A randomized study of fluorescence bronchoscopy versus white-light bronchoscopy for early detection of lung cancer in high risk patients

Omics-based deep learning approaches for lung cancer decision-making and therapeutics development

Lung cancer has been the most common and the leading cause of cancer deaths globally. Besides clinicopathological observations and traditional molecular tests, the advent of robust and scalable techniques for nucleic acid analysis has revolutionized biological research and medicinal practice in lung cancer treatment. In response to the demands for minimally invasive procedures and technology development over the past decade, many types of multi-omics data at various genome levels have been generated. As omics data grow, artificial intelligence models, particularly deep learning, are prominent in developing more rapid and effective methods to potentially improve lung cancer patient diagnosis, prognosis and treatment strategy. This decade has seen genome-based deep learning models thriving in various lung cancer tasks, including cancer prediction, subtype classification, prognosis estimation, cancer molecular signatures identification, treatment response prediction and biomarker development. In this study, we summarized available data sources for deep-learning-based lung cancer mining and provided an update on recent deep learning models in lung cancer genomics. Subsequently, we reviewed the current issues and discussed future research directions of deep-learning-based lung cancer genomics research.

A Review of Deep Learning Techniques for Lung Cancer Screening and Diagnosis Based on CT Images

One of the most common and deadly diseases in the world is lung cancer. Only early identification of lung cancer can increase a patient's probability of survival. A frequently used modality for the screening and diagnosis of lung cancer is computed tomography (CT) imaging, which provides a detailed scan of the lung. In line with the advancement of computer-assisted systems, deep learning techniques have been extensively explored to help in interpreting the CT images for lung cancer identification. Hence, the goal of this review is to provide a detailed review of the deep learning techniques that were developed for screening and diagnosing lung cancer. This review covers an overview of deep learning (DL) techniques, the suggested DL techniques for lung cancer applications, and the novelties of the reviewed methods. This review focuses on two main methodologies of deep learning in screening and diagnosing lung cancer, which are classification and segmentation methodologies. The advantages and shortcomings of current deep learning models will also be discussed. The resultant analysis demonstrates that there is a significant potential for deep learning methods to provide precise and effective computer-assisted lung cancer screening and diagnosis using CT scans. At the end of this review, a list of potential future works regarding improving the application of deep learning is provided to spearhead the advancement of computer-assisted lung cancer diagnosis systems.

Changes in the Heart Rate of Sniffer Dogs Trained for Detection of Lung Cancer

BACKGROUND

Lung carcinoma is one of the most common malignancies worldwide. At present, unfortunately, there are no markers that would allow early identification of this tumor in the preclinical or early clinical stage. The use of sniffer dogs has been reported to show some promise in early diagnosis of this type of cancer Aim: This study aimed to evaluate the possibility of utilizing changes in the heart rate of sniffer dogs (which increases when finding a positive sample) in tumor detection.

METHODS

This double-blinded pilot study included two sniffer dogs. A chest strap was fastened on the dog's chests for heart rate monitoring while they were examining samples and heart rate was recorded. Test parameters (sensitivity, specificity, positive and negative predictive values) were then calculated, evaluating performances based on (i) the dog's indications according to their training and (ii) the changes in their heart rates.

RESULTS

Calculation according to the dog's indications revealed an overall sensitivity of 95.2% accompanied by a specificity of 81.8%, a PPV of 93.7%, and an NPV of 85.7%, respectively. These results were not significantly different from those evaluated by heart rate; heart rate monitoring was, however, burdened with a relatively high proportion of invalid experiments in which heart rate measurement failed. When the method of calculation was changed from rounds to individual samples, the test parameters further increased.

CONCLUSIONS

This pilot study confirmed the hypothesis that heart rate increases in trained sniffer dogs when encountering samples from tumor-positive patients but remains unchanged when only negative samples are present. The reliability of results based on heart rate increase is similar to that obtained by a dog's indications and, if the limitation represented by technical issues is overcome, it could serve as a valuable verification method.

Current and Future Development in Lung Cancer Diagnosis

Lung cancer is the leading cause of cancer-related deaths in North America and other developed countries. One of the reasons lung cancer is at the top of the list is that it is often not diagnosed until the cancer is at an advanced stage. Thus, the earliest diagnosis of lung cancer is crucial, especially in screening high-risk populations, such as smokers, exposure to fumes, oil fields, toxic occupational places, etc. Based on the current knowledge, it looks that there is an urgent need to identify novel biomarkers. The current diagnosis of lung cancer includes different types of imaging complemented with pathological assessment of biopsies, but these techniques can still not detect early lung cancer developments. In this review, we described the advantages and disadvantages of current methods used in diagnosing lung cancer, and we provide an analysis of the potential use of body fluids as carriers of biomarkers as predictors of cancer development and progression.

Molecular predictive factors for progression of high-grade preinvasive bronchial lesions

RATIONALE

The outcome of precancerous bronchial lesions is not well known, and their management is subject to controversy. Many molecular alterations are present in preinvasive lesions, but none has been assessed to predict the evolution of the lesions.

OBJECTIVES

To analyze the outcome of high-grade precancerous lesions according to their molecular profile.

METHODS

Twenty-three severe dysplasia and 31 carcinoma in situ (CIS) lesions in 37 patients were monitored using repeated autofluorescence bronchoscopy over a 12-year period. Microdissection and polymerase chain reaction analysis were performed on paraffin tissue sections to assess loss of heterozygosity (LOH) and microsatellite instability on chromosome 3p, 5q, and 9p. Histology and molecular status at baseline were compared between 7 lesions that became invasive, 11 that relapsed after treatment, 17 that were eradicated with local treatment, and 19 that spontaneously regressed.

MEASUREMENTS AND MAIN RESULTS

Ninety-four percent of lesions that progressed or relapsed were CIS at baseline, whereas 79% of spontaneously regressing lesions were severe dysplasia (P < 0.0001). 3p and 9p LOH was more frequent in CIS than in severe dysplasia (P = 0.03). In the whole group of lesions as well as in the CIS group, 3p LOH was strongly associated with progression (P < 0.0001 and P = 0.02, respectively). Microsatellite instability was not associated with the outcome of the lesions. A therapeutic strategy based on the presence of 3p or 9p LOH would have led to overtreatment of six lesions but would have missed only 1 among the 18 progressing lesions.

CONCLUSIONS

Baseline histology and 3p LOH analysis appear to be useful in predicting the outcome of high-grade precancerous lesions.

Bronchial Intraepithelial Neoplasia/Early Central Airways Lung Cancer

BACKGROUND

An evidence-based approach is necessary for the localization and management of intraepithelial and microinvasive non-small cell lung cancer in the central airways.

METHODS

Material appropriate to this topic was obtained by literature search of a computerized database. Recommendations were developed by the writing committee and then reviewed by the entire guidelines panel. The final recommendations were made by the Chair and were voted on by the entire committee.

RESULTS

White light bronchoscopy has diagnostic limitations in the detection of microinvasive lesions. Autofluorescence bronchoscopy (AFB) is a technique that has been shown to be a sensitive method for detecting these lesions. In patients with moderate dysplasia or worse on sputum cytology and normal chest radiographic findings, bronchoscopy should be performed. If moderate/severe dysplasia or carcinoma in situ (CIS) is detected in the central airways, then bronchoscopic surveillance is recommended. The use of AFB is preferred if available. In a patient being considered for curative endobronchial therapy to treat microinvasive lesions, AFB is useful. A number of endobronchial techniques as therapeutic options are available for the management of CIS and can be recommended to patients with inoperable disease. In patients with operable disease, surgery remains the mainstay of treatment, although patients may be counseled about these techniques.

CONCLUSIONS

AFB is a useful tool for the localization of microinvasive neoplasia. A number of endobronchial techniques available for the curative treatment can be considered first-line therapy in inoperable cases. For operable cases, the techniques may be considered and discussed with the patients.

Optical Spectroscopy for the Classification of Malignant Lesions of the Bronchial Tree

Optical spectroscopy may be used for in vivo, noninvasive distinction of malignant from normal tissue. The aim of our study was to analyze the accuracy of various optical spectroscopic techniques for the classification of cancerous lesions of the bronchial tree. We developed a fiberoptic instrument allowing the measurement of autofluorescence spectroscopy (AFS), diffuse reflectance spectroscopy (DRS), and differential path length spectroscopy (DPS) during bronchoscopy. Spectroscopic measurements were obtained from 191 different endobronchial lesions (63 malignant and 128 nonmalignant) in 107 patients. AFS, DRS, and DPS sensitivity/specificity for the distinction between malignant and nonmalignant bronchial lesions were 73%/82%, 86%/81%, and 81%/88%, respectively. All three optical spectroscopic modalities facilitate an increase of the positive predictive value of autofluorescence bronchoscopy for the detection of endobronchial tumors. Even better results were obtained when the three spectroscopic techniques were combined.

Fluorescent bronchoscopy

Detection of clinically occult lung neoplasms may represent an opportunity for early curative intervention. Fluorescent bronchoscopy is a sensitive technique for detecting early endobronchial tumors that may be combined with CT scanning as part of a comprehensive lung cancer screening program. Identification and longitudinal follow-up of dysplastic endobronchial changes with fluorescent bronchoscopy should facilitate studies of chemoprevention and further knowledge regarding the natural history of these lesions. Analysis of bronchial epithelium with novel techniques such as genomic hybridization and gene expression arrays might provide even better predictors of progression of dysplastic endobronchial lesions.

Gene-promoter hypermethylation as a biomarker in lung cancer

Silencing of genes by aberrant promoter hypermethylation is now recognized as a crucial component in cancer initiation and progression. Highly sensitive assays have been developed to assess gene-promoter methylation in biological fluids. The detection of methylated genes in sputum could lead to the development of a screening test to non-invasively identify early cancer in high-risk people.

The curative potential of intraluminal bronchoscopic treatment for early-stage non-small-cell lung cancer

Bronchoscopic treatment modalities such as lasers, electrocautery, cryotherapy, photodynamic therapy, and brachytherapy are potentially curative for patients with very-early-stage non-small-cell lung cancer (NSCLC) in the central airways. Previously, studies had primarily focused on the effectiveness of surgery, surgical bronchoplasty, and photodynamic therapy. The cure rate of intraluminal bronchoscopic treatment is strongly related to the patient's functional status and tumor stage. Intraluminal tumors are curable bronchoscopically when they are accessible to the fiberoptic bronchoscope, strictly intraluminal, and superficial with visible proximal and distal tumor margins. Early-stage cancer infiltrating deeper into the bronchial wall may already harbor metastases to the regional lymph nodes; hence, curative intraluminal treatment is not feasible. The use of new diagnostic tools (eg, high-resolution computed tomography, autofluorescence bronchoscopy, and endobronchial ultrasound) may improve staging to select the category of patients in whom intraluminal bronchoscopic therapy with curative intent is appropriate. An accurate intraluminal tumor staging will improve our ability to exploit the curative potential of many bronchoscopic techniques for complete tumor eradication in patients with very-early-stage intraluminal NSCLC in their central airways. The use of bronchoscopic treatment as a less morbid alternative than surgical resection will benefit patients most when tumor is detected at the earliest stage possible.

Cigarette smoking, preinvasive bronchial lesions, and autofluorescence bronchoscopy

STUDY OBJECTIVES

Autofluorescence bronchoscopy (AFB), when used as an adjunct to standard white light bronchoscopy (WLB), enhances the bronchoscopist's ability to localize small neoplastic lesions, especially intraepithelial lesions. The current study was undertaken in order to define the population in which the rate of detection is higher using AFB.

DESIGN AND PATIENTS

Two hundred forty-four consecutive patients, who were symptomatic smokers or patients who previously had been treated for lung cancer or head and neck cancers, underwent WLB and AFB. All patients with endoscopic abnormalities underwent biopsies. Data concerning smoking history were prospectively registered.

RESULTS

We report the prevalence of high-grade or invasive lesions at the time of examination. On a lesion-by-lesion analysis, 92 low-grade lesions, 42 high-grade lesions (ie, moderate dysplasia, severe dysplasia, and carcinoma in situ), and 39 invasive carcinomas were diagnosed. There was no effect of age, gender, and age at smoking initiation on the prevalence of preinvasive or invasive lesions. The 10 patients who previously had undergone surgery for lung cancer and exhibited high-grade preinvasive lesions had a history of carcinoma of the epidermoid histologic type (p = 0.01). These 10 patients displayed multiple lesions in the bronchial tree (mean No. of lesions, 1.8 per patient). In current smokers, the prevalence of high-grade or invasive lesions were both related to the number of pack-years smoking had occurred (p = 0.01) and to the duration of smoking (p = 0.01). In contrast, the prevalence of preinvasive lesions in former smokers was related to a history of epidermoid carcinoma.

CONCLUSIONS

AFB should be recommended in patients with a history of epidermoid carcinomas of the lung. Current smokers with a prolonged smoking history appear to comprise a population in which the rate of detection of preneoplastic lesions is high with AFB.

Fluorescence bronchoscopy for early detection of lung cancer: a clinical perspective

The conventional method of bronchoscopy has only a 30% sensitivity to detect early stage cancer in the central airways. For patients with positive sputum cytology who clearly harbor early cancers, repeat and lengthy sessions of bronchoscopies are required for accurate localization of these lesions. This leads to a significant delay in obtaining the diagnosis, postponing an appropriate treatment and reduces the chance for cure. There are valid reasons for improving the detection rate of early stage lung cancers. The number of individuals at risk forms a large population, the outcome of patients treated with early stage cancer has been shown to be better and bronchoscopic treatments, e.g. photodynamic therapy and electrocautery, are currently alternatives for surgical resection. Finding more early stage cancers by screening the population at risk and accurate staging to enable treatment at the earliest stage feasible, may improve the dismal prognosis of many patients. This article deals with the clinical background and current problems in early detection of lung cancer and discusses our expectations regarding new developments in bronchoscopy for early detection, accurate staging and treatment of lung cancer.

Follow-up of bronchial precancerous lesions and carcinoma in situ using fluorescence endoscopy

Little is known about the natural history of precancerous bronchial lesions. Histological changes occurring in 416 bronchial intraepithelial lesions (104 high-risk subjects) were assessed over a 2-yr period, using repeated follow-up autofluorescence endoscopies. During the study, 6 of 36 normal epitheliums became dysplastic; 47 of 152 metaplasia evolved to low-grade dysplasia, two progressed to carcinoma in situ, and one to invasive cancer; 6 of 169 low-grade epithelial lesions progressed to a persistent severe dysplasia; 10 of 27 severe dysplastic lesions and 28 of 32 carcinoma in situ persisted or progressed, respectively (p = 0.0005, severe dysplasia versus carcinoma in situ 24 mo outcome). Carcinoma in situ appeared more frequent in patients with a prior history or concomitant cancer (p = 0.003). Persistence of smoking during the study did not influence high-grade lesion outcome. Progression of low-grade epithelial lesions during the study occurred only in patients with at least a high-grade lesion in another site at baseline (9 of 147 lesions, 6.1%). Our study suggests that low-grade epithelial lesions could be safely followed-up at 2 yr in patients without high-grade lesions at baseline, whereas severe dysplasia should be treated if they persist at 3 mo. Immediate treatment of carcinoma in situ appears warranted.

Review of recent advances in fluorescence bronchoscopy in early localization of central airway lung cancer

Centrally located lung cancers are radiologically occult until so far advanced as to have a low cure rate or require extensive resection for cure, but at a cost of high morbidity. These cancers represent about one-fifth of new lung cancers. Autofluorescence bronchoscopy appears to be an important tool in localizing premalignant and early malignant lesions in the large central airways, particularly when applied to high-risk patients. Applications include studies of molecular biology of premalignancy and early malignancy, chemoprevention studies, endobronchial therapy studies, localization of synchronous tumors, estimation of the extent of field cancerization, and better estimation of resection margins. Autofluorescence bronchoscopy appears to be significantly more sensitive than white light examination but has low specificity. This technology is likely to gain widespread use when evaluation of sputum for malignant changes is both more sensitive and specific, and when its application is demonstrated to reduce mortality in this important subgroup of non-small cell lung cancer patients.