Interobserver Variability and Diagnostic Performance of Gadoxetic Acid–enhanced MRI for Predicting Microvascular Invasion in Hepatocellular Carcinoma
Min J.H., Lee M.W., Park H.S., Lee D.H., Park H.J., Lim S., Choi S.-Y., Lee J., Lee J.E., Ha S.Y., Cha D.I., Carriere K.C., Ahn J.H.
Radiology 2020; 297:573–581. https://doi.org/10.1148/radiol.2020201940
Hepatocellular carcinoma (HCC) is the sixth most prevalent neoplasm and the third leading cause of cancer death [1]. The progression of HCC is rapid with high invasiveness, and among the current treatment strategies, radical surgical resection is the preferred treatment strategy in patients with localized disease and preserved hepatic function. Despite optimal surgical resection, the outcome of HCC patients remains poor, and the postoperative 5-year recurrence rate of HCC is approximately 70%, of which two-thirds of all recurrences occur within 2 years after surgery and <35% after liver transplantation [2-4]. Microvascular invasion (MVI) is a histopathologic feature of tumor aggressiveness, observed in 15-57% of HCC, and it is considered as a strong predictor that leads to a high recurrence rate and poor survival rate in HCC patients [5]. An accurate preoperative noninvasive evaluation of the MVI presence can contribute to the optimal treatment strategy and prognosis stratification in patients with HCC based on risk-benefit assessment [6]. Despite several improvements in imaging technique and radiological diagnosis, the preoperative prediction of MVI remains challenging in clinical practice and different imaging features have been proposed as predictors of MVI.
A recent study on the diagnostic performance of gadoxetic acid-enhanced magnetic resonance imaging (MRI) for predicting MVI in patients with diagnosis of HCC has been published by Min et al [7]. In this retrospective analysis, the Authors have evaluated the interobserver agreement and diagnostic performance in the preoperative MRI assessment of the presence of MVI in patients with surgically confirmed HCCs smaller than 5 cm. In their cohort of 100 patients, the most common cause of chronic liver disease was hepatitis B virus (88%) and 51 of 100 patients (51%) had liver cirrhosis. The mean tumor size was 2.8 cm ± 0.9 (62% <3 cm and 38% >3 cm) and MVI have been diagnosed in 39 of the 100 HCCs (39%) at histopathology. Considering the size of the tumor, HCCs >3 cm had a higher frequency of MVI than HCCs measuring 3 cm or less (55% vs 29%, respectively).
Imaging analysis was performed by eight independent fellowship-trained radiologists. Different imaging features for the prediction of MVI have been assessed, and among these nonsmooth tumor margin showed the highest sensitivity (38-90%) and peritumoral arterial phase hyperenhancement showed the highest specificity (84-98%) for all readers. The specificity (95-100%) and positive predictive value (67-100%) were higher compared with other combinations of imaging features for all reviewers (irregular rim-like enhancement in the arterial phase, peritumoral hepatobiliary phase hypointensity). The analysis revealed a fair-to-moderate overall interobserver agreement (k range: 0.24-0.41) regarding each MVI imaging feature or their combination and MVI probability between observers.
Diagnostic performance of each reader was modest for MVI prediction (area under the receiver operating characteristic curve [AUC] range, 0.60–0.74). The sensitivity for the diagnosis of MVI ranged from 15% to 69%, and the specificity ranged from 57% to 92%. Interestingly, when considering the size subgroup, the interobserver agreement for MVI probability was poor-to-fair and there were no differences in interobserver agreement for MVI according to the readers’ experience with regard to HCCs measuring 3 cm or less, while for HCCs larger than 3 cm, five more experienced readers showed higher sensitivity for the diagnosis of MVI. In the MVI probability categorization, more experienced readers showed higher agreement than less experienced readers.
According to recent studies, a considerable interobserver variability exists in the assessment of MVI in HCC imaged with MRI, even for more experienced radiologists. Despite evidences has reported good interobserver agreement for the preoperative prediction on MVI with MRI, in particular regarding the nonsmooth tumor margin peritumoral arterial phase hyperenhancement, peritumoral HBP hypointensity [8], and irregular rim-like enhancement in the arterial phase [9], currently the preoperative diagnostic performance of imaging features for prediction of MVI is still highly variable.
Therefore, other quantitative imaging analysis have been explored. Recent studies have reported that use of artificial intelligence and automated computerized image analysis, including radiomics, could improve the preoperative prediction of MVI and overcome the limits of the subjective assessment of MVI, but still need to be implemented and integrated in the clinical practice [10,11].
In conclusion, this study [7] reports a considerable interobserver variability and a limited diagnostic performance in the MRI assessment of MVI in HCC, even by experienced radiologists. More standardized imaging criteria and further researches evaluating the diagnostic performance of MVI imaging features and their combinations are warranted, in order to provide valuable information to guide a more objective clinical decision-making and appropriate therapeutic strategy for patients with HCC.
References:
- Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin (2018);68:394–424.
- Kardashian A, Florman SS, Haydel B et al. Liver transplantation outcomes in a U.S. multicenter cohort of 789 patients with hepatocellular carcinoma presenting beyond Milan criteria. Hepatology (2020).
- Vogel A, Cervantes A, Chau I et al. Hepatocellular carcinoma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol (2018);29:238–255
- Mazzaferro V, Sposito C, Zhou J et al (2018) Metroticket 2.0 Model for analysis of competing risks of death after liver transplantation for hepatocellular carcinoma. Gastroenterology (2018);154:128–139.
- Chong, HH., Yang, L., Sheng, RF. et al. Multi-scale and multi-parametric radiomics of gadoxetate disodium–enhanced MRI predicts microvascular invasion and outcome in patients with solitary hepatocellular carcinoma ≤ 5 cm. Eur Radiol (2021).
- Huang J, Tian W, Zhang L, et al. Preoperative Prediction Power of Imaging Methods for Microvascular Invasion in Hepatocellular Carcinoma: A Systemic Review and Meta-Analysis. Front Oncol (2020);10:887.
- Min JH, Lee MW, Park HS, Lee DH, Park HJ, Lim S et al. Interobserver Variability and Diagnostic Performance of Gadoxetic Acid-enhanced MRI for Predicting Microvascular Invasion in Hepatocellular Carcinoma. Radiology (2020) 297:3,573-581.
- Lee S, Kim SH, Lee JE, Sinn DH, Park CK. Preoperative gadoxetic acid-enhanced MRI for predicting microvascular invasion in patients with single hepatocellular carcinoma. J Hepatol (2017);67(3):526–534.
- Rhee H, An C, Kim HY, Yoo JE, Park YN, Kim MJ. Hepatocellular Carcinoma with Irregular Rim-Like Arterial Phase Hyperenhancement: More Aggressive Pathologic Features. Liver Cancer (2019);8(1):24–40.
- Ma X, Wei J, Gu D, et al. Preoperative radiomics nomogram for microvascular invasion prediction in hepatocellular carcinoma using contrast-enhanced CT. Eur Radiol 2019;29(7):3595–3605.
- He M, Zhang P, Ma X, He B, Fang C, Jia F. Radiomic Feature-Based Predictive Model for Microvascular Invasion in Patients With Hepatocellular Carcinoma. Front Oncol. 2020 Nov 5;10:574228.
Dr. Francesco Matteini is a second-year young radiology resident at the University of Palermo (Italy) and an ESGAR member. Dr. Matteini graduated at the University of Palermo in October 2018. He is actually involved in scientific researches centered on hepatobiliary topics, and specifically on CT and MR imaging of focal liver lesions.
Comments may be sent to