Impact of the extent of resection on the survival of patients with lower-grade gliomas using awake brain mapping

Purpose The aim of this study was to assess the effect of the extent of resection (EOR) of tumors on survival in a series of patients with lower-grade gliomas (LGGs) who underwent awake brain mapping. Methods We retrospectively analyzed 126 patients with LGGs in the dominant and non-dominant hemisphere who underwent awake brain surgery at the same institution between December 2012 and May 2020. Results The median progression-free survival (PFS) rate of patients with LGGs in the group with an EOR > 100 %, including supratotal resection (n = 47; median survival [MS], not reached), was signicantly higher than that in the group with an EOR < 100% (n = 79; MS, 43.1 months; 95% CI: 37.8–48.4 months; p = 0.04). In patients with diffuse astrocytomas and anaplastic astrocytomas, the group with EOR > 100 %, including supratotal resection (n = 25; MS, not reached), demonstrated a signicantly better PFS rate than did the group with an EOR < 100% (n = 45; MS, 35.8 months; 95% CI: 19.9–51.6 months; p = 0.03). Supratotal or gross total resection was correlated with better PFS in IDH-mutant type of diffuse astrocytomas and anaplastic astrocytomas (n = 19; MS, not reached vs. n = 35; MS, 40.6 months; 95% CI: 22.3–59.0 months; p = 0.02). By contrast, supratotal or gross total resection was not associated with longer PFS rates in patients with IDH-wild type of diffuse astrocytomas and anaplastic astrocytomas.


Introduction
Lower-grade gliomas (LGGs)-World Health Organization (WHO) grade II and grade III brain tumors are diffusely in ltrative tumors of the central nervous system [1,2].This term has recently replaced the speci cation "lowgrade" gliomas, which include grade II gliomas, in clinical practice.Overtime, LGGs typically progress to WHO grade IV tumors, glioblastomas (GBMs), which are malignant tumors with a median survival of only 7.8-31 months, even with aggressive treatment such as surgery and chemoradiotherapy [3,4].The classi cation of LGGs into molecular subtypes in the new WHO 2016 classi cation has important prognostic implications [5].In the 2016 classi cation, isocitrate dehydrogenase (IDH) 1 and 2 mutations are key genetic events in LGGs as well as in GBMs [5].

Surgical tumor resection is the initial rst-line treatment to control
LGGs, because it can decrease tumor volume and the risk for malignant transformation of LGGs into highly malignant gliomas.Many studies have provided evidence supporting maximum safe surgical resection and early surgical intervention, which prolong survival in patients with LGGs [6][7][8][9][10][11][12][13].Although there are no well-designed randomized controlled clinical trials demonstrating a correlation between a higher extent of resection (EOR) and a better clinical prognosis of LGGs, complete radiological resection of LGGs is the currently recommended approach.
The goal of complete tumor resection should be balanced against neurological disturbances including motor, language, and neurocognitive impairments, due to the in ltrative behavior of LGGs [14,15].To ensure both maximal safe resection and preservation of these neurological functions, the use of intraoperative awake brain mapping techniques has been proposed as the reference standard strategy for patients with LGGs [16][17][18][19][20]. Awake surgery allows for the maximum degree of resection while determining functional boundaries, using both cortical and subcortical functional mapping [17,21].When functional boundaries are observed within the tumor mass, resection can be subtotal or partial.If the functional boundaries lie outside the tumor region, gross total or supratotal resection can be achieved.Extended tumor resection beyond the margins of the abnormal magnetic resonance imaging (MRI)-veri ed area is known as supratotal resection [22,23].Because there is only limited evidence for an effect of supratotal or gross total removal on the control of the tumor and histological malignant transformation [22,24,25], the oncological impact of the EOR on survival is still uncertain.Furthermore, it is unclear how far around the tumor mass resection can be safely extended to prolong the survival of patients with LGGs.There is also concern regarding whether an improvement in survival due to massive tumor resection depends on the patient's molecular genetic status, such as IDH1/2 mutations in patients with IDH-mutant type or IDH-wild type of LGGs.
In the present study, we performed awake brain mapping with cortical and subcortical stimulation for LGGs in order to achieve supratotal resection whenever possible, while determining the functional brain tissue boundaries beyond the tumor margins.We assessed the effect of tumor EOR on survival in a series of 126 consecutive patients with LGGs who all underwent awake brain mapping.In addition, we performed a subanalysis to assess the association between clinical outcome and tumor EOR in the subtypes of LGGs classi ed according to the molecular genetic status after awake surgery.

Study design
This is a retrospective analysis of 126 consecutive patients with LGGs.All underwent awake surgery with intraoperative direct electrical stimulation at Nagoya University Hospital (Nagoya, Japan) between December 2012 and May 2020.Patient data on clinical information and outcomes were collected and analyzed.The Ethics Committee at Nagoya University Hospital approved this retrospective data evaluation and the experimental design of the study (approval number: 2020-0079).We obtained written informed consent prior to the surgical mapping procedure from all participants included in the study.

Intraoperative awake brain mapping technique
All 126 patients underwent awake brain mapping with direct electrical stimulation using an asleep-awake-asleep protocol, as previously described [17,23,[26][27][28][29].In brief, we performed a craniotomy under general anesthesia.In the rst stage, the tumor margins and cerebral sulcus and gyrus were identi ed using a neuro-navigational system, and letter tags were placed along the tumor boundaries on the cortical surface before the brain shifts occurred.Cortical mapping with direct electrical stimulation was then performed to detect the motor and language areas.We used a bipolar stimulator with a 2-mm diameter and a 5-mm interelectrode distance (Unique Medical, Osaka, Japan) to deliver a biphasic current (pulse frequency, 60 Hz: single pulse phase duration, 0.5 ms).The Neuromaster MEE1200 (Nihon Kohden, Tokyo, Japan) was used for intraoperative neurological monitoring during the awake surgery.
The stimulation intensity used for individual patients was determined by increasing the amplitude in 0.5-mA increments from a baseline of 1.0 mA until a functional response was elicited.Maximum individual intensities ranged from 2 mA to 8 mA.After determining the optimal stimulation threshold, cortical mapping was performed by applying electrical stimulations of the same intensity over the whole exposed cortical area surface.This stimulation threshold was then also used for subcortical mapping.Using strip electrodes, continuous digital electrocorticograms (ECoGs) were monitored to detect after-discharges during direct electrical stimulation and tumor resection.
For language mapping, the patients were asked to perform counting and picture-naming tasks, with the goal to identify the cortical language sites and subcortical bers using direct brain stimulations [23].The type of observed language disturbances was determined based on a detailed classi cation for language disorders, including speech arrest, anomia, dysarthria, anarthria, speech slowness, initiation trouble, perseveration, and paraphasia.
After the cortical mapping was complete, tumor resection was initiated using the subpial resection technique.
We removed the tumor to the level of the white matter, while frequently checking the patient's response using subcortical electrical stimulation.Subcortical mapping enabled us to determine the functional boundaries between the tumor and white matter pathways.Thus, tumor removal was accomplished according to the cortical or subcortical functional boundaries in all patients, while aiming for achieving supratotal resection whenever possible.After tumor resection, intraoperative MRI was routinely performed to assess the EOR of the tumor using a 0.4 Tesla vertical eld MR scanner (Aperto Inspire, Hitachi, Tokyo, Japan) set up in the operating room of the Brain THEATER at Nagoya University Hospital.

Degree of tumor resection and volumetric analysis
Volumetric analysis was performed using the iPlan® cranial planning software included in the BrainLAB iPlan® Cranial version 2.6 and 3.0 [30] (German HealthCare Export Group, Bonn, Germany).The pre-and postoperative tumor volumes were measured in all patients to estimate the EOR using contrast T1-weighted or FLAIR-weighted MRI data obtained before and after tumor removal.If the tumor was not enhanced or partially enhanced on MRI, the EOR was evaluated based on residual high-intensity lesions on FLAIR-weighted MRI.On the other hand, if the tumor was enhanced on MRI, the EOR of the tumor was calculated based on the residual enhanced tumor.
The EOR was calculated using the difference between preoperative and postoperative tumor volumes: (preoperative tumor volume -postoperative tumor volume) / preoperative tumor volume [6].Volumetric EOR was categorized as follows: supratotal resection, EOR > 100%; gross total resection, EOR = 100%; subtotal resection, EOR ≥ 90% to < 100%; and partial resection, EOR < 90%.A supratotal resection was de ned as tumor resection extending beyond the MRI-veri ed abnormal area or the complete removal of any abnormality, with the postoperative cavity volume being larger than the preoperative tumor volume [22,25,31].This was also de ned as a postoperative tumor cavity/preoperative tumor volume >100%.
Tumor progression was de ned as newly detected areas of contrast enhancement and/or an obvious increase in the FLAIR signal abnormality on follow-up MRI relative to the baseline postoperative MRI obtained within 72 hours of the operation.

Adjuvant therapy protocol
Further treatment of grade II gliomas that underwent total or supratotal resection was based on observation.
When more than 10% of the tumor was left, chemotherapy such as temozolomide (TMZ) was applied as adjuvant treatment.
For grade III gliomas, initial adjuvant treatment included focal external-beam radiation therapy by conventional radiation planning to approximately 60 Gy (±5% total dose), with daily concurrent TMZ at 75 mg/m 2 throughout the course of the radiation therapy.This was followed by adjuvant temozolomide according to the Stupp protocol [32].
Direct DNA sequencing for IDH1 and IDH2 mutations Direct sequencing of IDH1 and IDH2 was conducted as previously described [33].A 129-bp fragment spanning the sequence encoding the catalytic domain of IDH1, including codon 132, and a 150-bp fragment spanning the sequence encoding the catalytic domain of IDH2, including codon 172, were ampli ed for IDH sequencing.

Statistical analyses
All statistical analyses were conducted using SPSS version 27.0 (IBM Corporation, Armonk, NY, USA) for Windows (Microsoft Corporation, Redmond, WA, USA).Survival was estimated using the Kaplan-Meier method, and the log-rank test was used to assess survival differences among groups.Progression-free survival (PFS) was calculated from the day of the rst surgery till the occurrence of true tumor progression, death, or the end of the follow-up.Overall survival (OS) was calculated from the day of the rst surgery until death or the end of the follow-up.

Factors in uencing PFS in our cohort of patients with
LGGs during awake brain surgery were investigated using univariate and multivariate Cox proportional hazards models adjusted for major clinical prognostic factors, including age at diagnosis (>40 years vs. ≤40 years), histologic type (astrocytic vs. oligodendroglial), WHO grade (grade III gliomas vs. grade II gliomas), tumor location (frontal regions vs. other regions), IDH1 or IDH2 status (mutation or wild type), nal EOR (>100% vs. <100%), chemotherapy (+ vs. -), and chemoradiotherapy (+ vs. -).The covariates and the independent variables that showed signi cant differences in the univariate analysis were used for the analysis.The remaining factors in the multivariate Cox proportional hazards model (p < 0.05) were considered to be independent predictors of PFS.

Patients
Between December 2012 and May 2020, a total of 126 consecutive patients (74 males, 52 females) who underwent awake surgery for LGGs of the dominant and non-dominant hemisphere were enrolled in this study.The median follow-up time was 33.0 months (range, 0.3-89.9months).The patients' clinical characteristics are summarized in Table 1.The mean age at the time of awake surgery was 42.8 years (range, 17-76 years).The tumors were located in the left hemisphere in 88 cases (69.8%) and in the right hemisphere in 38 cases (30.2%).

Postoperative neurological outcomes
Table 2 shows the summary of transient or permanent postoperative neurological de cits, including motor and speech disturbances, according to main tumor location.During the postoperative course, 47 (37.3%) patients exhibited new transient speech disturbances, 27 (21.4%)developed transient motor disorders, ve (4.0%) exhibited permanent speech de cits, and nine (7.1%) had permanent motor de cits.Among the 73 patients with frontal tumors, 24 (32.9%)exhibited transient speech disturbances and 17 (23.3%)developed transient motor disturbances.Of these 73 patients, two (2.7%) demonstrated persistent speech de cits and three (4.1%) had permanent motor de cits.By contrast, three of the 27 patients with insular tumors (11.1%) exhibited permanent speech disorders and four (14.8%) had permanent motor de cits.

Survival and supratotal or gross total resection in LGGs
We analyzed whether supratotal (EOR > 100%) or gross (EOR = 100%) total resection affected the survival of our 126 consecutive patients.The Kaplan-Meier estimates for PFS according to EOR classes for all patients are shown in Figure 1.The median PFS rate of the patients in the group with an EOR ≥ 100 %, including supratotal resection (n = 47; median survival, not reached), was signi cantly higher than that in the group with an EOR < 100% (n = 79; median survival, 43.1 months; 95% CI: 37.8-48.4months; p = 0.04; Fig 1).There were no signi cant differences in the median PFS rate between the patients in the group with EOR >100 % (n = 15; median survival, not reached) and those in the group with EOR = 100% (n = 32; median survival, not reached; p = 0.1).Furthermore, there were no signi cant differences in the median OS rates based on the EOR.In patients with diffuse astrocytomas and anaplastic astrocytomas, the group with an EOR ≥ 100 %, including supratotal resection (n = 25; median survival, not reached), demonstrated a signi cantly better PFS rate than the group with an EOR < 100% (n = 45; median survival, 35.8 months; 95% CI: 19.9-51.6 months; p = 0.03; Fig 2).The median OS rates of patients with diffuse astrocytomas and anaplastic astrocytomas showed no difference between groups classi ed according to EOR.By contrast, neither the median PFS nor the median OS signi cantly differed across EOR classes in patients with oligodendrogliomas and anaplastic oligodendrogliomas.
Bene ts of supratotal or gross total resection for patients with IDH-mutant type of diffuse astrocytomas and anaplastic astrocytomas We next investigated the prognostic correlations between supratotal or gross total resection and IDH1/IDH2 mutations in diffuse astrocytomas and anaplastic astrocytomas.It is noteworthy that supratotal or gross total resection was correlated with better PFS in IDH-mutant type of diffuse astrocytomas and anaplastic astrocytomas (n = 19; median survival, not reached vs. n = 35; median survival, 40.6 months; 95% CI: 22.3-59.0months; p = 0.02).By contrast, supratotal or gross total resection was not associated with longer PFS in patients in IDH-wild type of diffuse astrocytomas and anaplastic astrocytomas.

We further established multivariate Cox proportional hazards models for the factors in uencing PFS in patients with
LGGs using the following prognostic factors: IDH1 or IDH2 status, nal EOR, and chemotherapy.These multivariate models were subsequently adjusted to estimate the HR associated with PFS in our patients.Notably, the multivariate models also showed that PFS rates were signi cantly related to IDH1 or IDH2 status (mutation vs. wild type, HR = 0.30, 95% CI: 0.15-0.60,p = 0.001) and nal EOR (≥100 % vs. <100 %, HR = 0.37, 95% CI: 0.16-0.87,p = 0.02; Table 3).

Discussion
Large observational studies based on the objective evaluation of the EOR for gliomas have shown that maximal resection is signi cantly associated with favorable clinical outcomes for WHO grade II gliomas [6][7][8][9][10][11]34].One retrospective study of 216 patients with WHO grade II gliomas demonstrated that patients with an EOR > 90% had a 5-year OS rate of 97%, whereas patients with an EOR < 90% had a 5-year OS rate of 76% [6].Jakola et al. reported on a retrospective population-based parallel cohort of WHO grade II gliomas in Norway, comparing two hospitals with distinct treatment strategies.One hospital strategy favored early surgical tumor resection, while the other preferred biopsy and observation for WHO grade II gliomas.Notably, this study revealed a signi cant increase in survival in patients in the "early surgical resection" group [9,10], with a 5-year OS rate of 74% compared to a rate of 60% in the "biopsy and observation" group.Furthermore, another study reported a signi cant survival bene t for patients under "early tumor resection" management compared to those under "biopsy management" (5-year OS: 82% vs. 54%) [11], at two different departments acting independently at the same university hospital.This suggests that glioma surgeons should aim for maximal safe resection by increasing the tumor EOR in order to prolong survival in patients with WHO grade II gliomas.A large European phase III clinical trial (EORTC-26951) found that the extent of surgery was signi cantly related to survival in WHO grade III anaplastic oligodendroglial tumors [35].Nomiya et al. retrospectively estimated the prognostic factors for 170 patients with WHO grade III anaplastic astrocytoma [36].The median survival times of their patients, classi ed into total resection, subtotal resection, partial resection, and biopsy-only groups, were 86.4,61.6, 22.9, and 23.4 months, respectively.The authors emphasized that the extent of surgery is the most powerful prognostic factor in the treatment of WHO grade III anaplastic astrocytoma.Kawaguchi et al. revealed the importance of surgical resection by investigating 124 consecutive patients newly diagnosed with WHO grade III gliomas [37].Among these patients without 1p/19q co-deletion, those with gross total removal had signi cantly longer median overall survival times than those without gross total removal (median survival: not reached versus 77 months).Our results indicate that, similarly, an EOR ≥ 100 %, including supratotal resection, is signi cantly associated with better PFS rates in patients with LGGs (WHO grade II and III gliomas) (p = 0.04, Fig. 1).These results support the notion that maximal safe resection is a crucial prognostic factor for improving the survival of patients with LGGs.
Duffau et al. rstly reported that supratotal resection improves the outcome of patients with WHO grade II gliomas who have undergone awake mapping, after a mean follow-up of 11 years [22].These results suggest that supratotal resection, extending beyond the abnormalities detected by FLAIR-weighted MRI, provides a survival bene t, because tumor cells might invade sites 10-20 mm away from the tumor boundaries on MRI [38].Since a greater EOR, such as that achieved by gross total or supratotal surgical tumor resection, could signi cantly increase survival in patients with WHO grade II gliomas, we tried to achieve supratotal resection of the functional boundaries whenever possible, with the aid of awake functional mapping.We previously reported on the e cacy of awake brain surgery for the supratotal resection of diffuse frontal LGGs, while motor, language, and neurocognitive functions are preserved [23].

Recently, Rossi et al. presented a retrospective review of 319 IDH-mutated
LGGs in which supratotal resection was signi cantly related to survival, independently of molecular subtypes and WHO grades [24].The authors found that PFS rates were signi cantly higher in patients with IDH-mutated WHO grade II and III astrocytomas or oligodendrogliomas who underwent supratotal resection than in those who underwent total resection.Moreover, supratotal resection was signi cantly associated with a reduced rate of malignant transformation and a better OS.The present study found that supratotal or gross total resection was signi cantly associated with better PFS in IDH-mutant type of diffuse astrocytomas and anaplastic astrocytomas, but not in IDH-wild type.Furthermore, our results revealed no signi cant correlations between PFS and supratotal or gross total resection in patients with oligodendrogliomas and anaplastic oligodendrogliomas.This difference in results might stem from a small number of patients with oligodendrogliomas (56 cases of WHO grade II and III oligodendrogliomas) in our series compared to the earlier study by Rossi et al. (180 cases of WHO grade II and III oligodendrogliomas).Moreover, WHO grade III anaplastic gliomas have been recommended to be treated by maximal safe resection followed by radiotherapy and adjuvant PCV (procarbazine, lomustine, and vincristine) [35,39].Considering that adjuvant chemotherapy is an independent prognostic factor for improved survival of WHO grade III anaplastic oligodendroglioma[40], the EOR might not play an important role in the long-term outcome of oligodendrogliomas and anaplastic oligodendrogliomas.In addition, our study did not reveal a signi cant effect of supratotal or gross total resection in patients with IDH-wild type of diffuse astrocytomas and anaplastic astrocytomas.Recent studies suggested that these IDH-wild types of astrocytic tumors belong to completely different entities and therefore require different tumor management [41].
Our study reveals that tumor resection with awake functional mapping is associated with a reduction in late severe permanent neurological de cits in patients with LGGs, despite an early increase in transient functional de cits.In our study, late permanent speech and motor disturbances were observed in 4.0% and 7.1% of patients, respectively.By contrast, early transient speech and motor disturbances were observed in 37.3% and 21.4% of patients, respectively (Table 2).Compared to the total postoperative neurological de cits, however, late permanent speech and motor disturbances in insular tumor cases were relatively common (11.1% and 14.8% of patients, respectively), which indicates that they present a surgical challenge regardless of the use of awake functional mapping.These observed transient neurological disturbances, due to the proximity of critical normal brain structures to the tumor cavity, usually disappeared within a few weeks or months after tumor resection.Awake mapping enables more extensive tumor resection for better tumor control, while reliably identifying critical normal brain structures so that permanent neurological de cits can be avoided.
Although the current study provides novel information regarding the impact of supratotal or gross total resection with awake mapping on the survival of patients with LGGs, our results are limited compared to those of prospective clinical trials, as retrospective studies may be in uenced by unrecognized biases.Most importantly, the survival improvement associated with EOR ≥ 100% may be due to biases from differential tumor aggressiveness in non-resectable or easily resectable portions of the brain.Furthermore, the present study was based on a small number of tumor cases; therefore, a larger cohort study is needed to further establish the effect of supratotal or gross total resection during awake surgery using cortical and direct axonal electrical stimulation.Thus, further accumulation of evidence for this surgical strategy for LGGs will help the improvement of the treatment of this disease and hopefully develop it into a novel therapy.

Conclusions
The present study demonstrates the signi cant association of tumor EOR with survival in patients with LGGs.It is noteworthy that supratotal or gross total resection signi cantly correlated with better PFS in IDH-mutant type of WHO grade II and III astrocytic tumors.In light of our nding that EOR did not correlate with PFS in patients with aggressive IDH-wild type of diffuse astrocytomas and anaplastic astrocytomas, we suggest treatments that are more intensive will be needed for the control of these tumors.

Figures
Figures

Table 3 .
Univariate and multivariate Cox proportional hazards models for the factors influencing PFS in patients with lower grade gliomas Abbreviation: PFS; progression free survival, WHO; World Health Organization, IDH; isocitrate dehydrogenase, EOR; extent of resection, HR; hazard ratio, CI; confidence interval