Reuse of Nasoseptal Flaps for Endoscopic Endonasal Skull Base Reconstruction

Structured Abstract

Introduction  Pedicled nasoseptal flap (NSF) placement is a critical component of skull base reconstruction after endoscopic endonasal approaches (EEAs). The effectiveness of NSF reuse has not been thoroughly studied. Prior reports using flaps harvested at one center and reused at another may have technical variability bias.

Methods  We identified patients who underwent both their initial and NSF-reused surgeries at Weill Cornell Medical College from 2004 to 2022 using a prospective database of all EEAs. Surgical pathology, intraoperative leak grade, use of cerebrospinal fluid (CSF) diversion and skull base coverage were examined. The primary outcome measure was occurrence of CSF leak.

Results  Fourteen patients (six women, eight men) underwent 14 first time and 14 revision operations with median age of 36.6 years (interquartile range [IQR]: 23.9–61.3) at the time of the NSF reuse. The median interval between the first NSF use and reuse was 70.6 months (IQR: 16.6–87). Eight patients were operated on for pituitary adenoma. Nonadenomas included three craniopharyngiomas and one case each of epidermoid, ependymoma, and chordoma. There were 16 high-flow, 8 low-flow intraoperative leaks, and 4 with no leak. CSF diversion was used in 24 operations. There were three postoperative leaks, one after a first operation and two after NSF reuse. All postoperative CSF leaks, whether first or second operations, occurred in cases with both high-flow intraoperative CSF leak and incomplete NSF coverage ( p  = 0.006).

Conclusion  NSF reuse is effective at preventing postoperative CSF leak. The primary predictors of leak are high-flow intraoperative leak and inadequate defect coverage with NSF, regardless of the operation number.

Introduction

Reconstruction of the anterior skull base after endoscopic endonasal approach (EEA) is paramount to prevent postoperative cerebrospinal fluid (CSF) leak. The goal is to restore a multilayered tissue barrier. The nasoseptal flap (NSF) is a pedicled vascularized tissue covering the skull base defect harvested from the mucoperiosteum and mucoperichondrium of the nasal septum, ^{1 2 3 4 5 6 7 8} which can be used in combination with other adjuncts to close a skull base defect. NSF has improved the success rate of the anterior skull base reconstruction, ² reduced the rate of CSF leak, ^{9 10} and maintained a low complication rate after EEA. ^{11 12 13}

The vascular supply of the NSF is provided by one or more posterior septal arteries, branches of the sphenopalatine artery, itself a terminal branch of the internal maxillary artery. Due to its robust blood supply, the NSF provides a natural, durable tissue layer that can effectively separate the cranial and nasal cavities, accelerate healing, and eliminate dead space. Factors previously associated with an increased risk of postoperative CSF leak are extended approaches, intradural localization of the lesion, intraoperative CSF leak, due to opening of the basal cisterns or the third ventricle, tumor pathology, hypercortisolemia, morbid obesity, prior radiation therapy, and reoperation. ¹

Whether reuse of the NSF is as successful at preventing postoperative CSF leak as when that flap is first used is not well studied. Prior reports have limited patients and have included patients operated at two different institutions, which introduces an uncontrolled variable, namely the surgeon performing the closure. For this reason, we identified a group of patients in which both the first and second surgeries were both done at our institution in which NSFs were reused, to control for the surgeon and their ability to successfully repair the skull base.

Methods

From a prospective database of all EEAs done at Weill Cornell Medical College (WCMC) from 2004 to 2022, we identified a series of patients requiring reoperation and reuse of the same NSF. We only included patients on whom we performed both the first and second operations, so that we could use the first operation as a control for the second one, it being performed by the same surgeons using the same technique. Patients whose NSFs were first placed at another institution were excluded. The study was approved by the WCMC Institutional Review Board. Electronic medical records were evaluated, noting demographics, operative notes, radiographic findings, and outcome. All data were independently evaluated by two researchers (D.M., E.J.B., or S.J.)

All patients had a pre- and postoperative magnetic resonance imaging (MRI) scans ( Fig. 1 ). The appearance of the NSF, including coverage of bony defect by NSF, and its vascularization were assessed independently by two investigators (D.M. and E.J.B.) using the sagittal, coronal, and axial T1-enhanced early postoperative MRI images as well as the operative notes. During data collection and analysis, incomplete coverage of NSF was documented in a blinded manner with respect to the occurrence of postoperative CSF leak. Flaps were not considered to be in optimal position if the flap was not covering the entire area of the skull base defect in any of the three planes or if the NSF was not directly opposed to the skull base, on sagittal plane on early postoperative MRI ( Fig. 2 , vertical and horizontal lines). Surgical details were highlighted, especially the occurrence of intraoperative CSF leak and the reconstruction technique. Intraoperative leaks were categorized into high flow (HF), low flow (LF), or none. We defined them based on the rate of leakage during surgery. HF was defined as visibly pulsating leakage, while LF was defined as steady dripping or seepage.

Fig. 1.

Sagittal sections through the sphenoid sinus on first postoperative MRI after EEA using NSF. The first two columns are the first seven patients as numbered in Table 1 ; the second two columns are the next seven patients. The first and third columns are images after the first EEA, while the second and fourth columns are images after the NSF reuse. The postoperative MRIs presented were performed within a week from surgery except for the four hash tagged cases where postoperative MRIs were delayed beyond a week. EEA, endoscopic endonasal approach; MRI, magnetic resonance imaging; NSF, nasoseptal flap.

Fig. 2.

Sagittal sections through the sphenoid sinus on first postoperative MRI after EEA showing ( A ) NSF that is not fully attached to the skull base with the relevant measurements of the distance from the anterior skull base to the NSF anterior edge (vertical line), as well as the distance from the sellar bone defect to the NSF (horizontal line), and ( B ) NSF that is well opposed to the skull defect and the bony edges. Both of the cases show fat in the sella and no tumor. EEA, endoscopic endonasal approach; MRI, magnetic resonance imaging; NSF, nasoseptal flap.

All tumors were classified according to the histopathological examination. Use of CSF diversion was also examined. The primary outcome measure was the incidence of postoperative CSF leak after surgery.

Surgical Technique

The surgical team consisted of a neurosurgeon (T.H.S.) and an otorhinolaryngologist (V.K.A., A.T., or A.K.). Lumbar drains (LDs) were placed intraoperatively in all cases in which a HF CSF leak was expected. LD was left in place for 24 hours, but it was extended if a CSF leak occurred postoperatively until resolved. The initial drainage rate was set at 10 mL/h. The drainage rate was then adjusted based on the patient's clinical response. External ventricular drains (EVDs) were placed if patients had hydrocephalus. The mucosal flap edges were identified by inspection and the flap was taken down by sharp and blunt dissection. This began at the distal end of the flap and moved proximally until the base of the sphenoid sinus was identified. Flap perfusion was not assessed intraoperatively using fluorescence. The flap was stored in the choana until the end of the resection. Defects after reoperations were reconstructed with same NSF as was used in the first operation, held in place with tissue sealant (either DuraSeal [Integra, United States] or Adherus [Stryker, United States]). Beneath the flaps, the methods of closure varied based on the flow rate of the leak and the preferred method of closure during that time period in our experience. Options included fat, fascia lata, button graft with either fascia lata or AlloMax (Davol, United States), gasket seal with Medpore (Stryker, United States) and either fascia lata or AlloMax (Davol, United States), Dura-Guard (Bio-Vascular, United States), Duraform (Natus Medical, United States), or HydroSet (Stryker, United States).

Data Analysis

Statistical analyses were performed using Statistica software (Version 13, StatSoft, Inc., Tulsa, Oklahoma, United States) and R programming. We compared continuous variables between two groups with the Mann–Whitney U test. The frequencies of categorical variables were compared using two-sided Fisher's exact test ( p  < 0.05 was significant). Clustering analysis was performed in R studio (version 1.2.5033) using the pheatmap package. Sankey plot was developed in R studio (version 1.2.5033) using the ggplot package.

Results

From a database of 1,590 EEA procedures, we selected all duplicate names to identify all reoperated patients. Of these, only cases in which the NSF was used in both operations were selected. The final study group consisted of 14 patients (6 women, 8 men) with median age of 33 years (interquartile range [IQR]: 21.7–54.8) at the time of the first surgery, and 36.6 years (IQR: 23.9–61.3) at the time of the second EEA ( Table 1 ). Pathology included pituitary adenomas ( n  = 8, one with Cushing's disease and one with acromegaly), craniopharyngiomas ( n  = 3), epidermoid ( n  = 1), ependymoma ( n  = 1), and chordoma ( n  = 1). The median time between the first and second surgeries was 70.6 months (IQR: 16.6–87). Thirteen patients were reoperated due to tumor recurrence, whereas one patient with a giant chordoma had a planned staged surgery prior to receiving proton beam radiation therapy. Sagittal sections through the sphenoid sinus on postoperative MRI after each procedure are presented in Fig. 1 . Early postoperative MRI was performed within 5 days in 23 of 28 operations. Detailed timing of postoperative MRI is provided in Table 1 .

Table 1. The 14 patients organized based on the interval between the first and second surgery.

									NSF appearance on post-op MRI
No. pathology	EEA	Age (y)	Sex	Interval (mo)	Intraop CSF diversion	Approach	Intraop CSF leak	POD	Coverage of bony edges	Opposed to skull	Additional reconstruction	Postop CSF leak
1. Macroadenoma	First	26	M	114	LD	TS	LF	103	Yes	Yes	Fat plus a Medpore buttress	No
	Second	36			LD	TS	No	3	Yes	Yes	Fat only	No
2. Macroadenoma	First	41	F	107	LD	TS, TC	LF	1	Yes	Yes	Nothing	No
	Second	49			No	TS, TC	No	12	Yes	Yes	Nothing	No
3. Macroadenoma a	First	56	M	98	No	TS	No	23	Yes	Yes	Fat plus a gasket with AlloMax	No
	Second	64			LD	TS	LF	5	No	No	Button with AlloDerm plus fat	No
4. Cushing's disease a	First	60	F	83	LD	TS, TC	LF	2	Yes	Yes	Fat only	No
	Second	67			LD	TS, TC	LF	3	No	No	Fat plus fascia lata	No
5. Macroadenoma	First	48	F	76	LD	TS, TP	HF	1	Yes	Yes	Fat plus a Medpore gasket with fascia lata	No
	Second	54			LD	TS, TP	HF	5	Yes	Yes	Medpore gasket with facia lata	No
6. Macroadenoma a	First	54	F	73	EVD	TS, TP	HF	7	Yes	Yes	Fat plus a Medpore gasket with fascia lata	No
	Second	60			LD	TS, TP	HF	2	Yes	Yes	Medpore gasket with facia lata	No
7. Epidermoid cyst	First	28	M	73	LD	TS, TP	HF	1	No	Yes	Fat plus Medpore gasket with fascia lata	No
	Second	33			LD	TS, TP	HF	11	No	Yes	Fat plus fascia lata	Yes
8. Craniopharyngioma a	First	4	M	69	EVD	TS, TP	HF	4	No	No	Fat plus Medpore gasket with fascia lata	Yes
	Second	10			LD	TS, TP	HF	4	Yes	Yes	AlloMax and fat	No
9. Craniopharyngioma	First	23	M	34	LD	TS, TP	HF	2	Yes	Yes	Medpore gasket with fascia lata	No
	Second	26			LD	TS, TP	HF	2	Yes	Yes	Button with AlloMax	No
10. Acromegaly	First	29	M	30	No	TS	LF	2	Yes	Yes	Fat plus a Medpore buttress	No
	Second	31			No	TS	No	2	Yes	Yes	Nothing	No
11. Craniopharyngioma	First	15	F	17	LD	TS	HF	4	Yes	Yes	Fat with Medpore buttress	No
	Second	16			LD	TS	HF	2	Yes	Yes	Fat plus Medpore buttress	No
12. Macroadenoma	First	65	M	16	LD	TS	HF	2	Yes	Yes	Button with AlloDerm plus fat	No
	Second	67			LD	TS	HF	2	No	No	Medpore gasket with AlloDerm covered with HydroSet	Yes
13. Pontine ependymoma	First	16	M	5	LD	Transclival	HF	1	No	No	Medpore gasket with fascia lata	No
	Second	16			LD	Transclival	HF	1	No	No	Medpore gasket with fascia lata	No
14. Chordoma	First	37	F	1	LD	Transclival	LF	2	Yes	Yes	Fat only	No
	Second	37			LD	Transclival	LF	3	Yes	Yes	Fat only	No

Abbreviations: EEA, endonasal endoscopic approach; EVD, external ventricular drain; HF, high flow; LD, lumbar drain; LF, low flow; mo, month; NSF, nasoseptal flap; POD, postoperative day; TC, transcavernous; TP, transplanum; TS, transsellar.

^aRadiation after first EEA.

Patients were divided into two groups, those with adenomas ( n  = 8) and those with other pathology ( n  = 6). The eight adenomas (four women, four men) had a median age of 51.3 years (IQR: 31.7–59.1) at the time of the first surgery, and 57 years (IQR: 39.3–66.1) at the time of the NSF reuse. The adenoma group was significantly older ( p  = 0.005) than the nonadenomas, who had median age of 19.8 years (IQR: 12.3–30) at the time of the first surgery, and 21.4 years (IQR: 14.7–34.5) at the time of the second. Four of the adenomas had comorbidities, three with diabetes mellitus. Use of chronic steroids was similar in both groups.

During first surgeries, eight adenomas had five intraoperative CSF leaks, of which three were HF, two were LF and three had no leak. These eight cases were closed with the following techniques beneath the NSF: nothing ( n  = 1), fat only ( n  = 1), fat plus a Medpore buttress ( n  = 2), fat plus a Medpore gasket with fascia lata ( n  = 2), fat plus a gasket with AlloMax ( n  = 1), button with AlloDerm plus fat ( n  = 1). Six patients had intraoperative CSF diversion (LD in 5 and EVD in 1). There were no CSF leaks in this group. For the six nonadenomas, there were six intraoperative CSF leaks, five HF and one LF. Although there was a trend for more HF CSF leaks in nonadenomas, this did not reach significance. These six cases were closed with the following techniques below the NSF: fat only ( n  = 1), fat with Medpore buttress ( n  = 1), fat plus Medpore gasket with fascia lata ( n  = 2), Medpore gasket with fascia lata ( n  = 2). All patients in this group had intraoperative CSF diversion (LD in 5 and EVD in 1). Of these first operations, there was one postoperative CSF leak in a craniopharyngioma (No. 8) with a HF CSF leak that stopped with a LD ( Fig. 3 ). Below the flap in this case, we used fat plus a gasket with Medpore and fascia lata.

Fig. 3.

Swimmer's plot demonstrating management of 14 patients after takedown and reuse of the same NSF (orange diamond). Time 0 is defined as the first surgery at our institution (green diamond). Each bar represents the f/u period for each patient. The color of each bar indicates either no or low-flow intraoperative CSF leak (light blue) or intraoperative high-flow CSF leak (dark blue). The symbols indicate patients who needed reoperation for the CSF leak (red X mark) or had LD for the CSF leak (orange X mark) or received radiation (yellow triangle). Patients are arranged from the top according to the order in Table 1 . CSF, cerebrospinal fluid; EEA, endoscopic endonasal approach; f/u, follow-up; LD, lumbar drainage; NSF, nasoseptal flap.

During the second surgeries, HF CSF leaks occurred at the same frequency as occurred during the first operations. In one case (No. 3), LF CSF leak occurred only during the second adenoma resection with no leak during the first procedure, and in three patients (No. 1, 2, 10) LF CSF leaks did not recur during the second surgery, even though they were present during the first surgery. In the eight adenomas, underneath the NSF closure was: nothing ( n  = 2), fat only ( n  = 1), fat plus fascia lata ( n  = 1), Medpore gasket with facia lata ( n  = 2), button with AlloDerm plus fat ( n  = 1), Medpore gasket with AlloDerm covered with HydroSet ( n  = 1). Six of these patients had intraoperative LDs placed. In the nonadenomas ( n  = 6), closure beneath the NSF was done the following way: fat only ( n  = 1), fat plus Medpore buttress ( n  = 1), fat plus fascia lata ( n  = 1), button with AlloMax ( n  = 1), Medpore gasket with fascia lata ( n  = 1), AlloMax and fat ( n  = 1). All of these patients had intraoperative LDs. Of these operations, there were two postoperative CSF leaks, both of whom had HF intraoperative leaks. Both required reoperation for repair. Of these, there was one leak in a patient with an epidermoid, which was closed with fat and fascia lata. The other one occurred in a patient with adenoma, who leaked after the second surgery since the HydroSet was not placed smoothly to cover the closure. He was brought back and the HydroSet was bolstered and he did not leak again ( Fig. 3 ). Neither necrosis of the reused flaps nor mucoceles or any other otolaryngological issues necessitating debridement were observed.

Of the three postoperative CSF leaks (No. 7, 8, 12), there were some commonalities. First, all occurred in the setting of a HF intraoperative leak. One postoperative CSF leak occurred after primary resection of a craniopharyngioma, which resolved with LD while the other two cases were observed in recurrent pituitary adenoma and epidermoid cyst and required reoperation. None of these had prior radiation. With regard to the NSF coverage of the skull base, there were 20 cases in which the flap covered the defect and was directly opposed to the skull base. None of these leaked. There were eight cases in which the NSF did not cover the skull base. Of these there were three postoperative CSF leaks (Fisher's exact p  = 0.017). In two of these cases, the flap was opposed to the skull base, one of which leaked. Postoperative CSF leak occurred only in cases in which both a HF CSF leak occurred and there was incomplete coverage of the skull base defect by the NSF ( p  = 0.006). In two cases (No. 7, 8), fascia was used as an adjunctive measure to cover the incomplete coverage of a bony defect by NSF, yet it failed to prevent the occurrence of CSF leak. Use of CSF diversion did not corelate with postoperative CSF leak.

A summary of our treatment algorithm and the relative risk of postoperative CSF leak is graphically represented in Fig. 4 . As is apparent, lower risk cases were sellar, first time operations, adenoma pathology, no or LF intraoperative CSF leak, no use of CSF diversions, with fewer layers of closure and flaps that fully cover the boney defect. In some adenomas with no intraoperative CSF leak, NSF was used for reconstruction since it was harvested at the beginning of the case in anticipation of a large CSF leak since the tumors being removed were quite large. Since we had the flap, we used it to close a large defect in the skull base despite there being no leak. Higher risk cases are suprasellar and posterior fossa, intradural tumors, reoperations, HF leaks, use of CSF diversion, closed with several layers of graft material and flaps that do not fully cover the boney defects.

Fig. 4.

Sankey plot showing the decision flow in management of skull base defects after EEA. The flow plot highlights the three patients in our cohort who had postoperative CSF leak. CSF, cerebrospinal fluid; EEA, endoscopic endonasal approach; NSF, nasoseptal flap; recon, reconstruction.

We then used unsupervised hierarchical clustering analysis (HCA) of select clinical characteristics to show what factors determine the reconstruction technique ( Fig. 5 ). Our HCA identified three separate clusters, which overlapped with the number of layers used for reconstruction. The first cluster represents the low-risk group and contained all the surgeries where one or two layers were applied. Only one patient in this group had a HF CSF leak and in almost all cases NSF was big enough to completely cover bony edges. The second cluster represents the high-risk group of surgeries with half pituitary adenomas, all having CSF diversion and intraoperative HF CSF leak. In half of these cases, the NSF was not sufficient to completely cover the bone defect. This group included two procedures after which CSF leak was observed. The third group consisted of procedures with three-layer reconstruction. There were more intraoperative HF CSF leaks than in the low-risk group. The coverage of bony edges by NSF was incomplete in some cases, and in one it resulted in postoperative CSF leak. Prior radiation occurred with similar frequency in each group. Our HCA indicates that intraoperative HF CSF leak and incomplete coverage of bony edges with NSF increase the risk of postoperative CSF leak.

Fig. 5.

Heatmap representation of unsupervised hierarchical clustering analysis of selected pre-, postoperative clinical characteristics and intraoperative findings in 14 patients reoperated with the NSF reuse. Each column represents one surgery, and they are grouped into three clusters according to unsupervised hierarchical clustering analysis. Red indicates “yes,” and blue indicates “no.” CSF, cerebrospinal fluid; NSF, nasoseptal flap.

Discussion

The risk of CSF leak after EEA ranges widely from 1 to 40%. ^{3 14 15 16 17} The risk of CSF leak when using free grafts to cover large skull base defects is very high. ^{3 18} The vascularized NSF has probably been the most effective innovation at lower postoperative CSF leaks after anterior skull base surgery. ^{2 9 10 11 12 13 14} Factors that increase the risk of CSF include obesity, hypercortisolemia, Cushing's disease, and radiation therapy. ¹⁹ NSF dehiscence and inadequate vascularization can give even worse results due to necrosis. ¹⁵ MRI enhancement of the flap is an indirect method of assessing whether circulation is preserved. ¹⁵ Complete coverage of the bony edges with vascularized NSF may preclude CSF leak even in cases of extended skull base defects. ^{5 20} More than 2-cm dural opening has been associated with NSF failure. ¹⁷

Reoperation has been suggested as one of the risk factors of postoperative CSF leak due to scar removal, poorer tissue vascularization, or shrinkage of the NSF during its takedown. ²¹ The level of scarring or healing can be heterogeneous and may differ between a redo surgery performed after a 5-year versus 1-month interval. Zanation et al conducted a two-center study on 28 patients to analyze the NSF viability and the risk of CSF leak after NSF reuse during 14 revision surgeries and 14 planned second-stage procedures. They observed intraoperative CSF leak in 20 patients and postoperative CSF leak in only 1 patient and had no cases of flap loss. However, their study included patients in whom the first operation was done at a different center. This introduces a bias based on technical variations in flap harvest and skull base closure between centers.

In our study, we only included patients in whom both the first and second operations were performed at our center. We found that the most important factors were not whether it was a reoperation, but rather whether there was a HF CSF leak and whether the flap was adequate to cover the entire defect. Hence, reoperation is only a contributing factor if the recovered flap is not adequate for the job. It is important to keep in mind that the NSF is only one of many layers, and that these other layers also contribute to reducing leak rates. In such circumstances, it is important to maximize the skull base closure with other graft materials. We have recently started using HydroSet (Stryker, United States) in such circumstances. ²² Likewise, LDs have been shown to be effective at reducing the rate of postoperative CSF leaks in patients at high risk. ^{23 24} Judicious use of LDs can help reduce leaks in cases with inadequate flap coverage. In our series, the presence of CSF diversion was not as important as flap coverage. Given the importance of flap coverage, if a re-harvested flap is not adequate to cover a defect, it may be prudent to harvest a second, new flap, which can be done on the contralateral side, or by accessing the lateral wall inferior and middle turbinates. ²⁵

While our study provides valuable insights into the use of NSF for endoscopic endonasal skull base reconstruction, the wide variability in additional reconstruction methods used besides the NSF in our small series makes the results difficult to interpret. This study suffers from the limitations inherent in a retrospective single institution series such as treatment bias based on the treating physicians' notion of the best strategy. The population presented is heterogenous and the goals of surgery can differ for each patient. Other limitations include the small sample size and the variability in repair methods employed beneath the NSF.

Conclusion

NSF reuse is effective at preventing postoperative CSF leak. The most important predictors of leak are the presence of HF intraoperative leak and inadequate defect coverage with the NSF, regardless of whether it is a first or second operation. If the re-harvested flap is inadequate to cover the defect, it may be prudent to harvest a new one from another location.