Article Text

Original article
Development and validation of the WASP classification system for optical diagnosis of adenomas, hyperplastic polyps and sessile serrated adenomas/polyps
  1. Joep E G IJspeert1,
  2. Barbara A J Bastiaansen1,
  3. Monique E van Leerdam2,
  4. Gerrit A Meijer3,
  5. Susanne van Eeden4,
  6. Silvia Sanduleanu5,
  7. Erik J Schoon6,
  8. Tanya M Bisseling7,
  9. Manon CW Spaander8,
  10. Niels van Lelyveld9,
  11. Marloes Bargeman10,
  12. Junfeng Wang11,
  13. Evelien Dekker1,
  14. Dutch Workgroup serrAted polypS & Polyposis (WASP)
    1. 1Department of Gastroenterology and Hepatology, Academic Medical Center, Amsterdam, The Netherlands
    2. 2Department of Gastroenterology and Hepatology, Netherlands Cancer Institute, Amsterdam, The Netherlands
    3. 3Department of Pathology, Vrije Universiteit Medical Center, Amsterdam, The Netherlands
    4. 4Department of Pathology, Academic Medical Center, Amsterdam, The Netherlands
    5. 5Division of Gastroenterology and Hepatology, Maastricht University Medical Center, Maastricht, The Netherlands
    6. 6Department of Gastroenterology and Hepatology, Catharina Hospital, Eindhoven, The Netherlands
    7. 7Department of Gastroenterology and Hepatology, Radboud University Medical Centre, Nijmegen, The Netherlands
    8. 8Department of Gastroenterology and Hepatology, Erasmus University Medical Centre, Rotterdam, The Netherlands
    9. 9Department of Gastroenterology and Hepatology, St. Antonius Hospital, Nieuwegein, The Netherlands
    10. 10Department of Gastroenterology and Hepatology, Medical Spectrum Twente, Enschede, The Netherlands
    11. 11Department of Clinical Epidemiology, Biostatistics & Bioinformatics, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands
    1. Correspondence to Dr Evelien Dekker, Department of Gastroenterology and Hepatology, Academic Medical Centre, Meibergdreef 9, Amsterdam 1105 AZ, The Netherlands; e.dekker{at}amc.uva.nl

    Abstract

    Objective Accurate endoscopic differentiation would enable to resect and discard small and diminutive colonic lesions, thereby increasing cost-efficiency. Current classification systems based on narrow band imaging (NBI), however, do not include neoplastic sessile serrated adenomas/polyps (SSA/Ps). We aimed to develop and validate a new classification system for endoscopic differentiation of adenomas, hyperplastic polyps and SSA/Ps <10 mm.

    Design We developed the Workgroup serrAted polypS and Polyposis (WASP) classification, combining the NBI International Colorectal Endoscopic classification and criteria for differentiation of SSA/Ps in a stepwise approach. Ten consultant gastroenterologists predicted polyp histology, including levels of confidence, based on the endoscopic aspect of 45 polyps, before and after participation in training in the WASP classification. After 6 months, the same endoscopists predicted polyp histology of a new set of 50 polyps, with a ratio of lesions comparable to daily practice.

    Results The accuracy of optical diagnosis was 0.63 (95% CI 0.54 to 0.71) at baseline, which improved to 0.79 (95% CI 0.72 to 0.86, p<0.001) after training. For polyps diagnosed with high confidence the accuracy was 0.73 (95% CI 0.64 to 0.82), which improved to 0.87 (95% CI 0.80 to 0.95, p<0.01). The accuracy of optical diagnosis after 6 months was 0.76 (95% CI 0.72 to 0.80), increasing to 0.84 (95% CI 0.81 to 0.88) considering high confidence diagnosis. The combined negative predictive value with high confidence of diminutive neoplastic lesions (adenomas and SSA/Ps together) was 0.91 (95% CI 0.83 to 0.96).

    Conclusions We developed and validated the first integrative classification method for endoscopic differentiation of small and diminutive adenomas, hyperplastic polyps and SSA/Ps. In a still image evaluation setting, introduction of the WASP classification significantly improved the accuracy of optical diagnosis overall as well as SSA/P in particular, which proved to be sustainable after 6 months.

    • COLONOSCOPY
    • COLORECTAL NEOPLASM
    • COLONIC POLYPS
    • COLORECTAL CANCER
    • CANCER PREVENTION
    View Full Text

    Statistics from Altmetric.com

    Video abstract

    Significance of this study

    What is already known about this subject?

    • Accurate endoscopic differentiation of colonic polyps would allow to resect and discard small colonic lesions and selectively resect diminutive premalignant lesions in the rectosigmoid.

    • Sessile serrated adenomas/polyps are common and premalignant.

    • Integrative classification systems for optical diagnosis of colonic lesions, including sessile serrated adenomas/polyps do not yet exist.

    What are the new findings?

    • We developed a new classification system for endoscopic differentiation of adenomas, hyperplastic polyps and sessile serrated adenomas/polyps with narrow band imaging (NBI); the Workgroup serrAted polypS and Polyposis (WASP) classification.

    • Use of the WASP classification significantly increased the overall diagnostic accuracy of optical diagnosis differentiating colonic polyps overall as well as sessile serrated adenomas/polyps in particular.

    • The pooled negative predictive value for optical diagnosis of diminutive neoplastic lesions was 91% 6 months after introduction of the WASP classification, when diagnosed with high confidence.

    How might it impact on clinical practice in the foreseeable future?

    • The WASP classification system could eventually contribute to realise a safe implementation of the ‘resect and discard’ as well as the ‘resect or leave in’ policy in daily practice.

    Background

    Colorectal cancer (CRC) arises from colorectal precursor lesions that often present as polyps. Detecting and resecting these lesions during colonoscopy decreases CRC incidence.1 Resecting all polyps, including all diminutive and small ones, also has its downsides considering complication risks and associated workload from gastroenterologists and pathologists.2 ,3 Compared with larger polyps (≥10 mm), small (6–9 mm) and diminutive (1–5 mm) polyps less often possess features of advanced neoplasia.4 Accurate optical diagnosis of small and diminutive polyps would allow to resect and discard these polyps without the need for histopathological assessment. Furthermore, all diminutive harmless hyperplastic polyps (HPs) in the rectosigmoid could be characterised and left in situ, resulting in safe and more cost-effective practice.3 ,5 ,6

    Colorectal polyps were historically subdivided into mainly adenomas (ADs) and HPs. ADs were assumed to be the only premalignant lesions, evolving to CRC via the traditional adenoma-carcinoma neoplasia pathway, while HPs were considered benign.7 For this reason international research has mainly focused on the endoscopic differentiation of ADs and HPs in order to increase the accuracy of optical diagnosis.3 ,8–12

    The recent classification of WHO subdivides serrated polyps into HPs, sessile serrated adenomas/polyps (SSA/Ps) and traditional serrated adenomas (TSAs).13 SSA/Ps and TSAs can develop to cancer via the alternative serrated neoplasia pathway, which has been suggested to cause approximately 15–30% of all CRCs.14–16 SSA/Ps comprise up to 12% of all polyps in an asymptomatic average risk population, while the prevalence of TSA is far smaller (<1%).17 Therefore it is also important that all SSA/Ps are detected, correctly differentiated and radically resected during colonoscopy.

    Endoscopic differentiation of SSA/Ps from ADs and HPs is difficult, even when advanced imaging techniques such as narrow band imaging (NBI) are used. As a result, validated classification systems to distinguish ADs, HPs and SSA/Ps using NBI do not currently exist.9 ,11 ,12 A recent study showed that with the use of the NBI International Colorectal Endoscopic (NICE) criteria, SSA/Ps are as often accounted as neoplastic as well as non-neoplastic polyps, which may limit the applicability of the NICE criteria in practice.18 In the past years, specific endoscopic features to characterise SSA/Ps have been described and could potentially be used to optimise the accuracy of optical diagnosis of colonic polyps, including SSA/Ps.19–22

    The aim of this study was to develop the first assessment method for optical diagnosis of small and diminutive colonic polyps which enables to differentiate ADs, HPs and SSA/Ps with the use of NBI and the NICE classification (the Workgroup serrAted polypS and Polyposis (WASP) classification) and to evaluate the improvement in diagnostic test accuracy among consultant gastroenterologists in the differentiation of these polyp subtypes after participation in a training module based on this classification system.

    Methods

    Development of the WASP classification

    Based on a stepwise approach using the NICE classification and the criteria described by Hazewinkel et al,19 an assessment method was developed for the endoscopic differentiation of ADs, HPs and SSA/Ps (figure 1): the WASP classification.11 According to this method, a colonic polyp is first assessed using the NICE criteria aiming to differentiate between polyps resembling HPs (type 1 polyps) and polyps resembling ADs (type 2 polyps).11 Using these criteria, the presence of at least one adenoma-like feature is sufficient to diagnose a type 2 polyp. These features are defined as (1) a darker colour than the surrounding mucosa, (2) prominent brown vessels or (3) an oval, tubular or branched surface pattern.

    Figure 1

    The WASP classification: method for optical diagnosis of hyperplastic polyps, sessile serrated adenomas/polyps and adenomas based on the NICE criteria and the Hazewinkel criteria in a stepwise approach. NICE, NBI International Colorectal Endoscopic; NBI, narrow band imaging; WASP, Workgroup serrAted polypS and Polyposis.

    Subsequently the diagnostic criteria for SSA/Ps as described by Hazewinkel et al19 are used to differentiate between SSA/Ps and HPs for type 1 polyps, and between SSA/Ps and ADs for type 2 polyps. The presence of at least two SSA/P-like features is hereby considered sufficient to diagnose an SSA/P. SSA/P-like features are defined as (1) a clouded surface, (2) indistinctive borders, (3) irregular shape or (4) dark spots inside the crypts.

    Training module development

    One expert gastroenterologist regarding colonic polyps and NBI (ED) and a research fellow (JEGIJ) developed a training module based on the WASP classification system. At first the broad content of the training was designed and subdivided into the following main items:

    • Clinical importance of the differentiation of small and diminutive colonic polyps

    • Differentiation between type 1 and type 2 polyps using the NICE criteria11

    • Characterisation of SSA/P using the Hazewinkel criteria19

    • Endoscopic differentiation of SSA/P, AD and HP via a stepwise approach

    • Training of endoscopic images of polyps with direct feedback (figure 2).

    Figure 2

    Representative screenshot of the training module: NBI image of SSA/P with direct feedback. NBI, narrow band imaging; SSA/P, sessile serrated adenoma/polyp.

    Endoscopic images of small and diminutive polyps were retrieved from a prospectively collected database of polyps, collected in patients with serrated polyposis syndrome during endoscopic surveillance.23 These endoscopic images were used to illustrate the separate criteria for endoscopic differentiation of polyp subtypes as well as for the final training with direct feedback (figure 2). Two gastroenterologists were asked to assess the comprehensibility and feasibility of a preliminary version of the module. Unclear segments were altered and missing information was added to the teaching module. The training module was developed using Microsoft PowerPoint (Microsoft Corporation, Redmond, Washington, USA). Participation in the module was intended to take approximately 20 min of time.

    First validation phase

    The first validation phase was designed to assess the added value of the WASP classification in the increase in accuracy of optical diagnosis of colonic polyps. In total 10 consultant gastroenterologists were asked to participate in the training. All endoscopists were familiar with advanced imaging techniques and the classification of polyp histology using the NICE criteria. Before participation in the training a set of 45 non-magnified, non-manipulated endoscopic images of polyps (15 SSA/Ps, 15 HPs and 15 ADs, based on histology) were presented. For all polyps, a high-resolution white light endoscopy (HR-WLE) image and a NBI image were presented simultaneously. For each polyp, the participant predicted polyp histology, accompanied with a high level or low level of confidence. Patient characteristics as well as polyp location and size were not presented. No feedback was given afterwards. Participants were unaware of the used proportions of polyp subtypes. After the training module, the same set of 45 polyp images was presented in a changed, random order and assessed by the participants under identical circumstances.

    Endoscopic images of small and diminutive polyps were retrieved from the same polyp database as used in the training module. Images of different polyps were selected to be used in the training module and in the validation phase. Only polyps with a HR-WLE and NBI image of sufficient quality were selected.

    Second validation phase

    The second validation phase was designed to assess the long-term learning effect of the introduction of the WASP classification as well as to evaluate the actual height of diagnostic performance, which could be compared with the Preservation and Incorporation of Valuable endoscopic Innovations (PIVI) statements.5 The same gastroenterologists were asked to participate in a second validation phase 6 months later. Participants predicted polyp histology of a new set of 50 non-magnified, non-manipulated endoscopic images of polyps, which consisted of 5 diminutive and 6 small SSA/Ps, 15 diminutive HPs and 1 small HP and 15 diminutive and 8 small ADs, comparable to proportions in daily practice.17 Again, for all polyps a HR-WLE image and a NBI image were presented simultaneously. For each polyp, the participant predicted polyp histology, accompanied with level of confidence. Patient characteristics as well as polyp location and size were not presented. Participants were unaware of the used proportions of polyp subtypes. No additional training was presented to the gastroenterologist before participation in this second validation phase. Images were retrieved from colonoscopies performed in the context of the Dutch population screening programme for CRC.

    Histopathological diagnosis

    For all polyps, the histopathological diagnosis was accounted as the reference standard. Polyps retrieved from the serrated polyposis database were assessed by two specialised GI pathologists (GAM and SvE) according to the revised Vienna criteria.24 Polyps retrieved during population screening were assessed by a specialised GI pathologist during daily practice and afterwards revised by GAM. Polyps were only included if both pathologists agreed on the diagnosis. The histopathological diagnosis was based on the morphological features on H&E staining. An SSA/P was defined as a serrated polyp with at least two irregular dilated crypts, including dilatation of the base of the crypts that often have a boot shape, L shape or inverted T shape.25 During revision the pathologists were blinded for patient characteristics, the endoscopic appearance of the lesions and for the diagnosis made by the other pathologist.

    Ethical approval

    This study was an educational endoscopic image evaluation study and therefore did not require revision by an institutional review board, as in agreement with the medical research involving human subjects act (WMO). This study was carried out in accordance with the Helsinki Declaration.26

    Statistical analysis

    This study was powered based on the improvement in diagnostic accuracy after training during the first validation phase. An improvement in diagnostic accuracy of 10% was assumed to be of clinical relevance. With 15% discordant pairs, a significance level of 5% and a power of 90%, 135 observations were needed per group for paired analysis.27 To satisfy for the fact that a variety of observers scored the same set of polyps an inflation factor was calculated using the formula 1+(n−1)ρ, which is commonly used for clustered matched-pair data.28 Assuming an intrarater correlation of 0.05 and using 45 observations per participant, the inflation factor was calculated to be 3.2 resulting in 432 observations (10 participants) needed for paired analysis.

    The diagnostic accuracy of optical diagnosis was calculated with the histopathological diagnosis as the reference standard. Generalised estimating equations with two levels of non-nested clusters were used to compare the diagnostic performance before and after training.29 Generalised estimating equations were also used to compare the diagnostic performance in both groups for polyps that were diagnosed with a high level of confidence. In order to calculate the diagnostic accuracy of optical diagnosis for polyp subtypes, answers were dichotomised. ADs and SSA/Ps were hereby considered neoplastic polyps, while HPs were considered non-neoplastic.

    The interobserver agreement in polyp diagnosis before and after training, as well as in the second validation phase, was calculated using Fleiss’ κ.30 A κ-value <0.20 was regarded as poor agreement; 0.21–0.40 as fair agreement; 0.41–0.60 as moderate agreement; 0.61–0.80 as good agreement and >0.81 as very good agreement.31 The pretraining and post-training agreements were compared using Wald statistics,32 by assuming that the κ before and the κ after training had a correlation of 0.3 since they are from the same patient group. STATA/IC V.10.1 (StataCorp; College Station, Texas, USA) and R V.2.15.0 (The R Foundation for Statistical Computing) were used for analysis. A p value <0.05 was considered statistically significant. The results of this study were reported in accordance with the standards for reporting of diagnostic accuracy statements for diagnostic accuracy studies as well as with the standards for reliability and agreement studies.33 ,34

    Results

    Participants

    Ten consultant gastroenterologists from nine hospitals were recruited to participate in this study. Five gastroenterologists were working in an academic hospital and five in a regional hospital, all from different parts of the country. All endoscopists but one were trained and accredited to perform colonoscopies in Dutch population screening for CRC, indicating adequate colonoscopy experience and quality.35 All endoscopists were familiar with advanced imaging techniques and the classification of polyp histology using the NICE criteria. In the first validation phase, gastroenterologists participated in the study during live sessions with two to five participants at a time. In the second validation phase gastroenterologists participated via a web-based programme.

    Diagnostic accuracy

    A comparison of the pooled overall diagnostic test accuracy before and after training is presented in table 1. The overall accuracy of optical diagnosis was 0.63 (95% CI 0.54 to 0.71) before training and increased significantly to 0.79 (95% CI 0.72 to 0.86) after training (16% improvement; 95% CI 9% to 22%; p<0.001). The individual overall accuracy of nine participants increased after training, while one participant showed no improvement (figure 3). For polyps diagnosed with a high level of confidence the pooled overall accuracy was 0.73 (95% CI 0.64 to 0.72) before training, which significantly increased to 0.87 (95% CI 0.80 to 0.95) after training (14% improvement; 95% CI 4% to 24%; p<0.01). The individual overall accuracy of eight participants increased after training, while the accuracy of two participants decreased (figure 3). For diminutive as well as small lesions the pooled overall diagnostic accuracy significantly improved after training, unaware of the confidence level of polyp diagnoses (table 1).

    Table 1

    First validation phase: comparison of the overall pooled diagnostic accuracy of optical diagnosis before and after training

    Figure 3

    Comparison of the individual accuracy and combined accuracy of optical diagnosis before and after participation in a training course about the use of the WASP classification. Data are presented for all polyps and for those polyps that were diagnosed with a high level of confidence.

    In table 2 the pooled accuracy before and after training is presented, dichotomised for the characterisation of different polyp subtypes. The pooled accuracy of optical diagnosis increased significantly after training for the characterisation of neoplastic polyps (ADs and SSA/Ps combined) (10% improvement; 95% CI 5% to 15%; p<0.001), SSA/Ps (12% improvement; 95% CI 5% to 17%; p<0.001) and ADs (11% improvement; 95% CI 4% to 14%; p<0.01). The individual accuracy increased for all 10 gastroenterologists after training regarding neoplastic polyps as well as SSA/Ps and for eight gastroenterologists regarding ADs (figure 3). For polyps diagnosed with high confidence similar improvement in pooled accuracy is shown for neoplastic polyps (8% improvement; 95% CI 2% to 15%; p<0.01), SSA/Ps (10% improvement; 95% CI 2% to 18%; p=0.02) and ADs (10% improvement; 95% CI 1% to 19%; p=0.03). The individual accuracy increased for nine gastroenterologists regarding neoplastic polyps as well as SSA/Ps and for eight gastroenterologists regarding ADs (figure 3).

    Table 2

    First validation phase: comparison of the dichotomised pooled diagnostic accuracy of optical diagnosis before and after training

    In table 3 the pooled overall accuracy as well as the pooled dichotomised accuracy of optical diagnosis from the second validation phase is presented. The pooled accuracy of optical diagnosis was 0.76 (95% CI 0.72 to 0.80) overall, 0.87 (95% CI 0.84 to 0.90) when dichotomised for SSA/Ps, 0.83 (95% CI 0.80 to 0.87) when dichotomised for ADs and 0.82 (95% CI 0.79 to 0.86) when dichotomised for neoplastic lesions. The pooled accuracy of optical diagnosis with high confidence was 0.84 (95% CI 0.81 to 0.88) overall, 0.91 (95% CI 0.88 to 0.94) when dichotomised for SSA/Ps, 0.90 (95% CI 0.87 to 0.93) when dichotomised for ADs and 0.89 (95% CI 0.85 to 0.92) when dichotomised for neoplastic lesions.

    Table 3

    Second validation phase: pooled overall and dichotomised diagnostic accuracy of optical diagnosis

    The pooled negative predictive value (NPV) of diminutive neoplastic lesions was 0.83 (95% CI 0.75 to 0.89), which increased to 0.91 (95% CI 0.83 to 0.96) for diminutive neoplastic lesions diagnosed with high confidence (see online supplementary table S1). In total 7 out of 10 gastroenterologists scored an NPV ≥90% for diminutive neoplastic lesions diagnosed with high level of confidence.

    Diagnostic agreement

    In the first validation phase the interobserver agreement of optical polyp diagnoses improved significantly from 0.32 (95% CI 0.28 to 0.35) at baseline to 0.58 (95% CI 0.55 to 0.62) after training about the use of the WASP classification (p<0.001). The interobserver agreement was considered moderate after training. In the second validation phase the interobserver agreement of optical diagnoses remained moderate with a Fleiss’ κ of 0.54 (95% CI 0.51 to 0.57). Interobserver agreement of optical diagnosis with high level of confidence could not be measured, since each individual diagnosed a different set of polyps with high confidence.

    Discussion

    Advanced imaging techniques such as NBI facilitate the endoscopic differentiation of colonic polyp subtypes. Classification systems based on NBI that include the characterisation of common and premalignant SSA/Ps however do not yet exist, hindering optimal practice.9 ,11 We developed the first integrative classification system for optical diagnosis of small and diminutive ADs, HPs and SSA/Ps (the WASP classification), combining the existing NICE classification and criteria by Hazewinkel et al19 in a stepwise approach.11 A structured training about the use of the WASP classification resulted in an overall significant increase in diagnostic accuracy of optical diagnosis for 10 consultant gastroenterologists predicting polyp histology (0.63 vs 0.79; 16% improvement; 95% CI 9% to 22%; p<0.001). For polyps diagnosed with a high level of confidence a similar increase in accuracy was found (0.73 vs 0.87; 14% improvement; 95% CI 4% to 24%; p<0.01). The accuracy for characterisation of SSA/Ps, ADs as well as SSA/Ps plus ADs (neoplastic polyps) also increased significantly after the training. The elevated diagnostic accuracy of optical diagnosis showed to be sustainable during long-term follow-up in an image evaluation setting. After 6 months the overall pooled accuracy was 0.76 (95% CI 0.72 to 0.80), which increased to 0.84 (95% CI 0.81 to 0.88) for polyps diagnosed with high confidence. The accuracy with high confidence was 0.91 (95% CI 0.88 to 0.94) for SSA/Ps and 0.89 (95% CI 0.85 to 0.92) for neoplastic lesions. The pooled NPV with high confidence of diminutive neoplastic lesions was 0.91 (CI 0.83 to 0.96), while 7 out of 10 individuals scored an NPV ≥90%. These results suggest an added value of the WASP classification system to existing characterisation methods.

    Several limitations in this study have to be acknowledged. First of all, still images were used to predict polyp histology. This does not fully resemble the optical diagnosis in daily practice, when polyps are assessed during real-time colonoscopy. This could potentially have led to selection bias, considering that it is not always possible to sufficiently visualise polyps during colonoscopy. On the other hand, real-time images of polyps might be easier to characterise if the polyp is sufficiently visualised, due to the opportunity to visualise the lesion from several angles, possibly improving the accuracy of optical diagnosis with high confidence compared with an image evaluation setting. Since the same polyps were assessed before and after training during the first validation phase, this will probably not have led to biased results considering the main objective of this study. A second potential shortcoming of this study is the fact that most images of polyps were derived from patients with serrated polyposis syndrome. The endoscopic features of polyps retrieved from this patient group could potentially differ from polyps retrieved from sporadic patients. There is however no reason to assume a difference in polyp phenotype of these polyps compared with their sporadic counterparts, and therefore we think that our data can be extrapolated to the general population.

    This study is performed in the context of a broad discussion regarding the advantages and disadvantages of current policy concerning the management of small and diminutive polyps detected during colonoscopy. Small and diminutive polyps rarely have features of advanced neoplasia, and resecting and discarding these polyps could result in a more cost-effective practice (resect and discard policy).3 ,6 Furthermore all diminutive non-neoplastic polyps in the rectosigmoid could be characterised and left in situ (resect or leave in policy). The PIVI statements are composed to safely implement these strategies in daily practice.5 According to the PIVI statements the agreement in assignment of postpolypectomy surveillance intervals using real-time endoscopic assessment of the histology of diminutive polyps should be ≥90% when compared with decisions based on pathology assessment in order to safely discard a polyp without histopathological assessment. Other expert groups have stretched the ‘resect and discard policy’ towards small lesions as well.3 Furthermore it is stated in the PIVI statements that an NPV for the diagnosis of premalignant polyps of ≥90% is mandatory in order not to resect diminutive benign polyps in the rectosigmoid. Many studies on validated criteria for polyp assessment have been performed in order to reach the goals of the PIVI statements, which showed varying results.3 ,8 ,10 ,36

    Kumar et al recently demonstrated the shortcomings of the NICE criteria, as characterisation of SSA/Ps is not included. As a result, premalignant SSA/Ps were not differentiated and equally often diagnosed as neoplastic as well as non-neoplastic polyps.18 Subsequently several experts stated that the resect and discard theory could only be implemented when proper endoscopic differentiation of SSA/Ps is possible.37 ,38 We developed a classification system that could help to differentiate neoplastic polyps, including SSA/Ps, from non-neoplastic polyps. We showed that a short training course on the use of the WASP classification significantly increases the overall accuracy of optical polyp diagnosis, resulting in a long-term overall diagnostic accuracy of 0.84 (95% CI 0.81 to 0.88) for polyps diagnosed with high confidence. An even larger accuracy with high confidence was reached differentiating SSA/Ps (0.91; 95% CI 0.88 to 0.94) as well as neoplastic lesions (0.89; 95% CI 0.85 to 0.92). Furthermore we showed that the combined NPV with high confidence of diminutive neoplastic lesions was 0.91 (95% CI 0.83 to 0.96), while 7 out of 10 individuals scored an NPV ≥90%. Although the accuracy of postpolypectomy surveillance intervals could not be evaluated in this study, introduction of the WASP classification seems promising, for the implementation of the ‘resect and discard policy’ for lesions <10 mm as well as for the implementation of the ‘resect or leave in policy’ for diminutive lesions in the rectosigmoid.

    Three recent studies independently evaluated the usefulness of a short training in the diagnostic accuracy differentiating ADs from HPs in participants with varying endoscopic expertise.27 ,39 ,40 These studies individually showed that a short training increased the accuracy of optical diagnosis. Our study showed an improvement in overall accuracy after training for nine participating gastroenterologists, while one (already relatively high-performing) gastroenterologist showed equal results. This indicates that the training course has a broad beneficial effect on an individual level. This effect can not solely be explained by an improvement in the use of the NICE criteria, since also the accuracy in the diagnosis of SSA/Ps increased for all 10 participants, resulting in a significant improvement overall.

    Future research is needed to assess diagnostic performance using the WASP classification during real-time colonoscopies in daily practice in order to verify if the WASP classification is truly suitable for the implementation of the ‘resect and discard’ as well as the ‘resect or leave in’ policies, for expert as well as non-expert gastroenterologists.3 Furthermore integrative classification systems should be developed and validated to endoscopically differentiate polyp subtypes with the use of other virtual chromoendoscopy techniques, such as high definition (HD) I-scan and flexible spectral imaging colour enhancement (FICE).

    In conclusion, we developed and validated the first classification system for optical diagnosis of small and diminutive ADs, SSA/Ps and HPs using NBI. Endoscopic differentiation using the WASP classification showed a significant increase in the accuracy of optical diagnosis overall among experienced gastroenterologists as well as the accuracy characterising neoplastic lesions and SSA/Ps in particular, which proved to be sustainable after 6 months. This is the first classification system that facilitates the characterisation of all prevalent polyp subtypes, including premalignant SSA/Ps and could eventually contribute to realise a safe implementation of targeted polyp treatment in daily practice.5 Before this could be realised, the diagnostic performance of the WASP criteria should be assessed in daily practice and compared with conditions as set by the PIVI statements.

    Acknowledgments

    The authors thank Dr Annekatrien Depla and Dr Henriette Tuynman for their participation in the validation phase of the WASP criteria and Ymkje Mirjam van der Brug for her overall assistance in the realisation of this study.

    References

    View Abstract

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Collaborators Jan J Koornstra; Iris D Nagtegaal; Workgroup serrAted polypS and Polyposis (WASP).

    • Contributors All authors have contributed to this article in the following way: Substantial contributions to the conception or design of the work, or the acquisition, analysis or interpretation of data. Drafting the work or revising it critically for important intellectual content. Final approval of the version published. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Conception and design: JEGIJ, ED. Data acquisition: all authors. Data analysis and interpretation: JEGIJ, JW, ED. Drafting the manuscript: JEGIJ, JW. Critical revision of the manuscript: all authors. Statistical analysis: JEGIJ, JW. Supervision: all authors.

    • Competing interests None.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement The developed computer based training course is available for everyone to use. Those interested can contact the corresponding author.

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.