Brain Cancer Progression: A Retrospective Multicenter Comparison of Awake Craniotomy Versus General Anesthesia in High-grade Glioma Resection : Journal of Neurosurgical Anesthesiology

Secondary Logo

Journal Logo

Clinical Investigations

Brain Cancer Progression: A Retrospective Multicenter Comparison of Awake Craniotomy Versus General Anesthesia in High-grade Glioma Resection

Chowdhury, Tumul MD, DM, FRCPC*,†; Gray, Kristen; Sharma, Mohit MSc, MD; Mau, Christine MD§; McNutt, Sarah§; Ryan, Casey§; Farou, Noa§; Bergquist, Patrick§; Caldwell, Catherine§; Uribe, Alberto A. MD, MS, MPHc; Todeschini, Alexandre B.; Bergese, Sergio D. MD, FASA∥,¶; Bucher, Oliver MSc#; Musto, Grace BSc#; Azazi, Emad Al MD, PhD*; Zadeh, Gelareh MD, PhD, FRCSC**; Tsang, Derek S. MD, MSc, FRCPC††; Mansouri, Seyed A. MD, MSc, FRCSC‡‡; Kakumanu, Saranya MD, FRCPC§§; Venkatraghavan, Lashmi MD, DNB, FRCA, FRCPC*

Author Information
Journal of Neurosurgical Anesthesiology: October 2022 - Volume 34 - Issue 4 - p 392-400
doi: 10.1097/ANA.0000000000000778


Gliomas are the most common primary brain tumor and, depending on World Health Organization (WHO) grade and molecular subtype, hold a poor prognosis, primarily due to rapid cancer progression and recurrence.1 Much research has been dedicated to understanding factors that affect glioma progression, such as age, sex, tumor grade, size and location, and adjuvant treatment regimens. As surgical resection is an established first-line treatment for gliomas, and anesthesia is an integral component of brain cancer surgery, we believe that the effect of anesthetics on brain cancer progression should be considered in discussions surrounding glioma outcomes.

Recently, in vitro and animal studies have suggested that certain anesthetics, such as the volatile agents, may stimulate glioma and other cancer cell activity by enhancing their migratory ability, upregulating proteins involved in adapting to hypoxic stress, modulating neuroendocrine stress responses to surgery, and suppressing the immune system.2–5 Other studies have assessed the effects of anesthetics on progression-free survival (PFS) and overall survival (OS) in human populations by specifically comparing different general anesthesia (GA) techniques.6–8 Use of total intravenous anesthesia (TIVA) with propofol appears to offer a survival benefit compared with volatile anesthesia in nonbrain cancer patients; however, similar studies in glioma patients have been inconclusive.9

Specific to treatment of brain tumors is the choice between resection with awake craniotomy and minimal sedation as opposed to craniotomy with GA. Some studies on glioma resection have reported improved PFS and OS in patients having awake craniotomy compared with GA.10–12 Easier intraoperative tumor identification, avoidance of eloquent brain areas, and administration of less perioperative opioid on average (the use of which has been associated with cancer metastasis) during awake craniotomy are potential explanations for these findings.13–15 In addition, awake craniotomy virtually always requires scalp injection of lidocaine or other amide local anesthetics, which have been linked to superior cancer survival outcomes.16–18 Finally, awake craniotomy has been shown to attenuate surgical stress responses, with lower visual analogue scores and lower plasma levels of cortisol, adrenocorticotropic hormone, norepinephrine, and phenylalanine/tyrosine ratio postoperatively compared with GA.19–21 Thus, it appears that using lower amounts of anesthetics, as during awake craniotomy, might be associated with improved survival outcomes for glioma patients compared with deeper sedation, as during GA.10 However, none of these studies were specific to high-grade glioma patients, and were limited by low participant numbers.

We conducted a multicenter retrospective study to assess the effect of anesthetic technique on PFS and OS in patients undergoing high-grade (WHO grade III and IV) glioma resection. We hypothesized that high-grade glioma patients undergoing awake craniotomy for tumor resection would have longer PFS and OS, improved postoperative pain control, and shorter length of hospital stay than patients receiving GA.


A multicenter retrospective cohort study was performed in patients who underwent surgical resection of high-grade glioma with either GA or awake craniotomy. Patient data were collected from Health Sciences Centre and CancerCare Manitoba in Winnipeg, Canada, which was the primary center (ethics approval #HS22458, January 21, 2019); secondary centers were Toronto Western Hospital and Princess Margaret Cancer Centre, Toronto, Canada (approval #19-5197, April 8, 2019), Penn State, Hershey, PA (approval #STUDY00014013, March 17, 2020), and Ohio State University Wexner Medical Center, Columbus, OH (approval #2019H0158, May 3, 2019). The decision between craniotomy with GA versus awake craniotomy was at the neurosurgeon’s discretion in consultation with the neuroanesthesiologist. A retrospective chart review was conducted at all sites for all eligible patients that underwent glioma resection between January 1, 2007 and December 31, 2017. Postsurgical follow-up data were collected up to May 31, 2019. Deidentified data at each site were saved to an electronic data base file, while identifying data, including name, date of birth, and personalized health numbers, were saved in a secure master file at each site. The deidentified data base file was forwarded to the primary center via secure email, where the pooled data were analyzed.

Eligibility Criteria

Patients aged 18 years or older at the time of inclusion and had been diagnosed with high-grade glioma (WHO grade III and IV) were eligible for recruitment into this study; subtotal and gross resections were included. Patients were excluded if they had received chemotherapy or radiotherapy before resection, if a high-grade glioma had been resected before the inclusion period, if the tumor was biopsied only, if the tumor was not of glial cell origin, or if the patient was pregnant.

Exposures and Data Collection

The study had 2 groups—a GA group that included patients who had glioma surgery with GA and endotracheal intubation and an awake craniotomy group that included patients who had glioma surgery with minimal sedation. The dosing and anesthetic technique for minimal sedation could vary from patient to patient, but would be light enough so as not to require an advanced airway for the entirety of the operative case. Data pertaining to anesthesia technique was obtained from the anesthesiology records. Other recorded variables included patient demographics (sex and age), WHO tumor grade, pathologic diagnosis, molecular diagnosis with respect to isocitrate dehydrogenase (IDH) and O(6)-methylguanine DNA methyltransferase (MGMT) mutation status, 3-dimensional tumor size, tumor location, and preoperative Karnofsky Performance Status (a scale used to measure the quality of life of cancer patients).22 Intraoperative data included date and duration of surgery, type and duration of anesthetic (defined as time from when the patient entered to exited the operating room), use of intraoperative magnetic resonance imaging (MRI), whether the patient received a blood transfusion, and extent of tumor resection. Tumor size was extracted from surgical and oncology records or, if missing from these records, calculated from the preoperative MRIs by a radiation oncologist (S.K.). Extent of tumor resection was defined by the neurosurgeon in the surgical report by intraoperative imaging, and confirmed with postoperative imaging reports (extracted from oncology records). Anesthetic agents considered were propofol, sevoflurane, desflurane and dexmedetomidine. Intraoperative complications, including respiratory, cardiovascular and neurological complications, nausea/vomiting, and conversion from awake craniotomy to GA, were recorded. Data on postoperative complications, including deep vein thrombosis, cerebrovascular accident, sepsis, new-onset seizures, new-onset neurological deficits, intracerebral hemorrhage and edema, cerebrospinal fluid leak, myocardial infarction, and pneumonia, were also collected. Narcotic consumption within 1 hour postoperatively was recorded. Postsurgery chemotherapy and radiotherapy start dates and regimens were extracted from the oncology records. Postoperative treatment regimens were determined by the oncology team, and were based on factors such as age, postoperative functional status, histology, IDH mutation and MGMT methylation status, and response to radiotherapy. Treatments were recorded up to date of first progression or date of death, whichever occurred first.


The primary outcome was postsurgical PFS; this was chosen because cancer progression is virtually guaranteed for high-grade glioma patients and is arguably the most problematic complication of the disease. Date of cancer progression was recorded as the date when oncology records indicated consensus within the team of oncologists that the patient’s cancer had progressed. This could be radiologic or clinical progression, or marked by a change in treatment regimen due to disease progression. Secondary outcomes were OS, length of hospital stay, and postoperative pain score. OS was recorded as time between surgery and death or the last follow-up, depending on whether or not the patient was alive at postoperative follow-up review. Hospital length of stay was defined as time between surgery and discharge from hospital. Pain score was measured using the Visual Analog Scale or the Numerical Rating Scale, depending on the center. For both scales, 0 to 4 was defined as no or minimal pain, 5 to 6 as moderate or tolerable pain, and 7 to 10 as severe or intolerable pain.23 Upon the patient’s arrival at the postanesthesia care unit, nurses recorded the pain score as a standard of care.

Statistical Analysis

Power calculations were conducted using simulations with 1000 replications. The median PFS of the GA group was assumed to be 15 months. Initially, the distribution of GA and awake craniotomy procedures was assumed to be 90% and 10%, respectively. Assuming an expected cohort size of 1400 patients, an alpha of 0.05, power of 0.80, and accounting for an assumed intraclass correlation of 0.03 between the 4 sites using a design effect approach, a hazard ratio (HR) of 0.6 would be required. At the conclusion of data collection from all centers, the study power was revisited using the known proportions of awake craniotomy and GA (21% and 79%, respectively), sample size (n=891), and median PFS in the GA group (6 mo). On the basis of this updated information, a HR of 0.64 would be detectable with an alpha of 0.05 and power of 0.8.

Statistical analyses were conducted using STATA version 14.2. Data are presented as median and 95% confidence interval (CI) or median (range) with interquartile range (IQR) as appropriate. Descriptive statistics were used to examine the frequency and distribution of patient and treatment characteristics among individuals that underwent GA or awake craniotomy with minimal sedation. Kaplan-Meier curves were used to describe PFS and OS between variables of interest. Log-rank tests were used to compare groups for significant differences, with P values ≤0.05 indicative of significance.

Separate Cox regression models were used to investigate associations between variables of interest and PFS and OS, respectively. Before model building, data were landmarked to 70 and 123 days postsurgery, respectively, to account for immortal time bias due to differences in radiotherapy and adjuvant chemotherapy duration.24–27 The landmark times were selected based on the latest progression or death noted in the data. Variables of interest were screened for inclusion in multivariable models by testing univariable associations with log-rank tests and Likelihood ratio testing. Variables with P values ≤0.2 were eligible for inclusion. These variables were screened further for potential multicollinearity using Cramer V when categorical variables had >2 rows and columns. Values ≥0.30 were considered indicative of multicollinearity. Multicollinearity among continuous variables was investigated using Spearman rank correlation coefficient, with values ≥0.8 indicative of multicollinearity. Point-biserial correlation and η2 were used to investigate whether continuous variables were multicollinear with binary and categorical variables. Values ≥0.8 and 0.25 were considered indicative of multicollinearity, respectively. Apart from anesthetic type (GA vs. awake craniotomy), which was included in all multivariable models regardless of its significance, variables were included only if their likelihood ratio test P-value was ≤0.05. Continuous variables were modeled using restricted cubic splines, with Akaike’s information criteria used to determine the number of knots included in the spline function. Scaled Schoenfeld residual plots were examined to assess the assumption of proportional hazards while sensitivity analyses with complete positive and negative correlation were used to assess independent censoring. The presence of outlying and influence observations were assessed using deviance residuals, and score and scaled score residuals, respectively.

Pain scores at the time of admission to the postanesthesia care unit were compared by anesthetic type overall, and by anesthetic type and treatment center. Comparisons were made using the Wilcoxon-Mann Whitney test with P values ≤0.05 indicative of significance. In addition, 2 secondary analyses were performed to compare the survival characteristics of TIVA versus volatile subgroups within the GA group, as well as dexmedetomidine versus no dexmedetomidine subgroups in the awake craniotomy group.


Between the 4 centers, 1895 patients’ charts were screened of which 891 met eligibility criteria for inclusion in the study (Fig. 1). Reasons for exclusion were: age below 18 (2 patients), WHO grade I or II glioma (149), not of glial cell origin (520), radiation therapy before surgery (22), did not undergo surgery (144), surgery date outside inclusion period (6), extent of resection was biopsy only (67), and missing primary or secondary outcome data (94). Of the study population, 704 (79.0%) received GA and 187 (21.0%) received awake craniotomy with minimal sedation.

Study flow chart. AC indicates awake craniotomy; GA, general anesthesia; WHO, World Health Organization.

Study Population and Preoperative Characteristics

The characteristics of the 2 groups are shown in Supplemental Digital Content 1 (Supplementary Table 1: Descriptive characteristics for the whole cohort, The 2 groups were similar with respect to age, sex, location of tumor, and preoperative Karnofsky Performance Status. The GA group had a larger proportion of WHO grade IV tumors compared with the awake craniotomy group (85.1% vs. 74.9%, respectively; P<0.001), and median (range) IQR tumor size was larger in the GA group (28.6 [0.4 to 360.0], IQR 46.2 cm3) than in the awake craniotomy group (23.6 [0.2 to 179.1], IQR 32.8 cm3; P<0.002). More GA patients were diagnosed with glioblastoma compared with awake craniotomy patients (85.1% vs. 75.9%, respectively; P<0.036). IDH and MGMT mutation status were unknown in the majority of the study population.

Anesthetic Agents and Perioperative Data

In the GA group, 36 (5.1%) patients received TIVA, 391 (55.5%) received volatile anesthetics and 277 (39.4%) received a combination of intravenous and volatile anesthetics. All 187 patients in the awake craniotomy group received TIVA with propofol, and 45 (24.1%) of these patients also received dexmedetomidine. Duration of anesthesia was shorter in the awake craniotomy group (180 [90 to 727], IQR 100 min) compared with the GA group (240 [73 to 860], IQR 124 min; P<0.001). Awake craniotomy was associated with more intraoperative complications compared with GA (9.1% vs. 5.1%, respectively; P<0.041). Median postoperative narcotic consumption was lower in the awake craniotomy group (0 [0 to 25], IQR 5) than in the GA group (2.5 [0 to 62], IQR 10; P<0.001). There were no significant between group differences with respect to use of intraoperative MRI, blood transfusion, and extent of tumor resection.

Postoperative Follow-up Data

In terms of postoperative cancer treatment, a larger subset of GA patients received ≥60 Gy of radiation compared with awake craniotomy patients (35.1% vs. 23.0%). Three hundred sixty-eight (52.3%) patients in the GA group received <60 Gy compared with 117 (62.6%) in the awake craniotomy group; 180 (11.4%) GA patients did not receive any radiation compared with 11 (5.9%) in the awake craniotomy group (P<0.001). Radiation data for some patients was missing (1.3% and 8.6% in the GA and awake craniotomy groups, respectively). Patients receiving GA had less adjuvant chemotherapy than those having awake craniotomy group. Two hundred fifty-nine (36.8%) GA patients did not receive any adjuvant chemotherapy compared with 48 (25.7%) awake craniotomy patients; 288 (40.9%) GA group patients received <6 cycles, compared with 89 (47.59%) awake craniotomy group patients, and 145 (20.6%) GA group patients had ≥6 cycles compared with 36 (19.3%) in the awake craniotomy group (P<0.035). Twelve (1.70%) patients in the GA group and 14 (7.49%) in the awake craniotomy group had chemotherapy data missing from their charts. Further details of postoperative cancer treatment are available in Supplemental Digital Content 1 (Supplementary Table 1,

Primary and Secondary Outcomes

The primary outcome, PFS, was not statistically different between groups. Median PFS in the awake craniotomy group was 0.54 (95% CI: 0.45-0.65) years compared with 0.53 years (95% CI: 0.48-0.60) in the GA group; HR 1.05; P=0.553 (Table 1, Fig. 2). OS was greater in the awake craniotomy group with a median survival of 1.7 (95% CI: 1.30-2.32) years compared with 1.25 (95% CI: 1.15-1.37) years in the GA group; HR 0.76; P<0.009 (Table 2, Fig. 2). Further details of PFS and OS are available in Supplemental Digital Content 2 (Supplementary Table 2: PFS and OS by anesthetic type, Multivariable analyses of the association between anesthesia type (GA or awake craniotomy) and PFS and OS did not show a significant difference for either outcome after controlling for other variables of interest (Supplemental Digital Content 3, Supplementary Table 3: Multivariable Cox regression for PFS and OS by anesthesia type, Median length of hospital stay was significantly shorter in the awake craniotomy group (2 [0 to 76], IQR 3 d vs. 5 [0 to 98], IQR 5 for awake craniotomy and GA groups, respectively; P<0.001). Pain scores were comparable between groups (Supplemental Digital Content 4, Supplementary Table 4: Pain score and length of hospital stay,

TABLE 1 - Univariable Cox Regression for Progression-free Survival
Hazard Ratio P 95% CI
Anesthesia (reference=general)
 Awake 1.05 0.553 0.89-1.24
 Age (10 y)* 1.36 <0.001 1.3-1.43
WHO grade (reference=III)
 IV 2.99 <0.001 2.42-3.69
Pathologic diagnosis (reference=anaplastic astrocytoma)
 Anaplastic oligodendroglioma 0.70 0.094 0.47-1.06
 Glioma 2.74 <0.001 2.12-3.53
IDH (reference=IDH-1)
 IDH-WT 1.74 <0.001 1.42-2.12
Preoperative KPS (reference=<70%)
 ≥70% 0.69 <0.001 0.59-0.80
Treatment group (reference=no RT)
 Chemotherapy+RT followed by adjuvant chemotherapy 0.35 <0.001 0.28-0.43
 RT alone followed by adjuvant chemotherapy 0.48 0.007 0.29-0.82
 Chemotherapy+RT alone 0.56 <0.001 0.43-0.73
 RT alone 0.39 <0.001 0.29-0.51
Adjuvant chemotherapy (reference=none)
 <6 cycles 0.76 <0.001 0.65-0.89
 ≥6 cycles 0.39 <0.001 0.32-0.48
MGMT (reference=positive)
 Negative 1.07 0.580 0.84-1.37
Tumor location (reference=infratentorial)
 Multifocal 1.75 0.039 1.03-2.97
 Supratentorial 1.47 0.126 0.90-2.42
Blood transfusion (reference=no)
 Yes 0.82 0.587 0.41-1.65
Intraoperative complications (reference=no)
 1-3 complications 1.12 0.446 0.84-1.49
Radiation (reference=none)
 <60 Gy 0.33 <0.001 0.26-0.42
 ≥60 Gy 0.31 <0.001 0.24-0.40
Chemotherapy+RT (reference=no)
 Yes 0.65 <0.001 0.55-0.77
Extent of resection (reference=gross total)
 Subtotal 1.09 0.252 0.94-1.25
Sex (reference=male)
 Female 0.98 0.751 0.85-1.13
 Tumor size (cm3) 1.00 0.623 1.00-1.00
 Duration of anesthetics (30 min) 0.96 <0.001 0.94-0.98
*Risk associated with every 10-year increase in age.
Risk associated with every 30-minute increase in duration of anesthetics.
CI indicates confidence interval; IDH, isocitrate dehydrogenase; KPS, Karnofsky Performance Status; MGMT, O(6)-methylguanine DNA methyltransferase; RT, radiation therapy; WHO, World Health Organization; WT, wild type.

Kaplan-Meier survival curve for general anesthesia and awake craniotomy. A, Progression-free survival (P=0.553). B, Overall survival (P=0.009).
TABLE 2 - Univariable Cox Regression for Overall Survival
Hazard Ratio P 95% CI
Anesthesia (reference=general)
 Awake 0.76 0.009 0.62-0.93
 Age (10 y)* 1.54 <0.001 1.45-1.64
WHO grade (reference=III)
 IV 3.99 <0.001 3.05-5.22
Pathologic diagnosis (reference=anaplastic astrocytoma)
 Anaplastic oligodendroglioma 0.87 0.599 0.51-1.48
 Glioma 3.90 <0.001 2.79-5.44
IDH (reference=IDH-1)
 IDH-WT 2.47 <0.001 1.92-3.17
Preoperative KPS (reference=<70%)
 ≥70% 0.60 <0.001 0.51-0.72
Treatment group (reference=no RT)
 Chemotherapy+RT followed by adjuvant chemotherapy 0.30 <0.001 0.24-0.38
 RT alone followed by adjuvant chemotherapy 0.44 0.004 0.25-0.77
 Chemotherapy+RT alone 0.53 <0.001 0.40-0.71
 RT alone 0.39 <0.001 0.28-0.54
Adjuvant chemotherapy (reference=none)
 <6 cycles 0.63 <0.001 0.53-0.75
 ≥6 cycles 0.30 <0.001 0.24-0.38
MGMT (reference=positive)
 Negative 1.47 0.013 1.09-2.00
Tumor location (reference=infratentorial)
 Multifocal 2.70 0.007 1.31-5.59
 Supratentorial 2.08 0.040 1.04-4.19
Blood transfusion (reference=no)
 Yes 1.03 0.935 0.49-2.17
Intraoperative complications (reference=no)
 1-3 complications 1.22 0.207 0.90-1.66
Radiation (reference=none)
 <60 Gy 0.28 <0.001 0.22-0.36
 ≥60 Gy 0.28 <0.001 0.22-0.37
Chemotherapy+RT (reference=no)
 Yes 0.54 <0.001 0.45-0.65
Extent of resection (reference=gross total)
 Subtotal 1.27 0.004 1.08-1.49
Sex (reference=male)
 Female 1.01 0.928 0.86-1.18
 Tumor size (cm3) 1.00 0.389 1.00-1.00
 Duration of anesthetics (30 min) 0.97 0.008 0.95-0.99
*Risk associated with every 10-year increase in age.
Risk associated with every 30-minute increase in duration of anesthetics.
CI indicates confidence interval; IDH, isocitrate dehydrogenase; KPS, Karnofsky Performance Status; MGMT, O(6)-methylguanine DNA methyltransferase; RT, radiation therapy; WHO, World Health Organization; WT, wild type.

A flow diagram showing landmark cohort selection is available in Supplemental Digital Content 5 (Supplementary Fig. 1: Selection of landmark cohorts for the study, The landmark analyses yielded similar results to analysis of the entire cohort. The PFS landmark cohort had a median PFS of 0.48 (95% CI: 0.43-0.57) years in the GA group compared with 0.52 (95% CI: 0.36-0.63) years in the awake craniotomy group; HR 1.07; P<0.475 (Supplemental Digital Content 6, Supplementary Table 5: Univariable and multivariable Cox regression for the PFS landmark cohort, Median survival in the OS landmark cohort was 1.18 (95% CI: 1.03-1.37) years and 1.83 years (95% CI: 1.24-2.35) in the GA and awake craniotomy groups, respectively; HR 0.81, P<0.056 (Supplemental Digital Content 7, Supplementary Table 6: Univariable and multivariable Cox regression for the OS landmark cohort,

Secondary Analyses

A Kaplan-Meier analysis comparing TIVA to volatile anesthetics within the GA group showed no significant difference in PFS (P<0.831) or OS (P<0.685) (Fig. 3). Further details are available in Supplemental Digital Content 8 (Supplementary Table 7: PFS and OS by GA type, Dexmedetomidine use in the awake craniotomy group was also not associated with any significant differences in PFS (P<0.312) or OS (P<0.900) (Supplemental Digital Content 9, Supplementary Table 8: PFS and OS by dexmedetomidine use in the awake craniotomy group, Finally, year of surgery had no effect on survival outcomes for either group (P<0.946) (Supplemental Digital Content 10, Supplementary Fig. 2: HRs describing the relationship between year of surgery for awake and GA and PFS,

Kaplan-Meier survival curve for general anesthesia type. A, Progression-free survival (P=0.831). B, Overall survival (P=0.685). TIVA indicates total intravenous anesthesia.


This study found no significant association between postresection PFS in high-grade glioma patients and anesthesia type (GA or awake craniotomy). OS was significantly longer in patients having awake craniotomy compared with those having GA in the univariable analysis. Notably, this significant association disappeared after controlling for other variables of interest, demonstrating a positive but nonstatistically significant association between awake craniotomy and OS. Awake craniotomy has been associated with improved OS in several studies, a finding predominantly attributed to larger extent of resection with awake craniotomy.12,28 Martino et al29 compared awake craniotomy with intraoperative electrical stimulation to craniotomy under GA without intraoperative electrical stimulation, and reported significantly longer PFS in the awake craniotomy group. However, other studies in the past decade have not reported a survival advantage of awake craniotomy compared with GA for glioma resection.10 As previously mentioned, it is worth noting that none of these studies were specific to high-grade glioma patients, some had potential for considerable confounding and patient numbers were relatively low in all of them.

Although the results of the current study were not markedly in favor of awake craniotomy, they do provide evidence that the awake technique may be superior in terms of OS. There are several possible explanations for this finding. Studies comparing GA and awake craniotomy for glioma surgery have shown significantly higher intraoperative plasma levels of epinephrine during awake craniotomy, but higher postoperative plasma levels of stress hormones, such as cortisol and phenylalanine/tyrosine ratio, after GA.19–21 It is plausible that a lower stress response during awake craniotomy confers a survival benefit for these patients. Furthermore, awake craniotomy allows for intraoperative identification of eloquent brain regions, thus helping to preserve neurological function and minimize functional decline; it also typically allows for greater percentage of tumor resection (although this was not reflected in our data).9,15 In addition, there is consistent use of lidocaine and other amide local anesthetics for scalp block when using minimal sedation, whereas local anesthetic use during GA may depend on individual neurosurgeon’s or neuroanesthetist’s preferences.15 There is evidence of longer PFS for high-grade glioma patients who receive a scalp block intraoperatively,17,18 with potential mechanisms including an opioid-sparing effect of local anesthesia, attenuation of the stress response, and a direct inhibitory effect of local anesthetics on glioma cells.16–18 Our study did not include data capture of scalp blocks or intravenous lidocaine use, and these would have been interesting and relevant variables. Related to this, awake craniotomy has been associated with reduced postoperative opioid consumption, which was also demonstrated in our study. Opioids have been shown to inhibit natural killer cell activity and have also been implicated in angiogenesis facilitating wound healing, suggesting a link to tumor progression.30–32

The landmark dates in this study were based on the latest time to progression for the PFS outcome (70 d), and the latest time to death for the OS outcome (123 d), within each of the different treatment categories. The purpose was to provide a more homogenous cohort by excluding patients that did not survive up to the landmark date, thereby correcting for differences in individuals that did not receive adjuvant therapy because they were too unwell as opposed to those that did not receive therapy because they did not want or require it. However, in our multivariable analyses this adjustment did not impact the association between anesthetic type and PFS or OS after accounting for other variables of interest.

Interestingly, there was no statistical difference in PFS or OS between TIVA and volatile GA subgroups. This contradicts multiple previous nonbrain cancer studies which have repeatedly demonstrated a survival benefit of TIVA with propofol over inhaled anesthesia.6–8 The in vitro prometastatic effects of volatile agents compared with various anticancerous effects of propofol have been proposed as potential mechanisms.2–5,9,33–35 However, the limited research conducted specific to glioma has not yielded similar results. In their retrospective cohort study, Dong et al36 found no significant difference in PFS and OS between sevoflurane and propofol anesthesia, although subgroup analysis did associate propofol with increased OS in patients with poor preoperative Karnofsky Performance Status. Cata et al37 compared the effects of isoflurane and desflurane alone and in combination with propofol on PFS and OS for glioblastoma patients, and found no statistical difference between any of the groups. Despite experimental data suggesting procancerous effects of volatile agents, our study did not identify any significant difference in PFS and OS between patients having anesthesia maintained with TIVA or a volatile agent. However, given the exceptionally small size of the TIVA group (36 patients) this is unsurprising.

Similarly, dexmedetomidine did not appear to influence survival outcomes, although this could also be due to the small size of this subgroup (45 patients). Dexmedetomidine is a selective α2-adrenergic agonist that is used during anesthesia for anxiolysis, analgesia and, primarily, for procedural sedation including its successful use for minimal sedation during awake craniotomy.38 In vitro and in vivo models of nonbrain cancers have also demonstrated antiapoptotoic and neuroprotective roles for dexmedetomidine, leading to increased metastatic burden.39,40 It is, however, possible that in vitro and in vivo results do not translate to humans, or that the interaction of dexmedetomidine with high-grade gliomas is not the same as that demonstrated with other cancer cell lines.

There are several important limitations to this study. First, as this was a retrospective chart review there was a moderate amount of missing data, particularly IDH and MGMT mutation status and postoperative radiotherapy regimens. Second, the potential risk for human error during data collection and transcription, in addition to poor quality of accessible data charted in the medical records, are also limitations linked to the retrospective nature of the study. Third, the lack of standardization of clinical care and of concentration and volume of anesthetics among all cases, as well as lack of randomization, may lead to uncontrolled and unrecognized biases such as unpredicted confounding factors and selection bias. Fourth, a general preference by clinicians to use GA rather than awake craniotomy led to an uneven number of participants in the 2 groups. Although these circumstances did not lead to statistically significant differences, there was variability across institutions in the treatment and management of patients included in the study. Patients with glioblastoma or WHO grade IV tumors tended to have more resections under GA, which may have contributed to inferior OS in this group. Fifth, despite conducting a multivariable model analysis incorporating many variables, we cannot predict immeasurable confounding factors that might influence the results. Sixth, the study did not collect detailed information related to surgical techniques or extent of tumor resection; therefore, we cannot exclude the possibility that this variability influenced the study’s findings. Seventh, time between surgery and first postoperative follow-up with oncology teams was not recorded, which could cause delay of identification of recurrence in some patients. Finally, the cause-of-death information was not available or collected in our study; all-cause mortality rather than cancer-related mortality was used as an outcome measure.


In summary, the current study did not demonstrate any significant difference in PFS in patients having glioma resection with awake craniotomy or GA. While it did suggest a moderate but significant improvement in OS in patients undergoing awake craniotomy when compared with GA, this association disappeared when controlling for other variables of interest. Although estimates of PFS and OS in this study can provide clinically relevant information to physicians who require an idea of clinical effectiveness when planning patient care, given the limitations of the study our results should not be generalized to medical centers with different patient profiles and different perioperative practices. Further randomized controlled trials examining the use of awake craniotomy with minimal sedation and GA for glioma resection should be conducted to achieve a more comprehensive understanding of the most favorable anesthesia management in this patient population.


1. Stupp R, Mason WP, van den Bent MJ, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med. 2005;352:987–996. doi: 10.1056/NEJMoa043330
2. Shi QY, Zhang SJ, Liu L, et al. Sevoflurane promotes the expansion of glioma stem cells through activation of hypoxia-inducible factors in vitro. Br J Anaesth. 2015;114:825–830. doi: 10.1093/bja/aeu402
3. Masoud GN, Li W. HIF-1α pathway: Role, regulation and intervention for cancer therapy. Acta Pharm Sin B. 2015;5:378–389. doi: 10.1016/j.apsb.2015.05.007
4. Lai R, Shan W, Zhou D, et al. Sevoflurane promotes migration, invasion, and colony-forming ability of human glioblastoma cells possibly via increasing the expression of cell surface protein 44. Acta Pharmacol Sin. 2019. doi: 10.1038/s41401-019-0221-0
5. Zhu M, Li M, Zhou Y, et al. Isoflurane enhances the malignant potential of glioblastoma stem cells by promoting their viability, mobility in vitro and migratory capacity in vivo. Br J Anaesth. 2016;116:870–877. doi: 10.1093/bja/aew124
6. Wu Z-F, Lee M-S, Wong C-S, et al. Propofol-based total intravenous anesthesia is associated with better survival than desflurane anesthesia in colon cancer surgery. Anesthesiology. 2018;129:932–941. doi: 10.1097/ALN.0000000000002357
7. Jun I-J, Jo J-Y, Kim J-I, et al. Impact of anesthetic agents on overall and recurrence-free survival in patients undergoing esophageal cancer surgery: a retrospective observational study. Sci Rep. 2017;7:14020. doi: 10.1038/s41598-017-14147-9
8. Yap A, Lopez-Olivo MA, Dubowitz J, et al., Group the GO-ARC. Anesthetic technique and cancer outcomes: a meta-analysis of total intravenous versus volatile anesthesia. Can J Anesth. 2019;66:546–561. doi: 10.1007/s12630-019-01330-x
9. Gray K, Avitsian R, Kakumanu S, et al. The effects of anesthetics on glioma progression: a narrative review. J Neurosurg Anesthesiol. 2020. doi: 10.1097/ANA.0000000000000718
10. Lu VM, Phan K, Rovin RA. Comparison of operative outcomes of eloquent glioma resection performed under awake versus general anesthesia: a systematic review and meta-analysis. Clin Neurol Neurosurg. 2018;169:121–127. doi: 10.1016/j.clineuro.2018.04.011
11. Incekara F, Olubiyi O, Ozdemir A, et al. The value of pre- and intraoperative adjuncts on the extent of resection of hemispheric low-grade gliomas: a retrospective analysis. J Neurol Surg A Cent Eur Neurosurg. 2015;77:79–87. doi: 10.1055/s-0035-1551830
12. Duffau H, Lopes M, Arthuis F, et al. Contribution of intraoperative electrical stimulations in surgery of low grade gliomas: a comparative study between two series without (1985-96) and with (1996-2003) functional mapping in the same institution. J Neurol Neurosurg Psychiatry. 2005;76:845–851. doi: 10.1136/jnnp.2004.048520
13. Heaney Á, Buggy DJ. Can anaesthetic and analgesic techniques affect cancer recurrence or metastasis? Br J Anaesth. 2012;109:i17–i28. doi: 10.1093/bja/aes421
14. Aich A, Gupta P, Gupta K. Could perioperative opioid use increase the risk of cancer progression and metastases? Int Anesthesiol Clin. 2016;54:e1–e16. doi: 10.1097/AIA.0000000000000112
15. Gupta DK, Chandra PS, Ojha BK, et al. Awake craniotomy versus surgery under general anesthesia for resection of intrinsic lesions of eloquent cortex—a prospective randomised study. Clin Neurol Neurosurg. 2007;109:335–343. doi: 10.1016/j.clineuro.2007.01.008
16. Leng T, Lin S, Xiong Z, et al. Lidocaine suppresses glioma cell proliferation by inhibiting TRPM7 channels. Int J Physiol Pathophysiol Pharmacol. 2017;9:8–15.
17. Zheng L, Hagan KB, Villarreal J, et al. Scalp block for glioblastoma surgery is associated with lower inflammatory scores and improved survival. Minerva Anestesiol. 2017. doi: 10.23736/S0375-9393.17.11881-X
18. Sung C-H, Tsuang F-Y, Shih C-C, et al. Scalp block is associated with improved recurrence profiles in patients undergoing primary glioma resection surgery. J Neurosurg Anesthesiol. 2019. doi: 10.1097/ANA.0000000000000664
19. Shinoura N, Yamada R. Stress hormone levels in awake craniotomy and craniotomy under general anesthesia. J Neurol Neurophysiol. 2014. doi: 10.4172/2155-9562.1000256
20. Klimek M, Hol JW, Wens S, et al. Inflammatory profile of awake function-controlled craniotomy and craniotomy under general anesthesia. Mediators Inflamm. 2009;2009:1–8. doi: 10.1155/2009/670480
21. Hol JW, Klimek M, van der Heide-Mulder M, et al. Awake craniotomy induces fewer changes in the plasma amino acid profile than craniotomy under general anesthesia. J Neurosurg Anesthesiol. 2009;21:98–107. doi: 10.1097/ANA.0b013e318192d4aa
22. Performance Scales: Karnofsky & ECOG Scores Practice Tools|OncologyPRO. Educational Portal for Oncologists. 2019. Available at: Accessed July 2, 2019.
23. Hawker GA, Mian S, Kendzerska T, et al. Measures of adult pain: Visual Analog Scale for Pain (VAS Pain), Numeric Rating Scale for Pain (NRS Pain), McGill Pain Questionnaire (MPQ), Short-Form McGill Pain Questionnaire (SF-MPQ), Chronic Pain Grade Scale (CPGS), Short Form-36 Bodily Pain Scale (SF). Arthritis Care Res (Hoboken). 2011;63(suppl 11):S240–S252. doi: 10.1002/acr.20543
24. Van Walraven C, Davis D, Forster AJ, et al. Time-dependent bias was common in survival analyses published in leading clinical journals. J Clin Epidemiol. 2004;57:672–682. doi: 10.1016/j.jclinepi.2003.12.008
25. Suissa S. Immortal time bias in observational studies of drug effects. Pharmacoepidemiol Drug Saf. 2007;16:241–249. doi: 10.1002/pds.1357
26. Beyersmann J, Gastmeier P, Wolkewitz M, et al. An easy mathematical proof showed that time-dependent bias inevitably leads to biased effect estimation. J Clin Epidemiol. 2008;61:1216–1221. doi: 10.1016/j.jclinepi.2008.02.008
27. Agarwal P, Moshier E, Ru M, et al. Immortal time bias in observational studies of time-to-event outcomes: assessing effects of postmastectomy radiation therapy using the National Cancer Database. Cancer Control. 2018;25:1073274818789355. doi: 10.1177/1073274818789355
28. Sacko O, Lauwers-Cances V, Brauge D, et al. Awake craniotomy vs surgery under general anesthesia for resection of supratentorial lesions. Neurosurgery. 2011;68:1192–1199. doi: 10.1227/NEU.0b013e31820c02a3
29. Martino J, Gomez E, Bilbao JL, et al. Cost-utility of maximal safe resection of WHO grade II gliomas within eloquent areas. Acta Neurochir (Wien). 2013;155:41–50. doi: 10.1007/s00701-012-1541-8
30. Forget P, De Kock M. Could anaesthesia, analgesia and sympathetic modulation affect neoplasic recurrence after surgery? a systematic review centred over the modulation of natural killer cells activity. Ann Fr Anesth Reanim. 2009;28:751–768. doi: 10.1016/j.annfar.2009.07.078
31. Stefano GB, Kream RM, Mantione KJ, et al. Endogenous morphine/nitric oxide-coupled regulation of cellular physiology and gene expression: implications for cancer biology. Semin Cancer Biol. 2008;18:199–210. doi: 10.1016/j.semcancer.2007.12.003
32. Bharati S, Chowdhury T, Bergese S, et al. Anesthetics impact on cancer recurrence: What do we know? J Cancer Res Ther. 2016;12:464. doi: 10.4103/0973-1482.148670
33. Iwasaki M, Zhao H, Jaffer T, et al. Volatile anaesthetics enhance the metastasis related cellular signalling including CXCR2 of ovarian cancer cells. Oncotarget. 2016;7:26042–26056. doi: 10.18632/oncotarget.8304
34. Yang N, Liang Y, Yang P, Yang T, Jiang L. Propofol inhibits lung cancer cell viability and induces cell apoptosis by upregulating microRNA-486 expression. Braz J Med Biol Res. 2017;50. doi: 10.1590/1414-431x20165794
35. Gao Y, Yu X, Zhang F, Dai J. Propofol inhibits pancreatic cancer progress under hypoxia via ADAM 8. J Hepatobiliary Pancreat Sci. 2019;26:219–226. doi: 10.1002/jhbp.624
36. Dong J, Zeng M, Ji N, et al. Impact of anesthesia on long-term outcomes in patients with supratentorial high-grade glioma undergoing tumor resection: a retrospective cohort study. J Neurosurg Anesthesiol. 2019. doi: 10.1097/ANA.0000000000000588
37. Cata JP, Hagan KB, Bhavsar SDO, et al. The use of isoflurane and desflurane as inhalational agents for glioblastoma surgery: a survival analysis. J Clin Neurosci. 2017;35:82–87. doi: 10.1016/j.jocn.2016.10.006
38. Prontera A, Baroni S, Marudi A, et al. Awake craniotomy anesthetic management using dexmedetomidine, propofol, and remifentanil. Drug Des Devel Ther. 2017;11:593–598. doi: 10.2147/DDDT.S124736
39. Lavon H, Matzner P, Benbenishty A, et al. Dexmedetomidine promotes metastasis in rodent models of breast, lung, and colon cancers. Br J Anaesth. 2018;120:188–196. doi: 10.1016/j.bja.2017.11.004
40. Wang C, Datoo T, Zhao H, et al. Midazolam and dexmedetomidine affect neuroglioma and lung carcinoma cell biology in vitro and in vivo. Anesthesiology. 2018;129:1000–1014. doi: 10.1097/ALN.0000000000002401

glioma; craniotomy; anesthetics; volatile; TIVA; propofol; progression; survival; cancer

Supplemental Digital Content

Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.