ESD is a complex interventional procedure involving several skills and diverse instrumentation. A satisfactory outcome for ESD requires expertise in multiple endoscopic techniques such as proper identification of the lesion, marking the periphery of the lesion, mucosal pre-cutting, submucosal dissection, and the management of bleeding and other complications. We believe that the artificial tissue simulator can be a bridge between instructional videos and animal model training. The trainees have the opportunity to practice each technique required to perform ESD and hone their skills until that point at which they are ready to attempt the animal model. The novel use of such a simulator in a training program has not been previously described.

A stepwise training process for learning ESD is widely accepted in Japan. The first step is the acquisition of basic endoscopic skills. Currently, prior to training, the novice has never performed any of the techniques necessary for the performance of ESD such as lesion marking, mucosal pre-cutting, submucosal dissection, and hemostasis of the submucosal vessels. Animal models, both ex vivo and in vivo [810], are expensive, inconvenient, and time consuming. The expense of hands-on training with live animals may limit the ability of some training programs to provide ESD training. One animal training session may cost nearly $300 (US) per trainee [10]. Additionally, the animal must be sacrificed afterward. The price of one ex vivo simulator with a pig stomach is $50 (US) [8]. Though the ex vivo animal models are less expensive, two hours of specimen preparation are necessary before the training session [8]. Training programs should be tailored around the program needs based on ethnicity, culture, and the working environment. With the animal model, it is possible that the novice will not be able to complete the ESD procedure in one session. Each step must be completed in sequence, and if one step is particularly challenging for the trainee, the following steps may not be completed. In this situation, use of an artificial tissue stimulator [11] (the cost is $20 (US)) at the beginning of training may save money and time, and limit the need for animal tissue. The commercial available resinoid stomach model has several port locations on which the artificial tissue can be mounted and used for multiple ESD practice sessions. The most important advantage of this novel artificial tissue model is that it provides an opportunity for inexpensive and multiple use by the novice, unlike animal models.

With practice using the artificial tissue model, the attendees felt more confident in performing each technique, and the ESD procedure became easier. It is imperative for ESD trainees to maintain a positive attitude and remain aware of their strengths and weaknesses to overcome their initial anxiety in learning the procedure. Using this training, the attendees were able to assess and improve their endoscopic manipulations. After watching an instructional video and performing ESD with the simulator, the attendees felt that the most difficult techniques were submucosal dissection and hemostasis with a mean difficulty score of 4.8. Yamamoto et al. previously reported that in a teaching program, submucosal dissection has been shown to be more difficult than mucosal incision, mostly because of uncontrolled haemorrhage [12]. The attendees in this study recognized the difficulty in performing hemostasis during ESD, and we had hoped that the red strands fixed on the lower layer, representing exposed submucosal vessels, would aid in practicing hemostasis. However, the attendees felt that in this regard the model was insufficient compared to the instructional ESD video. Obviously, expertise in hemostasis is necessary. Toeh et al. reported a 56.5 % rate of bleeding and perforation when 24 novice endoscopists used a live porcine model in an ESD training workshop in Hong Kong [13]. The attendees may gain the experiences and sufficiency training in the next step alive porcine model.

How to interpret the degree of the feasibility using this model for training? We observed the association between the decreasing difficulty and the feasibility. If the difficulty on performing each technique decreased more than one score after using the model, the attendees felt the feasibility is excellent (with a mean score more than 4) in performing such technique. For example, the attendees thought the simulator was feasible in simulated lesion marking (mean difficult score from 3.5?±?1.0 to 2.5?±?1.0; p?=?0.026) and submucosal dissection (4.8?±?0.5 to 3.8?±?1.3; p?=?0.037) because the techniques were easier to perform than just watching an instructional video.

Our study had a number of limitations. Because this was an in-house training project involving fellows from one institution, we had a very small sample size, precluding rigorous statistical analysis. Larger studies with an appropriate sample size and assessment of the clinical data are necessary. This pilot study provided training for almost all techniques used in ESD, but there were some differences from the use of an animal model. For instance, submucosal injections were more difficult to practice in the artificial tissue model. Novices should master their submucosal injection skills during polypectomy or EMR. Secondary, we believe that there is a long gap between learning from an instructional video and the skills required in an animal model. We did not assess performance scores in performing artificial tissue simulator in the very initial training step. It seems to help learning how to skill and proceed the multiple procedure orders employing varied instrumentation and the attendees did appear to improve confidence after the training. The third, this may raise the question of how many procedures should fellows do using artificial tissue simulator before the next step of ex-vivo animal model? When our fellows achieve success sessions in 3–5 artificial tissue simulator, they reach the learner level to performing the ex-vivo animal model. Whether those results will translate into better clinical performance on the next step animal models remains to be seen.