Introduction

Pituitary Adenomas (PA) account for 10–25% of intracranial tumors [11, 21]. Large and giant PA are less common and surgical treatment of these tumors may be challenging. Definitions of large pituitary adenoma (lPA) are lacking consistency [9], but in most recently series they were classified as lPA with a maximum diameter in any plane greater than 3 cm and a tumor volume larger than 10ccm [26, 38]. Giant pituitary adenoma (gPA) are commonly defined as larger than 4 cm in diameter [7, 24, 37] and having a volume greater than 10ccm [35]. Transcranial and transsphenoidal microsurgical approaches, simultaneously or consecutively have been advocated while endoscopic transsphenoidal surgery has become the standard treatment in the majority of large and giant PA in recent years [7, 24, 26, 33,34,35, 38, 46]. Endoscopic endonasal transsphenoidal surgery (EETS) has been reported to result in higher tumor resection rates, improved visual outcome, a higher percentage of pituitary function preservation, less postoperative complications and decreased hospital stay [16, 34, 49]. However, surgery of lPA and gPAs remains challenging [24] and treatment often requires multiple approaches not only because of their particular size, invasiveness and extrasellar extension but also because of encasement of neurovascular structures and multi-compartment growth. The disadvantage of endoscopic surgery is the lack of binocular stereoscopic vision, reduced depth perception, image distortion, reduced hand eye coordination and difficulties in estimating the correct size of objects. Three-dimensional (3D) endoscopy (3D-E) may have, at least from a technical point of view, the potential to overcome the disadvantages of two-dimensional (2D) endoscopy (2D-E). A few series show, that three-dimensional endoscopic pituitary surgery has similar results compared to 2D endoscopy [3, 22, 25, 51]. However, no data are published comparing the use of 3D-E and 2D-E for surgery of large and giant pituitary adenomas. Therefore, we aimed to analyze the results of our consecutive series of patients with lPA and gPA with regard to tumor resection, perioperative complications, mortality and morbidity as well as ophthalmological and endocrine outcome.

Material and methods

Study design

This is a retrospective, single center observational surgical case series, performed in a tertiary referral center including a consecutive series of patients with endoscopic endonasal transsphenoidal surgery for treatment of PA. The study was approved by the local ethics committee (Reference number 22748/2016/32). Data of all patients who underwent EETS were retrieved from a prospectively maintained database of sellar and parasellar tumors at the authors’ institution. Between November 2008 and January 2023, 480 patients with lesions of the anterior and central skull base underwent EETS. Patients with large and giant PA were included, while patients with other pathologies (craniopharyngiomas, chordomas, meningiomas, cysts etc.) were excluded from further analysis.

Patient’s characteristics (age, gender, endocrinological and ophthalmologic status) were continuously entered into the database. In addition, tumor variables were added (histology, tumor size, tumor volume, tumor extension, shape and growth direction, hemorrhagic apoplexy). The extension of the tumor into the intra-, para- and suprasellar space was assessed separately. For definition of parasellar growth, the Knosp criteria were applied [29]. The maximum diameter on preoperative MRI (either T1- or T2-weighted imaging; coronal, sagittal, or axial slices) was determined for each case. Preoperative tumor volume was analyzed using the Brainlab Elements software (Brainlab AG, Munich, Germany) based on thin sliced preoperative MRI data sets using the manual segmentation tool function by 2 authors (RG and DL).

Large pituitary adenomas (lPA) were defined as ≥ 3 cm and < 4 cm in at least one dimension and ≥ 10ccm in volumetric analysis. Giant pituitary adenomas (gPA) were classified as larger than 4 cm in diameter and more than 10ccm in volume. The median follow-up was 4 years.

Perioperative complications, such as postoperative hemorrhage (intracranial, nasal), cerebrospinal fluid (CSF) leak with the necessity for lumbar drainage, revision surgery or ventriculoperitoneal shunt as well as meningeal infection or pneumocephalus were analyzed.

Extent of resection was evaluated by comparing pre- vs. postoperative (8 to12 weeks) contrast-enhanced MR-imaging. Complete resection was defined as no residual tumor visible on postoperative MRI. If remaining tumor was detected on postoperative MRI, it was defined as partial resection without further subdivision.

Endocrinologic assessment

All patients had standardized preoperative endocrinological testing and re-testing 8 to 12 weeks after surgery. Any new deficit of any element of the pituitary axis compared to preoperative tests was considered as new deficit. Endocrinological remission was defined in accordance with the most recent consensus criteria [4, 18].

Ophthalmologic assessment

All patients underwent pre- and postoperative ophthalmologic investigation and change in visual acuity and visual fields was assessed.

Surgical treatment

All patients were operated by an interdisciplinary team of a neurosurgeon (R.G.) and otorhinolaryngologist (G.K.). The patients were operated using one of the 3 different endoscopy systems, 2D-HD, 3D-SD or 3D-HD. The visualization systems used during the EETS have been used in accordance with their availability, but not randomly selected. During the first years, the 2D-HD endoscopy system (Karl Storz, Germany) was used. Subsequently, 3D-SD (VisionSense vsii) and 3D-HD (VisionSense vsiii) (VisionSense, New York) became available, which were successively utilized. EETS was performed without procedural modifications over the entire time period. Surgery was performed using image guidance (Brainlab AG, Munich, Germany). All patients were observed in the ICU overnight and referred to the peripheral ward 1 day after surgery, in the absence of postoperative complications.

Statistical analysis

Statistical analysis was performed using SPSS software version 25.0 (Chicago, USA). Non-parametric statistical tests (Kolmogorov–Smirnov test and Kruskal–Wallis) were used to compare the two categories and their variables. Chi-square (X2) test and Wilcoxon test were applied to compare continuous variables between the different EETS-visualization technique groups. For the conducted analysis, p values less than 0.05 were considered to be statistically significant (α = 0.05).

Results

62 patients underwent endoscopic EETS for treatment of lPA or gPA. No patient died within 30 days of surgery. The patients’ age ranged from 23 to 88 years (median 57), 16 patients were female (25.8%) and 46 male (74.2%). In general, patient and tumor characteristics were found to be normally distributed (age, sex and tumor volume, growth direction and recurrence rate) although a ratio of 2.88:1 was detected for the two different endoscopic visualization groups (3D-E:2D-E, Kolmogorov–Smirnov test). Altogether, there was no significant difference of the patient´s preoperative characteristics such as age, sex, number of hormone secreting tumors and inactive adenomas, ophthalmologic and endocrine status. For tumor variables (largest diameter, volume, growth direction, tumor shape and hemorrhagic apoplexy) no significant differences were found (X2 test; Table 1).

Table 1 Patients and tumor baseline characteristics
Full size table

43 patients underwent EETS for treatment of an lPA (69.4%) and 19 patients (30.6%) for a gPA. The rate of patients with lPA and gPA was not statistically significant between groups. 46 patients (74.2%) underwent surgical resection using 3D-E and 16 patients (25.8%) using 2D-E. Of the 43 patients with lPA, 32 patients underwent EETS using 3D-E and 2D-E was used in 11 cases. Of the 19 patients, who suffered from a gPA, 14 were operated with 3D-E and 5 with 2D-E (Table 1).

Although neither the tumor volume nor the largest diameter was significantly different between the 3D-E and 2D-E group it is worth mentioning that in the 3D-E group less patients had a round-shaped PA and more patients had a multilobular tumor shape compared to the 2D-E group (Table 1). 5 patients underwent an extended approach during EETS for tumor resection in the 3D-E group (Table 2). However, extent of tumor resection did not significantly differ between the 3D-E and 2D-E visualization technique. EETS using 3D-E resulted in complete resection in 20 patients (43.5%) and partial resection in 26 of 46 patients (56.5%) compared to 7 patients (43.8%) with complete resection and 9 patients (56.3%) with partial resection in the 2D-E group, respectively (X2 test: p = 0.985, Table 3).

Table 2 Surgical specifications
Full size table
Table 3 Outcome
Full size table

The overall rate of preoperative ophthalmological deficits (visual acuity and visual field) was high in the whole cohort. In the 3D-E group, 32 of 46 patients (69.6%) had reduced visual acuity and 37 had deficits in visual field (80.4%). In the 2D-E group, the number of patients with ophthalmological deficits was even higher (visual acuity deficit was detected in 14 of 16 patients [87.5%] and deficit of visual field in 13 of 16 cases [81.3%]; Table 1). During follow-up, the postoperative ophthalmological assessment revealed that 30 of overall 46 patients (65.2%) had visual acuity improvement, 21 of 32 patients in the 3D-E group (65.7%) and nine of 14 patients in the 2D-E group (64.3%). Overall improvement of visual field was achieved in 31 of 50 cases (62.0%). 22 of 37 patients (accordingly 59.4%) presented with a better visual field in the 3D-E group compared to 9 of 13 patients (69.2%) in the 2D-E group (Tables 1 and 4). In 13 of all 62 patients (48.4%) postoperative visual acuity remained stable (9 [19.6%] in the 3D-E group and 4 [25.0%] in the 2D-E group), regardless of whether they had a visual deficit preoperatively or not. In 18 of all 62 patients (29.0%) the visual field remained stable postoperatively (15 [32.6%] in the 3D-E group and 3 [18.8%] in the 2D-E group), regardless of whether they had a preoperative deficit or not (Tables 1 and 4). Worsening of visual acuity occurred in 6 patients overall, all of which were in the 3D-E group (6/46 [13.0%]). 1 patient, who was also in the 3D-E group, suffered from postoperative deterioration of the visual field. However, there was no statistical significance for both complications compared to the 2D-E group (p = 0.376 and p = 0.777).

Table 4 Postoperative complications
Full size table

The overall number of perioperative complications was statistically significantly higher in the group of patients who underwent EETS using 3D-E (16 out of 46 patients) compared to one patient out of 16 in the 2D-E group (p < 0.05, X2 test). However, when all types of complications were assessed separately (postoperative hemorrhage, CSF leak, meningitis and deterioration of visual acuity and visual field) there was no statistically significant difference between both visualization groups (Table 4). The complication rate was not depended on tumor volume, tumor extension, intratumoral hemorrhage, endocrinological status or previous surgical treatment.

Intraoperative CSF leak occurred in 21 of 46 patients in 3D-E group (45.7%) and in 6 of 16 patients in 2D-E group (37.5%), without statistical difference (p = 0.571, X2 test). However, one of the most frequent complications was postoperative CSF leak, which was encountered in 8 (17.4%) and 1 (6.25%) patients in the 3D-E and 2D-E group, respectively. Eight patients (12.9%) were treated with lumbar drainage and all 9 patients (14.5%) underwent revision surgery (p = 0.276 and p = 0.357; X2 test). Postoperative epistaxis occurred in 5 out of 46 patients in the 3D-E group (10.9%). Meningeal infection was suspected or proven after lumbar puncture and culturing CSF in 4 of 62 patients (6.45%, Table 4).

New pituitary anterior lobe dysfunction was observed in 30 of 62 patients (48.4%, 8 patients [50.0%] in the 2D-E group and 22 patients [47.8%] in the 3D-E group). A new deficit of the posterior lobe was detected in 22.6% (14/62).

Discussion

Endonasal transsphenoidal endoscopic surgery is considered as state-of-the-art treatment for patients with pituitary adenomas [5, 43]. This holds especially true for large and giant PA with extrasellar extension and multilobular growth and/ or encasement of critical neurovascular structures [12, 20, 26, 31, 33,34,35, 46, 56]. Improvement in visualization technique and image resolution with high definition (HD) or 4 K displays [10, 48], the understanding of tumor growth patterns [55] and the detailed anatomical knowledge of critical neurovascular structures with improved outcome of patients [15, 52, 53] fostered a broader application of endoscopy over the recent years in patients with large and giant PA.

However, the lack of stereoscopic vision remains a major disadvantage of 2D endoscopy. Thus, 3D endoscopy, which enables stereoscopic vision with improved depth perception during endoscopic surgery, may help to overcome the limitations of 2D endoscopy. 3D-E may allow for improved dexterity and safety during surgical resection of large and giant PA while working behind and around neurovascular structures and identifying tissue planes for dissection. This can aid in recognizing layers of dissection and understanding neurovascular relationships especially in the suprasellar space [41, 54]. The first application of a rigid 3D endoscope in EETS was described in 2009 [51]. Over the recent years, 3D-E underwent further refinement and the development of endoscopes with smaller diameters and improvement in image quality with HD resolution resulted in 3D-E implementation in a few pituitary centers, which has been described in a limited number of reports [1,2,3, 13, 22, 27, 39, 41, 44, 47]. A few authors compared surgical results of 3D-E to 2D-E for primary or recurrent PA. They found similar extent of resection and complication rates compared to 2D-E and no difference for perioperative (estimated blood loss, operative time) and postoperative factors (length of stay, complications, and readmission rate) [3, 22, 27]. Barkhoudarian et al. found a significantly shorter total operative time, which was needed for adenoma resection in the 3D-E group independently of the surgical experience (residents or fellows). In particular, residents, who were inexperienced with endoscopy, were quicker to learn how to manipulate the 3D endoscope, especially for maneuvers within the sphenoid sinus [3]. However, a broad acceptance of 3D-E in EETS is missing. This may be due to periprocedural strain and dizziness, general discomfort, the need to wear polarization glasses or reduced color fidelity, although the learning curve for implementation of 3D-E was rather flat [14].

Extend of resection

To the best of our knowledge, no data are published for patients with large and giant PA who underwent EETS using 3D-E so far. We therefore analyzed a consecutive series of patients with lPA and gPA, who underwent EETS. All operations were performed by the same interdisciplinary surgical team, which excludes a surgeon-related bias. A recent definition of lPA was applied [9, 23]. Latest studies suggest a volume of ≥ 10ccm as a definition of giant pituitary adenomas, rather than the conventional definition of a tumor diameter larger than 4 cm. Thus, we analyzed large and giant PA together and the total mean tumor volume in this series was 18.1ccm, which therefore compares well with other studies, eg. Cusimano et al. (mean preoperative tumor volume was 19.95 ± 15.69ccm) [9]. The maximum tumor diameter in our patients was 6.4 cm, which was also comparable to the other studies [40, 56]. Other authors, namely Cappabianca et al. pointed out, that “size does not matter”; instead, attention should mostly be paid to the pattern of intracranial growth [5]. Moreover, hemorrhagic component, posterior extension and sphenoid sinus invasion were found to be significant predictors for the extent of resection [26]. In this series of patients with large and giant PA who were treated with 3D-E and 2D-E, complete resection was achieved in almost half of the cases (43.5% complete resection and 56.5% partial resection in 3D-E group compared to 43.8% complete and 56.3% partial resection in the 2D-E group). Therefore, no statistically significant difference was found. Our resection rate compares well with other series [20, 35, 40, 56].

Juraschka et al. describe a complete resection in 24.2%, near-total (≥ 90%) 16.7%, subtotal (70%–89.9%) in 36.4%, and partial (< 70%) in 22.7% after EETS for large and giant PA [26]. Cusimano et al. reported a 20.7% GTR rate and 90.6% average resection rate using a binostril EETS [9]. Gondim et al. recently published a series of patients with gPA (> 4 cm) with total removal of the tumor occurred in 38%, near-total removal in 18%, and partial removal in 44% [20]. In a larger series of 239 patients with gPA Chen et al. reported gross-total resection was achieved in 19.25%, near total in 23.43%, subtotal in 28.45%, and partial in 28.87% patients [7]. Yano et al. found a near-total resection of gPA in 16 of 34 (47.1%) cases. Near total resection was achieved significantly more often in anterior extension types and round tumor in superior extension types compared with multiple extension types [56]. Koutourousiou et al. achieved near-total resection (> 90%) in 66.7% in 2013 [31]. In a series of large and giant PA Fallah et al. achieved gross total resection in 82.5%, near-total resection in 12.5%, and subtotal resection 5% [12]. Komotar et al. conducted a systematic review of endoscopic transsphenoidal surgery for giant (≥ 4 cm) pituitary adenomas, and described GTR of 47.5% [30]. Thus, there is a remarkable discrepancy in the reported resection rates of large and giant pituitary adenomas depending on the postoperative definition of residual tumor.

Koutourousiou al. found the “true limitations of endoscopic endonasal surgery” are tumors with a multilobular configuration and extensions beyond the lateral wall of the cavernous sinus [31]. Moreover, the lateral extension of PA into the middle fossa is also a limitation of the endoscopic approach and can be a reason for incomplete resection [34]. In this series, 63% of all patients were found to have a parasellar tumor extension and 93.8% (3D-E group) and 88.7% (2D-E group) had a cavernous sinus invasion greater than Knosp 2. Micko et al. published a series including gPA where overall gross-total resection rates were 64% in round gPAs, 46% in dumbbell-shaped gPAs, and 8% in multilobular gPA [35]. They concluded that EETS is “a primary treatment modality to relieve mass effect in GPAs and extent of resection are dependent on gPA morphology” [35]. In the 3D-E group, 30.6% of patients had a multilobular tumor extension compared to 25% in the 2D-E group. 38.7% had a dumbbell shape tumor in the 3D-E group and 37.5% 2D-E group, which may explain the relatively high number of residual tumor, as it was shown by Micko et.al, that the neck-to-dome area ratio was of prognostic value for prediction of intraoperative tumor non-descent in dumbbell shaped adenoma [36].

Ophthalmological outcome

Our postoperative data showed that 30 of 46 patients (65.2%) had visual acuity improvement. In 31 of 50 cases (62.0%) visual field was improved after EETS, which is comparable to other series where improvements are reported to be between 60–90% [9, 12, 19, 20, 31, 42, 45, 56].

Perioperative complications

“Giant adenomas are not only difficult to resect but also have a greater risk of complications “ as constituted by Iglesias et al. [24]. In 2017, Nishioka et al. presented a surgical series of 128 giant nonfunctioning adenomas. Permanent surgical complications developed in 28 patients (22.0%) [40]. In our study the overall complication rate was 27.4% and therefore was comparable to the series from Nishioka et.al. One of the most frequent complications was postoperative CSF leak, which was encountered in 8 (17.4%) and 1 (6.25%) patients for 3D-E and 2D-E, respectively (overall in 14.5%). They were treated with lumbar drainage and underwent revision surgery, but without statistically significant difference between the 3D-E and 2D-E groups (X2 test: p = 0.276 and p = 0.357). The rate of postoperative CSF leaks in this cohort is in line with published reports, which postulated an incidence of 0–16% [6, 28, 32, 50, 57,58,59].

In previous studies, new anterior pituitary insufficiency following EETS of gPA was detected in 16–36% of cases [8, 12, 20, 31]. In our series, new pituitary anterior lobe dysfunction was observed in 30 of our 62 patients (48.4%). This relatively high number might be due to our relatively short follow-up period and the application of the latest diagnostic criteria.

Limitation of the study

A number of limitations have to be addressed. It is a retrospective single-center case series with a limited and an unbalanced number of patients per group (n = 16 in the 2D-E and n = 46 in the 3D-E group, respectively), with a ratio of 2.88:1. The majority of patients operated with 2D-E were treated at the beginning of the series and the application of the 3D-E and 2D-E systems varied later on according to their availability. We are well aware that each surgery increases the experience of the surgeon, especially in this long period of more than 14 years. However, the resection rate of the 3D and 2D endoscopy group are comparable. This may be due to the fact that even before starting to operate in an interdisciplinary team, both surgeons were trained over many years and had a large experience with regard to operative skills and also scientific evaluation of new technical developments during surgery [17]. In addition, we did not perform a side-by-side analysis during certain steps of surgery, which requires an increase of technical efforts during surgery and prolongs operation time.

Conclusion

In this series of patients with large and giant PA, the application of different 3D und 2D visualization systems during surgery did not show a significant difference for resection rate or for postoperative outcome. Our results show, that 3D-E is feasible for resection of large and giant PA.