CT-based emphysema score is associated with prolonged air leak after lung resection: a retrospective cohort study

Introduction

Prolonged air leaks (PALs) affect 6% to 19% of patients undergoing pulmonary resection for non-small cell lung cancer (NSCLC) (1-7). While most air leaks are small and resolve spontaneously, PALs can lead to significant patient morbidity and have been associated with worse pain, increased risk of pneumonia, empyema, and thromboembolism, and greater healthcare costs due to longer hospital stays and readmissions (8,9).

Several factors have been reported as risk factors for PAL, including smoking history, preoperative steroid use, lower forced expiratory volume in 1 second (FEV₁), non-fissure less technique, pathological tumor-node-metastasis (TNM) stage III/IV, male sex, low body mass index (BMI), upper lobe resections, advanced age, presence of pleural adhesions, lobectomy, and bilobectomy (5-8). Building on these risk factors, several predictive scores have been developed to estimate PAL risk (2,10-13). However, none of the existing models include computed tomography (CT)-based quantitative assessment of emphysema, which could offer valuable additional insights.

CT-based quantification of emphysema has been reported to be a predictor of PAL (14-18). One of the first studies on this topic, conducted in Japan in 2005 (n=62), concluded that CT-quantified emphysema scores were associated with duration of air leak after lobar lung resection, while FEV₁ was not (14). In 2016, investigators from Korea established the predictive value of combining CT-based emphysema scores and pulmonary function tests to assess the risk of post-operative pulmonary complications, including PAL (16). More recently, a 2018 study done on Japanese patients (n=248) identified an emphysema cut-point of 35% as a predictor of PAL, with a sensitivity of 87%, in patients undergoing thoracoscopic lobectomy (17).

Despite these reports, all prior studies have been limited by relatively small sample sizes, non-U.S. populations, and have predominantly focused on lobar resections (Box 1). Based on the hypothesis that emphysematous lung tissue is susceptible to structural compromise after lung cancer resection, the aim of this study was to investigate if greater pre-operative CT-quantified emphysema score is associated with increased risk of PAL after NSCLC resection. Furthermore, the study focused on two main objectives: (I) establishment of a standardized formula for CT-based emphysema quantification (emphysema score); and (II) validation of the association of higher emphysema score with increased risk of developing PAL in a large United States (US) cohort. Secondary objectives were: (I) to identify a clinically useful emphysema cut-point to preoperatively risk-stratify patients for PAL; and (II) to evaluate the applicability of emphysema score as a predictor of PAL in patients with reduced pulmonary function tests (PFTs). We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-946/rc).

Methods

Study population

Patients were identified from a prospectively maintained, single-institution database (RUSH University Medical Center, Chicago, IL, USA) of individuals who underwent curative-intent resection for NSCLC between 2010 and 2021 (n=596), with available preoperative chest CT scans for segmentation analysis. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Rush University Medical Center Institutional Review Board (IRB) (No. 24091203). Due to the retrospective nature of the study and minimal risk to participants, the Rush University Medical Center IRB waived informed consent for this study.

Eligible patients were 18 years or older and had pathologically confirmed stage I–IIIA NSCLC treated surgically during the study period. Patients were excluded if they had missing or poor-quality preoperative CT scans, duplicate entries, or had received neoadjuvant chemotherapy or radiation therapy, as these treatments can alter lung tissue integrity and potentially affect the risk of PAL. To maintain cohort homogeneity, patients who underwent pneumonectomy or bilobectomy were excluded, as were those who underwent a combination of wedge resection and lobectomy, in order to allow for adequate control of both procedure type and lobe operated upon in the adjusted analysis (Figure 1).

Figure 1 Flowchart demonstrating exclusion criteria. DAFS, Data Analysis Facilitation Suite.

Body segmentation analysis

Preoperative chest CT scans, obtained within 3 months prior to surgery, were processed using Data Analysis Facilitation Suite (DAFS) (19). DAFS is a proprietary image analysis platform designed for automated tissue segmentation and quantitative radiologic assessment. The software computes the number of voxels representing segmented body tissues and uses these data to derive quantitative metrics including tissue density [in Hounsfield units (HU)], surface area, and volume.

Preoperative CT scans were retrieved from the institutional Picture Archiving and Communication (PACS) system in Digital Imaging and Communications in Medicine Standard (DICOM) format. When multiple preoperative scans were available for a patient, we selected the scan closest to the date of surgery, provided it was of adequate quality. Scans that were grossly corrupted, had poor resolution, or incomplete visualization of the lung were manually excluded during an initial review. Once a scan was selected, it was processed using DAFS. In rare cases where the software returned a corrupted or incomplete output—indicating suboptimal scan quality—an alternative, higher-quality scan was used if available. This two-step review process helped ensure consistent quality across the dataset and minimized variability in emphysema quantification. When multiple preoperative scans were available for a patient, we selected the scan closest to the date of surgery, provided it was of adequate quality. Scans that were grossly corrupted, had poor resolution, or incomplete visualization of the lung were manually excluded during an initial review. Once a scan was selected, it was processed using DAFS. In rare cases where the software returned a corrupted or incomplete output—indicating suboptimal scan quality—an alternative, higher-quality scan was used if available. This two-step review process helped ensure consistent quality across the dataset and minimized variability in emphysema quantification.

Total lung volume and density were determined for the ipsilateral lung. Emphysematous lung was defined as lung tissue with an average density lung between −950 to −1,200 HU, in accordance with multiple prior studies (20-22). To determine the percentage of emphysematous lung tissue for each patient, the volume of lung with a density between −950 and −1,200 HU was divided by the total volume of the ipsilateral lung (Figure 2).

Figure 2 Schematic illustration of CT-based emphysema score calculation. The emphysema score is defined as the percentage of the ipsilateral lung with density between −950 and −1,200 HU. Left: depiction of total lung segmentation. Right: regions with density between −950 and −1,200 HU are shown in red to represent emphysematous areas. This figure is a visual conceptualization for illustrative purposes and does not display actual CT data. CT, computed tomography; HU, Hounsfield units.

Variables and outcomes

Demographic data collected included patient age, race, gender, and BMI. Preoperative variables collected included preoperative pulmonary function tests [predicted FEV₁, predicted diffusing capacity of the lung for carbon monoxide (DLCO)], smoking status (current, never, or ever smoker), history of cardiovascular disease, congestive heart failure (CHF), diabetes mellitus (DM), chronic obstructive pulmonary disease (COPD), prior cardiothoracic surgery, Eastern Cooperative Oncology Group (ECOG) score and clinical (T, N, and M) stage of the tumor (8^th edition of American Joint Committee on Cancer AJCC cancer staging system for NSCLC) (23). Operative data collected included surgical approach (open or minimally invasive), procedure laterality, and type of surgical resection (lobectomy, segmentectomy, or wedge resection). Postoperative variables collected included air leak lasting greater than 5 days, whether or not the patient was discharged with a chest tube, readmission within 30 days of discharge, and survival 30 days post-surgery. Our primary outcome of interest was the development of a PAL defined as an air leak lasting more than 5 days, based on prior studies (1,5,24,25). Additional outcomes assessed included major morbidity which was defined based on the Society of Thoracic Surgeons (STS) General Thoracic Surgery Database definition consisting of nine outcomes: tracheostomy, reintubation, initial ventilatory support greater than 48 hours, adult respiratory distress syndrome (ARDS), bronchopleural fistula, pulmonary embolus, pneumonia, unexpected return to the operating room, or myocardial infarction (MI) (26).

Statistical analysis

Continuous variables were analyzed using Student’s t-test for parametric data with equal variance, Welch’s t-test for parametric data with unequal variance, and the Wilcoxon rank-sum test for non-parametric data. Variables were considered non-normally distributed if the skewness was greater than 1 or less than −1. Variance equality was assessed using Levene’s test. Categorical variables were assessed using Fisher’s exact test for expected cell frequencies <5 and the Chi-squared test otherwise. Logistic regression models with a 95% confidence limit evaluated the association between emphysema and PAL, adjusting for age, gender, BMI, surgical approach, procedure type, pack-years, prior cardiothoracic surgery, tumor size, and lobe operated upon. Covariates included were selected based on existing literature (5-8). Predicted FEV₁ and DLCO were excluded from the multivariable model to prevent multicollinearity but were analyzed separately for their associations with PAL. Missing values were identified and labeled in Stata. In analyses where variables with missing data were included, Stata automatically excluded incomplete observations (listwise deletion) to ensure valid comparisons.

Univariable and multivariable logistic regression models were used to assess the association between emphysema and PAL. Due to the absence of a predefined threshold for emphysema in relation to PAL, an exploratory grid-search approach was employed whereby multiple cutpoints were created and systematically tested. Cut-points were based on different percentiles of emphysema, creating a binary variable at each threshold: low (< cut-point) and high (≥ cut-point) emphysema groups. Univariable and multivariable logistic regression models assessed the association of emphysema with PAL at each cut-point. To identify a practical threshold while maintaining sufficient statistical power for both groups, the goal of the grid-search was to identify the emphysema percentile at which the association between emphysema and PAL first showed significance (P<0.05), in both the univariable and multivariable logistic regression models.

A similar strategy has been described in statistical literature, where the threshold of an independent variable is identified as the point where its effect on the outcome first deviates from baseline, as determined by statistical evidence (27). This strategy aligns with how exploratory or hypothesis-generating studies are modeled, providing an objective method for threshold estimation in the absence of predefined cut-offs (28,29). This approach was selected over receiver operating characteristic (ROC)-based methods such as the Youden index, which in our dataset produced a cut-point at the 94^th percentile of emphysema scores—capturing only a small, extreme-risk subset and limiting clinical utility.

Internal validation of the 16% emphysema cut-point was performed using nonparametric bootstrap resampling (2,000 iterations). In each replicate, the full multivariable logistic model was re-estimated, and the odds ratio (OR) for emphysema ≥16% was recorded.

Furthermore, to evaluate the discriminative performance of our models, we computed the area under the ROC curve (AUC) for three logistic regression models: (I) a base model excluding the CT-based emphysema score; (II) a model incorporating the emphysema score as a binary variable (≥16%); and (III) a model incorporating the continuous emphysema score. Predicted probabilities were generated using the predict command, and AUCs with 95% confidence intervals (CIs) were calculated using the roctab command. Statistical comparisons between AUCs were performed using DeLong’s test via the roccomp command in Stata.

Comparison with established risk scores

To evaluate the relative performance of our CT-based emphysema score, we compared its predictive discrimination for PAL with that of the Pompili risk score, a previously validated multivariable clinical model (13). The Pompili score assigns 1 point each for male sex and FEV₁ <80%, and 2 points for BMI <18.5 kg/m², yielding a total score range of 0 to 4. Predicted probabilities from each model were generated using logistic regression, and ROC analysis was performed using roctab and roccomp in Stata. We also tested a combined model incorporating both the emphysema score and the Pompili score to assess potential additive value.

Additionally, the total cohort was divided into two groups: those with either FEV₁ or DLCO ≥60% and those with both FEV₁ and DLCO <60%. The association between CT-based emphysema score and PAL was then assessed in each group. The 60% threshold was chosen based on the American College of Chest Physicians (ACCP) guidelines, which recommend caution before proceeding with lobar resection in NSCLC patients who have either an FEV₁ or DLCO below this cutoff (30). All analyses were performed using Stata/IC 15.0 [StataCorp LLC, College Station, TX, USA, Stata (RRID:SCR_012763)].

Results

A total of 550 patients met inclusion criteria, of whom 58% (321/550) identified as female and 77% (425/550) white. The median age was 71 (IQR, 65–76) years, and the most common procedure was lobectomy (67%, 366/550). A total of 79% (432/550) of the resections were performed in a minimally-invasive fashion, and 9% (48/550) of the patients developed a PAL. Among patients who developed a PAL, the median duration of chest tube drainage was 9 (IQR, 6–15) days. Median length of stay (LOS) was 2 (IQR, 1–3) days, and 7% (36/550) of patients were discharged with a chest tube. Out of the 36 patients discharged with a chest tube, 33 patients (92%) had a PAL. The overall rate of major morbidity was 2% (11/550) while the 30-day mortality rate was 0.36% (2/550).

Cut-point for preoperative risk stratification

The optimal cut-point identified via grid-search method was 16%. There were 138 patients in the high-emphysema group [median emphysema score =22.20% (IQR, 19.1–26.13%)] and 412 in the low-emphysema group [median emphysema score =3.29% (IQR, 0.22–9.52%)]. Baseline characteristics of the patient cohort are outlined in Table 1. Representative coronal CT images from four patients across the emphysema score spectrum (3%, 16%, 26%, and 49%) are shown in Figure 3 to illustrate the visual variability and limitations of subjective assessment.

Figure 3 Coronal chest CT images from four patients with quantitatively assessed emphysema scores of (A) 3%, (B) 16%, (C) 26%, and (D) 49%. Despite a wide range in emphysema burden based on automated quantification, visual differences are subtle—particularly between low and intermediate scores—highlighting the limitations of subjective interpretation and the utility of quantitative CT analysis. CT, computed tomography.

This cut-point stratifies patients into high-risk (emphysema ≥16%) and low-risk (emphysema <16%) categories and predicts the risk of developing PAL with a sensitivity of 38% and a specificity of 76%, classifying 73% of patients correctly. When the binary variable for emphysema is incorporated into a multivariable ROC curve (Figure 4) which also included age, gender, BMI, surgical approach, procedure type, number of pack-years of smoking, history of cardiothoracic surgery, tumor size and lobe operated upon, it performs well as a predictor of PAL (AUC =0.70; 95% CI: 0.61–0.79; P<0.001). To test the internal stability of the 16% emphysema threshold, we performed bootstrap resampling with 2000 iterations using the full multivariable model. The bootstrapped OR for emphysema ≥16% was 2.13 (95% CI: 0.31–3.96; P=0.02), consistent with the original estimate. These findings support the robustness of the 16% threshold under data variation.

Figure 4 The ROC curve demonstrates the predictive performance of the multivariable model, including dichotomized emphysema percentage, for PAL. The AUC is 0.70 (95% CI: 0.61–0.79; P<0.001). AUC, area under the ROC curve; CI, confidence interval; PAL, prolonged air leak; ROC, receiver operating characteristic.

On comparing the predictive performance of the multivariable models, the base model (excluding emphysema score) yielded an AUC of 0.673 (95% CI: 0.586–0.760). Inclusion of the emphysema score as a binary variable (≥16%) increased the AUC to 0.699 (95% CI: 0.612–0.787), while inclusion as a continuous variable yielded an AUC of 0.692 (95% CI: 0.599–0.785). However, neither comparison reached statistical significance when compared to the base model (P=0.22 and P=0.41, respectively; DeLong’s test). These results suggest modest improvement in model discrimination with the addition of emphysema score (Figure 5).

Figure 5 ROC curves comparing multivariable logistic regression models for predicting PAL. (A) Base model excluding the CT-based emphysema score (AUC =0.673). (B) Model including the emphysema score as a continuous variable (AUC =0.692). (C) Model including the emphysema score as a binary variable using a 16% cut-point (AUC =0.699). Overall, both emphysema-based models showed modest improvement in predictive discrimination compared with the base model, but differences in AUC were not statistically significant (P=0.41 for continuous vs. base; P=0.22 for binary vs. base; DeLong’s test). AUC, area under the ROC curve; CT, computed tomography; PAL, prolonged air leak; ROC, receiver operating characteristic.

To assess how the CT-based emphysema score compares with established clinical risk models, we evaluated its performance against the Pompili multivariable score, as well as a combined model including both scores. The emphysema score alone yielded an AUC of 0.568 (95% CI: 0.479–0.658), while the Pompili score achieved a slightly higher AUC of 0.609 (95% CI: 0.518–0.699). A combined model incorporating both scores yielded an AUC of 0.626 (95% CI: 0.530–0.723). The difference among the three models did not reach statistical significance (P=0.10, DeLong test), suggesting that although modest, the emphysema score’s predictive ability is comparable to the more complex Pompili model.

Univariable logistic regression analysis using CT-based emphysema score as a continuous variable yielded an association with PAL (OR =1.03; 95% CI: 1.01–1.06, P=0.04). On multivariable regression, after controlling for age, gender, BMI, surgical approach, procedure type, number of pack-years of smoking, history of cardiothoracic surgery, tumor size and lobe operated, CT-based emphysema score (OR =1.04; 95% CI: 1.01–1.07; P=0.02), lobectomy (OR =2.59; 95% CI: 1.04–6.41; P=0.040) and history of cardiothoracic surgery (OR =3.50; 95% CI: 1.53–8.03; P=0.03) were associated with PAL (Table 2). CT-based emphysema score was also associated with greater risk of being discharged with a chest tube (OR =1.03; 95% CI: 1.002–1.07; P=0.040). However, there was no association of greater emphysema score with 30-day major morbidity (OR =1.02; 95% CI: 0.97–1.08; P=0.45) or 30-day mortality (OR =0.65; 95% CI: 0.25–1.70; P=0.38).

Upon analyzing CT-based emphysema score as a binary variable, univariable logistic regression demonstrated higher odds of developing PAL if the score was ≥16% (OR =1.91; 95% CI: 1.03–3.54; P=0.040). The multivariable model also demonstrated an association between emphysema score (OR =2.13; 95% CI: 1.05–4.34; P=0.040), lobectomy (OR =2.48; 95% CI: 1.006–6.13; P=0.048), and history of cardiothoracic surgery (OR =3.38; 95% CI: 1.48–7.69; P=0.004) with PAL (Table 3). In addition, the effect of greater emphysema score on the risk of being discharged with a chest tube remained significant (OR =2.27; 95% CI: 1.14–4.54; P=0.02).

Subgroup analysis by PFT

Predicted FEV₁ was associated with PAL (OR =0.98; 95% CI: 0.96–0.99; P=0.02) while predicted DLCO was not (OR =0.98; 95% CI: 0.97–1.00; P=0.07). Of the total cohort, 518 patients (94%) had either FEV₁ or DLCO ≥60%, while 32 patients (6%) had both FEV₁ and DLCO <60%. Emphysema score, analyzed as a continuous variable, was associated with PAL in the group with both FEV₁ and DLCO <60% (OR =1.08; 95% CI: 1.00–1.16; P=0.040), but not in the group where either FEV₁ or DLCO was ≥60% (OR =1.01; 95% CI: 0.98–1.05; P=0.45).

Among the 518 patients with normal pulmonary function (either FEV₁ or DLCO ≥60%), the median emphysema score was 7.1% (IQR, 0.64–15.2%). In the 32 patients with borderline pulmonary function (both FEV₁ and DLCO <60%), the median emphysema score was slightly lower at 6.6% (IQR, 1.4–25.7%). However, the proportion of patients with emphysema score ≥16% was notably higher in the borderline group (43.8%, 14/32) compared to the normal group (23.9%, 124/518).

Discussion

PAL is one of the most common postoperative complications following pulmonary resection for lung cancer and is associated with significant healthcare costs and patient morbidity (1-9,31). Prior studies have explored the association of CT-based quantitative emphysema with PAL, but these are smaller studies conducted abroad, investigating patients undergoing primarily lobectomy (14-17). Herein, we report the results from over 500 patients who had undergone lobar or sublobar resections for NSCLC.

Our investigation identified three risk factors associated with PAL in the multivariable analysis: emphysema score, lobectomy, and history of cardiothoracic surgery. It was found that patients with higher emphysema scores in the ipsilateral lung were over twice more likely to develop a PAL than those with lower values, after adjusting for potential confounders. This validates the findings of earlier studies and establishes CT-quantified emphysema score as a risk factor for PAL in our population (14-17). Lobectomy was also independently associated with an increased risk of PAL, consistent with prior studies (5-8). Compared to wedge resection, lobectomy involves greater parenchymal disruption, larger raw surface area, and a higher degree of anatomical dissection, all of which contribute to increased susceptibility to postoperative air leak. Additionally, a history of prior cardiothoracic surgery emerged as a significant predictor of PAL. Although detailed operative history and laterality were not consistently available, this association likely reflects the technical challenges posed by pleural or hilar adhesions from previous intrathoracic operations. These adhesions may increase the risk of parenchymal tearing or compromise seal integrity during resection, particularly when adhesiolysis is required.

Importantly, while prior studies typically quantified emphysema using values from both lungs, our approach focused specifically on the ipsilateral lung, recognizing that emphysema distribution can vary substantially between lungs, and that the operative side is more relevant for surgical outcomes (14-16). In patients identified as high risk for PAL based on preoperative emphysema score, several intraoperative adjustments may be considered. These include the use of staple-line buttressing or sealants, pleural tenting for upper lobe resections, minimizing the number of stapler firings across fragile parenchyma, and performing meticulous dissection to avoid unnecessary trauma. Surgeons may also conduct more rigorous intraoperative air-leak testing with immediate reinforcement of identified leaks. Postoperatively, strategies may include chest tube management with low suction or water seal, use of digital drainage systems to guide removal timing, and early ambulation with close monitoring for complications.

Although the identified emphysema score cut-point demonstrated moderate sensitivity (38%), its high specificity (76%) is of particular clinical value. A test with high specificity is ideal for ruling out patients who are unlikely to develop PAL. In surgical planning, especially when considering the use of resource-intensive intraoperative adjuncts, such as surgical sealants, staple-line reinforcement, or pleural tenting, the ability to confidently identify low-risk patients is crucial (1,32). These techniques, while potentially beneficial, lack definitive consensus regarding their effectiveness and are often applied inconsistently (1). When used in low-risk patients, they may result in unnecessary resource expenditure and extended operative time. Therefore, a single, objective, preoperative imaging-derived parameter such as the emphysema score, has potential to serve as a decision-support tool to guide the selective use of these adjuncts.

To contextualize the performance of the CT-based emphysema score, we compared it with the previously validated Pompili risk score, which includes four clinical parameters (13). Despite being a single imaging-derived variable, the emphysema score demonstrated comparable discriminative performance. When used in combination with the Pompili score, the model’s AUC increased slightly, though not significantly. These findings suggest that the emphysema score may capture aspects of pulmonary vulnerability not fully represented by clinical variables alone. Given its simplicity, objectivity, and preoperative availability, the emphysema score may serve as a practical adjunct or alternative to multivariable risk scores.

Moreover, our findings extend the relevance of emphysema score into a critical high-risk subgroup: patients with borderline pulmonary function, defined as both FEV₁ and DLCO <60%. An association between emphysema score and PAL was observed in the borderline pulmonary function group (FEV₁ and DLCO <60%) but not in patients with preserved pulmonary function (FEV₁ or DLCO ≥60%). This discrepancy may be explained by differences in emphysema burden between the two groups. Only 23.9% of patients in the normal PFT group had an emphysema score ≥16%, compared to 43.8% in the borderline group. Moreover, in the normal group, emphysema scores were generally low and showed less variability (IQR, 0.64–15.2%), limiting the score’s ability to separate patients at risk of PAL.

In contrast, the borderline group had a broader range of scores (IQR, 1.4–25.7%) and a higher proportion of patients with emphysema ≥16%, enhancing the score’s discriminative utility in that subgroup. Prior literature has established that patients falling below this threshold are at significantly increased risk for pulmonary complications and operative mortality following lung resection (30,33,34). However, clinical decision-making in this cohort is often fraught with uncertainty, as surgeons must balance the potential benefits of curative resection against the heightened risk of complications such as PAL. The results of this study suggest that within this borderline PFT group, the emphysema score may offer an additional layer of risk stratification. A patient with suboptimal lung function but a low ipsilateral emphysema score may not be at significantly elevated risk for PAL, which could provide reassurance and greater confidence to proceed with surgery. Conversely, a high score in this group may signal the need for enhanced intraoperative precautions or postoperative monitoring.

There are several limitations to this study. First, it is a single-institution, retrospective analysis, which may introduce bias due to institutional practices, demographic homogeneity, and surgeon experience. The consistency in perioperative care and surgical technique, while reducing variability, may limit generalizability. One limitation of our study is that the variable for prior cardiothoracic surgery did not distinguish between ipsilateral vs. contralateral procedures, or whether pleural space was entered. As such, the association observed may be driven primarily by ipsilateral lung resections, while inclusion of unrelated surgeries may dilute the strength of the association. Additionally, our dataset lacked several potentially relevant variables that could impact PAL risk, including preoperative FEV₁/forced vital capacity (FVC) ratios, serum albumin levels, corticosteroid use, the presence of pleural adhesions, details of intraoperative preventive measures, and the specific postoperative management of PAL.

Conclusions

The current findings suggest that CT-based emphysema score is associated with PAL following lung cancer resection, and that it may be a useful tool for preoperative risk stratification. In light of the significant patient morbidity and healthcare burden associated with PALs, the development of a single, objective, actionable parameter for risk stratification may be clinically useful. Future studies, with multi-center data and prospective designs, are needed to explore methods to integrate CT-quantified emphysema score into routine preoperative workflows while also establishing more robust clinical thresholds to guide treatment algorithms.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-946/rc

Data Sharing Statement: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-946/dss

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-946/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICJME uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-946/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. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Rush University Medical Center Institutional Review Board (IRB) (No. 24091203). Due to the retrospective nature of the study and minimal risk to participants, the Rush University Medical Center IRB waived informed consent for this study.

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. Aprile V, Bacchin D, Calabrò F, et al. Intraoperative prevention and conservative management of postoperative prolonged air leak after lung resection: a systematic review. J Thorac Dis 2023;15:878-92. [Crossref] [PubMed]
2. Rivera C, Bernard A, Falcoz PE, et al. Characterization and prediction of prolonged air leak after pulmonary resection: a nationwide study setting up the index of prolonged air leak. Ann Thorac Surg 2011;92:1062-8; discussion 1068. [Crossref] [PubMed]
3. Tong BC, Kosinski AS, Burfeind WR Jr, et al. Sex differences in early outcomes after lung cancer resection: analysis of the Society of Thoracic Surgeons General Thoracic Database. J Thorac Cardiovasc Surg 2014;148:13-8. [Crossref] [PubMed]
4. Kim WH, Lee HC, Ryu HG, et al. Intraoperative ventilatory leak predicts prolonged air leak after lung resection: A retrospective observational study. PLoS One 2017;12:e0187598. [Crossref] [PubMed]
5. Zheng Q, Ge L, Zhou J, et al. Risk factors for prolonged air leak after pulmonary surgery: A systematic review and meta-analysis. Asian J Surg 2022;45:2159-67. [Crossref] [PubMed]
6. Ponholzer F, Ng C, Maier H, et al. Risk factors, complications and costs of prolonged air leak after video-assisted thoracoscopic surgery for primary lung cancer. J Thorac Dis 2023;15:866-77. [Crossref] [PubMed]
7. Jin R, Zheng Y, Gao T, et al. A nomogram for preoperative prediction of prolonged air leak after pulmonary malignancy resection. Transl Lung Cancer Res 2021;10:3616-26. [Crossref] [PubMed]
8. Mueller MR, Marzluf BA. The anticipation and management of air leaks and residual spaces post lung resection. J Thorac Dis 2014;6:271-84. [Crossref] [PubMed]
9. Elsayed H, McShane J, Shackcloth M. Air leaks following pulmonary resection for lung cancer: is it a patient or surgeon related problem? Ann R Coll Surg Engl 2012;94:422-7. [Crossref] [PubMed]
10. Brunelli A, Varela G, Refai M, et al. A scoring system to predict the risk of prolonged air leak after lobectomy. Ann Thorac Surg 2010;90:204-9. [Crossref] [PubMed]
11. Lee L, Hanley SC, Robineau C, et al. Estimating the risk of prolonged air leak after pulmonary resection using a simple scoring system. J Am Coll Surg 2011;212:1027-32. [Crossref] [PubMed]
12. Attaar A, Winger DG, Luketich JD, et al. A clinical prediction model for prolonged air leak after pulmonary resection. J Thorac Cardiovasc Surg 2017;153:690-699.e2. [Crossref] [PubMed]
13. Pompili C, Falcoz PE, Salati M, et al. A risk score to predict the incidence of prolonged air leak after video-assisted thoracoscopic lobectomy: An analysis from the European Society of Thoracic Surgeons database. J Thorac Cardiovasc Surg 2017;153:957-65. [Crossref] [PubMed]
14. Ueda K, Kaneda Y, Sudo M, et al. Quantitative computed tomography versus spirometry in predicting air leak duration after major lung resection for cancer. Ann Thorac Surg 2005;80:1853-8. [Crossref] [PubMed]
15. Petrella F, Rizzo S, Radice D, et al. Predicting prolonged air leak after standard pulmonary lobectomy: computed tomography assessment and risk factors stratification. Surgeon 2011;9:72-7. [Crossref] [PubMed]
16. Shin S, Park HY, Kim H, et al. Joint effect of airflow limitation and emphysema on postoperative outcomes in early-stage nonsmall cell lung cancer. Eur Respir J 2016;48:1743-50. [Crossref] [PubMed]
17. Murakami J, Ueda K, Tanaka T, et al. Grading of Emphysema Is Indispensable for Predicting Prolonged Air Leak After Lung Lobectomy. Ann Thorac Surg 2018;105:1031-7. [Crossref] [PubMed]
18. Ozgur GK, Yavuz H, Cakan A, et al. Analysis of Factors Affecting Prolonged Air Leak and Expansion Failure in the Lung after Resection in Patients with Pulmonary Malignancy and Predictive Value of Preoperative Quantitative Chest Computed Tomography. Thorac Cardiovasc Surg 2025;73:418-26. [Crossref] [PubMed]
19. Voronoi Health Analytics. DAFS: Data Analysis Facilitation Suite – Uncover the arcane in medical imaging and unlock precision insights for a new era of healthcare. Accessed January 3, 2025. Available online: https://www.voronoihealthanalytics.com/dafs
20. Labaki WW, Xia M, Murray S, et al. Quantitative Emphysema on Low-Dose CT Imaging of the Chest and Risk of Lung Cancer and Airflow Obstruction: An Analysis of the National Lung Screening Trial. Chest 2021;159:1812-20. [Crossref] [PubMed]
21. Wang Z, Gu S, Leader JK, et al. Optimal threshold in CT quantification of emphysema. Eur Radiol 2013;23:975-84. [Crossref] [PubMed]
22. Atta H, Seifeldein GS, Rashad A, et al. Quantitative validation of the severity of emphysema by multi-detector CT. The Egyptian Journal of Radiology and Nuclear Medicine 2015;46:355-61.
23. Detterbeck FC. The eighth edition TNM stage classification for lung cancer: What does it mean on main street? J Thorac Cardiovasc Surg 2018;155:356-9.
24. Dezube AR, Dolan DP, Mazzola E, et al. Risk factors for prolonged air leak and need for intervention following lung resection. Interact Cardiovasc Thorac Surg 2022;34:212-8. [Crossref] [PubMed]
25. Chopra A, Hu K, Judson MA, et al. Association between Drainage-Dependent Prolonged Air Leak after Partial Lung Resection and Clinical Outcomes: A Prospective Cohort Study. Ann Am Thorac Soc 2022;19:389-98. [Crossref] [PubMed]
26. Fernandez FG, Kosinski AS, Burfeind W, et al. The Society of Thoracic Surgeons Lung Cancer Resection Risk Model: Higher Quality Data and Superior Outcomes. Ann Thorac Surg 2016;102:370-7. [Crossref] [PubMed]
27. Mallik A, Sen B, Banerjee M, et al. Threshold estimation based on a p-value framework in dose-response and regression settings. Biometrika 2011;98:887-900. [Crossref] [PubMed]
28. Busch EL. Cut points and contexts. Cancer 2021;127:4348-55. [Crossref] [PubMed]
29. Prince Nelson SL, Ramakrishnan V, Nietert PJ, et al. An evaluation of common methods for dichotomization of continuous variables to discriminate disease status. Commun Stat Theory Methods 2017;46:10823-34. [Crossref] [PubMed]
30. Brunelli A, Kim AW, Berger KI, et al. Physiologic evaluation of the patient with lung cancer being considered for resectional surgery: Diagnosis and management of lung cancer, 3rd ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest 2013;143:e166S-90S.
31. Varela G, Jiménez MF, Novoa N, et al. Estimating hospital costs attributable to prolonged air leak in pulmonary lobectomy. Eur J Cardiothorac Surg 2005;27:329-33. [Crossref] [PubMed]
32. Toloza EM, Harpole DH Jr. Intraoperative techniques to prevent air leaks. Chest Surg Clin N Am 2002;12:489-505. [Crossref] [PubMed]
33. Licker MJ, Widikker I, Robert J, et al. Operative mortality and respiratory complications after lung resection for cancer: impact of chronic obstructive pulmonary disease and time trends. Ann Thorac Surg 2006;81:1830-7. [Crossref] [PubMed]
34. Ferguson MK, Little L, Rizzo L, et al. Diffusing capacity predicts morbidity and mortality after pulmonary resection. J Thorac Cardiovasc Surg 1988;96:894-900.