Real-world prognostic model for malignant pleural mesothelioma.

Abstract

e20562 Background: Two mesothelioma prognostic models have been suggested: European Organisation for Research and Treatment of Cancer (EORTC) and Cancer and Leukaemia Group B (CALGB) models. Both were based on clinical trial patients enrolled in the mid-1980s to early-1990s. Several changes to mesothelioma management have been adopted since publication of these models, including improved surgical and palliative interventions and changes to systemic therapy. Methods: With ethics approval, we collected and analyzed the health data of malignant pleural mesothelioma (MPM) patients with histologically confirmed diagnosis treated at our institution between January 1991 and March 2019. The primary endpoint was overall survival (OS). Univariate analysis was used to identify significant predictors of survival, which were then used to construct a multivariate survival tree with complete case-analysis and bootstrapping. Patients were then stratified into three prognostic groups based on their median OS. Results: 337 patients were included in the study; 309 (91.7%) were dead at last follow-up. The median OS was 9.4 (8.1-11.5) months for the entire population. Eastern Cooperative Oncology Group (ECOG) performance status (PS), histology, white blood count, platelets, International Mesothelioma Interest Group stage, age and hemoglobin were independent predictors of survival. The final pruned survival tree was based on 285 patients and incorporated the first five predictors. Good, intermediate and poor prognostic groups had median OS of > 12 months, 6-12 months, and < 6 months, respectively. Factors associated with the prognostic groups were: good prognosis: ECOG 0-1, normal platelets, stage 1, 2 and epithelioid histology; intermediate prognosis: ECOG 0-1 with either stage 3, 4 and/or sarcomatoid or biphasic histology; poor prognosis: ECOG 2-4 regardless of other factors. Conclusions: In contrast to EORTC/CALGB, real world evidence generates these prognostic groups with face validity but will need independent validation. Our data does not account for recent advances including immunotherapy, and thus patients with non-epithelioid histology may have better survival than predicted.[Table: see text]