Comparison of Lung-RADS Version 1.1 and Lung-RADS Version 2022 in Classifying Airway Nodules Detected at Lung Cancer Screening CT

Abstract

Purpose

To compare the Lung Imaging Reporting and Data System (Lung-RADS) version 1.1 with version 2022 classification of airway nodules detected at lung cancer screening CT examinations.

Materials and Methods

This retrospective study included all patients who underwent a lung cancer screening CT examination in the authors’ health care network between 2015 and 2021 with a reported airway or endobronchial nodule. A fellowship-trained cardiothoracic radiologist reviewed these CT images and characterized the airway nodules by size, location, multiplicity, morphology, dependent portions of airway, internal air, fluid attenuation, distal changes, outcome at follow-up, and final pathologic diagnosis, if malignant. Sensitivity and specificity of Lung-RADS version 1.1 in detecting malignant nodules were compared with those of Lung-RADS version 2022 using the McNemar test.

Results

A total of 174 patients were included. Of these, 163 (94%) had airway nodules that were deemed benign, while 11 (6%) had malignant nodules. Airway nodules in the trachea and mainstem bronchi were all benign, while lobar and segmental airway nodules had the highest risk for lung cancer (17.2% and 11.1%, respectively). Of the 12 subsegmental airway nodules that were obstructive, three (25%) were malignant and nine (75%) were benign. Nodules with nonobstructive morphologies, dependent portions of airway, internal air, or fluid attenuation were all benign. Only 10 of the 92 (10.9%) patients with positive Lung-RADS by clinical report had cancer. Lung-RADS version 2022 resulted in higher specificity than version 1.1 (82% vs 50%, P < .001), without sacrificing sensitivity (91% for both).

Conclusion

Compared with the previous version, Lung-RADS version 2022 reduced the number of false-positive screening CT examinations while still identifying malignant airway nodules.

Keywords: CT, Lung, Primary Neoplasms, Pulmonary, Lung Cancer Screening, Lung-RADS, Nodule Risk, Airway Nodule, Endobronchial Nodule

© RSNA, 2024

Summary

Lung-RADS version 2022 provides an improved classification system of airway nodules compared with version 1.1, reducing the number of false-positive studies while still identifying malignant airway nodules based on various imaging features.

Key Points

• ■ In a comparison of the diagnostic performance of Lung-RADS version 1.1 to Lung-RADS version 2022 for airway nodules detected at lung cancer screening CT examinations, both versions exhibited similar sensitivity (91%), but version 2022 demonstrated improved specificity (82% vs 50%).

• ■ More than a quarter of the malignancies in this study occurred in subsegmental bronchi.

• ■ The majority of airway nodules are benign (94% in this study), with supportive features including location in the trachea or mainstem bronchi, multiplicity, nonobstructive morphologies, dependent portions of airway, internal air, and fluid attenuation.

Introduction

Airway nodules, also known as endobronchial nodules, have both malignant and benign causes (1–6). Mucus in the airway is commonly encountered on CT images and tends to be of low attenuation, may have a “bubbly” appearance from mixing with air, and often but not always occurs along dependent portions of the airway (1,7). Infection may cause tubular or branching nodular foci in the distal airways from mucus impaction, which are generally transient (1). If an endobronchial nodule persists, it may be neoplastic in nature. Benign tracheobronchial neoplasms are exceedingly rare (1,4–6). Malignant tracheobronchial neoplasms include squamous cell carcinoma, carcinoid tumors, and minor salivary gland tumors; these may present as a polypoid lesion, a focal sessile lesion, eccentric narrowing of the airway lumen, or circumferential wall thickening (1,4–6). Accurate differentiation between benign and malignant airway nodules based on CT imaging features is crucial for optimal management of patients undergoing lung cancer screening CT, particularly for reducing false-positive studies that may lead to unnecessary interventions such as bronchoscopy.

The American College of Radiology Lung Imaging Reporting and Data System (Lung-RADS) (8) is a quality assurance tool designed to standardize lung cancer screening CT reporting and management recommendations. The previous Lung-RADS version 1.1 classified all endobronchial nodules as category 4A (suspicious) lesions. However, the new Lung-RADS version 2022 introduced revised classification and management criteria for airway nodules, incorporating factors such as location and stability or growth. In the Lung-RADS version 2022 guidance table, airway nodules at baseline are classified as category 2 or 4A; a subsegmental airway nodule at baseline, new, or stable is classified as category 2; and a segmental or more proximal airway nodule at baseline is classified as category 4A. The guidance is somewhat ambiguous regarding subsegmental airway findings, noting that “if no underlying obstructive nodule is identified, these lesions may be classified as category 0 (likely infectious or inflammatory) or 2 (benign)”; however, no specific guidance is given regarding obstructive subsegmental airway nodules.

In this study, we compared the diagnostic accuracy of Lung-RADS version 1.1 with that of Lung-RADS version 2022 for airway nodules detected at lung cancer screening CT examinations. Additionally, we sought to evaluate the cancer risk associated with various airway nodule characteristics.

Materials and Methods

This retrospective Health Insurance Portability and Accountability Act–compliant study was approved by the institutional review board with a waiver of informed consent.

Patient Selection

We searched radiology reports of patients who underwent lung cancer screening CT within our health care network (which included two academic sites and one community site) between January 2015 and November 2021 to identify patients who had a reported airway nodule. A total of 9068 patients underwent 21 954 lung cancer screening CT examinations during this period. The lung cancer screening CT examinations were performed with multidetector helical (spiral) technique in a single breath hold, on full inspiration, and without the use of intravenous contrast medium. Axial images from the lung apices to the costophrenic sulci were acquired and viewed at 3.0-mm section thickness and 1.0-mm section thickness, with reconstruction intervals of 3.0 mm and 0.8 mm, respectively; multiplanar reconstruction was also utilized.

The search identified reports with “bronchus” or “endobronchial” in the Impression section of the report, yielding 299 reports in 251 patients. A thoracic radiologist (M.M.H.) with 7 years of posttraining experience reviewed the reports to verify that there was a true endobronchial nodule finding. This review yielded a total of 178 patients. Four patients were excluded because they had no follow-up in our system. The final study cohort consisted of 174 patients (Fig 1). Airway nodules that were described in the impression but were not the Lung-RADS category–defining lesion were reclassified as category 2 for the purposes of our analysis. For examinations with multiple airway nodules, only the nodule determining the overall category assigned in the report was considered in the analysis. Patient demographics were extracted from the electronic medical record.

Figure 1:

Cohort flow diagram shows inclusion and exclusion criteria.

Report Lung-RADS, Lung-RADS Version 1.1, and Lung-RADS Version 2022 Categories

The examination findings had been interpreted using Lung-RADS version 1.0 given the period in which the study was conducted. The Lung-RADS categories were extracted automatically from the clinical reports. Clinical report Lung-RADS categories were reclassified as “negative” (category 2) if the finding was not category-defining, and otherwise “positive” (category 4A). We calculated Lung-RADS version 1.1 and Lung-RADS version 2022 categories per imaging features (see below) using the guidance documents for those versions. For Lung-RADS version 2022, airway nodules were classified as category 2 if subsegmental in location and not obstructive, or they contained internal air, or there were multiple nodules.

Retrospective Determination of Airway Nodule Characteristics

A fellowship-trained cardiothoracic radiologist (A.K.D.) with 1 year of posttraining experience independently reviewed the images of each patient. The radiologist accessed the clinical reports to identify the airway nodule and was blinded to subsequent imaging and clinical diagnoses upon first imaging review. The radiologist characterized the airway nodule by size (the mean of the long- and short-axis diameters of the nodule), location (trachea, mainstem, lobar, segmental, subsegmental), multiplicity (single or multiple), morphology (obstructive, pedunculated, sessile, tubular or branching, weblike), dependent portions of the airway (yes or no), internal air (yes or no), fluid attenuation (yes, no, too small to assess), and distal findings (atelectasis, consolidation, tree-in-bud, none).

After review and characterization of the airway nodule was completed, the subsequent chest CT examination for each patient was reviewed to determine the outcome at follow-up of the described airway nodule. The outcome at follow-up was characterized as either resolved, decreased, stable, grew, or patient deceased.

Reference Standard

Electronic medical records were manually reviewed for each patient to determine whether a bronchoscopy was performed for further evaluation of the identified airway nodule. The final diagnosis of the airway nodule was characterized as benign if resolved or decreased on first subsequent follow-up, stable on first subsequent follow-up yet resolved or decreased on later follow-up, or with benign findings at bronchoscopy. The final diagnosis of the airway nodule was characterized as malignant if it grew at follow-up, if malignancy was determined with bronchoscopy or other biopsy, or if malignancy was presumed clinically based on aggressive CT imaging features and thoracic surgery clinical evaluation in the deceased patient.

Statistical Analysis

Study data were collected and managed using REDCap (Research Electronic Data Capture, version 11, REDCap Consortium) (9,10), a secure web-based software platform with electronic data capture tools hosted at the study institution. Study data were then analyzed in JMP Pro (version 17, SAS Institute). The frequency of lung cancer was summarized and stratified by airway nodule characteristics. Comparison of cancer rates across categorical variables was performed with the Fisher exact test; for size, the P value was derived from logistic regression. Receiver operating characteristic curve analysis was performed for lung cancer risk by nodule diameter, and the optimal cutoff was identified using the Youden index. Lung cancer risk was also stratified by report (Lung-RADS, Lung-RADS version 1.1, and Lung-RADS version 2022), and odds ratios were calculated. Comparison of sensitivity and specificity between Lung-RADS schemes was performed with a McNemar test. A P value of less than .05 was considered statistically significant.

Results

Characteristics of Patients and Airway Nodules

Characteristics of the 174 patients (98 males and 76 females; median age, 67 years [range, 55–79]) and airway nodules are presented in Table 1. The median nodule diameter was 5 mm (range, 2–29 mm). Of the 174 airway nodules, 18 (10.3%) were in the trachea, 41 (23.6%) in the mainstem bronchus, 29 (16.7%) in the lobar bronchus, 27 (15.5%) in the segmental bronchus, and 59 (33.9%) in the subsegmental bronchus. Airway nodules were more commonly multiple (112 [64.4%]), nondependent (134 [77.0%]), and without internal air (137 [78.7%]). The airway nodules demonstrated a range of morphologies, with sessile and tubular or branching being the most common. According to the original clinical interpretations, two (1.1%) of the 174 examinations were classified as Lung-RADS category 0, 10 (5.7%) as category 1, 72 (41.4%) as category 2, 12 (6.9%) as category 3, 73 (41.3%) as category 4A, two (1.1%) as category 4B, and three (1.7%) as category 4X. At follow-up, 111 (63.5%) of the 174 nodules resolved, 32 (18.4%) decreased, 24 (13.8%) were stable, six (3.4%) grew, and one (0.6%) was presumed lung cancer in the later deceased patient. Bronchoscopy was performed in 17 patients (9.8%), of which eight yielded benign findings (47.1%) and nine yielded malignant pathology (52.9%). A total of 11 airway nodules (6%) were diagnosed as lung cancer, while 163 (94%) were benign. Of the 11 malignant airway nodules, nine (81.8%) were diagnosed with bronchoscopy, one (9.1%) was presumed malignant clinically, and one (9.1%) was diagnosed based on biopsy of liver metastasis. Characteristics of the malignant airway nodules are presented in Table 2, and three examples of malignant airway nodules are presented in Figures 2–4. An example of a benign airway nodule is presented in Figure 5.

Table 1:

Characteristics of Patients and Airway Nodules

Table 2:

Characteristics of Malignant Airway Nodules

Figure 2:

Annual lung cancer screening CT image (cropped axial image, without contrast agent) in a 70-year-old male patient shows obstructive superior segmental right lower lobe nodule (arrow) with a mean diameter of 19 mm. Final pathology was adenosquamous carcinoma.

Figure 4:

CT images in a 65-year-old female patient show (A) obstructive left upper lobar nodule (arrow) with a mean diameter of 9 mm (initial lung cancer screening CT cropped axial image, without contrast agent) and (B) additional multifocal endobronchial mucus plugging (arrows) in the left lower lobe with distal atelectasis. Lung Imaging Reporting and Data System category 1 was assigned, although multiple endobronchial nodules were described, and a short-interval follow-up CT examination was recommended. Final pathology of the left upper lobe nodule was non–small cell carcinoma, favoring squamous cell carcinoma.

Figure 5:

(A) Lung cancer screening CT image (cropped axial plane image, without contrast agent) in a 66-year-old male patient shows a sessile nodule with internal air in the left mainstem to left upper lobe bronchus (arrow) with a mean diameter of 10 mm. The nodule was assigned as Lung Imaging Reporting and Data System category 4A in the clinical report. (B) Follow-up CT image shows the lesion is resolved.

Figure 3:

Initial lung cancer screening CT image (cropped axial image, without contrast agent) in a 75-year-old female patient shows obstructive superior subsegmental right lower lobe nodule (arrow) with a mean diameter of 8 mm. The nodule was assigned as Lung Imaging Reporting and Data System category 4A in the clinical report. Final pathology was squamous cell carcinoma.

Risk of Lung Cancer by Airway Nodule Characteristics

The risk of lung cancer in airway nodules, as stratified by airway nodule characteristics, is presented in Table 3. Airway nodules in the trachea and mainstem bronchi were all benign, while lobar and segmental airway nodules had the highest risk for lung cancer (five of 29 [17.2%]; and three of 27 [11.1%], respectively; P = .02). Multiple airway nodules were also most likely benign compared with single airway nodules (P < .001). Benign features included nonobstructive morphologies (P < .001), dependent portions of airway (P = .05), internal air (P = .07), and fluid attenuation (P = .001). Obstructive airway nodules had a higher risk of malignancy than nonobstructive airway nodules (11 of 31 [35.5%]) (P < .001). Of the 12 subsegmental airway nodules that were obstructive, three (25%) were malignant and nine (75%) were benign. Distal findings of either consolidation or tree-in-bud nodularity were more frequently associated with a malignant airway nodule than atelectasis or no distal finding (P < .001). Stable or growing airway nodules at follow-up had a higher risk of lung cancer (six of 24 [25.0%]; and four of six [66.7%], respectively; P < .001) compared with resolved or decreased airway nodules, none of which was malignant. Nodule size had an area under the receiver operating characteristic curve of 0.96 (95% CI: 0.91, 0.99; P < .001) for predicting lung cancer, and the optimal size threshold was 8 mm for a malignant airway nodule, with a sensitivity of 100% (11 of 11) and specificity of 82% (113 of 163). Benign airway nodules were overall smaller (median: 5.0 mm, mean: 5.6 mm, range: 2–16 mm) than malignant airway nodules (median: 18.0 mm, mean: 17.2 mm, range: 8–29 mm).

Table 3:

Risk of Lung Cancer, as Stratified by Airway Nodule Characteristics

Risk of Lung Cancer by Lung-RADS Category

The risk of lung cancer in airway nodules, as stratified by Lung-RADS category, is presented in Table 4. Based on the Lung-RADS category in the clinical report, the frequency of lung cancer in airway nodules was 0% (0 of two) in category 0 nodules, 10% (one of 10) in category 1 nodules, 0% (0 of 72) in category 2 nodules, 0% (0 of 12) in category 3 nodules, 9.6% (seven of 73) in category 4A nodules, 50.0% (one of two) in category 4B nodules, and 66.7% (two of three) in category 4X nodules. Upon reclassifying the report Lung-RADS to “negative” (category 2) versus “positive” (category 4A) according to Lung-RADS version 1.1, 10 of the 92 (10.9%) patients with positive Lung-RADS by report had cancer versus one of the 82 (1.2%) patients with negative Lung-RADS (P = .01). Upon reclassifying the report Lung-RADS to negative (category 2) versus positive (category 4A) according to Lung-RADS version 2022, 10 of the 40 (25.0%) patients with positive Lung-RADS by report had cancer versus one of the 134 (0.7%) patients with negative Lung-RADS (P < .0001).

Table 4:

Risk of Lung Cancer, as Stratified by Lung-RADS Category, and Diagnostic Accuracy of Lung-RADS

Comparison of Lung-RADS for Lung Cancer Detection

The diagnostic accuracy of Lung-RADS version 1.1 versus version 2022 is shown in Table 4. Sensitivity and specificity, respectively, of determining benignity versus malignancy among patients with at least one airway nodule detected were as follows: 91% (10 of 11) and 50% (81 of 163) for Lung-RADS version 1.1 and 91% (10 of 11) and 82% (133 of 163) for Lung-RADS version 2022 (P > .99 for sensitivity and P < .001 for specificity). Lung-RADS version 2022 led to 52 false-positive studies being reclassified as true-negative studies.

Discussion

In this study, we evaluated the diagnostic accuracy of the Lung-RADS classification system in predicting lung cancer in airway nodules detected at lung cancer screening CT examinations. Comparing Lung-RADS version 1.1 to the updated version 2022, we found that both versions exhibited similar sensitivity (91%) but that version 2022 demonstrated improved specificity (82% vs 50%) through the reduction of false-positive studies.

While the updated Lung-RADS version showed higher specificity, there is ambiguity in Lung-RADS version 2022 regarding how to classify a subsegmental airway nodule (8). In particular, while the guidance specifically categorizes nonobstructive subsegmental airway nodules as category 2, it does not specify a category for obstructive subsegmental airway nodules. We found that a substantial fraction (more than a quarter) of the malignancies in our study occurred in subsegmental bronchi. Thus, subsegmental location of an airway nodule does not guarantee benignity. Notably, of the subsegmental airway nodules that were obstructive in this study, 25% were malignant. Therefore, we would recommend that an obstructive airway nodule at any location be upgraded to Lung-RADS category 4A (or even category 4X).

We also examined the characteristics of airway nodules detected at lung cancer screening CT examinations and assessed their association with risk of lung cancer. Our findings demonstrate that the majority of airway nodules were benign (94% in this study), while a small proportion (6% in this study) were diagnosed as lung cancer. This is consistent with the results from the only other study of airway nodules detected on lung cancer screening CT examinations (96.3% were secretions, 3.7% were true lesions, but none was found to be malignant) (2).

In analyzing the risk of lung cancer based on airway nodule characteristics, we observed that nodules located in the trachea and mainstem bronchi were consistently benign. Multiple airway nodules were also more likely to be benign, with only one patient with multiple airway nodules being diagnosed with malignancy. In that case, mucus plugging occurred concurrently with a neoplasm. Benign features such as nonobstructive morphologies, dependent portion of the airway, internal air, or fluid attenuation further indicated a lower risk of malignancy. Some of these imaging features were also used in two previous studies to differentiate between mucus secretions and airway tumors on CT studies (2,7). Obstructive airway nodules were also high-risk: All 11 of the malignant airway nodules demonstrated obstructive morphology. Of the 11 malignant airway nodules, two were not determined to be the most worrisome nodule by the radiologist. These two cases highlight that airway nodules are typically perceived but may be at risk for cognitive or interpretation errors.

The present study had several limitations. First, it is a retrospective analysis with a small sample size. In particular, there was a small number of cancers. Second, because pathologic diagnosis was not obtained for most of the decreased and stable airway nodules at follow-up, some of these nodules may represent undiagnosed cancers. However, given that these patients were in a lung cancer screening program, they would be expected to have received regular follow-up, and any clinically significant lung cancer would likely have been diagnosed during the course of the study. Finally, our analysis was performed within a single health care network that includes two academic sites and one community site. Given the variability and radiologist discretion in aspects of Lung-RADS application, findings may differ at other institutions.

In conclusion, our study highlights the importance of assessing airway nodule characteristics for risk stratification in lung cancer screening CT examinations. The updated Lung-RADS version 2022 provides an improved classification system, reducing the number of false-positive findings while still identifying malignant airway nodules based on various imaging features. Thus, Lung-RADS version 2022 decreased false-positive studies that may have led to unnecessary interventions such as bronchoscopy. However, radiologists must remain wary of obstructive endobronchial lesions at any airway level because these have a significantly increased risk of malignancy. Future studies are warranted to confirm these findings in larger cohorts.