Predictive Model Able to Predict Colorectal Cancer Risk Among Average-Risk Persons

A new predictive model for advanced colorectal neoplasia in asymptomatic adults was able to estimate advanced colorectal neoplasia risk with high discrimination among average-risk persons.¹

Moreover, the model was able to identify a lower risk subgroup that may be screened non-invasively and a higher risk subgroup for which colonoscopy may be preferred. Especially during circumstances such as the coronavirus disease 2019 (COVID-19) pandemic, the new risk estimation model could help physicians determine whether an average risk patient’s specific risk indicates an at-home stool test would be a good screening option or suggests a colonoscopy may be the most appropriate option.

“Our model helps to refine where on the average risk continuum an individual falls,” study leader Thomas Imperiale, MD, of the Regenstrief Institute and the Indiana University School of Medicine, said in a press release.² “This information could be used to guide doctor-patient discussions about screening options, with the potential to increase patient acceptance of screening by giving them a choice correlated to their individual risk – true precision medicine. Studies have shown that giving individuals a choice increases screening uptake as many people look for alternatives to colonoscopy.”

The study, published in Gut, recruited average-risk 50- to 80-year-olds undergoing first-time colonoscopy screening in Indiana. Investigators measured sociodemographic and physical features, medical and family history, and lifestyle factors and linked them to the most advanced finding.

A risk equation was then derived from two-thirds of the sample and points were assigned to each variable to create a risk score. Scores with comparable risks were collapsed into risk categories and the model and score were tested on the remaining sample.

Of 3025 subjects in the derivation set (mean age 57.3 (6.5) years; 52% women), the prevalence of advanced colorectal neoplasia was 9.4%. The 13-variable model (c-statistic = 0.77) produced a total of 3 risk groups with advanced colorectal neoplasia risks of 1.5% (95% CI, 0.72%-2.74%), 7.06% (95% CI, 5.89%-8.38%), and 27.26% (95% CI, 23.47%-31.30%) in low-risk, intermediate-risk, and high-risk groups (P value < .001), containing 23%, 59%, and 18% of subjects, respectively.

In the validation set, comprised of 1475 subjects with an advanced colorectal neoplasia prevalence of 8.4%), model performance was found to be similar (c-statistic = 0.78), with advanced colorectal neoplasia risks of 2.73% (95% CI, 1.25%-5.11%), 5.57% (95% CI, 4.12%-7.34%), and 25.79% (95% CI, 20.51%-31.66%) in low-risk, intermediate-risk, and high-risk subgroups, respectively (P < .001), containing proportions of 23%, 59%, and 18%.

“The importance of colorectal cancer screening cannot be overstated,” said Imperiale. “A home annual FIT [fecal immunochemical test] testing, which looks for blood in the stool and is inexpensive, or stool DNA and blood testing every three years, are efficient ways to screen those at the low-risk end of the average risk population.

“Particularly during the COVID-19 pandemic, as we see people less willing to consider screening colonoscopies, having an accurate risk assessment tool to determine for whom other options are perfectly good and letting them know which options are suitable is essential. It also has the added benefits of enabling us to prioritize those who are in greatest need of colonoscopy while conserving potentially scare resources — from masks and other PPE (personal protective equipment) to the ancillary costs of anesthesia.”

References:

1. Imperiale TF, Monahan PO, Stump TE, Ransohoff DF. Derivation and validation of a predictive model for advanced colorectal neoplasia in asymptomatic adults. Gut. doi: 10.1136/gutjnl-2020-321698

2. Predicting colorectal cancer risk among average risk persons [news release]. Published November 10, 2020. Accessed December 3, 2020. https://www.regenstrief.org/article/tool-predicting-colorectal-cancer-risk/