Society: AGA

Background & Aims
Colonoscopy surveillance intervals are based on the predicted risk of metachronous colorectal cancer (CRC) after polyp removal. Due to the presence of co-existent findings at baseline colonoscopy, risk estimation per specific polyp subtype is difficult. To investigate the metachronous CRC risk, we evaluated the risk of patient groups with and without co-occuring findings.
Methods
Using screening colonoscopies performed after a positive fecal immunochemical test between 2014-2020 within the Dutch CRC screening program, we applied Cox regression analysis to evaluate the association between findings at baseline colonoscopy and metachronous CRC. Metachronous cancer cases, defined as CRCs diagnosed after 6 months of baseline colonoscopy, were collected from the Dutch Cancer Registry. Risk of subgroups based on polyp findings were compared to individuals without polyps in two different models. In model 1, individuals were classified into eight different subgroups based on the presence or absence of advanced adenomas and/or advanced serrated polyps. In model 2, we included presence of individual polyp characteristics in a multivariable analysis. Both models were adjusted for individuals’ sex and age. A hazard ratio >1.5 was considered clinically relevant. Advanced serrated polyps (ASP) were defined as serrated polyp ≥10mm, sessile serrated lesion with dysplasia, or traditional serrated adenoma.
Results
253,833 colonoscopies were included. Over a median follow-up of 36 months (IQR, 21-57), we identified 504 metachronous CRC cases. HR for CRC was 1.70 (CI95%, 1.07-2.69) for individuals with ASP without advanced adenomas (AA), 1.22 (0.96-1.55) for individuals with AAs without ASPs, and 2.00 (1.19-3.39) for individuals with ASPs and AA, compared to patients without polyps (Table 1). Individuals with non-advanced adenomas and/or non-advanced serrated polyps did not have an increased CRC risk. Model 2 showed that any villous adenoma (HR 2.732, 1.725-4.329), serrated polyp ≥10mm (HR 1.61; 1.06-2.45), sessile serrated lesion with dysplasia (HR 2.09; 1.12-3.88), traditional serrated adenoma (HR 2.668, 1.434-4.965), adenoma with high-grade dysplasia (HR 3.59; 2.17-5.94) and presence of at least five non-advanced adenomas (HR 2.13; 1.10-4.10) resulted in a clinically relevant increased risk for CRC, while any adenoma ≥10mm did not (HR 0.96; 0.74-1.25) (Table 2).
Conclusion
By evaluating high-quality screening colonoscopies with a median follow-up of 3 years and accounting for co-occuring findings, we observed an increased CRC risk in individuals that had ASPs with the presence of AAs, or individuals with ASPs without AAs. These findings could contribute to establish more restrictive post-polypectomy surveillance guidelines.

Table 1

Table 2

Background: Post-colonoscopy colorectal cancer (PCCRC) is defined as cancers diagnosed after a colonoscopy in which no cancer is found. The clinical features, prognosis, and risk factors of PCCRC remain poorly characterized.
Methods: In a large healthcare system (Mass General Brigham), we assembled a longitudinal cohort of patients who had undergone a colonoscopy between 2010-2018 by extracting detailed endoscopic, pathologic, and other clinical data from the electronic health records. We identified patients who were diagnosed with CRC between 6 and 48 months after a CRC-free index colonoscopy (PCCRC<4y) and at least 48 months after a CRC-free colonoscopy (PCCRC≥4y). We conducted chart review to categorize the causes of PCCRC according to the World Endoscopy Organization (WEO) recommendations and compared the anatomic subsites, stages, and survival between PCCRC<4y and PCCRC≥4y. We then matched up to 3 controls for each CRC patient based upon age, sex, race/ethnicity, and time and reason of index colonoscopy; and assessed potential risk factors for PCCRC using conditional logistic regression and accounting for missing data via multiple imputation.
Results: During a median follow-up of 5.6 years among 198,066 patients with a CRC-free index colonoscopy, we documented 121 incident PCCRC<4y and 86 PCCRC≥4y cases (females: 42% vs. 53%; mean age=67.7 vs. 70.1 years; median interval between index colonoscopy and diagnosis=2.2 vs. 6.0 years). According to the WEO categorization, 90% of PCCRC<4y cases likely occurred due to possible missed lesions or inadequate exam. Compared to PCCRC≥4ys, PCCRC<4ys had a more advanced stage at diagnosis (p=0.004) but no statistically significant difference in subsites (p=0.56). Among patients with a polyp finding in the index colonoscopy, PCCRC<4ys were more likely to arise in the same broad location (proximal colon, distal colon, and rectum) as prior polyps than PCCRC≥4ys, particularly in the proximal colon (57% vs. 18%). During a median follow-up of 4.6 and 3.3 years of PCCRC<4y and PCCRC≥4y patients, we documented 28 and 9 deaths, respectively; a lower 5-year survival was observed for PCCRC<4y than PCCRC≥4y (68% vs. 76%). By comparing PCCRC<4ys and PCCRC≥4ys and their matched controls (353 and 252 controls, respectively), we found that the presence of large (≥1 cm) index polyps was more strongly associated with higher risk of PCCRC<4ys than PCCRC≥4ys (p for heterogeneity=0.006), whereas no significant difference was found in other risk factors, including histology and multiplicity of index polyps and family history of CRC.
Conclusions: 90% of PCCRC<4y cases are avoidable via improved colonoscopy quality and lesion detection. PCCRC<4y had more unfavorable clinical profiles than PCCRC≥4y. Patients with large index polyps were at a particularly high risk of developing PCCRC<4y, likely in the same location.

Figure 1. Potential causes of PCCRC based on the WEO classification

Figure 1. Potential causes of PCCRC based on the WEO classification

Figure 2. Clinical features of PCCRC<4y and PCCRC≥4y

Figure 2. Clinical features of PCCRC<4y and PCCRC≥4y

Background and Aims: Colorectal cancer (CRC) is one of the commonest cancers worldwide. However, a majority of CRC demonstrate intrinsic resistance to immune checkpoint blockade (ICB) therapy. Cancer stem cells (CSCs) is a subpopulation of cancer cells considered as key players in chemoresistance, tumor relapse and metastasis. However, it is unclear whether CSCs might modulate antitumor immunity and its therapeutic implications.

Methods: Colorectal CSCs (POP66 and LS174T) were established from primary human CRC. Correlation between cancer stemness characteristics and immune checkpoint was examined by single cell RNA-seq (scRNA-seq) of CRC from azoxymethane/Dextran sodium sulfate (AOM-DSS)-treated mice model. TCGA cohort was used to examine correlation of CSC markers with immune checkpoints and SLC25A22. To test the function of SLC25A22 in immune evasion by colorectal CSCs, a colon stem cell-specific SLC25A22 knockout mice (Slc25a22^fl/flLgr5-Cre^ERT2) was established.

Results: Colorectal CSCs have remarkably high expression of immune checkpoints, including Galectin-9 (LGALS9) and PVR (CD155), as compared to a panel of CRC cells. In AOM/DSS-induced CRC, scRNA-seq revealed that Lgr5⁺ CSCs subpopulation have increased immune checkpoints expression compared to non-CSCs fraction. Corroborating these findings, CSCs markers positively correlated to immune checkpoint in TCGA CRC cohort. In support of this notion, co-culture of CSCs (POP66 and LS174T) with human CD8⁺ T cells impaired T cell proliferation and attenuated markers of T cell activity, including Granzyme B, IFN-γ and TNF-α. We next investigated if the blockade of SLC25A22, a driver of WNT-dependent CSCs, could reverse immunosuppressive phenotype of colorectal CSCs. SLC25A22 knockdown in CSCs inhibited stemness and immune checkpoints Galectin-9, PVR, and PD-L1. This in turn reversed suppressive effect of colorectal CSCs on CD8⁺ T cell proliferation and cytotoxic activation. We next established colon stem cell-specific SLC25A22 knockout mice, and demonstrated that SLC25A22 loss impaired AOM/DSS-induced tumorigenesis (P<0.01). Notably, scRNA-seq showed that SLC25A22 depletion decreased Lgr5⁺ CSCs in CRC, leading reduced intratumoral immune checkpoint expression. Assessment of tumor infiltrating immune cells with cytometry by time-of-flight (Cy-TOF) and flow cytometry validated that SLC25A22 knockout increased infiltration of CD8⁺ T cells, together with elevated expression of granzyme B, IFN-γ and TNF-α. Mechanistically, we revealed that SLC25A22-driven WNT signaling underlies the increased expression of immune checkpoints in colorectal CSCs.

Conclusion: Colorectal CSCs are drivers of immune evasion via the overexpression of immune checkpoints to antagonize intratumoral cytotoxic T cells. Targeting of SLC25A22 represents a potential therapeutic approach to reverse CSCs-mediated immunosuppression.

Background: Post colonoscopy colorectal cancers (PCCRCs) are an important colonoscopy quality indicator, but the methodology used to measure PCCRCs varies across jurisdictions. The World Endoscopy Organization (WEO) PCCRC methodology, developed using expert consensus, standardizes the approach to measuring PCCRC, but limitations persist. Recommendations for high-risk populations, including people with chronic inflammatory bowel disease (IBD) or recent colorectal cancer (CRC), often include surveillance colonoscopies at short intervals. In these individuals, cancers may be classified as PCCRCs when adherent to recommended surveillance intervals and as ‘detected CRCs’ when poorly adherent. Our aim was to explore variations in the WEO methodology to address indicator timeliness and impact of high-risk populations, using population-based data from Ontario, Canada.

Methods: We compared Ontario PCCRC rates using the WEO approach ((false-negative colonoscopies/false-negative + true-positive colonoscopies) x 100) using a look-forward (LF) (WEO approach, based on the year of the colonoscopy) and a look back (LB) (based on the year of the cancer diagnosis) method. False-negative colonoscopies refer to colonoscopies where a CRC is diagnosed within 6 to 36 months. True-positive colonoscopies refer to those where a CRC is diagnosed within 6 months. For people with multiple colonoscopies, only true-positive and false-negative colonoscopies closest to the cancer diagnosis date were included. We compared annual Ontario PCCRC rates from 2012-2017 using the LF method and from 2015-2020 using the LB method and examined timeliness using the 2020 reporting year. Using these same methods, overall PCCRC rates were calculated including and excluding those with IBD or multiple CRCs.

Results: From 2012 to 2017, using the LF method, false-negative, true-positive colonoscopies and annual PCCRC rates ranged from 547 to 614, 5238 to 5513 and 9.28% to 10.18%, respectively. From 2015 to 2020, using the LB method, false-negative, true-positive colonoscopies and annual PCCRC rates ranged from 521 to 640, 4921 to 5476 and 9.34% to 10.67% (See Table 1). The LB method provided more timely data for PCCRCs (LB: 100% in 2020 (reporting year) vs LF: 14% in 2020, 39% in 2019, 41% in 2018, 7% in 2017) and for colonoscopy (LB: 8% in 2020, 42% in 2019, 35% in 2018, 15% in 2017) vs LF: 100% in 2017) (Figure 1). In 2020 using the LB method, PCCRC rates dropped from 9.57% to 9.0% when persons with IBD were excluded and to 8.31% when only the first CRC per patient was included.

Limitations: Data from 2020 may be impacted by the COVID-19 pandemic.

Conclusion: Applying the LB method to the WEO methodology to measure PCCRCs allows more timely reporting and does not appear to impact PCCRC rates. Excluding high-risk populations does not appear to have a large impact on PCCRC rates at the population-level.

Table 1. Post colonoscopy colorectal cancer rates in Ontario, Canada from 2012-2020 measured with look forward and look back methods.

Table 1. Post colonoscopy colorectal cancer rates in Ontario, Canada from 2012-2020 measured with look forward and look back methods.

Figure 1. Timeliness of PCCRC diagnosis and colonoscopies using look forward (LF) and look back (LB) method using the example of the 2020 reporting year.

Figure 1. Timeliness of PCCRC diagnosis and colonoscopies using look forward (LF) and look back (LB) method using the example of the 2020 reporting year.

Background: Patients with inflammatory bowel disease (IBD) involving at least one-third of the colon have a significantly increased risk of developing colorectal cancer (CRC), with an estimated prevalence of 18% after 30 years of disease. To reduce the risk of CRC, many societies have recommended more regular screening intervals in addition to the implementation of improved technology for dysplasia detection. Historically, CRC screening for IBD involved white light endoscopy (WLE); however, in 2015, the SCENIC international guidelines strongly recommended the use of chromoendoscopy to improve dysplasia detection rates. The aim of our meta-analysis is to evaluate the efficacy of high definition white light endoscopy (HDWLE) versus dye-spraying chromoendoscopy (DCE) in detecting dysplasia.

Methods: A systematic review of the literature in PubMed, EMBASE, and Web of Science was performed from April 2018 to November 2022 using the same search criteria as our prior meta-analysis (Feuerstein et al. 2019). We screened these studies to ensure they adhered to our inclusion criteria (adult humans, IBD-related CRC screening, HDWLE and DCE arms). The primary outcome was the proportion of patients identified with dysplasia. Secondary outcomes were the proportion of patients with high-grade dysplasia (HGD) as well as colonoscopy duration. Using RevMan 5.4, we performed separate meta-analyses of randomized controlled trials (RCTs) and non-RCTs.

Results: There was a total of 14,240 studies assessed (142 new and 14,098 prior). Of these, nine were included in our analysis. In RCTs comparing DCE and HDWLE, dysplasia was detected in 19.6% (78/397) and 12.0% (48/401) of patients undergoing DCE and HDWLE screening, respectively, with a trend towards favoring DCE (RR 1.73, CI 0.98 – 3.07, p=0.06). Amongst the non-RCT studies, DCE (21.1%, 76/361) was more effective than HDWLE (109/1004, 10.9%) in identifying dysplasia (RR 2.62, CI 2.01-3.40, p < 0.01). When assessing HGD specifically, there was no difference between DCE and HDWLE in both RCT (RR 1.36, CI 0.31-6.05, p=0.69) and non-RCT data (RR 0.94, CI 0.37 – 2.43, p=0.91). Adenocarcinoma was rare in both RCTs and non-RCTs, being identified in 0.27% (2/758) and 0.28% (4/1405) of patients undergoing screening with DCE and HDWLE, respectively. Finally, total procedure time with DCE was on average 6 minutes longer than HDWLE (CI 2.24 – 11.19). Overall, the GRADE certainty of evidence for the RCT and non-RCT was low quality.

Conclusions: Based on the updated meta-analysis findings, non-RCT data suggests a robust benefit of DCE over HDWLE in dysplasia detection, while RCT data suggests a smaller potential benefit. Future studies are needed to assess the true benefit of DCE, especially in light of its increased procedural time without clear evidence of improved HGD detection.

Forest plots comparing the efficacy of DCE versus HDWLE in the detection of dysplasia and HGD

Forest plots comparing the efficacy of DCE versus HDWLE in the detection of dysplasia and HGD

Difference in procedure time in DCE versus HDWLE

Difference in procedure time in DCE versus HDWLE

Background: In contrast to screen-detected colorectal cancers (CRC) post-colonoscopy CRCs (PCCRCs) are most often detected at an advanced stage, which may negatively impact the efficacy of screening. By characterizing PCCRCs as interval or non-interval and assessing the most probable etiology, insights can be provided which may contribute in PCCRC prevention.
Methods: PCCRCs diagnosed after screening colonoscopy, performed between 2014-2016 after positive fecal immunochemical test (FIT) for CRC screening, were included. Demographic, endoscopic, and pathologic data were retrieved. PCCRCs were categorized, according to the WEO consensus statement, in interval PCCRC (CRC detected before the recommended surveillance) or non-interval PCCRC defined as type-A (CRC detected at the recommended surveillance colonoscopy), type-B (CRC diagnosed after the recommended surveillance interval) or type-C (CRC diagnosed in patients without surveillance). A root-cause analysis was performed for each PCCRC to determine the most probable etiology. Tumor stage distributions were compared between the PCCRC categories.
Results: In total, 432 PCCRCs were diagnosed in 116,362 participants undergoing screening colonoscopy after positive FIT. The 3-year PCCRC rate was 2.8%. Age, sex, and hemoglobin level were similar between participants that developed a PCCRC and those that did not. PCCRCs were classified as interval (n=214, 49.5%), non-interval type-A (n=82, 19.0%), non-interval type-B (n=130, 30.1%) and non-interval type-C (n=6, 1.4%). Most PCCRCs had as most plausible etiology a missed lesion with an adequate prior examination (n=202, 47.4%) or an incomplete resection of a previously identified polyp (n=107, 25.1%). Interval and non-interval type-C PCCRCs were more often diagnosed at a late stage (stage III or IV) (n=116, 54.2% and n=6, 100%, respectively), compared to non-interval type-A and type-B (n=18, 22.0% and n=53, 40.8%, respectively). The majority of non-interval type-B PCCRCs (n=98, 81.7%) were diagnosed in patients with an incomplete index colonoscopy and a recommendation for follow-up within six months: namely procedures with referral for polyp resection (n=60, 46.2%), without cecal intubation (n=26, 20.0%), or with insufficient bowel preparation (n=12, 9.2%).
Conclusion: In a FIT-based CRC screening program 50% of the PCCRCs were classified as interval. These were more often diagnosed at an advanced stage compared to non-interval type-A and type-B. This emphasizes the importance of high-quality index colonoscopy. Besides that, timely follow-up after incomplete index colonoscopy and large polyp removal should be recommended to reduce (non-interval type-B) PCCRCs.

Evelien Dekker

Academic Medical Center

Appearances