Complications in Endoscopic Endonasal Pituitary Adenoma Surgery: An Institution Experience in 310 Patients

Abstract

Objective  Endoscopic endonasal approaches (EEAs) have shown excellent results for majority of hypophyseal tumors. The aim of this study was to evaluate and report the complications of EEA in patients with pituitary adenoma (PA) who underwent surgery between 2013 and 2018.

Methods  We performed a retrospective review of 310 consecutive patients/325 procedures with PA treated with an EEA from May 2013 to January 2018. Minor complications including transient diabetes insipidus (DI) or new anterior pituitary hormone insufficiency in one axis and major complications including CSF leakage, hematoma needing reoperation, vascular damage, brain infection, new pan-hypopituitarism permanent DI, new visual impairment, neurological deficits, and mortality were recorded.

Results  We encountered 58 complications in 310 patients (18.7%) and 325 procedures (17.7%). Minor complications were 43 (13.9 and 13.2%) in 310 patients and in 325 procedures, respectively; whereas, major complications were 28 (9 and 8.6%, respectively). Total complications were associated with diameter group 2 (>30 mm), diaphragm sella violation, suprasellar extension, parasellar involvement, nonfunctional secretory type, and intraoperative arachnoid tearing.

Conclusion  EEA can be considered as a safe surgical treatment which has acceptable complications in the management of PAs.

Introduction

Surgical access to the sellar region has been a challenging issue for many years. A variety of approaches and techniques have been used to reach this region. ¹ After the introduction of the endoscope, surgery of the sellar region has achieved new dimensions. In recent studies endoscopic endonasal approach (EEA) has shown excellent results for majority of hypophyseal tumors. ² Complication rates in EEA for pituitary adenomas (PAs) are relatively low; they can still be significant, however ³ It seems that endoscopic techniques provide less complication in terms of gaining access to the tumor while tumor removal-related complications are still similar in both endoscopic and microscopic techniques. ⁴ In a large systematic review reporting nine endoscopic studies, the pooled complication rates for cerebrospinal fluid (CSF) leak and permanent diabetes insipidus (DI) were reported 2 and 1%, respectively. Also overall mortality rate was reported 0.24% (both due to vascular injury). ⁵ The aim of this study was to evaluate and report the complications of EEA in patients with PA who underwent surgery between 2013 and 2018.

Material and Methods

We performed a retrospective review of 310 consecutive patients/325 procedures with PA treated with an EEA approach from May 2013 to January 2018. We included all patients with a PA. Comprehensive endocrinologic and ophthalmologic assessments as well as imaging consisting of fine cut magnetic resonance imaging (MRI) and computed tomography (CT) scan with and without contrast administration were routinely done before surgery. An EEA was performed for all patients. The surgery was done by a single neurosurgeon (M.Z.) and a training resident with an expert ear nose throat (ENT) surgeon (S.M.S.) and his training fellow. These case series were the first EES series performed in our institute. Neuronavigation system (Optic Vision [Parseh Intelligent Surgical Systems Co., Parsis, IRI]) was loaded with pre-operative MRI and CT scan, and microvascular doppler probes were used in all the patients. In brief, initially partial posterior nasal septectomy was done and bilateral sphenoid ostia were located. The anterior wall of the sphenoidal sinus and the inner mucosa layer were removed to expose the sellar floor. Based on the tumor extensions, after drilling of the sellar floor, bone removal over the cavernous sinuses or tuberculum/planum sphenoidal was added to exposures. In cases without cavernous sinus involvement, bone drilling extended just to the junction of sellar and cavernous sinus dura so we could find carotid pulsation on Doppler study. If tumor was extended to the cavernous sinus, based on the tumor extension, cavernous sinus bone removal was done. Dura matter was opened and the three-hand binostril technique was used for tumor resection. The arachnoid layer was protected, if possible.

For weeping CSF leak, we used gelatin sponge or mucosal graft for dural reconstruction. When CSF fistula was obvious reconstruction was performed using fascia lata alone or in combination with fat graft. For large and high flow CSF fistula including opening two or more than two cisterns or opening of the third ventricle, the defect was reconstructed with three layers of fat graft, fascia lata, and pedicled nasoseptal flap (NSF). We did not use any artificial dural substitutes in any of our patients.

All patients visited the neurosurgery clinic in the first month after hospital discharge. Comprehensive endocrinologic and visual assessments were done 3 months after surgery. MRI with and without contrast administration was performed 3 months after surgery and every 12 months thereafter. Appropriate hormone replacement therapy was administered to patients with clinical and laboratory signs and symptoms of hypopituitarism.

Patients were distributed in two groups according to their age: less than 50-year-old and ≥50. Based on surgeon's capability in tumor resection, patients were divided into four groups: 1—more than 99% resection (gross total resection). 2—between 90 and 99% (near total resection). 3—between 70 and 90% (subtotal resection). 4—less than 70% (partial resection). Subtotal and partial resections were not considered to be complications. Suprasellar extension and diaphragm sella violation were defined as any tumor extension to the suprasellar region and through the diaphragm sella hiatus, respectively and it means as intracranial extension of the tumor. Parasellar involvement was described as any intracranial tumor extensions lateral to intra-arachnoid carotid arteries or optic apparatus.

In terms of tumor size, patients were divided into two groups: >30 mm or ≤30 mm. Arachnoid tearing and postoperative CSF leak were defined as any intraoperative CSF leakage during exposure or tumor resection and any postoperative CSF discharge from nose until 1 month, respectively. Diabetes insipidus was described as the elevated urine output more than 30 mL/kg/d, ⁶ urine specific gravity ≤1.005, urine osmolality less than 300 mOsm/kg, and serum osmolality more than 300 mOsm/kg. It was considered as temporary DI if the symptoms got relieved within 6 months. But in cases lasting for more than 6 months it was defined as permanent DI. ^{7 8 9} We defined visual impairment as any decline of at least one scale in visual acuity on standard Snellen chart or new defect in visual fields measured by the same optometrist at 3 months after surgery. New insufficiency of at least two axes of anterior pituitary hormones defined as pan hypopituitarism that we described it as a major complication, whereas one axis insufficiency was described as minor complication. Surgery-related mortality was defined as occurrence of a death attributable to surgery until 1 month after surgery. Patients were divided into two groups based on learning curve of the surgeon. As median (consequences and temporary) first 155 patients were classified as first experience group and the next 155 patients as second learning curve group. We aimed to explore neurosurgical complications; so, nasal complications were not included in this study. We divided our complications into two groups: 1—minor complications including transient DI or new anterior pituitary hormone insufficiency in one axis, 2—major complications including CSF leakage, hematoma needing reoperation, vascular damage, brain infection, new pan-hypopituitarism (pituitary hormone deficiencies in 2 or more axes), permanent DI, new visual impairment, neurological deficits, and mortality.

Statistical Analysis

SPSS version 18.0 (SPSS, Inc., Chicago, Illinois, United States) was used for statistical analysis. Baseline and postoperative values were presented as mean and standard deviation for continuous variables and frequency for categorical data. Categorical variables were analyzed with X ² or Fisher exact test. Means of continuous variables were analyzed with the Student t -test. Logistic regression analysis was done to reduce the effect of confounding factors. A p -value <0.05 was considered statistically significant.

Results

Among 310 consecutive patients with PA who were operated, 14 patients underwent reoperation and one patient had three successive surgeries. We had 325 procedures in 310 patients; so, we report the prevalence of complications in both datasets. Fortunately, our 15 revision surgery had no significant complication. Demographic characteristics of the patients are being shown ( Table 1 ).

Table 1. Demographic characteristics of patients.

Patient characteristic	Number (%) of 310
Gender
Male	146 (46.8%)
Female	165 (53%)
Age group
≤50 y	203 (63.5%)
< 50 y	107 (34.5%)
DM	44 (14.2%)
HTN	57 (18.4%)
Size
Macroadenoma	45 (14.5%)
Microadenoma	265 (85.5%)
Secretory function
Nonfunctional	147 (47.4%)
Acromegaly	109 (35.2%)
Cushing	28 (9%)
Prolactinoma	21 (6.8)
Mixed	5 (1.6%)
Diaphragm violation	113 (36.5%)
Suprasellar involvement	124 (40%)
Parasellar involvement	40 (12.9%)
KNOSP classification
stage 0	188 (60.8%)
stage I	38 (12.3%)
stage II	42 (13.5%)
stage III	21 (6.8%)
stage IV	21 (6.8%)
Presence of apoplexy	48 (15.5%)
Presence of cyst	51 (16.5%)
Surgery turn-primary	257 (83%)
Recurrent or residue	52 (17%)
Tumor resection
GTR	294 (94.8%)
NTR	2 (0.6%)
ST	5 (1.6%)
PR	9 (2.9%)
Arachnoid tearing	100 (32.3%)
Revision surgeries	14 (4.5%)

Total Complications

Overall, we encountered 58 complications in 325 procedures (17.7%). Minor complications were 43 (13.2%) in 325 procedures, whereas, major complications were 28 (8.6%). Table 2 demonstrates the rate of all complications in brief.

Table 2. Rate of different complications.

Complication	Number ( n )	% in 310 patients	% in 326 procedures
Minor complications	43	13.9%	13.2%
Major complications	28	9%	8.6%
Total	58	18.7%	17.7%
CSF leakage	8	2.6%	2.5%
DI
Total	44	14.1%	13.4%
Temporary	33	10.6%	10.1%
Permanent	11	3.5%	3.3%
Anterior pituitary insufficiency
Pan hypopituitarism	10	3.2%	3%
Minor insufficiency	4	1.3%	1.2%
Brain infections	2	0.6%	0.6%
Visual impairment	8	2.6%	2.4%
Neurological deficit	3	1%	0.9%
Required surgery hematoma	2	0.6%	0.6%
Mortality	4	1.2%	1.2%

Total complications were associated with many factors including diameter group 2 (>30 mm), diaphragm sella violation, suprasellar extension, parasellar involvement, diabetes mellitus, nonfunctional secretory type, intraoperative arachnoid tearing, gross total tumor resection, and surgeon experience group 1 (the first 155 patients). But, in multivariate analysis correlation between total complications rate and diabetes mellitus (DM) ( p  = 0.98), gross total tumor resection ( p  = 0.090), and surgeon experience group 1 ( p  = 0.061) was insignificant ( Table 3 ).

Table 3. Independent variables those were associated with complications.

Variable	p -Value (Univariate analysis)	p -Value (Multivariate analysis)
DM	0.016	0.098
Diameter group 2 (>30 mm)	0.001	0.001
Diaphragm sella violation	0.001	0.033
Suprasellar extension	0.001	0.010
Parasellar involvement	0.050	0.050
Nonfunctional adenoma type	0.018	0.023
Intraoperative arachnoid tearing	0.023	0.031
Gross total tumor resection	0.018	0.090
Experience group 1 (the first 155 patients)	0.030	0.061

CSF Leakage

Intraoperative arachnoid tearing was observed in 100 procedures (30.7% of 325 procedures). The rate of postoperative CSF leakage was 2.6% in 325 procedures. All of the leakages occurred within 2 to 7 days after surgery. Two patients were treated successfully with therapeutic lumbar punctures and six patients underwent revision surgery. Of note, in our cohort, we used NSF only in six cases as well as lateral nasal wall flap in one patient. In univariant analysis, we found no correlation between CSF leakage and our independent variables even with intraoperative arachnoid tearing or revision surgeries ( p >0.05).

Posterior Pituitary Insufficiency

Forty-four patients (14.1%) presented with postoperative DI, which in 33 cases (10.6%) was temporary, and permanent in 11 cases (3.5%). DI was related with the presence of pituitary apoplexy ( p  = 0.05) and intraoperative arachnoid tearing ( p  = 0.023) but in multivariate analysis correlation with apoplexy did not persist.

Brain Infections

Two patients (0.6%) developed postoperative meningitis. Microbiologic study of CSF in the first patient revealed Staphylococcus aureus positive culture which was treated with appropriate antibiotics successfully. Second patient encountered surgical site hematoma the day after surgery and was reoperated. Within the postoperative period the patient got Klebsiella pneumoniae meningitis and treated with broad spectrum antibiotics administration but despite negative results of CSF bacteriologic examinations, unfortunately the patient expired after 3 weeks because of extensive intraventricular hemorrhage.

Visual Impairment

In eight patients (2.6%) new visual impairments were developed. Among them seven patients encountered visual acuity aggravation and six patients with new visual field defects. One patient suffered from complete blindness of her left eye (no light perception). Inspecting procedure video revealed that probably it was due to extensive optic canal drilling. New visual impairment was associated with age group 2 (≥50) ( p  = 0.050), diaphragm sella violation ( p  = 0.022), suprasellar extension ( p  = 0.041), amount of tumor resection (near or gross total resection p  = 0.005), and nonfunctional secretory type ( p  = 0.021); but multivariate analysis revealed no significant association with age group ( p  = 0.075) and nonfunctional type ( p  = 0.053).

Anterior Pituitary Insufficiency

Ten patients (3.2%) encountered minor anterior pituitary insufficiency while pan-hypopituitarism was seen in four cases (1.3%). The incidence of these two complications correlated with diameter group 2 (>30 mm) ( p  = 0.003), presence of cyst ( p  = 0.041), gross total tumor resection ( p  = 0.017), and functional secretory type ( p  = 0.014).

Neurological Deficits

In three patients (1%) we encountered new neurological deficits. First patient presented ophthalmoplegia because of third and sixth nerve palsy due to surgical manipulation of the cavernous sinus. Second patient experienced right eye ptosis and left hemiparesis due to acute infarction in right anterior thalamus. Fortunately the patient fully recovered after 1 month. The third patient got drowsy within 1 week after surgery because of severe hyponatremia. Despite precautions in avoiding rapid correction of hyponatremia, she experienced extrapontine myelinolysis (EPM). We will present cases 2 and 3 in detail in the section of “unusual and very rare complications” later.

In univariant postoperative analysis new neurological deficits were correlated with age group 2 (>30) ( p  = 0.040) and gross total tumor resection ( p  = 0.016).

Hematoma Required Surgery

This was defined as a symptomatic or extensive hematoma that required reoperation. This complication occurred in two patients (0.6%). The first case developed by extensive surgical site hematoma without clinical presentation 8 hours after surgery. She was using antiplatelet agent (aspirin, 80 mg/d) until 5 days before surgery. The patient underwent surgical exploration and hematoma was evacuated. The second patient suffered blurred vision 1 day after surgery due to surgical site hematoma and immediately underwent surgical exploration and hematoma evacuation but in follow-up management he was affected with Klebsiella pneumonia meningitis and finally expired. In univariate postoperative analysis hematoma was associated with diameter group 2 (>30mm) ( p  = 0.050) and intraoperative arachnoid tearing ( p  = 0.033).

Vascular Injury

In this cohort we had an experience of one carotid artery injury in a young patient with a growth hormone (GH) adenoma. During bone removal over the right preclinoid carotid artery, it was injured by a Kerrison punch. Surgical site was buttressed with Surgicel using soft tamponade. Due to blood loss the operation was terminated. After 5 days, digital subtraction angiography (DSA) showed a pseudoaneurysm in clinoidal segment of left internal carotid artery (ICA) ( Fig. 1 ). This pseudoaneurysm was managed by endovascular treatment without any complication. Adenoma was successfully resected 2 days later.

Fig. 1.

A young man with a GH adenoma. ( a ) Coronal T1-weighted MRI with contrast showing macroadenoma on the right side of sella. ( b ) Digital subtraction angiography showing a pseudoaneurysm in clinoidal segment of right ICA*. ( c ) Postoperative coronal T1 MRI with contrast after 3 months with successful tumor resection. ICA, internal carotid artery; GH, growth hormone; MRI, magnetic resonance imaging.

Unusual and Very Rare Complications

There were four cases that we considered as “unusual and very rare complications.”

• Case 1: The first patient was a mix tumor of GH adenoma and small clival chordoma who experienced right eye partial ptosis and left hemiparesis just after surgery due to acute infarction in anterior right thalamus ( Fig. 2 ). During surgery, hemostatic agent (Gelfoam paste, GELITA Medical Co.) was injected into the right cavernous sinus to stop bleeding and at this time anesthesiologist reported sudden but transient bradycardia and hypotension. Pupils examination showed right eye anisocoria. We hypothesized that embolization of Gelfoam paste particles to the microvasculature could be the possible etiology of this event. Fortunately, the patient gained her full recovery after 1 month.

• Case 2: The second patient was a 36-year-old-woman with GH microadenoma who underwent EEA with success without any event during surgery. She was admitted again after 7 days from her surgery with drowsiness because of severe hyponatremia (serum Na level = 113 mEq/L). Despite precaution in avoiding rapid correction of hyponatremia, serum Na level reached to 128 mEq/L after 24 hours and she worsened with new clinical manifestations including general weakness, difficulty in swallowing and drooling, and scanning speech. MRI showed hyperintense signals in T2 and Flair in association with restriction in DWI sequences in bilateral basal ganglia, ( Fig. 3 ) typical for EPM. She was managed by a neurologist and gained her good recovery after 6 months.

• Case 3: The third case was a 53-year-old man with a nonfunctional recurrent giant PA with a maximal diameter of 49 mm and KNOSP grade I. Four days after surgery he developed midnight severe hypernatremia (serum Na level = 167 mEq/L), decreased level of consciousness, continued with bilateral aspiration pneumonia. Brain CT scan showed delayed ischemic infarction of both anterior thalami probably due to vasospasm ( Fig. 4 ). Finally, he got expired.

• Case 4: A 31-year-old woman, with nonfunctional PA and a familial history of hereditary mitochondrial myopathy underwent EEA. Twenty days after surgery, she came back to our emergency department with a stroke-like incident including right hemiparesis and dysphasia. MRI showed left hemisphere pan-cortical ischemia and sulcal effacement ( Fig. 5 ). The patient was evaluated and managed by a neurologist with Co-enzyme Q10, and L-Carnitine; and fortunately, she recovered.

Fig. 2.

(a) A patient with simultaneous GH microadenoma and ( b ) small mid-clival chordoma, in axial T1 with contrast MRI. ( c ) Intraoperative image showing chordoma tumor. ( d ) T2-weighted MRI in association with ADC map, ( e ) demonstrating an acute infarction in anterior right thalamus. ADC, apparent diffusion coefficient; MRI, magnetic resonance imaging.

Fig. 3.

( a ) Preoperative T1-weighted with contrast MRI shows a microadenoma on the right side of sella turcica. ( b –d) Early postoperative T2, Flair, and DWI, MR images respectively showing hyperintense signals and ( e ) hypointense signals in ADC map in bilateral basal ganglia; typical for EPM. ADC, apparent diffusion coefficient; DWI, diffusion-weighted imaging; MRI, magnetic resonance imaging.

Fig. 4.

( a ) Sagittal T1-weighted MRI with contrast demonstrating a recurrent giant PA. ( b ) Axial CT scan the day after surgery with pneumocephalus and pneumoventricle. ( c ) Ischemic infarction in bilateral thalami in axial CT scan 4 days after surgery. CT, computed tomography; MRI, magnetic resonance imaging; PA, pituitary adenoma.

Fig. 5.

( a ) Sagittal CT scans with contrast shows a pituitary adenoma with mild suprasellar extension. ( b–d ) T2, Flair, and T1 with contrast-weighted MRI 20 days after surgery showing left hemisphere pan-cortical ischemia and sulcal effacement. CT, computed tomography; MRI, magnetic resonance imaging.

Mortality

The overall mortality rate was 1.24% (four patients) which will be reviewed in detail.

The first patient was a 55-year-old male who had a nonfunctional recurrent PA with a maximal diameter of 70 mm and KNOSP grade III tumor. He underwent surgery and nasal septal flap was applied to reconstruct the sellar floor. Five days after surgery he encountered CSF leakage and pneumocephalus. There were no signs of infection and no positive CSF study. He underwent immediate exploration surgery and the surgeon encountered necrotic changes of applied nasal septal flap. So, it was resected and we reconstructed sellar floor again. But in postoperative period the patient expired because of septicemia and septic shock.

The second case was a 58-year-old woman with Cushing disease and underlying diseases including hypertension and DM. She had a microadenoma and got operated successfully with remission and a favorable postoperative period but within 2 weeks she encountered extensive right hand staphylococcal cellulitis in location of IV line and treated successfully. But unfortunately she developed myocardial infarction and died.

The third case was a 66-year-old man with a nonfunctional recurrent PA who is presented as the second case in the section “ Brain infections.” As we have mentioned above, the reason of death was gram-negative bacterial meningitis.

The last case was a 53-year-old man with a nonfunctional recurrent PA who is presented as case#3 in the section “ Unusual and very rare complications.”

The mortality rate was correlated with age group 1 (≤50 years) ( p  = 0.001) and surgery turn (residue or reoperation p  = 0.001).

Discussion

In this study we report our complications from a tertiary pituitary center with 325 EEA in 310 consecutive patients performed over 5 years specially focusing on rare and serious complications. The overall rate of complications was 17.7% in 325 procedures. Major complications occurred in 8.6% of procedures. Generally there is a wide range of overall complications rate in the studies from 8 to 26%. ^{10 11 12 13} These differences are clearly influenced by the authors' definition of the complications. Lofrese et al mentioned that their overall complication rate was 8% which was lower than in our study. ¹³ This could be due to the fact that they considered only four items as surgical complication while we had at least three items more than them (meningitis, DI, and adrenal insufficiency). The overall complications rate in Halvorsen et al study was 16.4% which was near to our study. ¹⁴

Diabetes Insipidus was the most common complication of our surgeries followed by CSF leakage. Some other studies reported DI as their most common complication like our series. ^{2 15 16} In univariate analysis our complications significantly correlated with different parameters including maximal tumor diameter (>30 mm), diaphragm sella violation, suprasellar and parasellar extensions, DM, nonfunctional secretory type, intraoperative arachnoid tearing, near and gross total resection, and experience group 1 (the first 155 cases). Some of these associations (DM, gross total tumor resection, and experience group 1) became insignificant when multivariate analysis was performed. It seems that these associations are expectable and most of them are suggested by the other studies. ^{2 10 15 17 18} Although we considered first 155 patients as group 1 who experienced more complications in univariate analysis, we are not sure that this point was the exact middle point of the learning curve. Because learning curve is not a simple linear curve, different factors could have impact on it. Serious complications such as carotid injury and stroke could be prevented using carotid Doppler sonography, neuronavigation, well-established collaboration with ENT team and being cautious during electrolyte imbalance corrections. For example in patient with right thalamus infarction, may be direct injection of Surgifoam into cavernous sinus was the etiology and it should be used cautiously in future cases. In patients with familial history of hereditary mitochondrial myopathy may be preoperative use of co-enzyme Q10, and L-Carnitine could prevent ischemic events. Hypernatremia could increase the risk of vasospasm by inducing diuresis and dehydration, so wise correction of it with collaboration by endocrinologist and nephrologist could make the outcomes better.

CSF Leakage

In our cohort, CSF leakage was the second common major complication (2.6%) and comparable to other reported studies that vary from 0.5 to 6%. ^{2 14 19 20} In a study conducted in Middle East by Alsaleh et al they reported that 20% of their endoscopic patients suffered from CSF leakage. ²¹ Except in two cases that had minor leakage and treated by medical treatment and lumbar punctures, we did early surgical exploration in six patients to reconstruct the skull base as soon as possible. We used on-lay fat graft and fascia lata in all revision surgeries. Among patients in whom a NSF was used (six cases), (1.9%), two patients had CSF leakage. In first case probable scenario was the damage to the septal artery during previous microscopic surgery. In the second case, during revision surgery we encountered shrinkage of the NSF probably due to inappropriate NSF installation in the first operation. Gondim et al used NSF only in patients with important CSF leakage who needed reconstruction surgery (3.1%). ¹ Previous studies recorded that NSF installation could be successful in up to 94% of cases and it could decrease CSF leak rate from >20% to <5%. ²² We found no correlation between CSF leakage and revision surgeries with independent variables of the study.

Hormonal Insufficiencies

DI was observed in 14.1% but it persisted as permanent DI in only 3.5% of patients. Also 3.2% of our patients experienced minor anterior pituitary insufficiency while 1.6% of them encountered pan-hypopituitarism (two or more axes hormonal deficiencies). Tabaee et al in a meta-analysis considering complications of endoscopic surgery concluded that anterior pituitary dysfunction rate in the EEA varies from 3 to 14.5% and DI ranged from 1 to 13.6%. ⁵ Hormonal complications in our study correlated with tumor diameter more than 30 mm, near and gross total tumor resection, functional secretory type, and intraoperative arachnoid tearing. It indicates that our effort to reach gross total resection (GTR) or near total resection (NTR), especially in larger tumors, and consequently extensive exploration of sellar and suprasellar cavity for residual tumor may be partially the leading cause of normal gland manipulation and subsequent hormonal deficiencies.

Meningitis

The rate of meningitis was 0.6% (two cases) in our cohort but none of them had CSF leakage. Our result was comparable with other studies ranging from 0.4 to 2%. ^{11 15 19 23} One of our two patients with meningitis developed positive CSF culture with Klebsiella pneumonia and unfortunately expired. Shibao et al mentioned that Staphylococcus epidermis was the most frequently detected species of both nasal cavity and sphenoid sinus. ²⁴ In a study by Kono et al, they could not identify any species as predominant germ among 11 patients with positive culture meningitis and there was a wide variety of microorganisms. ²⁵

Visual Impairments

The rate of ophthalmopathy in previous studies is reported up to 0.68%. ¹⁵ In this study 2.6% of the patients experienced postoperative worsening of the visual parameters (visual acuity or field) which were higher than our series in craniopharyngioma patients. ²⁶ One of our patients encountered left side blindness probably due to extensive drilling of the optic canal during surgery. Accurate explorations and continuous saline irrigation during drilling may reduce such injuries to optic nerves. In another case, surgical site hematoma leads to optic apparatus injury. Perhaps, quick diagnosis and surgical intervention in association with meticulous hematoma evacuation restore visual impairments.

Hematoma

We had two patients who required surgical intervention because of surgical site hematoma. Rate of postoperative hematoma in different studies varies from 0.4 to 1.2% ^{4 10 15} which in our cohort is in line with these studies.

Neurological Deficits

Three cases (1%) experienced new postoperative neurological deficits. Postoperative neurological deficits were correlated with age, tumor size greater than 30 mm, and gross total tumor resection. Fortunately, all patients recovered uneventfully. The rate of neurological complications varies among different studies but is usually presented up to 1% which our study shows is concordance with this rate. ^{12 16 18 19}

Mortality

As the most serious complication, the rate of surgical mortality was 1.2% (four patients). Most of the studies evaluating EEA for pituitary tumors including one large systematic review have reported mortality rates near to 1%. ^{12 27} In an interesting study by Ciric et al they sent 3,172 questionnaires to neurosurgeons to gather their experience about complications in transsphenoidal surgeries (both endoscopic and microscopic). ²⁸ In their national survey the mean operative mortality rate in groups with different learning curves was reported to be 0.9%. In our study mortality had significant relationship with repeated surgeries. It seems that meningitis, ischemic events, and systemic complications are among important etiologies that can lead to patients' death. Table 4 shows in detail the rate of complications reported in big case series.

Table 4. Literature review about complications of endoscopic TSS.

Author	Number of patients	CSF leakage	DI	Anterior pituitary insufficiency	Brain infection	Vision impairment	Hematoma	Vascular injury	Mortality
Cappabianca et al (2002) 15	146	2%	13.6%	3.42%	0.6%	0.68%	0.68%	0.6%	0.68%
Divitiis et al (2003) 10	N/A	2.1%	3.1%	14.5%	0.4%	0.4%	1.2%	0.4%	0.4%
Frank et al (2006) 4	381	1.2%	1.4%	3.1%	0	0	0.4%	0	0
Charalampaki et al (2009) 16	150	3.7%	5.9%	4.4%	N/A	N/A	N/A	0.7%	0.7%
Gondim et al (2010) 1	228	2.6%	6.3%	11.6%	0.6%	0.3%	0.6%	1%	1%
Zhou et al (2010) 19	375	0.5%	3.7%	N/A	N/A	0.6%	N/A	N/A	0
Halvorsen et al (2011) 14	506	0.4%	N/A	N/A	0.7%	1.4%	1.1%	0.3%	0
Mustafa berker et al (2012) 2	570	1.2%	4.6%	1.92%	0.7%	0	0	0.16%	0
Our study	310	2.6%	3.5%	3.2%	0.6%	2.6%	0.6%	0.3%	1.2%

Abbreviation: N/A, not applicable.

Limitations

For some uncommon complications such as meningitis, hematoma needing evacuation, vascular injury, and neurological events, because of low number of incidents in our cohort, precise analysis and relationships finding need a larger case series. Also, we did not include nasal complications to the study. Further prospective randomized investigations with larger sample sizes including nasal complications can help to find out the relationship of the complications with different variables precisely.

Conclusion

In our study complications were significantly correlated with different variables including maximal tumor diameter (>30 mm), diaphragm sella violation, suprasellar and parasellar tumor extensions, diabetes mellitus, nonfunctional adenoma type, intraoperative arachnoid tearing, near and gross total tumor resection, and experience group 1 (the first 155 cases). In multivariate analysis only correlation between total complications rate and diameter group 2 (>30 mm), diaphragm sella violation, suprasellar extension, parasellar involvement, nonfunctional secretory type, and intraoperative arachnoid tearing remained significant. Except for visual impairments, the rate of our complications is similar to other studies and is comparable with the literature. So, for PAs EEA as a safe surgical treatment with acceptable complications is advisable.