Virtual Library

Start Your Search

G.W. Chmielewski



Author of

  • +

    P2.06 - Poster Session/ Screening and Early Detection (ID 219)

    • Event: WCLC 2015
    • Type: Poster
    • Track: Screening and Early Detection
    • Presentations: 1
    • +

      P2.06-012 - A Model Incorporating Clinical, Radiographic, and Biomarker Characteristics Predicts Malignancy in Indeterminate Pulmonary Nodules (ID 2890)

      09:30 - 09:30  |  Author(s): G.W. Chmielewski

      • Abstract
      • Slides

      Background:
      The high false-positive rate associated with low-dose computed tomography (CT) lung cancer screening results in unnecessary testing, cost, and patient anxiety. We hypothesized that an algorithm incorporating clinical, radiographic, and serum biomarker data would be capable of differentiating benign from malignant pulmonary nodules.

      Methods:
      An institutional biorepository was used to identify 84 patients with ≤ 2 cm indeterminate pulmonary nodules identified on CT scan, including 50 patients with biopsy-proven, node-negative, non-small cell lung cancer (NSCLC) and 34 patients with benign, non-calcified, solitary pulmonary nodules. Clinical and radiographic data were collected from patient charts and imaging studies. Serum specimens were evaluated in a blinded manner for 55 biomarkers using multiplex immunoassays. Random forest analyses were used to generate a multivariate cross-validation prediction model incorporating clinical, radiographic, and serum biomarker data.

      Results:
      A total of 84 patients were identified with a median nodule size of 5 mm for benign nodules and 15 mm for NSCLC. Median smoking histories were 21 and 28 pack-years and patient age was 62 and 70 years, respectively. An algorithm incorporating serum biomarker profile (IGFBP-4, IGFBP-5, IL-10, IL-1ra, IL-6, SDF-1alpha, IGF-2), age, sex, BMI, COPD, smoking history, hemoptysis, previous cancer, nodule size, nodule location, spiculation, nodule type, and nodule count provided the optimal performance with a sensitivity 92%, specificity 65%, NPV 85%, and PPV 79%. This model performed with an overall accuracy of 81% with a cross-validated AUC=0.904.

      Conclusion:
      An algorithm incorporating clinical, radiographic, and serum biomarker characteristics may help differentiate benign from malignant pulmonary nodules. This model is currently being externally validated in a second-site patient cohort.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.