Impact of LI-RADS CT and MRI Ancillary Features on Diagnostic Performance: An Individual Participant Data Meta-Analysis. Journal Articles uri icon

  •  
  • Overview
  •  
  • Research
  •  
  • Identity
  •  
  • Additional Document Info
  •  
  • View All
  •  

abstract

  • Background A recent meta-analysis showed independent associations between most Liver Imaging Reporting and Data System (LI-RADS) ancillary features (AFs) and hepatocellular carcinoma (HCC), malignancy, and benignity. However, the impact of AFs on the diagnostic performance of LI-RADS remains unclear. Purpose To evaluate the impact of applying individual AFs on the diagnostic performance of CT and MRI LI-RADS using an individual participant data (IPD) meta-analysis. Materials and Methods Databases were searched for studies published from January 2014 to February 2023 that evaluated the diagnostic accuracy of CT and MRI for HCC in adults at risk for HCC using LI-RADS version 2014, 2017, or 2018. Observations were categorized according to LI-RADS major features, applying threshold growth when available, and excluding those previously treated or not meeting the composite reference standard (histopathologic analysis or imaging). Using a one-step approach, the IPD were pooled via bivariate mixed-effects models, accounting for clustering in participant-level and study-level random effects. The area under the receiver operator characteristic curve (AUC) for LI-RADS categories 1-5 and the positive predictive value (PPV), sensitivity, and specificity for LI-RADS category 5 (LR-5) observations were derived using three strategies: (a) major features only; (b) major features with each individual AF applied; and (c) similar to strategy 2 but allowing AFs favoring HCC in particular or malignancy in general to upgrade category LR-4 to category LR-5 when present. Comparisons were made using two-tailed z tests. Risk of bias was assessed using Quality Assessment of Diagnostic Accuracy Studies 2. Results Forty-six studies comprising 9257 observations (1098 CT, 8159 MRI) in 7811 adult participants (6792 male; mean age, 58.7 years ± 10.7 [SD]) were included. For all AFs, there were no differences in AUCs of LI-RADS categories 1-5 among strategies 1-3 (P value range, .65 to >.99). For category LR-5, there were also no differences among strategies 1-3 in the PPV, sensitivity, and specificity (P value range, .11 to >.99). Sensitivity analysis of only low-risk bias studies (nine of 46) yielded results consistent with primary analysis. Conclusion The application of individual AFs did not impact the overall diagnostic performance of CT and MRI LI-RADS compared with major features alone. © RSNA, 2025 Supplemental material is available for this article. See also the editorial by Searleman in this issue.

authors

  • Abedrabbo, Nicole
  • Lam, Eric T
  • McInnes, Matthew DF
  • Dawit, Haben
  • Kadi, Diana
  • Van Der Pol, Christian Balth
  • Salameh, Jean-Paul
  • Levis, Brooke
  • Naringrekar, Haresh
  • Lerner, Emily
  • Adamo, Robert G
  • Alabousi, Mostafa
  • Polikoff, Adam
  • Furlan, Alessandro
  • Tang, An
  • Kierans, Andrea S
  • Singal, Amit G
  • Arvind, Ashwini
  • Alhasan, Ayman
  • Song, Bin
  • Allen, Brian C
  • Reiner, Caecilia S
  • Clarke, Christopher
  • Ludwig, Daniel R
  • Diaz Telli, Federico
  • Piñero, Federico
  • Rosiak, Grzegorz
  • Jiang, Hanyu
  • Kwon, Heejin
  • Wei, Hong
  • Kang, Hyo-Jin
  • Joo, Ijin
  • Hwang, Jeong Ah
  • Min, Ji Hye
  • Song, Ji Soo
  • Wang, Jin
  • Podgórska, Joanna
  • Eisenbrey, John R
  • Bartnik, Krzysztof
  • Chen, Li-Da
  • Ronot, Maxime
  • Cerny, Milena
  • Seo, Nieun
  • Rao, Sheng-Xiang
  • Cannella, Roberto
  • Choi, Sang Hyun
  • Kim, So Yeon
  • Fraum, Tyler J
  • Wang, Wentao
  • Jeong, Woo Kyoung
  • Jing, Xiang
  • Kim, Yeun-Yoon
  • Kang, Zhen
  • Bashir, Mustafa R
  • Costa, Andreu F

publication date

  • July 2025