P11.32.a glioblastoma nomogram: predicting the location of first progression after long-course chemoradiotherapy based on a national cohort

BACKGROUND Glioblastoma (GBM) is known for its high local failure rate, as well as its migration, leading to these possible locations of disease progression: 1. Local-only; lesion of progression is connected to the initial tumour location 2. Non-local; lesion of progression is not connected to the initial tumour location 3. Combined; both local-only and non-local progression lesions are present An upfront prediction of progression location could potentially be used to select patients for differential treatment strategies. The aim of this retrospective national study was therefore to predict location of first progression using clinical variables. MATERIAL AND METHODS All adults with newly diagnosed GBM between 2014-2019 who started long-course (chemo)radiotherapy were included, if they had a radiotherapy-planning MRI made. The MRI of first progression was identified by retrospective review following RANO-criteria, and used to assess the location of contrast-enhancing progression. Median time to progression (TTP) and survival beyond progression were estimated using the Kaplan-Meier method (log rank test for comparison). To predict location of progression in subgroups of patients, we used a multivariable multinomial model (age, biological sex, performance status, tumour location & focality & laterality, MGMT-methylation status, and extent of surgery). RESULTS We included 940 patients (MGMT methylated n=376, unmethylated n=483, not reported n=81), of whom 760 had radiographic progression. Of those, progression location was local-only in 69%, non-local in 16%, and combined in 15%. Median (95% CI) TTP in months was shortest in the combined group (8.4, 7.3-9.4), followed by the local-only (9.2, 8.5-9.8) and non-local (13.9, 12.3-15.6) (p-value <.001). Survival beyond progression in months was shortest in the combined group (5.0, 3.8-6.2), followed by the non-local (6.0, 4.4-7.6) and local-only (9.0, 8.3-9.7) (p-value <.001). When comparing patients with MGMT-methylated and unmethylated tumours, the probability (95%CI) of local-only progression was .67 (.62-.73) vs. .73 (.69-.78), non-local was .21 (.17-.26) vs. .11 (.08-.14), and combined was .12 (.08-.15) vs. .16 (.12-.19). In the multivariable model, the probability (95%CI) of local-only progression varied from .49 (.29-.69) to .86 (.74-.97), of non-local from .04 (.00-.09) to .34 (.15-.53), and of combined from .05 (.01-.10) to .34 (.07-.60) between respective subgroups. CONCLUSION Location of progression was associated with time to progression and survival beyond progression. In our study population it was feasible to identify subgroups of patients with a high chance of local-only progression, who may potentially benefit most from radiation dose-escalation. Both patients with MGMT-methylated and unmethylated tumours had the highest chance for first progression located at the initial tumour site only.