The overall depiction rate of feeding arteries of CCRCC on 4D-CTA images was favorable (93.3%). Among 11 phases, the depiction rate was the highest (83.3%) in the 5th phase (delay, 23.4 s). Interlobar or arcuate arteries in the renal parenchyma tended to be less visible than segmental or lobar arteries surrounded by renal sinus fat.

Preprocedural CTA contributes to the identification of target vessels, which enables a reduction in contrast volume and fluoroscopy time during the procedure [7, 8, 15]. Although the advantages of preprocedural CTA have been shown in transcatheter arterial chemoembolization (TACE) for hepatocellular carcinoma (HCC) and prostatic artery embolization [7, 8, 15], CTA before TRAE has not been discussed in detail in the literature. Most studies on renal CTA have evaluated renal artery anatomy before surgery, such as partial nephrectomy or renal transplantation [16,17,18]. The target vessels evaluated in those studies were the renal artery trunk or segmental arteries. The detection rate of the segmental-level feeding arteries of RCCs in CTA ranged from 84 to 100% [16,17,18]. However, data about smaller renal artery branches (lobar or more distal-level branches) that require catheterization during selective TRAE procedures are scant. Since the visualization of distal-level feeding arteries should be more challenging because of their small diameter, the overall depiction rate of 93.3%, including those arteries by 4D-CTA, appears promising, suggesting that preprocedural 4D-CTA is useful for planning TRAE.

The 4D-CT images can be obtained using CT systems equipped with multi-row detectors ranging from 128- to 320-rows, which allow continuous or short-interval sequential scanning of a specific region for a period of time [12]. The 4D-CTA method involves a 4D-CT scan while administering contrast media to analyze the detailed vasculature and hemodynamics in various organs and lesions [7, 11,12,13]. In neurovascular settings, Willems et al. showed that 4D-CTA enables highly accurate identification of the feeding arteries of intracranial arteriovenous malformations [13]. In abdominal settings, Albrecht et al. investigated the impact of 4D-CTA before TACE for HCC [7]. They omitted stepwise extensive DSA during TACE procedures by introducing preprocedural 4D-CTA to detect tumor feeders [7]. This change in the workflow helped 4D-CTA to reduce the volume of intra-arterial contrast media by two-thirds and the radiation dose to operators by half during TACE [7]. The authors noted that 4D-CTA ensures sufficient contrast in the vessels owing to the inclusion of multiple time points, unlike conventional CTA which may miss the optimal arterial contrast timing depending on the examiner and patient [7]. The present study supports this hypothesis: renal 4D-CTA enabled post-processing selection of the optimal phase to visualize the feeding arteries of CCRCC.

Because the transit time of contrast media through the renal circulation is short, the renal vein is enhanced soon after the renal artery is enhanced [19]. Furthermore, during this brief period, the renal cortex and veins are intensely enhanced, obscuring the depiction of the small peripheral renal artery branches [9]. Small renal arterial branches cannot always be visualized in the renal artery phase of conventional dynamic CT [20]. In 4D-CTA with high temporal resolution, feeding arteries were favorably identified in specific phases where the CT values between the renal artery and cortex largely differed. Combining multiple phases further improved the identification, indicating that optimal timing for visualizing feeders varies among patients. Therefore, ensuring the imaging at the optimal time for visualizing feeding arteries is considered difficult by conventional CTA in which the delay time is fixed. Bolus tracking might help to optimize the timing of image acquisition and obtain an adequate renal arterial phase [21]. However, the present findings indicate that this is insufficient to reveal small feeding arteries because phases that are ideal and those with maximum arterial enhancement do not necessarily correspond.

Because our 4D-CTA protocol involved 11 sequential scans, we were concerned about high-radiation doses. Therefore, we configured scan parameters to minimize the radiation dose (tube voltage of 100 kV and AEC for tube current with SD35) while maintaining acceptable image quality, thus resulting in a DLP of 527 mGy.cm for 4D-CTA. The DLP value appears acceptable, compared with diagnostic reference levels of abdominopelvic contrast CT (600–1000 mGy.cm) in several countries [22]. Our results showed that the feeding arteries were most obvious during the 4th‒7th phases (delay, 21.3‒27.6 s). Thus, further dose reduction without impairing the ability to reveal feeding arteries might be achieved by limiting 4D-CT image acquisition to this time range.

In the present study population, cryoablation was planned and prepared based on the findings of 4D-CTA images. Treatment was planned in the same way as that based on conventional contrast-enhanced CT findings without a separate examination that causes extra radiation exposure. Thus, 4D-CTA did not confer any disadvantages in this respect.

This study had several limitations. It was a single-center study with a small number of patients. Renal tumors of histological subtypes other than CCRCC were excluded. The included tumors were relatively small in diameter and had a single or dual feeding artery. The present study retrospectively examined whether feeding arteries embolized during the procedure were visible on preprocedural 4D-CTA images. Therefore, a prospective study is needed to thoroughly evaluate the ability of 4D-CTA to detect feeding arteries. This would enable confirmation of whether or not potential feeders found on preprocedural 4D-CTA are true feeders using selective DSA from the corresponding position during the procedure. Because this study was still in the preliminary phase of assessing preprocedural 4D-CTA, it did not include changes in TRAE procedures such as omitting stepwise DSA. Thus, the contribution of 4D-CTA to a reduction in the contrast amount and radiation exposure during TRAE could not be examined. A comparative study is warranted to prove that preprocedural 4D-CTA contributes to such reductions. The area-detector CT system was not the latest version, so we had no access to the most recent deep-learning image reconstruction algorithms. We therefore applied model-based iterative reconstruction algorithms. Deep-learning reconstruction might improve image quality while minimizing radiation exposure [23].