Background Preoperative embolization of hypervascular brain tumors is frequently used to minimize intraoperative bleeding.
Objective To explore the efficacy of embolization using flat-detector CT (FDCT) parenchymal blood volume (PBV) maps before and after the intervention.
Materials and methods Twenty-five patients with hypervascular brain tumors prospectively received pre- and postprocedural FDCT PBV scans using a biplane system under a protocol approved by the institutional research ethics committee. Semiquantitative analysis, based on region of interest measurements of the pre- and post-embolization PBV maps, operating time, and blood loss, was performed to assess the feasibility of PBV maps in detecting the perfusion deficit and to evaluate the efficacy of embolization.
Results Preoperative embolization was successful in 18 patients. The relative PBV decreased significantly from 3.98±1.41 before embolization to 2.10±2.00 after embolization. Seventeen patients underwent surgical removal of tumors 24 hours after embolization. The post-embolic tumor perfusion index correlated significantly with blood loss (ρ=0.55) and operating time (ρ=0.60).
Conclusions FDCT PBV mapping is a useful method for evaluating the perfusion of hypervascular brain tumors and the efficacy of embolization. It can be used as a supplement to CT perfusion, MRI, and DSA in the evaluation of tumor embolization.
- CT perfusion
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Preoperative embolization of hypervascular tumors is frequently performed to minimize intraoperative bleeding.1–3 Meningiomas, hemangiopericytomas, and hemangioblastomas are the most common hypervascular tumors embolized preoperatively.1 ,3 ,4 Many of these hypervascular tumors have multiple blood supplies; preoperative embolization is usually difficult when attempting to achieve complete devascularization. Therefore, it is crucial to define the degree of embolization and the extent of ischemic areas in the tumor. DSA, CT perfusion, and MRI have been used to visualize the post-embolization appearance of the tumor blood supply and to assess the efficacy of embolization.5–8
Angiography systems equipped with flat-detector CT (FDCT) have emerged in recent years, and they provide high-quality, high-resolution CT-like images and also functional information, such as parenchymal blood volume (PBV) perfusion imaging.9–11 Although this newly developed technique has already been used in many cerebrovascular diseases,12–16 few studies have reported evaluation of hypervascular brain tumor embolization. Therefore, the purpose of this study was to evaluate the feasibility of FDCT PBV mapping in assessing the efficacy of hypervascular brain tumor embolization and characterizing post-embolization tumors.
Materials and methods
This prospective study was approved by the institutional research ethics committee, and informed consent was obtained from all the patients.
The inclusion criteria were the following: (1) presence of tumors with clear contrast enhancement on MRI, which could indicate an abundant blood supply; and (2) therapeutic need for embolization to reduce significant bleeding during surgery. Patients with renal impairment, pregnancy, allergy to iodinated contrast or other contraindication to DSA examination were excluded.
Twenty-five patients were recruited for this study in our institution between March 2014 and April 2015. Seven patients with meningiomas did not receive embolization owing to little blood supply from the external carotid artery, based on arterial angiography. A total of 18 patients (10 male and 8 female), aged between 22 and 65 years (mean age 49.67±12.44), underwent embolization and PBV examinations before and after embolization. Of these cases, 15 were meningiomas, two were hemangiopericytomas, and one was a hemangioblastoma (table 1).
DSA acquisition and semiquantitative FDCT PBV analysis
Angiographic examinations were performed to evaluate the vascular supply, using a biplane system (Artis Zee Biplane, Siemens Healthcare, Forchheim, Germany). FDCT PBV scans were acquired before and after embolization in the neurointerventional suite. A 5 F pigtail catheter was inserted at the aortic root. A 3D mask run was acquired first; then, the C-arm was rotated back to the initial position, and contrast material (iodixanol (Visipaque) 320 mg I/mL, GE Healthcare, Ireland) began to be injected through the catheter. To ensure that the contrast filling in the brain tissue reached a steady state, a second 3D fill run was triggered manually by monitoring the low frequency (2 frames per second) angiogram until superior sagittal sinus filling was observed. For both the mask and fill runs, the C-arm was rotated 200° in 6 s, with an angle increment of 0.5°. A total of 80 ml (dilution 50% with saline) of contrast material was delivered to the whole brain under an injection pressure of 600 pounds per square inch (psi) within 10 s at a rate of 8 mL/s.
Postprocessing of the 3D data to generate color-coded PBV maps was performed using commercial software (syngo NeuroPBV, Siemens Healthcare) on a clinical workstation (syngo X workplace, Siemens Healthcare). Three regions of interest (ROIs) were drawn on the pre- and post-embolized colorimetric perfusion maps for each patient. First, two freehand ROIs were delimited to the area of the maximal tumor diameter on axial and sagittal slices of pre-embolized DynaCT. The third ROI was delimited on healthy brain tissue as a reference region. Then, the three ROIs were converted on the corresponding slices of pre-embolized and post-embolized perfusion maps (figure 1). The PBV values were expressed as mL/1000 mL of cerebral tissue. The slice-wise relative PBV (rPBV) was expressed as follows: 1and the tumor perfusion index was calculated as follows: rPBV×tumor volume.
To evaluate the additional radiation dose exposure associated with the added FDCT PBV acquisition, the dose reports were archived and analyzed for all the patients.
The indication for embolization was based on a combination of tumor factors, such as the size, location, and vascular supply on angiography. These factors were determined by the surgeon and the neurointerventionalists subjectively, without reference to well-defined, standardized criteria.
After diagnostic angiography, the pigtail catheter was exchanged for an embolization catheter, which was inserted into the proper position at the feeding vessels through the same groin access. Embolization was performed using a pulse spray technique under a reverse real-time roadmap. The embolic agents used were polyvinyl alcohol particles, Gelfoam and Onyx. Embolization was complete when there was substantial cessation of forward flow or contrast stasis without reflux in the feeding vessel.
Patients who underwent perioperative embolization were randomly assigned to different operators for surgical resection within 24 hours. The operative data collected were the following: operative time, blood loss during the operation, and pathological features.
Data analysis was performed using IBM SPSS software, V.19.0.0, and the criterion for statistical significance was set at p≤0.05. Differences between pre- and post-embolization measurements of the rPBV were compared using the paired t-test. Correlations among blood loss, operating time and the post-embolic tumor perfusion index were determined using Spearman's correlation coefficients.
Preoperative embolization succeeded in all the 18 patients. The rPBV was decreased significantly from 3.98±1.41 before embolization to 2.10±2.00 after embolization (table 2).
After embolization, tumors were surgically removed from all patients except for one (patient 13), who had a petroclival meningioma and developed coma with contralateral limb paralysis after embolization. The remaining 17 patients were successfully treated. The blood loss and operating time correlated well with the post-embolic tumor perfusion index (table 3).
Thirteen patients (72%) with multiple feeders were not completely embolized. The post-embolization PBV maps provided the surgeons with a clear 3D view of the tumor spatial distribution, which still had a rich blood supply (figures 2 and 3). Complete embolization was performed in the remaining two patients (11%) with multiple feeders and three (17%) with a single feeder, resulting in obviously low perfusion on the postprocedural PBV maps (figure 1).
The recorded average dose–area product at the interventional reference point of one PBV acquisition was 10.25±0.34 mGy.cm2 and the effective dose (ED) was equivalent to 4.1 mSv, according to the Monte-Carlo conversion coefficient.17
Preoperative embolization of hypervascular brain tumors has been shown to be efficacious in reducing massive intraoperative bleeding and shortening surgical time as resection occurs safely without injury to adjacent brain tissue.2 ,18 However, with embolization it is difficult to achieve complete devascularization owing to the characteristics of multiple vessel supply to hypervascular tumors,19 ,20 especially when brain tumors are fed by the internal carotid artery. Although it is important to depict the degree of embolization and the possible residual blood supply in the tumor, accurate preoperative evaluation of the efficacy of embolization remains difficult. Conventional 2D DSA, perfusion CT, and perfusion-weighted MRI are commonly used to evaluate the efficacy of embolization.8 ,21 2D DSA can be useful for demonstrating the feeding vessels to the tumor, the site of tumor attachment, the patency of the sinuses, and arterial encasement. However, the study by Grand et al 22 showed that the degree of vascular embolization did not reflect tumor perfusion accurately. Some tumors remained well vascularized during surgery, which was not reflected during embolization. This discrepancy was difficult to differentiate with 2D DSA. Perfusion CT and perfusion-weighted MRI provided a better evaluation of the vascularization and degree of embolization,5 ,8 ,21 ,23 ,24 but the extremely long examination time and additional patient transportation have limited their widespread use. They are time-consuming, have also reduced patient safety, and increase the doses of radiation and contrast medium.
Like the development of FDCT, PBV perfusion imaging emerged in recent years and has been used in many ischemic cerebrovascular diseases.11 ,14 ,16 ,25 Similar to perfusion CT, FDCT PBV maps are calculated using time–density curve analysis from a three-dimensional rotational DSA during a steady-state contrast concentration in the cerebral vasculature, and PBV is used as an index to assess the perfusion state of the brain tissue.9 ,15 This new method offers semiquantitative measurement of PBV because the PBV technique has been validated in a number of clinical studies and it has been proved that PBV values correlate well with those from perfusion CT and MRI.15 ,16 ,26–28
We applied this new technique to evaluate the embolization of hypervascular brain tumors. First, we confirmed that the PBV maps could reflect tumor perfusion alteration accurately before and after embolization. After successful embolization, the post-embolization rPBV was significantly reduced, compared with the pre-embolization rPBV. With the decreased tumor perfusion index, the blood loss was significantly reduced. The post-embolic tumor perfusion index could be used to predict the blood loss during tumor resection and could direct the surgeon towards preoperational preparation of blood transfusion.
We also found that a lower post-embolic perfusion index resulted in a shorter operating time, which indicated that surgical resection benefited from better embolization. Second, the PBV maps provided intuitive information about the blood supply alteration inside the tumor caused by embolization. The area in which effective embolization occurred inside the tumor showed a significantly decreased perfusion status, which was defined by lower PBV values. In addition, semiquantitative perfusion analysis could assess the contribution of different feeding arteries to the tumor area by comparing the rPBV values before and after embolization. Selective embolization of certain feeding arteries would reduce the rPBV of this area (figure 2). Surgeons could take advantage of this information in surgical planning; they could predict in advance where the tumor would have massive bleeding and where little bleeding that could be quickly resected safely. The whole brain volume reconstruction of PBV could visualize any slice of perfusion imaging on transversal, sagittal, and coronary views; therefore, surgeons could identify the degree of blood supply in the tumor on multiplanar PBV maps.
In this study, although embolization reduced the tumor blood perfusion in 87% of patients with multiple feeders, the supply area of unembolized vessels was still highly perfused, which could be clearly seen on multiplanar PBV maps. Moreover, the fusion of PBV maps and angiographic images demonstrated clearly the feeding arteries to highly perfused regions (figure 3). According to the fusion images, surgeons could recognize the feeding arteries and block the residual tumor's blood supply during the early period of the operation to reduce intraoperative hemorrhage. Finally, FDCT PBV could be performed immediately and conveniently before and after embolization in the interventional suite so that the efficacy of embolization could be assessed when the patient was still on the operating table; inadequate embolization and decisions about further embolization could be made immediately after the assessment of FDCT PBV maps, thus avoiding the transportation of patients, saving time, and keeping surgical resection on schedule.
We also observed the ED, which was caused by FDCT PBV acquisition. It was similar to the published ED values of 3.6–5 mSv for perfusion CT imaging.29 ,30 However, PBV determination provided both perfusion imaging and angiographic information with a single simultaneous acquisition. For angiographic CT imaging, a second acquisition is needed, resulting in an additional ED of 3–5.5 mSv and 50–60 mL of contrast medium.16 We believe that the FDCT PBV maps have advantages over perfusion CT because they have a similar ED but less contrast is needed, and this method is more informative than perfusion CT imaging.
We also compared our IA aortic arch-injected FDCT PBV acquisition with IV injected FDCT PBV acquisition reported in other studies.26 ,28 The IV protocol typically requires injection of 80 mL of contrast agent but only a small amount circulates to the brain. In contrast, the IA injection requires only 40 mL of contrast agent, potentially reducing contrast-induced complications. Moreover, the IV protocol uses a prolonged, slow contrast injection and requires more time to reach the maximal opacification of the superior sagittal sinus. It results in a higher ED than our IA protocol. Therefore, our IA FDCT PBV protocol allowed for substantial dose reductions of the contrast agent and radiation exposure compared with the IV injection protocol, and was safer for the patients.
This study had some limitations. First, the IA FDCT PBV protocols required correct positioning of the angiographic catheter in the aortic arch and manual triggering of data acquisition when the contrast material reached the assumed ‘ideal steady state’ in the cerebral vasculature. The discrepancy caused by process variation could result in deviation of the PBV maps. However, an experienced operator would be able to avoid or reduce to a minimum this discrepancy. Second, between-subject differences might also have influenced the results. Patient individual factors, including age, sex, weight, cardiac output, vessel narrowing, cerebral microvascular dysfunction, etc, could lead to asymmetry of bilateral cerebral perfusion. The rPBV values were semiquantitative relative to the normal side. The abnormal perfusion of the other hemisphere would affect the accuracy of the rPBV values. Third, surgical resection was performed by random operators, which might explain the imperfect correlation (ρ=0.55–0.60) between the perfusion index and the operative data. A skilled surgeon would spend less time resecting the tumor with less blood loss. Also, the blood loss between tumors at different locations might not be the same. In a future study, recruitment of patients from a single surgeon and tumors at the same location might render the results even more robust.
FDCT PBV maps can relay information about the perfusion status of hypervascular brain tumors and the degree of embolization, and these maps can be used as a supplement in the evaluation of the efficacy of tumor embolization and the resection of tumors. It could serve as a helpful tool for surgical planning to program the precise location and thus enhance safety and improve patient outcomes.
Contributors H-DW and XZ contributed equally to the article. Study concepts: L-LW, XZ, H-DW. Study design: L-LW, XZ, H-DW. Data acquisition: Q-RZ, QW, S-JC, J-LD. Quality control of data and algorithms: J-LD, L-LW. Data analysis and interpretation: Q-RZ, QW. Statistical analysis: KH, S-JC. Manuscript preparation: L-LW, XZ, H-DW. Manuscript editing: KH. Manuscript review: XZ, H-DW.
Funding This work was supported by the National Natural Science Foundation of China (grant number 81471183) and by a grant from Jinling Hospital (grant number 2016006).
Competing interests None declared.
Patient consent Obtained.
Ethics approval This study was approved by the ethical committee of the Jinling Hospital, and the patients provided written informed consent before beginning treatment.
Provenance and peer review Not commissioned; externally peer reviewed.