Accepted Manuscript What is the Hang Up With Optical Diagnosis of Diminutive Colorectal Polyps? Sarah McGill, Swati G. Patel PII: S1542-3565(19)30377-5 DOI: https://doi.org/10.1016/j.cgh.2019.04.017 Reference: YJCGH 56442 To appear in: Clinical Gastroenterology and Hepatology Accepted Date: 5 April 2019 Please cite this article as: McGill S, Patel SG, What is the Hang Up With Optical Diagnosis of Diminutive Colorectal Polyps?, Clinical Gastroenterology and Hepatology (2019), doi: https://doi.org/10.1016/ j.cgh.2019.04.017. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
What is the Hang Up With Optical Diagnosis of Diminutive Colorectal Polyps? KEY WORDS: Colonoscopy, optical diagnosis, diminutive polyp, adenoma, sessile serrated polyp, competence AUTHORS: Sarah McGill 1 & Swati G. Patel 2,3 INSTITUTIONS: 1. University of North Carolina, Chapel Hill, North Carolina, United States 2. University of Colorado Anschutz Medical Center, Aurora, Colorado, United States. 3. Rocky Mountain Regional Veterans Affairs Hospital, Aurora, Colorado, United States. CONFLICTS OF INTEREST: None ABBREVIATIONS: Colorectal Cancer (CRC), American Gastroenterological Association (AGA), American Society for Gastrointestinal Endoscopy (ASGE), narrow-band imaging international colorectal endoscopic (NICE), Workgroup on serrated polyps (WASP), Computer-aided diagnosis (CAD) CORRESPONDING AUTHOR: Swati G Patel, MD MS 12631 E 17 th Avenue, Room 7614, Campus Box 158 Aurora, Colorado 80045 Swati.Patel@ucdenver.edu No writing assistance was provided for this manuscript.
Word Count: 1,949 AUTHOR CONTRIBUTIONS: S. McGill & S. Patel contributed equally to this manuscript and author order is listed alphabetically.
Over a decade ago, Yoshida and colleagues applied the push-button technology of electronic chromoendoscopy to colorectal polyps and differentiated adenomas from non-neoplasms (figure 1). 1 Since then, intense study has found that both experts and non-experts can perform this so-called optical diagnosis with high, not perfect, fidelity to pathology with tools already at the fingertips of endoscopists. 2 Yet universal histology for polyps remains the standard in the U.S., and optical diagnosis has not replaced it. Here we describe the current state of knowledge in the field related to diminutive ( 5 mm) polyps, and the steps that must be achieved before optical diagnosis can be put into routine clinical practice. Optical Diagnosis: Why and How Colonoscopy and polypectomy is the dominant colon screening modality in the United States 3 and in large part accounts for the progressive decline in colorectal cancer incidence and mortality over the last several decades. 4 Although colonoscopy is cost-effective when taking into account the high cost of colon cancer treatment, 5 the overall cost will continue to rise as colonoscopy uptake improves and more patients enter polyp surveillance protocols. Diminutive colorectal polyps are the most commonly found polyps during colonoscopy 6 and rarely harbor advanced histology or malignancy. 7 If diminutive polyp histology can be determined in real-time, polyps confidently diagnosed as adenomas can be resected and discarded, and non-neoplastic polyps in the rectosigmoid colon can be left in place, leading to enormous cost-savings. 8, 9 Endoscopists can always send polyps for pathology, if the optical diagnosis is not clear. Society Support of Optical Diagnosis Optical diagnosis is not incorporated into current U.S. practice guidelines for colorectal cancer screening. European and Japanese guidelines do recommend leaving in tiny polyps confidently
diagnosed as hyperplastic in the rectosigmoid colon, 10, 11 and we wonder if these abundant lesions are also routinely ignored by many, or most, experienced U.S. endoscopists. The idea of using imaging to guide management of both neoplastic and non-neoplastic polyps has been endorsed by major U.S. societies. An AGA Expert Panel concluded in 2015 that endoscopists who demonstrate competency and achieve set thresholds, discussed below, should use optical diagnosis to guide management of diminutive polyps in a setting where performance is regularly audited. 12 The American Society for Gastrointestinal Endoscopy (ASGE) initially published a Preservation and Incorporation of Valuable Endoscopic Innovations (PIVI) 13 statement to set standards for the realtime assessment of diminutive colorectal polyps. The PIVI statement suggested that for diminutive polyps characterized with high-confidence, endoscopists should achieve a 90% negative predictive value (NPV) for adenomas in the rectosigmoid colon and a 90% agreement in surveillance intervals compared to those dictated by histology. In 2015, the ASGE performed a meta-analysis of diagnostic accuracy studies. 14 Following this, it concluded that Narrow Band Imaging already met the set thresholds for clinical implementation when used by trained endoscopists with high confidence, and stated the organization s commitment to establish standards of practice, quality assurance and training programs for the new paradigm. Training on Polyp Differentiation, Gangnam-Style The ability to optically diagnose polyp histology is not innate and requires training and time. In the current study by Bae et al, 15 15 endoscopists at Seoul National University entered an intense training program they dubbed Gangnam-READI. Predictions were based on the Workgroup on serrated polyps (WASP) classification, 16 which uses the narrow-band imaging international colorectal endoscopic (NICE) classification 17 with four additional features to characterize sessile serrated polyps. Training began with a 30-minute audiovisual presentation with pre- and post-test and continued with 8
months of individualized feedback on their polyp predictions at colonoscopy, interactive seminars for low performers, and 3 months of email-based case study called Today s Polyp. These efforts bore fruit, though not all of the endoscopists reached the recommended thresholds for implementation. In the last phase of the study, all 14 endoscopists studied achieved the NPV threshold for leaving in rectosigmoid hyperplastic polyps, and nine of the fourteen achieved the threshold for surveillance interval concordance. Endoscopist Training & Performance Although early studies showed that endoscopists without prior expertise in optical diagnosis were unable to achieve the benchmarks set forth by the ASGE, 18-22 multiple recent studies, 23-25 including this one, 15 have shown that endoscopists who undergo high-quality training, as a group, can achieve the PIVI benchmarks. There are several considerations in deciding whether this data can translate to routine clinical practice. As Bae et al outline, a structured training program is a necessary prerequisite to successful optical diagnosis in vivo. Endoscopy practices would need to have access to validated training methods, dedicate resources to ensuring training is available to endoscopists and allow endoscopists time for training and re-training as needed. Furthermore, after training is complete, resources would need to be dedicated to ensuring performance is proctored to ensure endoscopists are meeting benchmarks. In addition to providing these resources to practicing endoscopists, gastroenterology training programs would need to incorporate optical diagnosis into training curricula. A main barrier to implementing an optical diagnosis strategy is likely the lack of resources needed to train endoscopists and monitor competence. It is important to note that although the participants achieved the PIVI thresholds as a group, not all endoscopists are able to achieve the thresholds. Studies to date have not been designed or powered to assess individual performance, thus most do not report individual endoscopist outcomes. Of
15, 18, 23, 25 the studies that do, only 17-59% of endoscopists were able to achieve both PIVI thresholds. Thus, endoscopy groups would need to decide which endoscopists can practice optical diagnosis and how to monitor performance over time to ensure ongoing competence. This variability in performance is likely a deterrent to universally adopting the strategy. Endoscopist, Patient and Payor Acceptance In our opinion, adequate training programs are not sufficient to inspire most physicians to embrace optical diagnosis shifts in payment models are likely also necessary to support this change. Fee-for-service models, in which endoscopists charge additionally for polypectomy and may have financial benefits from sending pathology, favor the present practice. Bundled reimbursement models, whereby the physician is paid a set fee for all associated costs of the colonoscopy, may provide needed incentives for change. Ideally, patients would also find the strategy worthwhile. Patients at a dermatology office understand that not all moles are premalignant and warrant removal. We may explain this similar concept at colonoscopy, additionally noting that high-confidence optical diagnosis of diminutive adenomas is suitable to calculate surveillance intervals. Studies on patient acceptance of an optical diagnosis strategy for diminutive polyps have shown varying degrees of approval. A study of hundreds of American colonoscopy patients found that twothirds of them were willing to have diminutive polyps discarded and not sent to histology. 26 Among those who were not willing, about half would want the risk of cancer to be zero in diminutive polyps to forego histology. By contrast at another U.S. center, two-thirds of patients were willing to pay out of pocket for histology of their diminutive polyps. 27 Medico-Legal Issues
Concerns over liability may loom large for physicians considering a resect-and-discard / diagnose-and-leave-in paradigm. The risk that a diminutive neoplasm not sent to pathology was in fact cancer is very remote. The risk that a diminutive rectosigmoid lesion confidently diagnosed as hyperplastic but in fact cancer, and left in the colon, is remoter still. Yet they are not theoretically nil. We feel these concerns are vastly overshadowed by the everyday hazards of colonoscopy missed mucosal surface area, undetected flat, subtle lesions, and remnant polyp at resection. Quality Assurance In addition to ensuring endoscopists reach and maintain proposed thresholds, colonoscopy quality programs would need to determine how to monitor quality metrics that are currently reliant on pathologic assessment of polyps. Systematic, high-definition photo-documentation is reliable to accurately assess adenoma detection rates. 28 This approach would also require periodic auditing, presumably by optical diagnosis experts, to ensure accurate reporting. Again, this is a resource that many institutions may not have access to. Cost Effectiveness Prior cost-effectiveness analyses have estimated cost savings ranging from 33 million dollars annually 8 to over a billion dollars of upfront savings. 9 However, it is unclear if these cost savings can be generalized to widespread application of optical diagnosis as these prior analyses relied on assumptions about optical diagnostic performance among experts in advanced imaging. 29-31 The proportion of highconfidence diagnoses is significantly lower among endoscopists naïve to advanced imaging compared to experts (70-78% 15, 22, 23 vs 80-90% 29, 30 ). Thus, there would be a larger proportion of polyps that require histologic analysis (and associated cost) than previously assumed. Furthermore, although endoscopists in recent studies are meeting the minimum 90% threshold in agreement of surveillance intervals recommended by the ASGE, the proportion is lower than that
reported by experts (90-91% 15, 22, 23 vs 93-98% 24, 29, 30 ). The vast majority of disagreement would bring patients back for sooner surveillance 23 in the optical diagnostic strategy and result in more frequent colonoscopies than if surveillance intervals were based on histology. As more endoscopists are trained in optical diagnosis, it is important to confirm that the strategy will be cost-effective if adopted in the US. Cost-effectiveness would also need to take into account the resources needed to train endoscopists, ensure that endoscopists maintain performance benchmarks and audit performance for quality purposes. Is Artificial Intelligence the Future of Optical Diagnosis? Even following vigorous training programs as described by Bae et al., physicians are variable in their competence to diagnose polyps optically. In addition, human experts are limited in their ability to serve as auditors for an optical diagnosis program and provide the new diagnostic gold standard via image review. Automated systems could potentially meet both of these challenges. Theoretically, computer-aided diagnosis (CAD) systems could give their own assessments of polyp pathology. Used regularly, such a system could act as a teacher, particularly for lower-performing physicians. If CAD systems could achieve high fidelity to histology and these diagnoses could be effortlessly archived along with high-resolution images, other hurdles such as physician acceptance and cost-effectiveness can be overcome. Recent studies employing deep convolutional neural networks augur that such a future may not be too distant. Most recently, researchers at Showa University in Japan, a group that has long pioneered optical diagnosis technique, took their CAD live into the colonoscopy suite and integrated it into the workflow. They imaged polyps with over 500x magnification with an endocytoscope, and a CAD system hooked directly into their processor. In the rectosigmoid colon, performance was excellent for
diminutive polyps and met PIVI thresholds (NPV 95.2%, CI 90.3-98%). Diagnostic accuracy was worse in the right colon, where the system over-predicted non-neoplasms. CAD programs tested ex-vivo had even better results. One that was trained on thousands of high-magnification polyp images and then tested on another set of such images yielded results favorable to those of imaging experts. 32 Similarly, results of a program that trained on hundreds of colonoscopy video and then tested on videos. 33 Conclusions In conclusion, the results from the Gangnam READI add to a robust body of literature that dedicated programs can train a majority but not all non-expert physicians to recommended levels of competency to forego histology and optically diagnose, then treat, colorectal polyps. Training programs are available for motivated physicians, but not yet standardized. Cost-effectiveness analyses estimate major cost savings when pathology for most diminutive polyps is avoided, however do not take into account the cost of training and maintenance of competence. Current payment models and assessments of quality in endoscopy disincentivize this paradigm shift. Artificial intelligence holds the potential to optimize accuracy of optical diagnosis while providing a means to efficiently audit performance, and even monitor quality without the histology of diminutive lesions. Although there is clear evidence that endoscopists can achieve the performance thresholds set forth by professional societies, the boost that real-time diagnosis needs to overcome the stated barriers may very well come from artificial intelligence support. Figure 1. Diminutive hyperplastic polyp (left) and adenomatous polyp (right) under electronic chromoendoscopy (Narrow band imaging). Image courtesy of Tonya Kaltenbach.
References 1. Machida H, Sano Y, Hamamoto Y, et al. Narrow-band imaging in the diagnosis of colorectal mucosal lesions: a pilot study. Endoscopy 2004;36:1094-8. 2. McGill SK, Evangelou E, Ioannidis JP, et al. Narrow band imaging to differentiate neoplastic and non-neoplastic colorectal polyps in real time: a meta-analysis of diagnostic operating characteristics. Gut 2013. 3. Steele CB, Rim SH, Joseph DA, et al. Colorectal cancer incidence and screening - United States, 2008 and 2010. MMWR Surveill Summ 2013;62 Suppl 3:53-60. 4. Edwards BK, Ward E, Kohler BA, et al. Annual report to the nation on the status of cancer, 1975-2006, featuring colorectal cancer trends and impact of interventions (risk factors, screening, and treatment) to reduce future rates. Cancer 2010;116:544-73. 5. Lansdorp-Vogelaar I, Knudsen AB, Brenner H. Cost-effectiveness of colorectal cancer screening. Epidemiol Rev 2011;33:88-100. 6. Lieberman D, Moravec M, Holub J, et al. Polyp size and advanced histology in patients undergoing colonoscopy screening: implications for CT colonography. Gastroenterology 2008;135:1100-5. 7. Ponugoti PL, Cummings OW, Rex DK. Risk of cancer in small and diminutive colorectal polyps. Dig Liver Dis 2017;49:34-37. 8. Hassan C, Pickhardt PJ, Rex DK. A resect and discard strategy would improve cost-effectiveness of colorectal cancer screening. Clin Gastroenterol Hepatol 2010;8:865-9, 869 e1-3. 9. Kessler WR, Imperiale TF, Klein RW, et al. A quantitative assessment of the risks and cost savings of forgoing histologic examination of diminutive polyps. Endoscopy 2011;43:683-91. 10. Ferlitsch M, Moss A, Hassan C, et al. Colorectal polypectomy and endoscopic mucosal resection (EMR): European Society of Gastrointestinal Endoscopy (ESGE) Clinical Guideline. Endoscopy 2017;49:270-297. 11. Tanaka S, Saitoh Y, Matsuda T, et al. Evidence-based clinical practice guidelines for management of colorectal polyps. J Gastroenterol 2015;50:252-60. 12. Lieberman D, Brill J, Canto M, et al. Management of Diminutive Colon Polyps Based on Endoluminal Imaging. Clin Gastroenterol Hepatol 2015;13:1860-6; quiz e168-9. 13. Rex DK, Kahi C, O'Brien M, et al. The American Society for Gastrointestinal Endoscopy PIVI (Preservation and Incorporation of Valuable Endoscopic Innovations) on real-time endoscopic assessment of the histology of diminutive colorectal polyps. Gastrointest Endosc 2011;73:419-22. 14. Committee AT, Abu Dayyeh BK, Thosani N, et al. ASGE Technology Committee systematic review and meta-analysis assessing the ASGE PIVI thresholds for adopting real-time endoscopic assessment of the histology of diminutive colorectal polyps. Gastrointest Endosc 2015;81:502 e1-502 e16. 15. Bae JH, Lee C, Kang HY, et al. Improved Real-Time Optical Diagnosis of Colorectal Polyps Following a Comprehensive Training Program. Clin Gastroenterol Hepatol 2019. 16. JE IJ, Bastiaansen BA, van Leerdam ME, et al. Development and validation of the WASP classification system for optical diagnosis of adenomas, hyperplastic polyps and sessile serrated adenomas/polyps. Gut 2016;65:963-70.
17. Hayashi N, Tanaka S, Hewett DG, et al. Endoscopic prediction of deep submucosal invasive carcinoma: validation of the narrow-band imaging international colorectal endoscopic (NICE) classification. Gastrointest Endosc 2013;78:625-32. 18. Ladabaum U, Fioritto A, Mitani A, et al. Real-time optical biopsy of colon polyps with narrow band imaging in community practice does not yet meet key thresholds for clinical decisions. Gastroenterology 2013;144:81-91. 19. Paggi S, Rondonotti E, Amato A, et al. Resect and discard strategy in clinical practice: a prospective cohort study. Endoscopy 2012;44:899-904. 20. Kuiper T, Marsman WA, Jansen JM, et al. Accuracy for optical diagnosis of small colorectal polyps in nonacademic settings. Clin Gastroenterol Hepatol 2012;10:1016-20; quiz e79. 21. Vu HT, Sayuk GS, Hollander TG, et al. Resect and discard approach to colon polyps: real-world applicability among academic and community gastroenterologists. Digestive Diseases & Sciences 2015;60:502-8. 22. Rees CJ, Rajasekhar PT, Wilson A, et al. Narrow band imaging optical diagnosis of small colorectal polyps in routine clinical practice: the Detect Inspect Characterise Resect and Discard 2 (DISCARD 2) study. Gut 2017;66:887-895. 23. Patel SG, Schoenfeld P, Kim HM, et al. Real-Time Characterization of Diminutive Colorectal Polyp Histology Using Narrow-Band Imaging: Implications for the Resect and Discard Strategy. Gastroenterology 2016;150:406-18. 24. Kaltenbach T, Rastogi A, Rouse RV, et al. Real-time optical diagnosis for diminutive colorectal polyps using narrow-band imaging: the VALID randomised clinical trial. Gut 2014. 25. Vleugels JLA, Dijkgraaf MGW, Hazewinkel Y, et al. Effects of Training and Feedback on Accuracy of Predicting Rectosigmoid Neoplastic Lesions and Selection of Surveillance Intervals by Endoscopists Performing Optical Diagnosis of Diminutive Polyps. Gastroenterology 2018;154:1682-1693 e1. 26. Rex DK, Patel NJ, Vemulapalli KC. A survey of patient acceptance of resect and discard for diminutive polyps. Gastrointest Endosc 2015;82:376-380 e1. 27. Vu HT, Sayuk GS, Gupta N, et al. Patient preferences of a resect and discard paradigm. Gastrointest Endosc 2015;82:381-384 e1. 28. Rex DK, Hardacker K, MacPhail M, et al. Determining the adenoma detection rate and adenomas per colonoscopy by photography alone: proof-of-concept study. Endoscopy 2015;47:245-50. 29. Ignjatovic A, East JE, Suzuki N, et al. Optical diagnosis of small colorectal polyps at routine colonoscopy (Detect InSpect ChAracterise Resect and Discard; DISCARD trial): a prospective cohort study. Lancet Oncol 2009;10:1171-8. 30. Rex DK. Narrow-band imaging without optical magnification for histologic analysis of colorectal polyps. Gastroenterology 2009;136:1174-81. 31. Rastogi A, Keighley J, Singh V, et al. High accuracy of narrow band imaging without magnification for the real-time characterization of polyp histology and its comparison with high-definition white light colonoscopy: a prospective study. Am J Gastroenterol 2009;104:2422-30. 32. Chen PJ, Lin MC, Lai MJ, et al. Accurate Classification of Diminutive Colorectal Polyps Using Computer-Aided Analysis. Gastroenterology 2018;154:568-575. 33. Byrne MF, Chapados N, Soudan F, et al. Real-time differentiation of adenomatous and hyperplastic diminutive colorectal polyps during analysis of unaltered videos of standard colonoscopy using a deep learning model. Gut 2019;68:94-100.