Diagnostic assessment of MRI with contrast dynamic liver using gadoxetic acid disodium (GD-EOB- DTPA) in the discrimination of hepatocellular carcinoma and non-hepatocellular carcinoma lesions according to 2018 lirads

Aim: In our study, we examined the diagnostic value of 2018 LI-RADS reporting system together with a diagnostic nomogram in the differentiation of HCC and non- HCC masses in contrast-enhanced dynamic liver MRI using the gadoxetate disodium (Gd-EOB-DTPA) which is a hepatocyte-specific agent. Materials and methods: Our study retrospectively started as a single-center after the approval of the Ethics Committee of the Faculty of Medicine of Zonguldak Bülent Ecevit University. Between January 2018 and August 2021, 788 liver contrast- enhanced dynamic MRI examinations obtained using the gadoxetic acid disodium (Gd-EOB-DTPA) were analyzed. 373 patients who did not meet the 2018 LI-RADS criteria were excluded from the study. The liver MRIs of the remaining 415 patients were examined by two observers and evaluated according to the 2018 LI-RADS reporting system. The cases were divided into 5 different groups according to the LI- RADS reporting system: patients with no hepatic mass and signs of chronic liver parenchymal disease were classified as LI-RADS 1, and patients with hepatic masses as LI-RAD 2-5. The conformity of the quantitative variables to the normal distribution was checked with histograms and Q-Q plots. These variables were summarized as mean ± standard deviation and median (smallest value to largest value), while qualitative variables were summarized as frequency (percentage). Quantitative variables suitable for normal distribution were evaluated with independent samples t- test or one-way analysis of variance (ANOVA) according to the number of groups, whereas quantitative variables not in accordance with the normal distribution were analyzed according to the number of groups with the Mann Whitney U test or the Kruskal Wallis test. Also, qualitative variables were analyzed with the Chi-square test or Fisher's exact test. Bonferroni correction was used for pairwise comparisons. The success of the variants in distinguishing between HCC and non-HCC was evaluated by ROC (receiver operating characteristic) curve analysis. Prospective selective multiple logistic regression analysis was performed to determine the most effective variables in determining HCC and p<0.05 values were considered statistically significant. Diagnostic nomogram was obtained by the statistical RStudio program, Durbin Watson and p tests. Results: In this study, MRIs of 788 patinets were analyzed (407 males and 381 females), mean age was 58.44 ± 13.52 (range 19 - 92). 373 (47.3%) patients were LI- RADS negative, 260 (33.0%) patients LI-RADS 1, 77 (9.8%) patients LI-RADS 2, 14 (1.8%) patients LI-RADS 3, 31 (3.9%) patients classified as LI-RADS 4 and 33 (4.2%) patients as LI-RADS 5. The study took into account 155 patients (108 males and 47 females) classified as LI-RADS 2-5. The mean age for these patients was 60.90 ± 12.64 (range 22 to 89). There was only one lesion in 108 (69.7%) patients, and their size was greater than 20 mm in 72 (46.5%) patients. 77 (49.7%) patients LI-RADS 2, 14 (9.0%) patients LI-RADS 3, 31 (20.0%) patients LI-RADS 4, and 33 (21.3%) patients LI- RADS It is classified as 5. The diagnosis was made according to MRI-clinical findings in 137 (88.4%) patients, whereas it was established with pathology in 18 (11.6%) patients. 90 (58.1%) patients were diagnosed as non-HCC mass, 6 (3.9%) patients as non-HCC malignancy, and 59 (38.1%) patients as HCC. LI-RADS was found to have the highest area under the ROC curve and the ideal cut-off point was determined as >3.5. LI-RADS also had the highest sensitivity, correct classification rate, and negative predictive value. Capsule enhancement had the highest selectivity and positive predictive value. The ideal cut-off value for AFP was >15 ng/mL and the ideal cut-off value for size was >20 mm. Prospective selective multiple logistic regression analysis was performed to determine the most successful variables in HCC. These variables were found as diffusion restriction, arterio-portal (A-P) wash-out, and LI-RADS 4&5. The risk of HCC was found to be 13.461 times higher in patients with diffusion restriction (OR: 13.461; 95% CI: 2.215 - 81.802; p=0.005) and was observed to be 8.934 times higher in patients with A-P wash-out (OR: 8.934; 95% CI: 1.306 - 61.100; p=0.026). Also, this risk in LI-RADS 4&5 patients was seen to be 138.980 times higher (OR: 138.980; 95% CI: 13.606 - 1419.605; p<0.001). In order to determine HCC, the most effective independent variables (diffusion restriction, A-P wash-out, LI-RADS 4&5) were considered in RStudio program, and a diagnostic nomogram was obtained. HCC risk was determined by the total score found in the nomogram. Conclusions: In the evaluation of liver MRI enhanced with Gd-EOB-DTPA as hepatospecific contrast material, the specificity and sensitivity of 2018 LI-RADS criteria is high. The correlation of dynamic contrasted liver MRI with follow-up and AFP is important in patients with chronic liver parenchyma disease. Using a diagnostic nomogram in determining HCC, both wash-out and diffusion restriction positivity contributed significantly to the LI-RADS classification in our study. Keywords: Hepatocellular carcinoma, HCC, LI-RADS, cirrhosis