Differential diagnosis of benign and malignant pulmonary nodules attached to the interlobar pleura

Introduction

Lung cancer remains a major public health challenge globally, characterized by consistently high incidence and mortality (1,2). As a non-invasive imaging modality, chest computed tomography (CT) has become indispensable in the early detection and diagnostic evaluation of lung cancer (3-5). Pulmonary nodules (PNs) are the most common radiological manifestation of early-stage lung cancer (6). Accurate differentiation between benign and malignant PNs is critical in clinical practice—ensuring timely intervention for malignancies while avoiding unnecessary procedures for benign lesions. However, the similarity in imaging appearance between benign and malignant PNs poses significant diagnostic challenges, frequently resulting in delayed or incorrect diagnoses (7). Consequently, a detailed analysis of CT-based discriminative features is essential to enhance diagnostic accuracy and improve patient outcomes.

Many researchers have extensively explored this topic and made significant progress (8-10). However, the distinction between benign and malignant PNs attached to the interlobar pleura (PNs-AIP) remains inadequately investigated. Given their specific anatomical location, PNs-AIP may exhibit unique imaging characteristics compared to non-pleural-associated nodules. Our clinical observations suggest that the retraction relationship between PNs-AIP and interlobar pleura (IP) may vary depending on the nodule’s pathological nature. We propose that the combination of conventional morphological characteristics with novel pleura-associated CT features—such as the nodule-pleura interface and associated pleural changes—could provide valuable diagnostic clues, which would improve the accurate classification of PNs-AIP as benign or malignant.

The primary goal of this study was to distinguish benign from malignant PNs-AIP through clinical and imaging features. Specific emphasis was placed on analyzing their spatial relationship with nearby IP structures on CT scans. By identifying robust discriminative features, we sought to enhance diagnostic accuracy and facilitate optimized clinical management strategies for PNs-AIP. We present this article in accordance with the TRIPOD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2456/rc).

Methods

Patients

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This retrospective study was granted ethical approval by The First Affiliated Hospital of Chongqing Medical University’s Institutional Review Board (approval No. 2019-062), and the need for informed consent was waived due to its retrospective nature. A consecutive series of 522 patients diagnosed with PNs-AIP from October 2017 to April 2025 were included in this study, all of whom met the predefined inclusion criteria: (I) had undergone chest CT scans; (II) exhibited PNs-AIP on chest CT images; and (III) received a confirmed diagnosis of PNs-AIP either via surgical pathology or by complete radiological resolution on follow-up CT scans within 2 years, with a follow-up interval of ≤6 months. The exclusion criteria were as follows: (I) poor imaging quality due to significant respiratory motion artifacts; and (II) any anti-tumor therapy before chest CT examination. The study included a total of 464 patients, with 198 in the benign PNs-AIP group and 266 in the malignant group. Benign cases included 130 inflammatory nodules confirmed by surgical pathology, 58 inflammatory nodules demonstrating complete resolution on follow-up CT, and 10 benign lung tumors confirmed through surgical intervention. All malignant cases were surgically confirmed and comprised 260 lung adenocarcinomas, 5 lung squamous cell carcinomas, and 1 small cell lung cancer. Figure 1 illustrates the patient enrollment process. Additionally, clinical data were systematically collected, encompassing age, sex, smoking history, respiratory symptoms, and history of malignant tumors. Furthermore, an external validation cohort was established, consisting of 116 consecutive patients enrolled from another center between May 2018 and April 2025, all of whom met the study criteria.

Figure 1 Patient inclusion flowchart. Center 1: The First Affiliated Hospital of Chongqing Medical University, Chongqing, China. Center 2: The Jinshan Affiliated Hospital of Chongqing Medical University, Chongqing, China. CT, computed tomography; PNs-AIP, pulmonary nodules attached to the interlobar pleura.

CT protocols

Non-contrast chest CT scans were performed using a scanner manufactured by Definition FLASH (Siemens Healthcare, Erlangen, Germany). Patients were positioned supine for craniocaudal CT acquisition during a single inspiratory breath-hold. The imaged volume spanned from the thoracic inlet to a point immediately below the costophrenic angle. Essential scanning specifications included: 110–120 kVp tube potential; 50–250 mAs tube current (employing automatic modulation); 0.5 s rotation time; 0.875–1.50 pitch range; 5 mm/5 mm axial slice thickness/interval; and 512×512 reconstruction matrix. For subsequent analysis, the images were reconstructed at thinner slices: 0.6–1.0 mm for the Siemens scanners. All final images utilized a sharp reconstruction kernel with lung window settings [width: 1,600 Hounsfield unit (HU); level: −600 HU].

CT image analysis

Two chest radiologists, with 17 and 8 years of experience in thoracic imaging respectively, independently reviewed the images while blinded to all clinical and pathological data. Final interpretations of the CT findings were reached through consensus discussions. The CT features of PNs-AIP were assessed according to the following predefined criteria:

• Location (attached to the right major fissure, attached to the right minor fissure, and attached to the left major fissure);
• Nodule size (the average of the maximum long-axis diameter and its perpendicular short-axis diameter);
• Shape [regular (circular or elliptical) or irregular (any other form, such as triangular, polygonal, elongated morphologies)];
• Marginal characteristics (lobulation and spiculation);
• Intra-nodular characteristics: density [subsolid (including pure ground-glass opacity and part-solid opacity) and solid] and bubble lucency (small low-density lucent bubbles are interspersed within the nodule);
• IP-associated CT characteristics: Figure 2 illustrates the representative images for IP-associated CT characteristics. Details are as follows:
  • IP thickening (Figure 2A).
  • Nodule-IP angle: defined as the angle existing between the nodule’s edge and the adjacent IP. This sign is categorized into two types and when the same type is presented in ≥ two planes among the axial, coronal, and sagittal planes, this type is determined as the final type: type I, both angles between the nodule and IP are acute; type II, any angle between the nodule and IP is either obtuse or right (Figure 2B-2D).
  • Nodule-IP positional relationship: defined as the spatial relationship between the nodule and the adjacent IP. This sign is categorized into four types: type I, the IP shifts towards the side opposite to the nodule; type II, the IP passes through the nodule; type III, the nodule adheres to the pleura without inducing pleural displacement; type IV, the IP retracts towards the nodule. In cases with multiple coexisting types, the classification is upgraded to the highest type present (e.g., when both type II and type IV are identified, the case is designated as type IV) (Figure 2E-2H).
  • Degree of IP retraction: applicable only to PNs-AIP with a type IV nodule-IP positional relationship, this parameter is defined as the ratio of H to L. Here, L denotes the projected length of the involved pleura, whereas H represents the maximum displacement of the pleura. The final value of the degree of IP retraction is determined by calculating the average of the measurements taken from the axial, coronal, and sagittal images (Figure 2I-2L).

Figure 2 Representative images for IP-associated CT characteristics. (A) Axial CT image shows IP thickening (red arrows). Axial CT images illustrate type I nodule-IP angles (B) and type II nodule-IP angles (C,D), respectively (red dotted lines). (E-H) Axial CT images show type I, II, III, and IV nodule-IP positional relationships, respectively (red dotted lines). (I) Schematic diagram elucidates the measurement method for the degree of IP retraction, expressed as the H/L ratio. Here, “L” denotes the projected length of the involved pleura, and “H” represents the maximum distance of pleural displacement. (J-L) The measurements of H (blue solid lines) and L (red solid lines) on axial, coronal, and sagittal images in the lung window setting, respectively. CT, computed tomography; IP, interlobar pleura.

For patients who underwent serial CT scans prior to surgery or whose nodules showed complete resolution on follow-up CT, the aforementioned imaging analyses were performed using the initial CT scan data.

Statistical analysis

Statistical analyses were conducted using the software SPSS 29.0.1.0 (IBM Corp., Armonk, NY, USA), MedCalc (MedCalc Software, Ostend, Belgium), and R (version 4.4.1, R Foundation for Statistical Computing, Vienna, Austria). The distribution of continuous variables was evaluated using the Kolmogorov-Smirnov test. Data conforming to a normal distribution were reported as mean ± standard deviation and analyzed with the independent samples Student’s t-test. Non-normally distributed continuous data were described as median [interquartile range (IQR)] and analyzed using the Mann-Whitney U test; categorical variables were presented as frequencies (percentages) and compared with the Chi-squared test. To determine the optimal diagnostic thresholds for age and nodule size, receiver operating characteristic (ROC) curve analysis was performed, with cut-off selection based on the maximization of Youden’s index. We set the threshold for statistical significance at a two-tailed P value of <0.05.

Variables showing significant differences between benign and malignant PNs-AIP in univariate analyses were included in a binary logistic regression model. Independent predictors were identified via forward conditional selection. The diagnostic performance of the model was evaluated using the area under the ROC curve (AUC). A comparison of the AUCs between models with and without IP-associated CT characteristics was performed for both the training and external validation cohorts using the DeLong test, with a two-tailed P value <0.05 considered significant. The incremental discriminative ability of the Model 2 was assessed by calculating the net reclassification improvement (NRI), and integrated discrimination improvement (IDI) with the PredictABEL package, with a two-tailed P value <0.05 considered significant.

Interobserver agreement was assessed using two statistical measures tailored to the data type. For categorical features, agreement was assessed using the simple Cohen’s Kappa coefficient, as the categories were nominal without an inherent order. For continuous variables derived from CT features, agreement was quantified using the intraclass correlation coefficient (ICC). For both the Kappa and ICC statistics, values greater than 0.75 were considered to indicate excellent agreement. Statistical significance for both measures was also set at P<0.05 (two-tailed).

Results

Consistency assessments

The interobserver agreement for all CT characteristics was excellent. For categorical variables, the Kappa values between the two observers were as follows: shape (0.919), lobulation (0.927), spiculation (0.922), density (0.935), bubble lucency (0.903), IP thickening (0.878), nodule-IP angle (0.953), nodule-IP positional relationship (0.978). For continuous variables, the ICC values between the two observers were as follows: nodule size (0.968) and degree of IP retraction (0.911). All ICC and Kappa values were statistically significant (P<0.05).

Comparison of clinical characteristics between benign and malignant PNs-AIP

Table 1 summarizes the clinical data of patients with benign and malignant PNs-AIP. Among the patients with benign PNs-AIP, ages ranged from 20 to 85 years with a median age of 56 [interquartile range (IQR): 17] years, whereas among the patients with malignant PNs-AIP, ages ranged from 29 to 81 years with a median age of 61 (IQR: 15) years. The optimal cut-off value for age in distinguishing between both groups was determined to be 56 years (AUC: 0.629; sensitivity: 72.6%; and specificity: 50.0%). Age ≥56 years and female gender were found to be more prevalent in patients with malignant PNs-AIP than those with benign PNs-AIP (P<0.05). However, no significant differences were observed regarding smoking history, respiratory symptoms, and history of malignant tumors between the two groups (all P>0.05). For patients with benign PNs-AIP showing complete radiological resolution on follow-up imaging, the duration from initial detection to full nodule resolution varied between 3 and 24 months, with a median time of 12 (IQR: 10) months. The number of follow-up CT scans was 1 to 4 times, with a median of 2 (IQR: 1) times.

Comparison of conventional morphological features between benign and malignant PNs-AIP

Table 2 summarizes the conventional morphological features of benign and malignant PNs-AIP. For patients with benign PNs-AIP, nodule sizes ranged from 2.13 to 27.50 mm, with a median size of 9.50 (IQR: 6.74) mm, whereas for patients with malignant PNs-AIP, nodule sizes ranged from 5.90 to 29.72 mm, with a median size of 14.68 (IQR: 9.58) mm. The optimal cut-off value for nodule size in diagnosing malignant PNs-AIP was identified to be 11.80 mm (AUC: 0.754; sensitivity: 69.9%; and specificity: 67.2%). Nodule size ≥11.80 mm, regular shape, lobulation, spiculation, subsolid density, and bubble lucency were more frequently observed in malignant PNs-AIP (all P<0.05). However, no significant difference was found in terms of location between the two groups (P>0.05).

Comparison of IP-associated CT characteristics between benign and malignant PNs-AIP

Table 3 summarizes the IP-associated CT characteristics of benign and malignant PNs-AIP. Type I nodule-IP angle and type IV nodule-IP positional relationship were more frequent in malignant PNs-AIP, whereas type II nodule-IP angle as well as types II and III nodule-IP positional relationship were more common in benign PNs-AIP. The degree of IP retraction of benign PNs-AIP ranged from 0.08 to 0.39, with a mean value of 0.19 (0.08), whereas for malignant PNs-AIP, the degree of IP retraction ranged from 0.11 to 0.42, with a mean value of 0.22 (0.10). The degree of IP retraction in malignant PNs-AIP was larger than that in benign PNs-AIP (all P<0.05) (Figures 3,4). However, no significant differences were found in IP thickening, and type I nodule-IP positional relationship between the two groups (all P>0.05).

Figure 3 Representative images for subsolid PNs-AIP patients. (A-D) Inflammatory nodule in a 76-year-old man. (A-C) Axial, coronal, and sagittal CT images in the lung window setting show an irregular subsolid nodule with the type II nodule-IP angle (red dotted lines) and type III nodule-IP positional relationship (blue dotted lines). (D) Photomicrography (hematoxylin and eosin staining; magnification ×200) confirmed inflammatory nodule. (E-H) Invasive adenocarcinoma in a 33-year-old woman. (E-G) Axial, coronal, and sagittal CT images in the lung window setting show a regular subsolid nodule with the type I nodule-IP angle (red dotted lines) and type IV nodule-IP positional relationship (blue dotted lines). (H) Photomicrography (hematoxylin and eosin staining; magnification ×200) confirmed invasive adenocarcinoma. CT, computed tomography; IP, interlobar pleura; PNs-AIP, pulmonary nodules attached to the interlobar pleura.

Figure 4 Representative images for solid PNs-AIP patients. (A-D) Invasive adenocarcinoma in a 56-year-old man. (A-C) Axial, coronal and sagittal CT images in the lung window setting show a regular solid nodule with the type I nodule-IP angle (red dotted lines) and type IV nodule-IP positional relationship (blue dotted lines). (D) Photomicrography (hematoxylin and eosin staining; magnification ×200) confirmed invasive adenocarcinoma. (E-H) Inflammatory nodule in a 35-year-old woman. (E-G) Axial, coronal, and sagittal CT images in the lung window setting show a irregular solid nodule with the type II nodule-IP angle (red dotted lines) and type III nodule-IP positional relationship (blue dotted lines). (H) Photomicrography (hematoxylin and eosin staining; magnification ×200) confirmed inflammatory nodule. CT, computed tomography; IP, interlobar pleura; PNs-AIP, pulmonary nodules attached to the interlobar pleura.

Binary logistic regression analysis

In this study, two binary logistic regression models were created to pinpoint key determinants for diagnosing malignant PNs-AIP. Model 1 indicated that age ≥56 years, female sex, nodule size ≥11.80 mm, regular shape, lobulation, spiculation, and subsolid density were identified as independent factors for diagnosing malignant PNs-AIP (Table 4). The AUC, sensitivity, specificity, and accuracy for the logistic regression function {P=1/[1+e ^{(4.299 − 0.946 × age ≥56 years − 1.012 × female sex − 1.158 × nodule size ≥11.80 mm − 1.020 × regular shape − 1.806 × lobulation − 0.901 × spiculation − 1.547 × subsolid density)}]} were 0.863 [95% confidence interval (CI): 0.829–0.893], 71.8%, 85.9%, and 79.3%, respectively. Model 2 indicated that age ≥56 years, female sex, nodule size ≥11.80 mm, regular shape, lobulation, spiculation, subsolid density, type I nodule-IP angle, and type IV nodule-IP positional relationship were identified as independent factors for diagnosing malignant PNs-AIP (Table 5). The AUC, sensitivity, specificity, and accuracy for the logistic regression function {P=1/[1+e ^{(5.295−0.845 × age ≥56 years − 1.027 × female sex − 0.997 × nodule size ≥11.80 mm − 1.096 × regular shape − 1.455 × lobulation − 0.898 × spiculation − 1.396 × subsolid density − 0.551 × type I nodule-IP angle − 1.516 × type IV nodule-IP positional relationship)}]} were 0.886 (95% CI: 0.853–0.913), 84.2%, 76.8%, and 80.8%, respectively. Model 2 demonstrated superior diagnostic performance for malignant PNs-AIP compared to Model 1. This improvement in malignancy prediction was supported by a statistically significant increase in the AUC (DeLong test), an NRI of 0.567 (95% CI: 0.392–0.743), and an IDI of 0.052 (95% CI: 0.030–0.073), all with P values <0.05, indicating the robust and significant enhancement of the model (Figure 5A).

Figure 5 Comparison of the areas under the receiver operating characteristic curves of the two models. (A) Training cohort. (B) External validation cohort. Model 1, binary logistic regression model including clinical and morphological nodular features to predict malignant PNs-AIP. Model 2, binary logistic regression model including clinical and morphological nodular features plus IP-associated CT characteristics to predict malignant PNs-AIP. CT, computed tomography; IP, interlobar pleura; PNs-AIP, pulmonary nodules attached to the interlobar pleura; ROC, receiver operating characteristic.

In the external validation cohort, Model 1 achieved an AUC of 0.859, with a sensitivity of 64.2%, specificity of 95.90%, and accuracy of 77.59%. Model 2 demonstrated superior performance, yielding an AUC of 0.903, sensitivity of 86.6%, specificity of 83.70%, and accuracy of 85.34%. Compared to Model 1, Model 2 exhibited improved discrimination for malignant PNs-AIP. This improvement was supported by a statistically significant increase in AUC (DeLong test), along with positive NRI 0.562 (95% CI: 0.221–0.903) and IDI 0.074 (95% CI: 0.027–0.120), all with P<0.05. These findings consistently indicate a meaningful enhancement in model performance (Figure 5B). The clinical and CT features of the validation cohort can be found in Tables S1,S2.

Discussion

As a distinct subtype of PNs, PNs-AIP are being detected with increasing frequency in clinical practice. Nevertheless, their differential diagnosis—particularly in predicting malignancy—remains poorly understood. Owing to this diagnostic challenge and their close anatomical relationship with IP, further investigation into the association between these nodules and IP may offer valuable insights for differentiation. Nevertheless, prior research has not comprehensively examined this relationship. To address this gap, our study performed a detailed analysis of clinical, conventional morphological, and IP-related CT features of both benign and malignant PNs-AIP, with the goal of identifying robust predictors of malignancy.

Regarding clinical characteristics, our results indicated that advanced age and female sex were predominant among patients with malignant PNs-AIP, which aligns with previously reported findings (11,12). The higher malignancy rates in older populations can be explained by a progressive decline in immune function and the lifetime accumulation of exposure to cancer-causing agents (13,14). In addition, the majority of malignant nodules in our cohort were adenocarcinomas. This is consistent with established evidence linking adenocarcinoma to female sex—a correlation also reinforced by our results.

Clinically, morphological analysis is the most important method used by radiologists to determine the nature of PNs-AIP. Despite considerable overlap in their CT presentations, benign and malignant PNs-AIP demonstrate distinct characteristics that are crucial for accurate differentiation. The present study showed that large size, regular shape, lobulation, spiculation, subsolid density, and bubble lucency were highly suggestive of cancerous nodules, which is consistent with previous findings (15-17). Furthermore, we developed Model 1 based on these clinical and conventional morphological characteristics. This model demonstrated moderate performance, with an AUC of 0.863 and an accuracy of 79.30% in the training cohort, and an AUC of 0.859 along with an accuracy of 77.59% in the external validation cohort.

However, the differential diagnosis information involved in Model 1 remains insufficient. Therefore, we further investigated the correlation between PNs-AIP and IP, and identified a set of novel IP-related imaging features to facilitate the differentiation of nodules with distinct pathological subtypes. Although the benign and malignant PNs in this study all adhere to the IP, they showed some significant differences in the nodule-IP interface. Firstly, we found that the type II nodule-IP angle occurred more frequently in benign PNs-AIP, whereas the type I nodule-IP angle was more prevalent in malignant PNs-AIP. For malignant PNs adjacent to the IP, as nodules grow, the distance to the IP diminishes. Upon reaching a certain size, they abut to the fissure, which was identical to the report of a previous study (18). Initially, the contact area is small, resulting in both angles between the nodule margin and the IP being acute. In contrast, benign PNs in our cohort were mostly inflammatory nodules. Due to inflammation’s tendency to involve the adjacent IP and the relatively large contact area with the fissure, an obtuse or right angle typically forms between the nodule margin and the IP.

Subsequent analysis revealed distinguishable anatomic spatial relationships between benign/malignant PNs-AIP and the IP on multiplanar reconstruction images. Malignant lesions typically demonstrated IP retraction toward the nodule (type IV), a phenomenon attributed to tumor stroma rich in fibroblasts secreting collagen and other matrix components that induce stromal contraction (19). Additionally, the surrounding inflammatory microenvironment promotes interactions between proliferating pleural cells, inflammatory infiltrates, and the nodule, facilitating IP retraction. In contrast, benign lesions exhibited either IP traversal through the nodule (type II) or IP adherence without fissural displacement (type III). Among inflammatory nodules, mild inflammation may spread via alveolar pores until constrained by the IP (20), forming straight nodule margins. Conversely, marked inflammation could transgress the pleural barrier, involving adjacent lung lobes (21,22). Although type IV IP-nodule relationships predominated in malignant lesions, certain inflammatory nodules also displayed this pattern—their IP retraction correlated with intra-nodular fibrous tissue proliferation (23). Notably, malignant PNs-AIP exhibited significantly greater degrees of IP retraction compared to their benign counterparts, a finding indicative of stronger pleural traction forces generated by malignant lesions. This differential retraction degree serves as a valuable discriminative sign between benign and malignant PNs-AIP, corroborating existing literature.

Furthermore, we constructed Model 2 through the integration of clinical characteristics, conventional morphological features of nodules, and IP-associated CT features. Beyond the independent predictors identified in Model 1, the type I nodule-IP angle and type IV nodule-IP positional relationship were further identified as additional independent predictors for malignant PNs-AIP. Binary logistic regression identified the type IV nodule-IP positional relationship as an independent predictor of malignant PNs-AIP. It demonstrated a high odds ratio (OR) of 4.556, indicating that this sign is a significant predictor for malignancy. In the training cohort, this integrated model achieved an AUC of 0.886 and an accuracy of 80.80%; in the external validation cohort, these values were 0.903 and 85.34%, respectively—demonstrating favorable effectiveness and generalizability. Additionally, the DeLong test, along with the NRI and IDI, consistently demonstrated that Model 2 exhibited superior predictive performance for malignant PNs-AIP compared to Model 1 in both the training and external validation cohorts. Thus, a comprehensive assessment of clinical parameters, conventional nodule morphology, and nodule-IP relationships is crucial for the differential diagnosis of PNs-AIP.

Our study has several limitations that merit consideration. First, the retrospective design inherently carries the risk of selection bias. Second, although the inclusion of benign nodules that resolved during follow-up (29.3%) may enhance the model’s ability to differentiate malignancy from active inflammatory lesions, it also leads to an underrepresentation of stable benign nodules. As a result, the model’s performance in distinguishing malignant from stable benign nodules requires further validation in cohorts enriched with such cases. Future studies will incorporate a broader spectrum of stable benign nodules and conduct detailed comparative analyses across benign subgroups to improve the model’s clinical applicability. Third, our malignant cohort was predominantly composed of adenocarcinoma (97.7%), which may influence the observed imaging phenotypes and potentially limit the model’s generalizability to nodules of other histological subtypes. Moreover, relevant molecular profiling data (e.g., EGFR/ALK status) were not available for this early-stage surgical cohort, restricting the investigation of underlying biological correlations. Fourth, the visual assessment of morphology-based CT features—such as the nodule-pleural interface angle and the degree of pleural retraction—is inherently subjective and labor-intensive. Developing an artificial intelligence or deep learning model for the automated detection, quantification, and integration of these specific IP-associated features represents a critical next step in advancing this research.

Conclusions

This research identifies clear clinical and radiological distinctions between benign and malignant PNs-AIP. Crucially, assessing the spatial relationship with the IP provides significant diagnostic value beyond conventional features, thereby allowing clinicians to better differentiate cases and optimize treatment approaches.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2456/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2456/dss

Funding: This work was supported by Natural Science Foundation of Chongqing, China (No. CSTB2025NSCQ-GPX1205) and Chongqing Medical Scientific Research Project (Joint Project of Chongqing Health Commission and Science and Technology Bureau) (No. 2026MSXM021).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2456/coif). All authors report that this work was supported by Natural Science Foundation of Chongqing, China (No. CSTB2025NSCQ-GPX1205) and Chongqing Medical Scientific Research Project (Joint Project of Chongqing Health Commission and Science and Technology Bureau) (No. 2026MSXM021). The authors have no other 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. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This retrospective study was granted ethical approval by The First Affiliated Hospital of Chongqing Medical University’s institutional review board (approval No. 2019-062), and the need for informed consent was waived due to its retrospective nature.

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. Leiter A, Veluswamy RR, Wisnivesky JP. The global burden of lung cancer: current status and future trends. Nat Rev Clin Oncol 2023;20:624-39. [Crossref] [PubMed]
2. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018;68:394-424. [Crossref] [PubMed]
3. Reck M, Dettmer S, Kauczor HU, Kaaks R, Reinmuth N, Vogel-Claussen J. Lung Cancer Screening With Low-Dose Computed Tomography. Dtsch Arztebl Int 2023;120:387-92. [Crossref] [PubMed]
4. Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, Fagerstrom RM, Gareen IF, Gatsonis C, Marcus PM, Sicks JD. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365:395-409. [Crossref] [PubMed]
5. Bonney A, Malouf R, Marchal C, Manners D, Fong KM, Marshall HM, Irving LB, Manser R. Impact of low-dose computed tomography (LDCT) screening on lung cancer-related mortality. Cochrane Database Syst Rev 2022;8:CD013829. [Crossref] [PubMed]
6. Sahu SP, Londhe ND, Verma S, Singh BK, Banchhor SK. Improved pulmonary lung nodules risk stratification in computed tomography images by fusing shape and texture features in a machine-learning paradigm. Int J Imaging Syst Technol 2021;31:1503-18.
7. Yuan X, Zhang J, Quan C, Cao J, Ao G, Tian Y, Li H. Differentiation of malignant and benign pulmonary nodules with first-pass dual-input perfusion CT. Eur Radiol 2013;23:2469-74. [Crossref] [PubMed]
8. Zhan Y, Song F, Zhang W, Gong T, Zhao S, Lv F. Prediction of benign and malignant pulmonary nodules using preoperative CT features: using PNI-GARS as a predictor. Front Immunol 2024;15:1446511. [Crossref] [PubMed]
9. Ma XB, Xu QL, Li N, Wang LN, Li HC, Jiang SJ. A decision tree model to distinguish between benign and malignant pulmonary nodules on CT scans. Eur Rev Med Pharmacol Sci 2023;27:5692-9. [Crossref] [PubMed]
10. Zhang X, Yang P, Tian J, Wen F, Chen X, Muhammad T. Classification of benign and malignant pulmonary nodule based on local-global hybrid network. J Xray Sci Technol 2024;32:689-706. [Crossref] [PubMed]
11. Wang C, Shao J, He Y, Wu J, Liu X, Yang L, Wei Y, Zhou XS, Zhan Y, Shi F, Shen D, Li W. Data-driven risk stratification and precision management of pulmonary nodules detected on chest computed tomography. Nat Med 2024;30:3184-95. [Crossref] [PubMed]
12. Chen D, Yang L, Zhang W, Shen J, Van Schil PEY, Divisi D, Seetharamu N, Gu J. Prevalence and management of pulmonary nodules: a systematic review and meta-analysis. J Thorac Dis 2024;16:4619-32. [Crossref] [PubMed]
13. Barta JA, Powell CA, Wisnivesky JP. Global Epidemiology of Lung Cancer. Ann Glob Health 2019;85:8. [Crossref] [PubMed]
14. Cui X, Han D, Heuvelmans MA, Du Y, Zhao Y, Zhang L, Groen HJM, de Bock GH, Dorrius MD, Oudkerk M, Vliegenthart R, Ye Z. Clinical characteristics and work-up of small to intermediate-sized pulmonary nodules in a Chinese dedicated cancer hospital. Cancer Biol Med 2020;17:199-207. [Crossref] [PubMed]
15. Zhao WH, Zhang LJ, Li X, Luo TY, Lv FJ, Li Q. Clinical and Computed Tomography Characteristics of Inflammatory Solid Pulmonary Nodules with Morphology Suggesting Malignancy. Acad Radiol 2025;32:1067-77. [Crossref] [PubMed]
16. Yao Y, Wang X, Guan J, Xie C, Zhang H, Yang J, Luo Y, Chen L, Zhao M, Huo B, Yu T, Lu W, Liu Q, Du H, Liu Y, Huang P, Luan T, Liu W, Hu Y. Metabolomic differentiation of benign vs malignant pulmonary nodules with high specificity via high-resolution mass spectrometry analysis of patient sera. Nat Commun 2023;14:2339. [Crossref] [PubMed]
17. Liao HF, Huang XT, Li X, Lv FJ, Luo TY, Li Q. Solitary lung adenocarcinoma: follow-up CT, pathological-molecular characteristics, and surgical prognosis for different morphological classifications. Insights Imaging 2023;14:209. [Crossref] [PubMed]
18. Lin RY, Lv FJ, Fu BJ, Li WJ, Liang ZR, Chu ZG. Features for Predicting Absorbable Pulmonary Solid Nodules as Depicted on Thin-Section Computed Tomography. J Inflamm Res 2021;14:2933-9. [Crossref] [PubMed]
19. Iwai M, Tulafu M, Togo S, Kawaji H, Kadoya K, Namba Y, et al. Cancer-associated fibroblast migration in non-small cell lung cancers is modulated by increased integrin α11 expression. Mol Oncol 2021;15:1507-27. [Crossref] [PubMed]
20. Aegerter H, Lambrecht BN, Jakubzick CV. Biology of lung macrophages in health and disease. Immunity 2022;55:1564-80. [Crossref] [PubMed]
21. Xiao YD, Lv FJ, Li WJ, Fu BJ, Lin RY, Chu ZG. Solitary Pulmonary Inflammatory Nodule: CT Features and Pathological Findings. J Inflamm Res 2021;14:2741-51. [Crossref] [PubMed]
22. Scherer PM, Chen DL. Imaging Pulmonary Inflammation. J Nucl Med 2016;57:1764-70. [Crossref] [PubMed]
23. Tsukui T, Wolters PJ, Sheppard D. Alveolar fibroblast lineage orchestrates lung inflammation and fibrosis. Nature 2024;631:627-34. [Crossref] [PubMed]