Abstract Background Hepatocellular carcinoma (HCC) is the most common primary liver cancer and a major cause of global morbidity and mortality (Tang et al. in Abdom Radiol 43:13–25, 2018). Transarterial chemoembolization (TACE) is a key treatment for unresectable HCC, improving survival and serving as a bridge to transplantation or surgery (Singal et al. in J Hepatol 72:250–261, 2020). Accurate imaging-based assessment of tumor response after TACE is critical for prognosis, treatment planning, and therapeutic decision making (Singal et al. in J Hepatol 72:250–261, 2020). LIRADS treatment response (TR) criteria are a standardized system used to assess the effectiveness of loco-regional therapies (like ablation or chemoembolization) on hepatocellular carcinoma (HCC) during follow-up imaging. Their clinical significance lies in providing clear, consistent, and reproducible terminology (Granata V, Grassi R, Fusco R, Setola SV, Belli A, Piccirillo M et al. in Int J Environ Res Public Health 18(7):3598, 2021). Methods This prospective study included 61 patients, with ages ranging from 43 to 71 years (mean age 61 years); the cohort consisted of 47 male patients and 14 female patients with 111 HCC lesions treated with TACE at the National Cancer Institute over a 12 month period. MRI evaluations within 1 month post-TACE used dynamic contrast-enhanced sequences, diffusion-weighted imaging (DWI), and subtraction imaging. Lesions were assessed for viability using LIRADS treatment response (LR-TR) criteria by three readers interpreting different protocols: reader (1) full MRI protocol, reader (2) abbreviated dynamic MRI, and reader (3) DWI with ADC maps. Statistical analyses included ROC curve evaluation for ADC cutoff values and inter-reader agreement using Cohen’s kappa ( k ). Results The reference standard full MRI protocol showed the highest diagnostic accuracy (94%), followed by abbreviated dynamic MRI (91%) and DWI with ADC maps (89%). An ADC cutoff of 0.95 × 10⁻ 3 mm 2 /s provided 88% sensitivity and 85% specificity for identifying viable lesions. Substantial to near-perfect inter-reader agreement was observed ( k values: 0.78 for reference standard full MRI protocol vs. dynamic MRI, 0.72 for reference standard full MRI vs. DWI). In a comparison between the reference standard full MRI protocol and each abbreviated protocol regarding the k agreement, there was considerable agreement between the abbreviated protocol (Subtraction) and the reference standard full MRI protocol with k = 0.982 ( p < 0.001). Conclusions Subtraction imaging improved diagnostic accuracy, sensitivity, and specificity, yielding comparable results to the reference standard full MRI protocol. It outperformed the abbreviated dynamic MRI and DWI providing higher reader confidence and agreement. Further studies are required to validate these findings and optimize MRI protocols for post-TACE lesion evaluation.
Elnafrawy et al. (Thu,) studied this question.