Lung Cancer Screening

Classification of Chest CT Lung Nodules Using Collaborative Deep Learning Model

Background

Early detection of lung cancer through accurate diagnosis of malignant lung nodules using chest CT scans offers patients the highest chance of successful treatment and survival. Despite advancements in computer vision through deep learning algorithms, the detection of malignant nodules faces significant challenges due to insufficient training datasets.

Methods

This study introduces a model based on collaborative deep learning (CDL) to differentiate between cancerous and non-cancerous nodules in chest CT scans with limited available data. The model dissects a nodule into its constituent parts using six characteristics, allowing it to learn detailed features of lung nodules. It utilizes a CDL submodel that incorporates six types of feature patches to fine-tune a network previously trained with ResNet-50. An adaptive weighting method learned through error backpropagation enhances the process of identifying lung nodules, incorporating these CDL submodels for improved accuracy.

Results

The CDL model demonstrated a high level of performance in classifying lung nodules, achieving an accuracy of 93.24%. This represents a significant improvement over current state-of-the-art methods, indicating the effectiveness of the proposed approach.

Conclusion

The findings suggest that the CDL model, with its unique structure and adaptive weighting method, offers a promising solution to the challenge of accurately detecting malignant lung nodules with limited data. This approach not only improves diagnostic accuracy but also contributes to the early detection and treatment of lung cancer, potentially saving lives.

Multi-kernel driven 3D convolutional neural network for automated detection of lung nodules in chest CT scans

The accurate position detection of lung nodules is crucial in early chest computed tomography (CT)-based lung cancer screening, which helps to improve the survival rate of patients. Deep learning methodologies have shown impressive feature extraction ability in the CT image analysis task, but it is still a challenge to develop a robust nodule detection model due to the salient morphological heterogeneity of nodules and complex surrounding environment. In this study, a multi-kernel driven 3D convolutional neural network (MK-3DCNN) is proposed for computerized nodule detection in CT scans. In the MK-3DCNN, a residual learning-based encoder-decoder architecture is introduced to employ the multi-layer features of the deep model. Considering the various nodule sizes and shapes, a multi-kernel joint learning block is developed to capture 3D multi-scale spatial information of nodule CT images, and this is conducive to improving nodule detection performance. Furthermore, a multi-mode mixed pooling strategy is designed to replace the conventional single-mode pooling manner, and it reasonably integrates the max pooling, average pooling, and center cropping pooling operations to obtain more comprehensive nodule descriptions from complicated CT images. Experimental results on the public dataset LUNA16 illustrate that the proposed MK-3DCNN method achieves more competitive nodule detection performance compared to some state-of-the-art algorithms. The results on our constructed clinical dataset CQUCH-LND indicate that the MK-3DCNN has a good prospect in clinical practice.

Shifting the Cancer Screening Paradigm: The Rising Potential of Blood-Based Multi-Cancer Early Detection Tests

Cancer remains a leading cause of death worldwide, partly owing to late detection which entails limited and often ineffective therapeutic options. Most cancers lack validated screening procedures, and the ones available disclose several drawbacks, leading to low patient compliance and unnecessary workups, adding up the costs to healthcare systems. Hence, there is a great need for innovative, accurate, and minimally invasive tools for early cancer detection. In recent years, multi-cancer early detection (MCED) tests emerged as a promising screening tool, combining molecular analysis of tumor-related markers present in body fluids with artificial intelligence to simultaneously detect a variety of cancers and further discriminate the underlying cancer type. Herein, we aim to provide a highlight of the variety of strategies currently under development concerning MCED, as well as the major factors which are preventing clinical implementation. Although MCED tests depict great potential for clinical application, large-scale clinical validation studies are still lacking.

MicroRNA-21 as a Diagnostic and Prognostic Biomarker of Lung Cancer: A Systematic Review and Meta-Analysis

Background: The relationship between microRNA-21 (miRNA-21) and pathogenesis of lung cancer is a considerable focus of research interest. However, to our knowledge, no in-depth meta-analyses based on existing evidence to ascertain the value of miRNA-21 in diagnosis and clinical prognosis of lung cancer have been documented.

Methods: We comprehensively searched all the literature pertaining to ‘miRNA-21’ and ‘lung cancer’ from four databases from the period of inception of each database until May 2020. Using specific inclusion and exclusion criteria, the literature for inclusion was identified and the necessary data extracted.

Results: In total, 46 articles were included in the meta-analysis, among which 31 focused on diagnostic value and 15 on prognostic value. Combined sensitivity (SEN) of miRNA-21 in diagnosis of lung cancer was 0.77 (95% confidence interval (CI): 0.72–0.81), specificity (SPE) was 0.86 (95% CI: 0.80–0.90), diagnostic odds ratio (DOR) was (95% CI: 12–33), and area under the SROC curve (AUC) was 0.87 (95% CI: 0.84–0.90). No significant correlations were observed between abnormal expression of miRNA-21 and gender, smoking habits, pathological type and clinical stage of lung cancer (P>0.05). In terms of overall survival (OS), univariate analysis (hazards ratio (HR) = 1.49, 95% CI: 1.22–1.82) revealed high expression of miRNA-21 as an influencing factor for lung cancer. MiRNA-21 was confirmed as an independent risk factor for poor prognosis in multivariate analysis (HR = 1.65, 95% CI: 1.24–2.19).

Conclusion: MiRNA-21 has potential clinical value in the diagnosis and prognosis of lung cancer and may serve as an effective diagnostic marker and therapeutic target in the future.

Referring high-risk individuals for lung cancer screening: A systematic review of interventions with healthcare professionals

OBJECTIVE

This systematic review described the effect of interventions aimed at helping Healthcare Professionals refer high-risk individuals for lung cancer screening. Primary outcomes included: patient outcomes such as lung cancer detection, screening for lung cancer, lung cancer treatments received and lung cancer mortality. Healthcare professionals' knowledge and awareness of lung cancer screening served as secondary outcomes.

METHODS

Experimental studies published between January 2016 and 2021 were included. The search was conducted in MEDLINE, CINAHL, ERIC, PsycARTICLES, PsycInfo and Psychology and Behavioral Sciences Collection. The quality of the included studies was assessed using the Mixed Methods Appraisal Tool and the level of evidence was assessed using the Scottish Intercollegiate Guidelines Network grading system.

RESULTS

Nine studies were included. Nurse navigation, electronic prompts for lung cancer screening and shared decision-making helped improve patient outcomes. Specialist screenings yielded more significant incidental findings and a higher percentage of Lung-RADS 1 results (i.e. no nodules/definitely benign nodules), while Primary Care Physician screenings were associated with higher numbers of Lung-RADS 2 results (i.e. benign nodules with a very low likelihood to becoming malignant). An increase in Healthcare Professionals' knowledge and awareness of lung cancer screening was achieved using group-based learning compared to lecture-based education delivery.

CONCLUSIONS

The effectiveness of Nurse navigation is evident, as are the benefits of adequate training, shared decision-making, as well as a structured, clear and well-understood referral processes supported by the use of electronic system-incorporated prompts.

A Comparative Study of Radiomics and Deep-Learning Based Methods for Pulmonary Nodule Malignancy Prediction in Low Dose CT Images

OBJECTIVES

Both radiomics and deep learning methods have shown great promise in predicting lesion malignancy in various image-based oncology studies. However, it is still unclear which method to choose for a specific clinical problem given the access to the same amount of training data. In this study, we try to compare the performance of a series of carefully selected conventional radiomics methods, end-to-end deep learning models, and deep-feature based radiomics pipelines for pulmonary nodule malignancy prediction on an open database that consists of 1297 manually delineated lung nodules.

METHODS

Conventional radiomics analysis was conducted by extracting standard handcrafted features from target nodule images. Several end-to-end deep classifier networks, including VGG, ResNet, DenseNet, and EfficientNet were employed to identify lung nodule malignancy as well. In addition to the baseline implementations, we also investigated the importance of feature selection and class balancing, as well as separating the features learned in the nodule target region and the background/context region. By pooling the radiomics and deep features together in a hybrid feature set, we investigated the compatibility of these two sets with respect to malignancy prediction.

RESULTS

The best baseline conventional radiomics model, deep learning model, and deep-feature based radiomics model achieved AUROC values (mean ± standard deviations) of 0.792 ± 0.025, 0.801 ± 0.018, and 0.817 ± 0.032, respectively through 5-fold cross-validation analyses. However, after trying out several optimization techniques, such as feature selection and data balancing, as well as adding context features, the corresponding best radiomics, end-to-end deep learning, and deep-feature based models achieved AUROC values of 0.921 ± 0.010, 0.824 ± 0.021, and 0.936 ± 0.011, respectively. We achieved the best prediction accuracy from the hybrid feature set (AUROC: 0.938 ± 0.010).

CONCLUSION

The end-to-end deep-learning model outperforms conventional radiomics out of the box without much fine-tuning. On the other hand, fine-tuning the models lead to significant improvements in the prediction performance where the conventional and deep-feature based radiomics models achieved comparable results. The hybrid radiomics method seems to be the most promising model for lung nodule malignancy prediction in this comparative study.

Predictive value of the hemoglobin, albumin, lymphocyte, and platelet (HALP) score and lymphocyte-to-monocyte ratio (LMR) in patients with non-small cell lung cancer after radical lung cancer surgery

Background

Examining the analytical worth of the preoperative hemoglobin, albumin, lymphocyte, platelet (HALP) score and lymphocyte-to-monocyte ratio (LMR) within diseased persons having non-small cell lung cancer (NSCLC) after radical lung cancer surgery.

Methods

Clinical data concerning 238 diseased persons with NSCLC who underwent radical lung cancer resection within Nantong Cancer Hospital between January 2009 and October 2015 had been looking back studied. ROC curve had been employed in regulating optimal critical worth of HALP and LMR that had been 48.00 and 6.30 singly. A 5-year amplification observed survival concerning diseased persons, and clinicopathological stuff assessed using statistics procedure. Kaplan Meier method, log rank test had been exploited from the point of view to analyze for surviving, and Cox regression analysis had been exploited for univariate and multivariate analysis. Eventually, a nomogram had been produced to examine the confirmation internally.

Results

Kaplan Meier survival assessment revealed top HALP class's overall survival (OS) was significantly higher than below HALP class's (P<0.001), and high LMR group's OS was also greater than below LMR class's (P=0.001). Patients possessing average continuance period of 4 years. Further stratified study revealed high HALP class possessed notable OS as compared below HALP class (P=0.0002), and top LMR class possessed considerable OS as compared to below LMR class (P=0.003) in lung adenocarcinoma. In non-adenocarcinoma, there was no substantial difference in OS between two classes (P>0.05). Preoperative HALP and LMR remained independent risk constituents for tumor progression (P=0.005, P=0.028), lymph node metastasis and level of differentiation also had a certain effect on tumor progression (P<0.05), according to Cox multivariate analysis. Rise in HALP and LMR will help diseased persons having NSCLC live longer. The nomogram's c-index in inside validation was 0.672 (95% confidence interval: 0.626-0.718).

Conclusions

Preoperative HALP versus LMR are independent predictive aspect within NSCLC diseased persons linked to clinicopathological features, and has a particular value in determining bodement.

Video-Based Education in Lung Cancer Screening

Lung cancer (LC) is the leading cause of cancer mortality in the USA; the American Cancer Society (ACS) estimates upwards of 220,000 new cases will be diagnosed this year. Recently, the Center for Medicare/Medicaid Services (CMS) agreed to cover LC screening with low-dose computed tomography (CT) for patients; however, CMS requires prior documentation of a shared decision-making (SDM) visit between the patient and the referring clinician to inform them about risks of screening. LC screening programs have begun to use YouTube for patient recruitment, education, and marketing of screening. The objective of this study is to shed light on the role of YouTube in lung cancer screening in terms of guidelines, screening options, target population, steps after screening, and risks and benefits of screening. We searched YouTube.com ™ to identify videos dealing with lung screening using the keywords: lung cancer screening. Videos without sound, uploaded before 2009, longer than 20 min, duplicate videos, and videos in a language other than English were excluded. This method yielded 123 videos that fit criteria. Videos were coded for inclusion of LC screening process, risks and benefits of screening, screening guidelines, risk factors for LC, and treatment options after LC diagnosis. One hundred twenty-three videos had a cumulative 261,261 views across all videos. A total of 38.7% of the videos included no mention of United States Preventive Services Task Force (USPSTF) or CMS guidelines for LC screening. Only 30% included any mention of the risks associated with screening: 14% mentioned false positives, 12% radiation, and 4% anxiety associated with screening. Ninety-two percent of all videos sampled were intended for patients, and the majority of videos were created by medical institutions (66%) and news channels (17%). Lung cancer screening videos on YouTube's platform have garnered a substantial amount of views. While all videos sampled highlighted the benefits of LC screening, the majority fail to discuss the risks associated with the screening process. Most videos were produced for marketing purposes rather than educational and therefore should not be used as a substitute for SDM visits.

Value of circulating tumor cells in the diagnosis and treatment of solitary pulmonary nodules

Background

Lung cancer has become the most common malignant tumor worldwide, with the highest rates of morbidity and mortality. The detection of circulating tumor cells (CTCs) can be simple, rapid, and minimally invasive, thus endowing them with a high value in the diagnosis of malignant tumors. We aimed to explore the correlation between CTCs in peripheral blood and benign or malignant solitary pulmonary nodules (SPNs).

Methods

A total of 223 patients with SPNs from January 2018 to May 2020 were recruited. During the same period, 20 healthy volunteers were recruited as controls. Venous blood samples were collected from participants for detecting CTCs using a folate receptor (FR)-positive cell detection kit, as well as tumor biomarkers.

Results

A significant difference in the level of CTCs were observed between the malignant SPNs group, the benign SPNs group, and the control group, which was markedly higher in the malignant SPNs group (10.48±3.49 FU/3 mL) than both the benign SPNs and control groups (6.38±0.53 and 4.45±1.21 FU/3 mL, respectively) (P<0.001). In addition, the level of CTCs was significantly higher in the benign SPNs group than in the control group (P=0.023). In particular, in the malignant SPNs group, patients older than 60 years (11.45±3.92 FU/3 mL) presented a notably higher level of CTCs than other patients (9.55±2.74 FU/3 mL). The patients were then classified according to the pathological subtypes of lung cancer. There was a significant difference in level of CTCs among patients with squamous cell carcinoma (9.10±1.94 FU/3 mL), adenocarcinoma (10.77±3.71 FU/3 mL), and adenosquamous cell carcinoma (11.78±2.61 FU/3 mL). Binary logistic regression analysis suggested that CTCs were an independent risk factor of malignant SPN (OR =3.698, 95% CI: 1.136–11.035, P=0.030). The sensitivity and specificity of CTCs in diagnosing malignant SPNs was significantly higher than tumor biomarkers (single or combined) [sensitivity =89.1%; specificity =92.3%; area under curve (AUC) (95% CI) =0.907 (0.861–0.942)].

Conclusions

Peripheral blood CTCs can be used in the diagnosis of malignant SPNs and are recommended for clinical application.

Evaluation of the association between vitamin D and lung cancer skin metastasis: A protocol for systematic review

BACKGROUND

This study aims to investigate the association between vitamin D (VD) and lung cancer skin metastasis (LCSM).

METHODS

The following databases will be retrieved from the beginning to the present of each database without language limitation: PUBMED, EMBASE, Cochrane Library, Web of Science, CBM, and CNKI. The reference lists of included trials and other sources will also be checked. Two researchers will independently undertake literature selection, data collection, and study quality evaluation. We will utilize a fixed or random-effect model to pool the data according to the heterogeneity test. The RevMan 5.3 software will be used to analyze the data and perform meta-analysis.

RESULTS

This study will summarize high quality study to explore the association between VD and LCSM.

CONCLUSION

The findings of this study will help to judge whether there is association between VD and LCSM.

ETHICS AND DISSEMINATION

No research ethical approval is required in this study, because it will only analyze published data. It is expected to disseminate through a peer-reviewed journal.

STUDY REGISTRATION

osf.io/ph2au.

Predictors of Lung Cancer Screening Utilization in a Population-Based Survey

PURPOSE

Annual low-dose CT (LDCT) screening in high-risk individuals has been recommended to detect lung cancer earlier and reduce mortality. The objective of this study was to identify demographic, financial, and health care factors associated with screening uptake in a population-based survey.

METHODS

Data from the Lung Cancer Screening Module and core modules of the 2017 Behavioral Risk Factor Surveillance System, a population-based survey administered via cell phone and landline, were analyzed to examine demographic, health, and financial factors associated with screening uptake among the 10 states that administered the screening module. Weighted frequencies and confidence intervals (CIs) were produced, and weighted Wald χ² tests were used to compare differences in screening utilization by patient characteristics. A multivariate logistic mixed-effects model was constructed, in which participant clustering by state was accounted for with a random intercept.

RESULTS

The uninsured were less likely to undergo LDCT screening (odds ratio [OR], 0.28; 95% CI, 0.12-0.65). LDCT screening uptake was higher for participants with chronic respiratory conditions (OR, 4.14; 95% CI, 2.33-7.35); those who were divorced, separated, widowed, or refused to answer (OR, 1.41; 95% CI, 1.05-1.86); those who had previous cancer diagnoses (OR, 1.90; 95% CI, 1.40-2.56); and those aged 65 to 69 years (OR, 1.23; 95% CI, 1.06-1.44) or 70 to 74 years (OR, 1.17; 95% CI, 1.00-1.37). Utilization also varied significantly across states.

CONCLUSIONS

Having a related health condition whereby participants were sensitized to the benefits of early screening (ie, another cancer diagnosis, presence of chronic respiratory conditions) and having insurance coverage were associated with higher LDCT screening uptake. Providers should engage LDCT-eligible patients through informed and shared decision making to increase preference-sensitive screening decisions.

Role of chest radiographs in early lung cancer detection

Background

Lung cancer is the second most common and the most fatal form of cancer. Although annual low-dose computed tomography is used as the primary method of cancer screening, it presents challenges regarding resources as well as potential health risks from radiation exposure. Chest radiography (CXR), though less effective, is used frequently and commonly. Moreover, often in clinical settings, CXR is the first imaging modality used; computed tomography is subsequently performed if abnormalities are detected on CXRs. This study examined whether controlling for distractors and time constraints, as well as side-by-side comparison of multiple CXRs in clinical settings can aid earlier detection of radiological abnormalities indicative of lung cancer lesions.

Methods

Thirty-two attending physicians in the Republic of Korea examined 1,750 radiographs of 50 lung cancer cases. Using “hot spot” technology, participants indicated the possible locations of cancer lesions on each radiograph. Subsequently, the same radiographs, cropped to focus the anatomical regions where lung cancers were diagnosed, were shown side-by-side to the participants. The participants were asked to identify the radiograph which first enabled the diagnosis of lung cancer and which first showed a possible lesion.

Results

Removal of systemic constraints alone significantly improved lesion identification by 221.72±9.69 days. Presenting radiographs side-by-side, cropped to relevant areas, had an additional significant and positive impact on cancer detection in both hidden and open areas on CXRs. Also, lesions were detected at smaller sizes and earlier than when actually diagnosed.

Conclusions

CXR with improved methods and settings provides an easily accessible and low-risk imaging method for earlier detection of lung cancer compared to current clinical imaging settings. Further, this study demonstrates the potential effectiveness of programs that allow side-by-side comparisons of cropped areas of multiple radiographs to detect radiological abnormalities.

DNA Methylation-Based Testing in Liquid Biopsies as Detection and Prognostic Biomarkers for the Four Major Cancer Types

Lung, breast, colorectal, and prostate cancers are the most incident worldwide. Optimal population-based cancer screening methods remain an unmet need, since cancer detection at early stages increases the prospects of successful and curative treatment, leading to a lower incidence of recurrences. Moreover, the current parameters for cancer patients’ stratification have been associated with divergent outcomes. Therefore, new biomarkers that could aid in cancer detection and prognosis, preferably detected by minimally invasive methods are of major importance. Aberrant DNA methylation is an early event in cancer development and may be detected in circulating cell-free DNA (ccfDNA), constituting a valuable cancer biomarker. Furthermore, DNA methylation is a stable alteration that can be easily and rapidly quantified by methylation-specific PCR methods. Thus, the main goal of this review is to provide an overview of the most important studies that report methylation biomarkers for the detection and prognosis of the four major cancers after a critical analysis of the available literature. DNA methylation-based biomarkers show promise for cancer detection and management, with some studies describing a “PanCancer” detection approach for the simultaneous detection of several cancer types. Nonetheless, DNA methylation biomarkers still lack large-scale validation, precluding implementation in clinical practice.

Early detection of the major male cancer types in blood-based liquid biopsies using a DNA methylation panel

BACKGROUND

Lung (LC), prostate (PCa) and colorectal (CRC) cancers are the most incident in males worldwide. Despite recent advances, optimal population-based cancer screening methods remain an unmet need. Due to its early onset, cancer specificity and accessibility in body fluids, aberrant DNA promoter methylation might be a valuable minimally invasive tool for early cancer detection. Herein, we aimed to develop a minimally invasive methylation-based test for simultaneous early detection of LC, PCa and CRC in males, using liquid biopsies.

RESULTS

Circulating cell-free DNA was extracted from 102 LC, 121 PCa and 100 CRC patients and 136 asymptomatic donors' plasma samples. Sodium-bisulfite modification and whole-genome amplification was performed. Promoter methylation levels of APC_me, FOXA1_me, GSTP1_me, HOXD3_me, RARβ2_me, RASSF1A_me, SEPT9_me and SOX17_me were assessed by multiplex quantitative methylation-specific PCR. SEPT9_me and SOX17_me were the only biomarkers shared by all three cancer types, although they detected CRC with limited sensitivity. A "PanCancer" panel (FOXA1_me, RARβ2_me and RASSF1A_me) detected LC and PCa with 64% sensitivity and 70% specificity, complemented with "CancerType" panel (GSTP1_me and SOX17_me) which discriminated between LC and PCa with 93% specificity, but with modest sensitivity. Moreover, a HOXD3_me and RASSF1A_me panel discriminated small cell lung carcinoma from non-small cell lung carcinoma with 75% sensitivity, 88% specificity, 6.5 LR+ and 0.28 LR-. An APC_me and RASSF1A_me panel independently predicted disease-specific mortality in LC patients.

CONCLUSIONS

We concluded that a DNA methylation-based test in liquid biopsies might enable minimally invasive screening of LC and PCa, improving patient compliance and reducing healthcare costs. Moreover, it might assist in LC subtyping and prognostication.

Knowledge-based Collaborative Deep Learning for Benign-Malignant Lung Nodule Classification on Chest CT

The accurate identification of malignant lung nodules on chest CT is critical for the early detection of lung cancer, which also offers patients the best chance of cure. Deep learning methods have recently been successfully introduced to computer vision problems, although substantial challenges remain in the detection of malignant nodules due to the lack of large training data sets. In this paper, we propose a multi-view knowledge-based collaborative (MV-KBC) deep model to separate malignant from benign nodules using limited chest CT data. Our model learns 3-D lung nodule characteristics by decomposing a 3-D nodule into nine fixed views. For each view, we construct a knowledge-based collaborative (KBC) submodel, where three types of image patches are designed to fine-tune three pre-trained ResNet-50 networks that characterize the nodules' overall appearance, voxel, and shape heterogeneity, respectively. We jointly use the nine KBC submodels to classify lung nodules with an adaptive weighting scheme learned during the error back propagation, which enables the MV-KBC model to be trained in an end-to-end manner. The penalty loss function is used for better reduction of the false negative rate with a minimal effect on the overall performance of the MV-KBC model. We tested our method on the benchmark LIDC-IDRI data set and compared it to the five state-of-the-art classification approaches. Our results show that the MV-KBC model achieved an accuracy of 91.60% for lung nodule classification with an AUC of 95.70%. These results are markedly superior to the state-of-the-art approaches.

Serum levels of candidate microRNA diagnostic markers differ among the stages of non-small-cell lung cancer

Circulating microRNAs (miRNAs) are promising markers for cancer diagnosis and prognosis. Numerous studies evaluating miRNAs as markers for non-small cell lung cancer (NSCLC) have been conducted in recent years; however, the majority of candidate markers proposed via individual studies were inconsistent and no marker miRNAs for the diagnosis of early stage NSCLC have been established. In the present study, miR-145, miR-20a, miR-21 and miR-223, which were previously reported as candidate diagnostic markers of NSCLC, were re-evaluated. The serum levels of these miRNAs were quantified in 56 patients with stage I-IV NSCLC using the TaqMan microRNA assays and separately compared the levels at each stage with those in 26 control patients. The level of miR-145 was significantly reduced in patients with NSCLC, regardless of clinical stage, and its level increased following tumor resection in patients with stage I-II disease. These results indicate that miR-145 is relevant as a diagnostic marker for stages I-IV NSCLC. Additionally, the levels of miR-20a and miR-21 demonstrated notable differences among patients at different clinical stages. These miRNAs distinguished patients in a number of, but not all, stages of NSCLC from cancer-free control patients. These results indicated that it is essential to analyze miRNA levels at each stage separately in order to evaluate marker miRNAs for NSCLC diagnosis.