Discriminating endoscopic features of sessile serrated lesions

To the Editor: Colorectal cancer (CRC) remains a leading cause of cancer-related death worldwide.[1] In addition to the classical adenoma-to-carcinoma model of colorectal carcinogenesis, the serrated neoplastic pathway accounts for 20%–30% of CRCs. Sessile serrated lesions (SSLs), the most common premalignant serrated polyp subtype, share many of the genomic characteristics of post-colonoscopy CRCs and are thus recognized as their precursor lesions. However, SSLs can be easily missed due to their flat shape and subtle endoscopic features. Sometimes they can also be difficult to distinguish from other types of polyps, especially hyperplastic polyps (HPs), which are believed to have no malignant potential.[2] Approximately 10% of polyps initially diagnosed as HPs endoscopically are reclassified as SSLs after histopathological examination, reflecting diagnostic difficulties in colonoscopy.[3,4] Current guidelines recommend "diagnose-and-leave" and "resect-and-discard" strategies for diminutive polyps, which might lead to SSLs being left in situ or diagnoses being missed,[5] raising the risk of recurrence and post-colonoscopy cancers. Furthermore, it remains debatable in the gastrointestinal community whether the risks of resecting SSLs outweigh the reduction in cancer risk. Any strategy that improves the diagnostic accuracy of SSLs could have a meaningful clinical impact. Recent studies have used white light endoscopy (WLE) and image-enhanced endoscopy (IEE) approaches such as narrow-band imaging (NBI) and chromoendoscopy (CE) to identify the specific characteristics of SSL. However, the results have been inconsistent. Thus, existing guidelines still lack definitive colonoscopic diagnostic features of SSL, and their accurate diagnosis and differentiation from HP and adenoma by endoscopy remain challenging. We therefore conducted a systematic review and meta-analysis to evaluate the endoscopic characteristics of SSL and to determine features that might help distinguish SSLs from HPs and adenomas in patients receiving CRC screening. The review registration was PROSPERO (CRD42018099945, 27/06/2018; updated on 04/05/2021). MEDLINE, Embase, and Cochrane Library databases were searched from their inception to December 2022 to identify potentially eligible studies, the detailed search strategy was provided in Supplementary Materials [https://links.lww.com/CM9/B519]. Cohort, cross-sectional, or case–control studies that compared the endoscopic findings of SSLs and other polyps based on histological examination were eligible. The full texts of potentially eligible articles were retrieved and reviewed for final eligibility as cohort, cross-sectional, or case–control studies comparing the endoscopic and histopathological assessments of SSLs and HPs/conventional adenomas. The Newcastle–Ottawa Scale (NOS) was used to assess the risk of bias of the included studies in terms of patient selection, controlling for confounders, and ascertainment of exposures. Data were extracted to evaluate associations between endoscopic characteristics and the diagnosis of SSLs from all included studies to conduct a random-effects meta-analysis. Reported odds ratios (ORs) and their corresponding standard errors (SEs) were used in the meta-analysis as the primary outcome. We delineated all investigated endoscopic features and present them in different groups according to the endoscopic modality. For each endoscopic feature, if two or more studies were included, data were extracted and pooled. The weight of the studies was estimated using the inverse variance method, and random variance between studies was estimated by using the DerSimonian–Laird method. Heterogeneity was evaluated by using the I2 method, and Egger's test was used to assess publication bias. Post hoc subgroup analysis was conducted based on the dysplasia status (with or without dysplasia). All meta-analysis were based on the number of lesions, where available. For endoscopic features included in ≥10 studies, random-effects meta-regression was used to evaluate the effects of lesion number, male proportion, average age, Eastern or Western countries, study type, adjustment for confounders, and average NOS for the combined association. The search identified 4230 articles. A total of 827 duplicates were removed, 3102 articles were excluded after title and abstract screening. The remaining 301 full-text articles were assessed, and 74 studies were finally included in the review. The 74 studies were conducted in 16 countries in North America, South America, Europe, Oceania, and Asia. WLE, NBI, CE, autofluorescence imaging (AFI), and endocytoscopy approaches were all reported. Among these studies, 44 studies used a cross-sectional design, 21 were cohort studies, and 9 were case-control studies. The median NOS score was 6 (range 5–9). The mean age of patients with SSL ranged from 41.9 years to 70.0 years (mean age 61.1 years). The proportion of male patients ranged from 31.2% to 97.0% (mean proportion 70.0%). Compared with HPs, SSLs were more commonly found in the right colon (proximal to the splenic flexure) (OR: 5.45, 95% confidence interval [CI]: 4.13–7.17) than in the left colon, and more common in the proximal colon (proximal to the sigmoid colon) (OR: 5.43, 95% CI: 4.26–6.93) than in the distal colon. SSLs tended to be larger than HPs (>5 mm vs. ≤5 mm; OR: 5.60, 95% CI: 3.82–8.22; >10 mm vs. ≤10 mm, OR: 5.93; 95% CI: 3.82–9.22). SSLs were more likely to have a Paris 0-II type than Paris 0-I morphology (OR: 2.33, 95% CI: 1.45–3.70) [Supplementary Figure 1, https://links.lww.com/CM9/B519]. Under WLE, SSLs were more likely to have mucus caps (OR: 8.48, 95% CI: 4.86–14.80), vague margins (OR: 2.71, 95% CI: 1.88–3.92), and cloud-like surfaces (OR: 8.19, 95% CI: 1.48–45.20) than HPs. There was no significant difference in the proportion of SSLs and HPs, with reddish color, irregular shape, or dark spots inside crypts [Supplementary Figure 1, https://links.lww.com/CM9/B519]. Under magnified NBI (ME-NBI), expanded crypt openings (ECOs), varicose microvascular vessels (VMVs), and thick branched vessels (TBVs) or dilated and branching vessels (DBVs) were more commonly seen with SSLs than with HPs (ORs: 5.09, 6.17, and 5.17, respectively; 95% CIs: 1.87–13.90, 1.57–24.30, and 1.81–15.80, respectively). The capillary pattern defined by Sano et al[6] did not distinguish SSLs from HPs (P = 0.22) [Supplementary Figure 1, https://links.lww.com/CM9/B519]. A type II pit pattern was less likely to be seen in SSLs than in HPs (OR: 0.22, 95% CI: 0.06–0.75), but a type II-O pit pattern was significantly more common in SSLs than in HPs (OR: 18.60, 95% CI: 8.45–40.70). A type II-L pit pattern was not a distinguishing feature of SSLs from HPs (P = 0.13) [Supplementary Figure 1, https://links.lww.com/CM9/B519]. Compared with conventional adenomas, SSLs were more likely to be proximal to the splenic flexure (OR: 2.35, 95% CI: 1.67–3.31), to be larger (>5 mm vs. ≤5 mm OR: 2.56, 95% CI: 1.54–4.35; >10 mm vs. ≤10 mm OR: 2.13, 95% CI: 1.14–4.00), and to have more Paris-0-II than Paris-0-I in morphology (OR: 3.33; 95% CI: 1.72–6.25). SSLs were also more likely to have mucus caps (OR: 26.20, 95% CI: 16.10–42.50) than conventional adenomas, although reddish color was not a distinguishing feature of either lesion (P = 0.67). Two studies compared SSLs with conventional adenomas and found that SSLs were more likely to have a type II-O pit pattern (OR: 540.40, 95% CI: 126.00–2315.00, P <0.01; I2 = 0; mean NOS 6.0). Thirteen studies investigated the endoscopic features of SSLs with and without cytological dysplasia. The proportions of SSLs with dysplasia larger than 5 mm and 10 mm were significantly higher than those of SSLs without dysplasia (>5 mm vs. ≤5 mm OR: 6.67, 95% CI: 3.23–14.3; >10 mm vs. ≤10 mm OR: 4.00, 95% CI: 2.00–8.33). SSLs with dysplasia were less likely to be Paris 0-II subtype compared with SSLs without dysplasia by WLE (OR: 0.38, 95% CI: 0.16–0.91). However, there were no differences in location, mucus cap, or reddish color between SSLs with or without dysplasia. Signs including a type II pit pattern and type II-O pit pattern were not distinguishing features of SSLs with and without dysplasia. Thus, our analysis confirmed that SSLs tended to be located more proximally than HPs or conventional adenomas and were more likely to be larger and flatter than HPs. We also identified some distinguishing features of SSLs under WLE and IEE. While only two out of six WLE signs were useful for identifying SSLs, three out of four ME-NBI signs and all three ME-CE signs were helpful for discriminating SSLs from HPs, suggesting enhanced potential of IEE for distinguishing SSLs from other lesions. Typical HPs measure less than 5 mm in diameter and are located in the distal colon, while typical SSLs are flat lesions with indistinct edges in the proximal colon. Our findings support 5 mm as a useful threshold for suspecting SSLs, rather than HPs. However, Anderson et al[7] found that a significant proportion of distal HPs were reclassified as SSLs when diagnostic histological features were present in at least three crypts, making location alone an unreliable predictor for histological diagnosis. Furthermore, the pale appearance and flat shape of SSLs justify a longer withdrawal time for detection, as observed in the Dutch screening cohort and the New Hampshire Registry study. Interestingly, the quality of bowel preparation did not seem to hamper detection of SSLs probably because the thicker mucus cap on SSLs may assist detection when preparation is unsatisfactory. Our findings are consistent with the recommendation that large proximal SSLs should be managed by individuals and centers with expertise, since their vague margin and flat morphology make endoscopic treatment more difficult and increase risk incomplete resection. It has been suggested that dysplastic SSLs have a greater potential for cancer progression, and limited studies have explored endoscopic features of SSL with dysplasia.[8] We found that dysplastic SSLs were more likely to be larger and have Paris-0-I morphology than SSLs without dysplasia, while location and current endoscopic signs including mucus cap, reddish color, and pit pattern were not helpful discriminators. Further studies on the endoscopic features of dysplastic SSLs by IEE are warranted to improve diagnostic accuracy of these particularly important lesions. Our study has several limitations. First, most of the included studies were cross-sectional without sufficient control of confounders. There was also significant heterogeneity between studies, including different patient groups and interventions used in the studies. Methodological considerations such as recruitment methods and risk of bias may also have contributed to the observed heterogeneity. Unfortunately, the lack of clinical and methodological details reported in the original studies precluded detailed heterogeneity analysis. Second, there were only a limited number of studies for certain endoscopic features, which may have reduced the power of our analysis. Third, NOS scores for some studies were low, especially in the domain of controlling for confounding factors. Although we performed meta-regression in an effort to explain the heterogeneity, most of the heterogeneities remain unexplained due to inherent properties of the published research. Interestingly, meta-regression revealed that SSLs in younger patients were more likely to have a mucus cap than HPs compared to older patients. This new finding requires cautious interpretation and further investigation. In conclusion, we synthesized current evidence on the distinguishing endoscopic features of SSLs in relation to HPs and conventional adenomas. Our meta-analysis highlighted several distinctive features of SSLs that might help develop novel integrated diagnostic models to inform endoscopic management. Funding This study was supported by CAMS Innovation Fund for Medical Sciences (CIFMS; No. 2022-I2M-1-003) and Peking Union Medical College (No. 2019zlgc0503). Conflicts of interest None.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.