Roles of different methods in measurement of distance between intracranial arteriovenous malforma-tion and eloquent cortex in prediction of postoperative neurological deficits

Objectives To explore the predictive values of different methods measuring the distance between lesion and eloquent cortex (LED) in surgery-related functional deficit (SFD) in intracranial arteriovenous malformations (iAVMs). Methods This study retrospectively reviewed and followed up the patients who underwent microsurgical resection of iAVMs at Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University from January to December 2017. Based on the difference (≥1) between the mRS (modified Rankin scale) at 3 months after surgery and that prior to surgery, 17 patients were categorized into the SFD group and 79 into non-SFD group. The eloquent cortex was firstly recognized according to the Spetzler-Martin(S-M)standard. Based on the presurgical T1-weight, T2-weight, time-of-flight, functional and diffuse tensor imaging MR, the LED and modified LED (MLED) were measured. Multivariate logistic analysis was performed to investigate the independent risk factors related to SFD at 3 months post surgery. The ROC (receiver operating characteristic) curve analysis was performed to explore the predicting values of different measuring methods in SFD. Results Multivariate logistic analysis based on 96 patients demonstrated that the S-M grading (OR=3.10), MLED (OR=0.66) and nidus type (diffuse and compact, OR=0.12) were independent risk factors for SFD (all P<0.05). For all iAVMs, both MLED and LED had good predictive value in SFD, and the AUC (area under a curve) values were 0.843 and 0.830 respectively (both P<0.05), while the S-M grading(AUC: 516, P=0.804) did not have predictive value. For compact iAVMs, LED (AUC=0.784) and MLED (AUC=0.726) could effectively predict the risk of SFD(all P<0.05). For diffuse iAVMs, MLED (AUC=0.833) could predict the risk of SFD (P=0.003), whereas LED (AUC=0.583, P=0.665) had poor predicting value. Conclusions The distance between iAVM and eloquent cortex seems to be an important factor to predict the surgical outcome. For compact iAVM, both LED and MLED have good predicting values, while only MLED shows predicting value for diffuse iAVM. Key words: Intracranial arteriovenous malformations; Neurological deficit; Multi-modeimage; Forecasting