Pattern-based volumetric CT quantification to predict radiation pneumonitis in patients with non-small-cell lung cancer who have diffuse parenchymal lung disease

Introduction

Definitive radiation therapy (RT) is a curative treatment option for non-small-cell lung cancer that is usually recommended for patients with early-stage disease who are medically inoperable, who refuse surgery, or who are high-risk surgical candidates (1). Radiation toxicity is one of the major concerns that arise in the treatment of lung cancer given the proximity of the lung, heart, major vessels, airways, spinal cord, and esophagus. Radiation pneumonitis (RP) is one of the pulmonary toxicities that limits the radiation dose and can lead to chronic pulmonary disease and even death. Previous studies have reported that significant RP occurs in roughly 10–40% of patients who received definitive RT for lung cancer. Age, sex, performance status, and pulmonary function were revealed as clinical risk factors for RP in those studies (2-6).

Underlying interstitial lung disease (ILD) is also a well-known significant risk factor for RP after RT. Especially, symptomatic and fatal RP occur more often in ILD patients (7-10). Some studies have suggested that proton beam therapy (PBT) is better than X-ray conformal RT in terms of the reduction of normal tissue dose (11-14). Even if patients have ILD, PBT is likely to be safer than X-ray conformal RT (15). However, RP remains a concerning complication of RT, including PBT, and makes it challenging for ILD patients to receive RT (16-18).

Some studies have attempted segmentation and quantitative analysis of ILD using computed tomography (CT) imaging and have evaluated the prognosis of ILD with the extent of ILD (19-23). Others have tried to predict cases of severe RP through quantitative analysis. Ye et al. quantified the amount of emphysematous tissue to evaluate the risk of RP (24), reporting that a smaller amount of emphysematous tissue was a significant risk factor for RP. To the best of our knowledge, there is no study predicting the severity of RP via pattern-based analysis using volumetric CT quantification. The aim of this study was therefore to quantify the extent of diffuse parenchymal lung disease (DPLD) relative to the whole-lung volume and determine which pattern(s) influence RP onset using texture analysis of pre-treatment chest CT scans. We present this article in accordance with the STROBE reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-7/rc).

Methods

Patients

This study was planned as a prospective study at a single tertiary referral hospital (Samsung Medical Center, Seoul, Republic of Korea). The institutional review board of Samsung Medical Center approved this study (No. 2020-12-060), and written informed consent was obtained from all patients before their inclusion. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Patients scheduled for proton therapy as a primary treatment for lung cancer were prospectively included from August 2018 to January 2020. Every patient was verified to be ineligible for surgery due to advanced age, impaired lung function, and/or poor performance status. For enrolled patients, the concurrent chemotherapeutic regimen included weekly paclitaxel combined with either cisplatin or carboplatin. A total of 54 patients who met the following inclusion criteria were enrolled: forced expiratory volume in one second (FEV₁) ≤1.0 L, and FEV₁ ≤50% predicted or diffusing capacity of the lungs for carbon monoxide (DLco) ≤50%, or pulmonary fibrosis. Diagnosis of underlying DPLD, such as chronic obstructive pulmonary disease (COPD), or ILD was confirmed by an experienced pulmonologist (H.Y). Specifically, idiopathic pulmonary fibrosis (IPF) was diagnosed according to the diagnostic criteria, which requires the exclusion of known causes and the presence of distinct features of CT and/or pathologic patterns (25,26). Figure 1 shows flow diagram of patient selection.

Figure 1 Flow diagram shows details of patient selection and analysis. CT, computed tomography; DLco, diffusing capacity of the lungs for carbon monoxide; FEV₁, forced expiratory volume in one second.

Diagnosis and staging for lung cancer

Every tumor was assessed through a physical examination, blood tests, pulmonary function test, chest X-ray, contrast-enhanced chest CT scan, whole-body ¹⁸F-fluorodeoxyglucose positron emission tomography/CT (PET/CT) scan, and brain magnetic resonance imaging (MRI). All tumors were staged based on the eighth edition of the American Joint Committee on Cancer tumor staging criteria (27). The pulmonary function test considered the following: (I) FEV₁; (II) forced vital capacity; (III) the FEV₁/forced vital capacity ratio; (IV) DLco; and (V) DLco divided by the alveolar ventilation.

Proton therapy

Four-dimensional simulation CT scans were acquired for treatment planning. The gross tumor volume was defined as the volume of the tumor identified based on all available clinical information. The internal target volume was delineated by combining all gross tumor volumes in each respiratory phase. The clinical target volume was generated with a 5-mm expansion from the internal target volume in all directions, then modified considering the adjacent anatomic structures. Elective irradiation of the clinically uninvolved lymph node was not allowed. The planning target volume was generated with 5-mm expansion of the clinical target volume. The percentage volume of lung receiving ≥20 Gy of the lung was maintained at ≤35%, and the mean lung dose was ≤20 Gy (28,29). Maximum doses to the spinal cord were not to exceed 50 Gy. The RayStation treatment planning system, version 6.2 (Raysearch Laboratories AB, Stockholm, Sweden) was utilized for treatment planning. PBT plans were calculated under the Monte Carlo algorithm or pencil beam algorithm (30). A fixed value of 1.1 was employed to account for the relative biological effectiveness. Daily image guidance was performed with cone-beam CT provided by VeriSuite (MedCom, Darmstdt, Germany) before each treatment session (30). At Samsung Medical Center, a proton therapy system (Sumitomo, Niihama, Japan) was utilized to implement either a continuous line-scanning method or a passive scattering method with wobbling for treatment (31).

Chest CT acquisition

CT scans were performed on a 64-channel multi-slice CT scanner (General Electric, Chicago, IL, USA; Siemens Healthcare, Munich, Germany; or Canon Medical Systems, Otawara, Japan). Images were obtained with the patient in a supine position with a voltage of 120 kVp and a slice thickness of 1–5 mm. With the General Electric scanner, standard and bone kernels were used. With the Siemens Healthcare scanner, B40f, B45f, and B50f kernels were used. With the Canon Medical Systems scanner, an FC04 kernel was used.

Radiologic analysis

Volumetric analysis was performed with pre-contrast images. Texture patterns were classified semi-automatically with the Aview COPD™ program (Corelinesoft, Seoul, Republic of Korea). First, the whole-lung volume was automatically calculated by the recognition of lung parenchyma, excluding pulmonary vessels or airways. After an overall analysis of the whole lung, each texture pattern analysis was completed. Texture patterns included emphysema, ground-glass opacities (GGOs), reticulation, and honeycombing (Figures 2,3 and Videos 1,2), and not only the absolute volume but also the percentage of the total lung volume affected were calculated for each. Discrimination between fibrosis and active inflammation is important for predicting the prognosis of ILD and treatment response. From a similar point of view, we hypothesized that the fibrotic component of ILD would have a greater impact on RP. In early stages of ILD, fibrosis presents reticular opacities on CT images. If airway dilatation and collapse of fibrotic alveoli progress, the areas show honeycombing. Therefore, the fibrotic component was calculated as a summation of the areas of reticulation and honeycombing (20,32-34). Automatic segmentation of texture patterns was precisely re-assessed by two radiologists (M.H.K. and H.Y.L.).

Figure 2 Images of lung cancer with underlying ILD in a 62-year-old man. (a,c) An axial non-contrast chest CT image shows subpleural GGOs, reticulation, and emphysema in both lungs. (b,d) Automatically categorized pattern-based map (purple, emphysema; blue, GGO; orange, reticulation). AEF is categorized as reticulation, which is a fibrosis-related pattern. The total lung volume was 4,668.8 cc. The total extent of ILD was 7.6% of the total lung volume (GGO, 1.0%; reticulation, 6.6%; honeycombing, 0.03%). Fibrosis is calculated as the summation of reticulation and honeycombing, which is 6.6% of the total lung volume. (e) Coronal post-contrast chest CT image obtained 2 months after proton therapy shows air space consolidation with GGOs, indicating radiation pneumonitis in the right upper lobe (white arrow). AEF, airspace enlargement with fibrosis; CT, computed tomography; GGO, ground-glass opacity; ILD, interstitial lung disease.

Figure 3 Images of lung cancer with underlying ILD in an 81-year-old man. (a) Axial non-contrast chest CT image shows subpleural GGOs, reticulation and emphysema in both lungs. (b) Automatically categorized pattern-based map (purple, emphysema; blue, GGOs; orange, reticulation). The total lung volume was 6,372.7 cc. The total extent of ILD was 18.3% of the total lung volume (GGOs, 1.7%; reticulation, 16.7%; honeycombing, 0.003%). Fibrosis is calculated as the summation of reticulation and honeycombing, which is 16.7% of the total lung volume. (c) Coronal non-contrast chest CT image obtained 5 months after proton therapy shows air space consolidation with GGOs, indicating radiation pneumonitis in the right middle lobe and right lower lobe (white arrow). CT, computed tomography; GGO, ground-glass opacity; ILD, interstitial lung disease.

Follow-up

Follow-up contrast-enhanced chest CT imaging was performed 1 month after the last day of treatment, then every 2–3 months for 2 years thereafter. When GGOs or air space consolidation developed on chest CT images following the termination of RT, RP was suspected. Other causes of pneumonia, such as infection or drug-induced pneumonitis, were excluded after additional laboratory testing, such as sputum culture or serology, or empirical use of antibiotics. After complete exclusion of other causes of pneumonitis, the RP grade was evaluated through the Common Terminology Criteria for Adverse Events version 5.0.

Statistical analysis

We divided the study group into two groups: patients with RP of grade <3, and patients with RP of grade ≥3. To examine the impact of biologically effective dose at α/β of 10 Gy (BED10) between the two groups (RP <3 vs. RP ≥3), patients were categorized into two groups: a low BED10 group (BED10 <100 Gy) and a high BED10 group (BED10 ≥100 Gy), according to previous literature (35). We compared patients’ demographics and clinical characteristics between the two groups using Student’s t-test, the Mann-Whitney U test, or Fisher’s exact test, as appropriate. The Mann-Whitney U test was used to compare the quantitative DPLD parameters between the two groups. Sample size calculations were conducted based on the assumption that, given a two-sided significance level of 0.05 and power of 0.80, the incidences of grade 3 or higher pulmonary toxicity would be 25% and 10% after X-ray treatment and PBT, respectively (15,36). We assumed an accrual rate of three patients per month and an additional follow-up period of 1 year. Therefore, the required sample size was 50 patients. Considering a 10% dropout rate, the total planned sample size was 55 patients. Univariable and multivariable logistic regression analyses were used to analyze independent risk factors for RP of grade ≥3. To find the optimal cutoff value which gives high probability that patients with greater value will have a severe RP, logistic regression analysis was performed for two groups based on the cutoff value while increasing the value by 1 from minimum to maximum. All statistical analyses were performed with SAS version 9.4 (SAS Institute, Cary, NC, USA). P<0.05 was considered to be statistically significant.

Results

Baseline characteristics

The baseline characteristics of patients and lung cancer are summarized in Table 1. A total of 54 patients were included, including 50 male patients (92.6%). The median age of the study population was 71.5 years (range, 57–87 years). T1, N0 was the most common TN stage, while 21 patients (38.9%) had T1-stage disease and 35 patients (64.8%) had N0-stage disease. Fifty-two patients (96.3%) had restrictions on physically strenuous activity. Fifty-one patients (94.4%) were current or previous smokers. Regarding underlying lung disease, we identified that COPD was diagnosed in 19 (35.2%) patients and IPF was diagnosed in 15 (27.8%) patients. Among 16 patients with non-IPF ILD, 14 patients (25.9%) were diagnosed with the combined pulmonary fibrosis and emphysema (CPFE), while the remaining 2 patients (3.7%) had ILD associated with rheumatoid arthritis (RA) or microscopic polyangiitis (MPA). The number of ports used in treatment was 2 in 17 patients (31.5%), 3 in 35 patients (64.8%), and 4 in 2 patients (3.7%), while the beam angles were individualized based on the patient’s tumor location and proximity to organs at risk.

Proton therapy

The median and mean clinical target volume were 133.4 and 183.8 cm³, respectively. The most frequently prescribed dose schedules for early-stage disease were 64 GyE in eight fractions (n=14, 25.9%) and 60 GyE in four fractions (n=13, 24.1%), respectively, while 66 GyE in 30 fractions was the most frequently adopted dose schedule for locally advanced disease (n=10, 18.5%; nine patients received concurrent chemoradiation therapy) (37). Overall, a median total dose of 64.0 GyE (range, 60.0–70.4 GyE) with a fraction dose of 7.0 GyE (range, 2.2–15.0 GyE) was delivered. Dose-volume parameters are summarized in Table S1. The percentage volume of lung receiving ≥5 Gy, ≥20 Gy, and mean doses to the lung were lower than the planning criteria (65%, 35%, and 20 GyE, respectively).

RP

The median follow-up period was 19.0 months (range, 1.1–35.9 months). A total of 43 patients (79.6%) were confirmed to have RP (Figures 2,3). Among all patients, 47 had RP of grade <3. Eighteen patients (33.3%) were asymptomatic. Every symptomatic patient was treated with oral steroid treatment. Seven patients had RP of grade ≥3, i.e., severe symptomatic RP requiring O₂ resuscitation. Among them, three patients expired after aggravation of RP.

A comparison of baseline characteristics of the patients and lung cancer cases in two separate groups is summarized in Table 2. Clinical characteristics were similar between the RP grade <3 and grade ≥3 groups. When FEV₁ was stratified into two categories (>50% vs. ≤ 50%), the numbers of patients who had low FEV₁ in the RP grade <3 and grade ≥3 groups were 14 (29.8%) and 1 (14.3%), respectively (P=0.39). The median age was 72 years in the RP grade <3 group and 69 years in the RP grade ≥3 group (P=0.47). The median tumor size was 3.2 cm in the RP grade <3 group and 3.3 cm in the RP grade ≥3 group (P=0.60), respectively, and the T, N stage was also not significantly different between the two groups (P=0.17 and 0.057, respectively). We identified the difference in pneumonia incidence between patients classified as usual interstitial pneumonia (UIP) and non-UIP. Among the 54 enrolled patients, 29 patients (53.7%) were classified as UIP, with 17.2% pneumonia incidence, while non-UIP patients had an 8% incidence of pneumonia. However, the difference was not statistically significant (P=0.31). Among the patients with RP grade ≥3, the proportion of patients with BED10 <100 Gy was higher (85.7%) compared to those with BED10 ≥100 Gy (14.3%), which suggested that higher BED10 did not have a negative impact on the occurrence of severe RP. Additionally, no significant differences were found for any of laboratory measurements.

Quantitative DPLD parameters

The median time taken to evaluate the whole lung pattern was 87 seconds (standard deviation, 16.4 seconds). Among 54 patients, the median whole-lung volume was 4,796.5 cc (range, 2,636.2–6,836.6 cc). The median volume of emphysematous tissue was 221.7 cc (range, 0.1–1,813.1 cc), which totaled 4.8% (range, 0.0–34.1%) of the whole lung. The median volume of total ILD components was 183.5 cc (range, 0.0–1,169.1 cc), which totaled 5.5% (range, 0.0–27.3%) of the whole lung. GGOs were the largest ILD component, with a median volume of 75.7 cc (range, 0.0–697.8 cc) or 1.5% (range, 0.0–16.7%) of the whole lung.

Total ILD components in the whole lung were significantly higher in the RP grade ≥3 group than the RP grade <3 group (P=0.047). The area of GGOs in the contralateral lung was larger in the RP grade ≥3 group than the RP grade <3 group (P=0.042). However, the rates of emphysema, reticulation, honeycombing, and fibrotic components were not significantly different between the two groups (Table 3).

Relationship among clinical parameters, quantitative DPLD parameters, and the risk of RP

In the univariable logistic regression analysis, female sex was the only clinical parameter significantly associated with severe RP (grade ≥3) [odds ratio (OR), 9; P=0.047] (Table 2). Among quantitative DPLD parameters, total ILD (OR, 1.1; P=0.02) and fibrotic components (OR, 1.1; P=0.04) in the whole lung were significantly associated with severe RP. However, honeycombing was not significantly associated with severe RP (P=0.14). Emphysema showed a negative association with severe RP, but not statistically significantly so (OR, 0.9; P=0.25). In the contralateral lung, GGOs (OR, 1.2; P=0.04) and reticulation (OR, 1.2; P=0.046) were significantly associated with severe RP (Table 3).

Multivariable analysis for severe RP

In the multivariable logistic regression analysis, we screened potential models with female sex and quantitative DPLD parameters, which were significantly associated with severe RP. Among the models, the “sex + total ILD component” model and “sex + total fibrotic component” model showed the best performance. In the first model, RP of grade ≥3 was associated with female sex and a high total ILD percentage [OR, 18.0 and OR, 1.2, respectively; area under the curve (AUC), 0.806]. In the second model, RP of grade ≥3 was associated with female sex and a high total fibrosis percentage (OR, 16.7 and OR, 1.2, respectively; AUC, 0.787) (Table 4).

Cutoff value of quantitative DPLD parameters for predicting severe RP

We analyzed optimal cutoff value of quantitative DPLD parameters for predicting severe RP. When ILD affected lung is more than 22% of whole lung volume, the predictive performance of severe RP was the best (AUC, 0.831; P<0.001). Fibrosis accounting for more than 8% of whole lung volume best predicted the severe RP (AUC, 0.774; P=0.02). AUC of logistic regression model using each cutoff value is shown in the graph (Figures 4,5).

Figure 4 Area under the curve of logistic regression model using cutoff value of total ILD component while increasing the value by 1 from minimum to maximum. ILD, interstitial lung disease.

Figure 5 Area under the curve of logistic regression model using cutoff value of total fibrotic component while increasing the value by 1 from minimum to maximum.

Discussion

Nowadays, CT imaging is used not only for radiologic diagnosis but also for the prediction of prognosis or treatment response. Some studies of the quantification of ILD have been performed to predict the prognosis of ILD. Carvalho et al. studied the quantification of ILD in a systemic sclerosis cohort and reported that patients with ILD affecting >29.6% of the total lung weight presented with significantly lower DLco and total lung volume compared to others (22). Marten et al. reported on computer-aided quantification of ILD associated with collagen vascular disease (23). Most quantitative analytic studies have used thresholding methods to extract regions with attenuation values above defined thresholds (38). It was difficult to distinguish between emphysema and airspace enlargement with fibrosis. Airspace enlargement with fibrosis is characterized by multiple thin-walled cystic lesions; however, it is a kind of histologically intensive hyalinized fibrosis (39,40). Therefore, this pattern should be classified as fibrosis, not emphysema. Our program is trained to classify airspace enlargement with fibrosis as a fibrosis-related pattern, which can be explained as one reason for its good predictive performance.

Meanwhile, when classifying each pattern of ILD involvement, interobserver reliability could not be guaranteed through visual assessment. Walsh et al. reported that κ values indicating interobserver agreement for honeycombing, traction bronchiectasis, and emphysema were 0.59±0.12, 0.42±0.15, and 0.43±0.18, respectively (41). Watadani et al. reported that the agreement score for the presence of honeycombing among 43 observers was moderate and the value was 0.40–0.58 (42). An automatic computed software was designed to assist the radiologists and solve the reliability problem. Furthermore, manual segmentation and visual assessment are time-consuming tasks. In the current study, the time taken to evaluate the whole lung pattern and calculate each amount was reduced to <2 minutes using an automatic computed system.

There are several previous studies about the relationship between ILD and RP in patients treated with conventional X-ray RT. Lee et al. reported that RP was observed in 60.0% (9/15) of patients with diffuse interstitial changes compared to 28.9% (13/45) of patients without interstitial changes (7). They also reported that interstitial changes raised the risk for RP of grade ≥3. Ueki et al. reported that ILD is a significant risk factor for RP of grade >2 and leads to worse survival outcomes (9). Some studies suggested that proton therapy is safer than photon RT for patients with lung tumors and even for patients with underlying ILD. In a recent study, treatment-related death occurred in 18.2% (4/22) of IPF patients after photon RT but none with proton therapy (15). The incidence of severe pulmonary complications was not significantly different between X-ray and proton groups, although it was observed less frequently in the proton group (40.9% vs. 12.5%). A meta-analysis comparing particle beam therapy and RT reported that the incidence of RP of grade >2 following particle beam therapy vs. conventional RT was 0.9% vs. 3.4% (P<0.001) (43). However, there were no studies using CT quantification for evaluation of RP in proton therapy group.

To the best of our knowledge, this study is the first to evaluate the risk of RP in patients with lung cancer and underlying ILD by volumetric pattern-based analysis. Furthermore, this study enrolled patients treated by PBT as it is more applied to patients with low pulmonary function and so more precise prediction for RP is needed. Lungs with more ILD affecting the parenchyma and more fibrosis are more susceptible to RP. Notably, lungs with more emphysematous tissue were less vulnerable to radiation, but the difference was statistically insignificant. Ye et al. also reported that an increased percentage of emphysematous tissue was associated with a significantly lower rate of consolidation changes 6 months after RT (24). The more emphysematous tissue there is, the less the normal lung tissue will be affected by RT.

Our study also examined about optimal cutoff value of total ILD component and total fibrotic component for predicting severe RP. Ueki et al. examined about relationship between RP incidence and severity of ILD which is visually assessed (9). They reported that the cumulative incidence of RP of grade >2 is significantly higher in meaningful ILD than others. They defined significant ILD if the interstitial changes involve more than 25% of each lobes using the scoring system proposed by Kazerooni et al. (44). Our optimal cutoff value of total ILD component was 22% which is similar with the previous study. However, the odds of severe RP increased linearly as the total ILD and fibrotic component increases and the data was crowded so we couldn’t find clear cutoff value to divide the study group into two pieces.

This concept can be applied not only to RP but also to other pulmonary diseases in which ILD may act as a risk factor. For example, the importance of drug-induced pneumonitis is emerging as the use of target agents or immune checkpoint inhibitors increases. For these drugs, early recognition of drug-induced pneumonitis is important because of the long-term exposure risk (45). Shibaki et al. reported that the incidence of immune-related pneumonitis was significantly higher in patients with pre-existing ILD than in patients without pre-existing ILD (29% vs. 9.9%, respectively). They also reported that pneumonitis occurred earlier in patients with pre-existing ILD compared to other patients (46). Therefore, our attempt at categorizing and quantifying DPLD using volumetric pattern-based analysis may pave the way to examine the risk of drug-related pneumonitis.

Our study had several limitations. First, this was a single-center study. Therefore, selection bias was inevitable. Second, the sample size was relatively small. Therefore, exact matching between the two groups was not feasible. However, there was no significant difference in patient or tumor characteristics between the groups. Third, CT scanners from multiple vendors were used in each patient. To minimize the variability associated with multiple-vendor CT scanning with different scan parameters and manufacturers, we performed quantitative analysis using advanced software incorporating deep learning-based technique (47). Fourth, we considered lung dose-volume constraints to be non-customized, while dose fractionation was individualized based on tumor size and location. Recent studies have reported that when planning stereotactic body radiotherapy (SBRT) for patients with ILD, dose-volume constraints were recommended to be set differently from those used in conventional radiotherapy plan (25,26). However, we did not consider those recommendations at the time of study design. Despite this, our study demonstrated that lung dose-volume parameters, including V5, V20, and mean lung dose, remained below the standard planning criteria (65%, 35%, and 20 GyE, respectively) in Table S1. Proof of concept was the primary objective of this study, so further well-designed prospective studies with larger sample sizes are warranted to verify the clinical utility of our results.

Conclusions

In conclusion, the total amount of ILD, especially the amount of fibrosis, affects the risk of severe RP after proton therapy for lung cancer in patients with underlying DPLD. Therefore, we can provide additional information on radiation dose adjustment by calculating what percentage of the lung parenchyma is occupied by ILD through volumetric CT quantification.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-7/rc

Data Sharing Statement: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-7/dss

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-7/prf

Funding: This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. NRF-2022R1A2C1003999); Future Medicine 20*30 Project of the Samsung Medical Center (#SMO1240791); and partly supported by Institute of Information & communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (MSIT) (No. RS-2021-II212068, Artificial Intelligence Innovation Hub).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-7/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 institutional review board of Samsung Medical Center approved this study (No. 2020-12-060), and written informed consent was obtained from all patients before their inclusion. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

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. National Comprehensive Cancer Network. Non-small cell lung cancer (Version 3.2025). Accessed 11 May 2025. Available online: https://www.nccn.org/professionals/physician_gls/pdf/nscl.pdf
2. Robnett TJ, Machtay M, Vines EF, et al. Factors predicting severe radiation pneumonitis in patients receiving definitive chemoradiation for lung cancer. Int J Radiat Oncol Biol Phys 2000;48:89-94. [Crossref] [PubMed]
3. Claude L, Pérol D, Ginestet C, et al. A prospective study on radiation pneumonitis following conformal radiation therapy in non-small-cell lung cancer: clinical and dosimetric factors analysis. Radiother Oncol 2004;71:175-81. [Crossref] [PubMed]
4. Park YH, Kim JS. Predictors of radiation pneumonitis and pulmonary function changes after concurrent chemoradiotherapy of non-small cell lung cancer. Radiat Oncol J 2013;31:34-40. [Crossref] [PubMed]
5. Shi A, Zhu G, Wu H, et al. Analysis of clinical and dosimetric factors associated with severe acute radiation pneumonitis in patients with locally advanced non-small cell lung cancer treated with concurrent chemotherapy and intensity-modulated radiotherapy. Radiat Oncol 2010;5:35. [Crossref] [PubMed]
6. Choi SH, Kim E, Heo SJ, et al. Integrative prediction model for radiation pneumonitis incorporating genetic and clinical-pathological factors using machine learning. Clin Transl Radiat Oncol 2024;48:100819. [Crossref] [PubMed]
7. Lee YH, Kim YS, Lee SN, et al. Interstitial Lung Change in Pre-radiation Therapy Computed Tomography Is a Risk Factor for Severe Radiation Pneumonitis. Cancer Res Treat 2015;47:676-86. [Crossref] [PubMed]
8. Onishi H, Yamashita H, Shioyama Y, et al. Stereotactic Body Radiation Therapy for Patients with Pulmonary Interstitial Change: High Incidence of Fatal Radiation Pneumonitis in a Retrospective Multi-Institutional Study. Cancers (Basel) 2018;10:257. [Crossref] [PubMed]
9. Ueki N, Matsuo Y, Togashi Y, et al. Impact of pretreatment interstitial lung disease on radiation pneumonitis and survival after stereotactic body radiation therapy for lung cancer. J Thorac Oncol 2015;10:116-25. [Crossref] [PubMed]
10. Peng J, Bie Z, Li Y, et al. Safety and efficacy of CT-guided percutaneous microwave ablation for stage I non-small cell lung cancer in patients with comorbid idiopathic pulmonary fibrosis. Eur Radiol 2024;34:4708-15. [Crossref] [PubMed]
11. Chang JY, Zhang X, Wang X, et al. Significant reduction of normal tissue dose by proton radiotherapy compared with three-dimensional conformal or intensity-modulated radiation therapy in Stage I or Stage III non-small-cell lung cancer. Int J Radiat Oncol Biol Phys 2006;65:1087-96. [Crossref] [PubMed]
12. Ohno T, Oshiro Y, Mizumoto M, et al. Comparison of dose-volume histograms between proton beam and X-ray conformal radiotherapy for locally advanced non-small-cell lung cancer. J Radiat Res 2015;56:128-33. [Crossref] [PubMed]
13. Wu CT, Motegi A, Motegi K, et al. Dosimetric comparison between proton beam therapy and photon radiation therapy for locally advanced non-small cell lung cancer. Jpn J Clin Oncol 2016;46:1008-14. [Crossref] [PubMed]
14. Hoppe BS, Nichols RC, Flampouri S, et al. Chemoradiation with Hypofractionated Proton Therapy in Stage II-III Non-Small Cell Lung Cancer: A Proton Collaborative Group Phase 2 Trial. Int J Radiat Oncol Biol Phys 2022;113:732-41. [Crossref] [PubMed]
15. Kim H, Pyo H, Noh JM, et al. Preliminary result of definitive radiotherapy in patients with non-small cell lung cancer who have underlying idiopathic pulmonary fibrosis: comparison between X-ray and proton therapy. Radiat Oncol 2019;14:19. [Crossref] [PubMed]
16. Nagano T, Kotani Y, Fujii O, et al. A case of acute exacerbation of idiopathic pulmonary fibrosis after proton beam therapy for non-small cell lung cancer. Jpn J Clin Oncol 2012;42:965-9. [Crossref] [PubMed]
17. Ono T, Hareyama M, Nakamura T, et al. The clinical results of proton beam therapy in patients with idiopathic pulmonary fibrosis: a single center experience. Radiat Oncol 2016;11:56. [Crossref] [PubMed]
18. Benveniste MF, Gomez D, Carter BW, et al. Recognizing Radiation Therapy-related Complications in the Chest. Radiographics 2019;39:344-66. [Crossref] [PubMed]
19. Chen A, Karwoski RA, Gierada DS, et al. Quantitative CT Analysis of Diffuse Lung Disease. Radiographics 2020;40:28-43. [Crossref] [PubMed]
20. Moon JW, Bae JP, Lee HY, et al. Perfusion- and pattern-based quantitative CT indexes using contrast-enhanced dual-energy computed tomography in diffuse interstitial lung disease: relationships with physiologic impairment and prediction of prognosis. Eur Radiol 2016;26:1368-77. [Crossref] [PubMed]
21. Park B, Park H, Lee SM, et al. Lung Segmentation on HRCT and Volumetric CT for Diffuse Interstitial Lung Disease Using Deep Convolutional Neural Networks. J Digit Imaging 2019;32:1019-26. [Crossref] [PubMed]
22. Carvalho ARS, Guimarães AR, Sztajnbok FR, et al. Automatic Quantification of Interstitial Lung Disease From Chest Computed Tomography in Systemic Sclerosis. Front Med (Lausanne) 2020;7:577739. [Crossref] [PubMed]
23. Marten K, Dicken V, Kneitz C, et al. Interstitial lung disease associated with collagen vascular disorders: disease quantification using a computer-aided diagnosis tool. Eur Radiol 2009;19:324-32. [Crossref] [PubMed]
24. Ye L, Zhang Y, Li T, et al. Emphysema quantification on computed tomography and its value in predicting radiation pneumonitis in lung cancer treated by stereotactic body radiotherapy. J Radiat Res 2019;60:371-9. [Crossref] [PubMed]
25. Kong FS, Moiseenko V, Zhao J, et al. Organs at Risk Considerations for Thoracic Stereotactic Body Radiation Therapy: What Is Safe for Lung Parenchyma? Int J Radiat Oncol Biol Phys 2021;110:172-87. [Crossref] [PubMed]
26. Saha A, Dickinson P, Shrimali RK, et al. Is Thoracic Radiotherapy an Absolute Contraindication for Treatment of Lung Cancer Patients With Interstitial Lung Disease? A Systematic Review. Clin Oncol (R Coll Radiol) 2022;34:e493-504. [Crossref] [PubMed]
27. Amin MB, Edge S, Greene F, et al. AJCC Cancer Staging Manual. 8th ed. Cham, Switzerland: Springer International Publishing; 2017.
28. Mehta V. Radiation pneumonitis and pulmonary fibrosis in non-small-cell lung cancer: pulmonary function, prediction, and prevention. Int J Radiat Oncol Biol Phys 2005;63:5-24. [Crossref] [PubMed]
29. Antonia SJ, Villegas A, Daniel D, et al. Durvalumab after Chemoradiotherapy in Stage III Non-Small-Cell Lung Cancer. N Engl J Med 2017;377:1919-29. [Crossref] [PubMed]
30. Noh JM, Yoo H, Lee W, et al. Prospective Study of Proton Therapy for Lung Cancer Patients with Poor Lung Function or Pulmonary Fibrosis. Cancers (Basel) 2022;14:1445. [Crossref] [PubMed]
31. Chung K, Han Y, Kim J, et al. The first private-hospital based proton therapy center in Korea; status of the Proton Therapy Center at Samsung Medical Center. Radiat Oncol J 2015;33:337-43. [Crossref] [PubMed]
32. Jacob J, Hansell DM. HRCT of fibrosing lung disease. Respirology 2015;20:859-72. [Crossref] [PubMed]
33. Johkoh T, Sakai F, Noma S, et al. Honeycombing on CT; its definition, pathologic correlation, and future direction of its diagnosis. Eur J Radiol 2014;83:27-31. [Crossref] [PubMed]
34. Cocconcelli E, Balestro E, Biondini D, et al. High-Resolution Computed Tomography (HRCT) Reflects Disease Progression in Patients with Idiopathic Pulmonary Fibrosis (IPF): Relationship with Lung Pathology. J Clin Med 2019;8:399. [Crossref] [PubMed]
35. Guckenberger M, Andratschke N, Dieckmann K, et al. ESTRO ACROP consensus guideline on implementation and practice of stereotactic body radiotherapy for peripherally located early stage non-small cell lung cancer. Radiother Oncol 2017;124:11-7. [Crossref] [PubMed]
36. Kim H, Yoo H, Pyo H, et al. Impact Of Underlying Pulmonary Diseases On Treatment Outcomes In Early-Stage Non-Small Cell Lung Cancer Treated With Definitive Radiotherapy. Int J Chron Obstruct Pulmon Dis 2019;14:2273-81. [Crossref] [PubMed]
37. Kim N, Noh JM, Lee W, et al. Clinical Outcomes of Pencil Beam Scanning Proton Therapy in Locally Advanced Non-Small Cell Lung Cancer: Propensity Score Analysis. Cancers (Basel) 2021;13:3497. [Crossref] [PubMed]
38. Ko JP, Naidich DP. Computer-aided diagnosis and the evaluation of lung disease. J Thorac Imaging 2004;19:136-55. [Crossref] [PubMed]
39. Watanabe Y, Kawabata Y, Kanauchi T, et al. Multiple, thin-walled cysts are one of the HRCT features of airspace enlargement with fibrosis. Eur J Radiol 2015;84:986-92. [Crossref] [PubMed]
40. Yamada T, Nakanishi Y, Homma T, et al. Airspace enlargement with fibrosis shows characteristic histology and immunohistology different from usual interstitial pneumonia, nonspecific interstitial pneumonia and centrilobular emphysema. Pathol Int 2013;63:206-13. [Crossref] [PubMed]
41. Walsh SL, Calandriello L, Sverzellati N, et al. Interobserver agreement for the ATS/ERS/JRS/ALAT criteria for a UIP pattern on CT. Thorax 2016;71:45-51. [Crossref] [PubMed]
42. Watadani T, Sakai F, Johkoh T, et al. Interobserver variability in the CT assessment of honeycombing in the lungs. Radiology 2013;266:936-44. [Crossref] [PubMed]
43. Chi A, Chen H, Wen S, et al. Comparison of particle beam therapy and stereotactic body radiotherapy for early stage non-small cell lung cancer: A systematic review and hypothesis-generating meta-analysis. Radiother Oncol 2017;123:346-54. [Crossref] [PubMed]
44. Kazerooni EA, Martinez FJ, Flint A, et al. Thin-section CT obtained at 10-mm increments versus limited three-level thin-section CT for idiopathic pulmonary fibrosis: correlation with pathologic scoring. AJR Am J Roentgenol 1997;169:977-83. [Crossref] [PubMed]
45. Lee HY. A Wake-Up Call for Immune Checkpoint Inhibitor-Related Pneumonitis. J Thorac Oncol 2021;16:1440-2. [Crossref] [PubMed]
46. Shibaki R, Murakami S, Matsumoto Y, et al. Association of immune-related pneumonitis with the presence of preexisting interstitial lung disease in patients with non-small lung cancer receiving anti-programmed cell death 1 antibody. Cancer Immunol Immunother 2020;69:15-22. [Crossref] [PubMed]
47. Hwang HJ, Kim H, Seo JB, et al. Generative Adversarial Network-Based Image Conversion Among Different Computed Tomography Protocols and Vendors: Effects on Accuracy and Variability in Quantifying Regional Disease Patterns of Interstitial Lung Disease. Korean J Radiol 2023;24:807-20. [Crossref] [PubMed]