Preoperative Surgical Risk Assessment by Evaluation of Tumor Blood Flow Using Perfusion Imaging by Computed Tomography in Transsphenoidal Surgery for Pituitary Adenoma

Background: Complete removal of large and giant pituitary adenomas (PAs) with suprasellar extension by transsphenoidal surgery (TSS) is sometimes dicult. Swelling and bleeding of a residual suprasellar tumor after incomplete removal via TSS can compress the surrounding neurovascular structures, resulting in postoperative complications. We measured PA blood ow by perfusion computed tomography (PCT) and assessed its usefulness in the preoperative prediction of postoperative hemorrhagic complications after TSS. Method: We performed PCT in 28 patients with PAs with suprasellar extension before TSS. Perfusion studies were performed with a 320-row multidetector computed tomography system. We measured cerebral blood volume (CBV) and cerebral blood ow (CBF) of the tumor (CBVt and CBFt) and normal-appearing white matter (CBVw and CBFw). Results: In all patients, CBVt was greater than CBVw. There was a statistically signicant difference between CBFt and CBFw (p < 0.0001). The patients were divided into 2 groups according to CBFt: high CBF (CBFt> mean value) and low CBF (CBFt< mean value). There were no statistically signicant differences between the 2 groups in terms of gender, age, maximum tumor diameter, suprasellar extension grade, and extent of resection. Two of the 7 patients in the high CBF group suffered from visual acuity deterioration and visual eld stenosis due to postoperative hemorrhage, while no patients in the low CBF group did (p = 0.014). Conclusions: PCT may be useful in the preoperative prediction of postoperative residual tumor hemorrhage, which can be a hazardous postoperative complication after TSS for large and giant PAs.


Introduction
Transsphenoidal surgery (TSS) is often the rst choice for the treatment of most pituitary adenomas, and the clinical outcomes are satisfactory when using either an endoscope or a microscope [4,10,11,16].
However, the surgical treatment of large and giant pituitary adenomas with suprasellar extension has been controversial. Some tumor characteristics such as a dumbbell-shaped adenoma, a multi-lobular adenoma, asymmetrical suprasellar extension, and a normal-sized pituitary fossa with a predominant suprasellar portion are not suitable indications for TSS [6]. Insu cient transsphenoidal resection puts patients at risk of postoperative bleeding, edema, and increased mass effect of the residual tumor.
Perfusion imaging by computed tomography (PCT) re ects blood ow and capillary condition, making it valuable in the assessment of brain tumors. PCT has been reported to be useful in differentiating malignant brain tumors, grading gliomas, and predicting prognosis after treatment [2,15,19]. Re ecting their neovascularization and increased angiogenic activity, high-grade gliomas have greater cerebral blood volume (CBV) and cerebral blood ow (CBF) compared with low-grade gliomas [2,19]. We measured the CBV and CBF of pituitary tumors to assess tumor vascularization and evaluated the association of elevated CBV and CBF with the incidence of postoperative hemorrhage after TSS for pituitary adenoma.

Patients
This retrospective study was approved by the institutional review board of the Shizuoka Cancer Center, and the requirement for informed consent was waived. All methods were performed in accordance with relevant guidelines and regulations.
We included 28 patients with pituitary adenomas treated by TSS at the Shizuoka Cancer Center from March 2015 to December 2019. There were 12 men and 16 women between 29 and 84 years of age (mean age: 56.8 years).

Image Acquisition
PCT was performed before surgery using a 320-row multidetector computed tomography system (Aquillion ONE; Canon Medical Systems, Tochigi, Japan). Seven seconds after intravenously injecting 50 mL (5 mL/s) of a nonionic contrast medium (Iomeron, 350 mg/mL; Bracco-Eisai, Tokyo, Japan) using a power injector, dynamic scanning was started, and tissue attenuation of the contrast medium was monitored.
Acquired data were analyzed using the delay-insensitive singular value decomposition plus perfusion algorithm on the computed tomography scanner console. The CBV, CBF, mean transit time, and time to peak were displayed automatically. Regions of interest were placed manually at the tumor and normalappearing white matter.
A total of 23 dynamic scans were performed for each patient, and radiation doses for PCT were 157.5 mGy per the volume computed tomography dose index and 2520 mGy cm per the dose length product.

Statistical Analysis
Analyses were performed using version 11 of the JMP® software (SAS Institute Inc., Tokyo, Japan). The differences in CBV and CBF between the tumor (CBVt and CBFt) and normal-appearing white matter (CBVw and CBFw) were compared by the Wilcoxon rank sum test. Clinical characteristics were compared using the chi-square test. P < 0.05 was considered statistically signi cant.
In all patients, CBVt was greater than CBVw. There was a statistically signi cant difference between CBVt and CBVw (p < 0.0001; Fig. 1A).
CBFt ranged from 28.4 to 133.2 (51.3 ± 27.7) mL/min/100 g, and CBFw ranged from 18.4 to 31.3 (24.1 ± 3.6) mL/min/100 g. There was a statistically signi cant difference between CBFt and CBFw (p < 0.0001; Fig. 1B). The patients were divided into 2 groups according to CBFt: high CBF (CBFt > mean value; Clinical characteristics, maximum tumor diameters, suprasellar extension grades [12,17], and incidence of postoperative hemorrhage are summarized in Table 1. In the high CBF group, there were 2 men and 5 women between 46 and 67 years of age (mean age: 50 years). In the low CBF group, there were 10 men and 11 women between 29 and 84 years of age (mean age: 59 years). Total tumor removal was achieved in 9 patients (43%) in the low CBF group and 1 patient (14%) in the high CBF group. There were no statistically signi cant differences between the 2 groups regarding the maximum tumor diameter, suprasellar extension grade, and extent of resection. Two of the 7 patients in the high CBF group suffered from visual acuity deterioration and visual eld stenosis due to postoperative hemorrhage, while no patients in the low CBF group did (p = 0.014).
A representative case with high CBFt is shown in Fig. 3. A preoperative gadolinium-enhanced coronal T1weighted image (Fig. 3A) shows a pituitary adenoma with suprasellar extension (Grade C). The scaled color maps for CBV (Fig. 3C) and CBF (Fig. 3D) show greater CBV and CBF of the tumor (6.6 and 77.5, respectively) than of the normal-appearing white matter (1.9 and 18.4, respectively). A postoperative gadolinium-enhanced coronal T1-weighted image (Fig. 3B) shows expansion of the tumor due to postoperative hemorrhage.

Discussion
There are 4 grades of pituitary adenoma with suprasellar extension: 1) Grade A, 0-to 10-mm suprasellar extension occupying the suprasellar cistern; 2) Grade B, 10-to 20-mm suprasellar extension with elevation of the third ventricle; 3) Grade C, 20-to 30-mm suprasellar extension occupying the anterior third ventricle; and 4) Grade D, > 30-mm suprasellar extension beyond the foramen of Monro or a Grade C tumor with lateral extension [12]. There is a 39.5% risk of residual or recurrent tumors after an initial TSS on imaging for Grade C and D adenomas [12]. Based on their prospective study, Honegger et al. reported that vertical intracranial extension was the strongest independent predictor of incomplete resection after TSS for pituitary adenomas with suprasellar extension [6]. A residual suprasellar tumor can cause postoperative hemorrhage, which can result in compression of the optic pathway and acute hydrocephalus [13,20,21]. Mortini et al. reported that complete giant pituitary adenoma removal was obtained in only 13 of 67 patients (19%) after an initial TSS [13]. Swelling and bleeding of a residual suprasellar tumor or sellar hematoma occurred in 8 of 85 patients (9%) treated by TSS. Saito et al. recommended that the intrasellar dead space and sellar oor not be reconstructed in patients undergoing subtotal or partial tumor removal during an initial TSS in an intentionally staged operation, allowing marked descent of the diaphragma sellae and suprasellar tumor, thus avoiding postoperative complications [17]. Zoda et al. reported that an open craniotomy might be selected as the initial operation in cases with the following factors: signi cant suprasellar extension, lateral extension, retrosellar extension, brain invasion with edema, rm tumor consistency, involvement or vasospasm of the arteries of the circle of Willis, and optic apparatus encasement or optic foramina invasion [21].
A combined simultaneous transsphenoidal and transcranial approach to large-to-giant pituitary adenomas has been adopted to maximize tumor excision and lower the risk of swelling and bleeding of the residual tumor [1,3,9]. Recently, Han et al. reported that an endoscopic endonasal TSS was their rst choice for surgical management of giant pituitary adenomas with enlarged sellae turcicae [5]. They also prepared for a simultaneous open craniotomy approach if a tumor had 1 or more of the following characteristics: dumbbell shape; irregular shape with signi cant subfrontal, temporal, or intraventricular extension; previous surgical treatment; suspicious brous consistency; or encasement of the optic apparatus and/or cerebral arteries [5]. Nagata et al. introduced a fully endoscopic combined endonasalsupraorbital keyhole approach for complicated parasellar lesions. They demonstrated that the intraoperative endoscopic view from the supraorbital approach showed the suprasellar pituitary tumor became hyperemic and swollen, with progression of tumor debulking from the transsphenoidal approach [14]. Because neurosurgeons cannot be aware of this phenomenon when using the transsphenoidal route alone, the swelling tumor compresses the surrounding neurovascular structures, resulting in postoperative complications.
In the present study, we demonstrated that pituitary adenomas had high CBVt and increased vascular density. Our results revealed that we could preoperatively identify patients with a high risk of postoperative hemorrhage from residual tumors based on high CBFt by PCT. Conversely, pituitary adenomas with low CBFt, even those with Grade C or D suprasellar extension, could be safely removed via TSS as an initial surgery.
Sakai et al. reported that arterial spin-labeled perfusion images from a 3T magnetic resonance scanner re ected the vascular density of nonfunctioning pituitary macroadenomas, which may be useful in the preoperative prediction of intra-and postoperative tumor hemorrhage [18]. PCT provides reliable information on tumor vasculature [2]. The linear relationship between attenuation changes on computed tomography and tissue concentration of contrast medium, as well as the lack of confounding sensitivity to ow artifacts, allow PCT to potentially offer a more accurate representation of tissue microvasculature than similar magnetic resonance perfusion studies [8]. Additionally, the use of PCT using a 320-row multidetector computed tomography system in the present study overcame the limitation of limited area of coverage by conventional PCT compared with magnetic resonance perfusion imaging.
The present study has limitations that are inherent to its retrospective design. In addition, our study included a relatively small number of cases of large and giant pituitary adenomas.

Conclusions
PCT may be useful in the preoperative prediction of postoperative residual tumor hemorrhage, which can lead to hazardous postoperative complications after TSS for large and giant pituitary adenomas. In

Consent for publication
Not applicable.

Availability of data and material
The datasets analysed during the current study are available from the corresponding author on reasonable request.

Competing interest
The authors declare that they have no competing interests.

Funding
This work has not been funded.
Author Contributions NH designed the study, and collected data with AU. NH, KM, and SD participated in statistical analysis. NH and ME interpreted results, and prepared and drafted the manuscript. All authors read and approved the nal manuscript. Figure 1 Box-and-whisker plots of CBV (a) and CBF (b) The horizontal bars inside of the boxes indicate medians. There were signi cant differences between both CBV and CBF of the tumor and white matter (p < 0.0001).