FIT Performance for CRC Screening Varies Widely

Although considered a single class, fecal immunochemical tests (FITs) vary in their ability to detect advanced colorectal neoplasia (ACN) and should not be considered interchangeable, new research suggests.

In a comparative performance analysis of five commonly used FITs for colorectal cancer (CRC) screening, researchers found statistically significant differences in positivity rates, sensitivity, and specificity, as well as important differences in rates of unusable tests.

“Our findings have practical importance for FIT-based screening programs, as these differences affect the need for repeated FIT, the yield of ACN detection, and the number of diagnostic colonoscopies that would be required to follow-up on abnormal findings,” wrote the researchers, led by Barcey T. Levy, MD, PhD, with University of Iowa, Iowa City, Iowa.

The study was published online in Annals of Internal Medicine.

Wide Variation Found

Despite widespread use of FITs for CRC screening, there is limited data to help guide test selection. Understanding the comparative performance of different FITs is “crucial” for a successful FIT-based screening program, the researchers wrote.

Levy and colleagues directly compared the performance of five commercially available FITs — including four qualitative tests (Hemoccult ICT, Hemosure iFOB, OC-Light S FIT, and QuickVue iFOB) and one quantitative test (OC-Auto FIT) — using colonoscopy as the reference standard.

Participants included a diverse group of 3761 adults (mean age, 62 years; 63% women). Each participant was given all five tests and completed them using the same stool sample. They sent the tests by first class mail to a central location, where FITs were analyzed by a trained professional on the day of receipt.

The primary outcome was test performance (sensitivity and specificity) for ACN, defined as advanced polyps or CRC.

A total of 320 participants (8.5%) were found to have ACN based on colonoscopy results, including nine with CRC (0.2%) — rates that are similar to those found in other studies.

The sensitivity for detecting ACN ranged from 10.1% (Hemoccult ICT) to 36.7% (OC-Light S FIT), and specificity varied from 85.5% (OC-Light S FIT) to 96.6% (Hemoccult ICT).

“Given the variation in FIT cutoffs reported by manufacturers, it is not surprising that tests with lower cutoffs (such as OC-Light S FIT) had higher sensitivity than tests with higher cutoffs (such as Hemoccult ICT),” Levy and colleagues wrote.

Test positivity rates varied fourfold across FITs, from 3.9% for Hemoccult ICT to 16.4% for OC-Light S FIT. 

The rates of tests deemed unevaluable (due to factors such as indeterminant results or user mistakes) ranged from 0.2% for OC-Auto FIT to 2.5% for QuickVue iFOB.

The highest positive predictive value (PPV) was observed with OC-Auto FIT (28.9%) and the lowest with Hemosure iFOB (18.2%). The negative predictive value was similar across tests, ranging from 92.2% to 93.3%, indicating consistent performance in ruling out disease.

The study also identified significant differences in test sensitivity based on factors such as the location of neoplasia (higher sensitivity for distal lesions) and patient characteristics (higher sensitivity in people with higher body mass index and lower income).

Levy and colleagues said their findings have implications both in terms of clinical benefits and cost-effectiveness of CRC screening using FITs.

“Tests with lower sensitivity will miss more patients with CRC and advanced polyps, and tests with higher sensitivity and lower PPV will require more colonoscopies to detect patients with actionable findings,” they wrote.

‘Jaw-Dropping’ Results

The sensitivity results are “jaw-dropping,” Robert Smith, PhD, senior Vice-president for cancer screening at the American Cancer Society, told Medscape Medical News. “A patient should have at least a 50/50 chance of having their colorectal cancer detected with a stool test at the time of testing.”

“What these numbers show is that the level that the manufacturers believe their test is performing is not reproduced,” Smith added.

This study adds to “concerns that have been raised about the inherent limitations and the performance of these tests that have been cleared for use and that are supposed to be lifesaving,” he said.

Clearance by the US Food and Drug Administration should mean that there’s essentially “no risk to using the test in terms of the test itself being harmful,” Smith said. But that’s not the case with FITs “because it’s harmful if you have cancer and your test doesn’t find it.”

By way of study limitations, Levy and colleagues said it’s important to note that they did not evaluate the “programmatic” sensitivity of repeating FIT testing every 1-2 years, as is generally recommended in screening guidelines. Therefore, the sensitivity of a single FIT may be lower than that of a repeated FIT. Also, variability in the FIT collection process by participants might have affected the results.

The study had no commercial funding. Disclosures for authors are available with the original article. Smith had no relevant disclosures.