Endoscopic Ultrasonography-Guided Tumor Ablation

With the introduction of curvilinear endosonoscopes, endoscopic ultrasonography (EUS) has achieved the role of a therapeutic modality as well as diagnostic procedure. EUS-guided tumor ablation is one such therapeutic modality. Various techniques of EUS-guided tumor ablation have been described, including radiofrequency ablation, photodynamic therapy, laser ablation, and ethanol injection. Most of the currently described techniques are experimental. Development and continuous improvement of devices, as well as establishment of indications for EUS-guided tumor ablations, are mandatory.

Introduction

When endoscopic ultrasonography (EUS) was first introduced in the 1980s, its primary role was a diagnostic tool. With the introduction of curvilinear-array endosonoscopes, the ability of performing fine-needle aspiration (EUS-FNA) became possible. Recently, the introduction of EUS-guided fine-needle injection has made possible the introduction of interventional EUS. The potential advantages of EUS-guided antitumor therapy are real-time imaging guidance, the ability to ablate tumor in poor surgical candidates, reduced morbidity compared with surgery, and the potential to be performed on an outpatient basis. This review focuses on EUS-guided radiofrequency ablation (EUS-RFA), EUS-guided photodynamic therapy (EUS-PDT), EUS-guided laser ablation, and EUS-guided ethanol injection to solid tumors. EUS-guided pancreatic cyst ablation, injection of antitumor agents, fiducial marker implantation, and brachytherapy.

Techniques of EUS-guided tumor ablation

EUS-RFA

Radiofrequency ablation (RFA) uses electromagnetic energy to induce thermal injury to the target tissue. In monopolar RFA, a closed-loop circuit includes a radiofrequency (RF) generator, an electrode needle, a dispersive electrode (ground pad), and the patient. The electrode delivers the energy to the tumor, resulting in a volume of high current density and localized heating. The ground pad closes the electrical current path; it is designed to disperse energy over a large area to reduce the possibility of thermal injury to the skin. In bipolar RFA, current oscillates between 2 interstitial electrodes, obviating a ground pad. It also confines current flow to the area between the electrodes and decreases the perfusion-mediated cooling, which results in faster and more focal heating of the target area. RF energy is the most well-studied ablation source, and is one of the safest and most predictable techniques for thermal ablation. Various EUS-RFA electrodes are shown in Fig. 1 .

EUS-RFA was initially described by Goldberg and colleagues in 1999. Under EUS guidance, a 19-gauge needle electrode with 1.0- to 1.5-cm tip was passed transgastrically to the pancreatic tail in 13 Yorkshire pigs. A 500-kHz monopolar RF generator supplied the electrosurgical current for 6 minutes, maintaining the electrode tip temperature of 90° ± 2°C. During RF application, a hyperechoic region with diameter around 1 cm appeared around the distal needle tip within 1 minute of achieving of 90°C tissue temperature. EUS needle withdrawal revealed spherical or conical hyperechoic lesions 1 to 1.5 cm in diameter. Follow-up EUS was performed on 2 pigs. EUS 9 days later demonstrated a spherical and hypoechoic pancreatic lesion 1.9 × 2.0 cm in diameter; the lesion was unchanged in size on EUS on day 14 but appeared more hypoechoic. EUS of the second pig on days 9 and 14 showed an elliptical, thick-walled fluid collection with a size of 2.7 × 2.8 cm containing low-grade echoes ( Fig. 2 ). Enhanced computed tomography (CT) scans in all pigs immediately after RFA revealed hypodense, nonenhancing foci measuring 8 to 10 mm in diameter with an enhancing rim. On CT scans from 6 pigs 14 days after RFA, 2 distinctive findings were noted. In 4 of 6 pigs, the previously visualized hypodense focus was not clearly delineated. In the other 2 pigs, a progressively enhancing rim 2 to 3 mm in thickness surrounding the coagulation focus of 1 to 1.2 cm could be seen. In pigs euthanized immediately after RFA, areas of treatment were visualized as 8- to 10-mm tan to brown lesions surrounded by normal pancreatic tissue; microscopic evaluation confirmed the presence of coagulation necrosis in treated tissues with sharp demarcation between treated and untreated tissue. In the 2 pigs euthanized 24 to 48 hours after the procedure, the gross findings were similar to those in pigs killed immediately. However, histopathology revealed a 1- to 2-mm watershed zone of an early inflammatory response surrounding the coagulated tissue. In the 6 specimens harvested 14 days after RFA, 2 patterns of tissue response were noted. In 2 pigs, a 1- to 1.2-cm cavity filled with necrotic tissue was identified; histology demonstrated an area of coagulative necrosis surrounded by a 2- to 3-mm fibrotic wall. In 4 pigs, a focus of brownish fibrotic tissue 5 to 8 mm long and 3 to 4 mm in diameter was seen invaginating the edge of normal-appearing pancreas. On histology, these foci were composed of fibrosis and scarring with residual areas of stromal and fat necrosis.

Correlation between imaging and gross pathology was excellent for lesions larger than 5 mm; the margin of difference was within 2 mm. Rim enhancement seen on CT images obtained immediately after RFA corresponded with interstitial hemorrhage. The rim enhancement seen at 14 days represented granulation tissue. Fours pigs showed resolution of previously visualized nonenhancing foci on CT; only residual fibrotic tissue was found in the ablated zone at pathology.

All pigs tolerated the RFA well, and no clinical evidence of distress was present in the surviving pigs. However, some complications were encountered. Three transmural gastric burns and one serosal small intestinal burn were attributed to improper electrode placement. There were no frank perforations of the gastrointestinal tract. One pig had elevated serum lipase, a focal pancreatitis, and a subsequent pancreatic fluid collection.

Modified RFA probes were tested subsequently. In 2008, Carrara and colleagues reported their experience with a hybrid cryotherm probe that combines bipolar RFA with cryotechnology. The heated probe is cooled by cryogenic gas (CO ₂in this probe), which increases the RF-induced interstitial devitalization, and thus compensated the reduced efficiency of bipolar RFA compared with monopolar RFA. The investigators performed EUS-RFA of various solid organs, documented the size of the ablation lesion measured immediately after ablation (T0), before euthanasia (T1) using EUS and immediately after euthanasia macroscopically (T2), and described histologic findings and complications.

For pancreas, 14 ablations were performed in 14 pigs. Energy input was 16 W, and cryogenic cooling was done by applying 650 psi of CO ₂. Application time ranged from 120 to 900 seconds. Seven pigs were euthanized at week 1, and the other 7 at week 2. Ablation resulted in elliptic lesions. When compared with the lesion size at necropsy, the ablated area was always overestimated by EUS, with a correlation coefficient of 0.89. A positive correlation between T0 and the duration of RF application was demonstrated. In addition, the correlation between T2 and the application time demonstrated a fixed ratio of 2.3 ( P <.0001) with a 1-week interval and 0.2 ( P = .01) with a 2-week interval. Although all pigs tolerated the procedure well and there was no mortality, there was 1 clinically symptomatic and laboratory-proven necrotic pancreatitis with peritonitis, 1 histologically and histochemically proven pancreatitis without clinical symptoms, 1 gastric wall burn, and 4 adhesions between the pancreas and the gut. It should be noted that longer application time resulted in greater variation in the lesion size.

The hybrid cryotherm probe was also evaluated on the porcine liver and spleen. For the liver, the histology after euthanasia revealed an area of liquefactive necrosis surrounded by coagulative necrosis and an inflammatory watershed zone composed of granulation tissue. The liver parenchyma surrounding the treated area was normal. The correlation coefficient for T1 versus T2 in the liver tissue was 0.71 ( P = .03) after removal of one potential outlier. The EUS significantly overestimated the lesion area T2. A positive trend of T1 over application time was noted ( r = 0.51, P = .1). For the spleen, the treated area showed a central hemorrhagic liquefactive necrosis surrounded by granulation tissue. The surrounding parenchyma was normal. The correlation coefficient for the correlation of T1 versus T2 in the spleen was 0.73 ( P = .04). There was a clear correlation of T2 and application time ( r = 0.75, P = .01). For both organs no complications, including the changes in laboratory tests, were observed.

Recently, an umbrella-shaped retractable electrode array, designed to provide a large area of coagulative necrosis, was investigated in pigs. EUS-RFA was performed in 5 normal pig livers. There was no difficulty in the transgastric deployment and retraction of the umbrella-shaped electrode. The ablated area evaluated by EUS 15 minutes after ablation showed a spherical and hypoechoic area with a diameter of 2.3 cm in all 5 pigs. On gross analysis, the mean ablated zone in the liver was 2.6 cm (range 2.5–2.7 cm). Histopathology confirmed the presence of coagulation necrosis in the ablated area ( Fig. 3 ). There was no evidence of a complication.

EUS-PDT

Photodynamic therapy (PDT) involves administration of a tumor-localizing photosensitizer, exposure of the target tissue to light of appropriate wavelength, and the generation of a highly cytotoxic oxygen species termed singlet oxygen. Antitumor effects of PDT derive from direct cytotoxic effects, damage to the tumor vasculature, and induction of inflammatory reaction leading to the development of systemic immunity.

A study with a porcine model evaluated the feasibility and safety of EUS-PDT. Porfimer sodium was injected into 3 pigs as the photosensitizer 24 hours before EUS-PDT. Under EUS guidance, the liver, the pancreas, and the kidney were punctured with a 19-gauge fine needle. A quartz optical fiber with a 1.0-cm cylindrical light diffuser was inserted through the needle and into the tissue. The tissue was illuminated with a 630-nm light to a total light dose of 50 J/cm. The animals were euthanized and examined after 2 days of observation. No signs of complication were seen during observation period. On pathological examination, gross ecchymosis was noted on the surface of the pancreas in one pig. The mean area of necrosis induced by EUS-PDT in the pancreas, the liver, the kidney, and the spleen were 3.6, 3.3, 3.2, and 8.5 mm ², respectively. The extent of complete necrosis was 100% only in the pancreas.

Another pilot study of EUS-PDT using the photosensitizer verteporfin on porcine pancreas was reported in 2008. Verteporfin is a photosensitizer commonly used in PDT for choroidal neovascularization secondary to advanced age-related macular degeneration. It has a shorter half-life of 4 hours and duration of photosensitivity of 5 days. In this study, 6 pigs were randomly divided into 3 groups with 2 pigs in each group; the first group was exposed to 10 minutes of 689-nm wavelength laser light at a light dose of 150 J/cm ², the second group to 15 minutes, and the third group to 20 minutes. Serum amylase, lipase, and renal and liver function tests were obtained at baseline and 4 days after the EUS-PDT. An abdominal CT with contrast was performed on day 4 to evaluate the pancreas for tissue effect. The pigs were euthanized on day 7 and the pancreas tail was harvested for pathologic examination. On CT, the PDT-induced pancreatic lesion was a low-attenuation focus in the pancreatic parenchyma. The mean diameter of the lesion after 10, 15, and 20 minutes of laser-light exposure on CT was 6.6, 9.4, and 26.3 mm, respectively. On gross pathology, the treated area appeared as a localized necrotic lesion, and the mean diameter corresponded with the time of exposure (15, 24, and 30.5 mm for 10, 15, and 20 minutes, respectively). Histology revealed a well-defined, solitary lesion that included areas of fat necrosis, granulation tissue, inflammation, and fibrosis. Except for 1 pig in the 10-minute group with a mild increase in serum amylase without clinical evidence of pancreatitis, no complication was encountered.

EUS-Guided Laser Ablation

There is one pilot study of EUS-guided neodymium:yttrium aluminum garnet (Nd:YAG) laser ablation of porcine pancreas. Under EUS guidance, a quartz optical fiber with a tip 300 μm in diameter was introduced to porcine pancreatic tail through a 19-gauge fine needle. An Nd:YAG laser with a wavelength of 1.064 nm was used, with an output power of 2 and 3 W and a total delivered energy of 500 and 1000 J on continuous mode. The pigs were followed up for 24 hours. For the same energy, the ablation area and ablation volume were increased when higher power was used. In detail, for the power setting of 2 W the mean ablation area was 49 and 67 mm ²in cases of set energies of 500 and 1000 J, respectively. For the power setting of 3 W the mean ablation area was 59 and 80 mm ²in the cases of set energies of 500 and 1000 J, respectively. For the ablation volume, the mean value was 314 and 460 mm ³in cases with set energies of 500 and 1000 J, respectively, and a power setting of 2 W. For a higher power of 3 W, the mean ablation volume was 428 mm ³in the cases with set energy of 500 J and 483 mm ³for the case with set energy of 1000 J. There was no major complication defined as clinically symptomatic and chemistry-proven pancreatitis with peritonitis. In 6 of 8 pigs, small peripancreatic fluid collections on pathologic examination were identified, without clinical signs. Serum amylase levels were increased in 7 pigs, and serum lipase levels were increased in all animals.

Recently, a successful EUS-guided Nd:YAG laser ablation of a hepatocellular carcinoma in the caudate lobe was reported. The optical fiber was inserted through a 22-gauge fine needle. Four needle insertions were performed to encompass the entire tumor, and for each illumination the laser was delivered at 5 W for 360 seconds and 1200 J per fiber. The total energy delivered was 7200 J in 36 minutes. The patient was discharged after 3 days uneventfully. CT obtained 24 hours after the procedure showed that the treated area was replaced by a homogeneous nonenhancing area. A subsequent CT 2 months after the procedure demonstrated uniform hypoattenuation without enhancement in the ablated zone.

EUS-Guided Ethanol Injection to Solid Lesions

The feasibility of EUS-guided ethanol injection of normal porcine pancreas was reported in 2005. In this study, under EUS guidance 98% ethanol was injected to the pancreas of 4 Yorkshire pigs, and 50% ethanol to another 4 Yorkshire pigs. The first animal injected with 1.0 mL of 98% ethanol developed pancreatic pseudocyst; subsequently the amount of ethanol injected was reduced to 0.5 mL. No animals showed signs of distress. All animals had elevation of serum amylase. For animals receiving 50% ethanol, pathology revealed a 2- to 6-mm area of necrosis, inflammation, and fibrosis. Two of those receiving 98% ethanol developed complication of pancreatitis: one developed pseudocyst and the other developed inflammatory colonic stricture. The mean diameter of treated pancreatic tissue after 98% ethanol injection was 18 mm. The investigators concluded that whereas 50% ethanol induces localized and self-limited changes, 98% ethanol causes more widespread and unpredictable pancreatitis.

A subsequent study was designed to determine the dose-response relationship of EUS-guided ethanol injection. Under EUS guidance, the investigators injected 2 mL of ethanol with concentrations of 0%, 20%, 40%, 60%, 80%, and 100% to each porcine pancreatic tail, which all animals tolerated well. Abdominal CT obtained on day 4 showed a hypodense area with a mean diameter of 19.4 mm in the pancreatic tail of the pigs that received 40%, 60%, 80%, and 100% ethanol. The changes were confined to the pancreas, without inflammatory changes or peripancreatic necrosis in the rest of the pancreas. Euthanasia was performed on day 7. No lesion could be identified in the pancreatic tail of the pigs injected with 0% and 20% ethanol. The porcine pancreas injected with 40%, 60%, 80%, and 100% ethanol demonstrated a visible coagulation area, with an increasing diameter in correlation to the concentration of ethanol. Histology confirmed the presence of coagulation necrosis in pigs injected with 40%, 60%, 80%, and 100% ethanol. The area of the pancreatic necrosis estimated by CT, measured on gross findings, and histology correlated with the concentration of ethanol. No correlation was found between the diameter of EUS image change and ethanol concentration.

Giday and colleagues evaluated the utility of contrast-enhanced EUS for visualization and monitoring of ethanol-ablated porcine pancreas, by modifying the injection technique described by Aslanian and colleagues with the addition of purified carbon particle solution to facilitate visualization of the injected area on postmortem examination. Reevaluation of the pancreas with EUS and subsequent contrast-enhanced EUS using activated perflutren lipid microspheres intravenous bolus infusion was done at 24 hours (1 pig), 48 hours (2 pigs), and 7 days (1 pig). Injection of microspheres improved the visualization of the hypovascular necrotic lesion. In addition, the diameter of the area of pancreatic necrosis on histology correlated well with that of the area with altered microperfusion detected by contrast-enhanced EUS.

Human cases of EUS-guided ethanol injection to solid tumors has been reported in hepatic metastasis, gastrointestinal stromal tumor, pancreatic insulinoma, and adrenal metastasis from non–small cell lung cancer.

The results of EUS-guided tumor ablation techniques on porcine models are summarized in Table 1 . The potential indications and limitations of the EUS-guided tumor ablation techniques are listed in Table 2 .

Table 1

Summary of EUS-guided tumor ablation techniques on porcine model

Technique	References	Year	N	Target Organ	Maximum Diameter of Ablated Area		Complications
Technique	References	Year	N	Target Organ	EUS (Immediately After Ablation)	Pathology	Complications
RFA	Goldberg et al	1999	13	Pancreas	15 mm	12 mm	Transmural gastric wall burns (n = 3) Intestinal serosal burn (n = 1) Pancreatitis with pancreatic fluid collection without symptoms (n = 1)
	Carrara et al ^a	2008	14	Pancreas	∼900 mm ²^b	∼4000 mm ²^b	Necrotic pancreatitis with peritonitis (n = 1) Pancreatitis without symptoms (n = 1) Gastric wall burn (n = 1) Adhesion between the pancreas and the gut (n = 4)
	Carrara et al ^a	2008	19	Liver (n = 10) Spleen (n = 9)	∼500 mm ²(liver) ^b ^c ∼600 mm ²(spleen) ^b ^c	∼400 mm ²(liver) ^b ∼500 mm ²(spleen) ^b	None
	Varadarajulu et al	2009	5	Liver	23 mm	27 mm	None
PDT	Chan et al	2004	3	Pancreas Liver Spleen Kidney	Results not given	14 mm ²(pancreas) ^b 9 mm ²(liver) ^b 20 mm ²(spleen) ^b 11 mm ²(kidney) ^b	Gross ecchymosis on the surface of the pancreas (n = 1)
PDT	Yusuf et al	2008	6	Pancreas	Results not given	30.5 mm ^d	None
Laser	Di Matteo et al	2010	8	Pancreas	22 mm ²^b	87 mm ²^b	Peripancreatic fluid collection (n = 6) Elevated serum amylase (n = 7) Elevated serum lipase (n = 8)
Ethanol injection	Aslanian et al	2005	8	Pancreas	11.7 mm ^c ^e	6 mm ^e	Pancreatic pseudocyst (n = 1) ^f Inflammatory colonic stricture (n = 1) ^f Elevated serum amylase (n = 8) ^f
	Matthes et al	2007	6	Pancreas	35 mm	27 mm	None
	Giday et al	2007	4	Pancreas	10 mm	10 mm	None