Incidence, etiology, and management of cerebrospinal fluid leaks following trans-sphenoidal surgery.

Abstract

The incidence of cerebrospinal fluid (CSF) leak following trans-sphenoidal surgery ranges from 0.5% to 15.0%. Factors predicting which patients are likely to develop postoperative leaks and optimal management of these patients are poorly defined. The objective was to determine 1) the incidence of CSF leak following trans-sphenoidal surgery; 2) demographic or intraoperative factors associated with postoperative leaks; 3) techniques and efficacy of postoperative leak management at Oregon Health and Science University, (Portland, OR). Retrospective chart review. Two hundred thirty-five trans-sphenoidal surgeries were performed on 216 patients between 1994 and 2001. Follow-up data were available for 217 operations (92.3%) performed on 202 patients (93.5%). Postoperative CSF leaks occurred in 6.0% (13 of 217) of patients. Leaks were more common in the setting of revision surgery versus primary surgery (14.6% vs. 4.0%, P =.0096), nonadenomatous disease versus pituitary adenoma (15.8% vs. 5.1%, P =.059), or if an intraoperative leak occurred (12.7% vs. 2.7%, P =.004). However, on multivariate analysis, only the presence of intraoperative leak (P =.008) and nonadenomatous disease (P =.047) were found to be independently associated with postoperative CSF leak. Size of adenoma was not associated with occurrence of postoperative CSF leak (6.4% for microadenoma vs. 4.2% for macroadenoma) on both univariate and multivariate analysis. There were 13 postoperative CSF leaks: 2 resolved with lumbar drainage and 11 required operative management (three required multiple procedures). Of the three patients who required multiple operations, two had hospital courses complicated by meningitis and postinfectious hydrocephalus and ultimately required ventriculoperitoneal shunts. Endoscopic re-exploration was successful in 87.5% (7 of 8) of cases. Cerebrospinal fluid leaks following trans-sphenoidal surgery occurred in 6.0% of cases. Nonadenomatous disease and presence of an intraoperative leak were independent predictors of a postoperative leak. Endoscopic re-exploration combined with packing was an effective technique in managing uncomplicated postoperative leaks. In the setting of meningitis and postinfectious hydrocephalus, more invasive techniques such as ventriculoperitoneal shunt may be necessary.