Does development of submucosal dissection models influence quality of training? Comparison of existing models

Abstract

Background and study aims

Use of endoscopic submucosal dissection (ESD) is growing, but access to it remains limited. The aim of this study was to compare the performance of various existing models and the progress made by students on them.

Methods

Four training models (bovine colon (ex-vivo), ex vivo porcine model, live porcine model, and an artificial model (Endogel)) were evaluated during a 1-week training course. Each participant was evaluated at the beginning (D1) and at the end of the training (D5). Learners performed a standardized ESD of 2 cm on the four models in a randomized order. Experts evaluated the ability of participants to perform ESD using the objective structured assessment of technical skill score (OSATS).

Results

Sixteen students were involved, the average age was 35.6 years (+/- 4.6) and they practiced endoscopy for 10 years (+/-5.3). The OSATS significantly increased in each model during the week, with mean scores increasing from 8.6 to 23.3, from 10.7 to 12.9, from 8.8 to 21.3 and from 8.2 to 12.5 for the bovine colon, ex vivo porcine model, live porcine model, and Endogel models, respectively.

Conclusions

Ex-vivo models are good models for learning ESD skills. The bovine colon model seems to be the most discriminating. Synthetic models should be reserved for novices.

 

MissingLinkText

Keywords

Endoscopy Lower GI Tract - Endoscopic resection (polypectomy, ESD, EMRc, ...) - Quality and logistical aspects - Training

Introduction

The indication for endoscopic submucosal dissection (ESD) is expanding across all segments of the digestive tract but its adoption in Western countries is still progressing at a slow pace. ESD necessitates a heightened level of endoscopic proficiency and mandates rigorous training, even for experienced endoscopists. The learning curve for colonic ESD is long. R0 resection rates during the initial learning phase range from 46% to 75%, accompanied by an elevated incidence of complications (20% perforations and 20% conversions to surgery) [1] [2] [3] [4].

Several factors have been identified to explain the rate of complications, including operator expertise and volume of the ESD center [5] [6]. Pre-practice utilization of animal models has been shown to decrease the complication rate and increase the en bloc resection rate. The European Society of Gastrointestinal Endoscopy (ESGE) recently issued a statement on ESD practice [7]. They recommend a minimum of 20 ESD procedures on models before transitioning to human subjects, with the objective of achieving at least eight en bloc complete resections in the last 10 training cases, and with no perforation [8].

A primary obstacle to wider adoption of ESD in Western countries is the challenge in accessing animal models. Live animal models offer an advantage due to their similarity to humans, allowing for replication of typical operating conditions encountered in real practice, including the potential to mimic respiratory movements and address common complications in real time. Ex vivo models, such as bovine colon and ex vivo porcine model, are also available but are considered less realistic. Use of animal models is constrained by ethical and societal considerations, the theoretical risk of zoonosis, their accessibility, cost, and ethical concerns about sacrificing animals after each session.

To address these challenges, ESD simulators have been developed to facilitate learning various stages of ESD. They have the potential to enhance skill acquisition and technique implementation, thereby reducing training costs and the number of animals sacrificed. The objective of this study was to compare different ESD models available in France during a national ESD training week in terms of user opinion and progression capability.

 

MissingLinkText

Methods

Description of ESD training week

Since 2021, the French society of endoscopy has organized an ESD training program. During the year, a theoretical course in ESD is offered to 16 trainees previously selected because of their motivation and the local possibility of ESD companionship. After the theory classes, a practical training course of 5 days is organized to learn from the expert during simulation on ESD models.

 

MissingLinkText

Organization of the evaluation

We designed a prospective evaluation of the practice of simulated ESD on four different ESD models during this week to compare the training potential and the user opinion on Day 1 and Day 5 of the training week for each model.

Four training models (bovine colon [ex vivo], ex vivo porcine model, live porcine model and an artificial model [Endogel]) were evaluated during this 1-week training course.

Each participant was evaluated at the beginning (D1) and at the end of their training (D5). Four groups of four participants were created. In each group, the same ESD knife (Splash M Knife, Flush Knife, Gold Knife, Dual J Knife) was used to train on the four ESD models. Each group rotated among the four models and was evaluated at D1 and D5 ([Fig. 1]). The running order was randomized at D1 and D5.

Fig. 1 ESD training organization and evaluation.

Between the two evaluations, the trainees had intensive ESD formation supervised by French ESD experts on animal models. This training was conducted on live porcine, bovine colon, and ex vivo porcine models and focused exclusively on ESD techniques, including circumferential incision, cutting, and dissection with or without traction. Under the supervision of ESD experts, students received guidance and correction as needed, covering aspects such as scope positioning and knife handling.

 

MissingLinkText

Standard ESD procedure

In each model, the trainees had to perform a 20-mm ESD. A simulated lesion was done by the expert in each model using a snare applied on the mucosa with force coagulation current in order to provide a whitish oval simulated lesion. Then, the trainees had to perform submucosal injection of the lesion (using a saline solution colored with indigo carmine), incision of the lesion to gain access to the submucosae, and the dissection phase without traction. They all had10 minutes on each model to perform the entire resection.

 

MissingLinkText

Evaluation of the trainees and models

Experts evaluated the ability of each participant using an objective structured assessment of technical skill (OSATS) [9]. We modified this score for endoscopic evaluation (Supplementary Table 1). Information about complete incision, submucosal access, and total dissection also was collected by the experts. For each model, each expert his impressions about conductivity, slide, softness, and visualization of the submucosae.

Level of satisfaction with each model was assessed using a numerical scale ranging from 0 (poor) to 10 (excellent).

At the end of the week, the expert graded a student's ability to perform ESD on a scale from 0 to 20 (Supplementary Table 2).

 

MissingLinkText

Statistical analysis

Quantitative variables were calculated as mean, (interval) or median (interquartile range), and qualitative variables were calculated as a percentage.

To evaluate the progression of the student on each model during the ESD training week, we first hypothesized that the training during the week did not differ between students. The normality was verified with the Shapiro-Wilk test. We then compared the difference between the OSATS obtained by the students on D1 and D5 on each of the models using the Student's paired t-test. The percentage of complete incision, submucosae access, and full dissection were also tested between D1 and D5 using with the Fisher's exact test.

Ability to perform ESD was evaluated using a visual analogue scale ranging from 0 to 20. Pearson's coefficient was used to assess the correlation between the note obtained on the visual analogue scale and the OSATS. Correlation was judged very strong from 1 to 0.9, strong from 0.9 to 0.7, moderate from 0.7 to 0.5, low from 0.5 to 0.3. and poor from 0.3 to 0. The alpha risk was set to 0.05.

 

MissingLinkText

Ethics

Use of animal models was approved by the local committee of Limoges simulation platform (EMIS, Limoges, France).

 

MissingLinkText

 

MissingLinkText

Results

Trainee characteristics

Sixteen students were involved, most of whom were male (11/16, 68%) with an average age of 35.6 years (+/- 4.6) and had been practicing endoscopy for an average of 10 years (+/-5.3). One-quarter of participants had never performed an ESD, 56.2% had performed fewer than 10, and one participant reported performing more than 20 ESDs.

 

MissingLinkText

Skills abilities and progression

The OSATS was evaluated for each participant at D1 and D5 and statistically improved on each model between D1 and D5 ([Fig. 2]).

Fig. 2 Progression of the OSATS endoscopic score during the ESD week on each model.

Mean OSATS at D1 and D5 with the ex vivo porcine model were 10.71 (SD 3.5) and 12.86 (SD 2.03), respectively, with a mean difference of -2.14 (SD = 3.44; 95% confidence interval [CI] -4.06 to -0.23; P = 0.036).

For the bovine colon, mean OSATS at D1 and D5 were 8.57 (SD 2.59) and 23.14 (SD 3.11), respectively, with a mean difference of -14.57 (SD 3.72; 95% CI -16.64 to -12.5; P < 0.001).

Mean OSATS at D1 and D5 with a live porcine model were 8.43 (SD 2.98) and 21.29 (SD 5.33), respectively, with a mean difference of -12.86 (SD 5.53; 95% CI -15.93 to -9.78; P < 0.001).

For artificial model, the mean OSATS at D1 and D5 were 8.19 (SD 3.49) to 12.5 (3.14), respectively, (SD 5.0, Δ -4.5; SD 4.6; 95% CI -7.5 to -2.5; P = 0.006). Overall performance and procedure flow significantly improved in all models, except for the artificial model. Results of each subitem in the OSATS are detailed in [Table 1].

Table 1 Detailed mean scores for each item on the objective structured assessment of technical skill (OSATS) across models.

	Respect for tissue		P value	Time and motion		P value	Instrument handling		P value	Control knob		P value	Flow operation		P value	Overall performance		P value
	D1	D5		D1	D5		D1	D5		D1	D5		D1	D5		D1	D5
Ex vivo porcine	3.0	3.8	0.041	2.6	3.6	0.017	2.7	3.9	0.001	3.7	3.5	0.56	3.1	4.6	0.0036	2.5	3.3	0.017
Bovine colon	2.5	4.1	0.0001	3.4	3.6	0.66	3.8	3.6	0.55	3.8	3.8	0.91	2.5	4.3	< 0.001	1.7	3.8	< 0.001
live porcine model	2.6	3.4	0.064	2.9	3.6	0.11	2.6	3.6	0.0033	3.9	3.5	0.26	2.4	3.6	< 0.001	2.3	3.4	< 0.001
artificial model	2.5	3.3	0.0057	2.7	3.6	0.044	3.3	3.9	0.089	3.4	3.8	0.35	2.6	3.2	0.16	2.4	2.8	0.21

The percentage of complete incision, submucosae access, and full dissection are presented in [Table 2].

Table 2 Skills abilities and progression on each model during the week.

	Complete incision			Submucosae access			Full dissection			Mean OSAT score
OSATS, objective structured assessment of technical skill.
	D1	D5	P value	D1	D5	P value	D1	D5	P value	D1	D5	P value
Ex vivo pig stomach	56.25%	56.25%	1	62.5%	31.25%	0.15	6.25%	0	1	10.71	12.86	0.036
Ex vivo cow colon	18.75%	93.75%	0.019	12.5%	81.25%	< 0.001	0%	0	1	8.57	23.14	< 0.001
live porcine model	68.75%	62.5%	1	37.5%	56.25%	0.48	6.25%	6.25%	1	8.43	21.29	< 0.001
artificial model	100%	87.5%	0.48	75%	75%	1	31.25%	18.75%	0	8.19	12.50	0.006

 

MissingLinkText

Experts’ impressions

The impression of conductivity, slide, softness, and visualization of submucosae were analyzed by the four experts (Supplementary Fig. 1). Experts rated each student at the end of the ESD week depending on their ability to practice ESD and the median score was 16 of 20 (IQR 14.5–17). This subjective appreciation correlates with the OSATS (ρ = 0.98; r² = 0.958; P < 0.001) ([Fig. 3]).

Fig. 3 Correlation between expert impression of the ability of the student and the OSATS score.

 

MissingLinkText

Satisfaction

Students overwhelmingly favored the live porcine model, with a median numerical scale rating of 8 (interquartile range [IQR] 7.5–8.7) ([Fig. 4]). Students identified the live pig model as the sole model with which they were capable of effectively managing bleeding complications. All participants perceived practicing on living animal models as beneficial to their education.

Fig. 4 Trainee satisfaction level with the different endoscopic submucosal dissection training models.

 

MissingLinkText

 

MissingLinkText

Discussion

Although ESD is the best option for obtaining R0 resection, beginning ESD on humans without adequate training is associated with low effectiveness and high mortality rates [10]. It requires multifaceted skill training, including simulation in various models to complete the long learning curve.

Our study demonstrated that the four simulation models tested are effective for obtaining a progression. All trainees acquired skills and, therefore, progressed during the training week. But the four models are not equivalent. In our study, a synthetic model allowed easy incision and submucosal access, even for beginners. The rate of achieving full dissection with this model exceeded that of other existing models, yet the progression in the OSATS on it was relatively modest. On each model, precision of the gesture assessed by the “respect for tissue” item improved at the end of the training week. The resection technique was statistically more fluid at the end of the week for all three animal models. The artificial model on each item showed a less significant difference between the beginning and the end of the training week. Consequently, based on our finding, the synthetic model does not seem sufficient to discriminate trainees depending on their progression. However, it emerged as the optimal choice for initial skill acquisition and introduction to ESD techniques. The advent of synthetic models raises questions about the actual need for animal sacrifice in training sessions and the potential for expanding the number of ESD practitioners [11]. However, limitations were noted, such as managing complications like bleeding and perforation, ensuring optimal visibility of the submucosal layer using gravity, and controlling air with insufflation and suction during the procedure. Recently, Mitsui et al. reported a high satisfaction and acceptability rate among experts using a new training model [12].

Animal models, particularly the bovine colon, presented higher difficulty levels in ESD. The bovine colon model appears to be the most effective in discriminating the progression of students, because it exhibited the highest delta observed on the OSATS. This may be attributed to the challenging accessibility of the submucosae in this model due to avascular tissue and the freezing process, which can compromise tissue quality. Although the bovine colon is not the preferred model according to our subjective evaluation by the students, it could potentially be the optimal choice due to factors such as accessibility, cost-effectiveness, and absence of additional animal sacrifices. This model holds promise for evaluating ESD progression in studies that compare different ESD training strategies.

A live porcine model had the best satisfaction for students and a good discrimination potential but it poses challenges for widespread implementation in training centers, especially due to factors such as animal sacrifice, cost, and facility requirements. Although it brings theoretical advantages such as bleeding, respiratory movement, perforation management, the potential for discrimination does not surpass the bovine colon and ethical concerns, animal welfare, and the theoretical risk of zoonosis need to be addressed with this model.

In our study, we tried to evaluate skill progression using the OSATS endoscopic score, derived from the surgical training skills evaluation. The OSATS is complex to evaluate but highly correlated to the expert impression of trainee capacity.

If the models are important to learn first ESD skills, supervision is mandatory for the first human procedure. If supervision by an expert was the standard until now, virtual supervision has already demonstrated its own benefit but could be improved by artificial intelligence (AI) [13] [14]. Recently, Cao et al. pioneered the creation of AI capable of recognizing different stages of ESD. Envisioning the future, there is optimism for development of an AI system that can guide practitioners on where to incise the lesion, thereby preventing complications and reducing the learning curve for ESD.

Our study has some limitations. First, we did not evaluate traction strategies in the four models while use of ESD is growing. Although all the models tested can be employed with traction [15], this study aimed to evaluated initial ESD performance without adding the complexity of traction strategy and its own learning curve. Second, the participants were evaluated on D1 and D5 by the same evaluators who trained them during the entire week, which could have biased the results. The intensive week of training was conducted using live porcine and ex vivo models and synthetic models were not utilized due to their cost. Although the results obtained on D1 and D5 may be subject to bias from this limitation, each student received an equal amount of training time on the animal models, which are inherently more challenging than synthetic models. The standardized structure of the intensive program likely helped mitigate this potential bias. The experts were also not blind during the final evaluation, and use of a modified OSATS for endoscopic evaluation can be criticized because it is not yet validated. Finally, this study demonstrated the high potential of an animal model to obtain student progression, but correlation between animal skill enhancement and first human procedures is still missing. That evaluation is ongoing in a trainee long-term follow-up study.

 

MissingLinkText

Conclusions

In conclusion, although synthetic models offer a valuable opportunity for early beginners to initiate ESD training, animal models remain essential for accurately measuring trainee progression. The cow colon model stands out for its discriminative capacity, accessibility, and absence of additional animal sacrifices. Although the living pig model remains the preferred option for trainees due to its realistic features, ethical considerations and cost concerns associated with it could raise questions in training workshops. Therefore, a balanced approach that integrates both synthetic and animal models may offer the most comprehensive and effective ESD training experience while addressing ethical and practical considerations.

 

MissingLinkText

 

MissingLinkText

Conflict of Interest

Conflicts of interest: CY: consultant and lectures for Abbvie, Takeda, Jansen, Amgen, Galapagos. JR: Honorarium for training sessions in endoscopy and endoscopic resection for Olympus, Cook Medical, MS: Honorarium for training session for Boston Scientific. Invitation for congress by Olympus, Cook, Cousin Medical, Boston Scientific, Pentax, AbbVie, MSD, Amgen, Norgine. MP: Honorarium for training sessions in endoscopy and endoscopic resection for Olympus, Cook Medical, Boston scientific, Pentax Medical. Honorarium for training sessions in endoscopic characterization with Norgine, Provepharm. Uegw invitation by AlfaSigma. Patent of our institution Hospices civils de Lyon for IPEFIX device JJ: ESD training sessions for Olympus, Fuji, Erbe, Pentax, Lumendi. Lectures for Abbvie, Janssen, Norgine The remaining authors have no conflicts of interest to declare.

Acknowledgement

The authors acknowledge the GRECO (Groupement de Recherches et d'Études en Chirurgie rObotisée).

• References

• 1 Spychalski M, Skulimowski A, Dziki A. et al. Colorectal endoscopic submucosal dissection (ESD) in the West - when can satisfactory results be obtained? A single-operator learning curve analysis. Scand J Gastroenterol 2017; 52: 1442-1452
• 2 Hsu W-H, Sun M-S, Lo H-W. et al. Clinical practice of endoscopic submucosal dissection for early colorectal neoplasms by a colonoscopist with limited gastric experience. Gastroenterol Res Pract 2013; 2013: 1-5
• 3 Shiga H, Ohba R, Matsuhashi T. et al. Feasibility of colorectal endoscopic submucosal dissection (ESD) carried out by endoscopists with no or little experience in gastric ESD. Digest Endosc 2017; 29: 58-65
• 4 Iacopini F, Bella A, Costamagna G. et al. Stepwise training in rectal and colonic endoscopic submucosal dissection with differentiated learning curves. Gastrointest Endosc 2012; 76: 1188-1196
• 5 Odagiri H, Yasunaga O. Complications following endoscopic submucosal dissection for gastric, esophageal, and colorectal cancer: a review of studies based on nationwide large-scale databases. Ann Transl Med 2017; 5: 189-189
• 6 Toyokawa T, Inaba T, Omote S. et al. Risk factors for perforation and delayed bleeding associated with endoscopic submucosal dissection for early gastric neoplasms: Analysis of 1123 lesions. J Gastro Hepatol 2012; 27: 907-912
• 7 Pimentel-Nunes P, Pioche M, Albéniz E. et al. Curriculum for endoscopic submucosal dissection training in Europe: European Society of Gastrointestinal Endoscopy (ESGE) Position Statement. Endoscopy 2019; 51: 980-992
• 8 Parra-Blanco A. Endoscopic submucosal dissection training with pig models in a Western country. WJG 2010; 16: 2895
• 9 Martin JA, Regehr G, Reznick R. et al. Objective structured assessment of technical skill (OSATS) for surgical residents: Objective structured assessment of technical skill. Br J Surg 1997; 84: 273-278
• 10 Rahmi G, Hotayt B, Chaussade S. et al. Endoscopic submucosal dissection for superficial rectal tumors: prospective evaluation in France. Endoscopy 2014; 46: 670-676
• 11 Lee DS, Lee GH, Kim SG. et al. Usefulness of a new polyvinyl alcohol hydrogel (PVA-H)-based simulator for endoscopic submucosal dissection training: a pilot study. Clin Endosc 2023; 56: 604-612
• 12 Mitsui T, Sunakawa H, Takayama S. et al. Development of a novel colorectal endoscopic submucosal dissection training model. Endoscopy 2023; 55: E808-E810
• 13 Cao J, Yip H-C, Chen Y. et al. Intelligent surgical workflow recognition for endoscopic submucosal dissection with real-time animal study. Nat Commun 2023; 14: 6676
• 14 Pioche M, Rivory J, Nishizawa T. et al. Randomized comparative evaluation of endoscopic submucosal dissection self-learning software in France and Japan. Endoscopy 2016; 48: 1076-1083
• 15 De Cristofaro E, Lafeuille P, Jacques J. et al. Non-animal endoscopic training models are also effective for simulation of endoscopic submucosal dissection with adaptive traction strategy. Endoscopy 2023; 55: E973-E974

Publication History

Received: 09 July 2024

Accepted after revision: 24 February 2025

Accepted Manuscript online:
26 May 2025

Article published online:
17 June 2025