Effects of cruciate embedding fascia-bone flap technique on grade II–III cerebral spinal fluid leak in endoscopic endonasal surgery

Under approval of the institutional ethics committee, we retrospectively collected data from consecutive patients diagnosed with sellar region lesions who underwent EES by the same group of senior surgeons in our institution from May 2015 to May 2020. All operations were in accordance with the general surgical indications for sellar regional lesions: neurological impairment due to lesional compression, progressive lesion growth, functioning adenoma resistant to drug therapy, and increased intracranial pressure induced by lesional mass effects. All patients were confirmed to have an intraoperative grade II or III CSF leak corresponding to Esposito’s Grade [20] (Fig. 1a), and the formation mechanism of their leaks was identical: skull base bone-dura defects.

In the reconstruction phase of the operation, patients who received CEFB were included as the CEFB group; patients with insufficient bone harvest for CEFB (this anatomical variation, encountered in the nasal phase of the surgical approach, has not been considered an influencing factor of postoperative CSF leak after EES [21, 22]) were treated with PNSF and served as the PNSF group. Data regarding age, sex, body mass index (BMI), diabetes, lesion volume, smoking, hypertension, surgical approach and pathology were analysed to verify the consistency of the baseline between the two groups.

Patients who simultaneously had grade III intraoperative CSF leak and \(\ge\) 3 high-risk factors for postoperative CSF leak (diabetes, age > 65, hydrocephalus, body mass index (BMI) > 30, adoption of extended approach [23]) received a combination of CEFB and PNSF to strengthen the reconstruction. As a supplemental case series, we also recorded and described the outcomes of these patients.

We excluded recurrent tumour patients due to their compromised tissue healing ability, as local cicatrization would impair the reconstruction [24], causing confounding fluctuations in the data. All patients enrolled were followed-up for at least 6 months, otherwise, they were excluded.

The two-surgeon, four-hand technique was used in all operations with a high-definition endoscope (Karl Storz, Tuttlingen, Germany). The standard endoscopic endonasal approach (EEA) was mainly adopted in cases of pituitary adenoma and Rathke cleft cyst, and the extended EEA (EEEA) was mainly used in operations on craniopharyngioma, meningioma, and invasive or giant pituitary adenoma. The multilayer closure strategy was applied in all operations.

The inlay parts in both groups were identical. Absorbable acellular dermal matrix (ADM) (Heal-all, ZH-BIO Inc., Yantai, China) was used as the first subdural inlay. It was placed to cover the margin of the residual diaphragma sellae (Fig. 1b). In cases of total loss of diaphragma sellae, the ADM was placed at the original site of the diaphragma sellae and meticulously leaned on structures such as the optic chiasm and hypophyseal stalk or tucked under the internal carotid artery (ICA). Then, an autologous fat graft harvested from the patient’s thigh was placed inside the sellar space or corresponding subdural space under the ADM to sustain the first layer (Fig. 1c).

Interrupted partial dural suturing was applied in all operations. The suturing was made with 5–0 nylon by the sliding-lock-knot technique as described previously [8]. The dural incision was preferably made in an “H”, “T”, “Y”, or inverted “U” shape to facilitate the multipoint suturing of 3–5 stitches (Fig. 1d).

In the CEFB group, the fascia lata onlay was implanted according to the following protocols. First, the fascia lata was tailored into a rectangle shape and placed over the dural defect either longitudinally (sagittal direction) or transversely (coronary direction). We preferred the longitudinal onlay in most cases for better convenience. Typically, in extended EEA, the frontal edge of the fascia lata reaches the planum sphenoidale, and the rear edge reaches the upper clivus, extending at least 10 mm beyond the bone defect edges. The lateral edges of the fascia should reach the lateral bone defect margin or slightly exceed it by no more than 2 mm (Fig. 1e).

In the PNSF group, the onlay of the fascia lata was performed in a conventional way by covering the bone defect with a redundancy of at least 10 mm.

In the CEFB group, the bone flap graft was harvested from the nasal septum, vomer, anterior wall of the sphenoid sinus, or bone septum inside the sphenoid sinus. The bone flap was countersunk and embedded under the edges of the bone defect as a rigid buttress to press the fascia lata onto the dural defect. It should be noted that the bone flap embedding was different from the circumferential wedging used in gasket-seal. In the CEFB procedure, the flap was wedged on 2 sides either longitudinally or transversely and placed vertically, crossing upon the fascia lata to finally form a cruciate embedding fascia-bone flap complex. The bone flap was trimmed and wedged with caution to avoid neurovascular injury. Typically, we transversely embed (vertical to the longitudinally placed fascia underneath) the bone flap under the lateral defect edges with an overlap exceeding 1–2 mm (avoiding compression on the optic nerve or ICA). Thus, the fascia could stretch out through the frontal and rear gaps between the bone flap and defect edge and be paved smoothly onto the skull base (Fig. 1f).

In the CEFB group, surplus fascia and fat graft were used to strengthen the whole construct (Fig. 1g). The out-stretching fascia was fixed with oxidized regenerated cellulose (Surgicel, Ethicon Inc., America) for apposition against the bone surface. The sphenoid sinus was filled with absorbable lactide caprolactone copolyesters (Nasopore, Polyganics, Groningen, The Netherlands) to sustain the fascia (Fig. 1h), and similarly, the nasal cavity was filled to reposition the mucosa back to the septum. No glue, balloon or iodoform gauze was used in the CEFB group (Fig. 1i).

In the PNSF group, a vascularized nasoseptal flap was harvested as described previously [14] to cover the fascia abundantly as the last layer. Surplus fat graft and iodoform gauze were inserted into the sphenoid sinus and nasal cavity to support PNSF coverage.

In both groups, the supine position was mandatory in the first postoperative 24 h. Then, the head of the bed was gradually elevated 20–30° per day unless a CSF leak was detected. The patient was allowed to get out of bed after no CSF leakage while in a persistent vertical sitting posture for 4 h. Lumbar drainage (LD) was not prophylactically used but was only performed when the patient had sustained CSF leak or a refractory intracranial infection. The nasal packing gauze in the PNSF group was removed 10 days after surgery.

Data regarding surgery duration, resection degree, grade/size of the CSF leak, intracranial infection, LD placement, LD duration, epistaxis, dysosmia, nasal discomforts, bed-stay time and hospitalization time were recorded for comparison. Nasal discomforts was introduced as a subjective symptomological item that was evacuated 10 weeks after surgery, including senses of dryness, pain, nasal obstruction, runny nose and nasal odour.

For continuous variables (age, BMI, lesion volume, leak size, surgical duration, etc.), Student’s t test was used for comparisons between the CEFB and PNSF groups. For categorical variables (sex, diabetes, smoking, hypertension, surgical approach, pathology, leak grade, CSF leak and infection, etc.), the Chi-squared test or Fisher’s exact test was used for comparison.

SPSS 25.0 was used for analyses. For the baseline data, a p value greater than 0.1 suggested that the difference in a variable between the two groups was uncertain. For the related indicators describing the effects of reconstruction, a p value less than 0.05 indicated the existence of a difference between groups.

Three patients were excluded from the study due to loss to follow-up (1 in CEFB and 2 in PNSF; total loss rate of follow-up 1.7%). None of them presented with CSF leak or meningitis in their last interview after surgery, but we then lost contact with them. The included patients were followed-up for at least 6 months. Finally, there were 110 patients enrolled in the CEFB group and 65 in the PNSF group.

For the preoperative baseline parameters, there were no significant differences between the 2 groups in age, sex, BMI, diabetes, smoking, hypertension, surgical approach or pathology of the lesions (p > 0.1) (Table 1). Despite the p > 0.1 result, due to the existence of type II errors, it cannot be simply asserted that the baselines of the two groups were completely consistent. However, the baselines of the two groups could be considered similar to a certain degree.

The grade II/III constituent ratio and size of intraoperative CSF leak (69/41 vs. 33/32; 18.55 ± 2.41 vs. 18.38 ± 3.09 mm², p > 0.1), incidence of postoperative CSF leak (2.7% vs. 3.1%, p > 0.1), intracranial infection (4.5% vs. 3.1%, p > 0.1) and LD placement (5.5% vs. 6.2%, p > 0.1) were similar between the CEFB and PNSF groups.

The CEFB group had a shorter LD duration (6.67 ± 2.16 vs. 10.50 ± 2.38 days, p < 0.01), bed-stay time (5.74 ± 1.58 vs. 8.83 ± 3.78 days, p < 0.01) and hospitalization time (10.49 ± 5.51 vs. 13.58 ± 5.50 days, p < 0.01) than the PNSF group. The CEFB was also associated with a lower occurrence of epistaxis (0 vs. 6.2%, p < 0.05) and nasal discomforts (0 vs. 7.7%, p < 0.01). However, the surgery duration of the CEFB group was longer than that of the PNSF group (2.62 ± 0.56 vs. 2.26 ± 0.62 h, p < 0.05), which was mainly ascribed to the meticulous trimming and embedding of the bone flap graft. All comparisons above are listed in Table 2.

During the process of bone flap embedding, a small amount of epidural haemorrhages often occurred but could be easily controlled by a gelatine sponge. No CEFB-related neurovascular injury occurred. One patient in the CEFB group and another in the PNSF group who experienced sustained postoperative CSF leak were cured by reoperation. The one in the CEFB group was caused by local curling of the fascia lata due to excessive compression of the bone flap. Another in the PNSF group was due to loosening of the PNSF. Other cases of postoperative CSF leak were healed by LD placement and postural confinement.

No fracture, dislocation or detachment of the CEFB construct was observed during the scheduled nasal debridement under endoscopy 3 weeks after surgery (Fig. 2a). The bone flap and fascia were found to be in place and firmly attached to the defect. Osteal structure reconstruction was assessed by pre- and postoperative CT scans (Fig. 2b–e). The long-term stability of the CEFB construct was examined by CT scan 3–6 months after surgery (Fig. 2f, g).

Data on subdivided characteristics are summarized in Table 3. In both grade II and grade III leak comparisons, the CEFB group showed similar effects as the PNSF group on the incidence of postoperative CSF leak (grade II leak: 2.9% vs. 3.0%; grade III leak: 2.4% vs. 3.1%), infection (grade II leak: 4.3% vs. 3.0%; grade III leak: 4.9% vs. 3.1%), and LD placement (grade II leak: 4.3% vs. 6.1%; grade III leak: 7.3% vs. 6.3%).

Among patients with preoperative hydrocephalus, the CEFB group yielded an outcome of 0 postoperative CSF leak out of 5 patients, while 1 leak occurred among 3 patients in the PNSF group.

There were 23 patients who had grade III intraoperative CSF leaks and multiple high-risk factors (as described in the Patients and group section) received the combined use of CEFB and PNSF. Ten patients had preoperative hydrocephalus. There was no postoperative CSF leak or LD placement, but 2 patients experienced postoperative infection.

The inlay materials contribute to reducing the leak size and CSF flow, eliminating dead space, buffering the impact of CSF pulsation, and alleviating CSF pooling or soaking [25, 26]. The key point of the CEFB inlay procedure is to finely adjust the volume of the fat graft to obtain optimal subdural tension fitting the buttress pressure of the wedged bone graft, generating appropriate tightness of the attachment between the fascia and dura. Deep suturing and knotting are no longer major issues in EES today, but literally “watertight” suturing is still technically challenging due to dural dehydration, fragility, or electrocauterization [8, 27]. Partial suturing is not sufficient to seal CSF leaks, but we deem it to still have the following benefits: 1. The dural defect is reduced and confined under the centre of the rigid buttress. 2. The dural interface is made available for onlay fascia attachment, avoiding direct contact between the fascia and inlay grafts. 3. The ADM and fat are anchored in place. 4. Intrasellar tension and compactness are increased [28].

The CEFB procedure shares the same concept of rigid reconstruction with the gasket seal [6], but with significant modifications. As the name addressed, gasket-seal focuses on the watertight closure of defect. The shape of the circumferentially wedged material used in gasket-seal must be highly matched to the shape of the bone defect [6, 29]. Artificial material is frequently used [6, 19, 30] because the autologous bone graft does not always perfectly fit. The core of the CEFB technique is the buttress pressure firmly holding the fascia lata onto the dura to promote tight attachment and mutual adhesion. First, the bone flap in the CEFB construct is embedded at only two sides of the bone defect instead of being fully circumferentially wedged. The size of the bone flap is qualified by adequate length on only one axis, which promotes the usage of autologous bone, and brings lower costs and risks of infection or rejection. In our study, 67.2% (133/198) of all patients achieved satisfactory autologous bone harvest. Second, the two-side embedding of the bone flap allows the fascia lata to stretch out through the gaps on the non-wedged sides and to be paved smoothly onto skull base. However, when gasket-seal are used, the fascia lata forms a “cauliflower leaf” shape [6, 29] due to the tight depression in the centre [6]. The tilted or curled edge of the fascia makes it difficult to provide smooth attachment. Third, partial dural suturing ensures correct epidural embedding of the bone flap, avoiding accidental misplacement into the subdural space. The firm contact between the sutured dura and fascia also assists adhesion formation. Finally, the gasket-seal is not ideal in cases of defects traversing two geometric planes [6]. In our EEEA cases with sufficient bone graft harvest, we wedged two separated bone flaps at different defect planes respectively (Fig. 3a). The fascia lata could then be supported evenly and held in place on angled planes.

Forces including brain gravity, CSF pulsation, and intracranial pressure are great concerns in skull base repair. These downwards forces are prone to compromising the reconstruction [28]. Hence, iodoform gauze packing, lumbar drainage, intranasal balloon, etc., are used as countermeasures [6, 30, 31]. The integrity of the CEFB construct is sturdy enough to resist these forces due to the rigidity of the firmly wedged bone flap. No dislocation or fracture of the bone flap was observed in any of our cases, even without iodoform gauze packing or lumbar drainage. These downwards forces might even be conjecturally helpful in inducing more solid compression of the CEFB structure and then strengthening the watertightness and attachment. This mechanism possibly contributed to the avoidance of CSF leaks among preoperative hydrocephalus patients and shortened the bed-stay time in the CEFB group. Especially in patients with preoperative hydrocephalus, both CEFB alone and in combination with PNSF exhibited a good outcome, with 0 postoperative CSF leaks in 15 patients.

The PNSF was a milestone in the development of skull base reconstruction due to its fast healing and long-term security of closure [14]. However, the harvesting of the PNSF necessitates a long incision on the nasal septum to mobilize an extensive piece of mucosa, which is then migrated to cover the skull base. The exposed donor site needs 6–12 weeks to be re-epithelialized [32]. Nasal complications related to this extensive shift of mucosa are not rare [15‐17, 33]. The nasal packing of iodoform gauze or a balloon regularly used in association with PNSF also influences mucosa regeneration and the patients’ subjective experience due to intranasal pressure and stimulation [15]. Garcia-Navarro reported that the PNSF did not make a significant difference in their gasket-seal practice, which raised the question of PNSF necessity [6]. In our study, we obtained similar outcomes in terms of postoperative CSF leak and infection between the CEFB and PNSF groups, and nasal complications were even reduced in the CEFB group. The effectiveness of CEFB is also similar to that of other representative reports on postoperative CSF leak [6, 9, 18, 19] (Table 4). Our data suggest that with appropriate CEFB use and multilayer reconstruction, the PNSF may not be the sole mandatory option. The advantages and disadvantages of the PNSF should be individually weighed. We did not attempt to replace the PNSF with the CEFB. On the contrary, the CEFB technique is not exclusive but highly compatible with the PNSF. In 23 patients with grade III CSF leak and multiple high-risk factors, the combination of the 2 techniques achieved 0 postoperative CSF leaks. This implied that with the additional protection from the PNSF, the CEFB flap might provide reliable security in high-risk cases.

First, harvesting the bone flap is not always sufficient or available due to anatomical variations or tumour invasion, especially in cases of oversized defects extending laterally. Second, coverage of the fascia under the wedging edges of the defect is limited. Several measures could be taken to improve these limitations: 1. During the surgical approach, any potentially usable bone structures should be preserved by avoiding excessive grinding of the microdrill. 2. The bone graft could be tailored into narrow strips and then wedged at intervals into the defect (Fig. 3b). 3. The wedging sides of the defect could be carefully enlarged to expose more dural surface for fascia attaching. 4. The dural defect could be minimized by improving the dural incision design and suturing. Finally, in the worst-case scenario, when all measures turn out to be unassured, the PNSF remains a trustworthy last resort.

Our study was retrospective and had only a limited number of patients. The establishment of a randomized control trial is difficult. The current follow-up period of 6 months was also not sufficient to perform long-term evaluation. We have to claim that due to the diversity of clinical conditions, our group set, baseline control, and outcomes comparison might be underpowered owing to the presence of type II error. The statistical analysis in our study should be viewed and interpreted with caution and only for reference. The emphasis of this article is the descriptive data and experience drawn from our surgical practice. A more rigorous study design and accumulation of cases are required for further assessment of the CEFB technique.