If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Dysplasia within sessile serrated polyps (SSPs) is difficult to detect and may be mistaken for an adenoma, risking incomplete resection of the background serrated tissue, and is strongly implicated in interval cancer after colonoscopy. The use of endoscopic imaging to detect dysplasia within SSPs has not been systematically studied.
Consecutively detected SSPs ≥8 mm in size were evaluated by using a standardized imaging protocol at a tertiary-care endoscopy center over 3 years. Lesions suspected as SSPs were analyzed with high-definition white light then narrow-band imaging. A demarcated area with a neoplastic pit pattern (Kudo type III/IV, NICE type II) was sought among the serrated tissue. If this was detected, the lesion was labeled dysplastic (sessile serrated polyp with dysplasia); if not, it was labeled non-dysplastic (sessile serrated polyp without dysplasia). Histopathology was reviewed by 2 blinded specialist GI pathologists.
A total of 141 SSPs were assessed in 83 patients. Median lesion size was 15.0 mm (interquartile range 10-20), and 54.6% were in the right side of the colon. Endoscopic evidence of dysplasia was detected in 36 of 141 (25.5%) SSPs; of these, 5 of 36 (13.9%) lacked dysplasia at histopathology. Two of 105 (1.9%) endoscopically designated non-dysplastic SSPs had dysplasia at histopathology. Endoscopic imaging, therefore, had an accuracy of 95.0% (95% confidence interval [CI], 90.1%-97.6%) and a negative predictive value of 98.1% (95% CI, 92.6%-99.7%) for detection of dysplasia within SSPs.
Dysplasia within SSPs can be detected accurately by using a simple, broadly applicable endoscopic imaging protocol that allows complete resection. Independent validation of this protocol and its dissemination to the wider endoscopic community may have a significant impact on rates of interval cancer. (Clinical trial registration number: NCT03100552.)
Sessile serrated polyps (SSPs) without dysplasia (SSP-ND) account for about 8% of colorectal polyps. They may take many years to develop dysplasia and can be managed effectively by complete endoscopic resection. SSPs with dysplasia (SSP-D), conversely, are high-risk lesions. Recent evidence suggests that they may progress rapidly to colorectal cancer (CRC), particularly because they occur in people aged similar to people with serrated pathway CRC. The dysplastic areas on the surface of SSP-D may mimic conventional adenomas, risking incomplete resection by an endoscopist who does not appreciate the background serrated tissue. Incomplete endoscopic resection of SSP-D leaves a high-risk lesion in situ with the molecular changes to rapidly progress to interval cancer. A reliable and accurate method to characterize dysplasia within SSPs by using endoscopic imaging has not been demonstrated previously in a large series.
New study findings
In 141 SSPs, a standardized endoscopic imaging protocol for dysplasia was accurate in 95.0% (95% confidence interval [CI], 90.1%-97.6%), with a negative predictive value of 98.1% (95% CI, 92.6%-99.7%) against the criterion standard of blinded GI specialist pathologists. The protocol is simple to use and broadly applicable to any endoscopist working in a unit with high-definition endoscopes. If independently validated, and with widespread dissemination to the endoscopic community, this protocol may help reduce rates of interval cancer via reliable identification and complete resection of SSP-D.
The Nature of SSPs
SSPs account for up to 8% of colorectal polyps in a screening population.
They are more common than adenomas, potentially because of a lack of appreciation of the boundaries of the lesion. Recent evidence suggests that they have a long dwell time (in the order of 17 years) before the development of dysplasia. In support of this, SSP-D are found in significantly older patients, with the age of patients with SSP-D closely matching the age of individuals with serrated pathway CRC. This provides strong evidence for SSP-D as a high-risk lesion with the potential for rapid progression to cancer.
and can be managed effectively by a meticulous withdrawal technique for endoscopic detection, accurate lesion characterization, delineation of the lesion boundaries, and complete endoscopic resection. SSP-D presents a unique challenge. It is established that adenomas are more easily detected than SSPs,
predominantly because of their more conspicuous surface structure. SSP-D has the potential to masquerade as adenomas. Therefore if only the dysplastic portion is appreciated and resected, then a potentially rapidly progressive residual is left in situ, and CRC may ensue. In this context, the ability to accurately recognize and characterize SSP-D to achieve complete endoscopic resection is a critical component of interventions to prevent serrated pathway CRC and potentially interval cancer.
have described the feasibility of using endoscopic imaging to detect dysplasia within SSPs, but no systematic study exists. In this study, we prospectively evaluated the utility of a simple, widely applicable endoscopic imaging protocol to detect dysplasia within SSPs compared with histopathology.
Format of the study
Over 41 months to January 2017, sequential SSPs ≥8 mm detected within the endoscopic resection practice of a single tertiary-care referral center were analyzed according to a standardized imaging protocol. Detailed patient, procedural, and lesion characteristics were recorded. Split-dose bowel preparation was used. Olympus high-definition colonoscopes were used throughout (Olympus 180/190 series, Olympus, Tokyo, Japan). Lesion size was determined with reference to an open snare of known diameter, as was any dysplastic area if present. The right side of the colon was defined as proximal to, and including, the hepatic flexure. For the purposes of this study, SSP-D describes the endoscopic determination of dysplasia within SSPs, and SSP-ND describes the endoscopic determination of no dysplasia within SSPs. Sessile serrated adenoma with dysplasia (SSA-D) and sessile serrated adenoma without dysplasia (SSA-ND) were used to describe the histologic determination of dysplasia within sessile serrated adenomas (SSA).
All procedures were performed by endoscopists with extensive experience in endoscopic resection or by an advanced endoscopy fellow under their direct supervision. There were 8 endoscopists in total. All patients provided written informed consent. Institutional review board approval was obtained. All authors had access to the study data and reviewed and approved the final manuscript.
Standardized imaging protocol
We created a standardized imaging protocol supported by our description of endoscopic features of dysplasia within SSPs
Lesions were considered as SSPs when 1 or more of the following features were identified: mucous cap, cloud-like or finely nodular surface, indistinct border, irregular shape, and lesion paler than the surrounding mucosa (under narrow-band imaging)
(Fig. 1). Once a lesion suspected to be an SSP was detected, the endoscopist completed a structured datasheet with the study coordinator. A detailed analysis of the surface of the lesion was made by using high-resolution endoscopic imaging with high-definition white-light (HDWL) and narrow-band imaging (NBI) (Olympus). In keeping with our previous experience,
magnification was not used nor was chromoendoscopy. A careful search was made for any demarcated area with an adenomatous (Kudo III/IV, NICE II) pit pattern among the serrated tissue. If detected, the lesion was described as SSP-D, and the features of this area were described in detail. The size and number of demarcated areas was recorded. Within demarcated areas, hyper-vascularity (in comparison to the background features of the overall lesion under NBI) and Kudo pattern were described (Fig. 2, Video 1, available online at www.giejournal.org). If a demarcated area was not detected, the lesion was labeled SSP-ND. The endoscopist was then asked to provide confidence that (1) the lesion represented an SSP and (2) the lesion did or did not contain dysplasia.
All lesions were resected endoscopically. Our common practice is to perform en bloc cold snare polypectomy for SSPs up to 10 mm. Larger lesions predicted as SSP-ND are resected by using piecemeal cold snare polypectomy or EMR. If there is evidence of dysplasia (SSP-D), we prefer EMR, resect the suspected dysplastic area en bloc, and submit it separately for histopathologic analysis.
Resected lesions were diagnosed as SSAs based on established criteria.
Hematoxylin-eosin–stained sections were available and were retrieved in all cases. All specimens were reviewed by 2 independent specialist GI pathologists blinded to the endoscopic diagnosis. The type of adenoma and the presence and grade of any cytologic dysplasia were recorded. Serrated dysplasia was not considered cytologic dysplasia as per current expert opinion.
For the initial analysis, histopathology was treated as the criterion standard.
For cases in which the histopathologic and endoscopic determination of dysplasia within an SSP were incongruent, a structured discussion between endoscopists and specialist pathologists was undertaken. A determination of the most likely correct diagnosis (SSA-ND or SSA-D) was decided by mutual agreement.
Statistical analysis was done by using SPSS version 23 (IBM, Armonk, NY) with a 2-tailed t test used for parametric continuous variables, Mann-Whitney U test for nonparametric continuous data, and the chi-square test for categorical variables. Only the largest demarcated area within an SSP-D was described if there was more than 1. Agreement between endoscopy and histopathologic analysis was determined by using 2 × 2 tables. For these analyses, histopathology was regarded as the criterion standard, before and after the structured discussion between specialist pathologists and endoscopists. The significance of P was regarded at <.05.
Over the study period, 144 SSPs were detected in 86 patients. Three SSPs in 3 patients were excluded (Supplementary Fig. 1, available online at www.giejournal.org). The mean age of patients was 64.0 years (standard deviation 15.4) and 57 of 83 (68.7%) patients were female. Forty-four of 83 (53.0%) patients had 1 SSP, 24 had 2 SSPs, 12 had 3 SSPs, 2 had 4 SSPs, and 1 had 5 SSPs included in the analysis.
Endoscopic features of SSPs
The median size of SSPs was 15 mm (interquartile range [IQR] 10-20). SSPs (1) were located in the right side of the colon in 77 of 141 cases (54.6%), (2) were most often Paris classification 0-IIa (119/141, 84.4%), (3) had a mucous cap (115/141, 81.6%), (4) had a finely nodular surface (84/141, 59.6%), had an indistinct border (95/141, 67.4%), and (5) were paler than the surrounding mucosa in the majority of cases (128/141, 90.8%). Endoscopic prediction that a lesion was an SSP was accurate in 141 of 141 cases (100%) compared with histopathology.
One hundred five of 141 lesions were predicted as SSP-ND by using the standardized imaging protocol, and 36 of 141 (25.5%) were predicted as SSP-D. SSP-ND were smaller than SSP-D (15 mm, IQR 10-20 vs 20 mm, IQR 15-30; P < .001) and were more likely to be paler than the surrounding mucosa under NBI compared with SSP-D (100/105, 95.2% vs 28/36, 77.8%; P = .004). SSP-D were more likely to contain a dominant Is component than SSP-ND (3/36, 8.3% vs 0/105, 0%; P = .027). Other endoscopic features of SSPs were similar between the groups (Table 1).
Table 1Characteristics of the lesions in the study including methods of resection split by endoscopic features of dysplasia by using the standardized imaging protocol
Of the 36 of 141 (25.5%) SSPs predicted as SSP-D by using the standardized imaging protocol, 27 of 36 (75%) were predicted with high confidence (Supplementary Table 1, available online at www.giejournal.org). Larger demarcated areas within SSPs were more likely to be predicted with high confidence (median 4 mm, IQR 2-5 vs 2 mm, IQR 1.5-3; P = .010). Features of the demarcated area were described as hyper-vascular in 31 of 36 (86.1%), Kudo III pit pattern in 24 of 36 (66.7%), and Kudo IV pattern in the remaining 12 of 36 (33.3%).
After initial histopathologic analysis, the standardized imaging protocol had a sensitivity of 92.9% (95% CI, 75.0%-98.7%), a specificity of 91.2% (95% CI, 83.9%-95.4%), and a negative predictive value of 98.1% (95% CI, 92.6%-99.7%). The accuracy was 91.4% (95% CI, 85.7%-95.1%).
Cases with disagreement between histologic and endoscopic determination of dysplasia within SSP
In 12 of 141 cases (8.5%), there was disagreement between endoscopic and histopathologic determination of dysplasia within specific SSPs. In each case, endoscopic and histologic images of the procedure were retrieved, and a structured discussion between endoscopists and the specialist histopathologists was undertaken (Fig. 3).
In 5 of 12 cases (41.7%), a consensus was reached that the endoscopic imaging was correct. Causes for histopathologic error were not errors in retrieval and/or processing of the specimen in 4 of 5 cases (80.0%), and a small area of dysplasia missed between tissue slices in 1 case of 5 (20.0%).
In 7 of 12 cases (58.3%), a consensus was reached that the histopathologic assessment was correct. Causes for endoscopic error were a demarcation area created by an alternative pathology in 3 of 7 cases (42.9%) (traditional serrated adenoma in 1 of 3 and a perineurioma in 2 of 3) (Fig. 4), colon morphology in 2 of 7 cases (28.6%), a borderline type II-O/III Kudo pit pattern in 1 of 7 cases (14.3%), and multifocal dysplasia with no transition point in 1 of 7 cases (14.3%).
After the results of consensus opinion were taken into consideration, the standardized endoscopic imaging protocol demonstrated a sensitivity of 93.9% (95% CI, 78.4%-98.9%), specificity of 95.4% (95% CI, 89.0%-98.3%), negative predictive value of 98.1% (95% CI, 92.6%-99.7%), and an accuracy of 95.0% (95% CI, 90.1%-97.6%) for the prediction of dysplasia within SSPs (Table 2).
Table 2Accuracy of the endoscopic assessment of dysplasia within serrated lesions ≥8 mm before and after discussion with the specialist histopathologists
Accuracy of endoscopic assessment of dysplasia (before discussion)
92.9% (95 CI, 75.0%-98.7%)
91.2% (95% CI, 83.9%-95.4%)
Negative predictive value
98.1% (95% CI, 92.6%-99.7%)
91.4% (95% CI, 85.7%-95.1%)
Accuracy of endoscopic assessment of dysplasia (after discussion)
93.9% (95% CI, 78.4%-98.9%)
95.4% (95% CI, 89.0%-98.3%)
Negative predictive value
98.1% (95% CI, 92.6%-99.7%)
95.0% (95% CI, 90.1%-97.6%)
SSA-ND, Sessile serrated adenoma without dysplasia as determined histologically; SSA-D, sessile serrated adenoma with dysplasia as determined histologically or by structured discussion; SSP-ND, sessile serrated polyp without dysplasia as determined endoscopically; SSP-D, sessile serrated polyp with dysplasia as determined endoscopically; CI, confidence interval.
En bloc cold snare polypectomy was the most common form of resection for SSPs (56/141, 39.7%). Piecemeal cold snare polypectomy (18/141, 12.8%) was principally used for large, nondysplastic SSPs. EMR was preferred for SSP-D (20/28, 71.4% vs 47/113, 41.6%; P = .005). Endoscopic resection was judged to be complete in all cases.
Recent data show that SSA-ND has a long dwell time before the development of dysplasia and progression to CRC.
has enabled their reliable detection, and they can be managed effectively by complete endoscopic resection. Conversely, SSA-D are high-risk lesions. Once they have acquired specific molecular changes (commonly hyper-methylation of MLH1
Although this may be of little consequence for SSP-ND, incomplete resection of SSP-D may be responsible for a significant proportion of interval CRC, perhaps primarily because of a failure to appreciate the nature or extent of the lesion. For example, SSP-D may be mistaken for adenomas with the endoscopist not appreciating the surrounding serrated component, which may then be left in situ. In this study we have demonstrated that a simple, broadly applicable, endoscopic imaging protocol can accurately determine the presence of dysplasia within SSPs in a large prospective cohort. This protocol therefore allows the suitably trained endoscopist to reliably recognize SSP-D, precisely delineate its extent, and ensure its complete endoscopic resection.
Contemporary colonoscopists with high adenoma detection rates report serrated lesion detection rates of approximately 8%,
suggests an accuracy of 0.93 (95% CI, 0.87-0.98) in discriminating SSPs from non-SSPs after training. Among the cohort of tertiary-care center endoscopists in this study, SSPs were correctly identified in all cases. We confirmed that the features of SSPs in the WASP classification
were common and can be used readily to identify SSPs, particularly a clouded or finely nodular surface, indistinct border, and a lesion paler than the surrounding mucosa by using NBI.
Modern endoscopic imaging allows reliable visualization to the level of the individual colon mucosal gland, permitting differentiation of so-called neoplastic pit patterns from non-neoplastic patterns.
conducted retrospectively among 41 SSP-D derived a set of endoscopic lesion characteristics (reddishness, double elevation, and central depression), which predicted dysplasia or serrated cancer. The inclusion of serrated cancer, use of still images, and derivation of highly subjective predictors limits the widespread applicability of this method of determining SSP-D.
We used features derived from our previous studies
in addition to our anecdotal observations to develop a standardized imaging protocol to determine dysplasia within SSP. Dysplasia commonly appears as a transition from type II open pits typical of SSP to type III or type IV pits typical of tubular or tubulovillous adenoma. The transition stands out to the observer and creates a demarcated area, particularly under NBI. This area often is hypervascular or dark (under NBI) compared with the surrounding serrated tissue. When the imaging protocol was applied prospectively to the 141 lesions in this study, the accuracy of determining dysplasia was very high (95%) as was the negative predictive value (98%).
Incomplete resection of a rapidly progressive high-risk lesion such as SSP-D is likely responsible for a proportion of interval cancers after colonoscopy. In this scenario, the endoscopist, distracted by the adenomatous pit pattern of an SSP-D, resects only the dysplastic portion, leaving part of a lesion with the mutations, allowing rapid progression to CRC in situ. The ability to determine accurately, by using endoscopic imaging, whether a given colon lesion is an SSP-D allows real-time appreciation of its high-risk nature as well as the full extent of the lesion, guarding against incomplete resection. It also allows selection of a safe method of endoscopic resection. Dysplastic lesions should be resected by established techniques, that is, EMR. In particular, the dysplastic focus should be removed en bloc and submitted separately for histopathology to avoid missing a focus of submucosal invasive cancer. At our center, we routinely perform piecemeal cold snare polypectomy for large SSPs without endoscopic evidence of dysplasia, aiming to avoid the risks of electrocautery, especially delayed bleeding from the right side of the colon. Prospective randomized studies (compared with EMR) are required to determine the safety and utility of this technique. The 5 patients who underwent piecemeal cold snare polypectomy for SSP-D in this study were elderly, and therefore avoidance of electrocautery risk was prioritized.
Another potential benefit of an accurate endoscopic imaging protocol to detect and characterize SSP-D concerns patients with multiple SSPs (including those meeting criteria for serrated polyposis syndrome), requiring several colonoscopic sessions to clear the colon. The majority of these lesions may take many years to develop the molecular changes required to become CRC. Endoscopic determination of dysplasia allows the endoscopist to prioritize high-risk lesions for resection.
Real-time analysis notwithstanding, histopathology remains an imperfect criterion standard in the study of dysplastic areas within SSPs. Dysplastic foci often are small (median 3 mm diameter in this study). These small foci may not be retrieved or may be destroyed during tissue processing. They may lie undiscovered between tissue sections, which can be as wide as 3 to 5 mm.
The majority of endoscopic errors in determining SSP-D were due to an alternative lesion mimicking the appearance of a demarcated area or the colon contour hiding the presence of a nodule. Overall, 10 of 12 cases (83.3%) with a discrepancy between endoscopic and histologic findings were endoscopically predicted as SSP-D, whereas histology determined SSA-ND. Although 5 of 10 such cases (50%) were later determined to be histologic error, the clinical relevance of these discrepancies should be played down because it is the ability to endoscopically determine a lack of dysplasia that matters clinically.
Given the accuracy of the standardized imaging protocol in predicting SSP-D among a group of expert endoscopists, independent validation of these findings is necessary before dissemination of learning materials to the wider endoscopic community. Complementary efforts should include dissemination of validated criteria to accurately identify SSPs. The support of international societies will be vital in this endeavor.
The strengths of this study include a large, previously unstudied cohort of patients with SSP, which were systematically described and prospectively collected. Involvement of multiple endoscopists and review of all specimens by 2 blinded specialist histopathologists minimized bias in analysis of the protocol and the histopathologic analysis. The standardized imaging protocol is simple and can be generalized widely because the majority of endoscopy units will have the equipment required to implement it.
The drawbacks of the study include the collection of cases from a single center. In addition, SSPs were detected within a tertiary-care resection cohort, potentially biasing the type of lesion seen in the everyday practice of a screening endoscopist. In particular, the mean size of SSPs in a screening populations is about 10 mm, according to a recent study,
as compared with the median size in this study of 15 mm. The experience described here with larger SSPs may be a stepping stone to understanding the problem and elucidating the imaging signs of dysplasia in smaller lesions. In addition, not only is SSP-D rare in screening populations (0.6% of all polyps in 1 study
), but if an SSP is small, the endoscopist is more likely to appreciate the lesion as a whole rather than be drawn to the dysplastic focus as such.
In conclusion, we have demonstrated that dysplasia within SSPs can be predicted accurately by using a standardized endoscopic imaging protocol. This simple protocol allows the appropriately trained endoscopist to characterize SSP-D fully–particularly the extent of the lesion–with a view to achieving complete resection. Given that SSP-Ds are strongly implicated in interval cancers after colonoscopy, independent validation of this protocol and its dissemination to the wider endoscopic community may have profound clinical consequences.
Supplementary Table 1Characteristics of dysplastic demarcated areas within 36 sessile serrated polyps with dysplasia detected within the study period, split by the endoscopist’s confidence in the presence of dysplasia
Total n = 36
Low confidence n = 9
High confidence n = 27
Endoscopic features of demarcated area
Size (median, IQR), mm
Hypervascular, no. (%)
Highest Kudo, no. (%)
SSA-D before discussion, no. (%)
SSA-D after discussion, no. (%)
IQR, Interquartile range; SSA-D, sessile serrated adenoma with dysplasia as determined histologically Before and after discussion refer to the discussion of cases between the endoscopist and the specialist pathologists.
DISCLOSURE: The Cancer Institute New South Wales provided funding for a research nurse and data manager to assist with the administration of the study. There was no influence from the institution regarding study design or conduct, data collection, management, analysis, or interpretation or preparation, review, or approval of the manuscript. All other authors disclosed no financial relationships relevant to this publication.
If you would like to chat with an author of this article, you may contact Professor Bourke at [email protected] .