Real-world mortality and the effect of comorbidities on survival in glioblastoma: A database study using computed phenotypes for rapid, reliable patient identification.

INTRODUCTION

Glioblastoma (GBM) continues to have poor prognosis with minimal improvement since the introduction of the Stupp protocol. While clinical data are often homogenous and not generalizable, real-world data can be used to identify more specific risk factors for poor outcomes.

METHODS

Utilizing our proprietary computerized phenotype, we identified adult GBM patients at the University of Florida (UF) and the OneFlorida+ CRN database from 2011 to 2021. The chi-square test was used to compare demographic variables between the 2 groups. Additionally, the effect of medical comorbidities and the Charlson Comorbidity Index on overall survival was evaluated.

RESULTS

A total of 380 and 1630 GBM patients were identified from UF and OneFlorida+, respectively. UF patients were older, with a lower percentage of minority races. Patients with seizures, hypertension, and dementia had the longest median survival, while those with liver disease and diabetes had the lowest median survival. Older age at diagnosis, presenting with paraplegia or hemiplegia, diabetes, liver disease, and hyperlipidemia were all found to be significant predictors of mortality on multivariate Cox analysis.

CONCLUSION

An estimated 20% of GBM patients participate in clinical trials, limiting the generalizability of results. It is imperative to continue to work towards obtaining high-volume real-world data to improve our understanding of this devastating condition. Here, we confirm that GBM cohorts can be identified in large databases and, using this real-world data, identified several comorbidities that confer worse or better survival times for patients with GBM. This data can be used to improve patient stratification and counseling.