Rationale for developing competence standards for optical diagnosis of diminutive colorectal polyps
Diminutive colorectal polyps, with a negligible risk of harboring cancer, constitute up to 60 % of all colorectal polyps [3 ]
[4 ]. The current management of polyps, including diminutive polyps, is to resect and submit them all for histological assessment. While the detection and removal of polyps contributes toward the reduction in colorectal cancer (CRC), histological assessment of these diminutive polyps results in substantial burdens and costs for colonoscopy units. These costs and burdens could be reduced by implementing an “optical diagnosis strategy” for diminutive colorectal polyps [5 ]. In this optical diagnosis strategy, endoscopists or artificial intelligence systems diagnose diminutive polyps during colonoscopy with high or low confidence. When these diminutive polyps are diagnosed with high confidence, they can be resected and discarded without histological evaluation (i. e. the “resect-and-discard strategy”). In addition, non-neoplastic lesions located in the rectum and sigmoid can be left in situ without resection, as they have no malignant potential (i. e. the “leave-in-situ strategy”) [6 ]
[7 ].
Though not yet achieved, the implementation of optical diagnosis for diminutive colorectal polyps into routine clinical practice remains an important goal as it would greatly reduce colonoscopy-associated costs [8 ]
[9 ]. The implementation of this strategy is therefore also endorsed by several international societies [10 ]
[11 ]
[12 ].
However, because misdiagnosis of diminutive lesions can result in inappropriate surveillance intervals and neoplastic lesions being left in situ, proficiency in optical diagnosis must be guaranteed before implementation. To address this, the ESGE Curricula Working Group for optical diagnosis training has comprehensively described the major training steps to achieve and maintain proficiency in optical diagnosis of diminutive colorectal polyps [1 ]
[2 ]. ESGE suggests that an endoscopist is competent after attending a validated training course, including an in vivo phase, and after reaching the endorsed competence standards during real-time colonoscopy. The currently endorsed PIVI competence criteria are however impractical and difficult to implement in daily practice. Developing alternative, easy-to-measure competence criteria might facilitate the implementation of the optical diagnosis strategy in clinical practice.
SODA competence standards for implementation of the optical diagnosis strategy for diminutive colorectal polyps
Leave-in-situ strategy
1 In order to implement the leave-in-situ strategy for diminutive colorectal lesions (1–5 mm), it is clinically acceptable if, during real-time colonoscopy, at least 90 % sensitivity and 80 % specificity is achieved for high confidence endoscopic characterization of colorectal neoplasia of 1–5 mm in the rectosigmoid. Histopathology is used as the gold standard.
Level of agreement 95 % (14 Strongly agree; 7 Agree; 0 Neither agree nor disagree; 1 Disagree; 0 Strongly disagree).
The panel took into account a number of considerations in the development of this SODA competence standard for implementation of the leave-in-situ strategy, which are detailed in the following sections.
Potential impact on short-term consequences
To establish safe and easy-to-measure competence standards for the leave-in-situ strategy for diminutive colorectal lesions, decisions should be made on clinically acceptable short-term consequences of implementing this strategy. When applying the leave-in-situ part of the optical diagnosis strategy, not recognizing neoplastic diminutive lesions could result in neoplastic lesions being left in the rectosigmoid, whereas not recognizing non-neoplastic diminutive lesions could result in unnecessary polypectomies of non-neoplastic lesions in the rectosigmoid.
In addition, it could be argued that the correct diagnosis of each polyp subtype is not of equal clinical importance. The clinical implications of incorrectly diagnosing neoplastic lesions in the rectosigmoid (i. e. leaving in situ a neoplastic lesion) and incorrectly diagnosing non-neoplastic lesions in the rectosigmoid (i. e. performing unnecessary polypectomy of a non-neoplastic lesion) are completely different. Therefore, from a patient safety point of view, correctly diagnosing non-neoplastic lesions may not be as clinically important as correctly diagnosing neoplastic lesions. However, from a time and cost-effectiveness point of view, correctly diagnosing non-neoplastic lesions is also important. Therefore, competence standards should take both scenarios into account.
To provide guidance for setting alternative competence standards for the optical diagnosis of diminutive colorectal polyps, a recent simulation study determined the relationship between the proportion of correctly optically diagnosed diminutive polyps and the above-mentioned short-term clinical consequences (i. e. [non]neoplastic lesions that would remain in situ in the rectosigmoid and surveillance interval agreement).
Short description of the simulation approach used [
16 ].
In this simulation approach, a virtual endoscopist/an artificial intelligence system performed optical diagnosis of diminutive polyps with a fixed diagnostic performance level (“strategy”) on two existing cohorts of patients who underwent colonoscopy in either a primary colonoscopy screening or fecal immunochemical test setting [18 ]
[19 ]. A total of 756 strategies were defined by systematically varying the proportion of correct optical diagnoses for each polyp subtype (i. e. adenomas, sessile serrated lesions, hyperplastic polyps). For each strategy the short-term clinical consequences (i. e. surveillance interval agreements, number of [non]neoplastic lesions left in situ in the rectosigmoid) were determined using Monte Carlo Sampling with 1000 repetitions.
This simulation study showed that the proportion of correctly diagnosed diminutive polyps (i. e. adenomas, sessile serrated lesions [SSLs], hyperplastic polyps [HPPs]) could be used as easy-to-implement competence standards for the performance of optical diagnosis of diminutive polyps. However, the question remained as to what an endoscopy society would consider clinically acceptable competence standards. Different proportions of correctly diagnosed diminutive polyps lead to different clinical consequences depending on the clinical setting and the surveillance guidelines used.
[Table 1 ] shows the short-term clinical consequences of implementing the leave-in-situ strategy using a range of settings for the proportion of correctly optically diagnosed adenomas and HPPs in a primary colonoscopy and fecal immunochemical test (FIT)-positive screening cohort. Results in this table are presented per 1000 individuals with at least one diminutive polyp. The impact on the number of diminutive neoplastic lesions that would remain in situ in the rectosigmoid and the number of unnecessary polypectomies that would be avoided (i. e. the number of diminutive non-neoplastic lesions that would be left in situ) is shown. For example, when, in a colonoscopy screening cohort, 90 % of all diminutive adenomas and 80% of all diminutive HPPs were diagnosed correctly by optical diagnosis, 26 diminutive neoplastic lesions would remain in situ in the rectosigmoid per 1000 patients (i. e. one diminutive neoplastic lesion in every 38 patients with a diminutive lesion), representing 8 % of all diminutive neoplastic rectosigmoid lesions. In addition, 324 unnecessary polypectomies would be avoided per 1000 patients, representing a 52 % reduction in unnecessary polypectomies (i. e. diminutive non-neoplastic lesions that would be left in situ).
Table 1
Impact of implementing the leave-in-situ strategy with different proportions of correctly optically diagnosed ≤5-mm adenomas and hyperplastic polyps (HPPs) in a primary colonoscopy screening and fecal immunochemical test (FIT)-positive screening cohort on the overall proportion of correctly diagnosed ≤5-mm polyps and the number of (non-)neoplastic lesions left in situ in the rectosigmoid, with histopathology as the gold standard. Results are presented per 1000 individuals with at least one ≤5-mm polyp.
Strategy[1 ]
Colonoscopy screening individuals
FIT-positive screening individuals
Correctly diagnosed ≤ 5-mm adenomas, %
Correctly diagnosed ≤ 5-mm HPPs, %
Correctly diagnosed ≤ 5-mm polyps, %
Neoplastic lesions of ≤ 5 mm left in situ, n (%)[2 ]
Non-neoplastic lesion of ≤ 5 mm left in situ, n (%)[3 ]
Correctly diagnosed ≤ 5-mm polyps, %
Neoplastic lesions of ≤ 5 mm left in situ, n (%)[2 ]
Non-neoplastic lesion of ≤ 5 mm left in situ, n (%)[3 ]
60
60
59
80 (26 %)
243 (39 %)
60
143 (26 %)
137 (39 %)
60
80
68
80 (26 %)
324 (52 %)
64
143 (26 %)
182 (52 %)
60
100
78
80 (26 %)
404 (65 %)
69
143 (26 %)
227 (65 %)
70
60
63
62 (20 %)
243 (39 %)
67
107 (20 %)
137 (39 %)
70
80
73
62 (20 %)
324 (52 %)
72
107 (20 %)
182 (52 %)
70
100
83
62 (20 %)
404 (65 %)
77
107 (20 %)
227 (65 %)
80
60
68
44 (14 %)
243 (39 %)
74
72 (14 %)
137 (39 %)
80
80
78
44 (14 %)
324 (52 %)
79
72 (14 %)
182 (52 %)
80
100
87
44 (14 %)
404 (65 %)
84
72 (14 %)
227 (65 %)
90
60
73
26 (8 %)
243 (39 %)
82
37 (8 %)
137 (39 %)
90
80
82
26 (8 %)
324 (52 %)
87
37 (8 %)
182 (52 %)
90
100
92
26 (8 %)
404 (65 %)
91
37 (8 %)
227 (65 %)
100
60
77
8 (2 %)
243 (39 %)
89
2 (2 %)
137 (39 %)
100
80
87
8 (2 %)
324 (52 %)
94
2 (2 %)
182 (52 %)
100
100
96
8 (2 %)
404 (65 %)
99
2 (2 %)
227 (65 %)
1 At each given proportion of correctly diagnosed diminutive adenomas and HPPs, we assumed that 40 % of all diminutive sessile serrated lesions would be diagnosed correctly with high confidence to ensure that the clinical consequences are a reflection of daily practice.
2 The number and proportion of all diminutive neoplastic lesions that would be left in situ in the rectosigmoid in a cohort of 1000 individuals.
3 The proportion and number of unnecessary polypectomies that would be avoided per 1000 individuals (i. e. diminutive non-neoplastic lesions that would remain in situ).
Table 1 s (see online-only Supplementary material) shows the impact on the number of diminutive neoplastic lesions that would remain in situ in the rectosigmoid and the number of unnecessary polypectomies that would be avoided if patients without a polyp were also included.
In addition, this simulation study showed that increasing and decreasing the proportion of high confidence diagnoses of all types of diminutive polyps by 10 % barely affected the clinical outcome (i. e. there was hardly any impact on the short-term clinical consequences). As the malignant potential of SSLs is increasingly being emphasized, in this simulation approach, the impact of correctly optically diagnosing diminutive SSLs was also assessed [21 ]
[22 ]. The simulation showed that the proportion of correctly diagnosed diminutive SSLs barely affects surveillance interval agreement and the number of (non)neoplastic lesions that would remain in situ. The fact that diminutive SSLs have little relevance is likely to be because of their low prevalence. In addition, individuals often have synchronous adenomas and/or large lesions found at colonoscopy that predominantly determine the surveillance intervals. However, the influence of SSLs on surveillance and the number of SSLs that would remain in situ might expand in the future as the detection rates of SSLs are currently increasing owing to increasing awareness of these subtle lesions among endoscopists [23 ].
Potential impact on long-term consequences
The impact of implementing the leave-in-situ strategy on the long-term consequences, such as increase in CRC incidence, CRC mortality, and cost-effectiveness, should be considered. Two studies have evaluated the implementation of the full optical diagnosis strategy (i. e. leave-in-situ and resect-and-discard) [9 ]
[10 ] ([Table 2 ]).
Table 2
Overview of optical diagnosis modelling studies: base-case assumptions, increase colorectal cancer (CRC) burden, and lifetime cost-savings when implementing the optical diagnosis strategy calculated per individual.
Study
Screening program
Base-case assumptions for optical diagnosis of diminutive colorectal polyps
Cost-savings per individual
Increased lifetime risk of CRC
Increased lifetime mortality risk from CRC
Sensitivity for neoplastic lesions
Sensitivity for non-neoplastic lesions
High confidence predictions
Vleugels et al. [9 ]
FIT screening
92 % adenomas; 91 % SSLs
88 %
76 %
€6
0.00466 %
0.00141 %
NICE (2017) Picot et al. [10 ]
[24 ]
FIT screening
NBI 91 %; FICE 81.4 %; iSCAN 96.2 %
NBI 81.9 %; FICE 85.0 %; iSCAN 90.6 %
78.6 %
NBI £6.11; FICE £887.70; iSCAN £8.49
NBI 0.005 %; FICE 0.02 %; iSCAN 0.004 %
NR
Kessler et al.[1 ]
[25 ]
Primary colonoscopy
90 %
90 %
100 %
$174
0.0076 %[2 ]
Hassan et al.[1 ]
[8 ]
Primary colonoscopy
94 %
89 %
83 %
$25
NR
NR
SSL, sessile serrated lesion; NICE, National Institute for Health and Care Excellence; FICE, flexible spectral imaging color enhancement; NBI, narrow-band imaging; NR, not reported.
1 Modelling studies that only incorporated the resect-and-discard part of the optical diagnosis strategy.
2 Missed CRCs.
Using the Adenoma and Serrated pathway to Colorectal CAncer (ASCCA) model, Vleugels et al. simulated a biennial FIT screening program, in which they compared an optical diagnosis strategy with the current strategy, submitting all diminutive polyps for histopathology assessment [9 ]. Implementation of an optical diagnosis strategy in a FIT-based screening program led to one CRC case, one additional CRC death in a cohort of 10 000 individuals, and a cost-saving of €6 per individual.
The National Institute for Health and Care Excellence (NICE) diagnostic guidance on “Virtual chromoendoscopy to assess colorectal polyps during colonoscopy” published in 2017 [10 ]
[24 ] modelled CRC risk and economic consequences of adopting the complete optical diagnosis strategy using a range of narrowed spectrum technologies (narrow-band imaging [NBI], flexible spectral imaging color enhancement [FICE], iSCAN). Very similar lifetime risks of CRC were reported using histopathology (3.025 %) versus narrowed spectrum-based optical diagnosis (3.020 %–3.045 %) to determine surveillance intervals. NBI and iSCAN were shown to be economically dominant to histopathology in all scenario analyses. Therefore, both modelling studies demonstrate that implementing the optical diagnosis strategy (which includes the leave-in-situ strategy) only marginally influenced long-term outcomes, such as CRC incidence and mortality, whilst saving costs.
Risk of leaving in situ a diminutive lesion with advanced neoplasia
The risk of cancer in diminutive polyps is very low and estimated to be roughly 1 in 3000 [26 ]. However, this estimate is not evenly distributed across all populations undergoing colonoscopy. In a pooled review consisting of five FIT cohorts and seven colonoscopy cohorts, a greater prevalence of advanced histology was seen in the FIT cohorts compared with the colonoscopy cohorts across all histological subtypes, although for CRC and villous histology the differences were not statistically significant [4 ]. In the FIT and colonoscopy cohorts, the pooled prevalence of advanced histology within diminutive polyps was 7.1 % and 1.5 % (P = 0.04). The pooled prevalence of CRC was 0.08 % in the FIT cohorts and 0.01 % in the colonoscopy cohorts (P = 0.37). This suggests that there may be a difference in the rates of advanced histology in diminutive polyps depending on the indication for colonoscopy.
Both the progression of untreated diminutive adenomas and the incidence of newly developed lesions should be considered when evaluating the incidence of advanced neoplasia (i. e. the risk of leaving in situ a diminutive lesion with advanced neoplasia). Longitudinal follow-up data on individuals with untreated diminutive adenomas are scarce. However, in the Japanese series of Sekiguchi et al. [27 ], the 5-year cumulative incidence of advanced neoplasia in individuals with untreated diminutive adenomas diagnosed by magnification was 1.4 % (95 %CI 0.5 %–3.4 %). Of the 508 untreated diminutive adenomas, none of the lesions progressed to advanced neoplasia during the follow-up period, all of the detected advanced neoplastic lesions (n = 21) were newly diagnosed in a different location from that in which the untreated diminutive adenoma was originally found.
A systematic review in 2017, including three studies with 327 patients, also showed that untreated diminutive adenomas have an indolent and benign course [7 ]. Only 0.6 % of all adenomas (2/340) developed into advanced adenomas in 2 to 3 years. Another Japanese study, that of Ninomiya, confirmed these low numbers when using magnification [28 ]. In their study with 706 patients with diminutive polyps on initial colonoscopy (excluding depressed lesions and Kudo V pit pattern), only two T1 cancers were detected on surveillance colonoscopy, and both were treated radically by endoscopic resection.
Data show that polyps are not equal in terms of their cancer risk. In particular, if there is a depressed area in the polyp, this substantially increases the risk that there might be a malignant component. Given the negligible risk that diminutive polyps harbor cancer, few data are available on this matter. Recently Oka et al. [29 ] presented a large study, in which the risk of submucosal invasion of diminutive adenomas was 0.19% (15/7801). Of these 15 lesions, 11 had a depressed area and only 0.01% of non-depressed diminutive adenomas (1/7687) had deep submucosal invasion. Therefore, if depressed adenomas were excluded from the leave-in-situ strategy, the risk of leaving in situ a cancerous lesion would be greatly reduced. Because of the higher prevalence of advanced histology that has been described, the panel encourages increased detection of non-polypoid and depressed lesions. These lesions should be resected and referred to the pathologist.
Pathological accuracy and reproducibility
Despite histology being the reference standard for differentiating between polyp subtypes, it is also hampered by some degree of misdiagnosis owing to error in sampling or retrieval, or pathology diagnosis [30 ]
[31 ]
[32 ]. Therefore, a 10 % error rate in the pathological discrimination between polyp subtypes may be assumed unless an enhanced reference standard is used, such as specialist review from multiple pathologists combined with cutting all available tissue. When replacing a pathological with an endoscopic diagnosis, it is preferable that this error rate should not be increased; most studies however have not used an enhanced reference standard.
Furthermore, as a starting point, most studies assume that every polyp is retrieved and available for histological analysis. However, even at expert centers, up to 10 % of polyps may be lost or destroyed after resection. The standard approach to lost or destroyed polyps is to assume that these are adenomas; however, reanalysis from the original DISCARD study suggests that, when taking this issue into account, the lower accuracy of optical diagnosis (versus a non-enhanced pathological reference standard) is balanced by the loss of polyps available for pathological diagnosis that have to be (over)called as adenomas. Therefore, in clinical practice, optical diagnosis and pathology may have similar accuracy on a “per polyp detected” basis [33 ].
Attitudes of endoscopists and patients with regards to a leave-in-situ strategy
Willems et al. performed an international survey among 808 endoscopists from nine endoscopy societies to evaluate their attitudes and practices with regards to a leave-in-situ strategy [34 ] In total, 63 % of the participants partly or completely agreed that diminutive polyps could be left unresected until the next screening colonoscopy because of the low associated cancer risk. Endoscopists were evenly split on the effects of leaving such polyps unresected, with about 50 % thinking that leaving diminutive polyps in place would increase the cancer risk of patients. Moreover, 52 % of endoscopists were already leaving diminutive polyps that appeared non-neoplastic in situ in their daily practice. These results are somewhat different from a survey by Gellad et al. in 2013, which reported that the majority of endoscopists would be somewhat agreeable to leaving diminutive polyps in place if guidelines were to support this practice [35 ].
Little is known regarding whether patients would accept a leave-in-situ strategy. The only evidence comes from von Renteln et al. who performed a study among 557 patients to investigate the valuable question of whether patients would find a leave-in-situ practice acceptable [36 ]. They found that approximately 50 % of individuals undergoing a routine colonoscopy would be agreeable to deferring resection of diminutive polyps until the next surveillance colonoscopy and participating in a trial to evaluate this approach.
Resect-and-discard strategy
2 In order to implement the resect-and-discard strategy for colorectal lesions of 1–5 mm, it is clinically acceptable if, during real-time colonoscopy, at least 80 % sensitivity and 80 % specificity is achieved for high confidence endoscopic characterization of colorectal neoplasia of 1–5 mm. Histopathology is used as the gold standard.
Level of agreement 100 % (17 Strongly agree; 5 Agree; 0 Neither agree nor disagree; 0 Disagree; 0 Strongly disagree).
The panel took into account a number of considerations in the development of this SODA competence standard for implementation of the resect-and-discard strategy, which are detailed in the following sections.
Potential impact on short-term consequences
When applying the resect-and-discard part of the optical diagnosis strategy, not recognizing neoplastic diminutive lesions could result in longer than appropriate surveillance intervals, whereas not recognizing non-neoplastic diminutive lesions could result in shorter than appropriate surveillance intervals. Again, it could be argued that the correct diagnosis of each polyp subtype is not of equal clinical importance. The clinical implications of incorrectly diagnosing neoplastic lesions (i. e. a longer surveillance interval) and incorrectly diagnosing non-neoplastic lesions (i. e. a shorter surveillance interval) are completely different. Therefore, competence standards should take both scenarios into account.
The aforementioned simulation study also determined the relationship between the proportion of correctly optically diagnosed diminutive polyps (adenomas and HPPs) and the surveillance interval agreement when implementing the resect-and-discard strategy in a primary colonoscopy and FIT-positive screening cohort ([Table 3 ]; Tables 2 s and 3 s ). Results in the tables are presented per 1000 individuals with at least one diminutive polyp. The impact on surveillance interval agreement is indicated by showing the proportion in agreement with the US Multi-Society Task Force on Colorectal Cancer surveillance guideline [37 ] and the proportion in agreement with the ESGE surveillance guideline [38 ]. For example, if 80 % of all diminutive adenomas and 80 % of all diminutive HPPs were to be diagnosed correctly with optical diagnosis in the primary colonoscopy screening program (i. e. 78 % of all diminutive polyps correctly diagnosed), the proposed surveillance intervals would agree with those determined by histological analysis for 80 % of the cases when using the US guideline and for 95 % of the cases when using the ESGE guideline.
Table 3
Impact of implementing the resect and-discard strategy with different proportions of correctly optically diagnosed ≤5-mm adenomas and hyperplastic polyps (HPPs) in a primary colonoscopy screening and fecal immunochemical test (FIT)-positive cohort on the overall proportion of correctly diagnosed ≤5-mm polyps and surveillance interval agreement with histopathology as the gold standard.
Strategy[* ]
Colonoscopy screening individuals
FIT-positive screening individuals
Correctly diagnosed ≤ 5-mm adenomas, %
Correctly diagnosed ≤ 5-mm HPPs, %
Correctly diagnosed ≤ 5-mm polyps, %
Surveillance interval agreement, %
Correctly diagnosed ≤ 5-mm polyps, %
Surveillance interval agreement, %
ESGE
US
ESGE
US
60
60
60
95
66
60
97
84
60
80
68
95
73
64
97
86
60
100
78
95
80
69
97
88
70
60
63
95
69
67
98
87
70
80
73
95
76
72
98
89
70
100
83
95
84
77
98
91
80
60
68
95
73
74
98
90
80
80
78
95
80
79
98
91
80
100
87
95
88
84
98
93
90
60
73
95
76
82
99
92
90
80
82
96
83
87
99
94
90
100
92
96
92
91
99
96
100
60
77
96
79
89
99
94
100
80
87
96
87
94
100
97
100
100
96
96
96
99
100
99
ESGE, European Society of Gastrointestinal Endoscopy (ESGE) post-polypectomy surveillance guidelines; US, US Multi-Society Task Force on Colorectal Cancer guidelines.
* At each given proportion of correctly diagnosed diminutive adenomas and HPPs, we assumed that 40 % of all diminutive sessile serrated lesions would be diagnosed correctly with high confidence to ensure that the clinical consequences are a reflection of daily practice.
[Fig. 1 ] shows the impact of implementing the resect-and-discard strategy for colorectal lesions of 1–5 mm with 80 % sensitivity and 80 % specificity for high confidence endoscopic characterization of 1–5 mm colorectal neoplasia on surveillance interval agreements. Different thresholds for the proportion of correctly diagnosed diminutive polyps again lead to different proportions in agreement with the surveillance guidelines depending on the guidelines and the clinical setting.
Fig. 1 Impact of implementing the resect-and-discard strategy for colorectal lesions of 1–5 mm with 80 % sensitivity and 80 % specificity for high confidence endoscopic characterization of 1–5 mm colorectal neoplasia on surveillance interval agreement using the ESGE surveillance guideline [38 ] and US Multi-Society Task Force on Colorectal Cancer surveillance guideline [37 ].
Potential impact on long-term consequences
Data on the long-term clinical consequences and cost-effectiveness of implementing the resect-and-discard strategy are limited to four modelling studies ([Table 2 ]). Two of these evaluated the implementation of the complete optical diagnosis strategy and demonstrated that implementation only marginally influenced long-term outcomes, such as CRC incidence and mortality, whilst saving costs [1 ]
[9 ]
[33 ]. Two studies exclusively evaluated the implementation of the resect-and-discard strategy [8 ]
[25 ]. In the modelling study by Kessler et al. [25 ], the estimated cost-savings of implementing the resect-and-discard strategy in a primary colonoscopy setting was $174 per individual, with a number needed to harm because of missed interval cancer of 7979. In the modelling study by Hassan et al., adoption of the resect-and-discard strategy in a primary colonoscopy screening program resulted in a saving of $25 per individual, without any meaningful effect on screening efficacy [8 ]. In summary, the modelling studies described above have provided evidence that implementation of the optical diagnosis strategy is associated with substantial cost-savings with negligible impact on patients’ cancer risk.
Data on the impact of varying the rate of high confidence diagnosis on long-term clinical consequences are limited to two modelling studies. These studies showed that varying the proportion of high confidence diagnoses did not increase the CRC burden [8 ]
[9 ]. They demonstrate that the cost-effectiveness of the optical diagnosis strategy appears to be dependent on the proportion of high confidence diagnoses [39 ]. Hassan et al. showed a linear relationship between the rate of high confidence prediction and the undiscounted savings projected in the US population, when implementing the optical diagnosis strategy in a colonoscopy screening program [8 ]. Assuming a 100 % (best-case scenario) and 50 % (worst-case scenario) proportion of high confidence diagnoses, the undiscounted benefit for the US population would be $40 million and $20 million, respectively. In another simulation study, varying the proportion of high confidence diagnoses from 50 % to 100 % led to cost-savings of €5 and €7, respectively, per individual in a biennial FIT screening program [9 ].
Risk of metastatic disease after resecting and discarding a diminutive lesion with cancer
The risk of metastatic disease after resecting and discarding a diminutive polyp in situ that contained cancer is very limited. In the study of Oka et al. [29 ], for seven diminutive invasive polyp cancers that were treated surgically, no lymph node metastases were seen.
Attitudes of endoscopists and patients with regards to a resect-and-discard strategy
When Willems et al. asked endoscopists in their worldwide survey about barriers to implementation of a resect-and-discard strategy, it appears the clinical uptake of resect-and-discard is low [34 ]. They found that only 15.8 % of the endoscopists used the resect-and-discard strategy in their current practice and 59.9 % thought that implementation of the resect-and-discard strategy was not feasible in its current form. Of all endoscopists, 44.6 % were afraid of making a wrong diagnosis, 53.8 % were concerned about potential medicolegal issues, and 58.3 % were afraid of assigning incorrect surveillance intervals to patients. These findings are similar to those of Soudagar et al. in 2016, where medicolegal concerns were the main barrier to implementation of the resect-and-discard strategy for the 105 gastroenterologists surveyed during a national conference in the USA [40 ]. While the consensus for most regions was that resect-and-discard was not feasible, 54 % of the European endoscopists showed an increased adoption of the strategy.
Rex et al. surveyed American colonoscopy patients and found that 66 % of them would accept a resect-and-discard strategy [26 ]. Of those unwilling to accept resect-and-discard, 50 % wanted an absolute zero chance of cancer in diminutive polyps and were willing to pay out of their own pocket for histological assessment of these diminutive polyps. Vu et al. further assessed this by approaching patients with the hypothetical question of whether they would be willing to pay approximately $150 for pathology rather than use the resect-and-discard strategy if the risk of a cancer in their polyp was 1 : 3000 [41 ]. Over two-thirds of patients would be willing to pay to have their diminutive polyp sent for pathological evaluation.
