Our ethics committee approved this study (approval number: KEN2211-029) and waived the need for informed consent due to its retrospective nature. However, opt-out consent was obtained for patients’ data use, and written informed consent was secured before treatment.

We included patients who underwent PCA or RAPN as initial treatment for T1a RCC between December 2017 and May 2022. Exclusion criteria were: (i) renal tumor biopsy or angiography without transcatheter arterial embolization (TAE) during the same hospitalization; (ii) PCA for multiple RCCs in one admission; (iii) treatment for other diseases during the same hospitalization; (iv) heparinization before RCC treatment; (v) metastatic lesions; (vi) recurrent RCC; (vii) enrollment in clinical trials; or (viii) insufficient data. Treatment strategies were decided at multidisciplinary team conferences, considering comorbidities, renal function, and patient preferences. PCA was used for patients unsuitable for surgery due to old age, comorbidities, or refusal.

The primary endpoint was the cost of PCA and RAPN. Secondary endpoints included procedure time, hospitalization duration, safety (major complications), and incremental cost-effectiveness (ICER). Costs were assessed from a hospital perspective. Complications were graded per the Clavien–Dindo classification [18], with major complications defined as grade III or higher.

Most PCA patients underwent TAE 1–3 days before ablation by interventional radiologists during the same hospitalization. TAE enhances RCC visualization under CT-fluoroscopy, minimizes the heat sink effect, and reduces bleeding and seeding risks. Indications for TAE were determined by consensus; it was not performed when tumors were easily identified on plain CT, exhibited poor vascularity, or in patients unable to use contrast media due to severe renal impairment or iodine allergy.

Under local anesthesia, a 4 F catheter was inserted via the common femoral artery. After selecting the renal artery, angiography was performed to identify tumor-supplying vessels. Tumor-feeding arteries were selectively catheterized using a 1.5–2.7 F microcatheter. Embolization was performed with iodized oil and absolute ethanol or a gelatin sponge, with coils used when necessary.

PCA was performed using an argon-based cryoablation system (CryoHit® or VISUAL ICE™, Boston Scientific, Marlborough, MA) with 17-gauge cryoprobes (IceRod™, IceSeed™, IceRod 1.5 Plus™, or Ice Sphere™, Boston Scientific) under local or general anesthesia and CT-fluoroscopy guidance in an interventional radiology suite. For RCC near vital organs such as the colon, hydrodissection was achieved by infusing a saline-2% contrast mixture to create a safe distance. The number of cryoprobes was decided by consensus. Typically, three were used for RCCs > 15 mm, and two for small RCCs.

After insertion, cryoablation was performed in two 10–15 min freezing cycles, separated by ≥ 2 min of passive thawing. After each cycle, CT imaging confirmed that the RCC was fully contained within the ice ball, ensuring an ablation margin of at least 6 mm [19]. If coverage was incomplete, additional freeze–thaw cycles were performed after repositioning the cryoprobes.

RAPN was performed by urologists under general anesthesia using the da Vinci Surgical System Si and Xi (Intuitive Surgical, Sunnyvale, CA, USA). A transperitoneal or retroperitoneal approach was chosen based on tumor location, size, and surgical history. Tumors were typically resected with total renal artery clamping under warm ischemia. Pyelopelvic suturing was performed with the urinary tract open; parenchymal sutures and a hemostat were used before cortical closure in most cases.

Cost terms were defined as in Table 1, representing “costs from the hospital perspective”. Total treatment cost was calculated by adding non-reimbursable supplies and medications (e.g., cryoprobes, gas, surgical instruments, disposables, iodized oil, and contrast media for hydrodissection); reimbursable supplies, medications, and transfusion costs (e.g., catheters and contrast media for TAE and other medication during hospitalization); depreciation of surgical and interventional radiology suites (including robotic and cryoablation systems and a unified CT and angiography system); and anesthesia equipment. Personnel costs for the operator, circulating nurses, surgical technologists, and radiological technologists were included. Total income (the hospital payment for each treatment) was calculated by adding procedure fees (based on insurance points), anesthesia fees, other fees (e.g., for transfusions and ICU admissions), reimbursable supplies and medications, and hospital income based on the Diagnosis Procedure Combination (DPC)/Per-Diem Payment Systems (PDPS). The hospital used activity-based costing, reflecting allocation based on specific medical procedures and resource usage.

Data on cryoprobes, catheters, surgical instruments, and procedure time were obtained from surgical records and the Data Warehouse. Gas usage (argon and helium) was sourced from the Medical Engineer Center database (one or two bottles of argon and 0.2 bottles of helium per case). Medications, transfusions, and anesthesia eligible for insurance claims were extracted from the DPC database. Treatment income was calculated per Japanese National Insurance as follows: TAE (code K-6153), 186,200 yen; PCA (code K773-4) at 528,000 yen; and RAPN (code K773-5) at 707,300 yen. Depreciation and labor costs were calculated based on cost price, frequency of use, and procedure time. Pathology fees for RAPN were excluded since biopsies for pathologic diagnosis before PCA were performed during separate admissions, thus excluded from the cost analysis.

Effectiveness was evaluated based on the frequency of complications, following a previous report [16]. Since the complication rate was low in this study, the ICER was calculated using the rates reported in a systematic review [20].

Data on patient age, sex, body mass index, comorbidities (diabetes, hypertension, and cardiovascular disease [CVD]), prior RCC treatment, single kidney status, antithrombotic drug use, tumor size, and R.E.N.A.L. nephrometry score were collected. Patients were classified according to ASA score and Charlson Comorbidity Index (CCI). These factors were statistically compared between the groups.

Descriptive statistics were used to summarize continuous variables (e.g., means and standard deviations) and categorical variables (e.g., frequencies and proportions). Outcomes were compared among three groups: PCA without TAE, PCA with TAE, and RAPN. These outcomes included total income, total cost, income-expenditure balance, procedural income, procedural cost, procedural income-expenditure balance, hospitalization  duration, and procedure time. The costs of PCA with and without TAE were compared to those of RAPN, and the differences and 95% confidence intervals were calculated.

To adjust for potential confounders, we used inverse probability weighted regression adjustment (IPWRA), which combines score weighting and regression adjustment to reduce bias. This method accounts for baseline differences and provides treatment effect estimates similar to those from randomized controlled trials. We used a treatment model to estimate the probability of each patient receiving a particular treatment, based on the covariates listed in Table 2. An outcome model was then used to estimate the outcomes while adjusting for these covariates, as well as single kidney status and hereditary renal cancer. By combining the two models, the IPWRA achieves double robustness, meaning that correct estimates can be obtained even if one of the models is misspecified [21].

If the IPWRA did not converge due to data limitations, we used Regression Adjustment (RA) as an alternative. The RA adjusts for confounding factors by including covariates in the regression model directly, thereby providing adjusted treatment effect estimates. Unlike the IPWRA, the RA lacks double robustness; however, it is computationally more stable and suitable as an alternative.

Similar comparisons were performed between the PCA (with and without TAE) and RAPN groups. Complication rates were compared using the chi-square test. The total cost and procedural costs for cases using two and three cryoprobes in the PCA were compared using Welch’s test. All analyses were conducted using Stata 18/MP4 (Stata Corp, College Station, TX, USA), and a P value of < 0.05 was considered statistically significant. Since this was an exploratory study, no adjustments were made for multiple comparisons [22]. A standardized mean difference (SMD) below 0.10 and a variance ratio (VR) between 4/5 and 5/4 were used to confirm covariate balance assumptions [23, 24].