Virtual Library

Start Your Search

Julian A. Villalba



Author of

  • +

    P38 - Pathology - Pathology/Staging (ID 108)

    • Event: WCLC 2020
    • Type: Posters
    • Track: Pathology, Molecular Pathology and Diagnostic Biomarkers
    • Presentations: 1
    • Moderators:
    • Coordinates: 1/28/2021, 00:00 - 00:00, ePoster Hall
    • +

      P38.02 - Reproducibility and Accuracy of Intra-Operative Assessment on Tumor Spread Through Air Spaces in Stage 1 Lung Adenocarcinomas (ID 1819)

      00:00 - 00:00  |  Presenting Author(s): Julian A. Villalba

      • Abstract
      • Slides

      Introduction

      STAS has been associated with worse prognosis in patients with early-stage lung adenocarcinoma, particularly those undergoing sublobar resection. Intra-operative consultation (IC) of STAS has been advocated by some surgeons to guide surgical management. However, to date, information regarding the reproducibility and accuracy of frozen section (FS) for STAS diagnosis is limited. Thus, we assessed inter-observer (IOA) and intra-observer agreement among pathologists evaluating STAS in FS, and its diagnostic yield for detecting STAS. Preliminary data were previously presented at a conference and we now report the results of more detailed analyses with a larger study cohort.

      Methods

      We retrospectively evaluated 100 T1 lung adenocarcinoma resections assessed by IC and with ample non-neoplastic lung parenchyma in FS slides. A consensus panel of 3 pathologists jointly reviewed all histology slides, documented the diagnosis on STAS (+/-) and artifacts (+/-) per published criteria (Kadota, 2015), and recorded several histological parameters in each case. Five pulmonary pathologists blinded to clinicopathologic data, independently reviewed FS slides in each case on two separate rounds, and recorded the presence of STAS and artifacts per same criteria. The second round occurred after a consensus conference in which selected cases with low IOA were discussed. IOAs on STAS were assessed using the Fleiss-kappa statistics and were correlated with several pathologic features including the presence of artifacts. Diagnostic yield was determined by comparing consensus diagnoses made by >3 observers in the FS slide with the final diagnosis of the consensus panel, and assessed by receiver operating characteristic curves (ROC).

      Results

      There was a high variability in the prevalence of STAS in FS reported by all observers on both rounds (mean: 33% and 35.4%, respectively). The Fleiss-kappa value for the STAS diagnosis in FS slides among the 5 observers was 0.45 in the first round and only slightly increased in the second round (0.51), while the intra-observer agreement on STAS diagnosis in FS between both rounds were relatively high ranging from 67.9 to 82.5% (mean: 76.8%). The sensitivity (44%) for detecting STAS in FS was low, while the specificity (91%) was high. Overall, FS showed a 71% accuracy for STAS diagnosis, a positive likelihood ratio of 5.0, a negative likelihood ratio of 0.6, and a ROC/AUC of 0.67 (95%CI: 0.56-0.78). IOA for the FS diagnosis of STAS on both rounds was significantly higher in low-grade tumors (p<0.0001; Fig. 1A), in those with no STAS based on the entire tumor assessment (p<0.0001; Fig. 1B), in those with no artifacts identified by more than 3 observers (p<0.0001; Fig. 1C), and in those STAS-positive cases with five or more STAS-clusters identified in FS compared to less than 5 clusters (p=0.009; Fig. 1D).

      Conclusion

      FS is highly specific but not sensitive for the diagnosis of STAS in Stage I lung adenocarcinoma. IOA on STAS was moderate in FS, and improved only marginally after a consensus conference, raising concerns about the global implementation of IC in STAS assessment. Multi-Institutional studies aimed to standardize criteria for STAS and artifacts, and to improve intraoperative techniques are warranted.

      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.