Meaningful progress towards a high-fidelity endoscopic submucosal dissection training simulator model
Article information
Endoscopic submucosal dissection (ESD) is widely accepted as a major curative treatment modality for early gastric cancer. With advancements in endoscopic equipment and diagnostic skills, the proportion of early gastric cancers amenable to curative ESD among all gastric adenocarcinomas has continued to increase. This trend indicates a growing demand for endoscopic specialists capable of performing ESD procedures.1
Teaching ESD skills to novices remains challenging. This challenge stems from the composition of the procedure, which involves multiple extremely delicate maneuvers. The conventional training method for ESD is on-the-job apprenticeship training performed on patients in the endoscopy room. Studies have shown that approximately 30 cases are needed to achieve an acceptable level of ESD proficiency through this method.2 This requirement poses a significant challenge for teaching hospitals, where securing sufficient cases for training purposes is not easy.
Simulator-based training has emerged as the most promising alternative to complement on-the-job apprenticeship training. Common endoscopic simulators include in vivo and ex vivo animal models, mechanical models, and computerized virtual reality models. For ESD procedural skills, the fidelity of electrosurgical knife-tissue interaction is critical. Therefore, ESD simulators using polyvinyl alcohol hydrogel (PVA-H)3,4 or konjac5 to replicate the gastric wall, along with animal models, have demonstrated superior effectiveness compared to other training modalities. Currently, no model has demonstrated sufficient validity to completely replace apprenticeship training. Therefore, a hybrid program appears to be the most appropriate approach. This program should combine initial intensive simulator-based training with subsequent on-the-job apprenticeship education.6 Such an integrated approach enhances both clinical relevance and educational effectiveness. The validation of simulator models is typically assessed through four different types of validity: face, content, construct, and transfer validity.7 However, when educators in training centers adopt new training models, feasibility plays a more important role than validity. Feasibility includes practical considerations, such as acquisition and maintenance costs, management resources, personnel requirements, and the time and workforce required for program implementation. Therefore, ethical and hygiene-related issues must be considered when using animal models. Despite their superior haptic realism, animal models are increasingly being replaced by mechanical simulators owing to these feasibility issues.
As of 2024, excluding animal simulator models, two synthetic gastric wall tissue models are available for ESD training at teaching hospitals. These models have demonstrated superior electrosurgical knife-tissue interaction fidelity: EndoGel (Sunarrow Kasei Co., Ltd.)3 using PVA-H and versatile training tissue, versatile training tissue (VTT) (KOTOBUKI Medical Inc.)5 using konjac. The VTT can be mounted on two different platforms. The first is the G-Master (KOTOBUKI Medical Inc.) setting frame,8 which manually reproduces various stomach locations through multiple adjuster parts. The second is the EndoCubot console unit (EndoRobotics Co., Ltd.), which automatically adjusts preset positions and simulates respiratory motion.
Studies examining the G-Master-VTT have demonstrated favorable content validity,8 face validity,8 and transfer validity.9 While no studies have primarily addressed construct validity, existing literature suggests satisfactory construct validity. A recent study by Toba et al.1 analyzed the learning curve of this model using the cumulative sum (CUSUM) method. This method is widely employed in medical procedure learning curve analysis. This study involved four trainees with substantial experience in diagnostic endoscopy. The lesions were confined to the anterior wall of the antrum, a location known for its relatively low technical difficulty, and had a resection size of 30 mm. Each trainee performed 30 procedures under expert supervision. The primary endpoint was resection speed (mm²/min) as an objective measure of improvement in the ESD technique. The study used the median resection speed for each trainee in 30 training sessions as the target value for the CUSUM analysis. The CUSUM curve showed three distinct phases for all trainees. The first phase was the initial learning phase, in which trainees experienced early difficulties and confusion. The second was the transitional phase or plateau period, during which trainees demonstrated an understanding of and adaptation to the technique, although this level of competence may still be insufficient for complex cases. The final phase was the mastery phase, which indicated technical proficiency. The transition from the first to the second phase required a median of 10 sessions. The transition from the second to the third phase required 19 sessions. No significant variations were observed among the four trainees.
The transition point of the 19 procedures from phase 2 to phase 3 was notably lower than the learning curve point of the 30 cases observed in actual gastric neoplasm ESD procedures. This difference can be explained by the simplified conditions used in this study. This study used only anterior wall lesions with a resection size of 30 mm. In contrast, actual cases involve multiple variables, including different locations, sizes, bleeding, and submucosal fibrosis. The number of required procedures is likely to increase for areas with higher difficulty, such as mid- or high-body areas. The addition of bleeding control elements also increased this number. However, excessive difficulty during the early training stages may not be desirable in novice education. To achieve optimal educational outcomes, educators need to calibrate the level of difficulty and training schedules according to trainee status and central resources. The inflection points identified in this study may provide useful information for educators to develop novice ESD training programs using this model.
Recent developments in ESD training models have led to significant progress. Excluding animal and computerized models, three mechanical models have demonstrated high fidelity: EndoGel, G-Master-VTT, and EndoCubot. The level of trainable skills increases in the order of EndoGel<G-Master-VTT<EndoCubot.
Notes
Conflicts of Interest
The authors have no potential conflicts of interest.
Funding
None.
Author Contributions
Conceptualization: GHL; Data curation: GHL; Formal analysis: GHL; Investigation: GHL; Methodology: all authors; Project administration: all authors; Resources: GHL; Supervision: all authors; Validation: all authors; Visualization: all authors; Writing–original draft: GHL; Writing–review & editing: all authors.