From the AGA Journals

Meta-analysis generally supports LI-RADS classification accuracy



Higher (more severe) Liver Imaging Reporting and Data System (LI-RADS) categories contained increasing proportions of hepatocellular carcinomas and overall malignancies, supporting the general reliability of the system, according to a systematic review and meta-analysis of 17 retrospective studies.

But 13% of LR-2 (“probably benign”) observations were actually hepatocellular carcinomas, as were 38% of LR-3 (“intermediate probability of malignancy”) observations, reported Christian B. van der Pol, MD, of McMaster University, Hamilton, Ont., and Christopher S. Lim, BBS, of Harvard Medical School, Boston, and their associates. Thus, clinicians should consider biopsy of many LR-3s, and LR-2s might need “more active management” than the currently recommended “return to surveillance,” including consideration for biopsy of solid LR-2 nodules measuring 1 cm or more, they wrote in Gastroenterology.

Histopathology confirmed that 93% of CT and MRI observations designated as LR-M (“definite or probable malignancy”) were indeed malignancies and that 36% were hepatocellular carcinomas,

The LI-RADS system, like its counterparts in breast and prostate imaging (BI-RADS and PI-RADS), classifies CT and MRI findings based on level of suspicion for malignancy. These categories include LR-M, LR-3, LR-2, LR-1 (“definitely benign”), LR-TIV (“definitely tumor in vein”), and LR-4 and LR-5 (“probably” and “definitely” hepatocellular carcinoma). However, CT and MRI interpretation is only as useful as it is accurate. To calculate actual percentages of hepatocellular carcinomas and overall malignancies within each LI-RADS category, the investigators analyzed aggregate data from studies found by searching MEDLINE, Embase, Cochrane CENTRAL, and Scopus during 2014-2018.

These 17 studies included 2,760 patients and 3,556 imaging observations. Pathology was the reference standard for LR-M, but for other LI-RADS categories, the researchers accepted strong clinical indicators of hepatocellular carcinoma, such as a 50% increase in lesion size within 6 months, or posttreatment recurrence of a previously confirmed malignancy. They classified observations as negative if they stayed stable in size for at least 12 months, spontaneously diminished in size, or disappeared without treatment.

In all, 94% and 97% of LR-5 observations were (respectively) hepatocellular carcinomas and other malignancies, as were 79% and 92% of LR-TIVs, 36% and 93% of LR-Ms, 74% and 80% of LR-4s, 38% and 40% of LR-3s, and 13% and 14% of LR-2s. No LR-1s were confirmed as malignant.

“Our data suggest biopsy of LI-RADS 3 observations should be considered in many patients, as a risk of 38% of HCC would usually provoke biopsy of a lesion elsewhere in the body,” the researchers wrote. They suggested consideration for biopsy of certain LR-2 lesions, but added that many “are small, perfusional alterations caused by arterioportal shunts, which are often not reported” and would be difficult or impossible to biopsy.

The study did not cover the most recent (2018) LI-RADS system, which featured several changes to simplify and better align it with American Association for the Study of Liver Diseases criteria, the researchers noted. They called for prospective studies to help confirm the accuracy of the LI-RADS system, particularly with regard to intermediate categories, such as LR-2.

The researchers disclosed no funding sources. Dr. van der Pol, Dr. Lim, and three other investigators reported having no conflicts of interest. Five researchers reported that they are members of the LI-RADS Steering Committee and four disclosed ties to pharmaceutical companies.

SOURCE: Van der Pol CB et al. Gastroenterology. 2018 Nov 13. doi: 10.1053/j.gastro.2018.11.020.

Next Article: