the efficacy of SpTR, supports the selection of a single, anatomically-based definition for SpTR.
Similar to our previous survey of 21 neurosurgical oncologists with experience performing SpTR for GBM, “GTR plus any resection of non-contrast enhanced disease” was the most commonly positively endorsed definition, with a similar fraction of each cohort agreeing or strongly agreeing that it was an acceptable definition for SpTR. However, as previously discussed, this definition would not standardize clinical research involving SpTR, provide a quantifiable benchmark for what can and cannot be classified as SpTR, or allow a neurosurgeon to determine when SpTR has been achieved using intraoperative navigation [16]. For these reasons and in light of the corpus of literature that demonstrates GBMs most often recur within 2 cm of the original contrast-enhancing tumor’s margins, we recommend that “GTR plus resection 1-2 cm beyond contrast enhancement” be implemented as a standardized definition of SpTR for GBM in clinical practice and future research studies [15, 21, 35–40]. A consensus of the AANS/CNS Tumor Section members who completed this survey support the adoption of this definition.
Similar to the results of surveying neurosurgical oncologists with expertise performing SpTR for GBM, the AANS/CNS Tumor Section cohort considered right frontal and right anterior temporal GBMs to be located in non-eloquent regions, amenable to SpTR, and appropriate to enroll in a clinical trial comparing SpTR to GTR. Of note, while the Tumor Section cohort found the right frontal GBM appropriate to enroll in a SpTR vs. GTR clinical trial by consensus, it was mildly more reticent to choose SpTR as the treatment they would personally perform relative to the right anterior temporal GBM; while 52 (69.3%) neurosurgeons thought SpTR was the best treatment for the right anterior temporal GBM, a smaller though still considerable fraction (n=44, 58.7%) would perform SpTR of the right frontal GBM. This trend does not appear to be associated with the relative perceived eloquence of the involved tissues but may be because the practice of performing lobectomy for GBM in the right temporal lobe is relatively well-known and supported by literature that demonstrates improved survival without increased frequency of post-operative deficits, while studies on the effect of SpTR specifically in the right frontal lobe are less common [41–43].
When previously surveyed, the 21 SpTR experts also agreed by consensus that the left anterior temporal GBM was amenable to SpTR vs. GTR randomization. This current cohort of more general neurosurgical oncologists, however, did not reach the 70% threshold for consensus. Compared to those with SpTR expertise (n=15, 78.9%), a significantly smaller fraction of neurosurgical oncologists from the Tumor Section designated SpTR as their chosen treatment for this GBM (n=30, 40.0%, p=0.004). This relative reluctance of the general neurosurgical oncologists to perform SpTR in this area may be related to the perceived eloquence of surrounding tissue, which includes Broca’s and Wernicke’s areas, visual fibers, and auditory cortex. Inadvertently injuring adjacent tissue involved in language could have a severe impact on quality of life, but such deficits have also been associated with decreased survival [44, 45]. These risks can be reduced by performing an awake craniotomy with speech mapping for the resection of left anterior temporal GBMs [12, 46, 47]. Of the 72 respondents who selected which intraoperative adjunctive methods they would employ during the resection of a left anterior temporal GBM, 32 (44.4%) indicated that they would use this technique; 53 of 76 survey respondents (72.6%) reported that they routinely perform awake craniotomies with speech mapping in their practice. Increasing this modality’s usage during the resection of left anterior temporal GBMs may encourage the wider neurosurgical community to consider performing SpTR for GBM in this region.
While this crowdsourcing survey is valuable for demonstrating consensus among the broader neurosurgical community on the preferred treatment of right frontal and anterior temporal GBMs, it is equally useful for revealing the anatomic locations of GBM where there is less agreement on optimal treatment. The proposed treatments for right insular GBM varied widely within the AANS/CNS Tumor Section cohort. While some felt that pursuing SpTR was reasonable, others recommended less aggressive interventions including LITT and SRS; even a simple majority could not agree on a treatment plan. Historically, neurosurgeons have been hesitant to resect insular tumors due to this cortex’s eloquence, complex shape, and relationship to the internal and middle cerebral arteries [48]. However, as the relationship between extent of resection and OS in insular high-grade gliomas has become clearer, increasing extent of resection while reducing morbidity using improved microsurgical and intraoperative adjunctive techniques has become increasingly recommended [6, 7, 48, 49]. Alimohamadi et al., for instance, demonstrated the potential of adjuncts including fMRI, diffusion tensor imaging (DTI), and awake craniotomy with intraoperative speech and motor mapping to yield high extent of resection without increased mortality or new major neurologic morbidities [50]. Hervey-Jumper and Berger reinforce that awake craniotomy with cortical and subcortical stimulation mapping during insular glioma resection results in lower rates of neurological deficit than resections performed without mapping (with: 4.3%; without: 19%; p=0.008) [48]. Again, further study and more widespread adoption of these techniques may encourage more aggressive intervention for insular GBMs and longer survival times for these patients.
Limitations
The individuals who completed this survey were recruited via email invitation and were not financially compensated. If these neurosurgeons chose to participate because they hold strong opinions about SpTR, the internal validity of the results may have been influenced by response bias.
Participants were asked to make decisions about the treatment of GBM patients using only anatomical information. However, when making decisions about surgical treatment of GBM, clinical factors are necessarily taken into account. The cases included in this study were taken from non-elderly and functional patients, which may limit the applicability of the results to older or less robust populations. In addition, extent of resection is often dependent intra-operatively on data from 5-ALA fluorescence or other adjuncts. This information was not provided on the survey in order to directly investigate the relationship between proposed treatment plans and anatomy alone; however, this constraint also limited the results’ generalizability.