Glioblastoma (GBM) is a highly aggressive brain tumor with a mean survival of 14 months. Literature has not well defined the impact of radiographic markers on prognosis due to limited patient data. To properly characterize the impact of radiographic and anatomical features of GBM on survival, we conducted the largest single-institution study at MD Anderson Cancer Center.

Using regular expression (RegEx) search and natural language processing (NLP) in Python, data extracted from the radiology information system (RIS) was used to identify 4425 patients from 2006-2021 with pathology-confirmed GBM. Overall survival (OS) was defined as time between first CT/MRI report and last. Tumors were characterized by single lobe involvement and multilobe involvement. Size per radiology report and laterality based on the patient's hand dominance were analyzed as well.

Single lobe analysis identified 579 frontal lobe tumors (OS 785.64±1034.64 days), 383 temporal lobe tumors (668.04±838.87), 195 parietal lobe tumors (OS 653.90±823.20), and 47 occipital lobe tumors with OS (498.091±456.29). T-testing, with no assumption of equal variance, showed a significant difference in OS between frontal tumors and occipital tumors (p<.001). ANOVA analysis showed significant difference in tumor size between single lobe groups, with temporal lobe tumors having the largest size (p<.001). Multilobe analysis via non-parametric t-testing showed that frontal lobe involvement had better OS than non-frontal lobe involvement (656.82±887.78 vs. 601.493±722.140 [p<.05]). Conversely, occipital lobe involvement was associated with significantly lower survival than non-occipital lobe involvement (516.57±622.26 vs. 655.12±848.01 [p<.01]). Laterality had no effect on OS (p=.698); however, the prevalence of dominant tumors was greater than non-dominant (85.3% vs 14.7%) Tumor size had no effect on differentiating short-term vs. long-term survivors (p=.563).

We demonstrate that tumor location and lobe involvement have a significant impact on patient survival. Frontal lobe involvement waassociated with better prognosiand occipital with worse. Tumor size wanot different between long-term and short-term survivors, suggesting other factorand biomarkerat play responsible for long-term GBM survivors.

The identification of various radiographic and anatomic markers of GBM can help assist physicians with clinical management and provide avenues for future research.