Radiofrequency ablation of hepatocellular carcinoma guided by real-time physics-based ablation simulation: a prospective study

Abstract

Objectives

To assess the safety and efficacy of radiofrequency ablation (RFA) guidance software that incorporated patient-specific physics-based simulation of each ablation volume.

Materials and methods

Patients referred for curative ablation of hepatocellular carcinoma (HCC) of 2–5 cm diameter were prospectively enrolled. RFA was performed under general anesthesia. Procedure planning and intraprocedural modifications were guided by computer simulation of each ablation. The segmented target (tumor with 5 mm margin) was registered to and superimposed on subsequent 3D multiplanar images. The applied RF energy was used to calculate a simulated ablation volume which was displayed relative to the electrode and segmented target, to depict any untreated target tissue. After each additional ablation, the software updated the accumulated simulated ablation volume in relation to the target. The primary endpoints were technical efficacy and rate of local tumor progression (LTP).

Results

Sixty-eight tumors were ablated during 57 procedures in 52 patients (68.3 ± 9.2 years old, 78.8% male); 15 (26.3%) had multiple lesions and 23 (39.1%) had prior HCC treatment. The mean tumor diameter was 2.73 (±0.64) cm. The intraprocedural simulation directed additional overlapping ablations in 75.9% of tumors. Technical success and efficacy were 100% at 3-month contrast enhanced CT or MRI follow-up after the single treatment session. Cumulative incidence function estimates for 1- and 2-year LTP were 3.9% and 20.2%, respectively.

Conclusion

This prospective study found computer-assisted guidance that simulated each ablation was both safe and efficacious. The low rate of LTP was similar to studies that employed stereotactic guidance and ablation confirmation, without requiring a second contrast enhanced study.

Keywords:

• Hepatocellular carcinoma
• computer-assisted therapies
• thermal ablation
• simulation

Introduction

Radiofrequency ablation (RFA) is a potentially curative minimally-invasive first-line therapy for unresectable small hepatocellular carcinoma (HCC) with low complication rates and rapid recovery [1,2]. The procedure entails placement of a needle electrode into the tumor and applying radiofrequency (RF) energy to heat and locally destroy the target tissue [1]. However, as tumors exceed 2 cm, the efficacy of a single ablation diminishes, with reported 2-year local tumor progression (LTP) rates of up to 30.2% [3–8].

LTP usually occurs at the edge of an ablation, reflecting an incomplete treatment due to a combination of tumor biology, technical limits, and imaging challenges [9]. Because HCC often grow ‘satellite’ extensions, ablation of an additional 5 mm of hepatic tissue beyond the enhancing tumor (the ablation margin) is necessary to reliably achieve LTP rates approximating the 6% 2-year LTP reported for surgical resection [3–5,10–13]. Achieving that margin is increasingly difficult in larger tumors, where imprecise electrode placement, and variation between the predicted and actual ablation volume may result in an inadequate margin [14,15]. The electrode manufacturer’s predicted ablation volume is imprecise because it is an estimation from ex-vivo data, which cannot incorporate patient-specific factors such as the actual power delivered, the cooling effect of adjacent blood vessels, or the tumor and local parenchymal perfusion [15].

In practice, the adequacy of the minimal ablation margin (MAM) is assessed by comparison of the viable (pre-ablation contrast-enhanced) tumor and the non-viable (post-ablation non-enhancing) ablated volume. While initially assessed by side-by-side image comparison, the quantitative measurement from registered and fused multiplanar images proved a superior predictor of LTP [4,7,16,17]. However, reliance on a post-procedure contrast-enhanced study and the time-consuming image registration is a practical limitation for guidance of intraprocedural modifications [12,18].

Ideally, the tissue damage produced after each ablation could be identified to allow intraprocedural direction for any additional overlapping ablations required to achieve the target margins; however, a repeat contrast-enhanced CT (CECT) after each ablation is often precluded by the risk of renal toxicity (which is elevated in cirrhotic patients on diuretics for ascites management [19,20]). A computer simulation may provide a depiction of each ablation volume without an additional administration of contrast [21,22]. Though an ablation simulation once took hours to compute, advances in computer hardware technology and numerical approaches have reduced this to minutes, permitting real-time intraprocedural application [23–26].

This study evaluated the safety and efficacy of Accublate™ RF Sim software (NE Scientific, LLC, Boston, MA) for RFA guidance. The software provided real-time 3D multiplanar graphical displays of patient-specific physics-based simulations for each ablation that were superimposed on the segmented tumor target. Independent simulations at each electrode position indicated the treated versus untreated target tissue and guided additional treatment if needed. This study evaluated the benefit of intraprocedural simulation guidance for RFA of HCC as assessed by technical success, clinical efficacy, and LTP rates.

Materials and methods

Patients

This prospective, single-arm study assessed the safety and efficacy of simulation-based guidance software for RFA of intrahepatic HCC. Patients eligible for enrollment were those referred for curative ablation of an HCC by the multidisciplinary Liver Tumor Clinic at a 400-bed academic cancer center between December 2019 and December 2021. Ablation was considered for tumors deemed unresectable due to comorbidity or location, or as a bridge to transplant. Selection generally adhered to the Barcelona Clinic Liver Cancer (BCLC) classification, though patients with higher grade disease were included if suitable for curative intent ablation.

The inclusion criteria required patients to be over 18 years of age, capable of providing written, informed consent, and have an expected survival of over 1 year. Patients had to be scheduled for RFA on a date when investigators were present, and the simulation software was available. The HCC diagnosis was confirmed by biopsy, alpha fetoprotein (AFP) >400 ng/ml, or Liver Imaging Reporting and Data System (LI-RADS) criteria; tumors were less than 5 in number and of 2-6 cm diameter [1,27]. Exclusion criteria were: tumor within 2 cm of another tumor or prior treatment site, tumor within 5 mm of first order biliary ducts, the presence of tumor in vein or extrahepatic metastases, contraindications to MRI, platelet count <30,000/ml, or an international normalized ratio >2. The final study population, consisted of 52 patients with 68 tumors ablated during 57 procedures (Figure 1, Tables 1 and 2).

Figure 1. Patient flow diagram.

This research was approved by the Dartmouth Hitchcock Health Institutional Review Board (STUDY02000081) and was performed in accordance with principles stated in the Declaration of Helsinki. Though the study was conceived and executed with technical input from the software developer, they were not involved in the procedures, data collection, image interpretation, or data analysis.

Simulation-guided radiofrequency ablation procedure

All procedures were performed on a dedicated interventional CT (Siemens Somatom Sensation AS 40, Siemens Healthcare AG, Munich, Germany) under general anesthesia, with full muscle paralysis to ensure optimal image registration. This facilitated a reproducible end-expiratory liver position during imaging and electrode placement [28]. All patients received perioperative intravenous antibiotic prophylaxis.

CT imaging (1 × 40 × l0.6 mm; 330s rotation time, 0.6 pitch, tube voltage 120 kV) was performed before, and at 45 and 110 s after intravenous injection of 110 ml of iohexol (Omnipaque 350, GE HealthCare, Chicago, IL) at 4 ml/s. The image series that best delineated the tumor was transmitted as 1.5 mm DICOM images to the Accublate^TM RF Sim software, running on a laptop in the interventional CT suite (Technical detail in Supplemental Appendix A) [23,24].

Following manual segmentation (1-2 min per tumor), the software added a user-designated 5 mm margin (Supplemental Figure A1a). Potential RFA electrode paths were assessed by positioning a virtual electrode on the software’s 3D multiplanar display. An outline of the ablation volume allowed optimization of the path in relation to coverage of the segmented target (Supplemental Figure A1b). After trajectory selection (1-3 min per tumor), the display could toggle from the oblique multiplanar views to the axial image to designate the desired skin entry site.

The LeVeen 15 ga needle guide (Boston Scientific, Inc, Marlborough, MA) was advanced freehand along the virtual electrode path, intermittently checked by low dose multiplanar CT images. For LIRADS 4 or M lesions, a coaxial 18 ga biopsy (Cook, Bloomington, IN) was obtained to confirm histopathology. The LeVeen multi-tined expandable (umbrella) electrode was then deployed, and a non-contrast CT confirming position was sent to Accublate for automatic rigid registration (elastix, Insight Segmentation and Registration Toolkit [29]) with the initial contrast enhanced study. If the registration was not visually satisfactory in 3 planes, manual adjustments could be performed using sliders on the graphical user interface (Supplemental Figure A3). When tumor abutted a non-expendable structure, the electrode was positioned first at the highest-risk position prior to hydrodissection, to optimize the accuracy of the simulation at that ablation site.

Manual designation of the tip and mid- electrode enabled software recognition, including the individual tine geometry, from which an expected ablation volume was superimposed on the segmented target (Figure 2a, Supplemental Figure A2). A key stroke toggled between the software simulation (based on nominal power) and the manufacturer’s map of the expected ablation volume. Energy was applied by the RF generator (RF3000, Boston Scientific, Inc, Marlborough, MA) per the manufacturer’s algorithm, and Accublate RF Sim sampled the power (W) and impedance (Ohms) each second to calculate the volume of coagulated tissue. The simulated ablation replaced the estimated volume outline (white outline to semiopaque green) (Figure 2b). Any target (tumor and margin) not covered by the simulated ablation was considered untreated target tissue.

Figure 2. Screenshots of the coronal graphical display of registered and fused images after electrode deployment, initial ablation, and reposition for overlapping ablation. (a) Initial expected ablation volume: The segmented tumor (red) and margin (yellow) from the initial CECT were superimposed on the registered unenhanced electrode study. After designation of electrode, software provided multiplanar orthogonal images that depict tumor, margin, and expected ablation volume from nominal energy application (white outline). (b) The simulated ablation volume, calculated from the applied RF generator energy iis depicted in semi-opaque green. The untreated volume (unshaded) is distinct. (c) Plan expected overlapping ablation volume: Reconstructed registered and superimposed unenhanced image of repositioned electrode; the electrode was designated, and expected additional ablation volume depicted (white outline). (d) Assess accumulated simulation ablation volumes: The additional simulated ablation volume (semi-opaque green) has been added to depict the accumulation of treated tissue.

Untreated target was addressed by repositioning the electrode for an overlapping ablation, if deemed feasible and safe. An unenhanced CT with the new electrode position was sent to the software, registered, and the electrode identified. The outline of the expected additional ablation (white outline) was again depicted, now relative to the target and simulated ablation volume from the first ablation (Figure 2c). Based on the application of RF energy at each new position, the simulated additional ablation volume was accumulated in the software’s internal tissue damage map to depict the total ablated volume (Figure 2d). Independent simulations were performed for each application of RF energy at each electrode position. The software’s internal tissue damage map accumulated all the simulated ablation volumes to display the cumulative effect of the treatment. For overlapping ablations, this version of Accublate RF Sim did not account for residual heat nor for alterations in tissue conductance properties due to a prior ablation. At any point, when the cumulative simulated ablations were superimposed on the target, the total simulated achieved ablation margin was clearly evaluable on the multiplanar images, and the calculated percentage of both the tumor and the margin encompassed by the simulated ablation was displayed numerically.

After tract ablation, a final non-contrast CT assessed immediate complications. Patients were admitted overnight, with clinical follow-up at 1-4 weeks. Follow-up Liver Tumor Clinic visits were scheduled every 3 months for at least 2 years, including serum AFP and a CECT or MRI. Clinical and imaging records were reviewed for complications, progression of disease, and additional interventions by hepatologists and one of four abdominal radiologists (5-19 years’ experience) performing a second read of the images (those physicians were unaware of patients’ participation in the study).

Oncological outcomes

The primary outcomes included technical success, technique efficacy, and local tumor progression (LTP), assessed for each tumor. Technical success was achieved if the treatment was completed per protocol and the target (tumor and 5 mm ablative margin) encompassed [1,12] In accordance with existing practice, residual or untreated tumor was first objectively evaluated at the 3-month follow-up contrast enhanced CT or MRI, at which time primary technique efficacy was assessed. Efficacy required complete ablation, defined as the absence of arterial enhancement (complete response by mRECIST and non-viable by LI-RADS® criteria) [1,27,30]. LTP was defined by viable tumor within or at the edge of the ablated volume after at least one study had documented treatment efficacy [1].

Secondary outcomes were hepatic recurrence (a new, non-local, intrahepatic HCC), and overall patient survival (from date of the first simulation-guided ablation). Adverse events (AE) that occurred within 30 days were classified by Society of Interventional Radiology Standards of Practice criteria [1,31].

Statistics

Sample size was calculated to detect a 70% improvement in 2-year LTP compared with an historical control rate of 27.2% (from similar risk profile population studies (tumor diameter of 2.4–2.9 cm, >5% with multiple tumors) (Table 3) [4,5,32,33]. For a type-I error rate of 0.05 and power of 0.8, with 10% loss rate, subjects with 68 tumors were enrolled.

Table 1. Baseline characteristics of hepatocellular carcinoma population treated by RFA with ablation simulation guidance: 57 procedures.

		n	%
Age (mean, standard deviation)		68.2 (±8.9)
Sex	male	46	80.7
ECOG performance status	0	39	68.4
	1	16	28.1
	2	2	3.5
Cirrhosis etiology	Hepatitis C	28	49.1
	Hepatitis B	3	5.3
	Alcohol	36	63.2
	MASH	19	33.3
	Other	2	3.5
Child-Turcotte-Pugh class	A	32	56.1
	B	18	31.6
	C	7	12.3
AFP	20–200	4	7.0
	200–400	2	3.5
	>400	2	3.5
Number of tumors treated*	1	43	75.5
	2	8	14
	3	4	7
	5	2	3.5
Within Milan criteria		46	80.7
Prior liver-directed treatments	Initial treatment (%)	Additional	total
RFA	14 (24.6)	21	35
MWA	2 (3.5)	4	6
Bland embolization	2 (3.5)	0	2
Radio-embolization	1 (1.7)	0	2
Radiation therapy	3 (5.3)	2	5
Resection	4 (7.0)	1	5
Total	26 (45.6)

AFP = alpha fetoprotein; ECOG = Eastern Cooperative Oncology Group; MASH = Metabolic dysfunction-associated steatohepatitis; MWA = microwave ablation; RFA = radiofrequency ablation.

*Only included in study if diameter >2 cm.

Time-to-events were analyzed using Kaplan-Meier methodology. LTP (per tumor) and hepatic recurrence (per procedure) were censored as of the last enhanced imaging study while survival (per patient) was censored at the last clinic visit in the medical record. Multivariate analyses employed the Fine and Gray competing risk model to account for deaths prior to events, and clustering for patients with multiple tumors. Two sample tests were used to compare LTP rates with historic controls. The Cox proportional hazards model was used for overall survival analysis. Statistical significance was set at p < 0.05. Analyses were performed in R software version 4.3.

Results

Of 61 consecutively consented patients, five were excluded after pre-procedural CECT (Figure 1). No exclusions occurred due to anatomical location of the tumor or failure of image registration due to the administration of artificial ascites. Simulation-guided RFA was performed according to protocol in 57 procedures in 52 unique patients to treat 68 tumors (Tables 1 and 2). Five patients were re-enrolled for ablation of an intrahepatic recurrence. At the time of the study procedure, 15 (26.3%) subjects had multiple tumors treated, but only those >2 cm were included. Eight (14%) patients had serum AFP > 20 ng/ml, and 23 (39.1%) had prior treatment of an HCC. Contrast administration was a concern in 44.1% due to elevated creatinine (n = 9, 15.7%) or diuretic treatment for ascites (n = 18, 31.6%). The 68 tumors had a median diameter of 2.73 cm (IQR 2.2, 3.1 cm). The LI-RADS^® 4 and M tumors were confirmed as HCC on biopsy.

Table 2. Baseline characteristics of hepatocellular carcinoma population treated by RFA with ablation simulation guidance: 68 tumors.

		n	%
Size (cm) (mean/range/IRQ)	2.73	2–5.4	2.2,3.1
> 3 cm		22	32.4
Solitary lesion		42	61.8
Subcapsular (within 1 cm)		48	70.6
Within 5 mm of > = 3 mm Vessel		18	26.5
Biopsy grade (differentiation)	Well	16	28.1
	Moderate	17	29.8
	Poor	5	8.8
	na	19	33.3
Location (Couinaud segment)	II	3	4.4
	III	13	19.1
	IV	7	10.2
	V	9	13.2
	VI	7	10.3
	VII	11	16.2
	VIII	16	23.5
LI-RADS^TM	4	19	27.9
	5	46	67.6
	M	3	4.4

IRQ = interquartile range; LI-RADS^TM= Liver Imaging and Reporting Data System (31); RFA = radiofrequency ablation.

Table 3. Outcomes after percutaneous ablation of HCC.

Study					Tumor		Sessions per patient	Technical efficacy %	LTP %
Author	Year	Type	n	Guide	Diameter	Multiple			1y	2y
Nakazawa [3]	2007	r	85	us	2.0	0	1.4	100	7.5
Shin [8]	2014	p	150	us	2.0	0	1.1	90.67	8.0
Jiang [5]^hc	2018	r	134	us	>2.0	15	1	100	13.4	30.2
Joo [19]	2022	r	172	ct	2.0	…	1.3	…	6.0*
Kim [33]^hc	2020	r	313	…	2.2	7.3	…	…	25.5	31.1
Liao [11]	2017	p	96	us	2.4	0	1.3	89	6.8	16
Di Costanzo [32]^hc	2015	p	70	us	2.6	10	1	97.4	17	24
Kim [4]^hc	2010	r	103	us	2.7	7	1	100	12.7	26.4
Solbiati [7]	2019	r	50	us	2.7		…	…	23.3
Current	2023	p	52	ct	2.7	24	1	100	3.9	20.1

Sessions = number of ablation procedures that comprised a course of treatment. Technical efficacy = patients without evidence for residual tumor at first follow-up study. LTP = local tumor progression (after a study without residual tumor).

^hc = historical control group; p = prospective, r = retrospective, n = number of subjects, … = not reported.

*HCC subpopulation.

The simulation showed typically a smaller volume of ablation compared to the manufacturer’s ablation map (Figure 3). For 51 (75.9%) tumors the simulation indicated incomplete target coverage and guided the targeting of additional overlapping ablations (median 3 (IRQ 2, 4)). The single session treatments were technically successful for all tumors. At 3-month contrast enhanced MRI or CT follow-up, technical effectiveness was demonstrated for 68 (100%) tumors.

Figure 3. Comparison of the predicted (a), simulated (b), and achieved ablation volume (c), adjacent to an internal marker (cyst—dashed white outline). Manufacturer’s predicted ablation volume (a) can toggle to the Accublate simulation (b) intraprocedurally. CECT at same level 2 days after ablation (c). Accublate simulation volume is smaller than manufacturer’s prediction, and more closely describes the shape. (Volumes in a and b are displayed superimposed on the electrode; in c, the cyst provides an internal reference to identify the similar level.)

LTP occurred in 9 (13.2%) of 68 ablated tumors at a median time-to-event follow-up of 20.3 months (IRQ 11.1, 26.4 months). The 1- and 2-year KM estimates of LTP were 3.9 and 20.9%, respectively (Figure 4). When adjusted for competing risk and clustering, these LTP rates were 3.5% and 16.7%, respectively. Unadjusted KM estimates were improved compared to the historic control rate at 1 year of 17.2% (p = .004) and 27.2% (p = .141), respectively. Subpopulation multivariate regression showed no significant risk factors for patients that had tumors >3 cm, within 5 mm of >3 mm vessel, AFP> 20, or prior interventions for HCC.

Figure 4. Cumulative Incidence Function of time to local tumor progression after RFA with ablation simulation guidance.

Intrahepatic recurrence occurred in 28 (49.1%) of 57 procedures at a median time-to-event imaging follow-up of 12.1 months (IRQ 5.7, 21.5 months). The time-to-event estimates for intrahepatic recurrence at 1 and 2 years were 36.0% and 62.0%, respectively (Supplemental Appendix B, Figure B1). The competing risks regression model identified patients with multiple tumors as having an increased risk with hazard ratio of 3.17 (95% CI 1.49–6.68, p = 0.003), manifest as a 1-year intrahepatic recurrence rate of 65.1%, versus 26.3% for those with a single tumor (p < .001).

Of the 52 patients, at median 24.5-month (IRQ 13.6, 32.0 months) clinical follow-up, six (11.5%) patients had received liver transplants, and 21 (40.4%) had died. The 1 and 2-year KM estimated overall survival after the ablation procedure was 60.6% and 58.1%, respectively (Supplemental Appendix B, Figure B2). Childs-Turcotte-Pugh class was associated with increased mortality (p = .041).

Length of stay was one day for 49 (86%) patients, 5 (8.8%) were discharged same day, and 3 (5.4%) stayed 2, 4, and 10 days. AEs occurred in 10 patients, 8 were mild-moderate, 1 severe, and 1 life-threatening [31]. The life-threatening AE was associated with the procedure; an 88-year-old woman became hypotensive with altered mental status, and CT identified a large subcapsular hematoma at the ablation site. She stabilized after a 2-unit transfusion in the ICU but had a prolonged recovery. The severe AE was also associated with the procedure, a re-admission for mental status changes due to difficult-to-control hypoglycemia.

Discussion

This prospective study evaluated the safety and efficacy of Accublate RF Sim, a physics-simulation computer-assisted ablation guidance software. The software superimposed a simulation of the ablated tissue volume(s) onto the target (tumor and margin) providing feedback on which target tissues were treated, and which remained untreated, without the need for IV contrast (Figures 2 and 3). The simulation software was safely incorporated into the workflow for RFA, yielding 100% technical efficacy after a single ablation session, and a low 2-year LTP rate of 20.2% for tumors of 2.7 cm median diameter after a single ablation session, comparing favorably to reported rates of 24 to 31% for similar populations (Table 3).

Multiple overlapping ablations are often employed to ensure coverage of larger or irregularly shaped targets [3]. The results of this study support the use of a simulated, fused ablation image to better direct overlapping electrode repositions versus a traditional mental overlay of the manufacturer’s estimate [4,7,8,16,17,25,34]. Despite the simplicity of this early version of the software (the simulation did not yet incorporate large vessel heat-sink or perfusion data), and the high-risk population, the outcomes were comparable to recent results of stereotactic guided ablation with MAM analysis.

Computer-aided graphic software with segmentation and registration capabilities was employed retrospectively for the analysis of the correlation between ablation margin and LTP [4,11,12]. However, adoption of routine intraprocedural assessment of ablation margin adequacy has been limited by the 10-30 min necessary to precisely register images [5,12,16, 35, 36]. In contrast, rigid image registration of images obtained during ventilator disconnection permitted an accurate, rapid (<40 s) alignment, and if needed, manual adjustments were performed in under two minutes [4].

A recent meta-analysis of intraprocedural stereotactic liver tumor ablation (computer-assisted visualization with robotic-guided targeting) demonstrated improved targeting accuracy, and a higher rate of complete ablation after the first treatment session (Pooled OR 1.94 (CI 1.18, 3.19)) [37]. The effectiveness of computer-assisted minimal ablation margin assessment for intraprocedural guidance was prospectively studied by Shin and colleagues [8]. Margin adequacy for RFA of 150 solitary, mean 2 cm diameter HCC was analyzed on a Leonardo workstation using HepaCare non-rigid registration (Siemens Healthcare), which led to additional ablations in 14% of patients to ensure 5 mm margins. In comparison with a conservatively treated group, the 1 year LTP was reduced from 17% to 4% (p = .006) [8].

Lachenmayer and colleagues, employed the stereotactic CAS-One IR system (CAScination AG, Bern Switzerland) to guide microwave ablation of 174 HCC with a mean 1.6 cm diameter [36]. Tumor segmentation and registration with the post-ablation CECT added a mean 14.3 min procedural time. Although insufficient ablative margins prompted immediate additional ablation in 10.9%, an additional 6.9% required a second ablation session for residual neoplasm at initial follow-up to achieve a 6-month technical efficacy of 96.3%. The 1 year LTP rate was 6.3% [36]. In the current study, intraprocedural guidance by simulation without a post-ablation intravenous contrast administration yielded comparable outcomes to the reported state of the art stereotactic guidance.

Ablation simulation remains investigational [38]. Clinical application was reported in a prospective feasibility trial of the RFA Guardian stereotactic guidance software [26]. The 3-min simulation demonstrated an average absolute error of 3.4 ± 1.7 mm when compared with the segmented ablation volumes on the 1-month follow-up CECT. Though deemed useful for ablation therapy planning, that study did not use the simulation for intraprocedural decision making, so the contribution to technical efficacy could not be assessed [26].

Simulation-based liver tumor ablation guidance addresses the difficultly of intraprocedural visualization of what has and has not been treated [38]. Simulation permits an assessment of the extent of treated tissue after, and potentially during, each electrode activation; and so, is a form of ablation confirmation software that can depict the margin achieved without additional contrast administration. Because any offset of the ablation volume due to an error in targeting is immediately displayed, the imprecision may be mitigated with an additional ablation.

Limitations of the current study included the convenience cohort without a comparator, the single applicator device, the single operator and single center, and heterogeneity of subsequent therapies. While consecutive patients were considered, the scheduled ablation date and availability of investigators determined participation. Though the LeVeen device has a particular umbrella tine configuration, the improved accuracy from simulation guidance (versus the manufacturer’s expected ablation volumes) can be extended to any RFA applicator with appropriate software modification and validation [15]. Comparison with prior studies was limited by the lack of prospective literature that assessed the impact of computer-assisted ablation, and variation in the definitions of a course of treatment and LTP.

The improvement in 1-year LTP was statistically significant, but that of the 2-year LTP was not. The former may have been artifactually low as logistic delays resulted in some ‘1-year follow-up’ studies obtained at 13-15 months, and some patients with non-local recurrence before LTP may have benefited from adjunctive chemotherapy (n = 6). And though the 26% improvement observed in the 2-year LTP was potentially clinically meaningful, this study lacked the statistical power to demonstrate significance. This was due to the small sample size which was based on a larger targeted reduction, and an unexpectedly high rate of the competing risks of mortality and transplant.

The Accublate RF Sim software provided an immediate computer-generated simulation of the achieved ablation volume after each activation for clear visualization of any untreated portion of the target. This informed the position of additional overlapping ablation(s) to ensure complete coverage during a single procedure, without an additional contrast enhanced scan. In this prospective study, the ablation visualization provided by the simulation software was associated with high efficacy and low LTP rates.

Prior publication

Early results of this study were presented in an abstract at the Society of Interventional Radiology 2022 Annual Meeting, Boston, MA: Abstract 168: Hoffer EK, Patel S, Borsic A. Real-time physics-based simulation of the ablation volume: early clinical outcomes in radiofrequency ablation of hepatocellular carcinoma. Published at: Journal of Vascular and Interventional Radiology 2022;33(6), Suppl S77–S78.

Acknowledgement

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Disclosure statement

Eric Hoffer, MD, is the Chief Medical Officer of NE Scientific, LLC. He has a profits interest agreement. Dr. Hoffer was responsible for study design, data analysis, and writing of the manuscript. Andrea Borsic, PhD, is the CEO, founder, and owner of NE Scientific, LLC, developer of the AccublateTM RF Sim computer guidance tool for ablative procedures. This research was sponsored by NE Scientific, LLC, through an NIH/SBIR grant, and may lead to the marketing of products. Dr. Borsic advised on study design, provided technical support, and reviewed the manuscript for technical accuracy.

Data availability statement

Data will be available on ClinicalTrials.gov site, or on request from the corresponding author.

Additional information

Funding

The study was supported by a National Cancer Institute of the National Institutes of Health/SBIR grant (R44CA189515) and was registered at ClinicalTrials.gov (NCT04152343) on 11/2/2019. No funding was contributed by nongovernmental sources.