Shrinkage of hepatocellular carcinoma after radiofrequency ablation following transcatheter arterial chemoembolization: Analysis of contributing factors

Objective This study was conducted to investigate tumor shrinkage and influencing factors in patients with hepatocellular carcinoma (HCC) from radiofrequency (RF) ablation following transcatheter arterial chemoembolization (TACE). Methods A total of 222 patients underwent combined sequential treatment of TACE and RF ablation for HCC at our institution between 2008 and 2014. Of those, 86 patients (men, 68; women, 18) who achieved compact iodized oil tagging and complete ablation were included for this retrospective study. We measured three-dimensional tumor diameters and calculated tumor volumes on pre-treatment CT/MRI and follow-up CT scans performed post-TACE, post-ablation, and 1 month post-treatment, respectively. To compare periodically generated tumor diameters and volumes, repeated measures analysis of variance (ANOVA) was applied. Multiple linear regression analysis was performed to identify factors impacting tumor shrinkage after RF ablation. Results Diameters and volumes of HCCs declined significantly in the immediate aftermath of RF ablation (i.e., between post-TACE and post-ablation CT scans) (p < 0.001, for both). Mean reduction rates in tumor diameter and volume immediately after RF ablation were 18.2 ± 9.1% and 44.4 ± 14.6%, respectively. Of note, tumors of left hepatic lobe and in subphrenic or perivascular locations showed lower rates of post-ablative volume reduction than those in counterpart locations (p = 0.002, 0.046, 0.024, respectively). Tumor size and liver function did not influence tumor shrinkage after RF ablation. Conclusion In patients with HCC, significant tumor shrinkage occurs immediately after RF ablation. The degree of shrinkage in response to ablative treatment seems to vary by tumor location.

Introduction Radiofrequency (RF) ablation is a common curative treatment for hepatocellular carcinoma (HCC) [1]. It is a locally applied thermal ablation technique intended to destroy tumor using heat [2]. Thermal ablation induces protein denaturation and dehydration, as well as contraction of collagen and tissue shrinkage [3,4]. In clinical and experimental studies, it has been noted that ablated tissues undergo involution [5][6][7][8]. However, it is not possible to clinically identify tumor shrinkage induced by RF ablation, because in CT studies, ablated tumor is indistinguishable from surrounding ablated tissue. Indeed, studies [9][10][11] conducted to date in this regard have measured ablation zones (including both tumor and surrounding ablated tissue), rarely measuring ablated tumor directly.
RF ablation combined with transcatheter arterial chemoembolization (TACE) has been lauded for its superior efficacy in the treatment of HCC, compared with RF ablation only [12][13][14][15]. This combined therapy provides a unique opportunity for the accurate assessment of tumor shrinkage itself on post-ablative CT. Iodized oil deposited via TACE is consequently visible as a high-attenuation nodule, easily discriminated from surrounding the low-attenuation ablation zone on post-ablative CT images [13,16]. One may then directly measure tumor diameters in both pre-and post-ablative CT scans to assess changes in size and volume.
We presumed that significant tumor shrinkage occurs immediately after RF ablation and various clinical and imaging variables in this setting contribute to significant tumor shrinkage immediately after RF ablation, thus we analyzed data from patients undergoing RF ablation following TACE oil tagging of HCC. Therefore, the purpose of this study was to investigate the nature of dimensional and volumetric changes achieved through RF ablation of HCC and to identify factors correlating with post-ablative tumor shrinkage.

Materials and methods
This study was approved by our institutional review board (Konkuk University Medical Center, KUH1140126). Due to its retrospective design, written informed consent was waived. We regularly adhered to all proposed standard terminology and reporting criteria for imageguided tumor ablation in the course of our work [17].
CT, with �5-mm safety margin [18]; and 6) multiplanar reformation images (coronal or sagittal) available for three-dimensional (3-D) tumor measurements. A flow chart of patient selection is shown in Fig 1. Each diagnosis of HCC was established at time of treatment using pre-treatment baseline CT or MRI studies done prior to TACE (within 1 month) and as stipulated by the American Association for the Study of Liver Diseases (AASLD) [19]. To accurately determine tumor margins, iodized oil must be retained and concentrated at its borders. Thus, we excluded those tumors showing poor or no iodized oil retention following TACE procedures. Tumors incompletely ablated were also excluded, owing to expectedly skewed rates of post-ablative volume change. In patients with multiple HCCs, only the largest eligible tumor was studied.

Combination treatment protocol for hepatocellular carcinoma
Our institutional protocol for combination TACE/RF ablation treatment of patients with HCC is described herein and is shown as a schematic in Fig 2. TACE treatments were undertaken on an inpatient basis using an interventional procedure room equipped with a commercially available digital subtraction angiography system (Axiom Artis dBA; Siemens Healthcare, Erlangen, Germany). Two experienced interventional radiologists (17 and 13 years, respectively) performed all TACE procedures. Upon completing celiac and superior mesenteric arteriography, hepatic artery angiography was performed using a 5-Fr catheter (Cook Medical, Bloomington, IN, USA). Right or left hepatic artery was then selectively catheterized (3-Fr MicroFerret; Cook Medical) to locate a tumor-directed feeder artery. Selective embolization was carried out thereafter, injecting a mixed emulsion of iodized oil (Lipiodol; Andre Guerbet, Aulnay-sous-Bois, France) and doxorubicin hydrochloride (Adriamycin RDF; Ildong Pharmaceutical, Seoul, Korea). Embolization continued until arterial flow stagnated and/or iodized oil was visualized within portal branch vessels. Gelatin sponge particles of 1-2 mm diameter (Gelfoam; Upjohn, Kalamazoo, MI, USA) were also infused. Once accomplished, angiography was again performed to assess the extent of vascular compromise and check for any residual tumor staining. Post-TACE CT obtained the next day served to verify that oil retention by tumor was adequate.  Percutaneous RF ablation took place within 0-3 days after TACE, performed by one of two radiologists (YJK and HSP with 18 and 10 years of experience in this setting, respectively). All procedures routinely involved local anesthesia with conscious sedation; both ultrasound and fluoroscopy guidance; and use of a 15-G or 17-G monopolar, internally cooled electrode bearing a 2-3 cm active tip (VIVA; STARmed, Goyang, Korea) and powered by a 200W generator (VIVA; STARmed). We used a 2-cm tip for smaller tumors (1-2 cm), reserving the 3-cm tip for larger-sized lesions. Depending on tumor size and configuration, single or multiple overlapping ablations were executed. The energy deposition algorithm applied reflected manufacturer's recommendations. Ablation was terminated once an entire tumor and surrounding hepatic tissue margins fell within in the ultrasound echogenic zone. To prevent bleeding or tract seeding, the electrode path was also cauterized during retraction [20]. Post-ablation CT was undertaken in the immediate aftermath to check for related complications (such as bleeding) and to gauge technical success. Each patient was then subjected to a multiphasic liver CT 1 month after RF ablation as the initial follow-up assessment of tumor remission.

Scan techniques
All CT examinations entailed use of a 64-MDCT (Somatom Definition [Siemens Healthcare]; LightSpeed VCT [GE Healthcare, Chicago, IL, USA]). The respective scanning parameters used for the two 64-MDCT systems were as follows: detector collimation, 64 × 0.6 mm and 64 × 0.625 mm; pitch, 0.984; and rotation time, 0.5 second. The reference tube current was set at 250 and 200 mAs at 120 kVp, with automated dose modulation. Axial images were reconstructed at section thickness /reconstruction intervals of 3 mm/3 mm (Somatom Definition) and 3.75 mm/3.75 mm (LightSpeed VCT). Coronal images were similarly reconstructed (3 mm/3 mm) using portal venous phase scan.
Pre-treatment, post-ablation, and 1-month follow-up assessments entailed multiphasic liver CT, providing unenhanced, late arterial, portal, and equilibrium phases. A total of 370 mg I/ mL of iodinated contrast medium, iopromide (Ultravist 370; Bayer Healthcare, Berlin, Germany), was administered via power injector for 30 seconds at a dose of 1.5 mL/kg (555 mg I/ kg) body weight, followed by injection of normal saline solution (30-40 mL). Late arterial, portal venous, and equilibrium-phase images were obtained at 25 sec, 70 sec, and 180 sec after the start of contrast administration. Post-TACE CT consisted of a pre-contrast scan only, without iodized contrast medium.
In nine patients who lacked pre-treatment CT scans, we used MRI scans for image analysis instead of CT scans. Liver MRI was performed via 1.5-T (n = 5, Signa HDxt; GE Healthcare) or 3-T (n = 4, Magnetom Skyra; Siemens Healthcare) superconducting system, using a 32-channel phased-array coil. Dynamic 3-D fat-saturated T1-weighed sequences were obtained after gadoxetic acid (Primovist; Bayer Healthcare) administration. Axial and coronal images were scanned at spatial resolutions of 1.1-1.8 mm and 2.7-to 5.2-mm section thickness in hepatobiliary phase, 20 min after the start of contrast medium injection.

Image analysis
Tumor measurement. Two clinically experienced abdominal radiologists (YJK and MHY) with years of cumulative expertise (18 and 7 years, respectively) measured 3-D diameters of each index tumor using pre-treatment CT/MRI and CT scans done post-TACE, postablation, and at the 1-month follow-up point. Imaging reviews were facilitated by PACS software (Centricity RA1000; GE Healthcare), shown on monitors at 2048 × 2560 spatial resolution. For precision in measurement, all images were displayed at one-by-one setting, and the PACS magnification function was used. Two reviewers initially screened images, selecting (by consensus) those that best depicted index tumors in axial, coronal, or sagittal planes. Actual 3-D diameters were generated from axial images (maximum diameter [Dmx] and perpendicular minimum diameter [Dmi]) and coronal or sagittal views (vertical [craniocaudal] diameter [Dv]) (Fig 3).
Measurements based on pre-treatment CT scans were obtained at a preset soft tissue window setting (width, 350 HU; level, 50 HU). To avoid the beam-hardening artifact due to iodized oil retention by index tumors or adjacent parenchyma, a preset bone window setting (width, 3000 HU; level, 500 HU) was adapted for post-TACE, post-ablation, and follow-up CT scans. In nine patients lacking pre-treatment CT scans as stipulated (within 1 month before TACE), MRI served for image analysis. Measurements were obtained from axial and coronal images in hepatobiliary phase of scanning, owing to the high contrast resolution between index tumor and liver parenchyma. Tumor volume was ultimately calculated by applying above 3-D diameters to the following equation: Rates of tumor diameter and volume reduction over time, from pre-treatment CT/MRI to CT scans performed post-TACE, post-ablation, and in follow-up, were calculated using the following formula: Reduction rate ¼ baseline tumor value À follow up tumor value baseline tumor value � 100 Factors impacting tumor shrinkage. To identify factors of potential influence in tumor shrinkage after RF ablation, we evaluated several clinical or laboratory parameters and some imaging features, including the presence of portal hypertension, tumor size, and tumor location. With the presumption that portosystemic collaterals are surrogate markers for portal hypertension [21], pre-treatment CT/MR images were duly screened for their presence. We also established several analytic variables in terms of tumor location, namely lobar (right vs left hepatic lobe), subcapsular (vs non-subcapsular), subphrenic (vs non-subphrenic), and perivascular (vs non-perivascular) sites. Subcapsular location was marked by hepatic capsular abutment of index tumor in axial or coronal images [22,23]. Subphrenic location similarly was equated with diaphragmatic abutment of index tumor in axial or coronal images [24,25]. Contact between index tumor and any first-or second-degree branches of portal vein or branches of hepatic vein �3 mm in axial diameter qualified as perivascular location [26][27][28].
Therapeutic outcomes. Local tumor progression, overall survival rate, and complications after combined TACE and RF ablation treatment were assessed. Local tumor progression is defined as the newly developed tumor at the margin of the ablation zone on follow-up images. Overall survival rate was analyzed using the interval between the RF ablation and either the death or the last visit to the hospital. Major and minor complications were assessed in accordance with the Society of Interventional Radiology guideline [29].

Statistical analysis
Tumor diameters and volumes and rates of shrinkage were individually expressed as mean ± standard deviation (SD). To assess dimensional/volumetric differences in tumors at various imaging time points, including pre-treatment, post-TACE, post-ablation, and 1 month post-treatment, repeated measures analysis of variance (ANOVA) was performed, applying the Bonferroni adjustment for multiple comparisons.
Student's t-test and the Kruskal-Wallis test were used to compare post-ablative rates of tumor volume reduction according to clinical and imaging features. The following patient variables were analyzed: age, gender, Child-Pugh score, serum alpha-fetoprotein, time interval (days) between TACE and RF ablation, portal hypertension (+/-), tumor size, and assorted tumor locations (lobar, subcapsular, subphrenic, or perivascular). Factors independently associated with post-ablative tumor shrinkage were identified by multiple linear regression method. The cumulative local tumor progression rates and overall survival rates were estimated by using the method of Kaplan-Meier. All computations relied on standard software (SPSS v17.0; SPSS Inc, Chicago, IL, USA), setting statistical significance at p < 0.05.

Results
Characteristics of the patient population are summarized in Table 1.
Diameters (3-D) and volumes of treated tumors and reduction rates achieved are recorded in Table 2.
However, these parameters did not differ significantly in periods between pre-treatment CT/MRI and post-ablative CT scans or between post-ablative and 1-month follow-up CT scans (p > 0.05).
Relations between post-ablative rates of tumor volume reduction and corresponding clinical or imaging features in patients with HCC are summarized in Table 3.
Results of multiple linear regression analysis are presented in Table 4. Tumor shrinkage after radiofrequency ablation Again, left lobar, subphrenic, and perivascular tumor locations emerged as variables independently associated with diminished rates of tumor volume reduction after RF ablation (p = 0.005, p = 0.009, p = 0.012, respectively).
There were treatment-related deaths and no major complications developed. There were 16 minor complications (18.6%): post-ablation syndrome (n = 10), diaphragmatic thermal injury (n = 3), perihepatic hematoma (n = 2), and pleural effusion (n = 1). All patients with minor complications were treated with analgesics. None of the patients experienced long-term hepatic dysfunction or hepatic failure after the procedure.

Discussion
Results of the present study demonstrate that significant declines in tumor volume occur immediately after RF ablation performed in patients with HCC, regardless of baseline tumor Table 2. Three-dimensional diameters and volumes of hepatocellular carcinomas and reduction rates achieved by radiofrequency ablation after transcatheter arterial chemoembolization.  size or liver function status. Mean diameter and volume reduction rates immediately following RF ablation were 18.2 ± 9.1% and 44.4 ± 14.6%, respectively. In addition, the degree of tumor shrinkage observed after RF ablation differed significantly according to tumor location, with left lobar, subphrenic, and perivascular locations showing significantly less tumor shrinkage than tumors at counterpart locations.

Pre-treatment CT/MR Post-TACE CT Post-ablation CT Follow-up CT
In an ex vivo study conducted by Brace et al. [6], ablative tissue contraction was investigated in liver and lung tissues during RF and microwave ablation. They reported hepatic tissue contraction of 2.9-4.8 mm (15-31%) after RF ablation, with more prominent shrinkage manifested in the peripheral coagulation zone. A clinical study pursued by Ganguli et al. [5]  similarly confirmed a 5.4 mm (21%) mean decline in renal tumor diameters immediately after RF ablation. In this instance, it was possible to assess post-ablative tumor shrinkage through changes in contour rather than direct visualization, because most of the renal tumors treated were exophytic in nature. Our patients with HCC likewise showed a mean decline of 3.2 mm (18.2%) immediately after RF ablation, aligning with the assorted investigations above.
Recently, Lee et al. [30] also assessed degrees of RF-and microwave-induced tissue shrinkage in liver tumors, while curiously reporting a trend (inconsistent with earlier data) toward minimal tissue swelling (1.3%) after RF ablation. Because the outlines of ablated tumors were still visible within treated zones in post-ablation MRI studies, they directly measured post-ablative tumor size as well. Unfortunately, this approach succeeded in only about 40% (18/44) of the patients studied, the few tumors (n = 18) evaluated and the wide variation in tumor contraction rate (1.31 ± 13.2%) constituting distinct drawbacks.
To prevent local tumor progression after RF ablation, complete tumor ablation is imperative, and assessment of treated margins is therefore crucial [18,31,32]. In current clinical practice, pre-and post-ablative images are compared side-by-side, using hepatic vessels and liver surface as landmarks to assess ablative margins [33]. Various techniques, such as image fusion, rigid or non-rigid image registration, and quantitative assessment, have been advanced during the past decade to improve post-ablative margin assessment [34][35][36][37]. However, tumor shrinkage following RF ablation (as shown herein) has not been addressed in any of the studies reported to date. Because such tumor shrinkage is problematic, leading to underestimation of original ablative zones, it should be considered in studies going forward.
The degree of tumor shrinkage after RF ablation may be impacted by tumor location. Heatsink effect is a known phenomenon whereby blood flow draws thermal energy away from targeted tissue, reducing coagulation volume [28,38]. It is highly plausible that shrinkage of perivascular tumors may be compromised in this way. Subphrenic tumors are often obscured on sonography due to overlapping lung tissue or ribs, which may hamper ablative efforts [24,39]. Nonetheless, only completely ablated tumors were allowed in this study, eliminating this  Tumor shrinkage after radiofrequency ablation possibility. In theory, continuous sliding of the diaphragm over a fixed subphrenic tumor may dissipate heat during ablative treatment, and we suspected that ventilatory convective heat loss would also come into play. Still, substantial tumor shrinkage (>30%) appears inevitable as an immediate effect of RF ablation, regardless of tumor location; and although Dodd et al. [40] have reported a significant inverse relation between the extent of RF ablation and the rate of portal venous flow, the presence of portal hypertension had no impact on degree of post-ablative tumor shrinkage in our study. Knowledge of this variability of tumor shrinkage according to tumor location would be helpful to perform RF ablation and interpret post-ablation imaging in clinical practice. This study has a number of limitations. Its retrospective design inherently implies selection bias. Furthermore, volumetric quantification software (still a clinical scarcity) was not engaged for the measuring of tumors. However, CT slice thickness and slice interval were minimized, and the PACS magnification function aided in measurement precision. Third, in nine patients MRI scans were used instead of pre-treatment CT scans for image analysis. The difference in the modality may have influenced the perceived tumor measurement. Finally, the combined use of TACE/RF ablation was needed to clarify tumor margins on post-ablative CT studies. The diminished circulatory heat loss achieved via TACE likely served to enhance RF ablative zones [13].
In conclusion, significant tumor shrinkage occurs immediately after RF ablation in patients with HCC. Tumor location is an important factor, significantly impacting the degree of postablative tumor shrinkage.