Virtual Library

Start Your Search

Jia-Jia Xu



Author of

  • +

    P1.11 - Screening and Early Detection (Not CME Accredited Session) (ID 943)

    • Event: WCLC 2018
    • Type: Poster Viewing in the Exhibit Hall
    • Track:
    • Presentations: 1
    • Moderators:
    • Coordinates: 9/24/2018, 16:45 - 18:00, Exhibit Hall
    • +

      P1.11-18 - A Classification-Based Machine Learning Method Reveals Exosomal miRNA Biomarkers for Patients with Pulmonary Ground Glass Nodule (ID 12462)

      16:45 - 18:00  |  Author(s): Jia-Jia Xu

      • Abstract

      Background

      Non-invasive detection of lung cancer is of critical importance but has proven challenging due to the rate of false-negative diagnosis with current tests. Plasma exosomes have been implicated as a non-invasive diagnostic source. However, little high throughput screening has been done in the early-stage lung cancer and problems such as bias of enrollment, less rigorous identification exists. This study aimed to reveal the plasma exosome-derived miRNA biomarkers for early-stage lung cancer patients, especially those with ground glass nodule (GGN).

      a9ded1e5ce5d75814730bb4caaf49419 Method

      Pre-operative and paired post-operative plasma samples from patients with solitary pulmonary nodule and healthy volunteers were prospectively collected. Finally 38 malignant nodules, 7 benign nodules and 5 healthy volunteers were enrolled. The malignant nodules included 9 pure GGNs, 11 mixed GGNs and 18 solid nodules. Exosomes were collected from 1mL plasma and were isolated with 3D Medicine EV isolation kit. Exosomal miRNA profiling was performed using miRNA-seq. And an exosomal miRNA diagnostic model for patients with malignant nodules was constructed by using support vector machine (SVM).

      4c3880bb027f159e801041b1021e88e8 Result

      In general, malignant nodules, benign nodules and healthy volunteers were indistinguishable based on overall clustering. Regarding to malignant nodules, pure GGNs and solid nodules could be separated under principal component analysis (PCA), and the mixed GGNs presented a transitional state between the pure GGNs and the solid nodules. Ultimately, a two-dimensional SVM diagnostic model for discriminating malignant and benign nodules was established. The optimal miRNA combination could reach an area under curve (AUC) of 0.96, with sensitivity and specificity of 94.7% and 91.7%, respectively.

      8eea62084ca7e541d918e823422bd82e Conclusion

      This preliminary analysis highlights the potential of exosomal miRNA based liquid biopsy for non-invasive detection of early-stage lung cancer. The SVM model seems could effectively distinguish pulmonary nodules, but needs further verified.

      fig.png

      6f8b794f3246b0c1e1780bb4d4d5dc53