Pituitary adenomas account for 15% of all intracranial neoplasms, making them the third most common pathology (1). Pituitary adenomas, recently renamed as pituitary neuroendocrine tumors (PitNETs) in the new classifications (2–4) are usually benign tumors, with a broad spectrum of biological and pathological characteristics (5–7). Surgery represents the first-line treatment for most pituitary adenomas (corticotroph, somatotroph and non-functional), except for most prolactinomas which are currently treated with dopamine agonists (8).

The transsphenoidal approach is the gold standard of surgical route for pituitary surgery (9–11), while the transcranial approach is considered only as a second-line surgical option, in well-selected patients with rare tumors extending anteriorly in the subfrontal area, laterally in the temporal fossa or encompassing the vessels (12, 13). The microscopic technique, initially developed by Cushing and successively taken forward by Dott, Guiot and Hardy (14–16), was progressively abandoned in most centers of excellence, leading to the transition towards the endoscopic technique in the late 1990s (10, 17–19). Nowadays, the endoscopic endonasal transsphenoidal approach is the mainstay in many centers worldwide, given its advantages in terms of quality of vision, tumor resection, endocrine outcome, sinus-nasal morbidity and length of hospital stay (17, 20, 21).

Despite advances in closure techniques, postoperative CSF (cerebrospinal fluid) leak remains the most common complication of the endoscopic endonasal transsphenoidal approach, occurring in around 10% of patients and requiring a specific second surgery. The rate of CSF leak-related meningitis is observed in approximately 5% of patients, increasing the length of hospital stay and medical costs (22, 23). Thus, a reconstruction strategy of the skull-base defect should be anticipated during preoperative planning and an accurate analysis of preoperative risk factors associated with CSF leak is essential. In previous studies, many preoperative risk factors have been reported such as BMI (body mass index) > 30, multiple surgeries, tumor size, extension and invasiveness, prior treatment with radiotherapy (24–29). In all these patients, the closure strategy should be rigorously considered before the sellar surgical step, in order to limit avoidable reconstruction failures (24, 30) and to modulate it according to the flow of the CSF leak observed during surgery (31–34).

Today, several protocols to reduce postoperative CSF leak have been proposed (33–40) but there is no consensus on the best closure strategy after endoscopic pituitary surgery. Moreover, a significant heterogeneity in the outcomes was reported (41). From the analysis of the first 1000 patients operated on with a mononostril endoscopic endonasal transsellar approach, we reported in 2014 a postoperative CSK leak rate <1% (18). Based on our substantial additional experience, we developed a gradual closure strategy that integrates individual preoperative risks of postoperative CSF leak with the operative findings.

The objective of the present study is to analyze the results of our closure strategy from a consecutive cohort of 3015 patients with pituitary adenomas, operated on by the same two senior expert neurosurgeons (S.G, B.B). The philosophy of this major surgical step is emphasized, highlighting the need to plan the closure step before entering the operative room.

This paper focuses on the evaluation of the closure step in pituitary surgery, based on a large series of 3015 adenoma patients treated with a mononostril endoscopic endonasal transsellar approach by the same expert surgical team. In PitNETs, invasion of the basal dura, cavernous sinus and/or diaphragm is encountered in up to 35% of patients (2), explaining the substantial occurrence of intraoperative leak requiring reliable watertight closure techniques. Initially the endoscopic technique was associated with high reported rates of postoperative CSF leak, ranging from 30 to 40% (47). With advances in endoscopic pituitary surgery and the development of new reconstruction techniques, optimized lower rates <10 % are currently reported (24, 40, 48–50). The present study demonstrates that our stepwise strategy is safe, reliable and effective. Applying our decision strategy, our rates of postoperative CSF leak and meningitis were 1% and 0.8% respectively, which compares favorably with the best rates previously reported (34, 51).

In pituitary surgery, the crucial step of closure should be anticipated. The neurosurgeon should keep in mind that the closure step already begins during the approach. Even if no intraoperative leak occurs, the overall concept is to preserve a good epidural space during the approach and to recreate a sellar floor whenever possible. Indeed, sellar floor closure is useful to identify bony landmarks in the event of a later second surgery and may limit the risk of postoperative CSF leak related to a secondary stall of the diaphragm during sneezing or blowing. During the approach, the sellar floor is opened is such a way that the bone opening is larger than the dural opening, in order to preserve the epidural space (34, 38, 52). When possible, an autologous bone graft is removed and preserved from the sphenoidal rostrum and/or one of a sphenoid sinus septation. If no bone graft is available, a synthetic resorbable polydioxanone (PDS) plate can be used with similar shaping and positioning (53). We choose to use the PDS material because of its stiffness, which is close to that of bone. We recommend the use of a resorbable sellar floor substitute to limit the risk of infectious complications.

In case of intraoperative CSF leak, our patients were divided into two groups on the basis of selected preoperative risk factors of closure failure (28, 29, 43). A specific gradual closure strategy was planned accordingly. Interestingly, other preoperative risk factors of closure failure have been reported recently by expert teams, such as suprasellar extension, chronic respiratory disease, type of sellar barrier, fibrous consistency, dumbbell-shape or lobulated asymmetrical configuration (42). A comprehensive analysis of all these factors should allow surgical management to be tailored to the individual patient.

In patients with no risk factor, the objective is to achieve a watertight closure and reconstruct the sellar floor. Different materials have been proposed for the reconstruction of skull-base defects. Autologous materials have been proposed, such as mucosal grafts from middle turbinate, fat grafts from abdominal region, muscle graft from lateral thigh, fascia grafts from fascia lata, lateral thigh or temporal muscle (31, 32, 48, 53–58). Heterologous biologic dural substitutes have been used, such as equine pericardium sheet (59) or human-derived acellular dermal matrix (60). Heterologous synthetic dural substitutes have also been proposed, such as polyester-silicone (61), resorbable polyglactin acid sheet (62), polytetrafluoroethylene (63) or collagen matrix (64, 65). At the end of the closure procedure, fibrin glue should is usually applied inside the sphenoidal cavity to fill the dead spaces (66–68). In our strategy, the key objective was to minimize the risk of postoperative leak while minimizing the proportion of patients treated by packing the sella or sphenoid sinus. Indeed, packing the sellar area with additional fibrous scar tissue may impact on postoperative MRI analysis and alter the quality of surgical landmarks in case of repeated surgeries. The choice of the closure technique depends on the flow intensity and on the leak location, as proposed by Conger et al. (34). In case of minimal diaphragm oozing, a simple treatment with a collagen sponge can be chosen (64, 66, 69). Collagen sponge has the great advantage to be easy to use. However, if the CSF is underestimated, this closure technique will be insufficient and a postoperative CSF leak will occur, requiring a second surgery. We therefore recommend spending a lot of time analyzing the intensity of CSF flow and adopting a safer alternative reconstruction strategy when in doubt. Focal low-flow intraoperative leak requires a stronger closure strategy: although an adequate epidural graft is more complex to perform, this strategy is well suited in this case, with a high rate of watertight closure without intrasellar packing. An epidural duraplasty can be performed, using a dural substitute, such as mucosa from a middle turbinate or fascia lata, held in place by a rigid buttress, as previously reported under different names, such as the gasket seal technique (34, 52, 70). However, the surgeon should be aware that if part of this duraplasty is positioned inside the intradural intrasellar space, the strategy will not prevent a postoperative leak. In our experience, the graft will paradoxically have a “gutter effect” with an increased risk of postoperative leak. Of note, in rare cases of focal diaphragm defect with a distended diaphragm bulging into the intrasellar space, a watertight closure can be elegantly obtained by coagulating the edges of the transdiaphragmatic orifice with bipolar forceps. In case of high-flow leak, the objective is to achieve a strong and persistent watertight closure with an intrasellar packing. Some authors recommend to use intrasellar fat and glue (33, 57, 71–75) with low rates of postoperative leak. In our experience, we prefer to use a fascia and muscle graft to achieve a two-layer intrasellar repair: the fascia is introduced into the sella and applied superiorly to cover the entire defect, in order to recreate a new diaphragm; the muscle graft is then positioned in the intrasellar compartment. Special care must be taken in the design of the muscle graft to avoid significant mass effect with compression of the optic chiasma.

In patients with risk factors, more complex strategies should be decided. In case of sellar floor lysis, bone may be missing, which is why no solid epidural buttress can be performed. Solutions with a buttress positioned within the sphenoid sinus have been developed. Some authors proposed a fat buttress to pack the sphenoid sinus (33, 57, 76, 77). The disadvantage of this technique is that the counterpressure is not applied in a targeted manner against the sellar floor and the fibrous scarring may complicate the surgical approach if further intervention is required. For these reasons, we prefer to use a Foley balloon catheter inflated within the sphenoid sinus for a few days, as previously described (78). Thus, the transient buttress is directly applied against the sella floor until the intrasellar muscle graft can no longer move. In more complex patients with giant tumors and large dural defects, a multilayer strategy with mucosal flap is usually recommended. Vascularized grafts have been a surgical revolution in preventing postoperative CSF leak, especially in patients with risk factors (24, 25, 48, 49, 79, 80). Different types of vascularized grafts have been described: unilateral or bilateral mucosal flaps originating from nasal septum or inferior turbinate; pericranium or temporal fascia flaps (45, 46, 48, 53, 54, 56, 81, 82). The choice of using vascularized flaps involves more complex skull base reconstruction techniques with multilayered closure and should balance the risk of postoperative CSF leak versus the morbidity of the flap itself (30, 83–85).

All 29 patients with postoperative CSF leak were reoperated with a dedicated upgraded closure strategy. Our management has evolved over the time. Indeed, the use of external lumbar drainage in these patients is associated with a higher risk of complication (86, 87). Of note, as previously reported, there is a lack of statistically significant improvement between patients with lumbar drains and patients with no lumbar drains and graded reconstruction strategies without lumbar drainage have been proposed (39, 88–90). Today, external lumbar drainage has been abandoned for the overwhelming majority of our patients treated with simple transsellar approach.

Postoperative CSF leak occurred in 18 patients with no intraoperative CSF leak identified. In 2 patients, the postoperative leak was obviously caused by inappropriate postoperative nose blowing. In 16 patients, the complication may have been due to intraoperative surgical misinterpretation, related to unnoticed diaphragm oozing.

Despite of our closure strategy, surgery failed in 11 patients with a well-identified intraoperative CSF leak, leading to an overall closure failure rate of 3.4%. Interestingly, most of the patients were treated before 2018, suggesting that this type of complication may decrease with endoscopic surgical experience, as previously proposed (91). Firstly, lack of experience with endoscopic techniques may have led to poor material selection or poor graft design, as previously reported (92–94). Secondly, the majority of these patients had at least one risk factor of closure failure. Thus, surgical failure was also explained by underestimated risk factors, such as severe obesity; BMI>40 is associated with increased intracranial pressure (44), which may affect the quality of closure (29). In patients with BMI>40 and focal sellar floor lysis, we now recommend the use of an additional Foley-catheter combined with repeated lumbar punctures, even if no CSF occurred during surgery. Closure failure was finally caused by poor operative conditions reducing the quality of vision and impacting on the analysis of CSF flow.

Our rate of postoperative meningitis was 0.8%, with a majority of aseptic meningitis. This result is in accordance with previous studies from pituitary centers of excellence (51, 95). In patients with no identified bacteria, antibiotic treatment was indicated on the basis of combined neurological symptoms (fever, meningeal syndrome) and analysis of repeated CSF examinations (cells increase, low glucose, high protein). These patients may have been overtreated. Nevertheless, all symptoms improved immediately after treatment was introduced, suggesting that a small bacterial inoculum was still present. In patients with septic meningitis, drug-sensitive Gram-positive positive organisms were predominant. All patients were treated by antibiotic therapy, with favorable outcome. Our results confirmed the data published by Jin et al. in a large retrospective study of 3242 patients (51). Interestingly, the vast majority of our patients (79.2%) with meningitis had a well-identified intraoperative leak, whereas a postoperative CSF leak was observed in only 20.8% of these patients. This result may suggest that the duration of CSF leak is not a strong predictor of postoperative meningitis. More data are needed to confirm this finding.

The main strength of this major study was the consecutive inclusion of all cases of adenoma patients treated with an endoscopic endonasal transsellar approach, reaching the substantial number of 3015 patients. Conversely, the inclusion of patients operated on by a single surgical team may be considered as a limitation. However, as mainly reported, the high expertise of surgical centers is essential (92, 96, 97). In this series, all patients were operated on by two experienced neurosurgeons treating >200 adenomas per year, after surgical indication was validated in a multidisciplinary meeting. If the expertise is not guaranteed, it can be hypothesized that outcome would become less favorable. Thus, at the beginning of the learning curve or if there is any doubt about a high-flow leak, safety should be paramount and intrasellar packing should be chosen. Due to the length of the study, our list of predictive factors is not exhaustive, which is a limitation of the present work. The main objective was to propose a graded closure strategy with excellent efficacy, based on our significant experience. Further studies with stronger decision-making paradigms, including more predictive factors, are needed in the future.

In conclusion, in pituitary surgery, the closure step should not be underestimated. By using a rigorous strategy, the postoperative leak rate can be reduced to 1% of patients. This crucial step must be planned before surgery and gently prepared during the surgical approach.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

In accordance with the French legislation, patient consent was not needed for this retrospective non-interventional study evaluating a routine care. An agreement was obtained after ethical acceptance of the study by the General Register of the Assistance Publique des Hôpitaux de Paris - Sorbonne University (registered under “CLOSPIT” Study, N°: 202211210146).

BB, AV, and SG contributed to conception and design of the study. BB, AJ, and AV organized the database. BB and AJ performed the statistical analysis. AV wrote the first draft of the manuscript. BB, AV, AJ, GR, and SG wrote sections of the manuscript. BB, AJ, GR, and SG contributed to interpretation of data for work. All authors contributed to manuscript revision, read, and approved the submitted version