Pituitary neuroendocrine tumors (PitNETs) are the second most common brain tumors, with an incidence of 3.9-7.4 cases per 100000 and a prevalence of 100 per 100000 in the general population [7]. The endoscopic trans-sphenoidal approach (eTSA) is currently the most advocated surgical method for PitNET due to its optimal panoramic visualization of the tumor and its surroundings. However, despite its many advantages, eTSA inherently possesses a risk of postoperative cerebrospinal fluid (CSF) leak through the skull base defect.

Over the past decades, numerous methods of skull base reconstruction have been developed, and a graded multilayered strategy is commonly employed. However, many of these methods appear overly aggressive [5,8,31], especially low-grade (grade 0-2) CSF leaks, often involving use of autologous fat grafts [29], fascia lata [16,21], and occasional peri-operative lumbar drainage (LD) [23,37].

Based on our institutional experience, we devised a straight-forward graded reconstruction method for PitNET cases with low-grade CSF leaks. In refining this approach, we specifically subcategorized grade 2 CSF leaks into two groups according to size of the diaphragmatic defect, as the conventional spectrum of grade 2 is too broad. In this study, we retrospectively reviewed the surgical outcomes of our reconstructive approach.

This study was conducted in accordance with the ethical standards set forth in the 1964 World Medical Association Declaration of Helsinki, as revised in 2008. The IRB of Seoul National University reviewed and approved the study protocol (IRB No. 2304-071-1422). The current study posed minimal risk to the subject and did not involve any identifiable personal data. To maintain the confidentiality of personal information, we anonymously processed all data collected from the individuals.

The development of a reliable postoperative skull base reconstruction technique has gained increasing importance as the use of eTSA continues to expand. Although several reconstruction algorithms for pituitary surgery have been reported, many appear complex, requiring multiple risk factors to be considered in decision-making, and often seem overly robust, particularly for low-grade CSF leaks. Numerous risk factors for postoperative CSF leaks following eTSA have been identified, including intraoperative CSF leaks [6], high BMI [10,15], Cushing’s disease [13] and repeated surgeries [25,26]. Many institutions incorporate these risk factors into their reconstruction algorithms to guide the choice of reconstruction method [2,4]. In contrast, our reconstruction algorithm is simple and intuitive, relying primarily on the intraoperative CSF leak grade [9] and the size of the diaphragmatic defect. Additional considerations included ICA exposure, a prior history of skull base surgery, or the presence of severe adhesion between the tumor and the diaphragm sellae, in which cases an NSF was applied.

The efficacy of NSF for skull base reconstruction has been well documented [39,40] and is frequently incorporated in highflow CSF leak (grade 3) reconstruction. Although our technique targets low-grade CSF leaks, we used the NSF primarily to protect the ICA when exposed during the removal of PitNET with cavernous sinus invasion. This step provided a physical barrier between the ICA and the nasal cavity to prevent wound breakdown and infection, thereby reducing the risk of potentially fatal complications such as carotid blowout [14,32]. Additionally, the NSF added an extra layer of reconstruction, contributing to a more stable skull base repair. As surgical techniques advance, enabling more aggressive exploration of the cavernous sinus, an area previously regarded as a “no man’s land”, there has been a marked increase in cases involving resection of the medial cavernous sinus wall, resulting in ICA exposure. Therefore, we consider ICA exposure an important factor when planning skull base reconstruction, even in low-grade CSF leak cases. NSF was also used in cases with a history of prior skull base surgery. Its primary role was to counter obstacles to healing caused by granulation tissue, which can hinder proper attachment of fibrin sealant patches. Moreover, it was used to seal potential undetected micro-injuries that may occur during tumor dissection from the diaphragm, particularly in the presence of adhesions from previous surgery, thereby promoting better healing.

For grade 2b cases, NSF was incorporated into rigid reconstruction to provide external coverage of HXA. This minimized the risk of infection by creating a physical barrier between HXA and the nasal corridor and supported mucosalization of the sphenoid sinus. We acknowledge that some grade 2b cases may not be suitable for rigid reconstruction with HXA due to thin or cracked mother bone or in institutions where HXA is not available. Although we advocate for rigid reconstruction in these patients, when HXA cannot be applied, we strongly recommend at least the use of NSF for reinforcement during reconstruction. Of note, although NSF is a safe and reliable reconstructive option, complications such as olfactory loss and septal perforation have been reported [20]. To avoid unnecessary complexity or excessive robustness in our reconstruction algorithm, NSF use was limited to cases where it was essential.

We subcategorized grade 2 intraoperative CSF leak into two groups by the size of diaphragmatic defect. This was done to create a more tailored and reproducible reconstruction algorithm in the wide spectrum of grade 2. An example of this can be seen in grade 2b where the size of the diaphragmatic defect was large (≥6 mm) without communication with the third ventricle. Although it should have been evaluated as grade 2, we found it necessary to treat the defect as a high-flow CSF leak requiring rigid reconstruction (Fig. 4). Therefore, by eliminating such confusions through subcategorization, we were able to stratify the risks accurately and tailor the reconstruction method specific to each case.

Numerous groups still use autologous fat in the sellar cavity when reconstructing low-grade CSF leaks to obliterate the sellar dead space [1,19,28,34]. Autologous fat is widely utilized due to the easy availability and relative safety. However, we did not use autologous fat in any of our cases because it can be associated with complications such as fat necrosis and lipoid meningitis [36]. Furthermore, harvesting fat grafts entails making an additional incision in the abdomen, which increases the overall surgery time and the recovery time for patients. We believe that obliterating the dead space in the sella is not essential, and the potential risk of complications can be avoided using our techniques. Instead, we preferred NSF over other membranous autologous grafts, such as fascia lata, in cases where an extra layer of reconstruction or protection was required. We believe that the well-vascularized nature of NSF could promote a better healing process than other avascular autologous grafts.

The indication for when to use LD in eTSA remains somewhat unclear, with some groups advocating its use to reduce peri-operative CSF leaks [23,24,37,41]. The general consensus appears to be to avoid the routine use of LD, as it poses a greater risk of complications compared to the benefit of lowering postoperative CSF leaks [30,38]. It should be reserved for selected reconstructions of high-flow CSF leaks [33]. Avoiding routine use of LD for low-grade CSF leaks can help shorten hospital stays and reduce overall healthcare costs.

There were two cases of meningitis, with an overall infection rate of 0.6%, which was comparable to previous studies reported to be between 1.7% and 1.8% [3,27,35]. Both cases of postoperative meningitis were diagnosed within 1 week of the surgery, each presenting with a consistent fever. Both cases were treated with antibiotics for 2 weeks and were discharged without further complications.

This study is limited by its retrospective and non-randomized design with outcomes based on a single surgeon’s experience. The absence of a comparison cohort prevents direct evaluation of the relative efficacy of the suggested reconstruction methods and limits the generalizability of the findings. Future studies are necessary to validate and refine the proposed strategy. Second, the study reviewed cases over a 5-year period (2017-2022). During this time, surgical techniques may have evolved as experience accumulated and feedback was incorporated. Third, although the mean follow-up duration was substantial at 2 years, issues that arise over longer periods may not have been identified. A 2-year period, however, is generally sufficient to detect postoperative CSF leaks or infections. Lastly, although specific reconstruction methods were established for each CSF leak grade, factors such as ICA exposure, prior skull base surgery, or the introduction of acellular dermal grafts created variations in reconstruction techniques within the same grade. These variations may have influenced the results. Even so, the methods tailored to each CSF leak grade were based on sound rationale, and we consistently adhered to our reconstruction protocol, which helps support the integrity of the findings.

The increasing adoption of eTSA emphasizes the need for reliable skull base reconstruction. We developed a simple, effective, and reproducible graded reconstruction method for low-grade CSF leaks in PitNET surgeries. Risk stratification was primarily based on intraoperative CSF leak grades and the size of the diaphragmatic defect. Furthermore, subcategorizing grade 2 leaks allowed for more precise risk assessment and facilitated tailored reconstruction strategies. Further evaluation with longer follow-up periods may enable refinement and optimization of this reconstruction algorithm.