Advanced adenomas help predict risk after negative surveillance

Patients with advanced adenomas or five or more nonadvanced adenomas on index colonoscopy had more than double the likelihood of advanced neoplasia at a subsequent surveillance colonoscopy despite a normal intervening exam, according to a prospective analysis published in Clinical Gastroenterology and Hepatology.

"Data on subsequent risk of neoplasia in patients with index adenomas and a first surveillance colonoscopy with no polyp findings are crucial when risk stratifying patients for appropriate surveillance interval recommendations. This is a common clinical challenge for endoscopists," noted Joseph C. Anderson, MD, of the Geisel School of Medicine at Dartmouth University in Hanover, NH, and colleagues. "The results of this study can aid endoscopists in decision-making surrounding interval follow-up of patients with a history of polypectomy on index colonoscopy and a normal first surveillance exam."

They analyzed 3,171 patients from the New Hampshire Colonoscopy Registry with no prior colorectal neoplasia who underwent an index colonoscopy followed by two surveillance examinations spaced more than 12 months apart. All patients had no polyps detected at the first surveillance colonoscopy, allowing assessment of residual risk after a negative exam.

Patients were stratified by index findings into four groups: no adenomas (n = 1,948); one to two nonadvanced adenomas (n = 846); three to four nonadvanced adenomas (n = 846); and advanced adenomas or five or more nonadvanced adenomas (n = 308). Advanced outcomes at second surveillance were defined as advanced adenomas, colorectal cancer, or large or dysplastic serrated polyps. 

Advanced outcomes occurred in 3.7% of patients with no index adenomas, 6.3% with one to two nonadvanced adenomas, 4.3% with three to four nonadvanced adenomas, and 6.2% with advanced adenomas or five or more nonadvanced adenomas. 

Compared with patients with no adenomas, those with advanced adenomas or five or more nonadvanced adenomas had 2.12 times the likelihood of advanced findings at second surveillance. Those with one to two nonadvanced adenomas had 1.67 times the odds, while those with three to four nonadvanced adenomas did not have a statistically significant increase in risk. 

When restricted to advanced adenoma or colorectal cancer, the association was stronger: patients with advanced adenomas or five or more nonadvanced adenomas had 3.47 times the likelihood of detection at second surveillance. 

Several clinically relevant modifiers were identified. Current smoking at the time of first surveillance was associated with approximately three times the likelihood of advanced findings. High-risk serrated polyps at index colonoscopy were also linked with increased risk. In contrast, higher adenoma detection rates at first surveillance were associated with lower risk, underscoring the importance of colonoscopy quality. 

Among patients with advanced adenomas or five or more nonadvanced adenomas, the mean interval from first surveillance to detection of advanced findings was approximately 54 months. This timeframe aligns with current US Multi-Society Task Force recommendations for five-year follow-up in higher-risk patients with a normal first surveillance colonoscopy.  

The findings indicate that a negative first surveillance colonoscopy does not eliminate risk in patients with high-risk index findings. In contrast, patients with three to four nonadvanced adenomas did not show an increased risk of advanced neoplasia, suggesting that intensified surveillance may not be necessary, although they remain at risk for recurrent nonadvanced adenomas. 

The study also highlights the impact of colonoscopy quality. Lower adenoma detection rates at first surveillance were associated with higher risk of advanced findings at follow-up, suggesting that missed lesions may contribute to subsequent neoplasia. 

The study had several limitations. Registry-based data may not capture all colorectal cancer diagnoses outside participating centers. Smoking status was self-reported, and the study population lacked racial diversity. Selection bias was also possible because inclusion required registry participation and complete follow-up. 

The researchers reported no conflicts of interest.