**1. Introduction**

Arteriovenous malformations (AVMs) in patients with Hereditary Hemorrhagic Telangiectasia (HHT; Osler's disease) are malformations in which arteries and veins are directly connected, due to the absence of intervening capillaries [1]. The most common clinical symptoms are spontaneous and recurrent epistaxis, as well as Telangiectasias (small AVMs) on the lips, tongue, buccal mucosa, face, chest, and fingers [2]. Larger AVMs become symptomatic in the lungs, liver, gastrointestinal tract, or brain; thus, complications from severe bleeding or shunting with possible consecutive cerebrovascular incidents may occur. Pulmonary arteriovenous malformations (PAVMs) are defined as pathologic communications between pulmonary arteries and pulmonary veins, resulting in a right-to-left

 Schneider, G.; Massmann, A.; Fries, P.; Frenzel, F.; Buecker, A.; Raczeck, P. Safety of Catheter Embolization of Pulmonary Arteriovenous Malformations— Evaluation of Possible Cerebrovascular Embolism after Catheter Embolization of Pulmonary Arteriovenous Malformations in Patients with Hereditary Hemorrhagic Telangiectasia/Osler Disease by Preand Post-Interventional DWI. *J. Clin. Med.* **2021**, *10*, 887. https://doi.org/ 10.3390/jcm10040887

Academic Editors: Hans-Jurgen Mager, Carmelo Bernabeu and Marco Post

Received: 18 January 2021 Accepted: 17 February 2021 Published: 22 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

shunt [3,4]. Larger shunts may result in hypoxemia manifesting with dyspnea, potentially increasing the risk of paradoxical cerebral embolization [5], and, in consequence, the risk of increased morbidity and mortality. Of the approaches to treating patients with PAVMs, catheter embolization, either with coils or vascular plugs [6], is considered the treatment of choice because of its high success rate and reduced invasiveness compared to lung surgery [7,8], and because embolization more favorably respects the unaffected lung parenchyma compared to surgical resection [9].

Reperfusion or recanalization of initially successfully treated PAVMs is the most common cause of recurrence after coil embolization [10–12]. However, interventionalists can minimize the risk of reperfusion by using dense "packing" techniques that result in the complete cross-sectional occlusion of feeding arteries [13]. Thus, PAVM embolization with Amplatzer vascular plugs (AVP) has been shown to achieve relatively high mid-term success rates in terms of recurrence or recanalization, even in bilateral treatment [14–17]. In general, a low reperfusion rate is noted in the long-term due to late re-opening [6,18].

The incidence of stroke in patients with HHT ranges between 9 and 18% [19,20]. Although the occurrence of clinically conspicuous stroke seems to be lower in patients with PAVMs treated with embolization therapy than in patients with untreated, persistent PAVMs [19–22], little is ye<sup>t</sup> known about the rate of clinically inconspicuous ischemic brain lesions associated with PAVMs. In contrast, procedure-associated, clinically inconspicuous ischemic brain lesions are common in up to 40% of patients undergoing supra-aortal endovascular procedures or neurovascular interventions, such as carotid stenting or endarterectomy [23,24].

To our knowledge, no data are ye<sup>t</sup> available on the incidence of peri-interventional cerebral ischemia occurring during catheter-based embolization of PAVMs. Therefore, the aim of our investigational study was to prospectively evaluate the incidence of periinterventional cerebrovascular incidents in patients with HHT referred for catheter-based embolization of PAVMs.

#### **2. Materials and Methods**

## *2.1. Patients*

This single-center, prospective study observational was approved by the institutional review board. Written informed consent for both catheter-based embolization and the use of imaging data was obtained from all patients or legal guardians.

All patients included in the study suffered from HHT, confirmed either by genetic testing or, in most cases, based on Curaçao criteria [1,25]. Independently of clinical presentation and symptoms, each included patient had at least one PAVM with a feeding artery diameter of at least 2 mm diagnosed by contrast-enhanced MR angiography (CE-MRA), and in a few cases, CT imaging.

Patients were ineligible for inclusion if they had a severe allergy to iodine contrast agents, significantly impaired renal function (GFR < 15 mL/min), and/or severely impaired blood coagulation (INR > 2) or platelet count (<50.000/dL). Likewise, patients were ineligible for inclusion if they were contraindicated for MRI (e.g., for implanted cardiac pacemakers).

#### *2.2. Embolization Technique*

Access through the right common femoral vein was obtained after local anesthesia of the groin region. A 7F sheath was inserted and the common pulmonary artery was probed with the help of a 5F pigtail catheter and a bentson guidewire. Diagnostic pulmonary angiograms were performed to locate and visualize the PAVMs. Afterwards, using a Rosen guidewire, a Cook White Lumax guiding catheter (Cook Medical) or a coaxial system consisting of a Neuron 6F Long Sheath and a Neuron 6F Select Catheter (Penumbra) was inserted, and selective catheterization of the segmental and subsegmental pulmonary artery feeding the PAVM was performed, using the coaxial system. Guidewires which might perforate the aneurysm sac were avoided. The number and diameters of the feeding

arteries of the evaluated PAVMs were identified after contrast medium injection. The PAVMs were classified as simple or complex based on the number of feeding arteries, as described elsewhere [26,27].

Depending on the size of the main feeding artery and the anatomical situation, either Nester-Coils (Cook, USA) or Amplatzer vascular plugs II/IV (St. Jude Medical) were used for embolization. These were introduced through a guiding catheter of appropriate size under a water seal. The diameter of the device was chosen to be approximately 30% larger than the size of the main feeding artery. The plug or coil was then placed as distally as possible in the feeding artery with sparing of the PAVM itself.

The choice of embolization device was made according to the length of the available landing zone, which is the distance between the PAVM and the first proximal pulmonary subsegmental artery. In the case of amplatzer vascular plugs, the position of the device was checked by Digital Subtraction Angiography (DSA) immediately after device placement. If the position of the device was deemed adequate and satisfactory, the device was released; otherwise, the device was retrieved and repositioned as necessary.

Post-embolization angiography was performed after satisfactory device placement to confirm the total occlusion of the PAVM.

Immediately before the procedure, each patient received IV injection of 2500 IE Heparin. The number of PAVMs treated in each patient ranged from one to eight, either treated in one intervention or across multiple interventions, depending on the duration and complexityoftheprocedureaswellasthe patient'sgeneralconditionofcompliance

#### *2.3. Pre- and Post-Interventional Pulmonary MRI*

during angiography.

Pulmonary CE-MRA to evaluate PAVMs before and after intervention was performed on a 1.5 Tesla (T) magne<sup>t</sup> (Magnetom Aera, Siemens Medical Systems, Erlangen, Germany) with a 16-channel phased-array coil. The imaging protocol consisted of dynamic, timeresolved, contrast-enhanced MRA, and high-resolution, pulmonary arterial- and early venous-phase, contrast-enhanced MRA sequences.

Time-resolved MRA was performed after injection of a small contrast bolus (0.025 mmol/kg of gadobenate dimeglumine [MultiHance ™, Bracco] or 0.05 mmol/kg of gadoteridol [ProHance ™, Bracco]). The sequence parameters were as follows: repetition time/echo time (TR/TE) = 2.7/1.0 ms, average field of view = 40 × 29 cm, slice thickness = 1.5 mm, 140–160 slices, BW = ±113 kHz. The temporal resolution of the sequence was 3 sec/dataset with a total of 72 slices. *k*-space sampling was performed via key-hole imaging (TWIST). The true spatial resolution was 1.2 × 1.2 × 1.5 mm3, which was interpolated to 0.7 × 0.7 × 1.0 mm<sup>3</sup> by zero-filling.

High-resolution, contrast-enhanced Angio 3D MRA was then performed using the timings established in the time-resolved study. Initially, breath-hold, non-contrast enhanced, T1-weighted, spoiled gradient recalled echo (FLASH 3D) images were acquired. The sequence parameters were as follows: TR/TE = 2.81/1.07 ms, average field of view = 40 × 29 cm, slice thickness = 1.3 mm, 140–160 slices, BW 540 kHz. The temporal resolution of the sequence was 2.2 s/dataset with a total number of up to 160 slices. The true spatial resolution was 1.3 × 1.3 × 1.5 mm3, which was interpolated to 1.1 × 1.1 × 1.3 mm<sup>3</sup> by zero-filling. Thereafter, the identical FLASH 3D sequence was repeated after injection of 0.075 mmol/kg of gadobenate dimeglumine or 0.15 mmol/kg of gadoteridol at a flow rate of 2 mL/s at end-inspiration, followed by a flush of 30 mL normal saline [28]. The scan time varied depending on patient size and the number of slices required. Likewise, the acquisition time varied with the size of the patient and the number of phase-encoded steps needed to maintain resolution. Iterative reconstruction was applied to provide an effective acceleration factor of approximately 4.0, which also varied slightly depending on the number of slices.

The first acquisition was the arterial phase of the pulmonary circulation, and possible shunts between the bronchial arteries and pulmonary veins were also visualized during this phase. Subsequently, a second full acquisition was performed in which normal pulmonary veins were visible. For all acquisitions, patients were instructed to hold their breath at end-inspiration. The total acquisition time for the entire MRA protocol ranged between 5 and 6 min. All examinations were performed as part of the daily clinical routine. Follow-up of all interventional procedures by means of CE-MRA was performed routinely, first at 3 months post-intervention and then at yearly intervals.

#### *2.4. Cerebral MRI*

Cerebral MRI was performed immediately before the embolization, as well as at the 4 h and 3-month post-embolization therapy. For detection of peri-interventional cerebral ischemic lesions, T2w imaging (T2-Turbo Spin Echo [TSE], slice thickness 3 mm, TR = 5000 ms, TE = 92 ms, BW 191 kHz) and Diffusion Weighted Imaging (DWI; echo planar imaging sequence, slice thickness 5 mm, TR = 6300 ms, TE= 89 ms, BW: 1132 kHz) using three different b-values (b = 0, 400, 800) with calculation of ADC maps were performed. Any new lesion occurring between the pre- and post-interventional cerebral MRI scans on either sequence was considered a new cerebrovascular incident associated with the intervention. Additional pre- and post-contrast T1-weighted TSE, FLAIR, and susceptibility weighted sequences were acquired as part of the initial screening to rule out cerebral AVMs, including micro-AVMs.

#### *2.5. Statistical Analysis*

The characteristics of all participants were transcribed into software (Excel, version2011; Microsoft, Redmond, Wash) for subsequent analysis. Central tendency was measured by the mean, while range and standard deviation were used to measure the dispersion of data.
