Adaptive radiofrequency ablation using an Octopus multipurpose electrode for small hepatocellular carcinoma adjacent to critical structures: a prospective single-center pilot feasibility study

Abstract

Purpose

This study aimed to assess the feasibility, safety, and preliminary oncologic outcomes of an adaptive radiofrequency ablation (RFA) strategy using the Octopus multipurpose (Octopus MP) electrode for small hepatocellular carcinoma (HCC) in anatomically challenging locations.

Methods

This prospective, single-center cohort study enrolled 76 patients with 83 small HCCs (≤3 cm) in at-risk locations, including tumors adjacent to heat-sensitive organs, central bile ducts, or major vessels and tumors in capsule-adjacent or subphrenic locations. The Octopus MP electrode incorporates integrated temperature monitoring and a central 19-gauge multipurpose needle for fluid instillation and real-time temperature sensing. An anatomy-driven protocol allocated tumors to combined percutaneous ethanol injection plus RFA (PEI-RFA) or temperature-monitored RFA (TM-RFA). The primary endpoint was the thermal injury rate, and secondary endpoints included technical success, technique efficacy, and local tumor progression (LTP).

Results

The per-protocol analysis included 45 patients with 51 tumors in the PEI-RFA arm and 31 patients with 32 tumors in the TM-RFA arm. Technical success and 1-month technique efficacy were 100% in both arms. No major complications occurred. Thermal injury was observed in three of 45 patients (6.7%) in the PEI-RFA arm and two of 31 patients (6.5%) in the TM-RFA arm; all events were minor and self-limited. Late clinically insignificant biliary strictures occurred in four of 45 patients (8.9%) in the PEI-RFA arm and one of 31 patients (3.2%) in the TM-RFA arm. Patient-based cumulative LTP rates at 12, 24, and 36 months were 4.44%, 7.13%, and 10.95%, respectively, in the PEI-RFA arm and 3.23%, 3.23%, and 7.41%, respectively, in the TM-RFA arm.

Conclusion

An anatomy-driven adaptive strategy using the Octopus MP electrode, which integrates established adjunctive ablation techniques within a single device, was feasible for small HCCs near critical structures and was associated with favorable safety and mid-term local control.

Graphical abstract

Introduction

Radiofrequency ablation (RFA) is an established curative treatment for small hepatocellular carcinoma (HCC; ≤3 cm), but conventional RFA has inherent limitations, including higher local tumor progression (LTP) rates than surgical resection and the risk of thermal injury to adjacent structures [1–4]. Although collateral thermal injury is rare, it has been reported during ablation of tumors abutting the diaphragm or heat-sensitive organs, including the colon, duodenum, bile duct, and gallbladder. In contrast, tumors adjacent to major vessels (≥3 mm) are limited primarily by the heat-sink effect, which can cause suboptimal ablation and a higher risk of LTP rather than thermal injury [2,5–8]. These anatomically challenging locations increase procedural difficulty and thermal injury risk [9], underscoring the need for techniques that maximize ablation efficacy while protecting surrounding structures.

Anatomically at-risk tumors can be grouped according to the dominant mechanism that limits conventional RFA: (1) oncologic risk from heat-sink effects in perivascular tumors [10]; (2) thermal risk in tumors abutting or near heat-sensitive structures, including the gallbladder, bowel, or bile ducts [5]; and (3) mechanical or access-related risk in capsule-adjacent or subphrenic tumors, where electrode placement and margin creation are constrained and where bleeding or tumor seeding may be a concern [5–7,11]. Because these mechanisms differ, a tailored strategy may be preferable to a uniform approach.

Adjunctive techniques, including artificial ascites, artificial pleural effusion, and cooling by endoscopic nasobiliary drainage, have been developed to improve the sonographic window and protect adjacent vital organs from RFA-induced thermal injury [12]. Nevertheless, device-based strategies that integrate real-time safety feedback and chemical augmentation may further improve feasibility in anatomically at-risk settings. The Octopus multipurpose (Octopus MP) electrode system combines two internally cooled radiofrequency (RF) electrodes with adjustable active tips and a central 19-gauge multipurpose needle, consisting of a multi-perforated applicator with an integrated thermocouple, for fluid instillation and real-time tissue temperature sensing [13]. Both functions are provided by the same single needle: fluid, such as saline or ethanol, is delivered through the side perforations, whereas the integrated distal thermocouple enables continuous real-time tissue temperature monitoring during RF energy delivery without requiring a separately inserted thermometric device. This integrated design enables two distinct therapeutic approaches: temperature-monitored RFA (TM-RFA), in which energy delivery is modulated according to the temperature at a critical interface [13], and combined percutaneous ethanol injection plus RFA (PEI-RFA), in which ethanol is used to augment ablation while allowing reduced RF energy delivery near vulnerable structures [14–16].

Therefore, this prospective cohort study evaluated the safety, defined primarily by thermal injury, and oncologic outcomes of an adaptive strategy using the Octopus MP electrode for small HCCs in anatomically at-risk locations.

Materials and Methods

Compliance with Ethical Standards

This prospective study was approved by the Institutional Review Board of Seoul National University Hospital (No. 2208-116-1351) and was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants before enrollment. The study was registered at ClinicalTrials.gov (NCT05666271).

Study Design and Patients

This prospective, single-center exploratory cohort study used protocol-guided, anatomy-driven treatment allocation and enrolled patients between August 2022 and March 2023. Approximately 200 consecutive referrals for RFA were screened, and 76 patients with 83 HCCs meeting the eligibility criteria were enrolled (Fig. 1).

Fig. 1. Study flowchart.

Study flow diagram illustrates patient enrollment, intervention allocation, procedural crossover, and outcomes. Of 200 consecutive referrals screened, 76 patients with 83 tumors met the eligibility criteria and were enrolled. Based on dominant anatomical risk and technical feasibility, 48 patients were initially allocated to combined percutaneous ethanol injection plus radiofrequency ablation (PEI-RFA) and 28 to temperature-monitored radiofrequency ablation (TM-RFA). Three patients crossed over from the PEI-RFA plan to TM-RFA because the 19-gauge multipurpose needle could not be placed, yielding a per-protocol cohort of 45 patients with 51 tumors in the PEI-RFA arm and 31 patients with 32 tumors in the TM-RFA arm. Within the TM-RFA arm, 22 patients (71.0%) underwent no-touch radiofrequency ablation, and nine patients (29.0%) underwent intratumoral puncture radiofrequency ablation. Technical success at the end of the procedure was 100% (76 of 76 patients). Primary technique efficacy after a single ablation session was 96.1% (73 of 76 patients); after immediate supplemental ablation in three patients, secondary technique efficacy was 100%. HCC, hepatocellular carcinoma.

Because the goal was to evaluate feasibility and safety in anatomically at-risk settings rather than to compare techniques, allocation to specific ablation approaches was nonrandomized and determined by a prespecified protocol based on tumor anatomy. Tumors were considered anatomically at risk when standard electrode placement and margin creation could be constrained, including in subcapsular or capsule-adjacent, subphrenic, pericholecystic, peribiliary, perivascular, and gastrointestinal-adjacent locations [9,17–20]. Tumors were categorized according to the dominant risk profile as follows: (1) Thermal at-risk tumors: Tumors with direct abutment or close proximity (≤5 mm) to heat-sensitive structures, including the gallbladder, bowel, stomach, or central bile duct [18,20,21]; (2) Oncologic at-risk tumors (heat-sink risk): Tumors with direct contact or close proximity (≤5 mm) to major vessels, including the portal vein, hepatic veins, or inferior vena cava, with a diameter ≥3 mm, where a clinically meaningful heat-sink effect and/or vascular injury risk was expected [19,22]; (3) Mechanical/access at-risk tumors: Capsule-adjacent tumors, defined as tumors with a tumor-to-capsule distance ≤10 mm, or subphrenic tumors, where electrode placement and safety-margin creation may be constrained and where bleeding or rupture risk can be increased [11].

Treatment allocation was stratified according to the dominant anatomical risk assessed during preprocedural planning. PEI-RFA was selected for thermal at-risk tumors, to permit reduced RF power at the critical interface, or for oncologic at-risk tumors, to mitigate the heat-sink effect, whereas TM-RFA was selected for mechanical/access at-risk tumors. For tumors adjacent to the gastrointestinal tract, either approach could be selected when adequate separation, with a target distance ≥5 mm, was achieved using artificial ascites, according to the following prespecified criteria: (1) PEI-RFA was preferred when a separation distance ≥5 mm could be reliably established using artificial ascites and the multipurpose needle could be positioned 3–5 mm from the gastrointestinal wall for focused ethanol delivery and interface temperature monitoring; (2) TM-RFA was selected when adequate hydrodissection (≥5 mm) was achievable but a stable intratumoral PEI configuration was technically unfeasible, for example because of tumor geometry, omental adhesion, or the needle trajectory required to reach the gastrointestinal interface; and (3) when neither ≥5 mm separation nor a safe PEI configuration could be achieved, the procedure was deferred or an alternative modality was considered. Intraprocedural crossover was permitted when the prespecified configuration could not be safely achieved.

Inclusion criteria

Patients aged 20–85 years with preserved liver function, defined as Child-Pugh class A or B, and 1–3 HCCs ≤3 cm each were eligible if HCC was confirmed by pathology or noninvasive imaging, including dynamic computed tomography (CT), magnetic resonance imaging (MRI), or contrast-enhanced ultrasonography, according to Korean Liver Cancer Association guidelines [23]. Patients were also required to have at least one tumor in an anatomically at-risk location as defined above [18,21].

Exclusion criteria

Patients were excluded if they had >3 total hepatic tumors; any single tumor >3 cm; Child-Pugh class C liver function; macrovascular invasion; severe coagulopathy, defined as a platelet count <50,000/mm³ or an international normalized ratio >1.5; extrahepatic metastasis; or inability to adhere to follow-up.

Sample Size Estimation

Because this was a descriptive pilot feasibility study without a predefined comparative hypothesis, no formal sample size calculation was performed. Enrollment of approximately 70–80 patients over an 8-month period was considered pragmatically feasible and adequate to provide preliminary estimates of safety and technical outcomes with reasonable precision for a single-center exploratory study.

Ablation Procedure

All procedures were performed at a tertiary referral center by a single operator (J.M.L.) with more than 20 years of experience in hepatic RFA, assisted by either a radiology resident or an abdominal clinical fellow. Tumors were assessed using contrast-enhanced CT or MRI within 6 weeks before the procedure. Tumor dimensions were measured in three orthogonal planes, and tumor volume was calculated using the ellipsoid formula. For each tumor, the anatomical relationship to adjacent critical structures was reviewed, and the shortest tumor-to-structure distance was recorded.

Ablation system and electrode

All treatments used a multichannel RF generator (VIVA RF System, STARmed, Goyang, Korea) and the Octopus MP electrode system (STARmed) [13]. The Octopus MP electrode system comprises two 17-gauge internally cooled RF electrodes with adjustable active tips and a central 19-gauge multi-perforated applicator with an integrated thermocouple at its distal tip, hereafter referred to as the “multipurpose needle.” This needle served two functions: (1) controlled instillation of saline or absolute ethanol through multiple side perforations along its distal shaft for uniform intralesional distribution and (2) continuous real-time tissue temperature sensing at a prespecified critical interface during RF energy delivery (Fig. 2).

Fig. 2. Octopus multipurpose electrode and ablation approaches.

A. The Octopus multipurpose radiofrequency (RF) ablation electrode (STARmed) is shown. The system comprises two internally cooled 17-gauge RF electrodes with adjustable active tips and a 19-gauge multipurpose needle, consisting of a multi-perforated applicator with an integrated temperature sensor. The inset diagram illustrates the integrated temperature sensor and multiple perforations for fluid instillation. B. Schematic illustration of combined percutaneous ethanol injection and RF ablation (approach 1) is shown. The multipurpose needle is inserted directly into the tumor to deliver ethanol, followed by RF ablation using electrodes placed at the tumor periphery. The integrated temperature sensor monitors tissue temperature near adjacent critical structures, with the gallbladder shown as an example. C. Schematic illustration of temperature-monitored no-touch RF ablation (approach 2) is shown. The two RF electrodes are positioned in the peritumoral liver parenchyma without puncturing the tumor. The temperature sensor monitors tissue temperature near the adjacent critical structure, with the diaphragm shown as an example, to help prevent thermal injury.

The adaptive treatment algorithm

Tumors meeting the anatomically at-risk criteria described above were treated using the prespecified adaptive strategy.

Arm 1: PEI-RFA

This approach was preferentially used for oncologic at-risk tumors, defined as perivascular tumors adjacent to major vessels ≥3 mm, and for selected thermal at-risk tumors, such as tumors adjacent to the gallbladder or central bile ducts, when reduced RF power at the critical interface was desirable. Ethanol induces chemical necrosis, which enables tumor destruction with reduced RF power and may lower the risk of thermal injury. Under fusion imaging guidance, two electrodes were placed at the tumor periphery, and the multipurpose needle was inserted through the tumor, with its distal tip positioned 3–5 mm from the nearest heat-sensitive structure, such as the gallbladder wall or bile duct, when feasible. This configuration focused ethanol delivery and temperature monitoring along the highest-risk interface. Absolute ethanol (99.5%; 1–5 mL, calculated as 0.5×tumor volume) was adjusted within ±20%–25% according to tumor location and injection resistance. The volume was reduced by approximately 20% for subcapsular or capsule-adjacent tumors to minimize the risk of ethanol leakage, capsular injury, and postprocedural pain and when high injection resistance indicated limited intratumoral diffusion capacity. Conversely, the volume was increased by up to 25% for perivascular tumors adjacent to vessels ≥3 mm to compensate for the heat-sink effect and accelerated ethanol washout by adjacent vascular flow and when low injection resistance suggested rapid interstitial dispersion (Figs. 2, 3). RFA was then performed immediately without electrode repositioning [24,25]. RF power was limited to approximately 30%–50% below the standard setting to prevent rapid carbonization from the early impedance rise associated with ethanol-induced protein denaturation and cellular dehydration, which lowers free water content but permits more sustained conductive heating and continued ethanol diffusion within the lesion [24,26]. In the PEI-RFA arm, RF power was reduced by approximately 30%–50% below the standard setting that would otherwise have been selected for a tumor of the same size. As a general guide, a larger reduction, toward 50%, was applied when the critical interface was closer (≤3 mm), and a smaller reduction, toward 30%, was applied when it was 3–5 mm away; however, this proximity-based scheme served as a starting framework rather than a strict rule. In the PEI-RFA arm, tumor puncture with the multipurpose needle was required in all cases to enable intralesional ethanol delivery; therefore, a no-touch configuration was not applicable to this arm.

Fig. 3. Combined ethanol injection and radiofrequency ablation for hepatocellular carcinoma adjacent to the common hepatic duct in a 58-year-old man.

A. Gadoxetic acid-enhanced T1-weighted arterial-phase image demonstrates a 1.6-cm nodule (arrow) with arterial-phase hyperenhancement in segment 1. B. Hepatobiliary-phase image shows the spatial relationship between the tumor (arrow) and the adjacent common hepatic duct (open arrow), which is located anterior to the lesion. C. Under ultrasound–MR fusion guidance, two radiofrequency electrodes (arrowheads) and the 19-gauge multipurpose needle (arrow) are positioned within the hyperechoic tumor. D. After instillation of 3 mL of absolute ethanol, the lesion (arrow) becomes markedly hyperechoic on ultrasound because of ethanol diffusion. E. After 3 minutes and 30 seconds of radiofrequency energy application at 60–80 W, the hyperechoic bubble cloud generated by tissue vaporization completely encompasses the virtual tumor marker (red circles), indicating adequate ablation coverage. F. Immediate postprocedural contrast-enhanced computed tomography (CT) demonstrates complete ablation of the index tumor with a sufficient ablative margin (arrow) and no evidence of bile duct injury. The heterogeneous appearance of the ablation zone reflects combined ethanol and thermal ablation. G. Fifteen-month follow-up contrast-enhanced CT shows no evidence of local tumor progression at the treated site (arrow), confirming durable local tumor control.

Arm 2: TM-RFA

This approach was preferentially used for mechanical/access at-risk tumors, including capsule-adjacent or subphrenic tumors without predominant perivascular risk, and for gastrointestinal-adjacent tumors when adequate separation (>5 mm) could be achieved and a PEI-RFA configuration was not selected. No-touch RFA was preferentially used to minimize the risk of tumor seeding in subcapsular tumors, whereas conventional intratumoral puncture was used when optimal peritumoral electrode placement was unfeasible. For no-touch RFA, electrodes were placed in the adjacent liver parenchyma under fusion imaging guidance (Figs. 2, 4); for tumor-puncture RFA, electrodes were placed within the target tumors. The multipurpose needle was advanced into the peritumoral parenchyma between the tumor and the diaphragm or abdominal wall, 3–5 mm from the capsule. Temperature was monitored continuously; RF power was suspended when the temperature approached 60°C and resumed at reduced power once the temperature decreased.

Fig. 4. No-touch radiofrequency ablation with intraprocedural temperature monitoring for recurrent hepatocellular carcinoma in a subphrenic location in a 68-year-old man.

A. Preprocedural contrast-enhanced computed tomography (CT) demonstrates a recurrent 1.3-cm arterial-phase hyperenhancing hepatocellular carcinoma (arrow) in the subphrenic portion of segment 4a. B. Ultrasound-CT fusion imaging confirms that the target tumor (yellow circle) is in close apposition to the diaphragm, classifying the lesion as mechanical/access at risk and supporting selection of temperature-monitored radiofrequency ablation according to the prespecified algorithm. C. Under ultrasound-CT fusion guidance, two internally cooled 17-gauge radiofrequency electrodes (arrows) are deployed in the peritumoral parenchyma without puncturing the tumor, consistent with a no-touch configuration. D. The 19-gauge multipurpose needle (arrow) is advanced into the peritumoral parenchyma between the tumor and the diaphragm to enable real-time tissue temperature monitoring at the tumor–diaphragm interface during radiofrequency ablation. E. During ablation, hyperechoic bubble formation (yellow dotted circle) indicates progressive coagulation encompassing the target lesion, while radiofrequency power is modulated according to temperature feedback from the multipurpose needle to protect the adjacent diaphragm. F. Immediate post-ablation CT shows that the hypoenhancing ablation zone (arrow) completely encompasses the recurrent tumor with an adequate ablative margin.

RFA protocol common to both approaches

Procedures were performed under conscious sedation administered and monitored by the procedural team, with continuous cardiorespiratory monitoring, including electrocardiography, noninvasive blood pressure measurement, and pulse oximetry. General anesthesia and dedicated anesthesiology assistance were not routinely required. Real-time ultrasound-CT/MRI fusion imaging guidance was used in all procedures. For 33 patients (43.4%) with suboptimal lesion conspicuity, perflubutane-enhanced ultrasonography (Sonazoid, GE Healthcare, Oslo, Norway) was used. For tumors near adjacent extrahepatic organs, artificial ascites was created using 5% dextrose, with a target separation of ≥5 mm when feasible, and the achieved separation was assessed in real time [19,27]. Artificial ascites was used to displace and thermally protect mobile extrahepatic organs, such as the gastrointestinal tract. Because artificial ascites cannot separate fixed intrahepatic structures, such as the central bile ducts or major vessels, or counteract the heat-sink effect, combined PEI was reserved for these settings and was not mandated when adequate separation from a mobile organ (≥5 mm) was achieved by ascites alone; in such cases, TM-RFA could be selected according to the algorithm. RF energy was delivered using switching monopolar and bipolar modes with impedance-based alternation, initiated at 60 W and increased by 10 W every 30 seconds [28,29]. Ablation continued for 3–12 minutes until the ablative zone encompassed the tumor with a target margin ≥5 mm when feasible, except where a full margin was not possible because of an adjacent critical structure. Tract ablation was performed during electrode withdrawal.

Post-ablation Assessment

Immediate evaluation was performed using multiphase contrast-enhanced CT or MRI. Ablation margins were assessed in consensus by an experienced radiologist with 25 years of hepatobiliary imaging experience and either a resident or a clinical fellow. A sufficient ablative margin was defined as ≥5 mm of nonenhancing tissue surrounding the tumor on all planes, except at interfaces with critical structures, where complete tumor coverage without a standard margin was accepted [30] (Figs. 3, 4). Technical success was defined as complete ablation without residual enhancement. Immediate retreatment was performed for residual enhancing tumor, as an absolute indication, or for margins <5 mm in noncritical areas, as a relative indication [30]. Patients were monitored for 12–24 hours before discharge.

Data Collection and Follow-up

Demographic, tumor, and procedural data were collected prospectively. Structured follow-up included clinical evaluations, laboratory tests, and dynamic CT or MRI at 1 month and every 3 months thereafter. The oncologic analysis was censored at the study data cutoff (median follow-up, 29.5 months; range, 0.8 to 36.4 months); for the five patients with late biliary strictures, additional clinical and imaging surveillance available beyond the analysis period (up to 43 months) was reviewed to confirm the stability of these findings.

Oncologic Outcomes and Definitions

The primary endpoint was safety, defined as the rate of thermal injury to nontarget structures. Thermal injury to nontarget structures was assessed on imaging and defined as radiologic evidence of ablation-related thermal damage, including any of the following: (1) wall thickening >3 mm involving the gallbladder, bowel wall (stomach, duodenum, or colon), central bile duct, or surrounding parenchyma; (2) focal enhancement abnormality, including delayed hyperenhancement or persistent hypoenhancement, at critical-structure interfaces; (3) diaphragmatic thickening >2 mm or signal-intensity alteration on MRI; (4) new biliary wall thickening or stricture with upstream ductal dilatation; or (5) new thrombosis or stenosis of an adjacent major vessel [5,6,31,32]. A standardized assessment template was used to document the presence, location, severity, and imaging features of any thermal injury. Severity was classified as minor, defined as reversible imaging changes without clinical consequence, or major, defined as imaging findings requiring intervention. Adverse events were classified according to timing and severity. Acute complications occurring within 30 days of the procedure were categorized as major, defined as requiring intervention, prolonging hospitalization, or causing significant morbidity, or minor according to standardized criteria [33]. Late imaging sequelae detected more than 30 days after the procedure during follow-up imaging were documented separately, including time to detection and clinical significance.

Secondary endpoints included technical success, technique efficacy, LTP, intrahepatic distant recurrence (IDR), distant metastasis, and recurrence-free survival (RFS). In this pilot study, these endpoints were preliminary assessments in a single-center, noncomparative cohort without a control group for formal comparative analysis. Technical success was defined as complete tumor coverage without residual enhancement on immediate post-ablation imaging at the end of the procedure [33]. Technique efficacy was defined as complete tumor ablation on 1-month follow-up imaging [30,33]. Residual tumor requiring a repeat session was counted as technical failure for the first session but could contribute to technique efficacy if complete ablation was confirmed on 1-month follow-up CT or MRI. LTP was defined as new tumor foci at the ablation margin after initial complete ablation, as confirmed by consensus review by two radiologists on follow-up imaging [33]. Recurrence outcomes were also assessed. IDR was defined as development of a new HCC nodule within the liver but separate from the ablation site. RFS was defined as the time from ablation to any recurrence.

Statistical Analysis

Given the pilot, noncomparative nature of this feasibility study, analyses were descriptive rather than hypothesis-testing. Continuous variables were reported as the mean±standard deviation or the median with interquartile range, and categorical variables were reported as frequencies and percentages. Safety and efficacy outcomes were summarized primarily for the overall cohort. All technique-specific results, including Kaplan-Meier curves and subgroup descriptions, are provided for descriptive purposes only and should not be interpreted as comparisons of the relative effectiveness of PEI-RFA and TM-RFA because treatment allocation was nonrandomized and driven by tumor anatomy. Time-to-event outcomes, including LTP, IDR, and RFS, were estimated using the Kaplan-Meier method with 95% confidence intervals (CIs).

Because only six LTP events occurred, exploratory univariate Cox proportional hazards models using the Breslow method for ties were used to evaluate potential predictors of LTP. Multivariable models were not pursued to avoid overfitting given the low event count [34]. Statistical significance was defined as P<0.05. Analyses were performed using MedCalc software version 22.0.30 (MedCalc Software, Mariakerke, Belgium).

Results

Baseline Patient and Tumor Characteristics

Seventy-six patients, including 59 men (77.6%), with 83 small HCCs were enrolled (mean age, 68.3±9.2 years; range, 42 to 83 years) (Fig. 1). Sixty-nine patients (90.8%) had a single tumor, and seven (9.2%) had two tumors; 52 patients (68.4%) were treatment-naïve, and 24 (31.6%) had recurrent disease after prior treatment. Baseline characteristics are summarized in Table 1.

Table 1.

Baseline patient and tumor characteristics

Characteristic	Total (n=76)	PEI-RFA (n=45)	TM-RFA (n=31)a)
Patient information
Age (year), mean±SD (range)	68.3±9.2 (42–83)	67.5±9.7 (42–83)	69.5±8.4 (52–82)
Male/Female	59 (77.6)/17 (22.4)	34 (75.6)/11 (24.4)	25 (80.6)/6 (19.4)
Clinical and laboratory parameters
Child-Pugh class, A vs. B	73 (96.1) vs. 3 (3.9)	42 (93.3) vs. 3 (6.7)	31 (100) vs. 0 (0)
α-fetoprotein (ng/mL), median (IQR)	3.8 (2.6–7.0)	4.2 (2.8–10.8)	3.6 (1.9–5.7)
PIVKA-II (mAU/mL)	24 (19–35)	26 (22–37)	22 (18–27)
Tumor characteristics (83 tumors)
No. of tumors	83	51	32
Single/Two tumors (patients)	69 (90.8)/7 (9.2)	39 (86.7)/6 (13.3)	30 (96.8)/1 (3.2)
Tumor size (cm), mean±SD	1.4±0.5	1.4±0.5	1.4±0.5
Tumor presentation: initial/recurrent (patients)	52 (68.4)/24 (31.6)	32 (71.1)/13 (28.9)	20 (64.5)/11 (35.5)
Anatomically challenging sitesb) vs. conventional location	82 (98.8):1 (1.2)	50 (98.0):1 (2.0)	32 (100):0 (0)
Thermal high-riskc)
Gallbladder	7 (8.4)	7 (13.7)	0
Central bile duct (> segmental)	5 (6.0)	5 (9.8)	0
GI tract (stomach, colon)	10 (12.0)	5 (9.8)	5 (15.6)
Oncologic high-risk (vessels ≥3 mm)	43 (51.8)	40 (78.4)	3 (9.4)
Mechanical high-risk
Subcapsular (≤10 mm from capsule)	39 (47.0)	21 (41.2)	18 (56.2)
Subphrenic (includes diaphragm proximity)	27 (32.5)	10 (19.6)	17 (53.1)
Ablation technique
No-touch RFA	22 (28.9)	0	22 (71.0)
Intratumoral puncture RFA	54 (71.1)	45 (100)	9 (29.0)

Values are presented as number (%) unless otherwise indicated.

No formal statistical comparisons were performed between protocol-defined approach groups because allocation was anatomy-driven and non-randomized.

Thermal at-risk=direct contact (0 mm) or ≤5 mm to gallbladder, stomach, bowel, or central bile duct; Oncologic at-risk=direct contact (0 mm) or ≤5 mm from major vessels ≥3 mm (portal/hepatic veins or IVC); Mechanical/access at-risk=capsule-adjacent (≤10 mm) or subphrenic.

PEI-RFA, percutaneous ethanol injection plus radiofrequency ablation; TM-RFA, temperature-monitored radiofrequency ablation; SD, standard deviation; IQR, interquartile range; GI, gastrointestinal; IVC, inferior vena cava.

^a)Three patients initially assigned to the PEI-RFA group were converted to TM-RFA due to technical inability to place the 19-gauge multipurpose needle, caused by omental interference (n=2) or treatment-induced fibrosis (n=1). ^b)”Anatomically challenging” locations include subcapsular, perivascular, subphrenic, and pericholecystic positions (tumor-based, n=83). ^c)Number of tumors with at least one anatomically at-risk feature among 83 tumors. At-risk categories are non-mutually exclusive; a single tumor may meet multiple categories.

Using the prespecified anatomical at-risk classification, 82 of 83 tumors (98.8%) met at least one at-risk criterion. Consistent with anatomy-driven allocation, the distribution of at-risk features differed markedly between the two arms (Table 1): oncologic risk, defined as proximity to vessels ≥3 mm, predominated in the PEI-RFA arm (78.4% vs. 9.4%), whereas subphrenic location predominated in the TM-RFA arm (53.1% vs. 19.6%).

Based on the dominant risk profile and technical feasibility, 48 patients were initially allocated to PEI-RFA and 28 to TM-RFA. Three patients crossed over from the PEI-RFA plan to TM-RFA because the 19-gauge multipurpose needle could not be placed, owing to omental interference in two patients and treatment-related fibrosis in one patient. Accordingly, the per-protocol cohort included 45 patients with 51 tumors treated with PEI-RFA and 31 patients with 32 tumors treated with TM-RFA. By protocol, all 45 patients in the PEI-RFA arm underwent tumor puncture with the multipurpose needle. In the TM-RFA arm, 22 of 31 patients (71.0%) underwent no-touch RFA, and nine of 31 patients (29.0%) underwent intratumoral puncture RFA because of anatomical constraints.

RFA Technical Parameters

Procedural parameters are presented separately for each arm in Table 2. Artificial ascites was used in 46 of 51 tumors (90.2%) in the PEI-RFA arm and 28 of 32 tumors (87.5%) in the TM-RFA arm, corresponding to 74 of 83 tumors overall (89.2%).

Table 2.

Summary of the technical parameters and procedural details of RFA

Parameter	Overall cohort	PEI-RFA group	TM-RFA group
	(n=76 patients, 83 tumors)	(n=45 patients, 51 tumors)	(n=31 patients, 32 tumors)
Adjunctive measure
Artificial ascites used	74 (89.2)	46 (90.2)	28 (87.5)
Ethanol injection volume (mL)	N/A	2.0 (1.5–3.0)	N/A
Ablation parameter
Total ablation time (min)	8.6±4.1	8.7±4.5	8.4±3.5
Total applied energy (kcal)	8.9±4.7	8.6±4.9	9.3±4.4
Mean ablation power (W)	71.9±16.5	72.1±17.0	71.6±15.9
Procedural outcome
Ablation zone volume (cm3)	32.1±18.6	31.0±20.3	33.9±15.6
Ablative margin achieveda)	80 (96.4)	48 (94.1)	32 (100)

Values are presented as number (%), mean±standard deviation, or median (interquartile range).

The combined PEI-RFA group included all procedures where ethanol was injected. The TM-RFA group included all procedures where ethanol was not used.

RFA, radiofrequency ablation; PEI-RFA, percutaneous ethanol injection plus RFA; TM-RFA, temperature-monitored RFA; N/A, not available.

^a)Ablative margin achieved was defined as a “yes” in the safety margin creation column of the dataset, indicating the creation of a non-enhancing ablative margin 5 mm around the tumor, except at interfaces with critical structures.

In the PEI-RFA arm, which included 45 patients with 51 tumors, all procedures were performed using tumor puncture with the multipurpose needle to enable intralesional ethanol delivery. The median ethanol dose was 2.0 mL (interquartile range, 1.5 to 3.0 mL). The mean total ablation time was 8.7±4.5 minutes, the total applied energy was 8.6±4.9 kcal, and the mean ablation power was 72.1±17.0 W.

In the TM-RFA arm, which included 31 patients with 32 tumors, 22 of 31 patients (71.0%) underwent no-touch RFA, and nine of 31 patients (29.0%) underwent intratumoral puncture RFA. The mean total ablation time was 8.4±3.5 minutes, the total applied energy was 9.3±4.4 kcal, and the mean ablation power was 71.6±15.9 W.

Procedural Outcomes, Technical Success, and Technique Efficacy

Technical success at the end of the procedure was achieved in 100% of patients in both arms (PEI-RFA, 45 of 45; TM-RFA, 31 of 31). After the initial ablation session, 73 of 76 patients (96.1%) overall achieved complete ablation without immediate supplemental ablation. Three patients underwent immediate supplemental ablation within 1 hour because post-ablation imaging suggested residual enhancement or an insufficient margin in a noncritical area; all three achieved complete tumor coverage, yielding secondary technique efficacy of 100%. Complete ablation on 1-month follow-up imaging was confirmed in all 76 patients (100%) in both arms.

The mean immediate ablation-zone volume was 31.0±20.3 cm³ in the PEI-RFA arm and 33.9±15.6 cm³ in the TM-RFA arm; the overall mean was 32.1±18.6 cm³. A prespecified ≥5-mm ablative margin was documented in 48 of 51 tumors (94.1%) in the PEI-RFA arm and 32 of 32 tumors (100%) in the TM-RFA arm, corresponding to 80 of 83 tumors overall (96.4%). The three tumors without a documented full margin in the PEI-RFA arm were all located at interfaces with critical structures, where complete tumor coverage without a standard 5-mm margin was accepted per protocol (Table 2).

Primary Endpoint: Safety and Adverse Events

No major complications, including bowel perforation, hemorrhage, or needle tract seeding, occurred in either arm.

PEI-RFA arm (45 patients, 51 tumors)

Minor complications occurred in seven of 45 patients (15.6%). The primary safety endpoint, thermal injury, was observed in 3 of 45 patients (6.7%): one case of asymptomatic gallbladder wall thickening and two cases of focal diaphragmatic thickening. All three cases resolved with conservative management. Late biliary sequelae, defined as clinically insignificant biliary strictures detected more than 12 months after the procedure, were identified in four of 45 patients (8.9%).

TM-RFA arm (31 patients, 32 tumors)

Minor complications occurred in three of 31 patients (9.7%). Thermal injury was observed in two of 31 patients (6.5%); both cases consisted of minor focal diaphragmatic thickening that resolved with conservative management. Late clinically insignificant biliary stricture was identified in one of 31 patients (3.2%). Across both arms, the five late biliary strictures appeared as focal narrowing of a segmental or peripheral bile duct adjacent to the ablation zone, with mild upstream ductal dilatation. None was associated with jaundice, cholangitis, or biliary intervention, and all remained clinically silent and radiologically stable on subsequent surveillance imaging performed beyond the primary analysis period (extended surveillance in these five patients, 38 to 43 months after ablation).

Because allocation was anatomy-driven and nonrandomized, these arm-specific rates are presented descriptively without formal between-arm statistical comparison.

Secondary Endpoint: LTP and Recurrence

LTP occurred in six of 76 patients overall. Kaplan-Meier estimates are provided separately for each arm in Table 3 and Fig. 5.

Table 3.

Cumulative incidence of LTP, IDR, and distant metastases at 12, 24, and 36 months, by treatment approach and overall (Kaplan-Meier estimates)

Endpoint	12 Months	24 Months	36 Months
LTP
Patient-based
PEI-RFA (n=45)	4.44 (1.15–16.54)	7.13 (2.29–21.07)	10.95 (4.05–27.60)
TM-RFA (n=31)	3.23 (0.46–20.82)	3.23 (0.46–20.82)	7.41 (1.83–27.38)
Overall (n=76)	4.05 (0–8.55)	5.65 (0.25-11.06)	10.18 (2.14-18.23)
Tumor-based
PEI-RFA (n=51)	4.10 (0.00–9.54)	6.40 (2.38–16.54)	9.80 (4.26–20.98)
TM-RFA (n=32)	3.13 (0.61–15.81)	3.13 (0.61–15.81)	7.30 (2.26–21.60)
Overall (n=83)	3.75 (0–7.91)	5.21 (0.22-10.19)	9.51 (1.94-17.09)
IDR	38.35 (26.78–49.53)	49.38 (36.58–60.94)	49.38 (36.58–60.94)
Distant metastases	6.98 (2.19–15.78)	9.94 (3.73–19.96)	9.94 (3.73–19.96)
RFS rates	47.93 (36.21–58.88)	36.10 (24.42–47.95)	32.09 (20.43–44.56)

Values are presented as percentage; numbers in parentheses indicate 95% confidence intervals. Estimated using the Kaplan-Meier method.

LTP is presented on both tumor-based and patient-based units; IDR and distant metastases are patient-based. Technique-specific estimates are provided descriptively; no formal between-approach comparisons were performed.

LTP, local tumor progression; IDR, intrahepatic distant recurrence; PEI-RFA, percutaneous ethanol injection plus radiofrequency ablation; TM-RFA, temperature-monitored radiofrequency ablation; RFS, recurrence-free survival.

Fig. 5. Kaplan-Meier analyses after radiofrequency ablation for hepatocellular carcinoma in anatomically at-risk locations.

Curves are shown for the overall cohort. Because treatment allocation was anatomy-driven and nonrandomized, approach-specific curves are not presented, and no between-approach inference is intended; approach-specific estimates are provided descriptively in Table 3. Shaded areas represent 95% confidence intervals (CIs). A. Patient-based cumulative local tumor progression (LTP)–free probability for the overall cohort (n = 76) is shown. The estimated LTP-free rates at 12, 24, and 36 months were 96.0%, 94.4%, and 89.8%, respectively. B. Tumor-based cumulative LTP-free probability for the overall cohort (n = 83 tumors) is shown. The estimated LTP-free rates at 12, 24, and 36 months were 96.3%, 94.8%, and 90.5%, respectively. C. Patient-based cumulative intrahepatic distant recurrence (IDR)–free probability at 12, 24, and 36 months is shown; the estimated IDR-free rates were 61.7%, 50.6%, and 50.6%, respectively. D. Patient-based recurrence-free survival at 12, 24, and 36 months is shown; the estimated rates were 47.9%, 36.1%, and 32.1%, respectively.

In the PEI-RFA arm, which included 45 patients with 51 tumors, patient-based cumulative LTP rates were 4.44% (95% CI, 1.15% to 16.54%) at 12 months, 7.13% (95% CI, 2.29% to 21.07%) at 24 months, and 10.95% (95% CI, 4.05% to 27.60%) at 36 months. Corresponding tumor-based rates were 4.10% (95% CI, 0.00% to 9.54%), 6.40% (95% CI, 2.38% to 16.54%), and 9.80% (95% CI, 4.26% to 20.98%), respectively.

In the TM-RFA arm, which included 31 patients with 32 tumors, patient-based cumulative LTP rates were 3.23% (95% CI, 0.46% to 20.82%) at both 12 and 24 months and 7.41% (95% CI, 1.83% to 27.38%) at 36 months. Corresponding tumor-based rates were 3.13% (95% CI, 0.61% to 15.81%) at both 12 and 24 months and 7.30% (95% CI, 2.26% to 21.60%) at 36 months.

Given the limited number of nonlocal recurrence events in each arm and the descriptive nature of this pilot analysis, IDR and distant metastasis are reported for the overall cohort (Table 3). IDR was the most common recurrence pattern, with cumulative incidence rates of 38.35% (95% CI, 26.78% to 49.53%) at 12 months and 49.38% (95% CI, 36.58% to 60.94%) at both 24 and 36 months. Distant metastases occurred at cumulative rates of 6.98% (95% CI, 2.19% to 15.78%) at 12 months and 9.94% (95% CI, 3.73% to 19.96%) at both 24 and 36 months. Recurrence-free survival was 47.93% at 12 months, 36.10% at 24 months, and 32.09% at 36 months.

Exploratory Analysis of Risk Factors for LTP

Given the low total number of LTP events (n=6) and the nonrandomized arm allocation, risk-factor analysis was performed in the pooled cohort rather than separately by arm to preserve statistical stability. In exploratory univariate Cox analysis, no variable showed a robust association with LTP. Tumor size showed a trend toward increased risk (hazard ratio, 1.67 per 1-cm increase; 95% CI, 0.95 to 2.93), whereas other covariates had wide CIs because of the limited event count. These results are hypothesis-generating only and should be interpreted with caution; arm-stratified risk-factor analyses were not feasible.

Discussion

In this prospective feasibility cohort enriched for small HCCs in anatomically at-risk locations, an adaptive strategy using the Octopus MP electrode was associated with a low rate of imaging-detected thermal injury: three of 45 patients (6.7%) in the PEI-RFA arm and two of 31 patients (6.5%) in the TM-RFA arm, all minor. The strategy also achieved 100% technical success and 100% technique efficacy at 1 month in both arms, with acceptable mid-term local control, as reflected by tumor-based cumulative LTP rates at 36 months of 9.80% in the PEI-RFA arm and 7.30% in the TM-RFA arm. The 100% technical success and 1-month technique efficacy should nevertheless be interpreted with caution. Complete ablation after a single session was achieved in 73 of 76 patients (96.1%), and three patients required immediate repeat ablation within 1 hour to achieve 100% secondary efficacy. In addition, these outcomes were obtained by a single highly experienced operator using intensive multimodality guidance; therefore, reproducibility may be lower in lower-volume settings.

Importantly, the individual components of this adaptive strategy are not novel. Combined PEI and RFA has been investigated since the early 2000s, and meta-analyses and comparative studies have shown improved local control compared with RFA alone, particularly for tumors larger than 3 cm or tumors in unfavorable locations [14,15,26]. Intraprocedural temperature monitoring, using separately inserted thermocouples, fiber-optic sensors, or remote thermistors, has likewise been used to reduce thermal injury during ablation near bile ducts, bowel, and other heat-sensitive structures and has been combined with intraductal chilled saline perfusion for peribiliary protection [35,36]. Adjunctive hydrodissection with artificial ascites or pleural effusion is also a well-established component of current practice [19,27,37].

Therefore, the contribution of the present study lies not in any single technique but in the integration of these techniques within a single device platform. This platform combines two cooled RF electrodes, intralesional fluid delivery with saline or ethanol, real-time tissue thermometry, and dual-electrode switching monopolar–bipolar energy delivery through one coordinated system. In this study, that system was applied within a prespecified anatomy-driven algorithm that matched each therapeutic mechanism to the dominant mechanism of failure at a given tumor location: heat-sink effects in perivascular tumors, thermal injury risk near heat-sensitive structures, and access constraints in capsule-adjacent or subphrenic tumors. This integrated approach streamlines the technical workflow compared with the use of multiple separate devices and sensors and allows the operator to switch between or combine chemical augmentation and thermal monitoring within the same setup without repositioning instruments.

Safety is a key barrier to percutaneous thermal ablation when tumors are adjacent to the gallbladder, central bile ducts, gastrointestinal tract, or diaphragm. In this cohort, which was enriched for small HCCs in anatomically at-risk locations, no major complications, such as gastrointestinal perforation, bile peritonitis, or clinically significant hemorrhage, were observed. The imaging findings categorized as thermal injury, including gallbladder wall thickening and focal diaphragmatic thickening, resolved with conservative management. Although all late biliary strictures were clinically insignificant and required no intervention, their incidence, five of 76 patients (6.6%), is not negligible in a small cohort and underscores the need for longer surveillance to confirm their benign natural history. Several protocol elements likely contributed to these safety outcomes. First, real-time temperature feedback permitted modulation or suspension of energy delivery when a critical interface approached 60°C. Second, ethanol augmentation was used in selected settings to enhance ablation while allowing reduced RF power near vulnerable structures, which may be particularly relevant for perivascular tumors susceptible to heat-sink effects [14,15,26,38]. Third, adjunctive separation and guidance techniques, including artificial ascites or pleural effusion, fusion imaging, and contrast-enhanced ultrasonography, supported accurate placement of electrodes and sensors in complex anatomy [39,40]. No-touch TM-RFA was preferentially applied in capsule-adjacent tumors when feasible to avoid direct tumor puncture [41,42]. Although the relative contributions of the device and these adjunctive techniques cannot be isolated in this single-arm design, these protocol elements collectively appear to have supported the favorable safety profile observed.

Oncologic outcomes should be interpreted in the context of the selected population and the absence of a concurrent control arm. Nevertheless, the observed LTP rates were within the range reported for conventional RFA in difficult locations, where prior series have reported LTP rates of approximately 10%–20% at 2 years [43,44]. These comparisons should be interpreted cautiously because the tumors in this study were small, with a mean diameter of approximately 1.4 cm, well below the 3-cm eligibility ceiling. Smaller tumors are intrinsically easier to encompass with an adequate margin and have a lower baseline risk of local progression independent of technique; therefore, favorable tumor size and biology, rather than the device or algorithm alone, likely contributed substantially to the observed local control. The adaptive algorithm was designed to match each therapeutic mechanism to the dominant mode of failure: ethanol augmentation for heat-sink and thermal injury risk and temperature monitoring for access-constrained, capsule-adjacent, or subphrenic tumors. These hypothesis-generating findings support further evaluation in controlled multicenter settings.

Alternative ablation modalities have been investigated for these challenging anatomies. Microwave ablation has shown comparable local control in perivascular settings; however, Makovich et al. reported complication rates exceeding 20% [20,45], which were higher than the minor complication rates observed in the present cohort (15.6% in the PEI-RFA arm and 9.7% in the TM-RFA arm). Cryoablation is another alternative with a less pronounced cold-sink phenomenon, although it carries specific risks, including cryoshock and bleeding [46]. In perivascular HCCs, cryoablation has yielded 2-year LTP rates of 14.6%, with transient vascular thrombosis in 6.9% of cases [47]. Although microwave ablation and cryoablation may offer specific advantages for perivascular tumors, RFA remains the most widely available and best-validated modality, particularly outside high-volume specialty centers. The Octopus MP electrode may extend the utility of RFA in high-risk scenarios by enabling combined PEI and real-time temperature monitoring, potentially reducing thermal injury while maintaining durable local control. Device selection should be individualized according to tumor anatomy and operator expertise. Clinically, these results suggest that selected small HCCs near critical structures may be treatable with percutaneous RFA when a structured safety protocol, appropriate adjuncts, and experienced operators are available. However, broader adoption will require external validation, standardized definitions of anatomical risk, and assessment of the learning curve and resource requirements. Because difficult-location RFA is strongly operator-dependent, particularly no-touch electrode placement, precise positioning of the multipurpose needle 3–5 mm from a critical interface, and real-time titration of energy according to temperature feedback, the favorable outcomes observed here may not be directly reproducible by less experienced operators. The learning curve for this integrated workflow should be characterized prospectively before broader adoption.

Several limitations should be acknowledged. First, this exploratory, single-arm, single-center study without a control group substantially limits attribution of outcomes to the Octopus MP electrode. The protocol incorporated several cointerventions, including fusion imaging, artificial ascites, no-touch placement, ethanol augmentation, real-time thermometry, and a highly experienced operator; therefore, the device’s independent contribution cannot be disentangled from these adjuncts, and the findings should be interpreted as demonstrating the feasibility of an integrated workflow rather than device-specific efficacy. Second, anatomy-driven allocation introduced confounding and limits inference about comparative effectiveness between PEI-RFA and TM-RFA. Third, the proximity-based definition of anatomically at-risk locations, defined as ≤5 mm for organ or vessel proximity and ≤10 mm for capsule adjacency, encompassed lesions with varying degrees of technical difficulty. This heterogeneity may have introduced differences in baseline procedural risk and complicates direct comparison with prior cohorts that used different at-risk definitions. In addition, selection bias is likely because tumors in which neither adequate separation nor a safe ablation configuration could be achieved were deferred or referred for alternative therapy and were not enrolled. Therefore, the reported safety and local-control estimates apply to technically feasible cases and may be more favorable than estimates in an unselected difficult-location population. Fourth, only tumors ≤3 cm were included, and performance in larger tumors remains unknown. Finally, because artificial ascites was used in nearly 90% of tumors and all procedures were performed by a single highly experienced operator at a high-volume tertiary center, these outcomes likely reflect advanced multimodality RFA in a highly specialized environment rather than the intrinsic effectiveness of the device platform. Generalizability to lower-volume centers and to operators earlier in the learning curve warrants dedicated study.

In conclusion, an integrated single-device platform that consolidates previously established adjunctive techniques, including intralesional ethanol augmentation, real-time tissue thermometry, and dual-electrode switching monopolar–bipolar energy delivery, within a prespecified anatomy-driven algorithm was feasible for small HCCs adjacent to critical structures. The approach was associated with a low rate of imaging-detected thermal injury and acceptable mid-term local tumor control. These findings should be regarded as hypothesis-generating rather than evidence of superior device efficacy, given the pilot, single-arm, noncomparative design. Controlled multicenter studies are needed to determine whether this integrated approach offers incremental clinical benefit over conventional RFA performed with separately deployed ethanol injection, thermocouple-based thermometry, and hydrodissection in comparable anatomically at-risk populations.

Key points

An anatomy-driven adaptive strategy using the Octopus multipurpose electrode, in which each tumor was allocated to either percutaneous ethanol injection plus radiofrequency ablation (PEI-RFA) or temperature-monitored radiofrequency ablation (TM-RFA) according to its dominant anatomical risk, was technically feasible for 83 small hepatocellular carcinomas (≤3 cm) in anatomically at-risk locations among 76 patients. No major complications occurred. Thermal injury was observed in 6.6% of patients, and all events were minor and self-limited, supporting the safety of this approach for tumors adjacent to critical structures. Technique efficacy at 1 month was 100%. Cumulative local tumor progression rates at 36 months were 11.0% in the PEI-RFA arm and 7.4% in the TM-RFA arm, suggesting durable local control with a favorable safety profile.