Multi-layered repair of high-flow CSF fistulae following endoscopic skull base surgery without nasal packing or lumbar drains: technical refinements to optimise outcome

Aims Post-operative CSF leak remains a significant problem following endoscopic skull base surgery, particularly when there is a high-flow intra-operative CSF leak. Most skull base repair techniques are accompanied by the insertion of a lumbar drain and/or the use of nasal packing which have significant shortcomings. Our aim was to review the results of a large series of endoscopic skull base cases where a high-flow intra-operative CSF leak rate was encountered and repaired to assess if modifications in technique could reduce the post-operative CSF leak rate. Methods A retrospective review of a prospectively maintained database of skull base cases performed by a single surgeon over a 10-year period was performed. Data regarding patient demographics, underlying pathology, skull base repair techniques and post-operative complications were analysed. Results One hundred forty-two cases with high-flow intra-operative CSF leak were included in the study. The most common pathologies were craniopharyngiomas (55/142, 39%), pituitary adenomas (34/142, 24%) and meningiomas (24/142, 17%). The CSF leak rate was 7/36 (19%) when a non-standardised skull base repair technique was used. However, with the adoption of a standardised, multi-layer repair technique, the post-operative CSF leak rate decreased significantly (4/106, 4% vs. 7/36, 19%, p = 0.006). This improvement in the rate of post-operative CSF leak was achieved without nasal packing or lumbar drains. Conclusion With iterative modifications to a multi-layered closure technique for high-flow intra-operative CSF leaks, it is possible to obtain a very low rate of post-operative CSF leak, without lumbar drains or nasal packing.


Introduction
The most significant advance in the management of skull base pathology over the last 20 years has been the introduction of endoscopic endonasal approaches (EEA) to lesions of the ventral skull base [25]. These approaches serve as an alternative to open craniotomy and may, in selected circumstances, lead to higher rates of gross total resection, improved neurological outcomes, lower complication rates and shorter hospital stays [8,18,24].
However, a major initial shortcoming of EEA for skull base pathology was the associated high rate of post-operative cerebrospinal fluid (CSF) leak; in early published series, the incidence of this complication often exceeded 20% [35]. CSF leak is the most common source of morbidity following endoscopic skull base surgery, and can lead to the development of life-threatening intracranial infection and pneumocephalus [19,32]. Patients undergoing expanded approaches, involving large arachnoid breaches and highflow intra-operative CSF leaks, are particularly susceptible to this complication [5,13]. The adoption of the nasoseptal flap (NSF) was instrumental in attempts to reduce the incidence of post-operative CSF leak, and it is now commonly used in skull base repair following EEA [9,25].
Multiple skull base repair protocols have been described, and a review of these techniques and their associated results has recently been published by our group [11]. In addition to the use of a synthetic or autologous dural substitute and a NSF, these techniques are often supplemented with intranasal packing or lumbar drainage (LD) [7,36,37]. Our current practice involves the use of autologous fascia lata and a NSF, secured with BioGlue® (CryoLife Inc., USA), without the requirement for any nasal packing or adjunctive lumbar drainage [11]. We have previously described the results of our skull base repair technique in 270 patients with all grades of intra-operative CSF leak, and reported a CSF leak rate of 1% in the latter third of the series [10]. This study focuses on the results of our skull base repair protocol following EEA when a high-flow intra-operative CSF leak was encountered, and describes the adaptions in our repair protocol that have led us to the current iteration of our skull base repair protocol.

Materials and methods
All endoscopic skull base cases performed by a single surgeon, where an intra-operative high-flow CSF leak was encountered between January 2012 and August 2022, were recorded prospectively in a clinical database and included in this study. Intra-operative CSF leaks were graded according to the Kelly grading system and cases with grade 3 leaks 'large diaphragmatic and/or dural defect' were included [5]. For all pathologies except pituitary adenomas, expanded approaches beyond the sella were used. Post-operative CSF leaks were diagnosed by the 'tilt' test in keeping with other studies in this area [3]. The incidence of post-operative meningitis and sinusitis was obtained via retrospective review of the medical records.
Our skull base repair technique has undergone several modifications over the 10-year period studied. Our most commonly used technique has involved the use of 2 layers of fascia lata, NSF and a dural sealant. This 'standardised technique' involves multi-layer closure with the placement of an inlay (intradural) layer of endogenous fascia lata, with a NSF placed over the bony and dural defects. Following this, a further layer of fascia lata is placed over the NSF, and the entire construct is secured with BioGlue® tissue sealant, applied extradurally, as described previously [11] (Fig. 1). Previously, we have also placed an onlay (extradural) fascia lata graft prior to placement of the NSF, which was then subsequently covered by the NSF and secured with BioGlue®.
The 'non-standardised technique', rather than being a single, unified technique, reflects a time period when nonstandardised techniques were used in the earlier part of the series and involved a variety of combinations of intradural placement of fat graft, synthetic dural substitute, Dura-Guard (Baxter, USA) or Duramend (Collagen Matrix, USA) or endogenous rectus sheath which was covered by a NSF and secured with BioGlue® or DuraSeal® (Integra LifeSciences, USA).
No nasal tampons or Foley catheters were used to buttress the skull base repair, and LDs were not utilised as an adjunct to any of the repair techniques.
All statistical analysis was performed in SPSS version 25 (IBM, USA). The χ 2 (with Yate's correction where necessary) and Fisher's exact tests were used for intergroup comparisons of categorical variables. Bonferroni corrections were used to correct for multiple comparisons. This project was approved by our institutional audit board and the requirement for individual patient consent was waived, as per institutional policy. The STROBE guidelines were followed in the drafting of this manuscript [33].

Results
During the study period (January 2012-August 2022), 541 EEAs were performed by a single surgeon. Of these, 142 EEA resulting in a high-flow CSF were performed. Table 1 provides a breakdown of the demographic details, underlying diagnosis and post-operative complications by repair technique implemented. Craniopharyngiomas (39%), pituitary adenomas (24%) and meningiomas (17%) accounted for the majority of cases in this series, with the remainder composed of other pathologies as described in Table 1.
The rate of post-operative CSF leak with our standardised technique was 4/104 (4%). This was significantly lower than the post-operative CSF leak rate associated with the nonstandardised technique which was 7/36 (19%). Table 2 demonstrates the results of univariate analyses of potential predictors of post-operative CSF leak. Use of a standardised repair technique, rather than the non-standardised technique, was the only factor associated with a significant decrease in the rate of post-operative CSF leak (4/106, 4% vs. 7/36, 19% Fisher's exact test, p = 0.006). Surgical approach used (i.e. transtubercular, transsellar, transclival) had no impact on the rate of postoperative CSF leak. The standardised technique was used significantly more commonly as the series progressed (χ 2 = 52.59, p ≤ 0.001), and the non-standardised technique was last used in 2020. Figure 2 illustrates these trends in the data.

Discussion
The results of this study demonstrate that with appropriate adaptions in skull base repair protocol, it is possible to significantly reduce the post-operative CSF leak rate following EEA, even in the presence of high-flow intra-operative CSF leaks. These results were achieved by means of iterative modifications to an existing repair protocol utilising multilayered closure including a NSF and without the requirement for nasal packing or prophylactic LD insertion. The post-operative CSF leak rate in the early part of our series using a non-standardised approach was 19%. After introduction of standardised techniques, the post-operative CSF leak rate dropped to 4%. On univariate analysis, the only factor found to be predictive of post-operative CSF leak in this series was the repair technique used: the use of standardised repair techniques was associated with a significant decrease in the rate of post-operative CSF leak when compared to a non-standardised repair technique.
Undoubtedly, the introduction of the pedicled NSF represented a major technical advance in the field of endoscopic skull base surgery, and facilitated the continued evolution of the indications for EEA to include extra-sellar skull base pathology, which may not otherwise have been possible due to the prohibitively high rates of CSF leak [9,20]. The widespread adoption of the NSF coincided with a reduction in the CSF leak rate from ~ 20 to ~ 5% in the published literature, and further developments since then have been largely iterative, with CSF leak rates following EEA with all degrees of intra-operative CSF leak rate varying between 2 and 4% [2,3,7,14,16,21,22,31,36]. The NSF has now been incorporated into a number of skull base repair techniques, often accompanied by adjuncts intended to provide additional support to the repair. Garcia-Navarro et al. previously described one such technique known as the 'Gasket Seal Closure' that resulted in a post-operative CSF leak rate of 4.5% [7]. In their publication, the authors argued that the  Our use of BioGlue™ (CryoLife, USA) tissue sealant, which has a demonstrably higher tensile strength than other tissue sealants, may potentially be the factor that obviated the need for a buttress to reinforce the skull base repair [26]. However, we did not compare this with other tissue sealants and therefore no definite conclusions can be made regarding the most optimal type of tissue sealant. It must also be noted that previous concerns have been raised in the literature regarding an increased rate of CSF fistulae and postoperative infection in association with the use of BioGlue®. We have previously published a study that demonstrated no increase in the rate of post-operative infection or sino-nasal morbidity when this product was used following endoscopic transsphenoidal surgery [6,12,30].
Prophylactic LDs are used by many skull base surgeons in the post-operative period with the aim of reducing the risk of post-operative CSF leaks [11]. Even surgeons who do not use LDs routinely in endoscopic skull base surgery tend to use it in all or at least in selected patients with highflow intra-operative leaks; a recently published multi-centre analysis of skull base repair protocols following EEA identified that prophylactic LDs were used in 41% of cases where a high-flow intra-operative CSF leak was encountered [4,7,28]. A randomised controlled trial (RCT) found the use of LDs to be beneficial in endoscopic skull base surgery with high-flow intra-operative leaks, reducing the risk of postoperative CSF leak from 21 to 8% [37]. However, LDs have also been associated with significant morbidity including low-pressure headache, nausea and vomiting in 13-63%; meningitis and other infections in 4-10%; radiculopathy; overdrainage with risk of subdural haemorrhage; and complications associated with decreased patient mobilisation [1,27,29]. Interestingly, a further RCT of LD vs. no LD following endoscopic skull base surgery with a high-flow intra-operative CSF leak demonstrated no difference in the rates of post-operative CSF leak, but a markedly increased rate of complications and length of stay for those patients assigned to the LD group [15]. Using our standardised technique, we have achieved a low post-operative CSF leak rate in patients with high-flow intra-operative CSF leaks, without the use of LDs in any of our patients. The lower post-operative leak rates in our series compared to the rates obtained in the RCT by Zwagerman et al. may be due to differences in repair technique [37]. However, the proportion of patients with posterior fossa pathology (such as chordoma and chondrosarcoma) was much lower in our series (7% vs 29%), which may act as a confounder given the technical difficulty often encountered in achieving watertight closure of extensive clival defects.
From our perspective, we feel that the placement of an inlay fascia lata graft as a dural substitute, beyond the margins of the bony/dural defect, is of critical importance. A fascia lata graft larger than the dural defect is placed intradurally, and opposed to the internal surface of the skull base. In our view, the placement of this inlay graft allows the natural brain/CSF pulsations to push the graft against the dura mater with each pulsation, as opposed to the risk of an extradural onlay graft migrating away from the defect. We have not observed any migration of the fascia lata away from the dural defect, which was the rationale for the creation of the 'Bilayer Button' by Luginbuhl et al. [22]. While the benefits of the NSF in the prevention of a post-operative CSF leak have been well documented previously, a nuance of the skull base repair technique we feel is particularly important is the direct apposition of endogenous tissues to each other, as in our view they adhere to each other much better than endogenous tissues to do synthetic materials such as dural substitutes [2,3,7,14,16,21,22,31,36]. Furthermore, although we cannot provide any empirical evidence in support of this assertion, we are of the view that one should not permit synthetic repair materials such as tissue sealant or haemostatic agents to intercede between the fascia lata and NSF, as these may act to prevent the adherence of these endogenous graft materials to each other. As mentioned previously, BioGlue® forms an extremely firm supportive layer once set, which counteracts the CSF pulsations which may otherwise lead to failure of the repair. Moreover, our repair technique does not involve the placement of any nasal packing or catheters. It has been observed that patients find the placement and Fig. 4 Multi-panelled figure demonstrating the repair of a defect in the tuberculum sella/ planum sphenoidale following the resection of a suprasellar craniopharyngioma. A Intra-operative endoscopic photograph demonstrating the skull base defect and view into the third ventricle following tumour resection. B Endoscopic photograph demonstrating the placement of an inlay fascia lata graft. Note that this graft is significantly larger than the dural defect. C Endoscopic photograph demonstrating the final position of the fascia lata graft, with the edges tucked in underneath the margins of the skull base defect. D Endoscopic photograph demonstrating placement of a pedicled nasoseptal flap to completely cover the skull base defect followed by the (E) placement of an onlay fascia lata graft which is in turn secured with (F) a layer of BioGlue. G Pre-operative T1-weighted MRI scan demonstrating the presence of a suprasellar craniopharyngioma with superior extension into the third ventricle. H Post-operative T1-weighted MRI scan demonstrating complete tumour resection and the enhancement of the pedicled nasoseptal flap overlying the skull base defect. I Diagrammatic representation of the Dublin technique as applied to approaches to sellar/ suprasellar pathology removal of nasal packing material to be the most uncomfortable part of the recovery from EEA to skull base lesions, and we agree with the developers of the 'Gasket Seal Closure' that the optimal skull base repair protocol should avoid this [23]. Finally, although there is some evidence to suggest that prophylactic lumbar drains are efficacious in the prevention of post-operative CSF leak, due to the potential for significant complications, we believe that their use is best avoided if possible [1,27,29,37].
The strengths of this study include its prospective nature, and that consecutive cases were included, limiting the influence of any selection bias. The study is further strengthened by the use of robust, validated technique for the grading of intra-operative CSF leaks [5]. The wide variety of pathology included (Table 1) serves to broaden the applicability of the results, and make them applicable to any EEA where a high-flow CSF leak is encountered. A potential confounding factor in this study was that increasing surgical experience, rather than any improvement in surgical technique, may have been responsible for the decrease in post-operative CSF leak. We specifically assessed for a potential influence of increasing experience on univariate analysis ( Table 2), but the only factor significantly associated with the incidence of this complication was the skull base repair technique. Moreover, the senior author had been practicing as an independent consultant neurosurgeon for 5 years at the time of the initiation of this series, and may have already overcome the learning curve associated with EEA [17]. Although it has been argued that continued improvement in outcome can be observed at the 'tail end' of the learning curve, the publication introducing this concept found no impact of further experience beyond 200 cases on the rate of post-operative CSF leak rates, and this finding has recently been replicated in a further study of over 1000 EEA procedures [13,34]. Undoubtedly, this study is limited by the fact that the results represent those of a single-surgeon, single-institution series, and these results may not be replicable in other settings. We intend to validate the results of the present study with a multi-centre, multi-surgeon study in the future.

Conclusions
With iterative modifications to a multi-layered closure technique based on the use of a vascularised NSF, it is possible to obtain very low rates of post-operative CSF leak, even when a high-flow intra-operative CSF leak is encountered. This was achieved without the use of LDs and/or nasal packing, suggesting that these invasive adjuncts may not be necessary.

Fig. 5
Multi-panelled figure demonstrating the repair of a clival defect following the resection of a chordoma. A Pre-operative T2-weighted MRI scan demonstrating the presence of an extensive clival chordoma, causing severe brainstem compression. B Post-operative T1-weighted MRI scan following contrast administration, demonstrating complete tumour resection with resolution of brainstem compression, along with enhancement of the nasoseptal flap. C Intraoperative endoscopic photograph following tumour resection, demonstrating a longitudinally extensive clival defect. The basilar artery and brainstem are clearly displayed. D Endoscopic view of the vertebrobasilar junction and emergence of the right anterior inferior cerebellar artery. E Intra-operative photograph demonstrating placement of the inlay layer of fascia lata within the dural/bony defect. F Final overview of the skull base repair, with a layer of BioGlue applied over an onlay fascia lata graft. G Diagrammatic representation of the Dublin technique as applied to approaches to clival pathology