Development of unenhanced CT-based imaging signature for BAP1 mutation status prediction in malignant pleural mesothelioma: Consideration of 2D and 3D segmentation

Our plain-English summary of Development of unenhanced CT-based imaging signature for BAP1 mutation status prediction in malignant pleural mesothelioma: Consideration of 2D and 3D segmentation, by X.J. Xie, S.Y. Liu, J.Y. Chen, et al (Lung Cancer, 2021).

The study

• Authors: X.J. Xie, S.Y. Liu, J.Y. Chen, Y. Zhao, J. Jiang, L. Wu, X.W. Zhang, Y. Wu, H. Duan, B. He, H. Luo, D. Han
• Journal: Lung Cancer
• Published: 2021;157:30-39
• Cited by: 20 articles, as counted by NIH iCite
• Read it: PubMed · Publisher

Abstract

Objectives: We aimed to explore the feasibility of 2D and 3D radiomics signature based on the unenhanced computed tomography (CT) images to predict BRCA1-associated protein 1 (BAP1) gene mutation status for malignant pleural mesothelioma (MPM) patients.

Materials And Methods: 74 patients with MPM were retrospectively enrolled (22 mutant BAP1, 52 wild-type BAP1 demonstrated by Sanger sequencing). The radiomic features were extracted respectively from the 2D and 3D segmentation of unenhanced pre-treatment CT images, and the dataset was randomly divided into training (n = 51) and test (n = 23) sets for radiomics model development and internal validation. The synthetic minority over-sampling technique (SMOTE) was used for data balancing in the training set. 2D or 3D features were sequentially selected by ICC > 0.8, correlation analysis (cut-value 0.7), univariate analysis or univariate logistic regression (LR), which were involved into multivariate LR for LR model construction. Following the comparison of the 2D and 3D models by the ROC analysis and Delong test for AUC, the calibration and clinical utility of 2D and 3D models were evaluated.

Results: 3D radiomic features showed better ICCs compared with 2D in both intra- (P < 0.001) and inter-observer (P < 0.001) analysis. 3D radiomic model based on selected features developed from a balanced training dataset presented a favorable predictive performance with AUC of 0.786 and 0.768 in the training and test sets, respectively. The predictive performance of 3D model was superior to 2D model (1 feature) both in the training (AUC 0.786 vs. 0.683, P = 0.036) and the test (AUC 0.768 vs.0.652, P = 0.441) set. The calibration curve and decision curves also indicate a better BAP1 prediction performance and clinical benefit for 3D model than that of 2D model.

Conclusion: The developed unenhanced CT-based 3D radiomics signature is potential as a noninvasive marker for predicting BAP1 mutation status.

Copyright © 2021. Published by Elsevier B.V.

In plain English

This retrospective study examined 74 patients with malignant pleural mesothelioma in Yunnan, China, who had pre-treatment CT without contrast in 2017 to 2019. Of these, 22 (29.7 %) had BAP1-mutated tumors and 52 did not. The authors built models from image features of one tumor slice (2D) or the whole tumor (3D). They trained the models on 51 patients plus computer-generated balancing cases and tested them on 23 others. 3D features were more repeatable when tumors were outlined again. The 3D model's area under the curve (AUC) was 0.786 (95 % CI 0.699 to 0.873) in training and 0.768 (0.508 to 1.0) in testing. The 2D model's was 0.683 (0.581 to 0.785) and 0.652 (0.310 to 0.993). Only the training difference was significant. No clinical factor differed significantly by BAP1 status. Tumor engulfing calcified pleural plaques was more common with the mutation: 5 of 22 versus 3 of 52 overall (significant), and 3 of 15 versus 2 of 36 in training (not significant). Individual asbestos exposure was not reported or analyzed. The authors concluded the 3D model is a potential tool for predicting BAP1 status, and they called for external validation.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Retrospective study developing and internally testing CT-based radiomics models (2D and 3D) to predict BAP1 gene mutation status
• Who was studied: Of 120 patients with pathologically confirmed malignant pleural mesothelioma who had pre-treatment chest CT between 2017 and 2019 at the 1st Affiliated Hospital of Kunming Medical University and the People's Hospital of Chuxiong Yi Autonomous Prefecture, China, 74 were included: 43 males and 31 females, mean age 55 (range 22 to 77). 22 had mutant BAP1 and 52 had wild-type BAP1. 51 were assigned to a training set and 23 to a test set.
• Exposure: The study looked at image features from unenhanced (no contrast) pre-treatment chest CT. Tumors were outlined either on the largest single slice (2D) or over the whole tumor (3D). BAP1 status was determined by sequencing tumor samples. Individual asbestos exposure was not reported or analyzed. In the discussion, the authors only supposed that most patients came from Dayao, an area of natural environmental asbestos exposure.
• Main results:
  • BAP1 mutation was found in 22 of 74 tumor samples (29.7 %); 52 (70.3 %) were wild-type.
  • 3D features were more repeatable than 2D features, both within one reader and between readers (P < 0.001 for each). 137 (11.1 %) of the 2D features and 385 (31.3 %) of the 3D features met the stability threshold.
  • 3D model (4 features): AUC 0.786 (95 % CI 0.699 to 0.873) in the training set, which had been balanced with synthetic cases (SMOTE), and 0.768 (95 % CI 0.508 to 1.0) in the test set. In the test set, sensitivity was 71.4 % and specificity 75.0 %.
  • 2D model (1 feature): AUC 0.683 (95 % CI 0.581 to 0.785) in the training set and 0.652 (95 % CI 0.310 to 0.993) in the test set.
  • The 3D model's AUC was higher than the 2D model's in both sets. The difference was statistically significant only in the training set (P = 0.036), not in the test set (P = 0.441).
  • No clinical factor (age, sex, family history of cancer, smoking, histologic subtype, stage) showed a statistically significant difference between mutant and wild-type BAP1 patients in the total and training sets (P > 0.05).
  • Calcified plaque engulfment on CT was more frequent in mutant BAP1 patients in the total set: 5 of 22 (22.7 %) versus 3 of 52 (5.8 %) in wild-type patients, Fisher's exact P = 0.046. The multivariate odds ratio was 7.031 (95 % CI 1.213 to 40.764). In the training set the counts were 3 of 15 (20.0 %) versus 2 of 36 (5.6 %), which was not statistically significant (P = 0.144). For that reason this feature was not included in the model.
• The authors' conclusion: The authors concluded that the radiomic model built from the more reliable 3D features is potentially a promising tool to predict BAP1 mutation status in MPM patients, and that it awaits further confirmation.
• Limitations the authors state:
  • The retrospective design and the rarity of the disease gave a limited and imbalanced dataset. External validation with more cases is needed.
  • Imaging data were not uniform because of different acquisition protocols.
  • The model was built only on unenhanced CT images. Prospective research with contrast-enhanced images is needed.

Topics

The National Library of Medicine indexes this study under subjects our site covers: Lung cancer, Mesothelioma.

Medical terms in this study, explained in our Lexicon: BAP1, Pleural Mesothelioma.

Related studies

• Intrahepatic cholangiocarcinoma development in a patient with a novel BAP1 germline mutation and low exposure to asbestos — G. Brandi, M. Deserti, et al. (2020)
• Malignant mesothelioma in situ: diagnostic and clinical considerations — E. Pulford, D.W. Henderson, et al. (2020)
• Current evidence base of FDG-PET/CT imaging in the clinical management of malignant pleural mesothelioma: emerging significance of image segmentation and global disease assessment — S. Basu, B. Saboury, et al. (2011)
• Germline mutation of Bap1 accelerates development of asbestos-induced malignant mesothelioma — J. Xu, Y. Kadariya, et al. (2014)
• Biological Mechanisms and Clinical Significance of BAP1 Mutations in Human Cancer — M. Carbone, J.W. Harbour, et al. (2020)
• Tumours associated with BAP1 mutations — R. Murali, T. Wiesner, et al. (2013)

Browse more research in Lung Cancer, or return to medical research.