An Algorithm for Sellar Reconstruction Following Transnasal Transsphenoidal Surgery: A Review of 300 Consecutive Cases

Objectives/Hypotheses: The transnasal, transsphenoidal (TNTS) approach to the sella is the workhorse endoscopic procedure for pituitary tumors. Various reconstruction techniques have been reported following TNTS surgery based on patient and lesion characteristics, ranging from no reconstruction to vascularized pedicled flaps. We review our institution’s experience with sellar reconstruction following TNTS, and propose an evidence-based algorithm.

Design: Retrospective review.

Setting: Tertiary academic medical center.

Participants: (Patients who underwent endoscopic TNTS surgery for sellar lesions between March 1, 2013 and August 31, 2016.

Main Outcome Measures: Patient demographic and clinicopathologic data were collected. Outcome measures included intraoperative and postoperative cerebrospinal fluid (CSF) leak rates, complication rates, and extent of resection (gross or subtotal).

Results: Three hundred consecutive patients were included in the analysis. The mean age during surgery was 49 ± 17 years, with 44.7% male patients. Pituitary adenomas accounted for the majority of lesions (78.3%), followed by Rathke’s cleft cysts (9.3%). 18% were recurrent lesions, and 3% had previous sellar radiation. Mean lesion size was 19.6 ± 10.7 mm. 43% had suprasellar extension, while 20% had cavernous sinus invasion. The intraoperative and postoperative CSF leak rates were 31.7% and 3.0%, respectively. 80% of lesions achieved gross total resection. In all cases of CSF leak, abdominal fat grafts were used, with 66% requiring additional rigid fixation with a bone graft or resorbable plate. Nearly 70% of cases were reconstructed with a free mucosal graft (FMG) to the sella only, with the remaining cases were reconstructed with a nasoseptal flap (NSF). 27% of nasoseptal flap reconstructions were kept in place with a Foley balloon in the sphenoid sinus. Multivariable logistic regression revealed that subtotal resection was associated with recurrent disease (OR 3.80, p = 0.002), suprasellar extension (OR 3.19, p = 0.008), and cavernous sinus invasion (OR 3.65, p = 0.001); intraoperative CSF leak was associated with recurrent disease (OR 2.00, p = 0.045); and postoperative CSF leak was associated with cavernous sinus invasion (OR 5.37, p = 0.016).

Conclusions: Based on this large series, we propose the following algorithm for sellar reconstruction: marsupialization for Rathke’s cleft cysts; FMG for no intraoperative CSF leak; fat graft ± rigid fixation and FMG for low-grade leaks; and fat graft ± rigid fixation and NSF for high-grade leaks.