Clinical and pathologic characteristics of long-term glioblastoma survivors.

Abstract

e14040 Background: Glioblastoma (GBM) is an aggressive brain tumor with a mean survival of 14 months. A small subset of patients live > 3 years and are classified as long-term survivors (LTS). To better understand factors influencing prognosis, we conducted a large retrospective single institution study using a novel method to identify LTS characteristics. Methods: Using a combination of 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 who presented at MD Anderson Cancer Center from 2006-2021 with a pathology-confirmed diagnosis of glioblastoma. Reports mentioning GBM were extracted using RegEx, and we defined overall survival (OS) as the duration between the date of the first CT or MRI report mentioning GBM and the last report on record. To verify the accuracy of this method, we performed a chart review where OS was defined as the date of pathological GBM diagnosis to the current date or date of death. Results: Cases with < 4 imaging reports were filtered to ensure high quality data yielding 2430 patients. The validation of NLP survival determination revealed a sensitivity and specificity of 94.7% and 97.6%, respectively and a mean absolute error percentage of 19.40% when tested against manual chart review. Mean OS of the GBM patients was 631 days. 38.9% (n = 946) of patients were female with average survival of 22.56±27.94 months while 61.1% (n = 1484) were male with an average survival of 19.55±25.96 months (p = 0.008). 61.1% (n = 1485) survived ≤1 year, 28.1% (n = 682) survived 1-3 years, and 10.8% (n = 263) survived ≥3 years. Average age for short-term survivors (STS) (< 1 year) was 52.42 ± 13.698 months and average age for LTS (> 3 years) was 54.75±13.965 months (p < 0.001). IDH mutants had a mean survival of 38.20 months while IDH wildtype patients survived 17.69 months (p < 0.0001). MGMT unmethylated patients had a mean survival of 13.70 months while methylated MGMT patients survived 23.99 months (p < 0.0001). Conclusions: Our OS statistics align with past literature. IDH mutants had better survival than wildtype and MGMT methylated tumors had better survival than unmethylated tumors. There was a significant difference in OS between male and female patients and in age between LTS and STS. Filtering patients with < 4 imaging reports removed poor quality patient data and likely left-skewed our data as the appearance of poor-quality patient data overlaps with those with exceptionally short survival. Our NLP method is a reliable method by which to large oncologic population trends. Understanding differences between LTS and STS can help with improving individualized treatment strategies. Our database is also well positioned for future studies, especially those relating the effects of radiation, chemotherapy, extent of resection, and imaging characteristics on survival.