Elsevier

Lung Cancer

Volume 139, January 2020, Pages 103-110
Lung Cancer

Joint use of the radiomics method and frozen sections should be considered in the prediction of the final classification of peripheral lung adenocarcinoma manifesting as ground-glass nodules

https://doi.org/10.1016/j.lungcan.2019.10.031Get rights and content

Highlights

Abstract

Objectives

To evaluate the diagnostic accuracy of radiomics method and frozen sections (FS) for the pathological classification of peripheral lung adenocarcinoma manifesting as ground-glass nodules (GGNs) in computer tomography (CT).

Materials and methods

A dataset of 831 peripheral lung adenocarcinoma manifesting as GGNs in CT were divided into two cohorts: training cohort (n = 581) and validation cohort (n = 250). Combined with clinical features, the radiomics classifier was trained and validated to distinguish the pathological classification of these nodules. FS diagnoses in the validation cohort were collected. Diagnostic performance of the FS and radiomics methods was compared in the validation cohort. The predictive factors for the misdiagnosis of FS were determined via univariate and multivariate analyses.

Results

The accuracy of radiomics method in the training and validation cohorts was 72.5 % and 68.8 % respectively. The overall accuracy of FS in the validation cohort was 70.0 %. The concordance rate between FS and final pathology when FS had a different diagnosis from radiomics classifier was significantly lower than when FS had the same diagnosis as radiomics classifier (46 vs. 87 %, P < 0.001). Univariate and Multivariate analyses showed that different diagnosis between FS and radiomics classifier was the independent predictive factor for the misdiagnosis of FS (OR: 7.46; 95%CI: 4.00–13.91; P < 0.001).

Conclusions

Radiomics classifier predictions may be a reliable reference for the classification of peripheral lung adenocarcinoma manifesting as GGNs when FS cannot provide a timely diagnosis. Intraoperative diagnoses of the cases where FS had a different diagnosis from radiomics method should be considered cautiously due to the higher misdiagnosis rate.

Introduction

Lung cancer is the most common cancer (11.6 % of total cancers) and the leading cause of death from cancers worldwide (18.4 % of total cancer deaths) [1], with adenocarcinoma being the most common histological type of lung cancer [2,3]. A new classification of lung adenocarcinoma was established in 2011, dividing lung adenocarcinoma into atypical adenomatous hyperplasia (AAH), adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA) and invasive adenocarcinoma (IAC) [4], consistent with the classification of lung adenocarcinoma published by WHO in 2015 [5]. In recent years, the detection rate of pulmonary nodules has been increased rapidly. GGNs is a special type of pulmonary nodule, and compared to solid nodules, the appearance of GGNs suggests a high risk of early lung cancer [6,7]. Except for several studies confirming GGNs as lung squamous cell carcinoma, the majority of malignant GGNs have been confirmed as early lung adenocarcinoma [8,9].

The progression of GGNs from AAH, AIS and MIA to IAC is a continuous and dynamic process accompanied by the increment of malignancy [10]. The intraoperative classification of early lung adenocarcinoma is of great significance in determining surgical strategy [11]. FS currently plays a main role in the intraoperative diagnosis and classification of early lung adenocarcinoma, guiding the surgery treatment. However, the classification accuracy of FS is unsatisfactory, and for some cases, it cannot give timely pathological classification due to the poor quality or sampling problems (FS deferrals), [[12], [13], [14], [15]]. This may lead to an inappropriate surgical strategy and even a second surgery for patients [11,14,15].

Radiomics, as an emerging technology, transforms medical images and information into quantitative data and then extracts many quantitative features that can be used as a non-invasive method for evaluating many tumor characteristics. Many studies have confirmed that radiomics has a high clinical application value [[16], [17], [18]]. Compared to the histopathological method, radiomics can be used as a “virtual biopsy” technique, can quantitatively analyze the entire tumor tissue, and has no sampling location limitation, thus providing richer and more complete tumor information for clinical diagnosis, treatment, and prognostic evaluation [[19], [20], [21], [22]]. The applications of radiomics in the field of lung cancer have been investigated widely, and many studies have confirmed its good prospect [[23], [24], [25], [26]].

In some studies, CT-based radiomics method was used to predict the pathological classification of lung adenocarcinoma manifested as GGNs [27,28], showing the usefulness of radiomics. However, those studies placed more emphasis on preoperative evaluation, and there is a lack of relevant studies focused on the comparison between radiomics method and FS to determine whether radiomics can benefit intraoperative classification. In this study, we compared the classification accuracy of radiomics and FS, and explored the value of joint consideration of those two methods for the intraoperative classification of peripheral lung adenocarcinoma manifesting as GGNs.

Section snippets

Data collection

This retrospective study was approved by the ethics committee of Shanghai Pulmonary Hospital, and the informed consent requirement was waived. A search of patients with peripheral lung adenocarcinoma manifesting as GGNs confirmed by surgery and preoperative CT from January 2017 to August 2018 in our hospital was performed.

The inclusion criteria were as follows: (i) had preoperative examination of CT within one month, (ii) maximum diameter of GGNs were ≤3 cm, (iii) preoperative CT layer

Baseline characteristics

AAH is generally considered benign and treated by follow-up strategy, however, surgical resection is usually recommended for AIS, MIA, and IAC, making AAH resection cases much lower than AIS, MIA, or IAC. Because of the clinical value of distinguishing AAH from AIS, MIA and IAC, AAH was separated into an independent group in this study. Through searching the data from January 1, 2017 to August 1, 2018, 4146 GGNs confirmed as early lung adenocarcinomas were obtained, including 247 cases of AAH,

Discussion

The timely and accurate intraoperative pathological classification of peripheral lung adenocarcinoma directly determines the optimum surgical strategy for patients, especially for the extension of lesions resection. Frozen section errors and deferrals have a significant clinical impact on the surgical treatment of patients with early lung adenocarcinoma, and a second surgery may be conducted after the initial surgery [11,14]. Limited resection is recommended for AIS and MIA, while lobectomy is

Conclusions

This study showed that radiomics classifier exhibited satisfactory classification performance in peripheral lung adenocarcinoma manifesting as GGNs in comparison with FS, and that it could serve as a reliable complementary method when FS can’t give timely pathological classification. In addition, different diagnosis between FS and radiomics was the independent predictive factor for the misdiagnosis of FS (OR: 7.46; P <  0.001). The intraoperative classification of cases where FS had a different

Funding

This work was supported by Natural Science Foundation of Shanghai [grant Number 19ZR1443100] and project of capacity construction of assisting departments of clinic in Shanghai municipal hospital [grant Number SHDC22015037].

The funders have no role in the study design, data collection, data analysis and writing the manuscripts.

Declaration of Competing Interest

The authors declare no potential conflicts of interest.

Acknowledgements

The authors would like to express their sincere gratitude to those contributing to this study who are not identified as authors, as they have lent assistance in the course of this research. We gratefully acknowledge the help of Ms. Liping Zhang and Ms. Yan Huang (Department of pathology, Shanghai Pulmonary Hospital, Tongji University School of Medicine), who helped to correct the diagnoses of the postoperative paraffin sections. Without their help, completion of this study and the manuscript

References (34)

  • X. Yang et al.

    CT-based radiomics signature for differentiating solitary granulomatous nodules from solid lung adenocarcinoma

    Lung Cancer

    (2018)
  • H. Harada et al.

    Functional advantage after radical segmentectomy versus lobectomy for lung cancer

    Ann. Thorac. Surg.

    (2005)
  • F. Bray et al.

    Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries

    CA Cancer J. Clin.

    (2018)
  • H. Ito et al.

    Nonfilter and filter cigarette consumption and the incidence of lung cancer by histological type in Japan and the United States:analysis of 30-year data from population-based cancer registries

    Int. J. Cancer

    (2011)
  • W.D. Travis et al.

    International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society: international multidisciplinary classification of lung adenocarcinoma: executive summary

    Proc. Am. Thorac. Soc.

    (2011)
  • J. Cho et al.

    Surgical resection of nodular ground-glass opacities without percutaneous needle aspiration or biopsy

    BMC Cancer

    (2014)
  • T. Kimura et al.

    Early image of squamous cell carcinoma in the peripheral lung field on high-resolution CT scan

    Nihon Kokyuki Gakkai Zasshi

    (2002)
  • Cited by (24)

    • Use of CT radiomics to differentiate minimally invasive adenocarcinomas and invasive adenocarcinomas presenting as pure ground-glass nodules larger than 10 mm

      2021, European Journal of Radiology
      Citation Excerpt :

      With the increasing research on artificial intelligence and radiomics, the identification of malignant lung nodules through the establishment of radiomics signatures has become a hot topic. Radiomics has been suggested as a “virtual biopsy” technique [10]. On CT images, the diameter, semantic features, and lymph node status help in the evaluation of the probability of invasiveness.

    View all citing articles on Scopus
    1

    These authors contributed equally to this work.

    View full text