Abstract
• Four contrast-enhanced MRI features were significantly associated with cholangiocyte phenotype HCC.
• A noninvasive cholangiocyte phenotype HCC predictive model was established based on MRI features.
• Patients with cholangiocyte phenotype HCC demonstrated a worse prognosis following hepatic resection.
Introduction
Hepatocellular carcinoma (HCC) is the most prevalent primary liver cancer, yet it exhibits significant biological heterogeneity. A particularly aggressive subset, known as the "cholangiocyte phenotype" HCC, is characterized by the expression of biliary markers such as CK7 or CK19. These tumors are associated with rapid growth, increased metastatic potential, and resistance to standard systemic therapies like sorafenib.
Currently, definitive identification of this phenotype requires invasive procedures like needle biopsy or post-operative pathological analysis. However, biopsies are prone to sampling errors due to tumor heterogeneity. This study aims to develop a noninvasive, preoperative method using contrast-enhanced MRI to identify these high-risk patients, thereby facilitating better surgical planning and personalized management.
Methods
This retrospective study analyzed 334 patients with solitary HCC from West China Hospital (2010–2022). The cohort was divided into a training set (n=232) and an independent validation set (n=102). Two radiologists independently assessed 24 MRI features.
The researchers employed logistic regression to identify significant predictors, leading to the development of the RNNN score. This simplified scoring system is based on four key imaging features:
- Rim arterial phase hyperenhancement (Rim APHE)
- Nodule-in-nodule architecture
- Non-smooth tumor margin
- Non-peripheral washout
Five-fold cross-validation was used to ensure model robustness, and Kaplan-Meier curves were utilized for survival analysis to compare Recurrence-Free Survival (RFS) and Overall Survival (OS).
Results
The RNNN score demonstrated strong predictive performance, achieving an Area Under the Curve (AUC) of 0.76 in the training set and 0.73 in the validation set. A score of 5.5 points was identified as the optimal threshold for high-risk classification.
Prognostic Implications
The model proved to be a powerful prognostic tool. Patients predicted to have the cholangiocyte phenotype exhibited significantly poorer outcomes:
- Median RFS: 926 days (predicted positive) vs. 1565 days (predicted negative).
- Median OS: 1504 days (predicted positive) vs. 2960 days (predicted negative).
Key Imaging Predictors
- Rim APHE: Identified as the strongest predictor of the aggressive phenotype.
- Nodule-in-Nodule: This feature likely indicates a "stepwise" evolutionary process where a sub-clone within the tumor acquires more malignant characteristics.
- Non-peripheral Washout: Interestingly, this feature acted as a protective factor, decreasing the likelihood of the cholangiocyte phenotype.
Discussion
The study highlights that macro-level MRI features can serve as reliable proxies for micro-level biological "stemness" and aggressive pathology. By identifying "red flags" like Rim APHE preoperatively, clinicians can anticipate a higher risk of recurrence.
However, several limitations remain. The single-center, retrospective design necessitates further external validation in diverse populations. Additionally, the combined definition of the phenotype (CK7 or CK19) might overlook subtle differences between these markers. Future research should explore whether patients identified via the RNNN score benefit from specific adjuvant therapies, such as immunotherapy, following resection.
Conclusion
The MRI-based RNNN score provides a noninvasive and effective means of predicting the aggressive cholangiocyte phenotype in HCC patients before surgery. This tool holds significant potential for improving prognostic stratification and guiding personalized treatment strategies in clinical practice.
Summary updated