Supramarginal Resection in Glioblastoma Guided by Artificial Intelligence

Glioblastomas are the most common and poorly prognostic primary brain neoplasms. Despite advances in surgical techniques and chemotherapy, the median survival time for these patients remains less than 15 months. This highlights the need for more effective treatments and improved prognostic tools. The globally accepted surgical strategy currently consists of achieving the maximum safe resection of the enhancing tumor volume. However, the non-enhancing peritumoral region contains viable cells that cause the inevitable recurrence that these patients face. Clinicians currently lack an imaging tool or modality to differentiate neoplastic infiltration in the peritumoral region from vasogenic edema. In addition, it is not always feasible to include all the T2-FLAIR signal alterations surrounding the enhancing tumor in the surgical planning due to the proximity of eloquent areas and the higher risk of postoperative deficits. However, the investigators have developed a model to predict regions of recurrence based on machine learning and MRI radiomic features that have been trained and evaluated in a multi-institutional cohort. The investigators aim to analyze whether an adjusted supramarginal resection guided by these new recurrence probability maps improves survival in selected patients with glioblastoma.

The SupraGlio-AI study aims to test the feasibility of the proposed AI-guided tailored supratotal resection for glioblastomas. The study will provide preliminary data on the accuracy of the AI model in predicting recurrence and the impact of using this information in surgical planning. This information will be crucial in determining the potential for a larger, randomized controlled trial in the future. The pilot study will also allow for refinement of the study design, intervention, and data collection processes before a larger-scale study is conducted. In addition to testing the feasibility and efficacy of the AI-guided tailored supratotal resection, this pilot study also has two secondary objectives: 1) Survival Analysis: The survival analysis will provide insights into the impact of using the AI model on patient outcomes and help determine the potential benefits of this approach. 2) Histopathological and Transcriptomic Analysis: The study will also include a histopathological and transcriptomic analysis of the tissue samples obtained from the high-risk regions defined by the AI model. This analysis will provide information on the molecular and cellular changes occurring in these regions and may offer insights into the underlying biology of glioblastoma recurrence. These data will inform the development of future studies aimed at improving patient outcomes. By incorporating these secondary objectives, this pilot study will contribute to a more comprehensive understanding of the potential benefits of using AI in guiding tailored supratotal resection for glioblastomas. The results will inform future research and potentially lead to the development of improved treatment approaches for patients with this type of brain tumor.