OBJECTIVES

: The EF-14 trial demonstrated that the addition of Tumor-Treating Fields (TTFields) to maintenance chemotherapy in newly diagnosed glioblastoma (GBM) resulted in statistically significant improvement in survival. EF-14 gave the highest 5-year survival observed in a GBM clinical trial to date. With this new potential for Long-Term Survival (LTS) in GBM, predictive modeling of LTS is now especially relevant for clinical and policy decision making. Previous studies have parametrically modeled the survival impact of TTFields, but their results suffer from poor in-sample fit, clinical inconsistency, and poor fit to long-term epidemiological data. Here we evaluate the LTS predictions from a novel, Responder Analysis-based method of parametric modeling applied to the EF-14 trial data.

METHODS

: Patients were classified as responders (R) or non-responders (NR) according to their Progression-Free Survival (PFS). If a patient’s PFS met or exceeded a given threshold time (TT), then they were classified as R. Otherwise they were classified as NR. Several TT’s were studied including 10, 12, 18, and 20 months. Parametric survival distributions were calibrated for each R/NR subgroup in each trial arm, producing four separate distributions. These were fit using Maximum Likelihood Estimation and Akaike Information Criterion for model selection. The long-term (greater than 5 year) conditional survival probabilities from each model were then compared to the epidemiological results from Porter et al. (2011).

RESULTS

: For every TT studied, the parametric distributions for the R-subgroups, despite showing significantly improved survival compared to NR’s, grossly underestimated conditional LTS probabilities compared to epidemiological results.

CONCLUSIONS

: This shows that parametric models trained exclusively on relatively short-term EF-14 trial data, even with responder analysis-based enhancements, suffer from considerable biases, limiting their usefulness for LTS modeling in GBM. Long-term epidemiology data and modeling approaches that incorporate this data are therefore necessary for informed decision making regarding novel treatments in glioblastoma.