Six PAE users from four institutions, with 4 to 13 years of experience in PAE, performing a total of 300 PAEs per year on average, gathered to discuss their specific techniques and propose a consensus imaging workflow. Reviews were organized for each user to present their imaging workflow and technique, allowing to highlight commonalities and differences and to share best practices, which were then tested by the other authors to feed the following discussions, resulting in an optimized imaging workflow and technique adopted by all centers (Fig. 2). This workflow was established with the objective of better visualizing the anatomy, better planning, guiding and assessing the procedure, while minimizing radiation exposure to patients and clinical staff.
A pre-operative MRI is recommended to confirm BPH, rule out prostate cancer, estimate zonal and total prostatic volumes and plan for PAE treatment, especially for smaller prostates.
To provide a complete overview of the pelvic anatomy, an initial bilateral non-selective fluoroscopic roadmap is recommended from the distal aorta just above the aortic common iliac bifurcation. This roadmap is helpful in understanding internal iliac arteries origin as well as in providing the length of the common and internal iliac arteries to support catheter selection . In our experience, hand-injected fluoroscopic roadmap of 50% diluted contrast provides sufficient image quality at this stage, and decreases radiation exposure compared to DSA.
PAE usually consists of a bilateral successive embolization, often starting from the left side because of the designed use of the Robert’s uterine and the Carnevale’s prostate catheters when performed through the femoral approach. For each side, a first proximal 5 seconds CBCT is acquired with the 5 French catheter in the internal iliac artery, above the bifurcation of the anterior and posterior branches . The following injection parameters are typically used: 22 to 26 cc of pure contrast, injected at 2 cc/s, with an X-ray delay of 6 to 8 seconds. This injection protocol ensures a good filling of the arteries during the entire spin, thus allowing an adequate visualization of both the arterial anatomy and the prostate parenchymal blush in a single CBCT (Fig. 3).
With the 180 degrees rotational DSA-like image provided by the CBCT spin, and with recent studies showing CBCT’s superiority to DSA for PAE planning [11, 12], we recommend skipping the typically acquired ipsilateral oblique DSA from the internal iliac artery  to limit patient radiation exposure, contrast medium use and reducing the procedure time. Another variant to further reduce radiation dose consists of acquiring the proximal CBCT bilaterally from the distal aorta, with 60cc of contrast injected at 6cc/s with an X-ray delay of 6 seconds. In our experience, this CBCT does not provide as much distal information of the prostatic arteries. Based on our experience, and given the lower prevalence of type V cases , we recommend starting straightforwardly with the internal iliac artery CBCT. It provides a rich source of anatomical information with the adequate level of distality and selectivity in most cases. In our practice, CBCTs can be acquired arms down or with arms on the chest with limited impact on image quality thanks to automatic exposure optimization (IGS5/740, GE Healthcare).
Proximal CBCT datasets should be analyzed carefully to identify arteries feeding the prostate, their pathways and non-target vessels. Prostatic arteries should be identified exhaustively and bilaterally to maximize treatment completion and reduce symptom recurrence risks . Advanced planning software such as Virtual Injection (Embo ASSIST, GE Healthcare) allows to simulate selective injections based on a proximal CBCT, facilitating the identification of prostatic arteries and non-target vessels (Fig. 4). Automatic segmentation of the pelvic and prostatic vasculature and of the arteries of interest, both prostate feeders and non-target vessels, along with their centerlines, facilitates procedure planning from table side and creates a 3D model for augmented fluoroscopy (Fig. 4).
During microcatheter navigation (≤ 2.4 French recommended), this 3D model is overlaid on the live fluoroscopy, providing a 3D roadmap that is and remains automatically registered to the patient despite gantry and table movements. In case of patient motion, the 3D roadmap registration can be adjusted from table side to better match the patient. Depending on the procedure stage, both the prostatic artery to select and the non-target arteries to avoid can be interesting to display on fluoroscopy. In addition to its benefits facilitating catheter guidance, the availability of the 3D roadmap allows to select the best working angulations to visualize vessel turns and bifurcations without using fluoroscopy thus with no radiation. In our experience, the 45 degree ipsilateral angle typically recommended should replaced by a systemic identification of the optimal working angle based on patient’s anatomy using 3D roadmap without fluoroscopy, which may be less steep than 45 degrees and thus more radiation effective. Multiple DSA runs can thus be avoided by using the CBCT-based 3D roadmap, resulting in both dose and contrast savings.
During microcatheter navigation, basic ALARA radiation best practices should also be enforced to minimize radiation exposure both to patients and medical staff: use of digital zoom at maximum FOV instead of magnification; rigorous collimation; low default dose settings and frame rates, e.g. 1 fps for DSA and 3.75 fps for fluoro, which can be increased when needed from table side; fluoroscopy storing instead of DSA when sufficient in terms of image quality; and avoiding steep angulations when possible to minimize scatter radiation and optimize image quality.
Once the microcatheter is in the desired location within the prostatic artery, a distal DSA in ipsilateral oblique view with 3-5cc of 50–70% contrast hand-injected or power-injected at 0.2 to 1 cc/s (depending on the prostate size, prostatic artery diameter and collaterals) is recommended to confirm no non-target vessels are being perfused, as well as to simulate the embolization treatment by visualizing the prostate perfusion. When further confirmation is needed, e.g. to clearly distinguish prostatic from rectal branches, or if both central gland and peripheral zone arterial branches are not observed, or for early experience PAE users, a distal 5 second CBCT is recommended with either hand or power injection with an 8–10 second delay to ensure good filling of both the prostate and any extra-prostatic structure (Fig. 5). In our experience, hand injection gives better control to fill the prostate, avoid reflux and obtain a strong injection to identify shunts. For hand injected CBCTs, half of the contrast is injected before the spin starts to optimize contrast uptake in the prostate and surrounding organs; the other half is injected during the spin rotation to confirm the local arterial anatomy and help evaluate risks of reflux in non target anatomies. Operators should not be present in the examination room during CBCT acquisition. If required due to manual injection preferrence, operators should stand behind lead shields to minimize their own radiation exposure. To avoid operator presence in examination room, power injection with PSI at 300 can help avoid reflux during CBCT acquisition.
Intra-arterial vasodilator should be used when navigating stenotic and/or tortuous anatomies, before DSA/CBCT acquisition and before embolic agents injection, with the aim of opening the prostatic vascular bed. Isosorbide mononitrate (5-10mg) and nitroglycerin (100µg) are typically used. Prostate enhancement can be better observed after vasodilator injection. Some shunts, mostly common to the internal pudendal artery, rectum and bladder, can also be more evident after vasodilator injection.
Once distal DSA/CBCT analysis confirms the microcatheter location for treatment, the embolic material is delivered proximally first, then deeper in the gland, following the PErFecTED technique [7, 21]. Postero-anterior incidence can be used for types I, II and III with the aim of reducing radiation exposure because they are longer and there is less risk of embolic agent reflux in these types of arteries. Ipsilateral oblique view is recommended for type IV because of the risk of proximal embolic agent reflux to the internal pudendal artery. Microspheres ranging from 100–500 µm have been used. Adverse events and complications seem to be more frequent with smaller particles due to deeper penetration and passing through anastomosis [7, 37]. During this time-consuming step of embolic agent injection, low dose fluoroscopy at 3.75fps can be used to monitor the embolization, in unsubtracted or subtracted mode, and stored for documentation purposes. Intermittent 0.5fps DSAs can be used as necessary. Collimation, digital zoom instead of magnification, and use of postero-anterior views are particularly recommended to reduce radiation in this step. A final 0.5fps DSA or stored fluoroscopy allows to control the success of the embolization, defined as total stasis. The catheter is then navigated to the other side where the workflow is repeated.