Interventional endoscopic ultrasound (EUS), including EUS-guided fine needle aspiration, has recently become popular. In particular, EUS-guided drainage has been developed as an alternative to percutaneous and surgical approaches.^1–3 The usefulness of EUS-guided biliary drainage, EUS-guided gallbladder drainage, EUS-guided drainage of pancreatic fluid collections, EUS-guided pancreatic duct drainage, and EUS-guided drainage of abdominal and pelvic collections have already been reported. Moreover, the utility of EUS-guided injection of various materials has also been described.^4–8 Above all, EUS-guided tumor ablation,^9–13 EUS-guided celiac plexus neurolysis (EUS-CPN), celiac plexus block, and celiac ganglion neurolysis (EUS-CGN) are expected to show a high degree of clinical usefulness.^14–18 EUS-CPN is well established as an alternative method for providing pain relief and reducing narcotic use, and has become quite popular because of the technical simplicity and safety. EUS-CPN has been assessed with a high level of evidence in the literature, and some papers have reported EUS-CGN as highly effective for pain relief compared to EUS-CPN. The effectiveness of EUS-CPN for pain relief is clear clinically, but remains insufficient and temporary. Patients need more effective options than pure ethanol injection. Injected ethanol infiltrates and rapidly spreads into surrounding tissue, requiring a relatively large amount for effective treatment. If the viscosity of the injected ethanol could be increased, effectiveness might also improve.

This study sought to identify a more effective method for CPN and tumor ablation therapy using ethanol and povidone iodine. We selected atelocollagen to increase the viscosity of ethanol and povidone iodine, as a well-known disinfectant that can be used even in patients who are allergic to ethanol. Both ethanol and povidone iodine are inexpensive and widely available around the world. We hoped to increase the clinical efficacy of their use for patients with this study.

EUS-guided injection was performed using a Medical Linear EUS system (Pentax, Tokyo, Japan) in animal models. The injection needle was a 19-G Expect (Boston Scientific, Marlborough, MA, USA). We applied two protocols in three sedated pigs. We used indigo carmine (Daiichi Sankyo, Tokyo, Japan), absolute ethanol (Kenei Pharmaceutical, Tokyo, Japan), 10% povidone iodine (Meiji Seika Pharmaceutical, Tokyo, Japan), and 1% and 3% atelocollagen implant (Koken, Tokyo, Japan).

Protocol 1

We performed EUS-guided injection of three kinds of ethanol into para-aortic tissue in a sedated pig. At more than 4 hours after injection, we checked ethanol injection sites after dissection. A blue area of indigo carmine staining would be apparent if the injected ethanol remained in place.

Protocol 2

We performed EUS-guided injection of a total of six kinds of reagents (two kinds of ethanol, three kinds of povidone iodine, and control atelocollagen) into the livers of two living pigs. After 2 weeks, we examined tissue damage to the liver in the two pigs. H&E staining of resected specimens was performed to evaluate tissue damage.

Protocol 1

We injected three kinds of reagents (including indigo carmine) in three separate areas of para-aortic tissue under EUS guidance (Fig. 1A). We can see the blue area of reagent 3 (75% ethanol: absolute ethanol 3.75 mL + 1% atelocollagen 1.25 mL + very small amount of indigo carmine) in the para-aortic tissue shown in Fig. 1B. Only reagent 3 was seen like blue gel, and still remained in place. Neither reagent 1 nor 2 was seen at the injection site at all.

Protocol 2

We made another two kinds of reagents (75% ethanol: absolute ethanol 2.25 mL + 3% atelocollagen 0.75 mL; 5% povidone iodine: 10% povidone iodine 1.5 mL + 3% atelocollagen 1.5 mL) as shown in Fig. 2. Ethanol plus 3% atelocollagen congeals easily, and so cannot be used, as shown in Fig. 2A. Conversely, 5% povidone iodine using 3% atelocollagen also becomes a hard gel that is difficult to aspirate and inject (Fig. 2B). Fig. 3 shows resected liver specimens from the two pigs. H&E staining shows the cytotoxic effects of the injected reagents (Fig. 4). As reagents 1 to 3 showed strong cytotoxic effects, we could not identify any viable cells in the injected area (Fig. 4A–4C). We also confirmed inflammation and infiltration into tissue surround the injected area by reagents 1 to 3 (Fig. 3A–3C, 4A–4C). However, viscosity of reagents 4 and 5 were high, resulting in round, brownish areas in the specimens (Fig. 3D, 3E). Both brownish areas exhibited clear, regular borders with greatly reduced infiltration into surrounding tissue compared to reagents 1 to 3. Cell death was seen only in the injected area, not in surrounding tissue (Fig. 4D, 4E). These findings were seen more with reagent 5 than with reagent 4. Only injected atelocollagen (reagent 6) showed no cytotoxicity for liver tissue (Fig. 4F).

In case of EUS-CPN or tumor ablation therapy, pure ethanol is frequently used in clinical situations.^14–18 Pure ethanol is safe, cheap, easy to get, and strongly cytotoxic. Ethanol injection therapies have thus become so commonplace around the world. However, ethanol is a liquid and easily infiltrates into surrounding tissue. After injection, ethanol rapidly percolates out and local cytotoxic effects are decreased. If the viscosity of the injected ethanol could be increased, ethanol would remain in the target area, and clinical effects would presumably be increased and last longer. We selected atelocollagen to increase viscosity. Atelocollagen is a type 1 collagen derived from the dermis of cows. Atelocollagen shows low antigenicity and high biocompatibility,¹⁹ and has seen use in a variety of applications. As another important point, some Japanese are allergic to ethanol, so new reagents are needed that are safe, cheap, easily obtained, and strongly cytotoxic. Finally, we selected povidone iodine. Protocol 1 examined the viscosity and penetration of different concentrations of ethanol mixed with atelocollagen. We could see the blue area of indigo carmine only with reagent 3 (75% ethanol: absolute ethanol 3.75 mL + 1% atelocollagen 1.25 mL + very small amount of indigo carmine). Because a high concentration of ethanol easily and rapidly percolates through tissues, we could not confirm blue areas of indigo carmine with reagents 1 and 2. If the concentration of ethanol is less than 75%, the cytotoxic effects will be decreased. Finally, we concluded that viscosity and stagnation were best for reagent 3 in this study. In protocol 2, we examined the cytotoxic effects of different concentrations of ethanol and povidone iodine. Degree of cytotoxicity was similar in reagents 1 to 3. Infiltration into surrounding tissue was strong, so the border of the injected area was irregular and unclear. In contrast, both reagents 4 and 5 (povidone iodine + atelocollagen) were only seen in the injected area. The borders of the injected areas for both reagents 4 and 5 were clear and regular. The influence of injected reagents 4 and 5 on surrounding tissue was weak. According to the results of protocols 1 and 2, we conclude that the cytotoxic effects, infiltration into surrounding tissue, and stagnation with reagent 3 were the best for EUS-CPN. Low infiltration into surrounding tissue and high viscosity and high stagnation with reagent 5 were considered the best for tumor ablation therapy. We think that EUS-guided fine needle injection for pancreatic neuroendocrine using reagent 5 should be safer and more effective than that using ethanol. We now have to confirm the safety and efficacy of this approach for humans in clinical trials.

In conclusion, we investigated the optimal concentrations of ethanol and povidone iodine using atelocollagen for EUS-CPN and tumor ablation therapy. For EUS-CPN, 75% ethanol mixed with 1% atelocollagen appears best. Povidone iodine mixed with 3% atelocollagen may be suitable for small tumor ablation therapy. Confirmation of the clinical usefulness of these methods in prospective trials is awaited.

This study is a collaboration with Boston Scientific.