A narrative review of diagnosis of lung cancer without a knife: challenges and opportunities

Introduction

Background

Lung cancer is the leading cause of cancer-related death around the world (1). It is also well established that early-stage lung cancer diagnoses are associated with lower mortality rates than disease which is identified in its later stages (1). The significant mortality of lung cancer necessitates reliable and accessible screening tests to begin more extensive workup and allow for the earliest diagnosis possible to promote overall survival. The current gold standard for screening is a low-dose computed tomography (LDCT). This screening test has many potential pitfalls, including false positive tests or incidental findings which may delay further workup.

The current gold standard for diagnosis of lung malignancy is a tissue biopsy to provide a histological diagnosis. However, lung biopsy can be associated with significant costs and carries risk of morbidity in some populations. To address this, the process of lung cancer diagnosis has evolved significantly over recent years with the development of multiple non-invasive approaches to securing a diagnosis. These advances have led to an increasing incidence of diagnoses made without traditional tissue biopsy through a variety of multimodal approaches.

Rationale and knowledge gaps

Patients with lung cancer diagnosed at later stages often have lower survival rates compared to their counterparts whose malignancy is identified at an earlier clinical stage (2,3). A retrospective cohort study from Yang et al. looked at over 5,000 patients to identify clinical and pathological characteristics of lung cancer patients and determine their survival rates based on the stage at which they were diagnosed (2). Their findings demonstrate the prognostic importance of the stage at which a lung malignancy is diagnosed (2). However, this study was limited in that all of the included patients came from the same institution. The National Cancer Institute (NCI) has also found this pattern of mortality. They describe 5-year survival rates for Lung and Bronchus Cancer as 64.7% for localized, 37.1% regional, 9.7% for distant, and 16.5% for the unstaged, showing the importance of earlier diagnosis for likelihood of survival (3). It is important to note that there are different survival rates between non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC). A study from Ganti et al. [2021] found that the overall 5-year survival for NSCLC was 26.4% compared to between 5% to 10% for SCLC (4,5). Through securing a diagnosis earlier in the natural course of their disease, patients are more likely to experience improved mortality through earlier curative surgical resection and radiotherapy (5). Despite these advantages, the costs and risks of securing a tissue diagnosis in the traditional manner must be taken into account.

Lung biopsy, while historically the cornerstone of diagnosing lung malignancy, incurs significant costs to both the patient and the healthcare system. The cost of biopsy varies widely depending on the technique and setting, with median costs ranging from approximately $1,028 for outpatient percutaneous biopsies to nearly $30,000 for inpatient surgical procedures (6). A well-powered retrospective cohort study from Chiu et al. described the measurable complication rates and costs of over 15,000 lung biopsy procedures, finding that these complications significantly increased healthcare expenditures (6). If complications occur during the procedure, the financial impact increases dramatically, with the average cost of a biopsy with complications rising to nearly $38,000 (6). Complications such as pneumothorax, bleeding, and the need for intubation are key drivers of overall costs, with complication-related expenses accounting for approximately 13% of total biopsy-related expenditure (6). The significant financial burden of lung biopsy motivates the need to pursue less expensive methods of diagnosis.

Another modality for diagnostic evaluation of pulmonary lesions is a robotic-assisted bronchoscopy (RAB). This procedure allows for minimally invasive sampling of suspicious pulmonary lesions using a bronchoscopic approach, which may help reduce procedure time and morbidity (7). However, this approach may be cost-prohibitive compared to conventional approaches. There are significant up-front costs associated with the initial purchase of required robotic equipment, as well as other costs such as maintenance, processing fees, and procedural expenses driven by the use of disposable instruments (7). These expenses may mean that RAB remains cost-prohibitive for many patient populations seeking a less-invasive approach to diagnosis. While recent advances in RAB demonstrate high yield clinic outcomes and a satisfactory safety profile, more research is still needed to determine the cost-effectiveness of RAB compared to conventional approaches (7).

The risks associated with lung biopsy are also clinically significant. Major complications, such as pneumothorax, occur in up to 8.4% of cases, while bleeding and respiratory compromise are less frequent but potentially severe (6,8). The likelihood of adverse events is also heightened in patients with smaller or deeply located lesions and those with underlying lung disease, such as chronic obstructive pulmonary disease (COPD). These risks often necessitate a careful, individualized assessment when considering biopsy, given that nearly 42% of patients may need multiple procedures (6). This consideration compounds both the risk and the overall cost. It is also worth noting that RAB, despite its significant cost, maintains a safety profile comparable to that of conventional biopsy (7). Altogether, the balance of diagnostic yield, financial cost, and patient safety underscores the need for advances in non-invasive diagnostic alternatives to traditional lung biopsy.

Objective

In this article, we will discuss the challenges associated with diagnosing lung cancer without a tissue biopsy, as well as the opportunities afforded through utilizing a variety of non-invasive methods. The aim of this article is to provide a general overview of available non-invasive methods for the diagnosis of new lung malignancy, with the hope that thoracic surgeons and other clinicians may use these approaches to diagnose their patients with the greatest accuracy and least invasive method available. We present this article in accordance with the Narrative Review reporting checklist (available at https://ccts.amegroups.com/article/view/10.21037/ccts-2025-1-66/rc).

Methods

To create a brief but thorough review of these methods, a literature search was performed. The authors utilized peer-reviewed publications from 1990 through present day available through PubMed for this review. The literature was reviewed for relevance to the topic at hand and whether it was the most recent addition to the scientific literature concerning that topic (Tables 1,2). These publications were then synthesized to create this review article.

Before diagnosis: LDCT screening

LDCT is the most widely used and well-established screening method for lung cancer around the world. Over the past two decades, several powerful studies have shown the utility and benefit of LDCT screening for lung cancer. In 2011, the National Lung Screening Trial, one of the first randomized controlled studies investigating the role of LDCT in lung cancer screening, demonstrated a 20% reduction in lung cancer mortality with screening LDCT compared to the current standard of care, chest radiography (9). In 2020, the NELSON Trial showed mortality benefits for high-risk individuals (male, previous or current smokers) who underwent volume computed tomography (CT) screening (10). The TALENT Study in 2024 demonstrated the utility of LDCT screening in never-smokers, with a positive LDCT having a 92.1% sensitivity rate and 84.6% specificity rate for lung cancer diagnosis (11). LDCT scans function as a reliable and appropriately sensitive screening modality to trigger further workup and an eventual diagnosis of lung malignancy. Despite the successes of LDCT as a screening modality, a final diagnosis cannot be made without the use of some other method, most often biopsy. The use of other tools in conjunction with LDCT for diagnostic purposes [especially artificial intelligence (AI) models] will be discussed later in this article.

The most commonly cited concern with LDCT for lung cancer screening is the rate of false-positive results and associated harm from unnecessary follow up interventions. However, financial analyses generally support LDCT as a cost-effective screening modality (12). Overall, LDCT is still considered the standard of care for screening of lung cancer in at-risk populations. Reliable screening tools such as LDCT allow for the identification of at-risk patients who may then undergo further workup for definitive diagnosis.

Positron emission tomography (PET)

PET with fluorodeoxyglucose (FDG) is a highly accurate modality for identifying malignancy due to the increased glucose metabolism (and therefore increased FDG metabolism) seen in highly metabolically active malignant cells. Due to this specificity, PET/CT imaging has been shown to have superior diagnostic accuracy in lung cancer when compared to LDCT (13). Reported sensitivities for lung cancer detection with PET imaging range from 83% to 100%, and specificities range from 63% to 90% in various reports (13). However, there is a risk of false negative results in lung cancer with low metabolic activity, such as carcinoid tumors or bronchioalveolar carcinomas. Additionally, nodules smaller than 1cm in diameter may be associated with false negative results as well (14).

Recent developments in PET biomarkers aim to develop novel radiopharmaceuticals that may increase the sensitivity and specificity of PET imaging. These include markers such as ¹⁸F-fluorothymidine, which is primarily used as an indicator of treatment response, 18F-fluoromisonidazole, which is used for imaging tumor hypoxia, and 11C-methionine, which may allow providers to distinguish between benign and malignant masses (15). While use of these markers is not yet routine practice, a multimodal approach to PET imaging shows promise for more sensitive and specific identification of lung malignancies.

Several factors challenge the diagnostic accuracy of PET imaging. One such factor is reduced sensitivity for small nodules (especially <7–8 mm) and microscopic disease, where limited spatial resolution and partial-volume effects cause underestimation of tracer uptake and falsenegative scans (16,17). Another is that FDG uptake is not always specific for malignancy, as infectious and inflammatory conditions (such as granulomatous disease, pneumonia, or sarcoidosis) can be intensely FDG-avid, producing false positives and lowering specificity. Interpretation is further complicated by motion artifacts from breathing and cardiac motion, and by the institutional availability of PET imaging. Though these challenges may reduce the prevalence of PET imaging, it is still widely used as a non-invasive measure for diagnosis of lung malignancies.

Lung nodule calculators

Lung nodule risk calculators are available for surgeons to utilize when trying to determine whether operative intervention is appropriate in a patient who has a nodule seen on imaging without definitive tissue diagnosis. These models may help the clinician stratify malignancy risk without utilizing invasive procedures, and in some cases to reliably diagnose a new malignancy.

The Brock University Model (also known as the “PanCan” model) was originally developed to stratify risk of malignancy for pulmonary nodules seen on LDCT screening. The model incorporates several variables, including age of the patient, sex of the patient, family history of lung cancer, presence of emphysema on CT, nodule size, nodule type, nodule location, nodule count, and presence of nodule spiculation to determine an estimated probability of malignancy for nodules detected on LDCT. This model is well-calibrated across external validation studies, and meta-analysis has demonstrated consistency and the highest accuracy of available risk stratification models, specifically when used for screening high-risk individuals (18,19). A study from Gonzalez Maldonado et al. looked at 1,159 participants from the German Lung Cancer Screening Intervention trial with 3,903 pulmonary nodules identified on LDCT to determine the discrimination ability (ability to distinguish malignant vs. benign nodules) and calibration (how well predicted risks match observed outcomes) of several malignancy prediction models (19). They found that screening-based models, such as the Brock University Model, performed best based on these metrics (19).

Many clinicians also use the Mayo Clinic model to predict the probability of malignancy of solitary pulmonary nodules found incidentally on imaging (as opposed to the Brock model, which is intended for use in nodules seen on screening LDCT). This model incorporates the patient’s age, smoking history, history of extrathoracic cancer, nodule diameter, nodule spiculation, and location in an upper lobe to provide an estimated malignancy probability. Systematic reviews and metanalyses have shown that this model gives a fair estimate of malignancy risk, but may be more applicable for specific populations such as smokers (20).

Compared to the Mayo Clinic Model, the Brock University Model is generally more reliable with an accuracy of approximate 84% in one study from Nair et al. (18). This may be due to the development of the Mayo Clinic Model in a study population with a much high prevalence of cancer than the screening population, leading to an overestimation of cancer risk (18). Although these models provide a structured framework for the estimation of malignancy potential, this clinical application of these models still varies greatly based on clinician experience and unconscious bias. It is critical that models are always used as a single data point in a larger discussion surrounding patient prognosis and treatment, and that best clinical judgement is always used when deciding whether to rely on the conclusions of these clinical models.

Bronchoscopy

Bronchoscopy plays a crucial role in the minimally invasive diagnosis of lung cancer by allowing direct visualization and targeted sampling of the airways. This method is especially important for the diagnosis of small or peripheral lesions that may others be technically challenging to sample. Traditional white light bronchoscopy (WLB) allows for inspection of the bronchial mucosa but has limited sensitivity, particularly for early or subtle lesions. Autofluorescence bronchoscopy (AFB) enhances detection by exploiting the differences in fluorescence between normal and dysplastic or malignant tissues, achieving higher sensitivity than WLB but with lower specificity (21,22). Combining AFB with WLB improves overall detection rates of precancerous and cancerous lesions (21). Narrow band imaging (NBI) further enhances diagnostic accuracy by improving mucosal contrast and highlighting abnormal microvascular patterns associated with malignancy. A single-center retrospective cohort study from Zhu et al. shows that NBI achieves superior sensitivity and specificity compared to both AFB and WLB, with NBI demonstrating sensitivity around 91.7% and specificity near 85% in diagnosing central lung cancer (23). Although their findings show a significant difference in sensitivity and specificity in this diagnostic method, the study is limited by its small sample size and single-center data (23). This technique also allows for real-time assessment of vascular patterns, which may correlate with histological cancer types to allow even more specificity in diagnosis. The integration of these bronchoscopic imaging modalities provides another powerful modality for the early and accurate identification of lung cancer while minimizing the need for invasive biopsy procedures.

RAB provides another bronchoscopic modality for diagnosis of suspicious pulmonary lesions. These procedures use an omnidirectional bronchoscope on a robotic arm which is guided by CT-scans performed prior to procedure. Key benefits of RAB over conventional bronchoscopy include enhanced reach in the peripheral lung, improved maneuverability, and superior procedural stability, which translate into higher diagnostic yields and fewer sampling errors for small or difficult-to-access lesions (6). Additionally, RAB allows for the integration multiple aspects of workup into one procedure, including visualization of lesions, sampling of suspicious nodules, and staging (6). The safety profile of RAB is similar to that of conventional bronchoscopy, highlighting its extensive benefits without significant drawbacks when compared to conventional bronchoscopy (6). However, as noted above, further work is needed to better characterize the cost-effectiveness of RAB. Despite the significant cost of the procedure, RAB has potential to provide great clinical yield with limited risks as a minimally-invasive approach to diagnosis of lung cancer. Other considerations for the feasible implementation of bronchoscopy include institutional availability and differences in outcomes based on operator experience and support. This modality is one that is likely only beneficial to patients in a resource-rich environment with providers who have adequate experience and support using bronchoscopy. As bronchoscopy becomes more widely available (particularly in healthcare settings with constrained resources), providers may more frequently be able to use bronchoscopy as a standard part of their practice.

Breath analyzers

One of the least invasive measures for diagnosing lung malignancy involves monitoring for changes in exhaled volatile organic compounds (VOCs) in breath tests. Recent research from Choueiry et al. has investigated detectable differences in VOCs correlated to oncogenesis when compared to those without malignant processes (24). The leading theory suggests that the increased metabolic rate found in malignant tissue produces an increased quantity of volatile byproducts, such as cyclohexane and xylene, and that these VOCs have potential to become markers for the early detection of lung cancer (24,25). However, clinical applications of these VOC breath test screenings are still severely limited by the availability of the instruments required for analysis, such as mass spectrometers, as well as a lack of standardization of sample collection (24). In addition, a reliable panel of known lung cancer-related VOC biomarkers is yet to be established (26). However, a study from Phillips et al. demonstrated that a panel of 22 specific VOC have 100% sensitivity and 81.3% specificity for identifying stage I lung cancer in patients with an abnormal chest radiograph (27). Further work is needed to confirm that this panel of VOCs could be widely applicable to a screening population.

Although breath analyzers using VOCs have great potential for clinical applications in the early and non-invasive detection of lung cancer, current evidence does not yet provide reliable parameters for clinical use. With further research in creating a reliable panel of known lung cancer-related VOCs and standardization of sample collection, this technique may offer surgeons a reliable and extremely non-invasive diagnostic modality. For the time being, this method remains complementary to histological diagnosis in most situations.

Biomarkers

Biomarkers play an important and expanding role in the non-invasive diagnosis of lung cancer. Liquid biopsy, which analyzes body fluids such as blood, offers a non-invasive alternative to access key tumor-derived materials for diagnosis and monitoring. Among the most studied biomarkers in liquid biopsy are circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA). CTCs are actual tumor cells shed into the bloodstream, while ctDNA consists of fragmented tumor-specific DNA released by these cells (28). These biomarkers allow detection of genetic mutations characteristic of lung cancer, providing insights into tumor characteristics that tissue biopsies may miss due to sampling limitations. Sequencing technologies have also been used to enhance the precision of mutation detection, and this information may then be used to create targeted therapeutic interventions. Clinical data suggest that liquid biopsies not only improve early detection of lung cancer, but also help guide treatment decisions, leading to improved outcomes for patients with NSCLC (28).

Beyond CTCs and ctDNA, other promising molecular biomarkers include microRNAs (miRNAs), circulating exosomes, and lung cancer-associated autoantibodies. miRNAs are small non-coding RNAs involved in gene regulation and are detectable in circulation. Specific expression profiles of miRNA have been linked to lung cancer presence and progression (29). Circulating exosomes are extracellular vesicles secreted by tumor cells that carry tumor-related proteins, RNA, and DNA. Autoantibodies generated against tumor-associated antigens can appear even before clinical symptoms, offering potential for early risk stratification and potentially diagnosis (30). While some of these biomarkers have demonstrated high specificity, their sensitivity varies, highlighting the utility of combining multiple biomarker types in panels to improve diagnostic accuracy (30). Collectively, these biomarkers represent a promising field for non-invasively diagnosing lung cancer at its earliest stages, or at the very least providing valuable insight into risk stratification for at-risk populations.

AI

The use of AI in the diagnosis of lung cancer is an emerging and evolving field. One of the most active fields of research involves radiomics, which is used to identify and extract quantitative features from medical imaging to be analyzed by AI models. These features are then used to identify, classify, and better characterize lung nodules to determine the risk of malignancy (31). These AI models, including deep learning modalities such as large language models, explainable AI, and super resolution, have so far demonstrated improved sensitivity and specificity compared with conventional visual assessment of pulmonary nodules (31). There is also a benefit to the patient with these AI models, as they require only non-invasive imaging to provide both a diagnosis and staging in many cases. Despite the apparent advantages of using AI as a non-invasive diagnostic modality, the risks include model overfitting, lack of generalization over multiple heterogenous populations, and difficulty in regulatory approval (31). While much of the work so far in validating AI models for use in cancer diagnosis shows significant promise, there remains work to be done to ensure safe and equitable use of these tools for the diagnosis of lung cancer.

Conclusions

The development of multiple minimally- and non-invasive modalities for the diagnosis of lung cancer presents both a challenge and an opportunity for thoracic surgeons. The methods discussed in this review, including advanced bronchoscopic techniques, AI, and liquid biopsy, may allow for more rapid and safe identification of lung malignancy while minimizing risk to patients. However, several challenges remain before these techniques become integral to the practice of thoracic surgery. These barriers include technical feasibility, institutional availability, cost effectiveness, and accuracy limitations as discussed. Despite these challenges, minimally- and non-invasive techniques demonstrate an evolving paradigm shift in the diagnosis of new lung malignancies.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the Narrative Review reporting checklist. Available at https://ccts.amegroups.com/article/view/10.21037/ccts-2025-1-66/rc

Peer Review File: Available at https://ccts.amegroups.com/article/view/10.21037/ccts-2025-1-66/prf

Funding: None.

Conflicts of Interest: Both authors have completed the ICMJE uniform disclosure form (available at https://ccts.amegroups.com/article/view/10.21037/ccts-2025-1-66/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.

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. Gierada DS, Black WC, Chiles C, et al. Low-Dose CT Screening for Lung Cancer: Evidence from 2 Decades of Study. Radiol Imaging Cancer 2020;2:e190058. [Crossref] [PubMed]
2. Yang P, Allen MS, Aubry MC, et al. Clinical features of 5,628 primary lung cancer patients: experience at Mayo Clinic from 1997 to 2003. Chest 2005;128:452-62. [Crossref] [PubMed]
3. National Cancer Institute. Cancer of the Lung and Bronchus - Cancer Stat Facts [Internet]. National Cancer Institute. 2024. Available online: https://seer.cancer.gov/statfacts/html/lungb.html
4. Ganti AK, Klein AB, Cotarla I, et al. Update of Incidence, Prevalence, Survival, and Initial Treatment in Patients With Non-Small Cell Lung Cancer in the US. JAMA Oncol 2021;7:1824-32. [Crossref] [PubMed]
5. Small Cell Lung Cancer Treatment (PDQ®) [Internet]. Cancer.gov. 2025 [cited 2025 Oct 2]. Available online: https://www.cancer.gov/types/lung/hp/small-cell-lung-treatment-pdq#section_1.20
6. Chiu YW, Kao YH, Simoff MJ, et al. Costs of Biopsy and Complications in Patients with Lung Cancer. Clinicoecon Outcomes Res 2021;13:191-200. [Crossref] [PubMed]
7. Bhandari BS, Jain A, Sharma S, et al. Robotic Bronchoscopy: A Comprehensive Review. J Respir 2024;4:128-39.
8. McLean AEB, Barnes DJ, Troy LK. Diagnosing Lung Cancer: The Complexities of Obtaining a Tissue Diagnosis in the Era of Minimally Invasive and Personalised Medicine. J Clin Med 2018;7:163. [Crossref] [PubMed]
9. National Lung Screening Trial Research Team. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365:395-409.
10. Jonas DE, Reuland DS, Reddy SM, et al. Screening for Lung Cancer With Low-Dose Computed Tomography: Updated Evidence Report and Systematic Review for the US Preventive Services Task Force. JAMA 2021;325:971-87. [Crossref] [PubMed]
11. Chang GC, Chiu CH, Yu CJ, et al. Low-dose CT screening among never-smokers with or without a family history of lung cancer in Taiwan: a prospective cohort study. Lancet Respir Med 2024;12:141-52. [Crossref] [PubMed]
12. Amicizia D, Piazza MF, Marchini F, et al. Systematic Review of Lung Cancer Screening: Advancements and Strategies for Implementation. Healthcare (Basel) 2023;11:2085. [Crossref] [PubMed]
13. McLoud TC. The role of PET scanning in the evaluation of lung carcinoma. Cancer Imaging 2003;3:83-4.
14. Hochhegger B, Alves GR, Irion KL, et al. PET/CT imaging in lung cancer: indications and findings. J Bras Pneumol 2015;41:264-74. [Crossref] [PubMed]
15. Zheng C, Miao J, Xu L, et al. Novel PET imaging biomarkers as predictors of postoperative recurrence in lung adenocarcinoma. BMC Cancer 2025;25:874. [Crossref] [PubMed]
16. Szyszko TA, Yip C, Szlosarek P, et al. The role of new PET tracers for lung cancer. Lung Cancer 2016;94:7-14. [Crossref] [PubMed]
17. Pellegrino S, Fonti R, Pulcrano A, et al. PET-Based Volumetric Biomarkers for Risk Stratification of Non-Small Cell Lung Cancer Patients. Diagnostics (Basel) 2021;11:210. [Crossref] [PubMed]
18. Nair VS, Sundaram V, Desai M, et al. Accuracy of Models to Identify Lung Nodule Cancer Risk in the National Lung Screening Trial. Am J Respir Crit Care Med 2018;197:1220-3. [Crossref] [PubMed]
19. González Maldonado S, Delorme S, Hüsing A, et al. Evaluation of Prediction Models for Identifying Malignancy in Pulmonary Nodules Detected via Low-Dose Computed Tomography. JAMA Netw Open 2020;3:e1921221. [Crossref] [PubMed]
20. Papalampidou A, Papoutsi E, Katsaounou PA. Pulmonary nodule malignancy probability: a diagnostic accuracy meta-analysis of the Mayo model. Clin Radiol 2022;77:443-50. [Crossref] [PubMed]
21. Zhang J, Wu J, Yang Y, et al. White light, autofluorescence and narrow-band imaging bronchoscopy for diagnosing airway pre-cancerous and early cancer lesions: a systematic review and meta-analysis. J Thorac Dis 2016;8:3205-16. [Crossref] [PubMed]
22. Andolfi M, Potenza R, Capozzi R, et al. The role of bronchoscopy in the diagnosis of early lung cancer: a review. J Thorac Dis 2016;8:3329-37. [Crossref] [PubMed]
23. Zhu J, Liu R, Wu X, et al. The value of narrow-band imaging bronchoscopy in diagnosing central lung cancer. Front Oncol 2022;12:998770. [Crossref] [PubMed]
24. Choueiry F, Barham A, Zhu J. Analyses of lung cancer-derived volatiles in exhaled breath and in vitro models. Exp Biol Med (Maywood) 2022;247:1179-90. [Crossref] [PubMed]
25. Keogh RJ, Riches JC. The Use of Breath Analysis in the Management of Lung Cancer: Is It Ready for Primetime? Curr Oncol 2022;29:7355-78. [Crossref] [PubMed]
26. Fan X, Zhong R, Liang H, et al. Exhaled VOC detection in lung cancer screening: a comprehensive meta-analysis. BMC Cancer 2024;24:775. [Crossref] [PubMed]
27. Phillips M, Gleeson K, Hughes JM, et al. Volatile organic compounds in breath as markers of lung cancer: a cross-sectional study. Lancet 1999;353:1930-3. [Crossref] [PubMed]
28. Ma L, Guo H, Zhao Y, et al. Liquid biopsy in cancer: current status, challenges and future prospects. Sig Transduct Target Ther 2024;9:336.
29. Bibikova M, Fan J. Liquid biopsy for early detection of lung cancer. Chin Med J Pulm Crit Care Med 2023;1:200-6. [Crossref] [PubMed]
30. Seijo LM, Peled N, Ajona D, et al. Biomarkers in Lung Cancer Screening: Achievements, Promises, and Challenges. J Thorac Oncol 2019;14:343-57. [Crossref] [PubMed]
31. Boubnovski Martell M, Linton-Reid K, Chen M, et al. Radiomics for lung cancer diagnosis, management, and future prospects. Clin Radiol 2025;86:106926. [Crossref] [PubMed]