Impact of Reference Standards on HCC-Confirmed Li-RADS Categories

A meta-analysis evaluating LI-RADS diagnostic accuracy studies reports that pathology-based reference standards were approximately four times more common than clinical reference standards. The study estimated the impact of pathology-based reference standards versus clinical reference standards that exclude pathologic findings on the percentage of hepatocellular carcinoma (HCC) confirmed for each LI-RADS category.

Using data from CT, MRI and contrast-enhanced ultrasound studies, the reference standard was classified as either pathology-based (explant, surgical resection or biopsy) or a clinical reference standard with pathology. Of the 32 studies included, there were 5,882 observations of which 4,002, or 68%, had HCC observations. Of these, 80% were diagnosed with a pathology-based reference standard and 20% with a clinical reference standard.

The authors also reported that observations confirmed as HCC where more than twice as likely to have a pathology-based reference standard. They suggest that requiring pathologic confirmation may introduce selection bias, particularly for the LR-2 and LR-3 categories that are not recommended for pathologic confirmation, due to the larger proportion of observations confirmed as HCC. Future studies can reduce this bias by including reference standards that are not pathology-based, especially for LR-2, LR-3 and LR-5 that are often not confirmed with pathologic examination.