Patient enrollment
This retrospective study was approved by the hospital’s institutional review board, and informed consent was obtained from all patients at the time of the procedure. From August 2008 to November 2019, all patients from a database of subjects who underwent transcatheter embolization of PAVM in a referral hospital were included in the study. Inclusion criteria were as follows: (1) patients who receive transcatheter embolization of PAVM, (2) patients had an initial and follow-up CT before and after embolization, and (3) treated PAVMs had follow-up confirmative angiography. Patients were excluded if they had complex PAVMs with multiple feeding arteries. Patient enrollment is summarized in Fig. 1. Clinical histories, physical examinations, CT images, and interventional procedural images stored in the Picture Archiving and Communication System (PACS) and electronic medical records were reviewed.
Embolization technique
After local anesthesia with 2 % lidocaine, venous access was obtained through the right common femoral vein. Subsequently, intravenous bolus administration of 3,000–5,000 U of heparin sodium followed. A 6-F guiding catheter (Envoy; Codman Neurovascular, Miami Lakes, Florida) was used in all cases. After right or left pulmonary angiography, the feeding artery was selectively catheterized with a 5-F hydrophilic catheter (Torcon NB Advantage; Cook Medical, Bloomington, Indiana) and coaxial microcatheter (Masters Parkway Soft; Asahi Intecc, Tokyo, Japan), if necessary. The type, size of vascular plugs (AMPLATZER Vascular Plugs [AVPs]; AGA Medical, Plymouth, Minnesota) and coils (Pushable: Tornado, Nester; Cook Medical, Bloomington, Indiana, Detachable: Concerto; Medtronic, Minneapolis, Minnesota), number of coils, and their deployment locations at the feeding artery or venous sac were determined by the operators from the selective angiography findings. In case of feeding artery embolization, the catheter was inserted into the feeding artery and advanced as close to the venous sac as possible. Feeding artery embolization was performed using an AVP oversized by 20–50 % or coils oversized by 20 % in relation to the feeding artery diameter. For venous sac embolization, a microcatheter was advanced into the venous sac via the feeding artery, and detachable coils oversized by 20–50 % in relation to the draining vein were deployed first. After a mesh frame of detachable coils was created to prevent migration, subsequent smaller pushable coils were deployed until there was complete cross-sectional occlusion of the venous sac and distal feeding artery. In all cases, completion digital subtraction angiography was performed after embolization to confirm vessel occlusion.
CT Examination
Initially and as a follow-up exam, all patients underwent contrast-enhanced chest CT, including with multidetector row scanners. To prevent air embolism during the CT scan all PAVM patients were referred to a highly skilled intravenous access team. The region scanned extended from the cervicothoracic junction to the upper abdomen, and images were acquired during a single breath-hold. For the examination, a total 80–100 mL of contrast medium was administered intravenously at a rate of 1.5–2 mL/s. Contiguous thin-section transverse CT images of the area of interest were reconstructed at 2.5-mm slice thickness without intervals by using dedicated soft-tissue kernels.
Follow-up CT was performed at approximately 6 and 12 months after embolization and then every 2 years after that. Rarely, if treatment was not completed in a single session in patients with multiple PAVMs, follow-up was performed earlier to evaluate previously treated lesions prior to additional sessions.
Angiographic confirmation
Follow-up angiography was performed in the following 2 situations. First, in case of the draining vein or venous sac size reduction of the PAVM was insignificant or enhancement in the venous sac persisted in the serial follow-up CT, the patient underwent angiography for suspicion of recanalization. Second, in patients with multiple PAVMs, embolization of all lesions could not be completed in a single session, and previously treated PAVMs were evaluated during subsequent embolization sessions. All pulmonary angiographies were first performed in the right or left pulmonary artery using an injector. Injection rates ranged from 10 to 15 mL/sec for volumes of 20–30 mL per injection, and frame rates were 3 frames per sec. For a more accurate evaluation, selective angiographies were followed at the segmental pulmonary artery with injection rates of 3–5 mL/sec for volumes of 10–15 mL or at its distal levels with careful manual injection.
Recanalization of the PAVM was evaluated during angiography and defined as a case where the feeding artery, venous sac, and draining vein were simultaneously enhanced due to either: (a) restoration of blood flow through a previously treated feeding artery or (b) collateral perfusion via a pulmonary artery to pulmonary artery collaterals.
CT Vessel Diameter analysis
On initial CT images, lobar location, type (simple or complex), and multiplicity of the PAVM were determined. Vessel diameters (draining vein, feeding artery, venous sac) were measured on initial and serial follow-up CT images by 2 radiologists. Measurements were made at the same location on all CT images, and the window level and width were unified to -500 and 2000. The boundary for pulmonary parenchyma was measured after sufficient magnification to clearly identify it. The feeding artery diameter was measured as close to the embolized segment as possible based on follow-up CT. The draining vein diameter was measured just distal to the venous sac before the confluence of adjacent parenchyma, avoiding any metal artifacts caused by the coils in the sac (Gamondes et al. 2016; Kawai et al. 2014). For the venous sac, the cross-sectional diameter of the thickest part perpendicular to the long axis was measured. However, when venous sac embolization was performed, the venous sac diameter could not be measured due to metal artifacts. The reduction rates of all vessel diameters were calculated by comparing the initial and last follow-up CT images using average values of measurements from 2 radiologists.
Data analysis
All PAVMs were divided into occluded and recanalized groups according to the angiographic results. Differences in the initial vessel diameter, vessel size reduction rate, period of angiography and CT follow-up, embolic materials (AVP or coils), and embolization locations (feeding artery or venous sac) in the 2 groups were examined by the t-test or Fisher’s exact test. Interobserver agreement was examined by analyzing the intraclass correlation of data sets of vessel diameter values measured by 2 radiologists. A p value < 0.05 was considered statistically significant. Statistical analysis was performed by using statistical software (Medcalc, version 19.4.1; Mariakerke, Belgium).
The area under the curves (AUC) and optimal cut-off values of vessel size reduction rates for recanalization was obtained by receiver operating characteristic (ROC) curve analysis, and the sensitivity and specificity of the diagnosis was compared to the existing 70 % sac/vein size reduction criteria.