Mo1662 PATIENT AND ENDOSCOPIC CHARACTERISTICS OF POST-COLONOSCOPY COLON CANCER: A CASE-CONTROL STUDY

Abstract

There is good evidence that screening programs are successful at reducing the risk of colorectal cancer (CRC). Colonoscopy is the gold standard, however it remains imperfect and colon cancers are still detected after screening or surveillance exams. Developing CRC within 10 years of a prior colonoscopy without cancer is considered post colonoscopy colorectal cancer (PCCRC). Possible explanations include missed lesions, new lesions, incomplete resection, and aggressive biology. The aim of this study was to compare patient characteristics and endoscopic findings of PCCRC cases with those from controls who did not develop CRC. Using a tumor registry dataset, we extracted all CRCs diagnosed at our medical center from 2012–2017. We identified those with prior colonoscopy without cancer within 10 years (index colonoscopy), which were considered PCCRC cases. Major exclusions were prior CRC or hereditary CRC syndromes, IBD, colonoscopies not done at our institution and incomplete polyp resection. Controls were collected from a cohort of all colonoscopies performed from 2003–2019. They never had CRC or high risk adenomas, had colonoscopy for screening indications, and were still followed in our medical record when they were identified (3/2019). The date of index colonoscopy, age and sex for cases was matched in a 3:1 fashion with controls and data were analyzed using conditional logistic regression. Index colonoscopies of 54 PCCRCs were matched with 162 controls. Patient characteristics were similar for both groups (age, sex, smoking, EtOH). The following features were significantly associated with risk of PCCRC: ever history of polyps (OR=3.84), ever history of serrated adenomas (OR=5.33), 2 v 0 polyps (OR=4.55), having a tubular adenoma (TA) (OR=3.53) or sessile serrated adenoma (SSA) (OR=8.49) on index colonoscopy, or having a follow-up recommendation from the endoscopist that did not correlate with the AGA guidelines (OR=5.40). 61% of PCCRC cases had polyps on their index colonoscopy in the same location as their PCCRC. Features that were not significantly associated with risk of PCCRC included: first degree relative with CRC, prep quality, proximal v distal location of polyp, 1 v 0 polyp, hyperplastic polyps, and polyp size (<5mm v 5-9mm). Not surprisingly, patients with TAs/SSAs or multiple polyps have higher risk of developing cancer, including PCCRC. We also show that when comparing TAs to SSAs, there is no significantly elevated risk of developing PCCRC. Given that patient and endoscopic characteristics do not differ between the groups, these data indicate that endoscopy-related factors such as missed or incompletely resected polyps may explain most PCCRC. Increased risk of PCCRC in patients who had incorrect follow-up recommendations should be further investigated.