Author of

• 25931

  +

  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

  • 26563

    +

    P1.11-17 - Discriminating Less Invasive Lesions of Early-Stage Lung Adenocarcinoma by Three-Dimensional Computed Tomography Analysis (ID 11151)

    16:45 - 18:00  |  Author(s): Kuniyoshi Hayashi
    • Abstract
    • Slides

    Background
    

    In the revised TNM classification for non-small cell lung cancer, the clinical T factor is specified by the maximum diameter and by the diameter of the solid component for subsolid nodules. However, as radiological measurement of the solid part is sometimes difficult, errors can occur among observers. If less invasive lesions can be truly predicted using computed tomography (CT) images, it will be very useful for determining treatment strategies. Hence, we investigated the ability to detect less invasive lesions in pathologically early-stage adenocarcinomas by evaluating the whole tumor based on three dimensional (3D) images from high-resolution CT (HRCT).

    a9ded1e5ce5d75814730bb4caaf49419 Method
    

    Among patients who underwent lung resections for primary lung cancer between February 2014 and December 2016 in our institution, we retrospectively reviewed 127 patients with pathological stage 0 or IA adenocarcinoma. All the lesions were divided into two groups: the less invasive group comprised adenocarcinoma in situ (AIS) and minimally invasive adenocarcinoma (MIA), and the invasive group comprised invasive adenocarcinoma. Radiological occupied lesion volumes (cm³) were semi-automatically calculated using 3D-CT volumetry, which included the data of CT values and voxels. The following six factors were also evaluated using voxel-based histogram analysis (VHA): % solid (the ratio of the volume with CT values above -300 Hounsfield units to the whole CT value), mean CT values, variance, kurtosis, skewness, and entropy. Using multivariate logistic regression analysis, the relationship between these seven variables and pathological less invasive lesions were analyzed to prepare an optimal model for detecting the less invasive group.

    4c3880bb027f159e801041b1021e88e8 Result
    

    Among 131 lesions, there were 39 lesions in the less invasive group (AIS/MIA = 16/23) and 94 in the invasive group. In univariate analysis, all the seven variables were significantly different between the two groups. Multivariate analysis using three variables revealed an odds ratio of 0.52 (95% confidence interval [CI]: 0.34-0.79, p = 0.002) for radiological lesion volume (cm³), 0.94 (95% CI: 0.89-0.99, p = 0.016) for % solid, and 1.58 (95% CI: 1.11-2.23, p = 0.01) for kurtosis. The optimal cut-off values were less than 8.2% for % solid, less than 5.8 cm³ for lesion volume, and greater than 3.6 for kurtosis. The area under the receiver operating characteristic curve was 0.92 (95% CI: 0.88-0.97) with the model, which achieved a 90% sensitivity and 84% specificity.

    8eea62084ca7e541d918e823422bd82e Conclusion
    

    Semi-automated objective discrimination of less-invasive lung adenocarcinomas can be achieved with high accuracy using VHA based on 3D-HRCT.

    6f8b794f3246b0c1e1780bb4d4d5dc53

    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.