Author of

• 18606

  +

  P3.03 - Poster Session with Presenters Present (ID 473)

  • Event: WCLC 2016
  • Type: Poster Presenters Present
  • Track: Mesothelioma/Thymic Malignancies/Esophageal Cancer/Other Thoracic Malignancies
  • Presentations: 1
  • Moderators:
  • Coordinates: 12/07/2016, 14:30 - 15:45, Hall B (Poster Area)

  • 18642

    +

    P3.03-036 - Prognostic Model for Mesothelioma Based on Cancer and Leukemia Group B (CALGB) Trials (Alliance) (ID 3976)

    14:30 - 14:30  |  Author(s): K. Yan
    • Abstract

    Background:
    Prognostic models play an important role in the design and analysis of mesothelioma treatment trials. The European Organisation for Research and Treatment of Cancer (EORTC) developed a well-known tool in 1998 to predict overall survival (OS) in patients with malignant mesothelioma. In this study, we built and assessed the performance of a new mesothelioma prognostic model OS using data from multiple CALGB clinical trials data.
    
    Methods:
    This study included 595 mesothelioma patients from fifteen completed CALGB treatment trials accrued between June 1984 and August 2009. We split the cohort of patients into two parts - 67% of patients as training and 33% as testing. We developed a Cox model using the training set with PS, age, WBC count, and platelet count as prognostic variables. To compare the EORTC and our new models, the concordance of predicted survival times and risk scores were estimated by concordance C (c-index) (Harrell et al. 1996) and AUC score at 6-months (Patrick et al. 2000). 95% confidence intervals were calculated for the c-index. Based on the prediction model fit from training set, we partitioned testing set patients into high-risk and low-risk groups using the median for their risk score values for the new model. For the EORTC model, the cut off of 1.27 from the original paper was used to assign the high-risk and low-risk groups. A Log-rank test was used to compare the survival curves of these two groups. We also compared our results with a model using PS alone.
    
    Results:
    For OS, the EORTC model c-index was 0.55 (0.52, 0.58) and P = 0.0007 comparing high- and low- risk patients for testing set. The new model c-index was 0.60 (0.56, 0.64), with P < 0.000001 for testing set. Using the new model, the median OS in the high-risk and low-risk groups in the testing set were 5.16 (4.70, 6.37) and 10.41 (7.95, 14.32) months, respectively. PS alone produced c-index of 0.55 (0.53, 0.57) and P = 0.0002 for testing set. The AUC scores at 6-months for testing set generated by EORTC and PS alone models are 0.62 and 0.66. The new model generated AUC scores at 6-months of 0.70.
    
    Conclusion:
    Our new model performs better than the EORTC model or PS alone for survival prognostication in patients with mesothelioma.