Quantitative CT Imaging Biomarkers of Interstitial Lung Disease Are Associated with Survival in Non–Small Cell Lung Cancer

Quantitative CT Imaging Biomarkers of Interstitial Lung Disease Are Associated with Survival in Non–Small Cell Lung Cancer | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Quantitative CT Imaging Biomarkers of Interstitial Lung Disease Are Associated with Survival in Non–Small Cell Lung Cancer Chunhui Qin, Kang Yu, Yuxin Li, Chunxiao Wang, Jiaheng Xu, Ling Liu, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8506344/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Background Interstitial lung disease (ILD) is a clinically important but often under-recognized comorbidity in patients with non–small cell lung cancer (NSCLC). Although ILD has been associated with adverse outcomes, objective and reproducible imaging biomarkers for prognostic assessment remain limited. This study aimed to evaluate the prognostic value of quantitative CT-derived ILD features in NSCLC. Methods In this retrospective cohort study, 324 patients with pathologically confirmed NSCLC and coexisting ILD were included. Automated CT-based segmentation was applied to quantify tumor volume and ILD-related components, including ground-glass opacity, emphysema, honeycombing, and fibrosis. Fibrosis was defined as the sum of reticulation, traction bronchiectasis, and honeycombing, with honeycombing analyzed as an independent subtype. Visual assessment was independently performed by two experienced radiologists. Cox proportional hazards regression was used to identify prognostic factors for overall survival (OS). Prognostic models were constructed using clinical variables, quantitative CT features, and their combinations. Results In multivariable Cox analysis, both honeycombing volume and total fibrosis volume were independently associated with poorer OS after adjustment for clinical covariates. Prognostic models incorporating quantitative CT features demonstrated improved discrimination compared with clinical variables alone. The combined model integrating clinical variables, quantitative CT features, and radiologists’ visual assessment achieved the best overall performance. Conclusions Quantitative CT imaging biomarkers of ILD are independently associated with survival in patients with NSCLC. Integrating objective imaging-derived ILD features with clinical evaluation may improve risk stratification and support individualized management in this patient population. Non–small cell lung cancer Interstitial lung disease Computed tomography Quantitative imaging Prognosis Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Lung cancer remains the leading cause of cancer-related mortality worldwide. Patients with comorbid interstitial lung disease (ILD) face even poorer outcomes due to reduced pulmonary reserve and increased treatment-related risks. Accurate prognostic assessment in this population is essential for guiding individualized treatment strategies. [ 1 , 6 , 7 ] Current approaches for ILD assessment rely primarily on visual scoring systems, such as the ATS/ERS and Fleischner criteria. However, visual evaluation is inherently subjective, with limited reproducibility and interobserver agreement. Quantitative CT-based measures, particularly those focusing on fibrosis and honeycombing, have shown promise as imaging biomarkers in ILD and lung cancer. Despite this, their role as independent prognostic factors in lung cancer has not been fully validated. [ 1 , 2 , 3 , 4 , 5 , 6 , 7 ] In parallel, deep learning (DL) has advanced applications in medical imaging, ranging from lesion detection to segmentation and prognostic modeling. While DL models demonstrate excellent performance, concerns about overfitting and generalizability remain, particularly when applied to survival prediction. Ensemble learning may overcome these limitations by combining complementary models for robust predictions. [ 6 , 7 ] Therefore, the aim of this study was to evaluate the prognostic significance of quantitative ILD features obtained from automated segmentation, including honeycombing and total fibrosis volume, in patients with lung cancer. We further compared traditional Cox regression, individual DL models, and ensemble strategies to establish an integrative prognostic framework. [ 1 , 6 , 7 ] Materials and Methods Study Sample This retrospective cohort study was approved by the Institutional Ethics Committee of the Fourth Affiliated Hospital of Harbin Medical University (Approval No. 2025 − 125).and conducted in accordance with the principles of the Declaration of Helsinki [ 8 ]. Written informed consent was waived due to the retrospective nature of the study. Between August 2020 and June 2024, a total of 5,016 patients with pulmonary lesions on chest CT were consecutively enrolled in this retrospective cohort study conducted. [ 9 ]. After excluding 244 patients with small cell lung cancer and 2,142 patients with other pathological types [ 10 ], 2,630 patients with non-small cell lung cancer (NSCLC) were included. Among them, 1,917 patients without interstitial lung disease (ILD) were excluded [ 11 ], leaving 713 NSCLC patients with ILD. Subsequently, 119 patients were excluded due to poor image quality, tuberculosis, pleural effusion, pneumothorax, atelectasis, prior lung surgery, radiotherapy, chemotherapy, targeted therapy, immunotherapy, or pulmonary metastases [ 12 ]. Furthermore, 270 patients were excluded because of missing follow-up information, follow-up time less than 12 months, or incomplete clinical/imaging data [ 13 ]. Finally, 324 patients were included in the analysis. Follow-up information was obtained from the hospital electronic medical record system and telephone interviews [ 14 ].The patient selection process is illustrated in Fig. 1 . Imaging Section CT Image Acquisition and Preprocessing All patients underwent chest CT examinations using multi-vendor and multi-model spiral CT scanners (GE, Siemens, Philips, etc.). The scanning parameters were as follows: tube voltage 120 kVp, automatic tube current modulation, slice thickness 1.25 mm, matrix size 512 × 512, pitch 1.0, and iterative reconstruction. To minimize the impact of scanner variability, all images were standardized and preprocessed, including intensity normalization and resampling to isotropic voxels[ 16 ]. ILD Diagnostic Criteria ILD diagnosis was based on routine radiology reports interpreted by experienced radiologists, in accordance with internationally accepted HRCT diagnostic criteria (ATS/ERS/JRS/ALAT guidelines and Fleischner Society recommendations)[ 17 , 18 ]. Thus, ILD inclusion and classification in this study followed the standards of routine clinical practice to ensure diagnostic consistency. Image Segmentation and Feature Calculation Image segmentation was performed using a deep learning model based on the U-Net architecture, which had been pretrained by a collaborating team (detailed network parameters and performance metrics were not disclosed). The model automatically segmented ground-glass opacities, emphysema, honeycombing, fibrosis, tumors, and normal lung tissue. All segmentation outputs were reviewed by experienced radiologists, and cases with substantial errors were manually corrected. Detailed model parameters were not publicly available, which represents a limitation of this study. Fibrosis is defined as the sum of reticulation, bronchial traction dilatation, and honeycombing. Honeycombing was considered an independent subtype of fibrosis and was segmented and analyzed separately. Subsequently, volumes of each ILD subtype, tumor, and the whole lung were calculated, and their relative proportions were derived to obtain quantitative ILD features. Visual Assessment Two experienced radiologists independently performed visual assessment of CT images while blinded to patients’ clinical and pathological information. The method was adopted from the Radiology pulmonary hypertension study (Dwivedi K et al., 2024)[ 19 ] and consisted of two components: 1) Ordinal grading: ILD-related imaging findings (such as honeycombing, ground-glass opacity, emphysema, and fibrosis) were graded on a four-point scale (0 = absent, 1 = mild, 2 = moderate, 3 = severe). 2) Binary classification: Presence of each imaging feature was recorded as absent (0) or present (1). Inter-observer agreement was evaluated using Cohen’s Kappa coefficient. The unweighted Kappa was 0.598, indicating moderate agreement, while the weighted Kappa was 0.800, suggesting good agreement for the ordinal grading[ 20 ]. Statistical Analysis Progression-free survival (PFS) was defined as the time from diagnosis to either distant metastasis or death, with patients censored at the last follow-up if no event occurred. All statistical analyses were performed using R software (version 4.3.0, R Foundation for Statistical Computing, Vienna, Austria) and Python (version 3.10, scikit-learn package). Categorical variables were summarized as frequencies and percentages, and compared using the chi-square test or Fisher’s exact test. Continuous variables were expressed as mean ± standard deviation for normally distributed data or median with interquartile range for non-normal data, and compared using the independent-sample t test or the Mann–Whitney U test, respectively. Both overall survival (OS) and progression-free survival (PFS) were analyzed. Kaplan–Meier survival curves were generated, and differences were compared using the log-rank test. Cox proportional hazards regression was applied to identify prognostic factors, with univariable analysis followed by multivariable modeling for significant variables. For model construction, five independent deep learning survival prediction models (DL1–DL5) were first developed as baseline models. Based on tabularized clinical and quantitative ILD features, XGBoost and Random Forest classifiers were constructed as traditional machine learning models. To further improve predictive performance and robustness, ensemble strategies were explored, including simple averaging, logistic regression-based fusion, and weighted fusion. All models were trained and evaluated with a 7:3 split into training and validation cohorts. Model performance was assessed by the area under the ROC curve (AUC), accuracy (ACC), sensitivity, specificity, and confusion matrix. Comparisons of AUC between models were conducted using the DeLong test. All statistical tests were two-sided, and p < 0.05 was considered statistically significant. Results Cohort Characteristics A total of 324 patients with NSCLC combined with ILD were included, comprising 226 in the training cohort and 98 in the validation cohort. The median age of the two cohorts was 67.0 years (IQR: 62.8–72.0 years), and approximately 43.8% were male. Smoking history, BMI, and major comorbidities (such as hypertension and diabetes) were similarly distributed between the cohorts (all p > 0.05). Regarding imaging features, tumor volume, total lung volume, and the distribution of ILD subtypes showed no significant differences. Visual scoring revealed comparable distributions of fibrosis, emphysema, and honeycombing between the training and validation cohorts. In terms of treatment, the proportions of patients receiving surgery, chemotherapy, and radiotherapy were similar between cohorts. Overall, no significant differences were observed in most clinical and imaging characteristics (all p > 0.05), except that total lung volume was significantly different (p = 0.02).Comprehensive baseline clinical and imaging characteristics are presented in Table 1 and Supplementary Table S1 . Table 1 Baseline characteristics of patients in the training and validation cohorts Characteristic Full Cohort (n = 324) Training (n = 226) Validation (n = 98) P value Age (y) 67.0 (62.8–72.0) 68.0 (63.0–72.0) 67.0 (62.0–72.0) 0.850 Sex 0.070 Female 142 (43.8%) 91 (40.3%) 51 (52.0%) Male 182 (56.2%) 135 (59.7%) 47 (48.0%) Smoking history 0.180 Never 167 (51.5%) 109 (48.2%) 58 (59.2%) Former 47 (14.5%) 34 (15.0%) 13 (13.3%) Current 110 (34.0%) 83 (36.7%) 27 (27.6%) BMI (kg/m²) 23.4 (21.3–25.9) 23.4 (21.1–25.6) 23.5 (21.6–26.3) 0.550 Hypertension 110 (34.0%) / 214 (66.0%) 70 (31.0%) / 156 (69.0%) 40 (40.8%) / 58 (59.2%) 0.100 Diabetes 49 (15.1%) / 275 (84.9%) 34 (15.0%) / 192 (85.0%) 15 (15.3%) / 83 (84.7%) 1.000 Emphysema 37 (11.4%) / 287 (88.6%) 29 (12.8%) / 197 (87.2%) 8 (8.2%) / 90 (91.8%) 0.260 Doctor 1 visual score 0.180 None 8 (2.5%) 7 (3.1%) 1 (1.0%) Mild 118 (36.4%) 79 (35.0%) 39 (39.8%) Moderate 121 (37.3%) 80 (35.4%) 41 (41.8%) Severe 77 (23.8%) 60 (26.5%) 17 (17.3%) Doctor 2 visual score 0.240 None 17 (5.2%) 15 (6.6%) 2 (2.0%) Mild 103 (31.8%) 68 (30.1%) 35 (35.7%) Moderate 138 (42.6%) 94 (41.6%) 44 (44.9%) Severe 66 (20.4%) 49 (21.7%) 17 (17.3%) Lung volume, mm³ 358.8.1(273.8–475.8) 379.0(282.4–487.6) 327.8 (236.5–436.1) 0.020 Surgery 232 (71.6%) / 92 (28.4%) 155 (68.6%) / 71 (31.4%) 77 (78.6%) / 21 (21.4%) 0.080 Note. SD, standard deviation; IQR, interquartile range; BMI, body mass index; ILD, interstitial lung disease. p-values were calculated using chi-square or Fisher’s exact test for categorical variables, and t-test or Mann–Whitney U test for continuous variables. Survival Analysis Univariable Cox Analysis Univariable Cox analysis identified multiple clinical and imaging features significantly associated with OS and PFS (Table S2). Among ILD features, honeycombing, fibrosis ratio and volume were significant predictors (Table S3). Collinearity Analysis Collinearity analysis showed strong correlations between several volume and ratio measures(Fig. 2 ). As expected, each pair of volume and ratio variables was highly correlated (e.g., GGO volume vs. GGO ratio, r = 0.89; tumor volume vs. tumor ratio, r = 0.82; honeycomb volume vs. honeycomb ratio, r = 0.97; emphysema volume vs. emphysema ratio, r = 0.93). Particularly, honeycomb volume and fibrosis total volume demonstrated extremely high correlations (r = 0.98), indicating that these two variables cannot be included simultaneously in multivariable Cox regression models. VIF(Table 2 ) analysis confirmed this observation, with elevated values for honeycomb and fibrosis variables, further supporting their mutual collinearity. Based on these findings, we retained honeycombing and fibrosis in separate models, while volume–ratio pairs were represented by only one variable in adjusted analyses. Table 2 Variance Inflation Factor (VIF) for Core Variables Variable VIF ggo_volume 4.93 ggo_ratio 4.93 tumor_volume 3.20 tumor_ratio 3.16 honeycomb_volume 206.95 honeycomb_ratio 17.40 fibrosis_total_volume 200.57 fibrosis_ratio 5.91 emphysema_volume 7.89 emphysema_ratio 7.86 Note. Variance inflation factor (VIF) analysis for core ILD- and tumor-related variables. VIF values greater than 5 indicate moderate collinearity, while values greater than 10 indicate severe collinearity. Multivariable Cox Analysis Honeycomb volume and total fibrosis volume were both significant predictors in univariable analyses. In multivariable Cox regression, each retained independent prognostic value when modeled separately; however, due to strong collinearity, neither remained significant when included simultaneously in the same model(Table 3 – 4 ). Table 3 Multivariable Cox regression analysis for honeycombing model Variable OS PFS HR(95% CI) P HR (95% CI) P age 1.01 (0.98–1.04) 0.44 1.00 (0.98–1.03) 0.93 sex 0.66 (0.40–1.09) 0.10 0.83 (0.53–1.31) 0.43 bmi 0.96 (0.91-1.00) 0.07 0.96 (0.92–1.01) 0.12 tumor_volume 1.02 (1.01–1.04 0.004 1.02(1.01–1.03) 0.01 tx_surgery 0.20 (0.12–0.34) < 0.001 0.23 (0.14–0.35) < 0.001 ki67 1.01 (1.00-1.02) 0.11 1.01 (1.00-1.02) 0.01 Adenocarcinoma 1.61 (0.96–2.70) 0.07 0.92 (0.58–1.47) 0.74 Squamous cell carcinoma 2.88 (0.78–10.63) 0.11 1.36 (0.39–4.81) 0.63 Large cell carcinoma 1.92 (0.80–4.57) 0.14 1.10 (0.48–2.52) 0.82 honeycomb_volume 1.06 (1.02–1.09) 0.001 1.05(1.01–1.08) 0.01 smoking_index 1.01 (1.00-1.02) 0.13 1.01 (1.00-1.01) 0.20 Note. HRs for volume variables are expressed per 10 cm³ increase. Table 4 Multivariable Cox regression analysis for fibrosis model Variable OS PFS HR(95% CI) P HR (95% CI) P age 1.01 (0.98–1.04) 0.44 1.00 (0.98–1.03) 0.94 sex 0.66 (0.40–1.09) 0.10 0.84 (0.53–1.31) 0.44 bmi 0.95 (0.91-1.00) 0.06 0.96 (0.92–1.01) 0.11 tumor_volume 1.02 (1.01–1.04) 0.004 1.02 (1.01–1.03) 0.01 tx_surgery 0.20 (0.12–0.34) < 0.001 0.23 (0.14–0.36) < 0.001 ki67 1.01 (1.00-1.02) 0.12 1.01 (1.00-1.02) 0.01 Adenocarcinoma 1.61 (0.96–2.69) 0.07 0.92 (0.58–1.47) 0.74 Squamous cell carcinoma 2.93 (0.79–10.82) 0.11 1.38 (0.39–4.87) 0.62 Large cell carcinoma 1.94 (0.81–4.62) 0.14 1.11 (0.48–2.54) 0.81 fibrosis_total_volume 1.06 (1.02–1.09) < 0.001 1.04 (1.01–1.08) 0.01 smoking_index 1.01 (1.00-1.02) 0.13 1.01 (1.00-1.01) 0.19 Note. HRs for volume variables are expressed per 10 cm³ increase. Kaplan–Meier Survival Analysis Kaplan–Meier survival analysis demonstrated that patients with higher tumor volume had significantly worse OS and PFS (log-rank p < 0.001 for both)(Figure S6).Honeycomb volume also showed significant stratification ability (OS: p = 0.020; PFS: p = 0.016), although confidence intervals overlapped partially(Fig. 3 ). In contrast, total fibrosis volume did not show significant separation in KM curves and was therefore presented in the supplementary materials. These findings highlight that tumor burden and honeycombing provide stronger unadjusted prognostic discrimination, while the prognostic effect of fibrosis emerges mainly in multivariable Cox regression due to collinearity.Kaplan–Meier curves for fibrosis total volume did not show significant separation and are therefore provided in the Supplementary Figure S1 . Although not significant in KM analysis, fibrosis remained an independent predictor in Cox regression after adjustment, likely due to confounding and collinearity effects. deep-learning survival models We developed five deep-learning survival models to predict overall survival (OS) in NSCLC patients with co-existing ILD. Their performance in the validation set is summarized in Table 5 : Model 3 (quantitative imaging + visual scores + clinical data) achieved the highest accuracy (0.898) and AUC (0.991).Model 4 (quantitative + clinical) and Model 5 (quantitative + visual + clinical, excluding tumor-related features) were similarly robust, with AUCs of 0.985 and 0.991, respectively.Model 1 (quantitative + visual) and Model 2 (quantitative ILD features only) yielded lower AUCs of 0.734 each.The ROC curves (Figure S7) show clear separation between models that incorporate clinical and visual information and those that do not. Notably, adding radiologists’ visual scores consistently improved model performance. Training curves for Model 3 (Fig. 4 ) demonstrate stable convergence, steadily increasing accuracy, decreasing loss, and a confusion matrix confirming high sensitivity and specificity. Table 5 Performance comparison of five DL-models Model Feature set Accuracy AUC Sensitivity Specificity 1 Quantitative + Visual scores 0.7551 0.734 0.945 0.20 2 Quantitative ILD features 0.7347 0.734 0.918 0.20 3 Quantitative + Visual + Clinical 0.8980 0.9907 1.000 0.60 4 Quantitative + Clinical 0.9286 0.9852 0.986 0.76 5 Quantitative + Visual + Clinical (excluding tumor-related) 0.9286 0.9907 0.986 0.76 Note. Accuracy and AUC were calculated on the validation set. Sensitivity and specificity were derived from the confusion matrices. Models 3–5 outperform Models 1–2. Adding physician visual scores increased AUC in both quantitative-only and quantitative + clinical comparisons. External validation is warranted. Fusion Models Given the exceptionally high AUCs (0.985–0.991) of individual deep-learning models—indicating a clear risk of overfitting—we expanded the benchmark to include Random Forest, XGBoost, and several other algorithms. Mean-probability and logistic-blending ensembles were then applied to recalibrate predictions. The resulting fusion models achieved a peak AUC of 0.994 and an AP of 0.964–0.985 on the validation set, substantially outperforming any single model while maintaining robust generalizability (Fig. 5 ). The predictive performance of the different models is summarized in Table S4. SHAP In the feature importance analysis (Fig. 6 ), tumor volume consistently ranked as the most influential predictor in both Random Forest and XGBoost models, followed by key clinical characteristics such as age, sex, BMI, and Ki67. Importantly, quantitative ILD-related features, including GGO volume, honeycombing volume, fibrosis total volume, and emphysema volume, were also among the top contributors. These findings indicate that, while tumor and clinical factors remain dominant predictors, ILD-associated quantitative imaging features provide additional prognostic information and may enhance risk stratification. Discussion In this study, we demonstrated that quantitative CT-derived features of interstitial lung disease (ILD), particularly honeycombing volume and total fibrosis volume, were independently associated with overall survival in patients with non–small cell lung cancer (NSCLC)[ 16 , 17 ]. Importantly, these associations remained significant after adjustment for established clinical covariates, indicating that background parenchymal abnormalities provide prognostic information beyond tumor-related factors alone. Patients with NSCLC and coexisting ILD represent a clinically distinct and vulnerable subgroup. Structural lung alterations such as honeycombing and fibrosis reflect irreversible parenchymal damage, reduced pulmonary reserve, and chronic inflammatory remodeling. These changes may adversely affect tolerance to surgery, chemotherapy, radiotherapy, and immunotherapy, thereby contributing to poorer long-term outcomes[ 19 , 20 ]. Our findings support the hypothesis that the extent of ILD burden, rather than its mere presence, plays a critical role in determining survival in this population[ 18 ]. Previous studies have mainly relied on visual assessment to evaluate ILD severity[ 28 ], which is inherently subjective and prone to interobserver variability. By contrast, quantitative CT analysis provides an objective and reproducible approach to characterize ILD extent. In the present study, quantitative ILD features complemented radiologists’ visual assessment, and models integrating both automated measurements and expert interpretation achieved improved prognostic stratification[ 30 , 31 ]. These results suggest that quantitative imaging biomarkers should be considered supportive tools that enhance, rather than replace, clinical judgment. Another notable finding of this study is the differential prognostic impact of specific ILD components. While both honeycombing and fibrosis were independently associated with survival in multivariable analysis, honeycombing showed stronger unadjusted discrimination in survival curves.[ 28 , 29 ] This observation is biologically plausible, as honeycombing represents end-stage fibrotic remodeling and irreversible architectural distortion, whereas fibrosis encompasses a broader spectrum of pathological changes. The strong collinearity between these features further underscores the importance of careful variable selection when modeling ILD-related prognostic factors. Several limitations of this study should be acknowledged. First, this was a retrospective single-center study, which may limit generalizability. Second, although automated CT segmentation was followed by manual quality control, segmentation inaccuracies—particularly for diffuse ground-glass opacities—cannot be fully excluded[ 21 , 22 ]. Third, external validation in independent cohorts was not performed and will be essential to confirm the robustness of our findings[ 22 , 25 ]. Finally, although overall survival was the primary endpoint, additional cancer-specific outcomes and treatment-related toxicity were not systematically analyzed. Despite these limitations, this study provides evidence that quantitative CT assessment of ILD offers clinically relevant prognostic information in NSCLC. By objectively characterizing the burden and pattern of background lung disease, quantitative imaging biomarkers may facilitate more accurate risk stratification and support individualized clinical decision-making. Future multicenter studies with external validation are warranted to further establish the role of ILD quantification in the management of lung cancer patients. Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work, the authors used ChatGPT (OpenAI) to assist with English language polishing and grammar correction. After using this tool, the authors carefully reviewed and edited the content as needed, and take full responsibility for the content of the published article. Declarations Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work, the authors used ChatGPT (OpenAI) to assist with English language polishing and grammar correction. After using this tool, the authors carefully reviewed and edited the content as needed, and take full responsibility for the content of the published article. Ethics approval and consent to participate This retrospective cohort study was approved by the Institutional Ethics Committee of the Fourth Affiliated Hospital of Harbin Medical University (Approval No. 2025-125). Written informed consent was waived due to the retrospective nature of the study. Consent for publication Not applicable. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Competing interests The authors declare that they have no competing interests. Funding This study was supported by the National Natural Science Foundation of China (Grant No. 62376078). Authors ’ contributions T.Z. conceived and supervised the study and acquired funding. C.Q. and T.Z. conceived and designed the study. C.Q. and K.Y. performed the formal analysis, investigation, and software development, and drafted the original manuscript. Y.L., C.W., J.X., L.L., J.C., Y.S., and A.L. contributed to data acquisition, resources, and critical revision of the manuscript. All authors read and approved the final manuscript. Acknowledgements Not applicable. References Goldstraw P, Chansky K, Crowley J, et al. The IASLC Lung Cancer Staging Project: proposals for revision of the TNM stage groupings in the forthcoming (eighth) edition. J Thorac Oncol. 2016;11(1):39–51. 10.1016/j.jtho.2015.09.009 . Rami-Porta R, Bolejack V, Giroux DJ, et al. The IASLC lung cancer staging project: the new database to inform the eighth edition of the TNM classification of lung cancer. J Thorac Oncol. 2014;9(11):1618–24. 10.1097/JTO.0000000000000334 . Chansky K, Detterbeck FC, Nicholson AG, et al. The IASLC Lung Cancer Staging Project: external validation of the proposals for revision of the TNM classification. J Thorac Oncol. 2017;12(7):110–23. Ettinger DS, Wood DE, Aggarwal C, et al. NCCN Guidelines Insights: Non–Small Cell Lung Cancer, Version 2.2024. J Natl Compr Canc Netw. 2024;22(5):329–79. 10.6004/jnccn.2024.0029 . Detterbeck FC, Boffa DJ, Kim AW, et al. Follow-up and surveillance of the patient with lung cancer after curative-intent therapy: ACCP guidelines. Chest. 2013;143(5 Suppl):eS437–54. Raghu G, Rochwerg B, Zhang Y, et al. Idiopathic pulmonary fibrosis (an update) and progressive pulmonary fibrosis in adults: an official ATS/ERS/JRS/ALAT clinical practice guideline. Am J Respir Crit Care Med. 2022;206(12):e1–57. 10.1164/rccm.202202-0399ST . Raghu G, Collard HR, Egan JJ, et al. An official ATS/ERS/JRS/ALAT statement: idiopathic pulmonary fibrosis: evidence-based guidelines for diagnosis and management. Am J Respir Crit Care Med. 2011;183(6):788–824. Lynch DA, Sverzellati N, Travis WD, et al. Diagnostic criteria for idiopathic pulmonary fibrosis: a Fleischner Society White Paper. Lancet Respir Med. 2018;6(2):138–53. Walsh SLF, Wells AU, Desai SR, et al. Interobserver agreement for the ATS/ERS/JRS/ALAT criteria for UIP pattern on CT. Thorax. 2016;71(1):45–51. King TE Jr, Pardo A, Selman M. Idiopathic pulmonary fibrosis. Lancet. 2011;378(9807):1949–61. Richeldi L, du Bois RM, Raghu G, et al. Efficacy and safety of nintedanib in idiopathic pulmonary fibrosis. N Engl J Med. 2014;370:2071–82. King TE Jr, Bradford WZ, Castro-Bernardini S, et al. A phase 3 trial of pirfenidone in patients with idiopathic pulmonary fibrosis. N Engl J Med. 2014;370:2083–92. Collard HR, Moore BB, Flaherty KR, et al. Acute exacerbations of idiopathic pulmonary fibrosis. Am J Respir Crit Care Med. 2007;176(7):636–43. Jacob J, Bartholmai BJ, Rajagopalan S, et al. Quantitative CT and outcome in idiopathic pulmonary fibrosis. Thorax. 2017;72(2):167–74. Humphries SM, Yagihashi K, Huckleberry J, et al. Idiopathic pulmonary fibrosis: CT-based texture analysis of extent of fibrosis. Radiology. 2020;296(1):209–18. Watadani T, Sakai F, Johkoh T, et al. Interobserver variability in the CT assessment of honeycombing in the lungs. Radiology. 2013;266:936–44. Goldin JG, Lynch DA, Strollo DC, et al. High-resolution CT findings in usual interstitial pneumonia: relationship to survival. Am J Respir Crit Care Med. 2008;177:433–9. Hansell DM, Bankier AA, MacMahon H, et al. Fleischner Society: glossary of terms for thoracic imaging. Radiology. 2008;246:697–722. Ardila D, Kiraly AP, Bharadwaj S, et al. End-to-end lung cancer screening with 3D deep learning on low-dose CT. Nat Med. 2019;25(6):954–61. Chen C, Qin C, Qiu H, et al. Deep learning for lung disease segmentation: a review. Med Image Anal. 2021;71:102024. Dwivedi K, Sharkey M, Delaney L, et al. Improving Prognostication in Pulmonary Hypertension Using AI-quantified Fibrosis and Radiologic Severity Scoring. Radiology. 2024;310(1):e231718. 10.1148/radiol.231718 . Uno H, Cai T, Pencina MJ, et al. On the evaluation of prognostic survival models. Stat Med. 2011;30(10):1105–27. 10.1002/sim.4154 . Harrell FE. Jr. Regression Modeling Strategies. 2nd ed. Springer; 2015. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas. 1960;20(1):37–46. Steyerberg EW. Clinical Prediction Models. 2nd ed. Springer; 2019. Travis WD, Brambilla E, Nicholson AG, et al. WHO Classification of Tumours: Thoracic Tumours. 5th ed. Lyon: IARC; 2021. Martinez FJ, Collard HR, Pardo A, et al. Idiopathic pulmonary fibrosis. Nat Rev Dis Primers. 2017;3:17074. Jo HE, Glaspole I, Grainge C, et al. Baseline characteristics of idiopathic pulmonary fibrosis: analysis from the Australian IPF registry. Eur Respir J. 2017;49:1601592. Kim HJ, Brown MS, Chong D, et al. Comparison of visual and quantitative assessment of interstitial lung disease extent. Eur Radiol. 2019;29(8):4374–82. Maldonado F, Moua T, Rajagopalan S, et al. Automated quantification of pulmonary fibrosis in HRCT. Am J Respir Crit Care Med. 2014;189(8):948–60. Sverzellati N, Lynch DA, Hansell DM, et al. American Thoracic Society–European Respiratory Society classification of the idiopathic interstitial pneumonias: advances since 2002. Radiology. 2015;277(3):e141–60. Flaherty KR, Wells AU, Cottin V, et al. Nintedanib in progressive fibrosing interstitial lung diseases. N Engl J Med. 2019;381:1718–27. World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 2013;310(20):2191–4. Additional Declarations No competing interests reported. Supplementary Files Supplementary.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 05 Feb, 2026 Editor invited by journal 08 Jan, 2026 Editor assigned by journal 07 Jan, 2026 Submission checks completed at journal 07 Jan, 2026 First submitted to journal 03 Jan, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8506344","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":586572080,"identity":"ea7f995a-5c15-4a91-a324-91d46fc81367","order_by":0,"name":"Chunhui Qin","email":"","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Chunhui","middleName":"","lastName":"Qin","suffix":""},{"id":586572081,"identity":"24b406b9-ba44-4407-b208-f941f7c8ca83","order_by":1,"name":"Kang Yu","email":"","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Kang","middleName":"","lastName":"Yu","suffix":""},{"id":586572082,"identity":"e38bb5a2-5930-4e59-8fb8-e3933b88b487","order_by":2,"name":"Yuxin Li","email":"","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yuxin","middleName":"","lastName":"Li","suffix":""},{"id":586572083,"identity":"3aa17b48-40dd-4a4a-be0f-2fe0be851e21","order_by":3,"name":"Chunxiao Wang","email":"","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Chunxiao","middleName":"","lastName":"Wang","suffix":""},{"id":586572084,"identity":"9a0e4a65-6a44-4785-b6df-91ddfb04cec3","order_by":4,"name":"Jiaheng Xu","email":"","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jiaheng","middleName":"","lastName":"Xu","suffix":""},{"id":586572085,"identity":"65fb7847-6a0f-4305-a181-6f4664320956","order_by":5,"name":"Ling Liu","email":"","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ling","middleName":"","lastName":"Liu","suffix":""},{"id":586572087,"identity":"455e1736-7189-4511-997b-e939ba3a8112","order_by":6,"name":"Jiahui Chen","email":"","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jiahui","middleName":"","lastName":"Chen","suffix":""},{"id":586572089,"identity":"71df334d-b297-41c9-a256-b18ee468def0","order_by":7,"name":"Yunjia Shuai","email":"","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yunjia","middleName":"","lastName":"Shuai","suffix":""},{"id":586572090,"identity":"afbfbd14-78b0-416f-a0dc-a41cf19a1610","order_by":8,"name":"Ao Li","email":"","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ao","middleName":"","lastName":"Li","suffix":""},{"id":586572091,"identity":"e8627278-d958-48f5-a1fa-612d90937b13","order_by":9,"name":"Tong Zhang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6klEQVRIiWNgGAWjYDACZhBhAMTs7QcffDCwsSNBC8+ZZMMZBWnJJFgnkWAmzfPhEGMDIYUGx3kPv+YpsMuTj0hIk7YxOMDMwH746Aa8Wg7zpVnOMEguNjzz8LB1jsEdPgaetLQb+LXwmBl8MGBO3NiekHg7x+AZM4MEjxlhLQkG9YkbGxIMpC0MDjM2EKHFGBi2hxPncyQYSTMQo0USaAvjDIPjiRtAgdxjkJbMRsgvfOfPGH/m+VOdOL8dGJU//tjY8bMfPoZXi8IBBjYJsAsPQEXY8CkHAfkGBuYPUMYoGAWjYBSMAuwAAI6DTimOLVklAAAAAElFTkSuQmCC","orcid":"","institution":"The Fourth Affiliated Hospital of Harbin Medical University","correspondingAuthor":true,"prefix":"","firstName":"Tong","middleName":"","lastName":"Zhang","suffix":""}],"badges":[],"createdAt":"2026-01-03 10:53:39","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8506344/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8506344/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102376241,"identity":"525eb2b0-9dd6-461d-8a94-7bd9442998ae","added_by":"auto","created_at":"2026-02-11 05:26:45","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":70083,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart showing the selection of patients for the study cohort. Exclusion criteria include prior lung surgery, radiotherapy, chemotherapy, targeted therapy, or immunotherapy, as well as poor image quality and other conditions.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8506344/v1/eeeb9b69054df8ea37083e14.jpeg"},{"id":102376227,"identity":"4acd02e0-c6c9-4257-8ae2-edb564c43106","added_by":"auto","created_at":"2026-02-11 05:26:35","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":346552,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelation heatmap of core variables (volumes and ratios). Strong correlations were observed within volume–ratio pairs and between honeycombing and fibrosis.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8506344/v1/de8ee6dbd66269496bcfddd4.png"},{"id":102376171,"identity":"edb0ce5e-3b1a-4e58-bf48-722cc988fd4d","added_by":"auto","created_at":"2026-02-11 05:26:29","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":75308,"visible":true,"origin":"","legend":"\u003cp\u003eKaplan–Meier survival curves for Honeycomb Volume. Patients were stratified into high and low groups based on median values. Both overall survival (OS) and progression-free survival (PFS) are shown. Shaded areas represent 95% confidence intervals.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8506344/v1/6793b34444ea72ee6f6ed401.png"},{"id":102376237,"identity":"54727c3a-0c66-4906-9e29-366807ae8164","added_by":"auto","created_at":"2026-02-11 05:26:43","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":81091,"visible":true,"origin":"","legend":"\u003cp\u003ePanel A shows the accuracy curve across epochs; Panel B shows loss curves (total/classification/regularization); Panel C shows the confusion matrix with counts and derived sensitivity/specificity.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8506344/v1/6c413c44511b31404c778231.png"},{"id":102376298,"identity":"963fead1-9e5e-4ac6-be3f-c8b41771f5b0","added_by":"auto","created_at":"2026-02-11 05:26:56","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1229321,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e(A)\u003c/strong\u003eROC analysis showed that individual deep learning models achieved very high AUCs (0.985–0.991), while Random Forest and XGBoost yielded moderate performance (AUC=0.824 and 0.735, respectively). Ensemble models (mean and logistic blend) demonstrated superior and stable performance with AUCs up to 0.994.\u003cstrong\u003e(B)\u003c/strong\u003ePR curve analysis indicated moderate performance of Random Forest (AP=0.565) and XGBoost (AP=0.447). Deep learning models achieved very high AP values (\u0026gt;0.950), and ensemble models (mean and logistic blend) further demonstrated robust performance (AP=0.964–0.985).\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8506344/v1/a21817b12e8d6851b9f8010c.jpeg"},{"id":102376301,"identity":"67c0fb05-2c6d-4a40-95ba-e79a94e01ce2","added_by":"auto","created_at":"2026-02-11 05:26:59","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":252913,"visible":true,"origin":"","legend":"\u003cp\u003eThe top 10 features identified by Random Forest (left) and XGBoost (right) are shown. Tumor volume and major clinical factors (e.g., age, sex, BMI, Ki67) demonstrated strong predictive power. Notably, quantitative ILD-related features (GGO volume, Honeycombing volume, Fibrosis total volume, Emphysema volume) also ranked highly, suggesting their potential prognostic value.\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8506344/v1/3d7e4dcb60f3bdab3f73fa32.png"},{"id":102376330,"identity":"8492e986-5d79-4b9d-87e7-659b1518115a","added_by":"auto","created_at":"2026-02-11 05:27:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2957280,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8506344/v1/4b439de8-376d-4b9e-a895-9e458d5638fb.pdf"},{"id":102376168,"identity":"adccd270-80ab-4114-8895-3f8b5e0b43f7","added_by":"auto","created_at":"2026-02-11 05:26:28","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":674932,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-8506344/v1/b92f0f3d8c56e12b3ab745ca.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Quantitative CT Imaging Biomarkers of Interstitial Lung Disease Are Associated with Survival in Non–Small Cell Lung Cancer","fulltext":[{"header":"Introduction","content":"\u003cp\u003eLung cancer remains the leading cause of cancer-related mortality worldwide. Patients with comorbid interstitial lung disease (ILD) face even poorer outcomes due to reduced pulmonary reserve and increased treatment-related risks. Accurate prognostic assessment in this population is essential for guiding individualized treatment strategies. [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eCurrent approaches for ILD assessment rely primarily on visual scoring systems, such as the ATS/ERS and Fleischner criteria. However, visual evaluation is inherently subjective, with limited reproducibility and interobserver agreement. Quantitative CT-based measures, particularly those focusing on fibrosis and honeycombing, have shown promise as imaging biomarkers in ILD and lung cancer. Despite this, their role as independent prognostic factors in lung cancer has not been fully validated. [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eIn parallel, deep learning (DL) has advanced applications in medical imaging, ranging from lesion detection to segmentation and prognostic modeling. While DL models demonstrate excellent performance, concerns about overfitting and generalizability remain, particularly when applied to survival prediction. Ensemble learning may overcome these limitations by combining complementary models for robust predictions. [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eTherefore, the aim of this study was to evaluate the prognostic significance of quantitative ILD features obtained from automated segmentation, including honeycombing and total fibrosis volume, in patients with lung cancer. We further compared traditional Cox regression, individual DL models, and ensemble strategies to establish an integrative prognostic framework. [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy Sample\u003c/h2\u003e \u003cp\u003e This retrospective cohort study was approved by the Institutional Ethics Committee of the Fourth Affiliated Hospital of Harbin Medical University (Approval No. 2025\u0026thinsp;\u0026minus;\u0026thinsp;125).and conducted in accordance with the principles of the Declaration of Helsinki [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Written informed consent was waived due to the retrospective nature of the study.\u003c/p\u003e \u003cp\u003eBetween August 2020 and June 2024, a total of 5,016 patients with pulmonary lesions on chest CT were consecutively enrolled in this retrospective cohort study conducted. [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. After excluding 244 patients with small cell lung cancer and 2,142 patients with other pathological types [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], 2,630 patients with non-small cell lung cancer (NSCLC) were included. Among them, 1,917 patients without interstitial lung disease (ILD) were excluded [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], leaving 713 NSCLC patients with ILD.\u003c/p\u003e \u003cp\u003eSubsequently, 119 patients were excluded due to poor image quality, tuberculosis, pleural effusion, pneumothorax, atelectasis, prior lung surgery, radiotherapy, chemotherapy, targeted therapy, immunotherapy, or pulmonary metastases [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Furthermore, 270 patients were excluded because of missing follow-up information, follow-up time less than 12 months, or incomplete clinical/imaging data [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Finally, 324 patients were included in the analysis. Follow-up information was obtained from the hospital electronic medical record system and telephone interviews [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].The patient selection process is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eImaging Section\u003c/h3\u003e\n\u003cp\u003eCT Image Acquisition and Preprocessing\u003c/p\u003e \u003cp\u003eAll patients underwent chest CT examinations using multi-vendor and multi-model spiral CT scanners (GE, Siemens, Philips, etc.). The scanning parameters were as follows: tube voltage 120 kVp, automatic tube current modulation, slice thickness 1.25 mm, matrix size 512 \u0026times; 512, pitch 1.0, and iterative reconstruction. To minimize the impact of scanner variability, all images were standardized and preprocessed, including intensity normalization and resampling to isotropic voxels[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eILD Diagnostic Criteria\u003c/p\u003e \u003cp\u003eILD diagnosis was based on routine radiology reports interpreted by experienced radiologists, in accordance with internationally accepted HRCT diagnostic criteria (ATS/ERS/JRS/ALAT guidelines and Fleischner Society recommendations)[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Thus, ILD inclusion and classification in this study followed the standards of routine clinical practice to ensure diagnostic consistency.\u003c/p\u003e \u003cp\u003eImage Segmentation and Feature Calculation\u003c/p\u003e \u003cp\u003eImage segmentation was performed using a deep learning model based on the U-Net architecture, which had been pretrained by a collaborating team (detailed network parameters and performance metrics were not disclosed). The model automatically segmented ground-glass opacities, emphysema, honeycombing, fibrosis, tumors, and normal lung tissue. All segmentation outputs were reviewed by experienced radiologists, and cases with substantial errors were manually corrected. Detailed model parameters were not publicly available, which represents a limitation of this study.\u003c/p\u003e \u003cp\u003eFibrosis is defined as the sum of reticulation, bronchial traction dilatation, and honeycombing. Honeycombing was considered an independent subtype of fibrosis and was segmented and analyzed separately.\u003c/p\u003e \u003cp\u003eSubsequently, volumes of each ILD subtype, tumor, and the whole lung were calculated, and their relative proportions were derived to obtain quantitative ILD features.\u003c/p\u003e\n\u003ch3\u003eVisual Assessment\u003c/h3\u003e\n\u003cp\u003eTwo experienced radiologists independently performed visual assessment of CT images while blinded to patients\u0026rsquo; clinical and pathological information. The method was adopted from the Radiology pulmonary hypertension study (Dwivedi K et al., 2024)[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e] and consisted of two components:\u003c/p\u003e \u003cp\u003e1) Ordinal grading: ILD-related imaging findings (such as honeycombing, ground-glass opacity, emphysema, and fibrosis) were graded on a four-point scale (0\u0026thinsp;=\u0026thinsp;absent, 1\u0026thinsp;=\u0026thinsp;mild, 2\u0026thinsp;=\u0026thinsp;moderate, 3\u0026thinsp;=\u0026thinsp;severe).\u003c/p\u003e \u003cp\u003e2) Binary classification: Presence of each imaging feature was recorded as absent (0) or present (1).\u003c/p\u003e \u003cp\u003eInter-observer agreement was evaluated using Cohen\u0026rsquo;s Kappa coefficient. The unweighted Kappa was 0.598, indicating moderate agreement, while the weighted Kappa was 0.800, suggesting good agreement for the ordinal grading[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Analysis\u003c/h2\u003e \u003cp\u003eProgression-free survival (PFS) was defined as the time from diagnosis to either distant metastasis or death, with patients censored at the last follow-up if no event occurred.\u003c/p\u003e \u003cp\u003eAll statistical analyses were performed using R software (version 4.3.0, R Foundation for Statistical Computing, Vienna, Austria) and Python (version 3.10, scikit-learn package). Categorical variables were summarized as frequencies and percentages, and compared using the chi-square test or Fisher\u0026rsquo;s exact test. Continuous variables were expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation for normally distributed data or median with interquartile range for non-normal data, and compared using the independent-sample t test or the Mann\u0026ndash;Whitney U test, respectively. Both overall survival (OS) and progression-free survival (PFS) were analyzed. Kaplan\u0026ndash;Meier survival curves were generated, and differences were compared using the log-rank test. Cox proportional hazards regression was applied to identify prognostic factors, with univariable analysis followed by multivariable modeling for significant variables. For model construction, five independent deep learning survival prediction models (DL1\u0026ndash;DL5) were first developed as baseline models. Based on tabularized clinical and quantitative ILD features, XGBoost and Random Forest classifiers were constructed as traditional machine learning models. To further improve predictive performance and robustness, ensemble strategies were explored, including simple averaging, logistic regression-based fusion, and weighted fusion. All models were trained and evaluated with a 7:3 split into training and validation cohorts. Model performance was assessed by the area under the ROC curve (AUC), accuracy (ACC), sensitivity, specificity, and confusion matrix. Comparisons of AUC between models were conducted using the DeLong test. All statistical tests were two-sided, and p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eCohort Characteristics\u003c/h2\u003e \u003cp\u003eA total of 324 patients with NSCLC combined with ILD were included, comprising 226 in the training cohort and 98 in the validation cohort. The median age of the two cohorts was 67.0 years (IQR: 62.8\u0026ndash;72.0 years), and approximately 43.8% were male. Smoking history, BMI, and major comorbidities (such as hypertension and diabetes) were similarly distributed between the cohorts (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.05).\u003c/p\u003e \u003cp\u003eRegarding imaging features, tumor volume, total lung volume, and the distribution of ILD subtypes showed no significant differences. Visual scoring revealed comparable distributions of fibrosis, emphysema, and honeycombing between the training and validation cohorts.\u003c/p\u003e \u003cp\u003eIn terms of treatment, the proportions of patients receiving surgery, chemotherapy, and radiotherapy were similar between cohorts. Overall, no significant differences were observed in most clinical and imaging characteristics (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.05), except that total lung volume was significantly different (p\u0026thinsp;=\u0026thinsp;0.02).Comprehensive baseline clinical and imaging characteristics are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics of patients in the training and validation cohorts\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFull Cohort (n\u0026thinsp;=\u0026thinsp;324)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTraining (n\u0026thinsp;=\u0026thinsp;226)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eValidation (n\u0026thinsp;=\u0026thinsp;98)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (y)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e67.0 (62.8\u0026ndash;72.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e68.0 (63.0\u0026ndash;72.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e67.0 (62.0\u0026ndash;72.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.850\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.070\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e142 (43.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e91 (40.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e51 (52.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e182 (56.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e135 (59.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e47 (48.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmoking history\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.180\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNever\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e167 (51.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e109 (48.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e58 (59.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFormer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e47 (14.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e34 (15.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e13 (13.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCurrent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e110 (34.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e83 (36.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e27 (27.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI (kg/m\u0026sup2;)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e23.4 (21.3\u0026ndash;25.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e23.4 (21.1\u0026ndash;25.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e23.5 (21.6\u0026ndash;26.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.550\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHypertension\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e110 (34.0%) / 214 (66.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e70 (31.0%) / 156 (69.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e40 (40.8%) / 58 (59.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiabetes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e49 (15.1%) / 275 (84.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e34 (15.0%) / 192 (85.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e15 (15.3%) / 83 (84.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEmphysema\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e37 (11.4%) / 287 (88.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e29 (12.8%) / 197 (87.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e8 (8.2%) / 90 (91.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.260\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDoctor 1 visual score\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.180\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNone\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8 (2.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7 (3.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1 (1.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMild\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e118 (36.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e79 (35.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e39 (39.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e121 (37.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e80 (35.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e41 (41.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSevere\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e77 (23.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e60 (26.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e17 (17.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDoctor 2 visual score\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.240\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNone\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e17 (5.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e15 (6.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2 (2.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMild\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e103 (31.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e68 (30.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e35 (35.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e138 (42.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e94 (41.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e44 (44.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSevere\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e66 (20.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e49 (21.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e17 (17.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLung volume, mm\u0026sup3;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e358.8.1(273.8\u0026ndash;475.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e379.0(282.4\u0026ndash;487.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e327.8 (236.5\u0026ndash;436.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.020\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSurgery\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e232 (71.6%) / 92 (28.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e155 (68.6%) / 71 (31.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e77 (78.6%) / 21 (21.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.080\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eNote.\u003c/b\u003eSD, standard deviation; IQR, interquartile range; BMI, body mass index; ILD, interstitial lung disease. p-values were calculated using chi-square or Fisher\u0026rsquo;s exact test for categorical variables, and t-test or Mann\u0026ndash;Whitney U test for continuous variables.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eSurvival Analysis\u003c/h3\u003e\n\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eUnivariable Cox Analysis\u003c/h2\u003e \u003cp\u003eUnivariable Cox analysis identified multiple clinical and imaging features significantly associated with OS and PFS (Table S2). Among ILD features, honeycombing, fibrosis ratio and volume were significant predictors (Table S3).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eCollinearity Analysis\u003c/h2\u003e \u003cp\u003eCollinearity analysis showed strong correlations between several volume and ratio measures(Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). As expected, each pair of volume and ratio variables was highly correlated (e.g., GGO volume vs. GGO ratio, r\u0026thinsp;=\u0026thinsp;0.89; tumor volume vs. tumor ratio, r\u0026thinsp;=\u0026thinsp;0.82; honeycomb volume vs. honeycomb ratio, r\u0026thinsp;=\u0026thinsp;0.97; emphysema volume vs. emphysema ratio, r\u0026thinsp;=\u0026thinsp;0.93). Particularly, honeycomb volume and fibrosis total volume demonstrated extremely high correlations (r\u0026thinsp;=\u0026thinsp;0.98), indicating that these two variables cannot be included simultaneously in multivariable Cox regression models. VIF(Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) analysis confirmed this observation, with elevated values for honeycomb and fibrosis variables, further supporting their mutual collinearity. Based on these findings, we retained honeycombing and fibrosis in separate models, while volume\u0026ndash;ratio pairs were represented by only one variable in adjusted analyses.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eVariance Inflation Factor (VIF) for Core Variables\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eVIF\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eggo_volume\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4.93\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eggo_ratio\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4.93\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003etumor_volume\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.20\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003etumor_ratio\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.16\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehoneycomb_volume\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e206.95\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehoneycomb_ratio\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e17.40\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003efibrosis_total_volume\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e200.57\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003efibrosis_ratio\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5.91\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eemphysema_volume\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7.89\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eemphysema_ratio\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7.86\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eNote.\u003c/b\u003eVariance inflation factor (VIF) analysis for core ILD- and tumor-related variables. VIF values greater than 5 indicate moderate collinearity, while values greater than 10 indicate severe collinearity.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eMultivariable Cox Analysis\u003c/h2\u003e \u003cp\u003eHoneycomb volume and total fibrosis volume were both significant predictors in univariable analyses. In multivariable Cox regression, each retained independent prognostic value when modeled separately; however, due to strong collinearity, neither remained significant when included simultaneously in the same model(Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMultivariable Cox regression analysis for honeycombing model\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eOS\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003ePFS\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHR(95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eP\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHR (95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eP\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eage\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.01 (0.98\u0026ndash;1.04)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.00 (0.98\u0026ndash;1.03)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.93\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003esex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.66 (0.40\u0026ndash;1.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.83 (0.53\u0026ndash;1.31)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.43\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ebmi\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.96 (0.91-1.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.96 (0.92\u0026ndash;1.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.12\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003etumor_volume\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.02 (1.01\u0026ndash;1.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.004\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.02(1.01\u0026ndash;1.03)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003etx_surgery\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.20 (0.12\u0026ndash;0.34)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.23 (0.14\u0026ndash;0.35)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eki67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.01 (1.00-1.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.01 (1.00-1.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAdenocarcinoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.61 (0.96\u0026ndash;2.70)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.92 (0.58\u0026ndash;1.47)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.74\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSquamous cell carcinoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2.88 (0.78\u0026ndash;10.63)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.36 (0.39\u0026ndash;4.81)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.63\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLarge cell carcinoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.92 (0.80\u0026ndash;4.57)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.10 (0.48\u0026ndash;2.52)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehoneycomb_volume\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.06 (1.02\u0026ndash;1.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.05(1.01\u0026ndash;1.08)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003esmoking_index\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.01 (1.00-1.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.01 (1.00-1.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.20\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eNote.\u003c/b\u003eHRs for volume variables are expressed per 10 cm\u0026sup3; increase.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMultivariable Cox regression analysis for fibrosis model\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eOS\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003ePFS\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHR(95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eP\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHR (95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eP\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eage\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.01 (0.98\u0026ndash;1.04)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.00 (0.98\u0026ndash;1.03)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003esex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.66 (0.40\u0026ndash;1.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.84 (0.53\u0026ndash;1.31)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.44\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ebmi\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.95 (0.91-1.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.96 (0.92\u0026ndash;1.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.11\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003etumor_volume\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.02 (1.01\u0026ndash;1.04)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.004\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.02 (1.01\u0026ndash;1.03)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003etx_surgery\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.20 (0.12\u0026ndash;0.34)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.23 (0.14\u0026ndash;0.36)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eki67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.01 (1.00-1.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.01 (1.00-1.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAdenocarcinoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.61 (0.96\u0026ndash;2.69)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.92 (0.58\u0026ndash;1.47)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.74\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSquamous cell carcinoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2.93 (0.79\u0026ndash;10.82)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.38 (0.39\u0026ndash;4.87)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.62\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLarge cell carcinoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.94 (0.81\u0026ndash;4.62)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.11 (0.48\u0026ndash;2.54)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003efibrosis_total_volume\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.06 (1.02\u0026ndash;1.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.04 (1.01\u0026ndash;1.08)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003esmoking_index\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.01 (1.00-1.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.01 (1.00-1.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.19\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eNote.\u003c/b\u003e HRs for volume variables are expressed per 10 cm\u0026sup3; increase.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eKaplan\u0026ndash;Meier Survival Analysis\u003c/h2\u003e \u003cp\u003eKaplan\u0026ndash;Meier survival analysis demonstrated that patients with higher tumor volume had significantly worse OS and PFS (log-rank p\u0026thinsp;\u0026lt;\u0026thinsp;0.001 for both)(Figure S6).Honeycomb volume also showed significant stratification ability (OS: p\u0026thinsp;=\u0026thinsp;0.020; PFS: p\u0026thinsp;=\u0026thinsp;0.016), although confidence intervals overlapped partially(Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). In contrast, total fibrosis volume did not show significant separation in KM curves and was therefore presented in the supplementary materials. These findings highlight that tumor burden and honeycombing provide stronger unadjusted prognostic discrimination, while the prognostic effect of fibrosis emerges mainly in multivariable Cox regression due to collinearity.Kaplan\u0026ndash;Meier curves for fibrosis total volume did not show significant separation and are therefore provided in the Supplementary Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e. Although not significant in KM analysis, fibrosis remained an independent predictor in Cox regression after adjustment, likely due to confounding and collinearity effects.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003edeep-learning survival models\u003c/h2\u003e \u003cp\u003eWe developed five deep-learning survival models to predict overall survival (OS) in NSCLC patients with co-existing ILD. Their performance in the validation set is summarized in Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e: Model 3 (quantitative imaging\u0026thinsp;+\u0026thinsp;visual scores\u0026thinsp;+\u0026thinsp;clinical data) achieved the highest accuracy (0.898) and AUC (0.991).Model 4 (quantitative\u0026thinsp;+\u0026thinsp;clinical) and Model 5 (quantitative\u0026thinsp;+\u0026thinsp;visual\u0026thinsp;+\u0026thinsp;clinical, excluding tumor-related features) were similarly robust, with AUCs of 0.985 and 0.991, respectively.Model 1 (quantitative\u0026thinsp;+\u0026thinsp;visual) and Model 2 (quantitative ILD features only) yielded lower AUCs of 0.734 each.The ROC curves (Figure S7) show clear separation between models that incorporate clinical and visual information and those that do not. Notably, adding radiologists\u0026rsquo; visual scores consistently improved model performance. Training curves for Model 3 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) demonstrate stable convergence, steadily increasing accuracy, decreasing loss, and a confusion matrix confirming high sensitivity and specificity.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerformance comparison of five DL-models\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFeature set\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQuantitative\u0026thinsp;+\u0026thinsp;Visual scores\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.7551\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.734\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.945\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.20\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQuantitative ILD features\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.7347\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.734\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.918\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.20\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQuantitative\u0026thinsp;+\u0026thinsp;Visual\u0026thinsp;+\u0026thinsp;Clinical\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.8980\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.9907\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.60\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQuantitative\u0026thinsp;+\u0026thinsp;Clinical\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.9286\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.9852\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.986\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.76\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQuantitative\u0026thinsp;+\u0026thinsp;Visual\u0026thinsp;+\u0026thinsp;Clinical (excluding tumor-related)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.9286\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.9907\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.986\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.76\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eNote.\u003c/b\u003e Accuracy and AUC were calculated on the validation set. Sensitivity and specificity were derived from the confusion matrices. Models 3\u0026ndash;5 outperform Models 1\u0026ndash;2. Adding physician visual scores increased AUC in both quantitative-only and quantitative\u0026thinsp;+\u0026thinsp;clinical comparisons. External validation is warranted.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eFusion Models\u003c/h2\u003e \u003cp\u003eGiven the exceptionally high AUCs (0.985\u0026ndash;0.991) of individual deep-learning models\u0026mdash;indicating a clear risk of overfitting\u0026mdash;we expanded the benchmark to include Random Forest, XGBoost, and several other algorithms. Mean-probability and logistic-blending ensembles were then applied to recalibrate predictions. The resulting fusion models achieved a peak AUC of 0.994 and an AP of 0.964\u0026ndash;0.985 on the validation set, substantially outperforming any single model while maintaining robust generalizability (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). The predictive performance of the different models is summarized in Table S4.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eSHAP\u003c/h2\u003e \u003cp\u003eIn the feature importance analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e), tumor volume consistently ranked as the most influential predictor in both Random Forest and XGBoost models, followed by key clinical characteristics such as age, sex, BMI, and Ki67. Importantly, quantitative ILD-related features, including GGO volume, honeycombing volume, fibrosis total volume, and emphysema volume, were also among the top contributors. These findings indicate that, while tumor and clinical factors remain dominant predictors, ILD-associated quantitative imaging features provide additional prognostic information and may enhance risk stratification.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we demonstrated that quantitative CT-derived features of interstitial lung disease (ILD), particularly honeycombing volume and total fibrosis volume, were independently associated with overall survival in patients with non\u0026ndash;small cell lung cancer (NSCLC)[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Importantly, these associations remained significant after adjustment for established clinical covariates, indicating that background parenchymal abnormalities provide prognostic information beyond tumor-related factors alone.\u003c/p\u003e \u003cp\u003ePatients with NSCLC and coexisting ILD represent a clinically distinct and vulnerable subgroup. Structural lung alterations such as honeycombing and fibrosis reflect irreversible parenchymal damage, reduced pulmonary reserve, and chronic inflammatory remodeling. These changes may adversely affect tolerance to surgery, chemotherapy, radiotherapy, and immunotherapy, thereby contributing to poorer long-term outcomes[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Our findings support the hypothesis that the extent of ILD burden, rather than its mere presence, plays a critical role in determining survival in this population[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e \u003cp\u003ePrevious studies have mainly relied on visual assessment to evaluate ILD severity[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e], which is inherently subjective and prone to interobserver variability. By contrast, quantitative CT analysis provides an objective and reproducible approach to characterize ILD extent. In the present study, quantitative ILD features complemented radiologists\u0026rsquo; visual assessment, and models integrating both automated measurements and expert interpretation achieved improved prognostic stratification[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. These results suggest that quantitative imaging biomarkers should be considered supportive tools that enhance, rather than replace, clinical judgment.\u003c/p\u003e \u003cp\u003eAnother notable finding of this study is the differential prognostic impact of specific ILD components. While both honeycombing and fibrosis were independently associated with survival in multivariable analysis, honeycombing showed stronger unadjusted discrimination in survival curves.[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] This observation is biologically plausible, as honeycombing represents end-stage fibrotic remodeling and irreversible architectural distortion, whereas fibrosis encompasses a broader spectrum of pathological changes. The strong collinearity between these features further underscores the importance of careful variable selection when modeling ILD-related prognostic factors.\u003c/p\u003e \u003cp\u003eSeveral limitations of this study should be acknowledged. First, this was a retrospective single-center study, which may limit generalizability. Second, although automated CT segmentation was followed by manual quality control, segmentation inaccuracies\u0026mdash;particularly for diffuse ground-glass opacities\u0026mdash;cannot be fully excluded[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Third, external validation in independent cohorts was not performed and will be essential to confirm the robustness of our findings[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Finally, although overall survival was the primary endpoint, additional cancer-specific outcomes and treatment-related toxicity were not systematically analyzed.\u003c/p\u003e \u003cp\u003eDespite these limitations, this study provides evidence that quantitative CT assessment of ILD offers clinically relevant prognostic information in NSCLC. By objectively characterizing the burden and pattern of background lung disease, quantitative imaging biomarkers may facilitate more accurate risk stratification and support individualized clinical decision-making. Future multicenter studies with external validation are warranted to further establish the role of ILD quantification in the management of lung cancer patients.\u003c/p\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eDeclaration of generative AI and AI-assisted technologies in the writing process\u003c/h2\u003e \u003cp\u003eDuring the preparation of this work, the authors used ChatGPT (OpenAI) to assist with English language polishing and grammar correction. After using this tool, the authors carefully reviewed and edited the content as needed, and take full responsibility for the content of the published article.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eDeclaration of generative AI and AI-assisted technologies in the writing process\u003cbr\u003e\u0026nbsp;\u003c/strong\u003eDuring the preparation of this work, the authors used ChatGPT (OpenAI) to assist with English language polishing and grammar correction. After using this tool, the authors carefully reviewed and edited the content as needed, and take full responsibility for the content of the published article.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis retrospective cohort study was approved by the Institutional Ethics Committee of the Fourth Affiliated Hospital of Harbin Medical University (Approval No. 2025-125). Written informed consent was waived due to the retrospective nature of the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was supported by the National Natural Science Foundation of China (Grant No. 62376078).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u003c/strong\u003e\u003cstrong\u003e’ contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eT.Z. conceived and supervised the study and acquired funding.\u003c/p\u003e\n\u003cp\u003eC.Q. and T.Z. conceived and designed the study.\u003c/p\u003e\n\u003cp\u003eC.Q. and K.Y. performed the formal analysis, investigation, and software development, and drafted the original manuscript.\u003c/p\u003e\n\u003cp\u003eY.L., C.W., J.X., L.L., J.C., Y.S., and A.L. contributed to data acquisition, resources, and critical revision of the manuscript.\u003c/p\u003e\n\u003cp\u003eAll authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eGoldstraw P, Chansky K, Crowley J, et al. The IASLC Lung Cancer Staging Project: proposals for revision of the TNM stage groupings in the forthcoming (eighth) edition. J Thorac Oncol. 2016;11(1):39\u0026ndash;51. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jtho.2015.09.009\u003c/span\u003e\u003cspan address=\"10.1016/j.jtho.2015.09.009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRami-Porta R, Bolejack V, Giroux DJ, et al. The IASLC lung cancer staging project: the new database to inform the eighth edition of the TNM classification of lung cancer. J Thorac Oncol. 2014;9(11):1618\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/JTO.0000000000000334\u003c/span\u003e\u003cspan address=\"10.1097/JTO.0000000000000334\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChansky K, Detterbeck FC, Nicholson AG, et al. The IASLC Lung Cancer Staging Project: external validation of the proposals for revision of the TNM classification. J Thorac Oncol. 2017;12(7):110\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEttinger DS, Wood DE, Aggarwal C, et al. NCCN Guidelines Insights: Non\u0026ndash;Small Cell Lung Cancer, Version 2.2024. J Natl Compr Canc Netw. 2024;22(5):329\u0026ndash;79. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.6004/jnccn.2024.0029\u003c/span\u003e\u003cspan address=\"10.6004/jnccn.2024.0029\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDetterbeck FC, Boffa DJ, Kim AW, et al. Follow-up and surveillance of the patient with lung cancer after curative-intent therapy: ACCP guidelines. Chest. 2013;143(5 Suppl):eS437\u0026ndash;54.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRaghu G, Rochwerg B, Zhang Y, et al. Idiopathic pulmonary fibrosis (an update) and progressive pulmonary fibrosis in adults: an official ATS/ERS/JRS/ALAT clinical practice guideline. Am J Respir Crit Care Med. 2022;206(12):e1\u0026ndash;57. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1164/rccm.202202-0399ST\u003c/span\u003e\u003cspan address=\"10.1164/rccm.202202-0399ST\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRaghu G, Collard HR, Egan JJ, et al. An official ATS/ERS/JRS/ALAT statement: idiopathic pulmonary fibrosis: evidence-based guidelines for diagnosis and management. Am J Respir Crit Care Med. 2011;183(6):788\u0026ndash;824.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLynch DA, Sverzellati N, Travis WD, et al. Diagnostic criteria for idiopathic pulmonary fibrosis: a Fleischner Society White Paper. Lancet Respir Med. 2018;6(2):138\u0026ndash;53.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWalsh SLF, Wells AU, Desai SR, et al. Interobserver agreement for the ATS/ERS/JRS/ALAT criteria for UIP pattern on CT. Thorax. 2016;71(1):45\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKing TE Jr, Pardo A, Selman M. Idiopathic pulmonary fibrosis. Lancet. 2011;378(9807):1949\u0026ndash;61.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRicheldi L, du Bois RM, Raghu G, et al. Efficacy and safety of nintedanib in idiopathic pulmonary fibrosis. N Engl J Med. 2014;370:2071\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKing TE Jr, Bradford WZ, Castro-Bernardini S, et al. A phase 3 trial of pirfenidone in patients with idiopathic pulmonary fibrosis. N Engl J Med. 2014;370:2083\u0026ndash;92.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollard HR, Moore BB, Flaherty KR, et al. Acute exacerbations of idiopathic pulmonary fibrosis. Am J Respir Crit Care Med. 2007;176(7):636\u0026ndash;43.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJacob J, Bartholmai BJ, Rajagopalan S, et al. Quantitative CT and outcome in idiopathic pulmonary fibrosis. Thorax. 2017;72(2):167\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHumphries SM, Yagihashi K, Huckleberry J, et al. Idiopathic pulmonary fibrosis: CT-based texture analysis of extent of fibrosis. Radiology. 2020;296(1):209\u0026ndash;18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWatadani T, Sakai F, Johkoh T, et al. Interobserver variability in the CT assessment of honeycombing in the lungs. Radiology. 2013;266:936\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoldin JG, Lynch DA, Strollo DC, et al. High-resolution CT findings in usual interstitial pneumonia: relationship to survival. Am J Respir Crit Care Med. 2008;177:433\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHansell DM, Bankier AA, MacMahon H, et al. Fleischner Society: glossary of terms for thoracic imaging. Radiology. 2008;246:697\u0026ndash;722.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArdila D, Kiraly AP, Bharadwaj S, et al. End-to-end lung cancer screening with 3D deep learning on low-dose CT. Nat Med. 2019;25(6):954\u0026ndash;61.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen C, Qin C, Qiu H, et al. Deep learning for lung disease segmentation: a review. Med Image Anal. 2021;71:102024.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDwivedi K, Sharkey M, Delaney L, et al. Improving Prognostication in Pulmonary Hypertension Using AI-quantified Fibrosis and Radiologic Severity Scoring. Radiology. 2024;310(1):e231718. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1148/radiol.231718\u003c/span\u003e\u003cspan address=\"10.1148/radiol.231718\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUno H, Cai T, Pencina MJ, et al. On the evaluation of prognostic survival models. Stat Med. 2011;30(10):1105\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/sim.4154\u003c/span\u003e\u003cspan address=\"10.1002/sim.4154\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHarrell FE. Jr. Regression Modeling Strategies. 2nd ed. Springer; 2015.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas. 1960;20(1):37\u0026ndash;46.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSteyerberg EW. Clinical Prediction Models. 2nd ed. Springer; 2019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTravis WD, Brambilla E, Nicholson AG, et al. WHO Classification of Tumours: Thoracic Tumours. 5th ed. Lyon: IARC; 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartinez FJ, Collard HR, Pardo A, et al. Idiopathic pulmonary fibrosis. Nat Rev Dis Primers. 2017;3:17074.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJo HE, Glaspole I, Grainge C, et al. Baseline characteristics of idiopathic pulmonary fibrosis: analysis from the Australian IPF registry. Eur Respir J. 2017;49:1601592.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim HJ, Brown MS, Chong D, et al. Comparison of visual and quantitative assessment of interstitial lung disease extent. Eur Radiol. 2019;29(8):4374\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaldonado F, Moua T, Rajagopalan S, et al. Automated quantification of pulmonary fibrosis in HRCT. Am J Respir Crit Care Med. 2014;189(8):948\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSverzellati N, Lynch DA, Hansell DM, et al. American Thoracic Society\u0026ndash;European Respiratory Society classification of the idiopathic interstitial pneumonias: advances since 2002. Radiology. 2015;277(3):e141\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFlaherty KR, Wells AU, Cottin V, et al. Nintedanib in progressive fibrosing interstitial lung diseases. N Engl J Med. 2019;381:1718\u0026ndash;27.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 2013;310(20):2191\u0026ndash;4.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Non–small cell lung cancer, Interstitial lung disease, Computed tomography, Quantitative imaging, Prognosis","lastPublishedDoi":"10.21203/rs.3.rs-8506344/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8506344/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eInterstitial lung disease (ILD) is a clinically important but often under-recognized comorbidity in patients with non\u0026ndash;small cell lung cancer (NSCLC). Although ILD has been associated with adverse outcomes, objective and reproducible imaging biomarkers for prognostic assessment remain limited. This study aimed to evaluate the prognostic value of quantitative CT-derived ILD features in NSCLC.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eIn this retrospective cohort study, 324 patients with pathologically confirmed NSCLC and coexisting ILD were included. Automated CT-based segmentation was applied to quantify tumor volume and ILD-related components, including ground-glass opacity, emphysema, honeycombing, and fibrosis. Fibrosis was defined as the sum of reticulation, traction bronchiectasis, and honeycombing, with honeycombing analyzed as an independent subtype. Visual assessment was independently performed by two experienced radiologists. Cox proportional hazards regression was used to identify prognostic factors for overall survival (OS). Prognostic models were constructed using clinical variables, quantitative CT features, and their combinations.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eIn multivariable Cox analysis, both honeycombing volume and total fibrosis volume were independently associated with poorer OS after adjustment for clinical covariates. Prognostic models incorporating quantitative CT features demonstrated improved discrimination compared with clinical variables alone. The combined model integrating clinical variables, quantitative CT features, and radiologists\u0026rsquo; visual assessment achieved the best overall performance.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eQuantitative CT imaging biomarkers of ILD are independently associated with survival in patients with NSCLC. Integrating objective imaging-derived ILD features with clinical evaluation may improve risk stratification and support individualized management in this patient population.\u003c/p\u003e","manuscriptTitle":"Quantitative CT Imaging Biomarkers of Interstitial Lung Disease Are Associated with Survival in Non–Small Cell Lung Cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-11 05:23:09","doi":"10.21203/rs.3.rs-8506344/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2026-02-05T18:51:22+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-01-08T16:44:52+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-07T06:12:14+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-01-07T06:10:05+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Cancer","date":"2026-01-03T10:43:46+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8623a513-63b5-40a5-b902-6b0d4266d083","owner":[],"postedDate":"February 11th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-02-11T05:23:10+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-11 05:23:09","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8506344","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8506344","identity":"rs-8506344","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}