Time trends in glioblastoma multiforme survival: the role of temozolomide

In 2005, maximum safe surgical resection, followed by radiotherapy with concomitant temozolomide (TMZ), followed by adjuvant TMZ became the standard of care for glioblastoma (GBM). Furthermore, a modest, but meaningful, population-based survival improvement for GBM patients occurred in the US between 1999 (when TMZ was first introduced) and 2008. We hypothesized that TMZ usage explained this GBM survival improvement. We used national Veterans Health Adminis-tration (VHA) databases to construct a cohort of GBM patients, with detailed treatment information, diagnosed 19972008 (n 1645). We compared survival across 3 periods of diagnosis (19972000, 20012004, and 20052008) using KaplanMeier curves. We used proportional hazards models to calculate period hazard rate ratios (HRs) and 95 confidence intervals (CIs), adjusted for demographic, clinical, and treatment covariates. Survival increased over calendar time (stratified log-rank P .0001). After adjusting for age and Charlson comorbidity score, for cases diagnosed in 20052008 versus 19972000, the HR was 0.72 (95 CI, 0.640.82; p-trend .0001). Sequentially adding non-TMZ treatment variables (ie, surgery, radiotherapy, non-TMZ chemotherapy) to the model did not change this result. However, adding TMZ to the model containing age, Charlson comorbidity score, and all non-TMZ treatments eliminated the period effect entirely (HR 1.01; 95 CI, 0.861.19; p-trend 0.84). The observed survival improvement among GBM patients diagnosed in the VHA system between 1997 and 2008 was completely explained by TMZ. Similar studies in other populations are warranted to test the generalizability of our finding to other patient cohorts and health care settings.