Background: Dural involvement is an important consideration in assessment of cranial base tumors dictating resectability and prognosis. Preoperative as well as intraoperative clues are valuable but not always correct. We evaluated a consecutive series of craniofacial resections at our center to correlate radiologically suspected dural involvement vis-à-vis intraoperative assessment and eventual pathology.
Methods: We conducted a retrospective analysis of cases of skull base tumors where potential dural involvement was considered. We recorded the preoperative radiological impression (contrast-enhanced magnetic resonance imaging) regarding dural involvement (normal, extradural, intradural, parenchymal disease), intraoperative impression (normal, adherent, subdural, parenchymal disease), and final histology (normal, reactive, tumor). We also recorded instances where the dura was resected and/or inadvertently breached and the incidence of postoperative cerebrospinal fluid leak and meningitis.
Results: One hundred twenty-seven cases were evaluated. Transcranial approaches were performed in 68 cases. Nineteen percent (24 cases) were endoscopic procedures. Dural resection was performed in 38 cases (30 being proven pathologically). The incidence of cerebrospinal fluid leak was 4.7%. The sensitivity, specificity, positive predictive value, and negative predictive value of magnetic resonance assessment were 34.5%, 97.9%, 83.3%, and 83.2%, respectively, providing an overall accuracy of 84%, and those for intraoperative dural adherence were 84.6%, 85.6%, 44%, 97.6%, and 85.5.%, respectively.
Conclusions: Preoperative magnetic resonance imaging, although a good modality for imaging the disease extent, may not always identify the extent of dural involvement. Intraoperative assessment therefore becomes very important especially when it is unequivocally normal. Both should be used to ensure accurate treatment strategies and tailor the need for dural resection.