A retrospective study of 127 colon cancer cases preceded by conventional abdominal CT has found radiologists missed the cancers a fifth of the time.

Although the scans were performed for reasons other than detecting colon cancer, the authors call for training radiologists in several clinical confounders that, in the study, correlated with the misses.

Clinical Radiology published the study, conducted in Israel at Tel Aviv University and the Chaim Sheba Medical Center, online July 4.

Lead author Eyal Klang, MD, and colleagues reviewed the cases of consecutive patients diagnosed with colorectal cancer at colonoscopy from 2006 to 2015 who also underwent abdominal CT (with and without oral and/or intravenous contrast) a year or less before the colonoscopy.

They looked at the original readers’ frequency of undetected tumors, then had the images independently interpreted by two radiologists (the “study readers”) who were alerted to cancer diagnosis but blinded to tumor location.

The researchers found that the original interpreters missed 25 of 127 colon cancers (19.7 percent), while the study readers found the tumors in all but eight cases (6.3 percent).

They further found that shorter tumor length contributed to the misses, as did the absence of several clinical conditions. These included fat stranding (subcutaneous swelling), vessel engorgement and swollen stomach lymph nodes.

“[A] high rate of missed colon cancers in non-targeted CT was demonstrated,” Klang et al. conclude. “Clinicians should be aware that if no tumor was reported in abdominal CT report, cancer has not been ruled out.”

Education on colon tumors “should be emphasized in the training process of radiologists, as it may improve their practices,” they note. “These findings also have medico-legal importance.”

The authors also recommend raising clinicians’ awareness of abdominal CT’s limitations.