Quantitative analysis of anatomical relationship between cavernous segment internal carotid artery and pituitary macroadenoma.

Cavernous segment internal carotid artery (CSICA) injury during endoscopic transsphenoidal surgery for pituitary tumor is rare but fatal. The aim of this study is to investigate anatomical relationship between pituitary macroadenoma and corresponding CSICA using quantitative means with a sense to improve safety of surgery.In this retrospective study, a total of 98 patients with nonfunctioning pituitary macroadenomas undergoing endoscopic transsphenoidal surgeries were enrolled from 2005 to 2014. Intercarotid distances between bilateral CSICAs were measured in the 4 coronal levels, namely optic strut, convexity of carotid prominence, median sella turcica, and dorsum sellae. Parasellar extension was graded and recorded by Knosp-Steiner classification.Our findings indicated a linear relationship between size of pituitary macroadenoma and intercarotid distance over CSICA. The correlation was absent in pituitary macroadenoma with Knosp-Steiner grade 4 parasellar extension.Bigger pituitary macroadenoma makes more lateral deviation of CSICA. While facing larger tumor, sufficient bony graft is indicated for increasing surgical field, working area and operative safety.

Introduction

Endoscopic transsphenoidal surgery is one of standard techniques for pituitary adenoma. Among reported complications of endoscopy procedure, injury to the internal carotid artery (ICA) is a fatal complication with an incidence of 0.25% to 3.8%.[ Parasellar ICA and sella turcica occupy the central portion of surgical field, and being familiar with surrounding structures is absolutely necessary. Several cadaveric and imaging studies have reported the feasibility of identified bony landmarks and normal ICA course during endoscopic transsphenoidal surgery.[ Nevertheless, studies discussing the anatomical relationship between pituitary adenoma and parasellar ICA are limited.

Preoperative assessment of association between pituitary adenoma and actual ICA course is important to minimize risk of ICA injury. Due to flattened view of 2-dimensional endoscopic surgery, modification of ICA nomenclature has been adopted for endoscopic surgery.[ DePowell et al[ proposed a modified ICA nomenclature suitable for both transcranial and endoscopic approaches and separated the cavernous segment ICA (CSICA) into 3 portions—C3-C4 bend, C4 segment, and C4-C5 bend.

Knowing anatomy of 3 portions of CSICA and their relationship with pituitary adenoma is suggested to assist surgeons in avoiding intraoperative ICA damage. Nevertheless, studies that examine the correlation of anatomic characteristics with CSICA in endoscopic perspective are limited. The aim of this study is to quantitatively analyze and examine the anatomical relationship between pituitary adenoma and corresponding parasellar ICA, named by DePowell's modified endoscopic classification.

Methods

Patient demographics

From January 2005 to December 2014, 129 endoscopic transsphenoidal operations for pituitary adenomas were performed in our institution (Tri-Service General Hospital, National Defense Medical Center). All procedures were approved by Taiwan neurosurgical society. For each tumor, functional status and permanent diagnosis were confirmed by hormone testing and histopathology respectively. Patients with nonfunctioning pituitary macroadenomas were eligible. Those with hormonally active tumors, previous pituitary surgery, previous pituitary radiosurgery, and incomplete MRI studies were excluded. A total of 98 patients were enrolled, including 65 males and 33 females with a mean age at the time of surgery of 51.18 years (range 13–87 years).

MRI acquisition

Preoperative MRI was performed on each patient using a 1.5-T MR unit (Vision Plus, Siemens). The MRI protocol was described as following: TR 2140 ms, TE 30 ms, TI 420 ms, matrix size 256 × 256, section thickness 2 mm, and intersection gap 0.21 mm. Routine images of the whole brain, including spin echo T1-weighted images, spin echo T2-weighted images, and fluid-attenuated inversion recovery (FLAIR) images were obtained. Spin echo contrast-enhanced T1-weighted images were obtained in the coronal, sagittal, and axial planes after intravenous gadolinium (Gd) administration (0.1 mmol/kg body weight).

Imaging measurement

Size of pituitary macroadenoma was assessed by 2 parameters on T1-weighted Gd-enhanced images: dimension of tumor in the coronal plane of median sella turcica; height of tumor in the midsagittal plane.

Distances between medial walls of bilateral cavernous segment internal carotid arteries (CSICAs) in 4 coronal planes from anterior to posterior: level 1, at the optic strut; level 2, at the convexity of carotid prominence; level 3, at the level of median sella turcica; level 4, at the dorsum sellae (Fig. 1).

Figure 1

MRI of 2 patients with pituitary macroadenomas. For each patient, intercarotid distances (horizontal lines) were measured between medial walls of bilateral cavernous segment ICAs in 4 coronal levels: (A and a) at the optic strut; (B and b) at the convexity of carotid prominence; (C and c) at the median sella turcica; (D and d) at the dorsum sellae. Grading of parasellar extension was estimated by Knosp–Steiner classification on coronal contrast-enhanced images. (E) Patient 1, grade 3 on the left side and grade 0 on the right side. (e) Patient 2, grade 1 on the left side and grade 2 on the right side. ICA = internal carotid artery, MRI = magnetic resonance imaging.

For each patient, the coronal dimension of tumor in the coronal plane of median sella turcica and measured distances between medial walls of bilateral CSICAs in 4 coronal levels were divided into right and left components separately by extension of septum pellucidum. Definitions of distances measured in this study were shown in Table 1 and Fig. 2.

Table 1

Definitions of distances measured in this study.

Figure 2

Three orthogonal MRA images for pituitary macroadenoma. MRA = magnectic resonance angiography.

Differences between left-sided and right-sided CSICA lateral deviation (distance measured from medial wall CSICA to midline) in 4 coronal levels.

Grading of parasellar extension of each tumor was evaluated by Knosp–Steiner classification on both sides in the coronal plane of median sella turcica.

Imaging analysis

The anatomical relationship between pituitary macroadenoma and CSICA was analyzed by following perspectives: (1) association between tumor size (including coronal dimension and sagittal height) and distances between bilateral CSICAs in 4 coronal levels; (2) association between tumor size and the difference between left-sided and right-sided CSICA lateral deviation in 4 coronal levels; (3) association between coronal dimension of tumor on one side and ipsilateral ICA lateral deviation in 4 coronal levels. The impact of different grading of parasellar extension was analyzed.

Statistical analysis

All data were analyzed using the SPSS statistical program (version 20, IBM), and statistical significance was set at P < 0.05. The data of these measurements were presented as mean values ±SD of the mean. The difference in various parameters was analyzed with an independent samples t test and significant test. Pearson correlation coefficient was measured for the assessment of associations between continuous variables. While facing 2 predictive variables at the same time, standardized Regression Coefficients were used to assess the strength of the relationship.

Results

Size of pituitary macroadenomas

The height of macroadenoma in the midsagittal plane was 26 ± 9.4 mm (range 9.3–56.6 mm). The dimension of macroadenoma in the coronal level of median sella turcica was 26.26 ± 6.68 mm (range 15.5–48.5 mm). Each component was 13.9 ± 4.39 mm (range 6–27.9 mm) on the left side and 12.36 ± 4.43 mm (range 4.2–23.8 mm) on the right side. Results of radiographic measurements of 98 pituitary macroadenomas were shown in Table 2.

Table 2

Results of radiographic measurements in 98 pituitary macroadenomas.

Measured distances over CSICAs

Distances between medial walls of bilateral CSICAs at the optic strut (D1), at the convexity of carotid prominence (D2), at the median sella turcica (D3) and at the dorsum sellae (D4) were 16.36 ± 3.18 mm (range 9.5–25.2 mm), 24.22 ± 3.72 (range 16.4–36.6 mm), 23.69 ± 4.05 mm (range 12.9–35.5 mm), and 23 ± 4.39 mm (range 12.9–35 mm) respectively.

Distances between medial wall of left CSICA and midline (left-sided CSICA lateral deviation) at the optic strut (D1-L), at the convex of carotid protuberance (D2-L), at the median sella turcica (D3-L) and at the dorsum sellae (D4-L) were 8.51 ± 2.11 mm (range 4.1–13.5 mm), 12.71 ± 2.44 mm (range 7.5–18.9 mm), 12.39 ± 2.72 mm (range 5.3–18.8 mm), and 11.97 ± 2.72 mm (range 4.9–18.8 mm). On the right side, the distances in 4 coronal planes were D1-R: 7.93 ± 2.33 mm (range 3.5–17.2 mm), D2-R: 11.46 ± 2.44 mm (range: 6.4–18.4 mm), D3-R: 11.31 ± 2.7 mm (range 2.1–19.7 mm), and D4-R: 11.05 ± 3.04 mm (range 2.1–20.2 mm).

Differences between left-sided and right-sided CSICA lateral deviation at the optic strut (D1-dif.), at the convex of carotid protuberance (D2-dif.), at the median sella turcica (D3-dif.), and at the dorsum sellae (D4-dif.) were 2.79 ± 1.67 mm (range 0.3–7.6 mm), 2.99 ± 2.34 (range 0.1–10.5 mm), 3.07 ± 2.59 mm (range 0.1–12.2 mm), and 3.14 ± 2.68 mm (range 0.1–12.2 mm), respectively.

Anatomical relationship between tumor and CSICA

The coronal dimension of pituitary macroadenoma was significantly associated with distances between medial walls of bilateral CSICAs in 4 coronal planes. The correlation coefficients were 0.547 at the level of optic strut; 0.634 at the level of convexity of carotid prominence; 0.667 at the level of median sella turcica and 0.638 at the level of dorsum sellae (Fig. 3, A–D). There was also a significant association between the height of pituitary macroadenoma and medial walls of bilateral CSICAs in 4 coronal planes (Fig. 3, A–D). The coronal dimension and height of pituitary macroadenoma were poorly correlated with difference between right-sided and left-sided CSICA lateral deviation in 4 coronal levels (Fig. 4).

Figure 3

Regression analysis between size of pituitary macroadenoma (including coronal dimension and sagittal height) and intercarotid distances in 4 coronal levels. Relationship between coronal dimension of tumor and intercarotid distances at the optic strut (A, correlation coefficient (r) = 0.547); at the convexity of carotid prominence (B, r = 0.634); at the median sella turcica (C, r = 0.667); at the dorsum sellae (D, r = 0.638). Relationship between sagittal height of tumor and intercarotid distances at the optic strut (a, r = 0.452); at the convexity of carotid prominence (b, r = 0.391); at the median sella turcica (c, r = 0.454); at the dorsum sellae (d, r = 0.483).

Figure 4

Regression analysis between size of pituitary macroadenoma (including coronal dimension and sagittal height) and differences of left-sided and right-sided ICA lateral deviation in 4 coronal levels. Relationship between coronal dimension of tumor and differences of bilateral ICA lateral deviation at the optic strut (A, correlation coefficient (r) = 0.126); at the convexity of carotid prominence (B, r = 0.218); at the median sella turcica (C, r = 0.138); at the dorsum sellae (D, r = 0.174). Relationship between sagittal height of tumor and differences of bilateral ICA lateral deviation at the optic strut (a, r = 0.008); at the convexity of carotid prominence (b, r = 0.067); at the median sella turcica (c, r = 0.086); at the dorsum sellae (d, r = 0.080). ICA = internal carotid artery.

For each side of macroadenoma, the strong association exists between dimension of tumor on 1 side and ipsilateral ICA lateral deviation in 4 coronal levels. The standardized regression coefficients were 0.502 at the level of optic strut; 0.615 at the level of convexity of carotid prominence; 0.691 at the median sella turcica and 0.666 at the level of dorsum sellae (Fig. 5).

Figure 5

Regression analysis between coronal dimension of tumor on the one side and ipsilateral ICA lateral deviation in 4 coronal levels. Relationship between coronal dimension of tumor on the one side and ipsilateral ICA lateral deviation at the optic strut (A, standardized regression coefficient (R) = 0.502); at the convexity of carotid prominence (B, R = 0.615); at the median sella turcica (C, R = 0.691); at the dorsum sellae (D, R = 0.666). ICA = internal carotid artery.

Impact of grading of parasellar extension

Bilateral sides of parasellar extension were recorded for each patient. On the left side, amount of tumors with Knosp–Steiner grade 0, 1, 2, 3, 4 were 20, 40, 20, 11, and 7. On the right side, amount of tumors with grade 0, 1, 2, 3, 4 were 35, 31, 19, 5, and 8, respectively.

Coronal dimension of tumor and ICA lateral deviation in the 4 coronal levels with different grading of parasellar extension were shown in Table 3. Within each grading of parasellar extension, the relationship between coronal dimension of tumor and ICA lateral deviation in 4 coronal levels were expressed as Pearson correlation efficiencies (Table 4). Within grade 4 of parasellar extension, coronal dimension of tumor on one side is not correlated with ipsilateral ICA lateral deviation (P > 0.05).

Table 3

Radiographic measurements for pituitary macroadenomas with different grades of parasellar extension.

Table 4

Correlation coefficients between tumor size and lateral deviation of ICA for pituitary macroadenomas with different grades of parasellar extension.

Discussion

Being familiar with position of cavernous segment internal carotid artery (CSICA) within limited surgical field is important while performing endoscopic transsphenoidal approach. Intercarotid distances between patients with pituitary macroadenomas and healthy individuals were compared and significantly wider intercarotid distances were observed for those with pituitary macroadenomas.[ Although quantitative analysis of distances between CSICAs while facing endoscopic transsphenoidal approaches were conducted, these studies failed to discuss the direct association between CSICA morphology and pituitary adenoma.[ Measurements from cadaveric study and those from image study existed significant difference.[

Commonly accepted classification of ICA is based on direction of antegrade flow and surrounding anatomy through dorsolateral transcranial approach, and the most used form, proposed by Bouthillier et al,[ separates the ICA into 7 segments. The cavernous segment ICA is further divided into 5 portions, described by Debrun et al[: (1) anterior ascending portion; (2) junction of anterior ascending and horizontal portion; (3) horizontal portion; (4) junction of horizontal and posterior ascending portion; (5) posterior ascending portion.

Traditional ICA nomenclature is ineligible and inappropriate for endoscopic transsphenoidal approach because of flat nature of 2-dimensional endoscopic view. Several modified classifications are announced for better understanding of ICA course through 2-dimensional endoscopic view.[

DePowell et al[ proposed 1 revised classification of ICA, suitable for both transcranial and endoscopic approaches, as universal language between clinicians. Its description of the CSICA, including C3–4 bend, C4 segment, and C4–5 bend, is used in our study to represent different portion of intracavernous ICA while performing the endoscopic transsphenoidal approach. Four coronal levels of ICA measurement represent different portion of CSICA, respectively. ICA in the coronal level of optic strut, corresponding to DePowell's end of C4-C5 bend, is the site of junction of CSICA and clinoid segment ICA. Level of carotid prominence, corresponding to DePowell's end of C4 segment and beginning of C4-C5 bend, points to the transition from horizontal portion to anterior ascending portion. Level of median sella turcica, corresponding to DePowell's median of C4 segment, represents the horizontal portion. Level of dorsum sellae, corresponding to DePowell's end of C3-C4 bend and beginning of C4 segment, approximates the transition from posterior ascending portion to the horizontal portion.

Increased intercarotid distances between CSICA in patients with pituitary adenomas were reported by previous studies.[ The result of our study elaborate on intercarotid distances being linearly proportional to size of pituitary macroadenomas—larger tumor causing wider intercarotid distances in 4 coronal levels of measurement, whole sella turcica. It is postulated that progressive enlargement of pituitary adenoma causing expansion of sella turcica and lateral deviation of cavernous sinus structure, including CSICA. Considering asymmetrical growth of pituitary macroadenoma may cause unpredicted CSICA lateral deviation, we divide the coronal dimension of tumor and measure intercarotid distances into right-sided and left-sided components. Existence of linear relationship between coronal dimension of tumor on one side and ipsilateral ICA lateral deviation in 4 coronal levels is also observed. Otherwise, differences between left-sided and right-sided CSICA lateral deviation in 4 coronal levels are not related to tumor size. By the above stated findings, we make 2 conclusions. The first, the amount of lateral deviation of CSICA over whole sella turcica is directly proportional to size of pituitary adenoma. The second, increased size of pituitary adenoma would not increase asymmetry of CSICA lateral deviation.

Cavernous sinus invasion by pituitary adenoma is considered as another issue with an incidence of 6% to 10%.[ Preoperative investigation for cavernous sinus invasion by pituitary adenoma is important to reduce operative morbidity. Although intraoperative finding is the gold standard diagnosis for cavernous sinus invasion, multiple radiographic characteristics are suggested as predictors for cavernous sinus invasion preoperatively with different sensitivity and specificity.[ Researches had highlighted that increased size of pituitary adenoma accompanied with higher risk of cavernous sinus invasion.[ However, the impact of cavernous sinus invasion on CSICA morphology is still unknown. In this study, Knosp–Steiner classification is employed to categorize the grading of parasellar extension by pituitary macroadenoma. Grading of parasellar extension on both sides of pituitary adenomas in the coronal plane of median sella turcica is recorded and its association with tumor size and CSICA lateral deviation is analyzed.

By the result of the study, increased grading of parasellar extension of pituitary macroadenomas accompany with larger coronal dimensions and further CSICA lateral deviation. Pituitary adenoma with grade 4 parasellar extension does not get linear relationship between tumor size and ICA lateral deviation. Preoperatively getting information about size of pituitary adenoma, Knosp–Steiner grading of parasellar extension is important and helpful in predicting intraoperative CSICA morphology.

Limitations of this study include small patient population, retrospective nature, and interobserver variability. In the future, larger patient number and prospective studies are needed to confirm the accuracy of this study.

Conclusion

Linear relationship exists between size of pituitary adenoma and intercarotid distance over CSICA—bigger tumor mass makes more lateral deviation of CSICA. While facing larger tumor, sufficient bony graft is indicated for increasing surgical field, working area, and operative safety. However, it is worth noting that this relationship does not exist in tumors with Knosp–Steiner grade 4 parasellar extension.