Influencing factors and prediction of growth heterogeneity in solid nodule non-small cell lung cancer based on artificial intelligence: a prospective study

Introduction

Lung cancer remains the most prevalent malignant tumor in China, accounting for the highest mortality rate among cancers (1). With the widespread adoption of low-dose computed tomography (CT) screening, the detection rate of pulmonary nodules has increased significantly. Based on density differentiation on CT, pulmonary nodules can be divided into solid nodules (SNs) and subsolid nodules (SSNs) (2). Smaller SNs generally have a lower probability of malignancy; once malignancy is confirmed, their growth rate is faster and their prognosis is worse than that of SSNs (3,4). The diagnosis of benign and malignant SN has been reported previously, but little is known about the growth pattern of malignant SN (5,6).

Volume doubling time (VDT) reflects the exponential growth of tumor cells. Traditionally, the VDT of malignant SN was believed to be less than 400 days (7,8). However, recent studies have identified an indolent growth pattern in some malignant SN, with VDT even exceeding 600 days (9). VDT is a critical parameter for assessing tumor growth and prognosis in lung cancer (10,11). Yankelevitz et al. (12,13) quantified the impact of delayed follow-up on patient prognosis during early lung cancer diagnosis and treatment by analyzing data from The International Early Lung Cancer Action Program (I-ELCAP), the National Lung Screening Trial (NLST), and the International Association for the Study of Lung Cancer (IASLC). They found that delay reduces the cure rate of lung cancer, with a decrease of approximately 1.0% in 10-year cure rates for every 1 mm increase in tumor diameter. For tumors with rapid growth (defined as VDT <60 days in the study) and larger baseline diameters, the decrease in cure rate caused by delay was more significant, whereas the impact on cure rates for moderate (60 days ≤ VDT <120 days) and slow growing (120 days ≤ VDT <240 days) tumors was relatively small. The above research results suggest that time is a prognostic factor for rapidly growing malignant SN. Given that faster-growing solid lung cancers are associated with worse prognoses, the British Thoracic Society has proposed clinical diagnosis and treatment guidelines based on the VDT of nodules (10). Further diagnostic investigation (biopsy, imaging or resection) must be offered for patients with nodules showing clear growth (use a ≥25% volume change to define significant growth) or a VDT of <400 days. For nodules with inert growth (VDT >400 days), regular follow-up is performed according to the patient’s preferences (14,15).

Therefore, identifying rapidly growing nodules through various initial examinations and promptly removing them at the earliest can significantly improve the prognosis of patients with solid nodular lung cancers. This study explores the clinical, imaging, pathological, and genetic characteristics that affect nodule growth, predicts rapidly growing nodules, and provides theoretical support for personalized management of solid nodules. We present this article in accordance with the TRIPOD reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-516/rc).

Methods

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by institutional ethics board of Cancer Hospital, Chinese Academy of Medical Sciences (No. NCC-014122). Informed consent was taken from all the patients.

Patients

We prospectively collected patients with pathologically confirmed non-small cell lung cancer (NSCLC) between February 2020 and February 2025. The inclusion criteria were as follows: (I) patients with chest thin-layer CT showing suspected malignant solid nodules (<3 cm); (II) patients who had undergone preoperative thin-layer CT follow-up for at least 60 days or had experienced significant growth (>25%) within 60 days and not received any other treatment; (III) NSCLC diagnosed through surgical pathology. The exclusion criteria were as follows: (I) patients who underwent targeted therapy, chemotherapy, or other adjuvant therapies before surgery; (II) the pathological results did not accurately match the chest CT results; (III) poor CT image quality; and (IV) lack of clinical, radiological, or pathological data (Figure 1). Finally, A total of 250 consecutive patients with 250 nodules were enrolled.

Figure 1 Flowchart of the study population. CT, computed tomography; SN, solid nodule.

Artificial intelligence segmentation and diagnosis of solid nodules

This study used the Deep Wise Multimodal Research Platform (https://keyan.deepwise.com; Beijing Deepwise & League of PHD Technology Co., Ltd., Beijing, China) to achieve intelligent segmentation and diagnosis of pulmonary solid nodules. By setting standard lung window parameters (window level −600 HU, window width 1,600 HU) for automated detection and segmentation, the research process includes the following key steps.

A total of 416 cases of lung SNs automatically identified by the system (based on 1,166 follow-up data from 416 patients) were stratified and validated by two senior radiologists (with 4 and 25 years of chest CT diagnostic experience, respectively). By adopting a layer-by-layer cross-sectional CT image verification method, automatic segmentation lines were required to accurately locate the edges of the nodules, and the top and bottom contours must meet the anatomical boundary standards. After resolving the differences through negotiation, 159 cases (13.6%) with follow-up data that did not meet the segmentation criteria were excluded, and 250 cases of solid nodules were included. The overall segmentation accuracy of the system reached 86.4% (1,007/1,166), which was slightly lower than the performance in previous studies on pure ground-glass nodules, but it still showed superior segmentation efficacy (16).

Based on the deep self-supervised fine-grained benign and malignant classification model developed in the early stage, the diagnostic performance is improved through three core technologies:

• Self-supervised pre training: extracting universal features using large-scale unlabeled data;
• Fine grained network architecture: optimizing feature extraction for problems such as smooth surface and blurry texture features of sub centimeter solid nodules;
• Attention mechanism: enhance the model’s ability to characterize small lesions (<8 mm).

In previous studies, the AUC of the model in the internal test set was 0.964 [95% confidence interval (CI): 0.942–0.986], with an accuracy of 93.4%; sensitivity, 96.5%; and, specificity, 90.8%; in the external test set, the AUC was 0.945 (95% CI: 0.910–0.979); accuracy, 91.1%; sensitivity, 97.7%; and, specificity, 86.0% (6). This study continued the previous high-risk population screening strategy and embedded the optimized model into Deep Wise system forming a full process solution of “intelligent detection~ layered verification~ risk stratification~ dynamic follow-up”. For specific technical details, please refer to Appendix 1.

CT scanning technique

All CT scans were performed using a 64 row detector CT system (GE Medical System: Optima CT 660, LightSpeed VCT, or Discovery CT 750 HD); Toshiba Aquilion Medical System. To eliminate respiratory artifacts, the subjects were instructed to fully inhale and hold their breath for image scanning. The core parameters are set to: tube voltage of 120 kVp; automatic tube current modulation (range, 200–350 mA); noise index 13; the spacing is 0.992 or 0.984; rack rotation time 0.5 s; slice thickness 1.0–1.25 mm, reconstruction interval 0.8 mm. Use standard window width and position: lung window (window width: 1,600 HU, window position: −600 HU); Mediastinal window (window width 360 HU, window position 60 HU). In contrast enhanced scanning, non-ionic contrast agent iopromide injection (iodine concentration: 300 mg/mL) was injected at a dose of 80–90 mL and a flow rate of 2.5 mL/s, and image acquisition was performed 35 seconds after injection.

Analyses of clinical and imaging characters

Patient clinical data included age, gender, smoking status (never smoked, quit smoking, smoking), symptoms, history of other malignant tumors, chronic obstructive pulmonary disease (COPD) history, other respiratory diseases and serological tumor markers including: carcinoembryonic antigen (CEA), neuron-specific enolase (NSE), cytokeratin 19 fragment antigen 21-1 (CYFRA21-1), recombinant pro-Gastrin Releasing Peptide (ProGRP), squamous cell carcinoma antigen (SCCA).

Quantitative features acquisition: the maximum, minimum diameters and average density (represented by the average CT value) of the nodules, nodule volume and VDT were automatically calculated and recorded.

The SN diameter, mean density, volume, mass and VDT were automatically calculated using the Deep Wise system. The mean diameter was defined as the average of its maximal length and maximal orthogonal diameter on transverse CT sections. The VDT were calculated for growing SNs as log₂⋅T/ log(Xf/Xi), where Xf and Xi are the final and initial volumes, respectively, and T is the interval between the final and initial CT scans (16,17). According to VDT, all SNs in this study were divided into two groups: the rapid growth group (VDT ≤200 days) and the slow growth group (VDT >200 days).

Qualitative features acquisition: (I) location: right upper, right middle, right inferior, left upper, left inferior lobe; (II) presence of deep lobulation, such as nodules with arc chord distance/chord length >0.4 on CT images indicating deep lobulation (Figure 2); (III) morphology: round/oval, irregular; (IV) boundary: clear and smooth, clear and irregular, and blurry; (V) ground glass halo sign; (VI) vacuoles, cystic cavities, and cavities: the presence of gas containing components in the lesion indicates the presence of vacuoles or cystic cavities, with those larger than 5 mm defined as cystic cavities; Void: a >5 mm cavity formed by bronchial drainage in the necrotic area of the tumor; (VII) whether there is calcification; (VIII) truncation of peripheral blood vascular or crossing through nodules, and whether there is vascular convergence (Figure 3A,3B); (IX) truncation of peripheral bronchial or crossing through nodules (Figure 3C,3D); (X) the relationship between nodules and pleura: direct contact, record the maximum diameter of the contact surface; indirect contact refers to pleural traction sign; (XI) the presence of emphysema or pulmonary fibrosis on CT.

Figure 2 Schematic diagram of leaf division sign. Find the most prominent protrusion of the nodule on the image, which is the “peak” position of the lobes, and then determine the most concave points on both sides of the protrusion. These two concave points, together with the starting point of the protrusion, form an arc (red line). Connect two concave points, and the length of this line segment is the chord length (dashed line). Draw a perpendicular line from the convex starting point to the chord length, and the length of the perpendicular line is the chord arc distance (blue line). When the ratio of chord arc distance to chord length is greater than 0.4, it is defined as the deep lobed sign.

Figure 3 The relationship between nodules and bronchial vascular bundles. (A) Vascular crossing sign: lung markings or enhanced vascular shadows can be seen through the mediastinal window, passing through the nodules. (B) Vascular truncation sign: lung markings or enhanced vascular shadows can be seen at the edge of nodules in the lung and mediastinal windows. (C) Bronchial truncation sign: bronchus can be seen truncated at the edge of the nodule on the lung window. (D) Bronchial crossing sign: bronchus can be seen passing through the nodule on the lung window, either unchanged in shape, narrowed, or dilated (the above imaging features were indicated by blue arrows in the figure).

One radiologist searched for and recorded the clinical data and operation notes of all the patients. Two radiologists identified the SNs and evaluated their morphological features on multiplanar reconstruction images of the initial and final preoperative CTs scans. Disagreements were resolved through consultations.

Pathological and genetic analysis

The pathological diagnosis and subtype categorization of SNs were confirmed based on the new World Health Organization (WHO) pulmonary tumor classification, 2021 edition. Surgical samples served as the basis for the final pathological assessments. Pathological diagnosis results were mainly obtained from the reports released by the Pathology Department of our research center. Senior pathologists specializing in chest pathology were consulted for the final diagnosis in case of uncertainty.

The following pathological features were recorded, including: pathological types, histological subtypes of lung adenocarcinoma, pathological maximum diameter, invasion of visceral pleura, vascular tumor thrombus, nerve invasion, spread through air spaces (STAS), intrapulmonary dissemination, and pathological staging.

Routine genetic testing was performed on specimens diagnosed with adenocarcinoma by pathology, including amplification refractory mutation system polymerase chain reaction (PCR) method, lung cancer second-generation sequencing panel, 520 gene second-generation sequencing large panel, covering epidermal growth factor receptor (EGFR), Kirsten rat sarcoma viral oncogene homolog (KRAS), v-raf murine sarcoma viral oncogene homolog B1 (BRAF) mutations and their mutation sites, anaplastic lymphoma kinase (ALK), recombinant c-ros oncogene 1, receptor tyrosine kinase (ROS1), rearranged during transformation (RET) gene, etc.

Statistical analysis

All statistical analyses were conducted using SPSS software (Version 27.0; IBM Corp., Armonk, NY, USA). Continuous variables underwent assessment via independent samples t-tests or Wilcoxon rank-sum tests, depending on data distribution normality. Categorical variables were compared using χ² tests or Fisher’s exact tests for small-sample categories. Measurement data are summarized as mean ± standard deviation for normally distributed variables or median (25th, 75th percentiles) for nonparametric distributions. Count data are expressed as percentages (%) across study cohorts. Statistical significance was defined at P<0.05. Variables exhibiting P<0.20 in univariate screening were incorporated into multivariable logistic regression models to identify independent risk factors. Model discriminative capacity was quantified using receiver operating characteristic (ROC) curve analysis. To mitigate confounding bias, propensity score matching (PSM) was employed, balancing feature distributions between rapid- and slow-growth cohorts. A 1:1 nearest-neighbor matching algorithm without replacement enforced a caliper width of 0.005. This approach optimized intergroup similarity and statistical power across stratified categories, with balance adequacy verified through χ² and Fisher’s exact tests.

Results

Clinical features

Among the 250 included patients, there were 103 males and 147 females (mean age: 58.3±9.9 years) with a median preoperative follow-up period of 75.5 (37.0, 273.3) days. Twenty patients were smoking and 55 patients had quitted smoking. Most patients had no symptoms (88.0%). Two patients had a history of both COPD and bronchiectasis. 46 patients had a history of malignancy. Further, 185 patients were examined for tumor markers, and the positive results were shown in Table 1. Additionally, 62 and 188 patients were examined with plain and contrast-enhanced CT at the first diagnosis, respectively. Pathological tumor-node-metastasis (TNM) staging was performed after surgery, with 233, 15, 1 and 1 case in the T1, T2, T3, and T4 stages, respectively. Furthermore, 17 (6.8%) of these nodules presented lymph node metastasis (N1 stage, n=7; N2 stage, n=10). There were three cases of distant metastases (Table 1).

CT imaging characters affecting the growth of malignant SNs

Deep Wise workstation (periodically updated software version) was used for automatic detection and segmentation of the SNs and the VDTs were calculated with it. The average VDT for all 250 malignant SNs was 518.1 days, and the median VDT was 313.5 (163.8, 664.5) days. All SNs in this study were divided into two groups based on the VDT: the rapid growth group (n=84, 33.6%) and the slow growth group (n=166, 66.4%), with a median VDT of 104.0 (70.3, 165.0) and 517.5 (313.8, 874.5) days, respectively.

Male sex, smoking history, and NSE positive patients typically showed faster growth of SNs. The rapidly growing SNs had a larger average diameter, higher CT value, larger volume, more deep-lobulation and spiculation signs, more vascular and bronchial truncations, and fewer vascular and bronchial crossings. The clinical and imaging characters at the first diagnosis before surgery were included in the univariate and multivariate logistic regression analysis. The smoking history, CT value, and the deep lobulation sign were the risk factors for the rapid growth of nodules. The area under curve (AUC) value of the ROC curve for predicting the growth rate of SNs with the multivariate regression model was approximately 0.704 (0.636, 0.771), with a sensitivity of 65.5% and specificity of 70.5% (Figure 4, Table 2).

Figure 4 ROC curves of multivariate logistic regression models for predicting the growth rate of solid nodular lung cancer. The smoking history, CT value, and the deep lobulation sign were the risk factors for the rapid growth of nodules. The AUC value of the ROC curve for predicting the growth rate of SNs with the multivariate regression model was approximately 0.704 (0.636, 0.771), with sensitivity of 65.5% and specificity of 70.5%. AUC, area under curve; CT, computed tomography; ROC, receiver operating characteristic; SN, solid nodule.

Pathological factors affecting the growth of malignant SNs

Among the 250 nodules, there were 231 cases of adenocarcinomas; 11, squamous cell carcinoma; 5, adenosquamous carcinoma; 2, pleomorphic carcinoma; and, 1, large-cell neuroendocrine carcinoma. The median VDT of squamous cell carcinoma, large cell neuroendocrine carcinoma, adenosquamous carcinoma, pleomorphic carcinoma, and adenocarcinomas, were 75.0 (44.0, 85.0), 118.0, 248.0 (151.0, 1,453.6), 319.0 (165.0, 473.0), and 343.0 (181.0, 702.0) days, respectively, with significant statistical differences (P<0.001).

Rapid growing nodules had a lower degree of differentiation and a larger pathological maximum diameter than the slow-growing nodules (P=0.004, 0.01). Univariate and multivariate logistic regression analyses indicated that the pathological histology type and degree of differentiation were risk factors for rapid growth (P=0.009, 0.006) (Table 3).

Gene mutations analysis of malignant SNs

In this study, 168 SNs were subjected to genetic testing after surgery. Among them, 75.6% harbored genetic mutations. The EGFR, KRAS, TP53, ALK, BRAF, ROS1, and RET gene mutation accounted for 43.4%, 22.6%, 3.6%, 2.4%, 2.4%, 1.8%, and 1.2%, respectively. According to the nodes with VDT =200 and 400, the VDT stratification was further refined, and the nodules were divided into rapid growth, moderate growth, and inert growth groups (n=61, 42, and 65, respectively). There were no statistically significant differences in the gene mutations between the two groups (Table 4). Statistically significant differences were observed in the occurrence of gene mutations among the different pathological types (P=0.01). Therefore, to balance the confounding factors of pathological types, the following text only explored the gene mutation status of adenocarcinomas. Gene mutations in adenocarcinomas identified in this study are shown in Table 5. There were no statistically significant differences in gene mutations between any two of the three growth groups.

Next, according to whether the pathological components contained high-grade components (micropapillary, solid, cribriform, and other complex gland components), the adenocarcinomas were divided into high-grade and non-high-grade components. Statistically significant differences were observed between the groups with and without high-grade TP53 and ALK mutations (P=0.005, 0.03). Subsequently, according to whether it was a mucinous adenocarcinoma, they were divided into 28 cases for the mucinous group and 133 cases for the non-mucinous group, respectively. The exon 21p.L858R and 19 mutations of EGFR and KRAS were significantly different between mucinous and non-mucinous adenocarcinomas (P=0.02, 0.01, <0.001).

Discussion

This study comprehensively investigated the clinical, imaging, pathological, and genetic factors influencing the growth dynamics of malignant SNs in NSCLC. By stratifying nodules based on VDT, we identified key predictors of rapid growth and provided novel insights into personalized management strategies. The findings align with and extend previous research, while also highlighting critical areas for future exploration.

Notably, in the past two years, scholars have reported the inert growth mode of some SNs. Here, we present examples of rapidly growing and indolent growth of solid nodular lung cancer (Figures 5,6). The research results of Hammer et al. and Zhang et al. showed that approximately half of the malignant SNs did not grow or showed inert growth with VDT >600 days, which is consistent with our findings (9,18). Our results have confirmed the heterogeneity of SN growth rates, averaged at 313.5 days in our cohort, with 66.4% of the nodules exhibiting indolent growth (VDT >200 days). These findings reinforce the clinical utility of VDT in risk stratification, as emphasized by the British Thoracic Society guidelines (14).

Figure 5 Examples of rapid growth in malignant SNs. On May 29, 2023, a 65-year-old male was found a nodule in the dorsal segment of the lower lobe of the right lung during a chest thin-layer CT examination at our hospital. The baseline size was approximately 17 mm × 16 mm and the volume was approximately 2,196.21 mm³. The patient underwent PET-CT examination on June 7, 2023, with increased uptake and SUVmax of 6.8. On September 13, 2023, a chest thin-layer CT examination was performed, with a final size of approximately 22 mm × 21 mm and a volume of approximately 5,608.65 mm³, and a VDT of 79 days. On October 17, 2023, a thoracoscopic right lower lobectomy was performed, and the pathology showed poorly differentiated adenocarcinoma, mainly in solid form. Genetic testing showed a mutation in exon 6 of the TP53 gene. CT, computed tomography; PET, positron emission tomography; SN, solid nodule; SUVmax, maximum standardized uptake value; TP53, tumor protein 53; VDT, volume double time.

Figure 6 Examples of inert growth in malignant SNs. On November 1, 2017, a 41-year-old male was found a nodule in the anterior segment of the left upper lobe during a chest thin-layer CT scan at our hospital. The baseline size was about 7 mm × 6 mm and the volume was about 194.33 mm³. The patient underwent chest thin-layer CT scans on November 5, 2020, July 13, 2022, and February 10, 2023. The final size was about 10 mm × 7 mm and the volume was about 297.6 mm³, with a VDT of 1,806 days. The volume growth curve showed slow growth from 2017 to 2021, with a larger slope in 2022 and relatively faster nodule growth. On March 3, 2023, an anatomical partial enlargement resection of the upper tongue segment of the left upper lobe was performed. Pathology showed poorly differentiated adenocarcinoma of the upper tongue segment of the left lung, with a sieve-like structure and mucus secretion. Genetic testing showed RET gene translocation mutation and SOX exon 1 mutation. CT, computed tomography; RET, rearranged during transformation; SN, solid nodule; SOX, SRY-related HMG-box genes; VDT, volume doubling time.

VDT in the solid part is a risk factor that affects the prognosis of patients with lung adenocarcinoma. The studies by Park, Setojima, and Nakahashi et al. (11,19,20) showed that the cut-off values for VDT with poor recurrence free survival were 400, 215, and 300 days, respectively. Therefore, preoperative assessment of its growth rate is crucial for the treatment and prognosis of SNs. Our previous studies were been relatively mature in predicting the nature of SNs using CT images; however, few studies have predicted their growth (5,6,21). We hope to use preoperative chest CT examination, which is the most important examination method, to predict the speed of its growth and assist in clinical practice. Smoking history, higher CT values, and deep lobulation were independent risk factors for rapid growth. The association between smoking and accelerated tumor progression is well documented owing to its role in promoting genomic instability and inflammation (22,23). The correlation between higher CT values (reflecting denser tumor tissue) and rapid growth may indicate increased cellular proliferation or reduced intratumoral necrosis, a hypothesis supported by the pathological observations of poor differentiation in rapidly growing SNs (P=0.004). Deep lobulation, a marker of invasive growth patterns, further underscores the importance of morphological features in predicting tumor behavior.

Poorly differentiated tumors and those with larger pathological diameters were strongly associated with rapid growth, which is consistent with the biological premise that dedifferentiation enhances proliferative capacity. The pathological histological type and degree of differentiation were risk factors for rapid growth: squamous cell carcinoma (median VDT =75.0 days) and large cell neuroendocrine carcinoma (VDT =118.0 days) demonstrated the fastest growths, aligning with their known poor prognoses (24,25). Due to the fact that 92.4% (231/250) of cases in this study were adenocarcinoma and the number of other pathological types was relatively small, there may be some differences.

NSCLC has several tumor driver genes. In this study, 75.6% of patients had gene mutation. Of which EGFR gene mutations accounted for the majority (43.4%), consistent with previous studies (26,27). Intriguingly, although TP53 and ALK mutations are enriched in adenocarcinomas with high-grade components, they do not directly correlate with VDT stratification (28,29). In our study, there was no significant difference in the growth of solid nodular lung cancer between patients with mutations in any gene. This suggests that genetic alterations may drive histological aggressiveness rather than growth rate, implying that VDT and molecular profiling should be combined to optimize therapeutic decisions. To explore the molecular mechanisms of nodule growth, future studies using transcriptome, protein regulation, and single-cell sequencing methods will be conducted.

There are several limitations in this study. Firstly, there were differences between preoperative CT scans and follow-up plans for patients. Secondly, squamous cell carcinoma and large cell carcinoma had a higher degree of malignancy, progressed faster, and usually had a larger volume during diagnosis. These cases underwent a series of procedures upon admission, such as puncture, chemotherapy, and targeted therapy. Therefore, these situations needed to be excluded. Thirdly, the Deep Wise system was influenced by different CT scanning protocols, and the variability of its volume measurement was also taken into account. Due to inaccurate segmentation by the AI system, some nodules were excluded, resulting in a reduction in the sample size of this study. Fourthly, quality measurement could simultaneously reflect the volume and density of SN. However, due to the lack of analysis on the variability of quality measurements in the Deep Wise system and the impact of contrast agents on quality measurements, SN growth was not defined by an increase in quality. Finally, the prognoses of enrolled patients were not estimated, which will also be considered in the future.

Conclusions

This study deepens our understanding of the heterogeneity of SN growth in NSCLC. It illustrated the relatively inactive growth of some solid nodular lung cancers. By identifying smoking history, CT values, and deep lobulation as key predictive factors for rapid progression, this study provides a theoretical basis for nodule risk management. Squamous cell carcinoma and poorly differentiated tumors would accelerate the nodule growth. Gene mutations (TP53, ALK) would drive the differentiation of NSCLC, but did not regulate their growth rate. In order to explore the molecular mechanism of nodule growth, future studies such as transcriptome, protein regulation, and single-cell sequencing methods will be conducted. These insights provide a theoretical basis for personalized management of solid nodular lung cancer to improve the prognosis of patients with solid pulmonary nodules.

Acknowledgments

We would like to thank Editage (www.editage.cn) for English language editing.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-516/rc

Data Sharing Statement: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-516/dss

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-516/prf

Funding: This work was supported by Noncommunicable Chronic Diseases-National Science and Technology Major Project (grant Nos. 2024ZD0520000 and 2024ZD0520002), the Fundamental Research Funds for the Central Universities, Peking Union Medical College (grant No. 3332024207), Beijing Natural Science Foundation (grant No. 7222148), National Natural Science Foundation of China (grant No. 81971616).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-516/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by institutional ethics board of Cancer Hospital, Chinese Academy of Medical Sciences (No. NCC-014122). Informed consent was taken from all the patients.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Zheng R, Zhang S, Zeng H, et al. Cancer incidence and mortality in China, 2016. J Natl Cancer Cent 2022;2:1-9. [Crossref] [PubMed]
2. Wyker A, Sharma S, Henderson WW. Solitary Pulmonary Nodule. StatPearls. Treasure Island (FL) ineligible companies. Disclosure: Sanjeev Sharma declares no relevant financial relationships with ineligible companies. Disclosure: William Henderson declares no relevant financial relationships with ineligible companies; StatPearls Publishing. Copyright © 2025, StatPearls Publishing LLC. 2025.
3. Sun K, You A, Wang B, et al. Clinical T1aN0M0 lung cancer: differences in clinicopathological patterns and oncological outcomes based on the findings on high-resolution computed tomography. Eur Radiol 2021;31:7353-62. [Crossref] [PubMed]
4. Ahn Y, Lee SM, Choe J, et al. Prognostic Implications of the Volume Doubling Time of the Solid Component in Lung Adenocarcinomas Manifesting as Part-Solid Lesions on Chest CT. AJR Am J Roentgenol 2025;224:e2432470. [Crossref] [PubMed]
5. Liu J, Qi L, Wang Y, et al. Development of a combined radiomics and CT feature-based model for differentiating malignant from benign subcentimeter solid pulmonary nodules. Eur Radiol Exp 2024;8:8. [Crossref] [PubMed]
6. Liu J, Qi L, Xu Q, et al. A Self-supervised Learning-Based Fine-Grained Classification Model for Distinguishing Malignant From Benign Subcentimeter Solid Pulmonary Nodules. Acad Radiol 2024;31:4687-95. [Crossref] [PubMed]
7. Jiang B, Han D, van der Aalst CM, et al. Lung cancer volume doubling time by computed tomography: A systematic review and meta-analysis. Eur J Cancer 2024;212:114339. [Crossref] [PubMed]
8. Hammer MM. Comparison of volume-based and diameter-based assessment of solid nodules in lung cancer screening. Thorax 2025;80:579. [Crossref] [PubMed]
9. Hammer MM, Gupta S, Byrne SC. Volume Doubling Times of Benign and Malignant Nodules in Lung Cancer Screening. Curr Probl Diagn Radiol 2023;52:515-8. [Crossref] [PubMed]
10. Liu M, Mu J, Song F, et al. Growth characteristics of early-stage (IA) lung adenocarcinoma and its value in predicting lymph node metastasis. Cancer Imaging 2023;23:115. [Crossref] [PubMed]
11. Nakahashi K, Shiono S, Nakatsuka M, et al. Prognostic impact of the tumor volume doubling time in clinical T1 non-small cell lung cancer with solid radiological findings. J Surg Oncol 2022;126:1330-40. [Crossref] [PubMed]
12. Yankelevitz DF, Yip R, Henschke CI. Impact of Duration of Diagnostic Workup on Prognosis for Early Lung Cancer. J Thorac Oncol 2023;18:527-37. [Crossref] [PubMed]
13. Yankelevitz DF, Yip R, Henschke CI. Impact on Prognosis of Stage I Non-Small Cell Lung Cancer Secondary to Delays in Diagnostic Workup. Radiology 2024;313:e240420. [Crossref] [PubMed]
14. Callister ME, Baldwin DR, Akram AR, et al. British Thoracic Society guidelines for the investigation and management of pulmonary nodules. Thorax 2015;70:ii1-ii54. [Crossref] [PubMed]
15. Creamer AW, Horst C, Verghese P, et al. Incidence rate and malignancy risk in new nodules in a lung cancer screening programme. Eur Radiol 2025;35:7156-66. [Crossref]
16. Qi LL, Wu BT, Tang W, et al. Long-term follow-up of persistent pulmonary pure ground-glass nodules with deep learning-assisted nodule segmentation. Eur Radiol 2020;30:744-55. [Crossref] [PubMed]
17. Qi LL, Wang JW, Yang L, et al. Natural history of pathologically confirmed pulmonary subsolid nodules with deep learning-assisted nodule segmentation. Eur Radiol 2021;31:3884-97. [Crossref] [PubMed]
18. Zhang R, Tian P, Qiu Z, et al. The growth feature and its diagnostic value for benign and malignant pulmonary nodules met in routine clinical practice. J Thorac Dis 2020;12:2019-30. [Crossref] [PubMed]
19. Park S, Lee SM, Kim S, et al. Volume Doubling Times of Lung Adenocarcinomas: Correlation with Predominant Histologic Subtypes and Prognosis. Radiology 2020;295:703-12. [Crossref] [PubMed]
20. Setojima Y, Shimada Y, Tanaka T, et al. Prognostic impact of solid-part tumour volume doubling time in patients with radiological part-solid or solid lung cancer. Eur J Cardiothorac Surg 2020;57:763-70. [Crossref] [PubMed]
21. Liu J, Qi L, Wang Y, et al. Diagnostic performance of a deep learning-based method in differentiating malignant from benign subcentimeter (≤10 mm) solid pulmonary nodules. J Thorac Dis 2023;15:5475-84. [Crossref] [PubMed]
22. Herzog C, Jones A, Evans I, et al. Cigarette Smoking and E-cigarette Use Induce Shared DNA Methylation Changes Linked to Carcinogenesis. Cancer Res 2024;84:1898-914. [Crossref] [PubMed]
23. Du M, Xin J, Zheng R, et al. CYP2A6 Activity and Cigarette Consumption Interact in Smoking-Related Lung Cancer Susceptibility. Cancer Res 2024;84:616-25. [Crossref] [PubMed]
24. Jiang X, Liu MW, Zhang X, et al. Observational Study of the Natural Growth History of Peripheral Small-Cell Lung Cancer on CT Imaging. Diagnostics (Basel) 2023;13:2560. [Crossref] [PubMed]
25. Karita R, Suzuki H, Onozato Y, et al. A simple nomogram for predicting occult lymph node metastasis of non-small cell lung cancer from preoperative computed tomography findings, including the volume-doubling time. Surg Today 2024;54:31-40. [Crossref] [PubMed]
26. Chen K, Bai J, Reuben A, et al. Multiomics Analysis Reveals Distinct Immunogenomic Features of Lung Cancer with Ground-Glass Opacity. Am J Respir Crit Care Med 2021;204:1180-92. [Crossref] [PubMed]
27. Li H, Sun Z, Li Y, et al. Disparate genomic characteristics of patients with early-stage lung adenocarcinoma manifesting as radiological subsolid or solid lesions. Lung Cancer 2022;166:178-88. [Crossref] [PubMed]
28. Feng Y, Qian R, Cui D, et al. Mutant TP53 promotes invasion of lung cancer cells by regulating desmoglein 3. J Cancer Res Clin Oncol 2024;150:312. [Crossref] [PubMed]
29. Wang C, Zhang S, Ma B, et al. TP53 mutations upregulate RCP expression via Sp1/3 to drive lung cancer progression. Oncogene 2022;41:2357-71. [Crossref] [PubMed]