Reconstructive Outcomes of Multilayered Closure of Large Skull Base Dural Defects Following Open Anterior Craniofacial Resection

Abstract

Introduction  Standardized reconstruction protocols for large open anterior skull base defects with dural resection are not well described. Here we report the outcomes and technique of a multilayered reconstructive algorithm utilizing local tissue, dural graft matrix, and microvascular free tissue transfer (MVFTT) for reconstruction of these deformities.

Design  This study is a retrospective review.

Results  Eleven patients (82% males) met inclusion criteria, with five (45%) having concurrent orbital exenteration and eight (73%) requiring maxillectomy. All patients required dural resection with or without intracranial tumor resection, with the average dural defect being 36.0 ± 25.9 cm ² . Dural graft matrices and pericranial flaps were used for primary reconstruction of the dural defects, which were then reinforced with free fascia or muscle overlay by means of MVFTT. Eight (73%) patients underwent anterolateral thigh MVFTT, with the radial forearm, fibula, and vastus lateralis comprising the remainder. Average total surgical time of tumor resection and reconstruction was 14.9 ± 3.8 hours, with median length of hospitalization being 10 days (IQR: 9.5, 14). Continuous cerebrospinal fluid drainage through a lumber drain was utilized in 10 (91%) patients perioperatively, with an average length of indwelling drain of 5 days. Postoperative complications occurred in two (18%) patients who developed asymptomatic pneumocephalus that resolved with high-flow oxygen therapy.

Conclusion  A standardized multilayered closure technique of dural graft matrix, pericranial flap, and MVFTT overlay in the reconstruction of large open anterior craniofacial dural defects can assist the reconstructive team in approaching these complex deformities and may help prevent postoperative complications.

Introduction

Open anterior craniofacial resection (OACR) was first described by Smith et al in 1954, becoming the gold standard for the management of anterior skull base tumors. ¹ As it classically involves a bifrontal craniotomy to expose the frontal sinus and ethmoid and sphenoid roofs, ² endoscopic approaches to the anterior skull base have now largely supplanted OACR due to their less invasive nature, reduced operative time, and decreased length of hospitalization. ^{3 4 5 6} Indications for OACR are now primarily in the management of endoscopically unresectable disease from orbital soft tissue invasion, intracranial extension, or infratemporal fossa or pterygopalatine space involvement of disease. ³ In cases where tumors involve dura or extend intracranially, formal reconstruction is necessary to prevent potentially life-threatening intracranial complications, such as cerebrospinal fluid (CSF) leak, tension pneumocephalus, and meningitis. ⁷ Goals of reconstruction of the skull base include creation of a watertight division between the intracranial and extracranial compartments, elimination of intracranial dead space, and in instances of concurrent orbital exenteration, frontal craniectomy, or maxillectomy, restoration of the facial skeleton, and soft tissue contour. ^{8 9 10 11 12} While multilayered reconstruction of the skull base has been proven to decrease complications compared to single layer closure following endoscopic resection, ⁶ reconstructive outcomes following OACR with concurrent dural or intracranially extending tumor have been much less described. This is most likely due to the low incidence of anterior skull base malignancies being amenable to OACR, as well as frequent grouping of open anterior skull base with middle and lateral skull base reconstructive outcomes, making interpretation of the results difficult. ^{9 10 11 12} The aim of this study is to describe a reconstructive algorithm utilizing a multilayered technique of dural graft matrix, pericranial flap, and microvascular free tissue transfer (MVFTT) overlay for large surface area and volume deformities of the anterior skull base following OACR with dural and/or brain resection, as well as detail the postoperative complications.

Materials and Methods

Following institutional review board approval, a review of all patients that underwent OACR by bifrontal craniotomy or subcranial approach and required dural or intracranial tumor resection between August 2015 and January 2020 was performed. Data collected included sociodemographics, prior tumor treatment, tumor characteristics, surgical details, length of hospitalization, postoperative complications, and adjuvant therapies rendered. Tumor characteristics included histopathology and staging based on AJCC eighth edition. ¹³ Surgical details recorded were surface area of dura resected, resection of intracranial tumor, need for concurrent orbital exenteration or maxillectomy, MVFTT donor site, local or biologic tissue options used, and presence and duration of continuous lumbar drainage. Precisely, 30-day postoperative complications examined included CSF leak, tension pneumocephalus, and meningitis or encephalitis.

At the authors' institution, the reconstructive algorithm for repairing OACR deformities is depicted in Fig. 1 . Dural defects are preferentially repaired by using a dural graft matrix (DuraGen, Integra, Plainsboro, New Jersey, United States). The dural graft matrix is sutured by using 3–0 Nurolon (Johnson & Johnson Medical NV, Belgium) in a running fashion to ensure as close to a watertight closure as possible. This is followed by a pericranial flap overlay, which is similarly sutured in place. If the pericranial flap or its vascular pedicle are unavailable due to the extent of resection or prior harvest, then vascularized fascia of the MVFTT is used in its place as an overlay to the dural graft matrix repair. Placement of either fasciocutaneous, musculo-fascio-cutaneous, or musculo-fascial only MVFTT is then performed to augment the separation between the intracranial and extracranial spaces, cover exposed bone, supply volume for any potential spaces created, as well as resurface any mucosal or epithelial surfaces such as the external orbit or oral cavity, if indicated ( Figs. 2–4 ). For smaller soft tissue volume deficits without facial skeleton loss, MVFTT donor sites preferred by the authors include the volar forearm or anterolateral thigh (ALT). If there is loss of facial skeletal support, then the osteocutaneous fibula flap or forearm flap/ALT flap with polyetheretherketone (PEEK) implant can be considered to supply both bony structural and soft tissue reconstructive needs. In the reconstruction of larger soft tissue volume deformities, the ALT or subscapular system flaps are used in the authors' practices.

Fig. 1.

Proposed reconstructive algorithm for open anterior craniofacial resection deformities with dural or intracranial tumor resection. cALT, chimeric anterolateral thigh; PEEK, polyetheretherketone; SG, skin graft; VL, vastus lateralis.

Fig. 2.

Patient with large squamous cell carcinoma of the paranasal sinuses and orbit requiring open anterior craniofacial resection and dural resection. ( A ) Intraoperative photo prior to tumor resection. ( B ) Operative defect with dura resection completed. *Indicates area of pericranial flap and duraseal already in place. ( C ) Anterolateral thigh free flap with underlying vastus lateralis muscle in place at the conclusion of the procedure. ( D ) A 6-month postoperative outcome following radiotherapy showing well-healed soft tissue reconstruction.

Fig. 3.

Patient with primary squamous cell carcinoma of the paranasal sinuses requiring OACR with dural resection. ( A ) Resected anterior skull base with orbital exenteration. ( B ) OACR defect with dural graft matrix and duraseal in place. *Demonstrates pericranial flap elevated and prepared for dural overlay reconstruction. ( C ) Frontal view of the soft tissue reconstruction using a vastus lateralis free flap and skin graft. ( D ) A 8-month postoperative outcome following chemoradiotherapy. OACR, open anterior craniofacial resection.

Fig. 4.

Patient with primary squamous cell carcinoma of the ethmoid sinus requiring open anterior craniofacial resection with dural resection. ( A ) Resected anterior skull base with medial maxillectomy and orbital exenteration. ( B ) Frontal view of the intraoperative deformity with pericranial flap inset (*). ( C ) Immediate postoperative outcome with anterolateral thigh in place. ( D ) A 20-month post-treatment outcome following chemoradiotherapy.

All patients undergo computed tomography scanning of the brain on postoperative day 1 to evaluate for pneumocephalus or retained fluid, and another just prior to lumbar drain removal, with serial scans performed thereafter only if neurologic symptoms manifest during the course of hospitalization. Continuous postoperative CSF drainage using a lumbar drain is performed according to surgeon preference and was not uniform over the study time period. In those patients that did undergo drainage, CSF was drained at 10 mL per hour until the postoperative day 3, at which point drains were clamped and then removed 24 to 48 hours afterward depending on development of any concerning symptomatology (e.g., headache or altered status).

Results

Eleven patients met inclusion criteria. Males represented 83% of patients, with an average age of 48 years ( Table 1 ). Three (27%) patients were current smokers. Four (36%) patients had history of prior radiotherapy to the primary site. The most common pathologic indication for OACR was squamous cell carcinoma (64%), with the average dural defect rendered being 36.0 ± 25.9 cm ² (median = 40.0 cm ² , interquartile range [IQR]: 17.5–52.2 cm ² ). Five (45%) patients required concurrent orbital exenteration and eight (73%) required concurrent maxillectomy. Mean length of surgery including both resection and reconstruction was 14.9 ± 3.8 hours (median = 13.8 hours, IQR: 12–18.3 hours). All patients underwent a multilayered reconstruction as described above, with nine (82%) undergoing combined dural graft matrix, pericranial flap, and MVFTT, and two (18%) undergoing dural graft matrix and MVFTT with overlay vascularized fascia due to loss of pericranial tissue. All patients incurred active CSF leaks intraoperatively. Ten (91%) patients had lumbar drain placed intraoperatively with continuous drainage requiring an average of 5 days (IQR: 4.25–5 days) postoperatively.

Table 1. Open anterior craniofacial resection patient and operative characteristics.

Patient	Age	Sex	Pathology	TNM stage	Prior XRT	Reconstruction method	MVFTT donor site	Surface area of dural defect (cm 2 )	Intracranial tumor resection	Other resections	Duration of LD (d)	Complications	Adjuvant XRT administered	Length of follow-up (mo)	Vital status
1	29	M	ACC	T4N0M0	Y	DGM PF MVFTT	ALT + VL	56.25	Y	OE, M	4		Y	12.2	Dead of disease
2	40	F	SCC	T3N2M0	N	DGM PF MVFTT	Fibula	15	N	M	6		Y	19.5	Alive + NED
3	58	M	SCC	T4N0M0	N	DGM PF MVFTT	AMT + RF	40	N	OE, M	5		Y	20.3	Alive + disease
4	44	M	SCC	T4N0M0	N	DGM PF MVFTT	ALT + VL	40	N	OE, M	5		Y	20.5	Alive + disease
5	39	M	SCC	T4N0M0	N	DGM PF MVFTT	ALT + VL	15	N	M	5	Pneumocephalus	N	1.2	Dead other cause
6	53	M	SCC	T4N0M0	Y	DGM PF MVFTT	ALT + VL	100	Y		11	Pneumocephalus	N	2.1	Dead of disease
7	60	M	SCC	T4N0M0	Y	DGM MVFTT	ALT + VL	25	N	M	3		N	27.5	Alive + disease
8	65	F	SCC	T4N0M0	N	DGM PF MVFTT	ALT + VL	40	N	OE, M	5		Y	16.8	Alive + NED
9	41	M	Paraganglioma	N/A	N	DGM MVFTT PEEK	Radial forearm	12	N		4		N	23.7	Alive + NED
10	33	M	Solitary fibrous tumor	N/A	N	DGM PF MVFTT	ALT + VL	40.9	N		0		N	13.3	Alive + NED
11	67	M	SNUC	T4N0M0	Y	DGM PF MVFTT	ALT	12	N	OE, M	5		N	3.1	Alive + NED

Abbreviations: ACC, adenoid cystic carcinoma; ALT, anterolateral thigh; AMT, anteromedial thigh; DGM, dural graft matrix; M, maxillectomy; MVFTT, microvascular free tissue transfer; NED, no evidence of disease; OE, orbital exenteration; PEEK, polyetheretherketone; PF, pericranial flap; RF, rectus femoris; SCC, squamous cell carcinoma; SNUC, sinonasal undifferentiated carcinoma; VL, vastus lateralis.

The median length of hospitalization was 10 days (IQR: 9–13). Two (18%) patients developed asymptomatic pneumocephalus detected postoperatively on routine postoperative CT scans. No patient developed postoperative complications of CSF leak, meningitis, wound dehiscence, encephalitis, or naso-oro-cutaneous fistula. There were no microvascular thrombotic complications in our cohort. The average length of postoperative follow-up was 17 months (IQR: 8–20). One (9%) patient with schizophrenia was discharged to a mental health facility where he passed away on postoperative day 35 of unclear reasons, and an autopsy was refused by his family. Five (45%) were alive with no evidence of recurrent disease, three (27%) were alive with recurrent disease, and two (18%) died of their disease.

Discussion

OACR has been proven to be effective in managing locally extensive anterior skull base malignancies through wide exposure and visualization of the skull base, maximizing complete and en bloc resection of tumors. ¹⁴ The most common pathologies requiring OACR are squamous cell carcinoma (28.8%), adenocarcinoma (16.1%), and esthesioneuroblastoma (11.6%), with 5-year overall/disease-free survival reported to be 44.4/49.9%, 51.5/53.1%, and 77.8/64.3%, respectively. ¹⁵

While the effectiveness of OACR is not controversial, reconstructive algorithms to minimize postoperative complications have been less elucidated. This is largely due to the low incidence of procedures performed but is also due to frequent grouping of different reconstructive modalities together (local, regional, and MVFTT) as well as anterior skull base defects with middle and posterior deformities, making subgroup analysis of reconstructive modalities impossible. Overall complication rates following OACR range between 32 and 46%, ^{4 15 16} but more importantly, intracranial complications range between 11 and 29% ^{17 18} and are associated with dura or brain resection. ¹⁵ Therefore, watertight and robust reconstruction of these open dural deficits to separate the intracranial contents from the external environment is of paramount importance, and has classically been achieved with local reconstructive techniques such as pericranial or temporalis flaps. ¹⁶ The use of these local flaps is an excellent option for anterior skull base reconstruction due to their close proximity, but they often lack sufficient volume for dead space reconstruction and may not have sufficient surface area to repair larger dural deficits. Additionally, any loss of the flap or error in its inset leaves little room for prevention of CSF leak or other complications without other tissues available for reinforced coverage. ¹⁹ Not surprisingly, the results of local flap reconstruction for OACR defects have demonstrated a 27% rate of intracranial complications ^{15 16} and postoperative CSF leaks between 8 and 20%. ^{15 16 20 21 22}

Following introduction of MVFTT in OACR reconstruction, CSF leak rates have seen a drastic improvement but reported rates remain between 5 and 8.5%. ^{16 18 23} MVFTT is effective in reconstructing complex, large skull base defects due to the ability to tailor the flap to specific defect volumes, select specific tissue components required, as well as withstand postoperative radiotherapy. ^{9 17 24} Unfortunately, the rate of CSF leak following MVFTT is difficult to quantify as much of the literature does not adequately compare the outcomes of OACR with or without dural or intracranial tumor resection, and in reports where dural resection is listed, dimensions of the deformities are not available. This distinction is important as it has been demonstrated that OACR with dural resection is much more likely to result in postoperative CSF leak than in cases without. In a retrospective study of 38 OACR cases, Moiyadi et al encountered CSF leak postoperatively in 7.9% of cases, where dural resection was performed and in 0% of those without dural resection. ²² Recently, Gill et al, in an investigation of MVFTT outcomes of anterior skull base reconstruction, demonstrated a 0% CSF leak rate in cases where dura was left intact and a 22% rate in those requiring dural resection. ²³

In this study, all patients underwent OACR with significant dural defects created (36.0 ± 25.9 cm ² ). These defects, in the authors' opinion, are not amenable to local flap reconstruction alone, and a reconstructive algorithm was created to streamline the decision process in management. Dural deformities are primarily reconstructed with dural graft matrix and, if available, pericranial flap overlay. Dural graft matrix has been well described in dural reconstruction but has been mainly used in endoscopic repairs of defects less than 2 cm, with reported CSF leak rates of 1.5 to 3%. ^{25 26} Performance of this multilayered technique for primary dural reconstruction may aid in reinforcing imperfect dural sutures and provide extra watertight coverage. In cases where no pericranial flap is available to reinforce the dural graft matrix, then the subcutaneous fascia of a fasciocutaneous free flap or investing fascia of a chimeric muscle flap (e.g., vastus lateralis muscle) can serve as a vascularized overlay graft. It is the authors' preference to augment the dural reconstruction with free adipose or muscle tissue to further buffer the separation of intra- and extracranial compartments. The choice of tissue employed for augmenting the dural reconstruction is ultimately dependent on the size of the external defect and whether or not there is need for facial skeleton reconstruction. In cases where large volume soft tissue and external epithelial coverage is required, such as naso-orbito-frontal resections with or without orbital exenteration, larger chimeric flaps such as the musculocutaneous ALT can be employed. In cases of orbital preservation with minimal volume or surface area deformities, adipofascial forearm or ALT flaps may be ideal. Lastly, when there is need for facial skeleton reconstruction (frontal bar or maxilla), those with smaller concurrent soft tissue deformities may be amenable to PEEK implant and adipofascial MVFTT or osteocutaneous fibula MVFTT. Larger volume soft tissue deformities with facial skeleton deficiencies may benefit from PEEK implant and musculocutaneous MVFTT, or an osteomyocutaneous flap such as one based on the subscapular system. Using this algorithm, we were able to achieve postoperative CSF leak and fistula rates of 0%. Intracranial complications were limited to two (18%) cases of asymptomatic pneumocephalus that resolved with high-flow oxygen therapy alone.

The multilayered closure of skull base deformities is not a novel concept. It has been employed and proven extensively in the endoscopic skull base literature. Tabaee et al compared Gasket seal with inlay of fat to Gasket seal without inlay fat and demonstrated postoperative CSF leak rates of 8.8 and 33%, respectively. ²⁷ The translation of this concept to large open anterior skull base deformities has not been well described; however, Weber et al in 2007 proposed a reconstructive algorithm for skull base deformities with MVFTT donor site choice determined by volume of the defect (small through large), and a separate designation for defects with recalcitrant CSF leak. ¹⁸ Although this algorithm addresses the importance of volume-based reconstruction, it does not incorporate treatment decisions necessary for dural and bony replacement and does not employ a specific multilayered reconstructive approach.

The most common complication in our series was asymptomatic pneumocephalus (18%) detected on routine postoperative head computed tomography scan. This rate is much higher than that reported in the MVFTT skull base reconstruction literature, which is approximately 3%. ^{17 18 28} This metric is difficult to compare, however, as most reports document tension pneumocephalus, and the rate of asymptomatic pneumocephalus is not described. As 10 (91%) patients had continuous lumbar drainage postoperatively, this may be an explanation for our higher rate of pneumocephalus. In a review of 161 patients that underwent anterior skull base resection, perioperative tension pneumocephalus (3.1%) was found to be significantly associated with continuous lumbar drainage. ²⁸ Placement of a lumbar drain at our institution is surgeon and situation dependent. The benefit of continuous CSF drainage postoperatively is that decreased intracranial CSF pressure for the first few days allows for healing without any potential leak. While this is at perceived odds with the risk of developing tension pneumocephalus, if the dural defect is adequately closed, there generally should no longer be a risk of continuous in-flow of air. Our lumbar drain protocol did not have to be adjusted for low pressure symptoms (e.g., headache or nausea) in our patient population. Future prospective investigation of variable lumbar drainage in the setting of large dural defect reconstruction and MVFTT is warranted.

Important limitations in this study include retrospective design and small patient population. As this limitation is a function of the infrequency of OACR with dural resection, our series compares favorably to many in the literature while also reporting key relevant data such as dural resection surface area. All cases were performed at a single institution, which may limit the applicability of results globally, but considering all reconstructive surgeons (A.D.H., D.J.H., and A.T.H.) came from different institutional training backgrounds, this supports the adoption of a specific reconstructive algorithm in order to streamline and maximize postoperative outcomes.

Conclusion

Multilayered reconstruction of OACR defects with dural resection utilizing dural graft matrices, local flaps, and MVFTT may mitigate concerning intracranial complications such as CSF leak and fistula. Adoption of a reconstructive algorithm for these defects can produce uniform results among reconstructive surgeons. Further multi-institutional investigation of this algorithm and of the role of continuous lumbar CSF drainage would be beneficial.

Funding Statement

Funding None.