Language mapping during AC to guide tumor resection is a widespread approach for preserving language function while achieving maximal tumor resection and improving the postoperative quality of life [15]. Even with the most efficient approach, many patients experience permanent partial or total language dysfunction following AC [16, 17, 20–23], and it remains unclear which patients are more susceptible to surgery-related language impairments. We found that of the four subtypes, type II tumors and tumors that were close to the posterior SLF/AF were most likely to result in permanent language impairment postoperatively.
Key time point for predicting the postoperative language status
In this study, we defined permanent postoperative language impairment as impairment that remained at the 3-month follow-up, based on previous related studies [28–30]. Transient language impairment is a common complication after AC [15]. In this study, the percentage of patients who had a transient language deficit is consistent with those of previous studies [16, 18, 20]. Moreover, the number of patients with transient language deficits gradually increased within the first 3 days after AC and subsequently decreased after the third day following AC. The transient language deficits may be associated with edema or a transient retraction injury [31]. The time of peritumoral edema onset was significantly short to induce functional impairment on the day of tumor resection [32–34] in cases of worse postoperative language function compared with the preoperative status. The presence of new-onset infarctions was another reason for postoperative language impairment. In this study, no patient exhibited new-onset infarction, based on postoperative MRI performed within 72 h after surgery. To avoid new brain infarctions, we applied pledgets soaked in papaverine to the drainage vessels in some patients, as recommended by Berger et al. [35]. Hence, language function was damaged during AC in these patients. All patients who did not experience language impairment on the day of tumor resection recovered their language function. However, over 50% of patients who experienced language impairment on the day of tumor resection did not recover after 3 postoperative months. Thus, the key time point to predict long-term language prognosis is the day of tumor resection. If language deficit is detected on the day of tumor resection, systemic rehabilitation training [36, 37] for language function should be implemented to accelerate patient recovery.
Risk factors associated with permanent surgery-related language impairments
In this study, type II tumors, shorter distance between the glioma and the SLF/AF, and IDH mutation were negative factors in patients with surgery-related language impairment.
Type II tumors were associated with the worst language prognosis and longest recovery time, which could be due to two reasons. First, compared with the anterior and inferior SLF/AF, the posterior SLF/AF has the lowest capacity to reorganize and compensate for damaged functions [38]. This vulnerability of the anterior and inferior SLF/AF could be compensated by the actions of the uncinate fasciculus, inferior longitudinal fasciculus, and inferior frontal-occipital fasciculus [39, 40]. Because type II tumors are close to or directly invaded the posterior SLF/AF, language impairment easily results when the posterior SLF/AF is accidentally transected during tumor resection. Second, the sensitivity of the bipolar electrical stimulator for subcortical structure identification is not as high as that for cortical identification [41]. This phenomenon was also identified in a previous study on motor-related subcortical structure mapping [24]. Regarding the regions of tumor resection in language-eloquent areas, the recommended safety region for tumor resection is > 5 mm from positive sites [42, 43]. However, the diameter of the bipolar stimulator is 5 mm, and the effective stimulating depth is 2.5 mm (Figure S3). Hence, even if we identify positive sites on the SLF/AF with a bipolar stimulator, the EOR exceeds the safety margin. This could explain why shorter distance from the tumor to the SLF/AF was a risk factor.
Moreover, we did not expect that IDH mutation would be unfavorably associated with surgery-related language impairment. In previous studies, IDH mutation was considered a favorable factor for total resection [44, 45]. However, the mutation was not beneficial for subcortical fiber protection, such as in the removal of type II tumors under the current and standardized subcortical stimulation protocol (the reasons are described above). Regarding gliomas with IDH mutations, although many studies indicate the associated benefits of longer survival outcomes after gross total resection [45–47], the focus should be on the preservation of language function in cases where the tumor has invaded the posterior SLF/AF.
Furthermore, patients with type IV tumors required longer time for language recovery than those with types I and III. This finding indicates that transcortical resection for insular gliomas may have more serious postoperative complications than cortical glioma resection. However, no difference in the ratio of permanent language deficit was found between types I, III, and IV, supporting the notion that transcortical insular glioma resection via AC and language mapping is a safe and reliable approach [48, 49].
Intraoperative epileptic status in the four tumor subtype groups
We found that patients with types I and IV tumors were susceptible to ISE during functional mapping. Our findings are consistent with those of a previous study that demonstrated that the occurrence of ISE was related to tumor location in the frontal lobe [30]. For type IV tumors, all insular gliomas were removed by transcortical resection, and most of the removed cortices were taken from the frontal lobe. This could explain why patients with type IV tumors were significantly more susceptible to ISE. Therefore, for types I and IV, we recommend caution when increasing stimulation intensity during functional mapping and the administration of anti-epilepsy drugs [30] before AC to avoid ISE.
Value of our language recovery classification
Our classification method revealed that AC is the optimal approach to preserve language function. However, if tumors have invaded the posterior central gyrus or supramarginal gyrus (above the lateral fissure level) and are close to the posterior SLF/AF (within 2.75 mm), there is an increased risk of postoperative permanent language impairment. Conservative tumor resection under a neuronavigation guiding system should be performed in patients with type II tumors to avoid postoperative language impairment. Moreover, our classification provides a new preoperative decision component that could help patients and their families better understand the risk of language deficit and possible surgical outcomes. Additionally, our classification suggests that anti-epilepsy drugs should be regularly administered to patients with brain tumors to avoid ISE.