Radiomics and molecular analysis: Bridging the gap for predicting hepatocellular carcinoma prognosis

Abstract

This editorial examines a recent study that used radiomics based on computed tomography (CT) to predict the expression of the fibroblast-related gene enhancer of zeste homolog 2 (EZH2) and its correlation with the survival of patients with hepatocellular carcinoma (HCC). By integrating radiomics with molecular analysis, the study presented a strategy for accurately predicting the expression of EZH2 from CT scans. The findings demonstrated a strong link between the radiomics model, EZH2 expression, and patient prognosis. This noninvasive approach provides valuable insights into the therapeutic management of HCC.

Key Words: Hepatocellular carcinoma; Computed tomography; Radiomics; Enhancer of zeste homologue 2 expression; Non-invasive imaging

Core Tip: This editorial discusses a study that combined radiomics based on computed tomography (CT) with molecular analysis for predicting the expression of the gene enhancer of zeste homolog 2 (EZH2) in patients with hepatocellular carcinoma (HCC). The radiomics model revealed a strong correlation with both EZH2 expression and patient survival, providing a noninvasive approach for obtaining key molecular insights. This innovative approach may enhance the prediction of HCC prognosis and may guide the implementation of personalized treatment, marking a major advancement in cancer diagnostics and precision medicine.

TO THE EDITOR

The evolving field of hepatocellular carcinoma (HCC) research is experiencing a shift with the convergence of radiomics and molecular analysis. A recent study by Yu et al[1], published in the World Journal of Clinical Cases, titled “Computed tomography based radiomics predicts the fibroblast-related gene EZH2 expression level and survival of hepatocellular carcinoma”, exemplifies this innovative approach. This pioneering research highlights the predictive capacity of computed tomography (CT)-based radiomics to estimate the expression of the fibroblast-related gene enhancer of zeste homolog 2 (EZH2) and its correlation with survival outcomes in HCC patients. Radiomics, which focuses on extracting quantitative features from medical images, is gaining traction as a powerful, non-invasive tool for characterizing tumor phenotypes[2]. In their study, Yu et al[1] explored how CT-based radiomics could predict EZH2 expression levels in HCC, traditionally assessed through invasive methods such as biopsies. By analyzing CT images from HCC patients, the researchers extracted high-dimensional radiomics features that were strongly associated with EZH2 expression, a gene known for its role in epigenetic regulation and cancer progression. EZH2 is implicated in driving tumor aggressiveness and is linked to poorer outcomes across several cancer types, including HCC (Figure 1). Using machine learning techniques, the study developed a model to predict EZH2 expression based on radiomics features. This model, validated with robust statistical metrics, demonstrated its ability to reliably differentiate between high and low EZH2 expression groups. Crucially, the study found a significant correlation between predicted EZH2 expression and overall survival in HCC patients — those with higher EZH2 expression, as predicted by the model, exhibited worse survival outcomes. This reinforces EZH2 as a vital prognostic marker, particularly in HCC, where its expression is associated with aggressive tumor behavior and poorer prognosis[3]. One of the standout aspects of this study is its demonstration of radiomics as a non-invasive alternative to molecular profiling. While biopsy-based methods remain effective, they are invasive and not always feasible. CT-based radiomics offers a significant advantage, allowing clinicians to derive similar insights into tumor biology from routine imaging. This could fundamentally change how HCC is managed, offering an alternative for predicting patient prognosis and guiding treatment decisions without needing invasive procedures[1]. The ability to integrate radiomics into clinical practice opens new doors for personalized medicine in HCC. By merging imaging data with molecular insights, clinicians can better understand the tumor phenotype, leading to more precise prognostication and individualized treatment plans. Also, this approach facilitates non-invasive monitoring of disease progression and response to therapy, ultimately enhancing patient outcomes[4]. This study builds on existing research that has demonstrated the utility of radiomics across various cancers, including HCC, where it has been used to predict microvascular invasion and to provide prognostic insights following radical resection[5]. The absence of intra-tumoral tertiary lymphoid structures has also been linked to worse prognosis and mTOR signaling activation in HCC[6], further underscoring the importance of identifying molecular and cellular markers that can inform treatment decisions. By showing that CT-based radiomics can reliably predict EZH2 expression and patient survival, Yu et al[1] add critical evidence to the potential of radiomics to serve as a prognostic tool in HCC. In sum, the integration of radiomics with molecular analysis represents a major leap forward in HCC management. This non-invasive assessment of tumor biology enhances prognostic accuracy and has the potential to guide personalized treatment strategies. As radiomics continues to evolve, its application in clinical practice promises to revolutionize cancer diagnostics and patient care. Yu et al[1] have made an invaluable contribution, demonstrating the transformative potential of radiomics in enhancing the precision and effectiveness of HCC treatment.

Figure 1 Mammalian target of rapamycin signaling pathway and its interaction with EZH2 expression in hepatocellular carcinoma: Mechanisms and implications. Illustration of the mammalian target of rapamycin (mTOR) signaling pathway and its interaction with EZH2 expression in hepatocellular carcinoma (HCC). The figure highlights key components of the mTOR pathway, including upstream regulators (e.g., PI3K/AKT) and downstream effectors that influence cell proliferation, survival, and metabolism. EZH2, a critical epigenetic regulator, is shown to be modulated by mTOR signaling, contributing to tumor progression, invasion, and poor prognosis in HCC. The interplay between mTOR activation and elevated EZH2 expression suggests a mechanistic link that promotes cancer aggressiveness, making these pathways potential therapeutic targets. Arrows indicate signaling cascades and molecular interactions involved in HCC development. HCC: Hepatocellular carcinoma; IRS: Insulin receptor substrate; PI3K: Phosphoinositide 3-kinase; PTEN: Phosphatase and tensin homologue; PDK1: 3-Phosphoinositide-dependent kinase 1; AKT: Protein kinase B; mTOR: Mammalian target of rapamycin; EZH2: Enhancer of zeste homolog 2.

THE ROLE OF RADIOMICS: A TRANSFORMATIVE APPROACH IN HCC MANAGEMENT

Radiomics involves extracting large amounts of quantitative features from medical images to decode the tumor phenotype. By leveraging advanced algorithms and machine learning techniques, this approach provides a non-invasive alternative to traditional molecular profiling, offering critical insights that influence treatment decisions and patient monitoring[7]. In HCC, radiomics has demonstrated significant potential in providing molecular information that can improve patient outcomes. The recent study by Yu et al[1] underscores the impact of radiomics in predicting the expression of EZH2 and correlating this with patient survival in HCC cases. By analyzing CT images, the researchers developed a radiomics model that accurately predicts EZH2 expression, which is crucial given its role in HCC development and prognosis[3]. Radiomics has proven its utility in various aspects of HCC management. For instance, Xia et al[5] demonstrated that a CT-based radiomics model could predict microvascular invasion, a key factor influencing prognosis. Gu et al[4] took this further by developing a multi-view radiomics feature fusion model that revealed distinct immuno-oncological characteristics and clinical outcomes in HCC, highlighting the versatility of radiomics in clinical settings. The integration of radiomics with other clinical and molecular data enhances prognostic accuracy and informs personalized treatment strategies. For example, Xie et al[2] introduced a clinical-radiomic-pathomic model for prognosis prediction in HCC patients post-radical resection, demonstrating how combining diverse data types can provide a more comprehensive understanding of the tumor microenvironment and improve patient outcomes. Radiomics also offers a significant advantage over traditional biopsy methods, which are invasive and sometimes impractical. Its non-invasive nature enhances patient comfort and broadens the scope of diagnostic and prognostic assessments. Liu et al[8] demonstrated the power of radiomics in optimizing treatment strategies by using a radiological-clinical nomogram to predict the benefits of sequential therapy in patients with unresectable HCC. Overall, radiomics represents a transformative tool in HCC management, delivering valuable molecular insights through non-invasive imaging techniques. As the field continues to evolve, the integration of radiomics with clinical and molecular data promises to revolutionize cancer diagnostics and care, paving the way for more precise and personalized treatment strategies.

IMPLICATIONS FOR CLINICAL PRACTICE

The study by Yu et al[1] illustrates the transformative potential of radiomics in enhancing HCC prognosis and informing personalized treatment plans. However, several limitations of this study warrant consideration to provide a comprehensive view of its potential shortcomings and areas for improvement. One of the primary limitations is the sample size. Although the authors included enough patients to demonstrate the radiomics model’s predictive capability, a larger cohort would improve the robustness and generalizability of the findings. Expanding the sample size could validate the model across diverse populations and different stages of HCC, ensuring its applicability to a broader patient demographic. The study’s retrospective design presents another limitation. Retrospective analyses are vulnerable to biases, including selection bias, as the patients included may not represent the entire population of HCC patients. Additionally, relying on pre-existing CT images may limit the study to specific imaging protocols, potentially affecting the reproducibility of the results in different clinical settings or with various imaging equipment. The methodological approach to feature extraction and model development also raises concerns. While advanced machine learning techniques were used, the choice of radiomics features and algorithms significantly influences the outcomes. If the selected features do not sufficiently capture tumor heterogeneity or biological variability in EZH2 expression, the predictive model may lack precision. Future studies should consider using a wider range of features and employing more rigorous validation techniques to strengthen the model. Interobserver variability in CT image interpretation is another potential limitation. Different radiologists may interpret the same images differently, leading to variability in radiomics feature extraction. Addressing this issue by standardizing imaging protocols and accounting for interobserver variability could improve the reliability of the results. Moreover, the biological validation of the radiomics model could be expanded. While the study correlates radiomics features with EZH2 expression, further exploration into the biological mechanisms connecting radiomics data to EZH2 expression could provide deeper insights. Integrating histological data, such as fibroblast presence or immune cell infiltration, could enrich the understanding of how these features relate to tumor biology and patient outcomes. The study also demonstrates a correlation between EZH2 expression and survival, but it does not explore the functional implications of this relationship. Investigating how EZH2 influences treatment response or resistance in HCC could offer valuable information for clinical practice, guiding more tailored treatment approaches and improving patient management. Generally, while the study by Yu et al[1] provides important insights into the potential of CT-based radiomics for predicting EZH2 expression in HCC, addressing these limitations in future research will be critical to improving the robustness, applicability, and clinical utility of radiomics in oncology.

In addition, radiomics offers a non-invasive alternative to traditional molecular profiling. The study by Yu et al[1] highlights how a radiomics model can predict the expression of the fibroblast-related gene EZH2, which is associated with cancer progression and patient survival in HCC. The ability to derive such critical molecular insights from CT images represents a significant advancement in cancer diagnostics and patient management. One of the key implications of this study is the potential for radiomics to be seamlessly integrated into clinical workflows. By providing additional molecular insights, radiomics can help clinicians develop more personalized treatment plans, thereby improving patient outcomes. This non-invasive approach also reduces the need for traditional biopsy methods, which, while effective, are often invasive and not always feasible. Radiomics can offer similar or even superior prognostic information, enhancing patient comfort and compliance. Furthermore, Yu et al[1] pave the way for further research into the integration of radiomics with other molecular markers in HCC and other cancers. Radiomics’ potential goes beyond predicting EZH2 expression; it can also monitor treatment response and disease progression. This is particularly valuable in HCC, where early and accurate prediction of patient outcomes is critical. Previous studies have demonstrated radiomics’ utility in various cancers, including predicting microvascular invasion in HCC[5] and prognosis following radical resection[2]. Future research should focus on validating these findings in larger, multi-center cohorts to ensure their robustness and generalizability. Additionally, exploring radiomics’ application in monitoring treatment response and disease progression presents a promising avenue for future studies. The integration of radiomics with clinical and molecular data will provide a more comprehensive understanding of the tumor microenvironment and patient prognosis[4]. Briefly, the integration of radiomics with clinical and molecular data represents a major advancement in HCC management. Yu et al[1] demonstrate the potential for radiomics to revolutionize cancer diagnostics and patient care by offering valuable molecular insights through non-invasive imaging techniques. As research continues to validate and refine these models, radiomics is poised to play an increasingly central role in personalized cancer treatment and patient management, ultimately improving clinical outcomes and the quality of life for patients with HCC and other cancers.

CONCLUSION

The integration of CT-based radiomics with molecular analysis marks significant progress in HCC research. This innovative approach, which predicts EZH2 expression levels and patient survival, offers a promising non-invasive tool for improving cancer prognosis and guiding personalized treatment strategies. As this field continues to evolve, the potential for radiomics to transform clinical practice and enhance patient outcomes becomes increasingly evident.