— Study finds FIT misses more cancers and precancerous lesions in women at standard threshold
October 14, 2026 • 3 min read
At the standard positivity threshold, fecal immunochemical testing (FIT) is likely to miss more colorectal cancers (CRCs) and advanced precancerous lesions (APLs) in women versus men, a retrospective analysis suggested.
Among more than 30,000 screening participants, men had higher fecal hemoglobin levels and FIT sensitivities for CRCs and APLs compared with women, reported Thomas F. Imperiale, MD, of Indiana University School of Medicine in Indianapolis, at the American College of Gastroenterologyopens in a new tab or window annual meeting in Nashville, Tennessee.
Fecal hemoglobin levels for CRCs were a mean 955 ng/mL for women compared with 1,541 ng/mL for men (P=0.03), while mean levels for APLs were 174 and 230 ng/mL, respectively (P<0.001).
At a threshold of 100 ng/mL -- the standard positivity threshold for an FIT test -- there was a higher sensitivity for both CRCs and APLs for men versus women:
- CRC: 77.4% vs 62% (P=0.03)
- APLs: 25.8% vs 20.2% (P<0.001)
However, specificity was slightly lower for men than women (94.2% vs 95.3%, P<0.001).
In order to meet the CRC sensitivity for men, the threshold in women would have to be lowered to 12 ng/mL, reducing specificity to 75.7%. To meet APL sensitivity, the threshold in women would have to be reduced to 65 ng/mL, resulting in a specificity of 92.8%.
Additionally, to reach 95% specificity would require thresholds of 120 ng/mL for men and 94 ng/mL for women.
After his presentation, Imperiale was asked if he had concerns about lowering the threshold for FIT in women in order to detect more adenomas and cancers.
"It's an open [question] because it needs consideration of the downstream effects," Imperiale noted. "But, if the goal were to equal cancer sensitivity detection, then the threshold should be different. Exactly what that should be would depend on which test is used, because there is going to be variation among those tests."
He added that much would depend on what is considered important in a screening program, adding that a survey of population-based screening programs indicated that maximizing sensitivity was a primary goal.
"But, other factors were [important] as well," Imperiale said. "So, it's a trade-off, and I think understanding performance and long-term outcomes is what is most important."
In explaining the rationale behind this analysis, Imperiale noted that FIT -- the most frequently used CRC screening tool worldwide -- has an average sensitivity of 79% and a specificity of 94%, with test characteristics varying depending on threshold and the brand of test.
How performance varies by sex is "less clear," he said. "We know the prevalence and distribution of colorectal cancer and advanced precancerous lesions differs between males and females, but do these differences warrant different FIT thresholds?"
This retrospective analysis used data from the DeeP-C opens in a new tab or windowand BLUE-C opens in a new tab or windowstudies -- validation studies for multi-target stool DNA testing.
Imperiale and colleagues included 30,165 study participants (mean age 63.4 years, 53.4% women), of whom 0.54% had CRC and 9.6% had APLs.
There was no significant difference by sex in fecal concentration levels for CRCs or APLs among those with nonadvanced neoplasia, nonneoplastic findings, or normal colonoscopies.
While there was no difference in lesion size between sexes, there was a difference in location. A higher proportion of women had proximal CRCs versus men (48.1% vs 31%), with a much lower sensitivity for CRC at the 100-ng/mL threshold (50% vs 80.8%, P=0.01). Women also had a higher proportion of proximal APLs (61.2% vs 53.6%), with a lower APL sensitivity (12.8% vs 17.7%, P=0.01).
In a multivariable analysis, male sex, current or former smoking, a body mass index ≥30, larger lesion size, and distal/rectal location were independently associated with FIT sensitivity for advanced neoplasia.
Imperiale acknowledged that the analysis was limited since it involved a single brand of test, a single threshold, and a single application of a test, "the benefit of which increases with programmatic application."