Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma

Microvascular invasion (MVI) in hepatocellular carcinoma (HCC) poses significant challenges in the treatment and management of patients. MVI is a histological feature indicative of tumour cells infiltrating blood vessels near the primary tumour. This feature is closely associated with increased recurrence rates and poorer outcomes post-surgery. Therefore, it is essential to develop reliable preoperative predictive models to guide treatment decisions and improve patient prognosis. A recent study published in Insights into Imaging discusses a comprehensive MVI prediction strategy incorporating conventional ultrasound (US), Sonazoid-enhanced contrast-enhanced ultrasound (CEUS) and biochemical indicators.

Conventional Ultrasound and MVI Indicators

Conventional US is a fundamental tool for evaluating liver tumours, providing information on tumour morphology and vascularity. Features such as an obscure tumour boundary and the presence of an intratumoural artery have been independently associated with MVI. An obscure boundary on B-mode US suggests invasive growth patterns that disrupt the normal liver parenchyma, a characteristic often correlated with aggressive tumour behaviour. Additionally, intratumoural arterial flow detected using pulsed Doppler US is indicative of increased angiogenesis, a process that supports tumour growth and metastasis. While these indicators offer valuable information, their standalone predictive power is limited due to moderate specificity.

Recent studies have demonstrated that the presence of an intratumoural artery, a hallmark of tumour neovascularisation, correlates with macrotrabecular-massive subtypes of HCC, known for their poor prognosis. Detecting these features through conventional US provides an accessible and non-invasive method for initial MVI risk stratification. However, incorporating advanced imaging modalities and biochemical markers is necessary to improve the specificity and sensitivity of predictions.

Enhanced Predictive Power with Sonazoid-CEUS

Sonazoid-enhanced CEUS represents a significant advancement in liver imaging, providing detailed real-time assessment of tumour vascular phases. The unique properties of Sonazoid, including its Kupffer-phase imaging, make it particularly valuable for assessing liver lesions. Kupffer cells are liver-resident macrophages that uptake Sonazoid microbubbles, allowing clear differentiation between healthy liver tissue and tumours. The complete clearance of Sonazoid contrast in the Kupffer phase has been identified as a crucial indicator of MVI. This feature suggests impaired Kupffer cell function, associated with tumour progression and an immunosuppressive microenvironment.

The predictive power of CEUS in determining MVI stems from its ability to reveal subtle changes in tumour perfusion and clearance. The dynamic imaging provided by Sonazoid-CEUS captures these characteristics, which may not be evident in conventional US alone. This combination enhances the accuracy of preoperative MVI assessment, facilitating more targeted treatment plans.

Adding Kupffer-phase observations to the predictive model significantly boosts the overall performance, as evidenced by a substantial increase in area under the curve (AUC) metrics during validation. This improvement demonstrates that while conventional US highlights basic structural and vascular features, Sonazoid-CEUS offers deeper insights into tumour behaviour and surrounding tissue interaction.

Biochemical Indicators and Model Integration

Biochemical markers have long been a cornerstone of tumour diagnostics. Alpha-fetoprotein (AFP) and protein induced by vitamin K absence-II (PIVKA-II) are two commonly used markers in liver cancer assessment. Among these, PIVKA-II has shown superior specificity for predicting MVI. Elevated levels of PIVKA-II are linked to tumour growth driven by p53 pathway alterations and are associated with the aggressive nature of certain HCC subtypes.

Integrating PIVKA-II with imaging findings refines the predictive model, addressing the limitations of imaging and biochemical assessments when used independently. In the context of MVI prediction, PIVKA-II complements the imaging data by indicating the biological behaviour of the tumour. A study involving patients with single HCC ≤ 5 cm demonstrated that a combined model using conventional US, CEUS features and PIVKA-II levels outperformed models relying solely on either modality. This model achieved AUC values above 0.90 in both training and validation cohorts, underscoring its robustness and clinical relevance.

The combined model’s balanced sensitivity and specificity highlight its potential for clinical adoption. It provides a practical, non-invasive means for early MVI detection, allowing for timely and aggressive treatment strategies, such as wide-margin resections or adjunct therapies to improve patient survival rates. The integration of multimodal data ensures that predictions are not only accurate but also reflective of different aspects of tumour pathology.

Preoperative identification of microvascular invasion in HCC is crucial for optimising surgical outcomes and reducing recurrence risk. The combination of conventional ultrasound, Sonazoid-enhanced CEUS and biochemical markers such as PIVKA-II presents a comprehensive, non-invasive strategy for MVI prediction. This model's high predictive accuracy offers clinicians a reliable tool for tailoring treatment plans to patient-specific risks, ultimately improving long-term survival. Further research and external validation will bolster the utility of this approach, paving the way for its integration into standard preoperative evaluation protocols.

Source: Insights into Imaging

Image Credit: iStock