Free cookie consent management tool by TermsFeed



life long learning

April 2024

LI-RADS CT and MRI Ancillary Feature Association with Hepatocellular Carcinoma and Malignancy...
Journal Watch by Pedro Coroas da Rocha

LI-RADS CT and MRI Ancillary Feature Association with Hepatocellular Carcinoma and Malignancy: An Individual Participant Data Meta-Analysis
Dawit H, Lam E, McInnes MDF, et al.
Radiology. 2024;310(2):e231501.


As of 2022, hepatocellular carcinoma (HCC) ranks as the third leading cause of cancer death worldwide, with a relative 5-year survival rate of approximately 18% (1,2).CT, MRI, and in some instances, CEUS, play an essential role in the initial assessment of HCC (3).

The LI-RADS was created by the American College of Radiology to establish uniformity in imaging reports of patients at risk for HCC (4). The LI-RADS diagnostic framework is based on assigning major criteria and ancillary features (AF) to a final lesion score, which reflects a probability of HCC and overall malignancy (4). AFs can favor HCC, malignancy in general, or benignity by upgrading or downgrading by one level, the category assigned, except in specific cases (upgrading to LI-RADS 5).

Although major criteria have been considerably investigated, there is little evidence of the contribution of AF to LI-RADS diagnostic performance. This individual participant data meta-analysis aimed to evaluate the association of LI-RADS's AFs with HCC and malignancy.

 This review included 20 studies composed of 3091 CT/MRI liver observations from 2456 adult participants (mean age, 59 years ± 11 [SD]; 75.3% men and 24.7% women).

Eight of nine AFs favoring malignancy in general were independently associated with HCC and/or malignancy, the exception being fat-sparing in a solid mass. Corona enhancement was associated with malignancy only, and US visibility as a discrete nodule and iron sparing in a solid mass were associated with HCC only. The strongest associations were found for diffusion restriction (OR = 14.5) and mild-moderate T2 hyperintensity (OR = 10.2).

Four of five AFs favoring HCC were independently associated with HCC, with mosaic appearance being also associated with malignancy. Blood products in mass were not associated with HCC, nor malignancy in general. Non-enhancing capsule had the strongest association with HCC (OR = 3.5).

Finally, parallels blood pool enhancement, marked T2 hyperintensity, and hepatobiliary phase isointensity were negatively associated with HCC and malignancy, with the first two having the strongest association with benignity (OR = 0.02 and OR = 0.06, respectively). Iron in mass more than liver was associated with decreased odds of malignancy. Size stability of 2 years or more, size reduction, and undistorted vessels did not show a negative association with HCC or malignancy.

The authors suggest that mandatory incorporation of some AFs could improve the diagnostic performance of LI-RADS, given the strong associations between most AFs and HCC, malignancy, and benignity.


Despite LI-RADS version 2018 assigning equal weight to each ancillary feature (AF), this meta-analysis identified variations in the degree of association between individual AFs and HCC or malignancy. This finding could also implicate a different rationale whenever an observation presents both malignancy and benignity favoring AFs.

The review also demonstrated that some AFs had stronger associations with HCC than some major criteria. For example, non-enhancing capsule (OR = 3.5) had a stronger association with HCC than size and enhancing capsules. Therefore, the authors suggest that upgrading a LI-RADS 4 observation through ancillary features with robust independent associations with HCC could potentially increase the sensitivity of LR-5.

Lastly, seven AFs showed no evidence of association with their predicted outcomes, possibly due to low positivity rates, sample size, and lack of applicability. So, the overall impact of these AFs on the diagnostic efficacy of LI-RADS is probably minimal.

 Regarding limitations, the present study was retrospective, with heterogeneous data, resulting in different sample sizes for each AF. The main results were derived by excluding threshold growth from the major criteria to maximize sample size and robustness of the models. Additionally, many of the included studies were susceptible to bias, either in terms of patient selection or in the flow and timing domain.

In conclusion, Dawtit et al. highlighted data for refining the LI-RADS classification, revealing that most AFs were independently associated with malignancy, HCC, or benignity. Further investigation is necessary to determine the significance of each AF and how these features can be effectively integrated into the LIRADS algorithm.


(1) Vogel A, Meyer T, Sapisochin G, Salem R, Saborowski A. Hepatocellular carcinoma. Lancet 2022;400(10360):1345–1362.

(2) Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer statistics, 2022. CA Cancer J Clin 2022;72(1):7–33.

(3) Philips CA, Rajesh S, Nair DC, Ahamed R, Abduljaleel JK, Augustine P. Hepatocellular Carcinoma in 2021: An Exhaustive Update. Cureus 2021;13(11):e19274.

(4) American College of Radiology Committee on LI-RADS®. LI-RADS CT/MRI v2018. Available at Accessed on March 27, 2024.


Pedro Coroas da Rocha is a second-year Radiology Resident at the "ULS de Santo António" in Porto, Portugal. He completed his undergraduate medical training at the School of Medicine of the University of Minho in Braga. He has a wide range of interests in diagnostic imaging, including abdominal and urogenital radiology.

Comments may be sent to prochao97(at)