CSF Rhinorrhoea After Endonasal Intervention to the Skull Base (CRANIAL) - Part 1: Multicenter Pilot Study

Background CRANIAL (CSF Rhinorrhoea After Endonasal Intervention to the Skull Base) is a prospective multicenter observational study seeking to determine 1) the scope of skull base repair methods used and 2) corresponding rates of postoperative cerebrospinal fluid (CSF) rhinorrhea in the endonasal transsphenoidal approach (TSA) and the expanded endonasal approach (EEA) for skull base tumors. We sought to pilot the project, assessing the feasibility and acceptability by gathering preliminary data. Methods A prospective observational cohort study was piloted at 12 tertiary neurosurgical units in the United Kingdom. Feedback regarding project positives and challenges were qualitatively analyzed. Results A total of 187 cases were included: 159 TSA (85%) and 28 EEA (15%). The most common diseases included pituitary adenomas (n = 142/187), craniopharyngiomas (n = 13/187). and skull base meningiomas (n = 4/187). The most common skull base repair techniques used were tissue glues (n = 132/187, most commonly Tisseel), grafts (n = 94/187, most commonly fat autograft or Spongostan) and vascularized flaps (n = 51/187, most commonly nasoseptal). These repairs were most frequently supported by nasal packs (n = 125/187) and lumbar drains (n = 20/187). Biochemically confirmed CSF rhinorrhea occurred in 6/159 patients undergoing TSA (3.8%) and 2/28 patients undergoing EEA (7.1%). Four patients undergoing TSA (2.5%) and 2 patients undergoing EEA (7.1%) required operative management for CSF rhinorrhea (CSF diversion or direct repair). Qualitative feedback was largely positive (themes included user-friendly and efficient data collection and strong support from senior team members), demonstrating acceptability. Conclusions Our pilot experience highlights the acceptability and feasibility of CRANIAL. There is a precedent for multicenter dissemination of this project, to establish a benchmark of contemporary practice in skull base neurosurgery, particularly with respect to patients undergoing EEA.


INTRODUCTION
T he endonasal transsphenoidal approach (TSA) has developed into the approach of choice for resecting pituitary adenoma and most sellar masses. 1,2 More recently, the expanded endonasal approach (EEA) has bolstered endoscopic access to the skull base, allowing resection of many diseases extending beyond the sella alone, including large pituitary adenomas, craniopharyngiomas, Rathke cleft cysts, meningiomas, and clival chordomas. 3,4 Despite the benefits that these minimally invasive approaches afford, cerebrospinal fluid (CSF) rhinorrhea remains a frequent complication, [5][6][7] with potentially serious consequences, including meningitis, pneumocephalus, low-pressure headaches, and prolonged admission. 6,8,9 Arguably, the most important determinant for the development of CSF rhinorrhea is the skull base repair technique used intraoperatively. 4 Other risk factors for postoperative CSF rhinorrhea include previous cranial radiotherapy or surgery; tumor size and infiltration; high-flow intraoperative CSF leak; dural defect size; increased body mass index (BMI, calculated as weight in kilograms divided by the square of height in meters); and surgeon experience. 4,5,7,[10][11][12] A vast array of options and combinations is available for repairing the skull base, including direct closure of the dura using sutures or clips; dural reconstruction using autologous fascia or synthetic materials; vascularized flaps (e.g. nasoseptal and turbinate flaps); avascular grafts (e.g. fat grafts); synthetic grafts; and tissue glues (e.g. fibrin glues). 4,[12][13][14][15] These repair constructs are often supported by buttresses (e.g. septal bone or titanium mesh), nasal packing (e.g. Merocel packs [Medtronic Inc., Minneapolis, Minnesota, USA]), and lumbar drains. 4,13,14 The choice of repair can be graded in response to numerous factors, such as tumor (type, size, hydrocephalus), defect (size, extent of intraoperative arachnoid breach), patient (BMI, sinonasal disease) and operation (approach, primary or revision). 14,16 Previous observational studies suggest that there may be a role for nasoseptal flaps in the context of high-grade intraoperative CSF leak (high-flow leaks with large dural defects). 17,18 In addition, a recent randomized controlled trial suggests that perioperative lumbar drain use combined with nasoseptal flap repair (in the context of dural defects >1 cm 2 and high-flow intraoperative CSF leak), significantly decreases CSF rhinorrhea rates. 19 However, overall, there is a lack of comparative evidence and consensus as to the optimal reconstruction technique; this is the case in high-flow and low-flow intraoperative CSF leaks, small and large dural defects, and primary and revision surgery. 14 Resultantly, there is considerable heterogeneity in skull base repair protocols (largely based on surgeon opinion) 14 with complementary variations in CSF rhinorrhea rates: generally up to 5% for TSA and generally up to 20% for EEA (although as high as 50% in some EEA case series). 4,7,8,[20][21][22][23] CRANIAL (CSF Rhinorrhoea After Endonasal Intervention to the Skull Base) is a prospective multicenter observational study seeking to determine 1) the scope of the methods of skull base repair and 2) the corresponding rates of postoperative CSF rhinorrhea in contemporary neurosurgical practice in the United Kingdom and Ireland. 24 The project is a collaboration between 3 principal bodies: students and junior doctors via NANSIG (Neurology and Neurosurgery Interest Group), neurosurgical specialty trainees via the BNTRC (British Neurosurgical Trainee Research Collaborative), and skull base consultants (neurosurgery and ear nose and throat) via the CRANIAL steering committee. Thus far, 29 centers (of the 40 adult and pediatric neurosurgical centers in the United Kingdom and Ireland) have been recruited to join the project, with each center having a local team of consultants, trainees, junior doctors, and medical students.
Before national dissemination, the project was piloted at selected centers. The usefulness of piloting multicenter studies before scaling is well established and includes assessing protocol feasibility, logistic planning, refining data collection and recruitment instruments, and increasing the investment of key stakeholders. 25 The CONSORT (Consolidated Standards of Reporting Trials) statement 26 has recently extended its guidelines to include feasibility projects, recognizing their role in refining methodologies and processes before definitive multicenter studies. In the context of previous BNTRC studies, reflection on pilot experiences has proved formative in streamlining recruitment, study setup, and data collection before expansion. 27 In this article, the feasibility, acceptability, and practicality of the proposed CRANIAL study are assessed. We present preliminary data collected and outline our experience, the successes, and the challenges in establishing a scalable version of the CRANIAL study.

METHODS Design
A multicenter prospective observational cohort study design was implemented across multiple tertiary academic neurosurgical units in 2 phases. 28  The project was registered as a service evaluation at each center, garnering approvals from audit departments (and Caldicott guardians when required). The local team consisted of consultant lead(s) with overall project responsibility, trainee lead(s) in charge of data collection, and on occasion, student lead(s) for additional support. The STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement was used in the preparation of this article. 29 Eligible patients included those of all ages undergoing TSA for sellar tumors and EEA for skull base tumors. 28 The TSA was defined as surgical access to the sella alone (transsphenoidal), whereas the EEA was defined as acquiring surgical access to an area beyond the sella (e.g. transtubercular or transclival). 24 Exclusion criteria were patients undergoing transcranial surgery and those with a history of preoperative CSF rhinorrhea. Case selection was nonconsecutive because of pauses in collection for data proforma amendments and attaining extra approvals (e.g. information governance approvals when requested). small diaphragmatic defect), grade 2 (moderate leak with obvious diaphragmatic defect), or grade 3 (large leak typically created as a part of EEA). 16 Primary outcomes were 1) methods of intraoperative skull base reconstruction used and 2) postoperative CSF rhinorrhea requiring intervention (CSF diversion and/or operative repair).
Local teams submitted data to a secure Web-based central database hosted by Castor Electronic Data Capture (https://www. castoredc.com/). All data were collected within 30 days of operation. Data points collected by medical students or junior trainees were confirmed with operating surgeons or senior members of the team before final submission into the Castor Electronic Data Capture system. 24 To facilitate accurate and standardized discussion of skull base repair techniques, supportive materials were provided: skull base repair taxonomy, illustrations, and clear definitions. 24 In addition to the data outlined earlier, qualitative data were collected from local pilot trainee leads with an open question "Tell us about your experience during the CRANIAL project e the positives and challenges." This information, along with the procedural experience of the management committee overseeing the project, informed a set of iterative changes to the project.

Data Validation
Data validation was performed in all 3 centers to audit quantitative data accuracy. This process involved an independent data validator (who did not collect local data) who reviewed data sets for several enrolled cases, selected randomly. This data validator was from the hospital in which the data were collected. The targets for validation were a secure and accurate record of Castor identification records with corresponding medical record numbers; no case/ data duplication; and data accuracy >95%.

Data Analysis
Pooled quantitative data (from phase 1 and 2) were analyzed using Microsoft Excel version 16.41 (Microsoft, Redmond, Washington, USA) to present descriptive statistics. The data were used to create tables summarizing demographic, tumor, and operative characteristics. Tumor characteristics, intraoperative skull base repair technique, and dural defect size with associated intraoperative CSF leak grade are graphically shown. Six months follow-up data were not available or complete in most cases because of the recency of data collection and were excluded. Qualitative feedback from local pilot leads from the phase 1 centers was analyzed in terms of content using NVivo software (version 12.6.0). Deductive coding was performed by an independent author (D.Z.K.). Codes were used to generate themes, which in turn were organized into the following categories according to content analysis: 1) pilot positives and 2) pilot challenges.

RESULTS
Data were collected on 187 patients across the tertiary neurosurgical centers, between November 2019 and July 2020 inclusive ( Figure 1). There were no duplicates in cases/data in the records audited for data validation. All centers fulfilled the >95% accuracy target per case.

Skull Base Reconstruction and Support
Skull base reconstruction included the use of dural repair, dural replacement, glues, hemostatic agents, grafts, and pedicled flaps. Compiled EEA and TSA repair technique frequencies per preoperative and operative risk factors for CSF leak are shown in Table 3. Figures 2 and 3 show the heterogeneity of repair technique frequency per center.

Qualitative Feedback
Qualitative feedback was collected from 4 pilot leads in phase 1 of the study, informing improvements for phase 2. The content analysis generated 14 codes, refined into 6 themes (Supplementary Table 1). These themes were then categorized into "positives" and "challenges." Three principal positives were highlighted. First, the data collection interface was complimented: the Castor software was described as "really simple to use, speeds up data collection and is enjoyable to use", and the organization of the data proforma (via logic trees) facilitated efficient data entry: "Not overwhelming the user with all the unnecessary questions (and only loading them if needed)" and the "flow is logical." This collection process was complemented by electronic medical record systems at all pilot centers, allowing pilot leads to establish flexible routines: "15 minutes work a week" with "all electronic notes making the data collection very straight forward." In addition, supportive materials provided to local teams were applauded for their usefulness (sample audit registration forms, study protocol, practical step-bystep guide, and skull base methods explanatory diagrams). Comments included "excellent diagrams explaining the technical nuance of skull base surgery" and "registration was easy because I had a template to follow." Pilot leads were generally met with receptiveness from senior colleagues; one pilot lead organized a meeting with senior operating members of the team who "amended her operation notes to specifically mention the things I need to collect data on." This strategy allowed efficient data collection and consistent data verification.
Local team engagement is crucial to the effective execution of the project. Lapses in this engagement have the potential to present challenges; one pilot lead highlighted "op notes contain limited information, often standardized text" and that it can be "difficult to get a hold of consultants or StR's [specialist trainees] to check with them the data points that need to be checked with them or that weren't clear." Several approaches were adopted in response to this challenge; one pilot lead met with operating surgeons early on to adapt operative notes to include additional CRANIAL data points (e.g. CSF leak grade and dural defect size), whereas another pilot lead compiled data points needing verification into a table for weekly verifications with operating surgeons. Moreover, the volume and complex nature of the data posed a challenge initially. There was heterogeneity in the definitions and categorization for different skull base repair techniques across centers; to address this, a taxonomy diagram, definitions set, and explanatory illustrations were generated as discussed earlier. 24 Specific data points were adjusted and clarified based on feedback (e.g. "size of skull base defect" was refined to "max diameter of dural defect," with categorical answer options [and a "not available" option for instances where this was difficult to ascertain]). Concerns over future compliance with detailed follow-up data were raised; these data points were rationalized and many were made optional to capture primary outcomes without overloading data collectors. The final set of challenges were concerning the future of the project in the context of the coronavirus disease 2019 (COVID-19) pandemic and its impact on endonasal surgery. Guidance for a significant reduction in the amount of endonasal skull base cases was released just after the completion of pilot phase 1 data collection. Resultantly, COVID-19 Crelated data points were added to the data proforma for Phase 2 piloting (reported elsewhere).

Principal Findings
This pilot study has shown the acceptability and feasibility of the current CRANIAL protocol. 28 Acceptability is shown through qualitative feedback from local pilot leads, which was largely positive (user-friendly and efficient data collection, felt supported by central CRANIAL team and seniors). Challenges were addressed iteratively (production of supportive materials and adaptations of data proforma), again met positively by pilot leads. Moreover, feasibility is highlighted through the successful registration and execution of the study at 12 tertiary neurosurgical centers, with high-quality data collected on 187 patients.
As expected, most of these endonasal cases were pituitary adenomas (n ¼ 142/187, 76%) and the most common approach was TSA (n ¼ 159/187, 85%). Although our pilot sample is too small to make conclusions, it is interesting to note the array of repair techniques used. The most common skull base repair techniques used were tissue glues (Tisseel, Adherus, Duraseal, Bioglue, and Evicel) in 132/187 cases (71%), and grafts (most commonly fat graft and Spongostan) in 94/187 cases (50%). These repairs were most frequently supported by nasal packs (Nasopore, Merocel, and Bismuth Soaked Ribbon Gauze) in 125/187 cases (67%). Nasoseptal flaps were used in only 41/187 cases (22%) and lumbar drains were used in 200 /187 cases (11%). Adjuvant conservative and medical prevention of CSF rhinorrhea were equally variable (most commonly laxatives and avoiding straining). The incidence of confirmed postoperative CSF rhinorrhea was 6/159: (3.8%) of TSA cases and 2/28 (7.1%) of EEA cases. In all of these cases, the initial intraoperative skull base repair techniques were heterogeneous. Four of these cases with postoperative CSF rhinorrhea did not have intraoperative CSF leak detected, suggesting occult intraoperative leak. This finding is described in other case series, with some investigators advocating for universal sellar repair or use of routine intrathecal fluorescein to address this. 30,31 Findings in the Context of Literature In our pilot analysis, the encountered postoperative CSF rhinorrhea rates are in line with the array of rates cited in the literature. For TSA, the occurrence of CSF rhinorrhea is generally between 2% and 5% 7,8,20,21 but has been recorded as high as 10% via meta-analysis. 32 Occurrence in EEA is even more diverse (likely reflecting casespecific variations in exact approach), with rates generally ranging from 5% to 20% but as high as 50%. 4,22,23 Risk factors for postoperative CSF rhinorrhea include increased BMI, intraoperative CSF leak (especially if high flow), previous cranial radiotherapy, previous skull base surgery, tumor size, local tumor infiltration, dural defect size, and surgeon experience. 4,5,7,[10][11][12]33 However, potentially the most important determinant for the development of CSF rhinorrhea is related to skull base repair technique used intraoperatively. 4,16 The heterogeneity in skull base repair techniques suggested in our pilot study is echoed in the literature, reflecting the general lack of comparative evidence to guide practice. 14 Practically, many centers use graded repair protocoledependent factors such as dural defect size and CSF leak flow volume. 34 In our series, CSF diversion was used more in the context of tumors >1 cm in diameter, EEA, and high-grade  Table 3). Similar patterns are noted for the use of vascularized flaps, dural replacement grafts, rigid buttresses, and nasal packing on a basis of such CSF leak risk factors ( Table 3). Several noncomparative studies 17,18,23,34 have suggested that in the context of large skull base defects (>3 cm) and/or high CSF flow (via the opening of the ventricle or arachnoid cistern), the use of nasoseptal flaps decreases resultant postoperative CSF rhinorrhea. Some investigators advocate for graft-based reconstruction (fat, fascia, and collagen sponge) in this context, 35,36 whereas others describe a multifaceted approach combining various techniques (e.g. fat, collagen sponge, rigid buttress, and nasal packs) with or without lumbar drain for high-flow leaks with large dural defects. 16,34 The only level 1 evidence supporting practice is a recent randomized controlled trial that found (in the context of dural defects >1 cm 2 and high-flow intraoperative CSF leak repaired with a nasoseptal flap) that the use of perioperative lumbar drain significantly decreased postoperative CSF rhinorrhea rates (P ¼ 0.017; odds ratio, 3.0; 95% confidence interval, 1.2e 7.6). 19 For smaller defects and minor/no CSF leak, fat, fascia, and avascular mucosal grafts are described. 23,36 Other repair protocols support the use of collagen sponge and titanium mesh buttress for such cases. 16 More generally, some surgeons champion dural closure or dural replacements, 37,38 with others suggesting that these have little impact in the context of nasoseptal flap use. 39 Similarly, high-level evidence for postoperative CSF leak repair is equally scarce, with lumbar drains and endonasal direct pedicled flap or graft repair frequently reported. [40][41][42] There is widespread variability in skull base repair protocols; this is the circumstance in both high and low CSF flow situations, and in both prevention and repair CSF rhinorrhea. 14

Limitations
There are several limitations to the study, calling for a tempered assessment of findings. Firstly, because of the pilot nature of this study, results are of small sample size, particularly with respect to EEA. Cases were not necessarily collected consecutively and because of recency of cases, follow-up is limited to the immediate postoperative period (the national project will include up to 6 months of follow-up per case). Data points are purely observational and across the context of multiple centers. Practically, data point verification was sometimes a challenge logistically for junior members of the team, although ways to mitigate this have been presented and will be useful when scaling up this project. One such data point was dural defect, which was not recorded in approximately 30% of cases and in the context of TSA was recorded to include sellar dura (a defect that may not confer the same risk of postoperative CSF leak as dural/arachnoid defects elsewhere).

CONCLUSIONS
Our pilot experience highlights the acceptability, feasibility, and scalability in the CRANIAL project procedures. Early results suggest heterogeneity in methods used for skull base repair. There is a clear precedent for establishing a benchmark of contemporary practice in skull base neurosurgery in the United Kingdom and Ireland via multicenter dissemination of this project.