Hepatocellular carcinoma (HCC) presents significant challenges in clinical management due to its high recurrence rates and variable prognosis. Key pathologic features, including microvascular invasion (MVI), poorly differentiated pathologic grade (poor PG) and satellite nodules (SNs), contribute to tumour progression and treatment resistance. While histologic examination provides definitive insights, it remains invasive and sometimes incomplete. To address this gap, gadoxetic acid-enhanced MRI offers a noninvasive alternative for preoperative assessment. A newly developed scoring model based on imaging characteristics enables the prediction of high-risk pathologic features and early recurrence, potentially improving treatment planning and patient outcomes.
Development of the MRI-Based Scoring Model
Gadoxetic acid-enhanced MRI was used to assess imaging features in patients with HCC before surgical resection. A retrospective study across three medical centres analysed 366 patients, including a training dataset and two external validation datasets. Imaging characteristics such as tumour size, irregular morphology, intratumoural arteries, peritumoural enhancement in the arterial phase and low peritumoural signal intensity were identified as significant predictors of high-risk pathologic features. Multivariable logistic regression was employed to develop a scoring model, termed the Image Score (I-score), which demonstrated high accuracy in predicting adverse pathologic features. The model was validated externally and exhibited strong predictive performance across different patient cohorts.
Recommended Read: TACE-RFA vs Surgery for Late-Recurrent HCC: A Comparative Analysis
The study identified specific imaging features that correlated with high-risk pathologic attributes. The presence of irregular tumour morphology, a maximum tumour diameter exceeding 4.0 cm and intratumoural arteries was strongly associated with increased risk. Peritumoural enhancement in the arterial phase and low peritumoural signal intensity were also found to be key indicators. These features were incorporated into the I-score, which assigned weighted values based on their predictive significance. In the training dataset, the scoring model achieved an area under the receiver operating characteristic curve (AUC) of 0.93, indicating high diagnostic performance. In external validation datasets, AUC values of 0.86 and 0.84 reinforced the robustness of the model in diverse clinical settings.
Clinical Application in Predicting Early Recurrence
Beyond pathologic prediction, the I-score was evaluated for its prognostic utility in predicting early recurrence within two years of surgery. In an outcome dataset, the I-score was identified as an independent predictor of recurrence, alongside factors such as MVI and postoperative adjuvant transarterial chemoembolisation (PA-TACE). Patients with higher I-scores demonstrated significantly increased recurrence rates. A combined model incorporating the I-score, MVI and PA-TACE status exhibited superior prognostic accuracy compared to individual predictors, suggesting that integrating MRI-based scoring with clinical data enhances risk stratification and treatment decision-making.
Patients classified as high-risk by the I-score exhibited significantly poorer outcomes in terms of recurrence rates. The hazard ratio for early recurrence was 5.2 for patients with elevated I-scores, underscoring the model’s utility in identifying those at greatest risk. The study further demonstrated that patients who did not undergo PA-TACE postoperatively had higher recurrence rates, highlighting the potential role of adjuvant therapy in mitigating disease progression. The integration of MRI-based scoring with clinical factors such as PA-TACE status and MVI could refine therapeutic strategies and aid in optimising post-surgical management.
Advantages and Limitations of the Scoring Model
The MRI-based scoring model offers several advantages, primarily by providing a comprehensive, noninvasive assessment of high-risk pathologic features. The integration of peritumoural signal intensity and enhancement characteristics adds diagnostic value, improving sensitivity and specificity in predicting tumour aggressiveness. Additionally, the model’s ability to predict early recurrence provides clinicians with valuable prognostic insights, enabling more tailored treatment strategies.
However, certain limitations exist. The study’s retrospective nature and variability in MRI acquisition parameters across centres may introduce inconsistencies. Although external validation strengthened the model’s reliability, prospective studies with larger and more diverse patient populations are needed to confirm its broader applicability. Furthermore, the exclusion of patients undergoing MRI with non-specific gadolinium agents limits the generalisability of findings to all HCC cases. Continued refinements incorporating additional imaging features and biomarkers may enhance predictive accuracy in future applications.
The development of a gadoxetic acid-enhanced MRI-based scoring model represents a significant advancement in the preoperative evaluation of hepatocellular carcinoma. By enabling noninvasive prediction of high-risk pathologic features and early recurrence, the I-score provides a valuable tool for clinicians in treatment planning and prognosis assessment. The model’s integration of key imaging indicators allows for improved identification of patients at high risk of recurrence, facilitating timely intervention.
As the clinical landscape of HCC management evolves, the use of imaging-based predictive models could play a pivotal role in guiding surgical and postoperative decisions. Further research and prospective validation are essential to refine and expand the clinical utility of the I-score. With continued advancements, the implementation of noninvasive MRI-based models may significantly improve patient outcomes through personalised therapeutic strategies and more informed clinical decision-making.
Source: Radiology
