Advances in the prediction of spread through air spaces with imaging in lung cancer: a narrative review

Introduction

According to Global Cancer Statistics 2020, lung cancer remains one of the leading causes of cancer death (1). Studies have proved that spread through air spaces (STAS) is an independent risk factor for poor prognosis in lung cancer patients and affects recurrence-free survival (RFS) and overall survival (OS) after surgery. Accordingly, accurate prediction of STAS status before the operation is of great significance for the selection of surgical plan for lung cancer patients. Compared with sublobectomy, lobectomy can improve the prognosis of patients (2). Postoperative pathology is the gold standard for diagnosing STAS, but the selection of clinical treatment regimens for STAS patients lacks timeliness. Over the years, significant emphasis has been placed on diagnosing STAS in lung cancer by preoperative bronchoscopy, percutaneous transthoracic needle biopsy (PTNB) and intraoperative frozen section (FS) pathology. Preoperative bronchial cytology exhibited a limited ability to predict the STAS status of tumors. Besides, percutaneous biopsy yielded a low sensitivity for the diagnosis of STAS and can cause local infection, pneumothorax, vascular injury, hematoma and other complications. The time for intraoperative FS pathological diagnosis of STAS is limited, and it is often challenging for pathologists to definitively diagnose the existence of STAS (3-5). Therefore, noninvasive imaging methods that comprehensively reflect tumor characteristics have been widely researched. This article reviews the recent imaging-related findings on predicting STAS in lung cancer. We present the following article in accordance with the Narrative Review reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-22-2593/rc).

Methods

A non-systematic review of the latest literature to predict the status of STAS in lung cancer patients was carried out. Relevant articles in English available in the PubMed database as at 27 June 2022 were included. Search terms included (“STAS”) AND (“lung cancer” OR “CT” OR “¹⁸F-FDG PET/CT” OR “radiomics” OR “predicted” OR “prognosis”) (Table 1).

Discussion

Definition

In 2013, Onozato et al. (6) proposed the definition of “tumor islands” as clusters of free tumor cells in the alveolar spaces. Studies have shown that they are associated with a poor 5-year RFS rate for patients with lung adenocarcinoma (ADC). Active intervention is needed for patients with “tumor islands”. In 2015, Kadota et al. (2) did a study on 411 patients with stage I lung ADC 2 cm or less and defined the STAS, which indicated the presence of tumor cells within the air spaces in the lung parenchyma beyond the edge of the main tumor in the form of micropapillary clusters, solid nests, or single cells. Warth et al. (7) retrospectively analyzed the data of 569 patients with stage I-IV lung ADC and found that approximately half of the lung ADC had focal or diffuse STAS and tumor cell clusters with a length of less than 3 alveolar spaces from the main lesion (at least 5 cells in each cluster) were defined as focal STAS, otherwise, it was diffuse STAS. Based on the aforementioned studies on lung ADC, in the same year, the World Health Organization (WHO) officially identified STAS as the fourth type of invasion of lung ADC (8), in addition to the other three types of invasions (non-lepidic growth, stromal infiltration, vascular and pleural infiltration). In addition, some studies had found that STAS was not only present in lung ADC (9-12). In 2017, Lu et al. (9) retrospectively analyzed 445 cases of stage I-III lung squamous cell carcinoma (SCC) and found that STAS was present in one-third of them. With more advanced clinical staging, the frequency of the appearance of STAS increased. Subsequent studies (10-12) have shown that STAS is also present in small cell lung carcinoma (SCLC), lung pleomorphic carcinoma (PC), and lung neuroendocrine tumors (NETs). Therefore, an objective definition of STAS can improve the understanding of the correlation between STAS and clinical prognosis.

Clinical significance

Since 1960, lobectomy has been the standard surgical treatment method for stage I non-small cell lung carcinoma (NSCLC) (13). In 2006, Okada et al. (14) conducted a multicenter trial and found that for patients with a peripheral cT1N0M0 non-small cell lung cancer of 2 cm or less, the 5-year disease-free survival (DFS) rate and OS rate for patients after sublobar resection were similar to those who underwent lobar resection, but the postoperative lung function of patients that underwent sublobar resection was significantly better. Therefore, sublobar resection should be considered as an alternative for stage IA non-small cell lung cancer 2 cm or less, even in low-risk patients. However, an increasing body of evidence suggests that for STAS-positive stage I lung ADC patients, sublobectomy yields a higher risk of recurrence than lobectomy, suggesting that STAS is associated with a poor prognosis in patients undergoing sublobectomy (15-17). Zhong et al. (18) found that the positive rates of STAS in lung ADC with solid nodules and subsolid nodules were 34.9% and 10.8%, ground glass opacity (GGO) components in the subsolid nodules with positive STAS were independently associated with good prognosis of patients. Kadota et al. (2) showed that STAS and surgical margins less than 1.0 cm from the tumor edge were important risk factors for local recurrence after sublobectomy of early lung cancer, when the resection margin was greater than 2.0 cm from the tumor edge, the STAS status was not correlated with local tumor recurrence, therefore, it is recommended that when sublobectomy is performed for patients with early lung cancer, a wider surgical margin should be set if STAS is involved. Chen et al. (19) showed that for patients with stage IA lung ADC who underwent sublobectomy combined with STAS positivity and patients with stage IB lung ADC combined with STAS positivity, adjuvant chemotherapy could achieve a better prognosis. In addition, some scholars have found that STAS is associated with poor prognosis of other types of lung cancer. Yanagawa et al. (20) retrospectively analyzed the clinicopathological data of 220 patients with lung SCC and found that for patients with stage I lung SCC, the 5-year RFS and OS rates of STAS-positive patients were significantly lower than STAS-negative patients, in patients with stage II and III lung SCC, STAS is not associated with recurrence or survival rate. Toyokawa et al. (10) analyzed the data of 30 patients with lung SCLC and found that patients with lung SCLC were prone to STAS, with a positive rate of 83%. Yokoyama et al. (11) studied the clinical and pathological data of 35 patients with lung PC and found that the RFS and OS of STAS-positive patients were significantly lower than STAS-negative patients, STAS was an independent risk factor for poor clinical prognosis by multivariate analysis. Aly et al. (12) studied the prognosis of 487 patients with pathological stage I-III lung NETs and found that STAS was related to an earlier occurrence of distant metastasis and a higher incidence of lung cancer-specific death. In summary, STAS positivity affects RFS and OS rates in lung cancer patients. Therefore, accurate prediction of STAS for lung cancer in the preoperative period is of great significance for selecting the optimal surgical approach (lobectomy or sublobectomy) and postoperative adjuvant therapy.

Computed tomography (CT) signs

Interestingly, a number of studies have indicated that some CT morphological characteristics of lung cancer are correlated with STAS. A retrospective study by Kim et al. (21) reported that STAS was more common in lung ADC with solid nodules, and the consolidation tumor ratio (CTR) was an independent predictor of STAS, with a threshold of 90%, the sensitivity of predicting STAS positivity was 89.2%, and the specificity was 60.3%. de Margerie-Mellon et al. (22) studied subsolid nodules and found that the overall size of the lesion, the size of the solid component, and the percentage of the solid component were positively correlated with STAS positivity. Toyokawa et al. (23) found that a positive pleural notch sign and negative ground-glass opacities were positively correlated with STAS positivity. Li et al. (24) studied 578 patients with lung ADC and constructed a multivariate logistic regression prediction model that included age, the maximum diameter, and the solid component ratio, with an area under the curve (AUC) value of 0.801 for predicting STAS. Qi et al. (25) found that CTR, ground glass ribbon, pleural connection, and absence of cystic airspaces were independent predictors of STAS, and the AUC value of STAS predicted by CTR was 0.760. Given that the evaluation of CT signs mostly relies on radiologists’ experience, there is a certain degree of subjectivity, and different inclusion and exclusion criteria may lead to different conclusions (26). Therefore, the value of predicting STAS status using CT may be limited, and the results would need further verification.

18-fluorine-fluorodeoxyglucose positron emission tomography/CT (¹⁸F-FDG PET/CT)

It is well-established that ¹⁸F-FDG PET/CT can provide information on tumor morphology and the anatomical relationship with the adjacent structures and reflecting the metabolic activity of the tumor, and thus suggesting the tumor invasiveness (27). Suh et al. (28) did a study on 387 patients with NSCLC 2 cm or less. By combining preoperative imaging information and intraoperative frozen pathological results, a decision-making flowchart for predicting STAS was constructed, including the percentage of GGO on CT, the ratio of tumor/liver standard uptake values on ¹⁸F-FDG PET/CT, the intraoperative histopathology of lepidic predominance, yielding positive sensitivity, specificity, and negative predictive values of 79.3%, 68.5% and 89.5%, respectively. Wang et al. (29) retrospectively analyzed the preoperative ¹⁸F-FDG PET/CT-related data of 127 patients with early-stage lung ADC, including SUVmax, SUVmean, SUVpeak, metabolic tumor volume (MTV), and total lesion glycolysis (TLG), diameter and computed tomography volume (CTV), STAS risk prediction model developed by multivariate logistic regression analysis showed that SUV and TLG were valuable indicators for predicting STAS, the specificity and accuracy of the prediction model were 88.6% and 71.1%, respectively. Nishimori et al. (30) retrospectively analyzed the preoperative ¹⁸F-FDG PET/CT-related parameters of 52 patients (55 lesions) to predict the STAS status of clinical stage I lung ADC, the results showed a significant variation in SUVmax between STAS-positive and STAS-negative groups, while there was no significant difference in MTV and TLG, receiver operating characteristic (ROC) curve analysis showed that when the SUVmax threshold was set to 2.48, the SUVmax prediction sensitivity was 89.5%, the specificity was 52.8%, and the AUC value was 0.74. Current studies have shown that SUVmax is an important indicator for predicting STAS. However, it should be noted that these studies were mostly retrospective, highlighting the possibility of bias during case inclusion, and these study results were not validated. Accordingly, the potential of ¹⁸F-FDG PET/CT in predicting STAS and evaluating the prognosis of patients requires further studies.

Radiomics

Based on the analysis of CT and MRI images, radiomics extracts internal quantitative features and converts them into data for further mining in order to obtain the hidden biological values from the data (31). Over the years, some scholars have attempted to use radiomics to study STAS. Chen et al. (32) performed image segmentation and delineation on the lesions of 233 patients with stage I lung ADC and extracted important radiomics features to construct a model to predict STAS status, the AUC values of the internal and the external validation models were 0.63 and 0.69, respectively, indicating that CT-based radiomics has a definite value in the preoperative prediction of STAS in stage I lung ADC. Jiang et al. (33) performed univariate logistic regression analysis to screen the radiomics characteristics associated with STAS in 462 lung ADC patients and constructed a radiomics-based random forest (RF) machine learning model to predict STAS status, the 12 radiomics characteristics and patient age were found to be significantly correlated with STAS, and the RF model yielded an AUC value of 0.754 for predicting STAS. Zhuo et al. (34) performed radiomics feature extraction on 212 cases of tumor lesion area and peritumoral area (5, 10, 15 mm), combined with clinical features and made use of multivariate logistic regression to establish a radiomics nomogram to predict STAS, the maximum diameter of the solid component and mediastinal lymph node enlargement were identified as independent predictors of STAS, the AUC value of the training and validation sets were 0.98 and 0.99, respectively. Qi et al. (35) conducted a retrospective study of 216 lung ADC patients, based on the morphological signs of the lesion and the intratumoral and peritumoral radiomics characteristics, the ability of the single model and the comprehensive model to predict the STAS of lung ADC was compared, the AUC values of the comprehensive model training set, internal validation set and external validation set were 0.907, 0.897 and 0.909, respectively, which were significantly higher than those of the single model. Indeed, prediction models based on radiomics require strict procedures, including detection, segmentation, feature extraction and selection, and the operation process is complex and time-consuming (36). The clinical utility of CT-based radiomics in predicting STAS needs to be further investigated.

Deep learning (DL)

DL, also known as deep neural network learning, has shown great prospects in predicting the invasiveness of lung ADC by constructing multi-hierarchical nonlinear processing units to perform multi-level feature learning and improve the accuracy of disease diagnosis (37). Tao et al. (38) performed lesion segmentation and delineation on 203 cases of NSCLC patients with preoperative enhanced thin-slice CT, extracted radiomics features, and constructed a clinicopathological/CT model, a conventional radiomics model, a computer vision model, three-dimensional (3D) convolutional neural network (CNN) model and a combination model composed of 3D CNN model and clinicopathological/CT model to predict STAS status of NSCLC, the results showed that the 3D CNN model yielded superior performance than other models, with AUC values of 0.93 and 0.80 in the training and validation sets. At present, only one study has been conducted on the prediction of lung cancer STAS based on DL, and it is widely thought that its prediction efficiency has great room for improvement. In the future, better neural network models have to be constructed for improving the prediction efficiency.

Conclusions

STAS is a key invasion type of lung cancer, not only found in lung ADC but also in lung SCC, lung SCLC, lung PC, and lung NETs. For patients with early STAS-positive lung cancer, lobectomy is shown to have positive impact on patient prognosis. Although the gold standard for STAS detection and diagnosis is based on histopathological approach, preoperative CT and ¹⁸F-FDG PET/CT have certain values in predicting the status of STAS, thus assist clinicians to select the optimal surgical approach and postsurgical treatment. Meanwhile, the prediction of STAS based on radiomics and DL represents a future research direction. At present, the mechanism of STAS remains controversial. Little is currently known about whether STAS is a key invasive pattern of lung cancer, or whether STAS can be used as a reference indicator in TNM staging of lung cancer, highlighting the need for further research. Most of the previous studies were based on retrospective analysis, and their findings were not externally validated. Accordingly, their clinical practicability needs to be further explored.

Acknowledgments

The authors thank Dr. Zhengwei Zhang of the Department of Pathology, Second Affiliated Hospital of Navy Medical University, Shanghai, China for his selfless and valuable assistance in this article.

Funding: This research was funded by National Key R&D Program of China (Nos. 2022YFC2010000 and 2022YFC2010002), Key Program of National Natural Science Foundation of China (No. 81930049), National Natural Science Foundation of China (No. 82171926), Shanghai Science and Technology Innovation Action Plan Program (No. 19411951300), Clinical Innovative Project of Shanghai Changzheng Hospital (No. 2020YLCYJ-Y24), and Program of Science and Technology Commission of Shanghai Municipality (No. 21DZ2202600).

Footnote

Reporting Checklist: The authors have completed the Narrative Review reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-22-2593/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-22-2593/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-22-2593/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
2. Kadota K, Nitadori JI, Sima CS, et al. Tumor Spread through Air Spaces is an Important Pattern of Invasion and Impacts the Frequency and Location of Recurrences after Limited Resection for Small Stage I Lung Adenocarcinomas. J Thorac Oncol 2015;10:806-14. [Crossref] [PubMed]
3. Medina MA, Onken AM, de Margerie-Mellon C, et al. Preoperative bronchial cytology for the assessment of tumor spread through air spaces in lung adenocarcinoma resection specimens. Cancer Cytopathol 2020;128:278-86. [Crossref] [PubMed]
4. Cao L, Jia M, Sun PL, et al. Histopathologic features from preoperative biopsies to predict spread through air spaces in early-stage lung adenocarcinoma: a retrospective study. BMC Cancer 2021;21:913. [Crossref] [PubMed]
5. Walts AE, Marchevsky AM. Current Evidence Does Not Warrant Frozen Section Evaluation for the Presence of Tumor Spread Through Alveolar Spaces. Arch Pathol Lab Med 2018;142:59-63. [Crossref] [PubMed]
6. Onozato ML, Kovach AE, Yeap BY, et al. Tumor islands in resected early-stage lung adenocarcinomas are associated with unique clinicopathologic and molecular characteristics and worse prognosis. Am J Surg Pathol 2013;37:287-94. [Crossref] [PubMed]
7. Warth A, Muley T, Kossakowski CA, et al. Prognostic Impact of Intra-alveolar Tumor Spread in Pulmonary Adenocarcinoma. Am J Surg Pathol 2015;39:793-801. [Crossref] [PubMed]
8. Travis WD, Brambilla E, Nicholson AG, et al. The 2015 World Health Organization Classification of Lung Tumors: Impact of Genetic, Clinical and Radiologic Advances Since the 2004 Classification. J Thorac Oncol 2015;10:1243-60. [Crossref] [PubMed]
9. Lu S, Tan KS, Kadota K, et al. Spread through Air Spaces (STAS) Is an Independent Predictor of Recurrence and Lung Cancer-Specific Death in Squamous Cell Carcinoma. J Thorac Oncol 2017;12:223-34. [Crossref] [PubMed]
10. Toyokawa G, Yamada Y, Tagawa T, et al. High Frequency of Spread Through Air Spaces in Resected Small Cell Lung Cancer. Anticancer Res 2018;38:1821-5. [PubMed]
11. Yokoyama S, Murakami T, Tao H, et al. Tumor Spread Through Air Spaces Identifies a Distinct Subgroup With Poor Prognosis in Surgically Resected Lung Pleomorphic Carcinoma. Chest 2018;154:838-47. [Crossref] [PubMed]
12. Aly RG, Rekhtman N, Li X, et al. Spread Through Air Spaces (STAS) Is Prognostic in Atypical Carcinoid, Large Cell Neuroendocrine Carcinoma, and Small Cell Carcinoma of the Lung. J Thorac Oncol 2019;14:1583-93. [Crossref] [PubMed]
13. Yang HX, Woo KM, Sima CS, et al. Long-term Survival Based on the Surgical Approach to Lobectomy For Clinical Stage I Nonsmall Cell Lung Cancer: Comparison of Robotic, Video-assisted Thoracic Surgery, and Thoracotomy Lobectomy. Ann Surg 2017;265:431-7. [Crossref] [PubMed]
14. Okada M, Koike T, Higashiyama M, et al. Radical sublobar resection for small-sized non-small cell lung cancer: a multicenter study. J Thorac Cardiovasc Surg 2006;132:769-75. [Crossref] [PubMed]
15. Shiono S, Endo M, Suzuki K, et al. Spread Through Air Spaces Is a Prognostic Factor in Sublobar Resection of Non-Small Cell Lung Cancer. Ann Thorac Surg 2018;106:354-60. [Crossref] [PubMed]
16. Kadota K, Kushida Y, Kagawa S, et al. Limited Resection Is Associated With a Higher Risk of Locoregional Recurrence than Lobectomy in Stage I Lung Adenocarcinoma With Tumor Spread Through Air Spaces. Am J Surg Pathol 2019;43:1033-41. [Crossref] [PubMed]
17. Ren Y, Xie H, Dai C, et al. Prognostic Impact of Tumor Spread Through Air Spaces in Sublobar Resection for 1A Lung Adenocarcinoma Patients. Ann Surg Oncol 2019;26:1901-8. [Crossref] [PubMed]
18. Zhong Y, Xu Y, Deng J, et al. Prognostic impact of tumour spread through air space in radiological subsolid and pure solid lung adenocarcinoma. Eur J Cardiothorac Surg 2021;59:624-32. [Crossref] [PubMed]
19. Chen D, Wang X, Zhang F, et al. Could tumor spread through air spaces benefit from adjuvant chemotherapy in stage I lung adenocarcinoma? A multi-institutional study. Ther Adv Med Oncol 2020;12:1758835920978147. [Crossref] [PubMed]
20. Yanagawa N, Shiono S, Endo M, et al. Tumor spread through air spaces is a useful predictor of recurrence and prognosis in stage I lung squamous cell carcinoma, but not in stage II and III. Lung Cancer 2018;120:14-21. [Crossref] [PubMed]
21. Kim SK, Kim TJ, Chung MJ, et al. Lung Adenocarcinoma: CT Features Associated with Spread through Air Spaces. Radiology 2018;289:831-40. [Crossref] [PubMed]
22. de Margerie-Mellon C, Onken A, Heidinger BH, et al. CT Manifestations of Tumor Spread Through Airspaces in Pulmonary Adenocarcinomas Presenting as Subsolid Nodules. J Thorac Imaging 2018;33:402-8. [Crossref] [PubMed]
23. Toyokawa G, Yamada Y, Tagawa T, et al. Computed tomography features of resected lung adenocarcinomas with spread through air spaces. J Thorac Cardiovasc Surg 2018;156:1670-1676.e4. [Crossref] [PubMed]
24. Li C, Jiang C, Gong J, et al. A CT-based logistic regression model to predict spread through air space in lung adenocarcinoma. Quant Imaging Med Surg 2020;10:1984-93. [Crossref] [PubMed]
25. Qi L, Xue K, Cai Y, et al. Predictors of CT Morphologic Features to Identify Spread Through Air Spaces Preoperatively in Small-Sized Lung Adenocarcinoma. Front Oncol 2021;10:548430. [Crossref] [PubMed]
26. Ledda RE, Milanese G, Gnetti L, et al. Spread through air spaces in lung adenocarcinoma: is radiology reliable yet? J Thorac Dis 2019;11:S256-61. [Crossref] [PubMed]
27. Volpi S, Ali JM, Tasker A, et al. The role of positron emission tomography in the diagnosis, staging and response assessment of non-small cell lung cancer. Ann Transl Med 2018;6:95. [Crossref] [PubMed]
28. Suh JW, Jeong YH, Cho A, et al. Stepwise flowchart for decision making on sublobar resection through the estimation of spread through air space in early stage lung cancer. Lung Cancer. 2020;142:28-33. [Crossref] [PubMed]
29. Wang XY, Zhao YF, Yang L, et al. Correlation analysis between metabolic tumor burden measured by positron emission tomography/computed tomography and the 2015 World Health Organization classification of lung adenocarcinoma, with a risk prediction model of tumor spread through air spaces. Transl Cancer Res 2020;9:6412-22. [Crossref] [PubMed]
30. Nishimori M, Iwasa H, Miyatake K, et al. 18F FDG-PET/CT analysis of spread through air spaces (STAS) in clinical stage I lung adenocarcinoma. Ann Nucl Med 2022;36:897-903. [Crossref] [PubMed]
31. Kumar V, Gu Y, Basu S, et al. Radiomics: the process and the challenges. Magn Reson Imaging 2012;30:1234-48. [Crossref] [PubMed]
32. Chen D, She Y, Wang T, et al. Radiomics-based prediction for tumour spread through air spaces in stage I lung adenocarcinoma using machine learning. Eur J Cardiothorac Surg 2020;58:51-8. [Crossref] [PubMed]
33. Jiang C, Luo Y, Yuan J, et al. CT-based radiomics and machine learning to predict spread through air space in lung adenocarcinoma. Eur Radiol 2020;30:4050-7. [Crossref] [PubMed]
34. Zhuo Y, Feng M, Yang S, et al. Radiomics nomograms of tumors and peritumoral regions for the preoperative prediction of spread through air spaces in lung adenocarcinoma. Transl Oncol 2020;13:100820. [Crossref] [PubMed]
35. Qi L, Li X, He L, et al. Comparison of Diagnostic Performance of Spread Through Airspaces of Lung Adenocarcinoma Based on Morphological Analysis and Perinodular and Intranodular Radiomic Features on Chest CT Images. Front Oncol 2021;11:654413. [Crossref] [PubMed]
36. Aerts HJ. The Potential of Radiomic-Based Phenotyping in Precision Medicine: A Review. JAMA Oncol 2016;2:1636-42. [Crossref] [PubMed]
37. Zhao W, Yang J, Sun Y, et al. 3D Deep Learning from CT Scans Predicts Tumor Invasiveness of Subcentimeter Pulmonary Adenocarcinomas. Cancer Res 2018;78:6881-9. [Crossref] [PubMed]
38. Tao J, Liang C, Yin K, et al. 3D convolutional neural network model from contrast-enhanced CT to predict spread through air spaces in non-small cell lung cancer. Diagn Interv Imaging 2022;103:535-44. [Crossref] [PubMed]