Scoring Tool Predicts Colorectal Cancer Risk in Younger Adults

Although colorectal cancer (CRC) incidence rates have steadily declined since population-based screening was introduced in the 1990s, CRC rates among patients younger than 50 years, referred to as early-onset colorectal cancer (EOCRC), have steadily risen. However, CRC screening is only recommended for patients older than 45 years despite that half of new EOCRC cases being diagnosed in individuals younger than 45 years.

“Although we know that EOCRC cases have been steadily increasing over the past few decades, we don’t know exactly what the underlying cause is,” explains Carole Macaron, a staff physician in Cleveland Clinic’s Gastroenterology, Hepatology & Nutrition Department. “Prior studies have identified several factors — sex, race, family history of CRC, smoking and alcohol consumption history, obesity, diet, diabetes and hyperlipidemia — impacting risk. We felt that if we could better understand how these factors weigh into a patient’s risk, we could use them to help identify patients younger than 45 who could benefit from earlier CRC screening.”

Dr. Macaron is senior author of a recent study in Digestive Diseases and Sciences, which looked to develop a model estimating the likelihood of advanced colorectal neoplasia (AN) in adults age < 45 years. The group first identified which factors were significantly associated with risk of detecting AN. Then, using those findings, they created an internally validated model and prediction score based on easily obtainable clinical factors.

Study design

The group performed a cross-sectional analysis of adults between 18 and 44 years who underwent a colonoscopy at Cleveland Clinic between January 2011 and December 2021. After excluding patients with high-risk indications for colonoscopy, 9,446 patients were included in the study.

Of these, 3,681 (39%) patients were male, and 5,765 patients (61%) were female. The mean age of study participants was 36.8 ± 5.4, and most patients were white (81.8%) and non-Hispanic (89.4%). The mean BMI was 28.7 ± 7.1. Most patients (80.3%) had no family history of CRC, though 595 (6.3%) had one first-degree relative with CRC younger than 60 years of age and 238 (2.5%) had one first-degree relative 60 years of age or older. There were 3468 (39.7%) patients who reported current or former tobacco use and 6489 (70.9%) patients reported current or former alcohol use. There was a history of hyperlipidemia in 1,146 patients (12.3%), and 236 patients (2.5%) had a history of diabetes. Additionally, 735 (7.8%) patients reported using aspirin.

The group detected AN in 346 patients (3.6%). Of these, 174 patients were found to have a tubular adenoma (TA) ≥ 10 mm, 90 (26.0%) had a tubulovillous adenoma (TVA), 26 (7.5%) had an adenoma with high-grade dysplasia (HGD), 33 (9.5%) had a sessile serrated polyp (SSP) ≥ 10 mm, 15 (4.3%) had a traditional serrated adenoma (TSA), and 8 (2.3%) were found to have invasive adenocarcinoma.

Developing the scoring tool

“We used a multivariate logistic regression model to identify clinical predictors of AN,” says Dr. Macaron. “We found significant association between AN and several variables.”

These included: BMI (P = .0157), former tobacco use (P value = .0009), current tobacco use (P value = .0015), first-degree relative with CRC (P value < .0001) and other family history of CRC (P = 0.0117).

“Interestingly, the sex of a patient did not have a significant association, so it was left off our final model,” says Dr. Macaron.

In the validation set, the model showed moderate discrimination (c-statistic 0.645). Overall, the estimated AN rates ranged from 1.8% in patients with a score of 1 to 22% in patients with a score of 12.

"We found that the risk increases significantly at a score of 9," explains Dr. Macaron. "Patients with a score of 8 have a risk of about 6%, but that jumps to 14% at a score of 9."

Narrowing the gap in CRC screening

Dr. Macaron says that although the tool has been internally validated, it still needs to be prospectively validated. Ultimately, the team hopes the tool can help bridge the gap in current CRC screening recommendations by providing a tailored screening strategy for younger patients who appear to be at high risk and may benefit from earlier screening.

“One of the aspects of this tool that we’re so excited about is that the variables we focused on are all easily available to primary care physicians,” explains Dr. Macaron. “Most patients aren’t regularly meeting with a gastroenterologist, but PCPs can be safety nets and help identify patients who may be at risk. The variables we looked at – BMI, smoking and alcohol consumption history, and family history – are all right there in every patient’s chart. There’s also the potential to convert the model into an online calculator that could be incorporated into a medical center’s electronic medical records.”

Looking ahead at next steps

Although Dr. Macaron expects the screening recommendations to eventually be reconsidered as they relate to age, she believes the scoring tool can help identify younger at-risk patients in the meantime.

“To our knowledge, this is the first U.S. study to create an easy-to-calculate risk prediction score for adults younger than 45 years of age,” she says. “While it’s still early, and our tool needs to be prospectively validated, we believe the initial results are promising, and we’re excited about its potential to assist with risk stratification in an increasingly at-risk age group.”