Virtual Library

Start Your Search

Isaac Kah Siang Ng

Author of

  • +

    P2.13 - Targeted Therapy (Not CME Accredited Session) (ID 962)

    • Event: WCLC 2018
    • Type: Poster Viewing in the Exhibit Hall
    • Track:
    • Presentations: 1
    • Moderators:
    • Coordinates: 9/25/2018, 16:45 - 18:00, Exhibit Hall
    • +

      P2.13-27 - Development, Internal Validation, and Calibration of a Risk Score to Predict Survival in Patients with EGFR Mutant NSCLC (ID 13610)

      16:45 - 18:00  |  Author(s): Isaac Kah Siang Ng

      • Abstract


      The emergence of molecularly targeted therapy has transformed the clinical landscape of metastatic non-small cell lung cancer (NSCLC). We profiled a local cohort of EGFR mutation-positive, advanced NSCLC patients who were treated with first-line EGFR tyrosine kinase inhibitors (TKIs) and sought to develop a nomogram-based prognostic model for predicting clinical outcomes in this patient subgroup.

      a9ded1e5ce5d75814730bb4caaf49419 Method

      A cohort of 199 EGFR mutation-positive, advanced NSCLC patients was retrospectively analyzed, with the inclusion criteria as follows: at least one sensitizing EGFR mutation, diagnosed with metastatic disease at diagnosis or incurable disease recurrence, and subsequently treated with first-line EGFR TKI for at least one-month duration. The Mnet machine learning algorithm was used to select variables for inclusion into the Cox model, with an alpha of 0.06 and gamma of 3. Internal validation was performed by plotting the bootstrapped (n = 200 bootstrap samples) time-dependent AUC according to the method described in Uno et al. (2007), and the predictive power of the model was assessed by computing Harrell’s C and Somers’ D concordance statistics. Internal calibration was evaluated using a plotted graph of observed vs predicted mortality risks of five risk groups (“low risk”, “medium-low risk”, “medium risk”, “medium-high risk”, and “high risk” quintiles), as well as the Kaplan-Meier survival estimates for each quintile.

      4c3880bb027f159e801041b1021e88e8 Result

      The resultant prognostic nomogram included the following variables - total white blood cell (WBC) count, hemoglobin levels, serum LDH levels, Neutrophil/Lymphocyte Ratio (NLR), ethnicity (Chinese vs non-Chinese), Karnofsky-Performance Status (score of ‘90-100’ or ‘70-80’ vs ‘0-60’), Charlson Comorbidity Index (≥3, 2, or 1 vs 0), neurological symptoms, brain, lung/pleural and adrenal metastases. The time-dependent AUC at 6, 12 and 18 months were 62.8%, 65.5%, and 67.4% respectively and the Harrell’s C and Somers’ D concordances were 0.6342 and 0.2684 respectively. The 12-month observed vs predicted overall survival (OS) probabilities were 0.56 vs 0.61, 0.64 vs 0.66, 0.55 vs 0.69, 0.77 vs 0.71, and 0.87 vs 0.73 respectively for each of the five risk groups described earlier, suggesting good calibration. Survival curves for the five risk groups also segregated well on visual inspection, and was corroborated by log-rank P < 0.001.

      8eea62084ca7e541d918e823422bd82e Conclusion

      We herein report a novel nomogram-based risk scoring system that incorporates biochemical, immune and clinical variables to provide fairly robust OS estimation in EGFR mutation-positive, advanced NSCLC patients treated with first-line EGFR TKIs.