P1.01-39 Is There an Association Between Intracranial Progression and Overall Survival or Neurologic Death in Patients with EGFR Positive NSCLC?

Abstract

Many studies have examined the incidence of Brain Metastases (BrM) in EGFRm+ NSCLC, and their impact on radiographic and survival outcomes. We have examined the incidence of ND and predictive factors in a retrospective cohort. The clinical significance of ICP remains controversial; patients who develop new lesions may be treated with further locoregional treatment such as stereotactic radiosurgery (SRS). This study examines the association between ICP and survival or ND, to help define appropriate endpoints for future trials. All patients treated for BrM from EGFRm+ NSCLC between 2004 and 2016 were identified from an institutional registry. Clinical data regarding demographics, progression events, pattern of progression, and treatment were extracted from medical records and verified with imaging reports. A multivariable competing-risks model was constructed to test the association between ICP at 6, 12 and 18 months and the outcome of ND. A separate model tested the association between ICP and OS. 198 patients were included in the study. Median age was 61 years, 67% of patients were female, 46% had BrM at NSCLC diagnosis. Median DS-GPA was 2.5. Median OS for the group was 20 months, and the 5-year cumulative incidence of neurologic death was 40%. 111 patients (56%) had ICP and median intracranial PFS was 16 months; 29% had ICP within 6 months and 33% between 6 and 12 months, 18% at 12-18 months and 21% at >18 months. ICP was due to new BrM in 51%, progression of existing BrM in 34% and leptomeningeal carcinomatosis in 15%. Among patients with ICP, earlier progression was associated with ND (HR 2.48, p=0.01, see table). In addition, pattern of ICP was significant; patients with leptomeningeal spread (HR 3.74, p<.001) had higher risk of ND than those with new BrM or progression of existing lesions. In a multivariable model, early ICP, pattern of ICP and more initial BrM were independently associated with higher risk of ND and also shorter OS.Tabled 1Regression models of ND and OSCovariateUnivariate Model of Neurologic DeathMultivariable Model of Neurologic DeathMultivariable Model of Overall SurvivalHR (95% CI)P valueHR (95% CI)P valueHR (95% CI)P valueIntracranial Progression < 6 months 6-12 months 12-18 months >18 months2.48 (1.21-5.08) 1.56 (0.75-3.23) 1.61 (0.72-3.57) ref0.013 0.24 0.254.88 (2.13-11.19) 2.69 (1.06-6.80) 3.85 (1.44-10.29) ref<.001 0.036 0.00711.22 (5.11-24.64) 5.31 (2.47-11.42) 4.82 (2.16-10.77) ref<.001 <.001 <.001Pattern of Progression New lesion Progression of existing BrM Leptomeningeal Spreadref 1.29 (0.70-2.35) 3.74 (1.96-7.15)0.41 <.001ref 1.12 (0.55-2.26) 4.70 (2.11-10.48)0.76 <.001ref 1.54 (0.93-2.57) 3.19 (1.64-6.20)0.10 <.001Number of Initial BrM 1-4 5-10 >10ref 1.92 (0.95-3.86) 2.94 (1.60-5.39)0.07 <.001ref 2.43 (1.05-5.62) 2.57 (1.11-5.90)0.038 0.027ref 3.27 (1.61-6.62) 3.74 (1.96-7.15)0.001 <.001First Line CNS Treatment WBRT SRS Systemic aloneRef 0.69 (0.38-1.24) 0.58 (0.27-1.23)0.22 0.15Ref 1.17 (0.49-2.75) 0.88 (0.42-1.87)0.73 0.88Ref 1.10 (0.57-2.11) 1.28 (0.70-2.34)0.79 0.43EGFR Exon 19 vs 210.73 (0.44-1.23)0.240.71 (0.40-1.24)0.230.72 (0.52-1.00)0.05BrM at Diagnosis0.93 (0.54-1.60)0.781.45 (0.77-2.73)0.260.93 (0.65-1.32)0.69 Open table in a new tab Among patients with brain metastases from EGFRm+ NSCLC, shorter time to intracranial progression was associated with higher rates of neurologic death, as was the pattern of progression. This supports the validity of studies using intracranial progression as an endpoint, particularly if pattern of disease is taken into account.