Virtual Library

Start Your Search

Cheng Lu



Author of

  • +

    MA11 - Biomarkers of IO Response (ID 912)

    • Event: WCLC 2018
    • Type: Mini Oral Abstract Session
    • Track: Immunooncology
    • Presentations: 1
    • Moderators:
    • Coordinates: 9/25/2018, 10:30 - 12:00, Room 203 BD
    • +

      MA11.03 - Interaction of Tumor Infiltrating Lymphocytes and Cancer Nuclei Predicts Response to Nivolumab in Non-Small Cell Lung Cancer (NSCLC) (ID 14143)

      10:40 - 10:45  |  Author(s): Cheng Lu

      • Abstract
      • Presentation
      • Slides

      Background

      Immune checkpoint inhibitors, particularly drugs targeting the Programmed death-1 (PD-1) pathway, are promising agents in NSCLC. These drugs however are effective in only a small subset of patients. Programmed death Ligand-1 (PDL1) expression in the tumor predicts response to these agents but is not an optimal biomarker because of spatial and temporal heterogeneity associated with PDL1. PD-L1 is upregulated in response to inflammation in the tumor and strongly correlates with Tumor-infiltrating lymphocytes (TILs). In this work, we evaluated whether quantitative measurements relating to the spatial interplay and arrangement of TILs and cancer nuclei from diagnostic biopsy tissue slide images (H&E) was predictive of response to Nivolumab.

      a9ded1e5ce5d75814730bb4caaf49419 Method

      Tumor biopsies of total 82 NSCLC patients previously treated with Nivolumab from two different institutions were employed in this study. The RECIST criteria was used to define response. [vv1] The 492 features characterizing the global interaction of TILs and cancer cells through graph interplay metrics are extracted from tumor regions delineated by two expert pathologists to interrogate the difference of phenotypes. Top 5 features were learnt on learning set by random forest classifier from one institution (n=32) and independently validated on patients from a second site (n=50).

      4c3880bb027f159e801041b1021e88e8 Result

      The most predictive features comprised of difference of characteristic path length between lymphocyte graph and cancer nuclei graph and cosine similarity between lymphocyte node and cancer nuclei node based on their node centrality index. The random forest classifier yielded an area under the receiver operating characteristic curve (AUC) of 0.76 on the training cohort and 0.68 on the validation set (Figure 1).til40.png

      8eea62084ca7e541d918e823422bd82e Conclusion

      Our results showed that quantitative measurements relating to the spatial interplay and arrangement of lymphocyte and cancer nuclei from H&E slide images were predictive of response to Nivolumab in NSCLC. Additional independent multi-site validation of these features is needed.

      6f8b794f3246b0c1e1780bb4d4d5dc53

      Only 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, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      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.