Presurgical fMRI Affecting the Extent of Resection and Outcome of the Supratentorial Gliomas; a Single-center Experience From a Developing Country

Background Functional magnetic resonance imaging (fMRI) is one of the available non-invasive imaging to map the different brain areas, which has been used during the current years. In this study, we aimed to evaluate the effect of fMRI on decision-making, the extent of resection, and the outcome of the patients with supratentorial gliomas. Methods This prospective study conducted from 2014 through 2017 to investigate how presurgical fMRI can help the neurosurgeon during glioma surgery. Baseline demographic and clinical data were collected, and standard fMRI protocol was used for each patient.


Introduction
Intra-axial brain tumors like gliomas, especially slow-growing ones, are still challenging due to the higher postoperative morbidities and unacceptable consequences. The tumors close to or within the eloquent areas are more controversial in terms of operation, owing to various degrees of postoperative complications and decreased quality of life (1)(2)(3). In this regard, many neurosurgeons choose an observational approach until the neurologically deteriorated situation, or may prefer suboptimal resection that is accompanied by some other disadvantages such as wounded glioma (4).
During the last decades, advanced neuroimaging has helped neurosurgeons select the best approach to achieve the maximal extent of tumor resection, minimal postsurgical de cits, and improved survival(5-7).
Functional magnetic resonance imaging (fMRI) is one of the available imaging has been used during the current years. This modality identi es local hemodynamic changes in the brain tissues showing neuronal activities of the functional regions. In general, fMRI uses the blood oxygen level-dependent (BOLD) technique to detect the cerebral areas with a dropped deoxyhemoglobin level during the speci c tasks(8).
During the functional tasks, neuronal activities lead to raised regional cerebral blood ow causing increased oxygen saturation. Then, high deoxyhemoglobin levels have been detected in T2 * -weighted imaging as hyperintensity, re ecting the eloquent regions (7).
It can map the cerebral regions non-invasively before and during the surgery. Hence, it may change the neurosurgeon's opinion in terms of the optimal approach, and when and where to stop (9). In this study, we aimed to evaluate the effect of fMRI on the extent of resection and the outcome of the patients with supratentorial gliomas.

Patients
The institutional review board approved the study, and each patient signed informed consent documents before we proceeded. Before surgery, patients underwent BOLD fMR imaging in a 3Tesla MR imager.
Forty-one consecutive patients (29 male and 12 female patients, mean age of 48.14 ± 22.1 years) with potentially resectable gliomas were evaluated prospectively from April 2014 to August 2017. All patients were operated by a neurosurgeon with more than 10 years experiences, and they were followed at least for a year. The tumor types were classi ed according to the World Health Organization (WHO) grading system resulted in 16 low-grade gliomas (LGG) and 25 high-grade gliomas (HGG) that all were newly diagnosed. All patients underwent postoperative conventional brain MRI within rst 48 hours to estimate the EOR.
All the patients were cooperative and performed different essential functional tasks in association with the tumor location. There were not any confounded fMRI due to the artifact, because in patients with nonquali ed fMRI, the imaging was repeated. Relevant Table 1 presented the anatomical and clinical data of the patients. Patients were asked to perform three different types of functional scans depending on the tumor location. Finger and toe-tapping were performed to show motor center activation and sponge squeezing for assessing the sensory function. Besides, Language task was implemented using the previously developed task by neuroimaging and analysis group (NIAG) (10). For this case, due to the level of literacy and academic education, we run the reverse word reading task (RWR) to evaluated language functional area. The subject was presented with Persian word while letters were being presented in the left to right alignment and she was asked to read the word silently. Preprocessing sections were applied rstly by brain extraction tool (BET) and continued by fMRI expert analysis tool (FEAT) for motion correction, slice timing correction, spatial smoothing, temporal ltering and intensity normalization.
For evaluating the effect of fMRI on decision making and the estimated extent of resection (EOR), we showed the conventional brain MRI to a group of four experienced neurosurgeons rst, and ask them about their approach and estimated EOR. Then, we showed them conventional MRI along with fMRI, and ask them for same questions. Finally, we compared their opinions with postoperative EOR.

Results
A total of 41 patients with newly diagnosed supratentorial gliomas met inclusion criteria, including 29 males and 12 females ranging in age from 21 to 72. All patients, except one, were right-handed. The most common presenting symptom was headache (85.3%), and seizure (70.7%). Only 29.2% of all cases re ected neurological de cits as presenting symptoms.
All patients had preoperative KPS of more than 70, and 63.5% had KPS more than 90 who were evaluated by the preoperative fMRI. HGG accounts for 61% of all, and the rest of them were LGG predominately in the left hemisphere. Table 1 shows all the patient features in detail.
Although presurgical fMRI led to a greater extent of resection (EOR) in 51.2% and decreased the EOR in 31.7% of patients, the differences between the estimated EOR before and after viewing fMRI were not signi cant statistically (p = 0.132).
The outcome was assessed in terms of three different indices, including language, motor, and KPS. In 11 out of 24 patients (45.83%) with eloquent areas tumors, the estimated EOR by the neurosurgeon utilizing only conventional images compared to conventional besides functional sequences changed according to the fMRI. In cases with non-eloquent tumors, fMRI did not affect the EOR.
Our results revealed a signi cant association between eloquent areas tumor and the EOR (p = 0.001 and Spearman's correlation coe cient = 0.51). Changes in the EOR after adjusting the fMRI were seen in 11 patients, including 5 LGG and 6 HGG, but there was no signi cant relationship between the EOR and tumor histology (p = 0.620).
According to our results, in patients that the neurosurgeon decided to increase the EOR after the fMRI evaluation, non-signi cant language deterioration in short-term follow-up (one month after surgery) occurred. However, ultimately they found normal language function in a 1-year follow-up (p < 0.05). The motor index had a similar situation to the language one as well. In 12 out of 24 (29.2%) presenting with motor de cits before admission, utilizing the fMRI by the neurosurgeon caused the EOR change. After one year, motor de cits improved in 6 out of 12, and in the other 6, de cits were constant.
For the patients in whom fMRI led to changed neurosurgeon's mind about the EOR, postsurgical KPS dropped signi cantly till month one, whereas it reached out to the presurgical level or even better in a 1year follow-up.
Patients who experienced the higher EOR by fMRI usage were analyzed independently. Eight out of 41 (19.5%) experienced higher EOR only after accessing the neurosurgeon to fMRI. Among these patients, postoperative KPS and motor indices re ected dropped levels accompanied by an xed language index level in the short-term period, whereas all indices improved after a long-term follow-up. There was a signi cant association between increasing EOR and all indices in a 1-year follow-up.
After one year, four patients with HGG died, including one with non-eloquent gliomas and three with eloquent tumors. In two out of four, fMRI ndings resulted in less EOR than what the neurosurgeon had estimated without the fMRI.

Discussion
In this prospective study, we assessed preoperative fMRI's role as an advanced neuroimaging in surgery of supratentorial gliomas. Nowadays, advanced perioperative neuroimaging is used widely to help neurosurgeons minimize iatrogenic complications, postoperative neurological de cits, then increase the EOR and improve the overall survival (OS) of the patients as much as possible.
Although there was non-signi cant, presurgical fMRI changed the neurosurgeon's mind about the EOR in our patients. Consistent with our results, Bartsch  EOR in gliomas has a prominent role in overall survival and progression-free survival (14). One of the preoperative fMRI goals is to help the surgeons increase the EOR as much as possible. Our results indicated a change in the EOR in more than 50% of all patients that statistically was signi cant in 45.83% of patients with eloquent areas gliomas. These data were in concordance with Lorenzen et.al. (12) (14).
Given the results mentioned above, patients in whom the neurosurgeon changed the EOR according to the fMRI ndings showed worse short-term outcomes in terms of language, motor, and KPS, but all patients improved after a long-term period. Twenty-four patients presented with motor de cits and fMRI changed the EOR in half of them, led to the worst postoperative motor score. 50% of them reached the baseline score, and the other found motor improvement even better than the baseline within the rst year after surgery. There was no report of permanent neurological de cits in our study, whereas Lorenzen et al. reported 28.5% postoperative de cits, in which 14.2% were permanent (12). Furthermore, Kapsalakis et al. reported 29.9% new postsurgical de cits and/or worsening of the presurgical de cits that were persistent in 13.8% (14). Similar to our results, Vysotski et al. revealed more improvement in postoperative neurological de cits in patients who underwent presurgical fMRI (9).
Parallel to Castellano et al., 19.5% of our patients found a higher EOR using the fMRI (15). In these cases, KPS and motor indices were aggravated in short-term follow-up along with unchanged language score (non-signi cant), whereas all three indices improved in the following year signi cantly. Our data were in concordance with Vysotski et al. and Castellano et al. ndings [12,13] . Short-term deterioration may result from relying on the fMRI ndings leading to resection of the tumor as close as the eloquent regions. The fMRI limited the EOR in 31.7% that 15.3% of them diagnosing as HGG in eloquent cortex died after 1-year follow-up.
In future studies, a larger sample size with only eloquent areas gliomas as well as longer duration in the follow-up period is recommended. The ndings of presurgical fMRI in association with intraoperative fMRI and intraoperative cortical and subcortical electrophysiological stimulation could provide more valuable results.
In conclusion, we postulate that preoperative fMRI should be considered preoperatively, especially in the eloquent areas gliomas disregarding the tumor grade. Moreover, it can lead to a positive outcome in patients with supratentorial gliomas via giving useful data about the relation of the tumor and vital centers of the brain, although it may result in an unfavorable short-term outcome.

Declarations
Funding: Not applicable Con icts of interest/Competing interests: Not applicable Ethics approval: The institutional review board approved the study, and the ethical number is IRSINA23041393.

Consent to participate:
Each patient signed informed consent documents before we proceeded.

Contributions:
SEK and AA conceived and designed research. AM and KKY conducted experiments. MASA contributed analytical tools. MS analyzed data. AP wrote the manuscript. All authors read and approved the manuscript.