Glioblastomas and Temporalis Muscle

Glioblastoma (GBM) is recognized as the most prevalent and aggressive primary brain tumor, presenting significant challenges for both clinicians and researchers.1 In the United States, the incidence of GBM is approximately 3.19 cases per 100,000 people, accounting for about 15% of all brain tumors.2 Unfortunately, patients diagnosed with GBM often face a dismal prognosis, with median survival typically extending just over a year, even with aggressive treatment regimens.1 The complexity of GBM—characterized by its heterogeneous nature3 and diverse molecular subtypes—complicates effective treatment and underscores the urgent need for robust prognostic tools.1 Traditional prognostic indicators, such as the Karnofsky Performance Status (KPS) and the Neurological Assessment of Neuro-Oncology (NANO), have long served as cornerstones of clinical assessment.4 While valuable, these tools are heavily reliant on subjective clinician evaluations, which can lead to considerable interobserver variability. Such subjectivity may obscure the true clinical picture and fail to capture the multifaceted biological underpinnings of GBM, ultimately impacting treatment decisions and patient outcomes. To address these limitations, there is a growing interest in the application of quantitative imaging techniques, particularly radiomics, to enhance prognostic assessments. Radiomics involves the extraction and analysis of a multitude of features from medical images, allowing for a more objective and comprehensive evaluation of tumor characteristics.5 By leveraging advanced imaging modalities, such as magnetic resonance imaging (MRI), radiomics can uncover insights into the tumor microenvironment, potentially revealing patterns that traditional metrics might overlook.5 A particularly promising area of research is the assessment of the temporalis muscle, located within the temporal fossa of the skull. This muscle is a reliable indicator of lean muscle mass and is universally visible on clinical brain MRI, making it convenient for assessing muscle quality and body composition.6 Recent studies suggest that metrics such as temporalis muscle thickness and cross-sectional area may serve as objective indicators of prognosis in GBM patients.4, 7 However, these morphological indices provide limited information regarding muscle quality and are often subject to measurement inaccuracies and high interrater variability due to the irregular morphology and thin structure of the temporalis muscle. Thus, it is essential to move beyond conventional morphological and volumetric metrics in disease prognosis. Radiomics, through the extraction of high-dimensional quantitative features, offers deeper insights into the pathophysiological changes occurring within both the temporalis muscle and the tumor itself. This approach mitigates the limitations associated with traditional metrics, such as measurement errors and interrater variability, which are particularly pronounced in irregularly shaped structures. Consequently, integrating temporalis muscle radiomics with tumoral features and clinical profiles enhances prognostic precision and supports the development of tailored therapeutic strategies in precision medicine. Moreover, radiomics-based prognostic tools show promise in assisting clinical practices for treating GBM patients. However, incorporating radiomic analysis into routine diagnostic protocols necessitates seamless collaboration among various disciplines, including radiologists, oncologists, data scientists, and software engineers. Standardizing radiomic feature extraction protocols and developing user-friendly software interfaces for healthcare professionals are critical steps toward the clinical implementation of these tools, facilitating their integration into existing hospital infrastructures. As radiomics becomes more embedded in clinical practice, it has the potential to significantly enhance diagnostic accuracy, inform treatment decisions, and ultimately improve patient outcomes. Despite the potential of radiomics, several limitations warrant consideration. One key concern is the variability in patient demographics, such as age, which may influence the predictive value of radiomic features. Future studies should aim to account for demographic differences and explore age-adjusted models to enhance the robustness of prognostic assessments. Another challenge lies in the current reliance on manual segmentation of imaging data. This process is not only time-consuming but also susceptible to variability among raters, potentially impacting the reliability of radiomic analyses. Advancements in automatic segmentation, particularly through the application of machine learning and deep learning techniques,8 could significantly enhance the reproducibility and scalability of radiomic assessments. Notably, Zapaishchykova et al recently introduced a deep learning-based pipeline for standardized and reproducible auto-segmentation of the temporalis muscle on brain MRI scans,6 which holds promise as a valuable tool for future research endeavors. In conclusion, the potential of temporalis muscle radiomic features as prognostic biomarkers in GBM is significant. Incorporating radiomics of both the temporalis muscle and the tumor into clinical practice could greatly enhance patient care by enabling more precise prognostication and personalized treatment planning for GBM patients. Future research should aim to validate these findings in larger, more diverse cohorts while exploring the integration of additional imaging modalities and advanced machine learning algorithms. As the fields of radiomics and machine learning continue to progress, they hold the promise of uncovering novel biomarkers and therapeutic targets, offering renewed hope for patients facing this devastating disease. Dan Sun reported being employed at Tempus AI, Inc. The other authors declare no conflict of interests.