Clinical Staging Of Lung Cancer Research Articles

Automatic Extraction of Lung Cancer Staging Information From Computed Tomography Reports: Deep Learning Approach.

BackgroundLung cancer is the leading cause of cancer deaths worldwide. Clinical staging of lung cancer plays a crucial role in making treatment decisions and evaluating prognosis. However, in clinical practice, approximately one-half of the clinical stages of lung cancer patients are inconsistent with their pathological stages. As one of the most important diagnostic modalities for staging, chest computed tomography (CT) provides a wealth of information about cancer staging, but the free-text nature of the CT reports obstructs their computerization.ObjectiveWe aimed to automatically extract the staging-related information from CT reports to support accurate clinical staging of lung cancer.MethodsIn this study, we developed an information extraction (IE) system to extract the staging-related information from CT reports. The system consisted of the following three parts: named entity recognition (NER), relation classification (RC), and postprocessing (PP). We first summarized 22 questions about lung cancer staging based on the TNM staging guideline. Next, three state-of-the-art NER algorithms were implemented to recognize the entities of interest. Next, we designed a novel RC method using the relation sign constraint (RSC) to classify the relations between entities. Finally, a rule-based PP module was established to obtain the formatted answers using the results of NER and RC.ResultsWe evaluated the developed IE system on a clinical data set containing 392 chest CT reports collected from the Department of Thoracic Surgery II in the Peking University Cancer Hospital. The experimental results showed that the bidirectional encoder representation from transformers (BERT) model outperformed the iterated dilated convolutional neural networks-conditional random field (ID-CNN-CRF) and bidirectional long short-term memory networks-conditional random field (Bi-LSTM-CRF) for NER tasks with macro-F1 scores of 80.97% and 90.06% under the exact and inexact matching schemes, respectively. For the RC task, the proposed RSC showed better performance than the baseline methods. Further, the BERT-RSC model achieved the best performance with a macro-F1 score of 97.13% and a micro-F1 score of 98.37%. Moreover, the rule-based PP module could correctly obtain the formatted results using the extractions of NER and RC, achieving a macro-F1 score of 94.57% and a micro-F1 score of 96.74% for all the 22 questions.ConclusionsWe conclude that the developed IE system can effectively and accurately extract information about lung cancer staging from CT reports. Experimental results show that the extracted results have significant potential for further use in stage verification and prediction to facilitate accurate clinical staging.

Concordance of Clinical and Pathologic Nodal Staging in Resectable Lung Cancer

Clinical staging of lung cancer may not reliably predict nodal disease, and its accuracy in The Society of Thoracic Surgeons General Thoracic Surgery Database is not described. Among anatomic pulmonary resections for stages I to III lung cancer with complete clinical and pathologic staging (2012-2017), the accuracy of invasive mediastinal staging (IMS) was compared with noninvasive mediastinal staging only. Accuracy, defined as concordance between clinical and pathologic nodal status, was examined using logistic regression to determine factors associated with clinical nodal (cN) accuracy. Variation in accuracy across centers was recorded and categorized. We included 39,516 patients with stages I to III pulmonary cancer (adenocarcinoma, 66%; squamous, 23%; neuroendocrine, 5%; mixed, 3.3%; other, 2.4%), of whom 90.4% had cN0 disease. IMS was performed in 32.4%. The IMS group had more central tumors (14.8% vs 6.0%, P < .001) and cN1-2 (15.7% vs 6.8%, P < .001). Nodal accuracy was 79.8%. Although IMS had a lower nodal accuracy for cN0-2 disease (74.6% vs 82.6%, P<.001), IMS had higher accuracy when comparing patients with cN1-2 disease (53.9% vs 46.9%, P < .001). In multivariable analysis central tumors (odds ratio, 0.47; 95% confidence interval, 0.43-0.51) and >cN0 disease (odds ratio, 0.25; 95% confidence interval, 0.22-0.29) were associated with lower accuracy. Accuracy of IMS in the top 20 centers was 94.4% and in the bottom 20, 70.9%. Staging accuracy in lung cancers selected for initial resection declines with >cN0 and central tumors. Noninvasive staging in tumors without cN involvement misses nearly 20% of cN1-2. Center-specific accuracy is a target for quality improvement.

CT-defined Visceral Pleural Invasion in T1 Lung Adenocarcinoma: Lack of Relationship to Disease-Free Survival.

BackgroundPathologic visceral pleural invasion (pVPI) leads to upstaging from T1 to T2. However, it is unclear whether the CT features for pVPI can be reliably used as a clinical T2 descriptor for preoperative staging.PurposeTo validate the diagnostic accuracy and analyze the prognostic value of CT findings for the prediction of pVPI in patients with resected node-negative lung adenocarcinoma.Materials and MethodsThis retrospective cohort study included clinical T1N0M0 adenocarcinomas resected between 2009 and 2015. The diagnostic CT findings suggestive of pVPI were evaluated by a thoracic radiologist. The accuracy of diagnostic CT findings in relation to pVPI and accuracy for disease-free survival (DFS) were evaluated by using test performance metrics and multivariable Cox regression analysis, respectively.ResultsThe authors analyzed 695 patients (median age, 63 years; 411 women). Data for pVPI were not available in six patients. The accuracy of CT features for pVPI ranged from 62.7% (432 of 689 patients) to 72.3% (498 of 689 patients). Positive predictive values ranged from 44.1% (173 of 392 patients) to 56.4% (88 of 156 patients), which indicated that about half of the CT-based predictions were false-positive. Multivariable Cox regression models showed that none of the combinations of CT findings were independent predictors of DFS (adjusted hazard ratios, 1.40, 1.48, 1.06, and 1.21 for each combination; P > .05 for all). In addition, pVPI was not an independent prognostic factor (adjusted hazard ratio, 1.27; P = .26), whereas age and clinical T category were independent prognostic factors in all Cox models (P < .05 for all).ConclusionCT features of pathologic visceral pleural invasion (pVPI) have an accuracy of 62.7%-72.3%. CT features of pVPI were not independent prognostic factors for disease-free survival in clinical T1 lung adenocarcinomas. This argues against the use of CT features of visceral pleural invasion as T2 descriptors in the clinical staging of lung cancer.© RSNA, 2019Online supplemental material is available for this article.See also the editorial by Nishino in this issue.

Impact of health care organization on surgical lung cancer care

ObjectivesOrganization and governance of national healthcare might play an important role in decision-making and outcomes in patients with lung cancer. Both Denmark and the Netherlands have a high level of healthcare but a different financial coverage, governance and level of centralization. By using both national databases we analyzed the consequences of these differences on patterns of care and outcomes with a focus on morbidity, mortality and clinical staging. Materials and methodsGeneral numbers on both healthcare systems were requested. All patients who had surgery for lung cancer from 2013 to 2016 were included. Mortality, morbidity and clinical staging were analyzed for patients with NSCLC without metastases, only one operation and no neo-adjuvant therapy. ResultsIn 2016 annual budget as share of gross national product was 10.4% for both countries. In Denmark 4 hospitals performed lung surgery in 2016, compared to 43 hospitals in the Netherlands. We included 4030 Danish and 8286 Dutch patients. In the subgroup 30-day mortality was 1.5% in Denmark compared to 1.9% in the Netherlands. The percentage of patients with a complicated course was 24.4% and 34.8% respectively (p < 0.05). Accuracy between cTNM and pTNM was 53.0% in Denmark and 52.9% in the Netherlands. ConclusionSurgery for lung cancer is at a high level in both countries, reflected by low mortality-rates. Centralization has been implemented successfully in Denmark, which might explain the lower rate of patients with a complicated post-operative course, although different definitions preclude firm conclusions. In both countries correct clinical staging of lung cancer remains a challenge.

A promising tool for T-staging of lung cancer: Is convex probe endobronchial ultrasound superior to computed tomography in detecting invasion of mediastinal and hilar vessels?

In the present study, we aimed to compare performance of convex probe endobronchial ultrasound and computed tomography in detecting vascular invasion of mediastinal and hilar lesions. Medical data of a total of 55 patients (47 males, 8 females; mean age 59.6±7.7 years; range, 29 to 76 years) who underwent convex probe endobronchial ultrasound for diagnosis and staging of lung cancer in a tertiary care hospital between May 2016 and December 2017 were retrospectively analyzed. The presence of vascular invasion was determined according to two main criteria: visualization of the tumor tissue within the vessel lumen and loss of vessel-tumor hyperechoic interface. All available contrast enhanced computed tomography images were retrospectively re-evaluated by a blinded radiologist. The intra-rater agreement between convex probe endobronchial ultrasound and computed tomography was analyzed. The sensitivity, specificity, positive and negative predictive values, and accuracy of both modalities were calculated. A total of 65 vessel-tumor interface areas of 55 patients were analyzed. Almost all mediastinal and hilar vascular structures including pulmonary arteries and veins, aorta, superior vena cava and its branches, and left atrium with pulmonary veno-atrial junctions could be easily assessed by convex probe endobronchial ultrasound. The intra-agreement of both modalities in detecting vascular invasion was k=0.268 (p=0.028). In nine patients with a surgical confirmation, the sensitivity, specificity, positive and negative predictive values, and accuracy values were 100%, 33.3%, 75.0%, 100%, and 77.7%, respectively for convex probe endobronchial ultrasound and 66.6%, 33.3%, 66.6%, 33.3%, and 55.5%, respectively for computed tomography. Convex probe endobronchial ultrasound can be used to detect vascular invasion alone or in conjunction with contrast-enhanced computed tomography. Hence, a T4 lesion would be better differentiated from T3 in clinical staging of lung cancer.

Automatic Detection and Staging of Lung Tumors using Locational Features and Double-Staged Classifications

Lung cancer is a life-threatening disease with the highest morbidity and mortality rates of any cancer worldwide. Clinical staging of lung cancer can significantly reduce the mortality rate, because effective treatment options strongly depend on the specific stage of cancer. Unfortunately, manual staging remains a challenge due to the intensive effort required. This paper presents a computer-aided diagnosis (CAD) method for detecting and staging lung cancer from computed tomography (CT) images. This CAD works in three fundamental phases: segmentation, detection, and staging. In the first phase, lung anatomical structures from the input tomography scans are segmented using gray-level thresholding. In the second, the tumor nodules inside the lungs are detected using some extracted features from the segmented tumor candidates. In the last phase, the clinical stages of the detected tumors are defined by extracting locational features. For accurate and robust predictions, our CAD applies a double-staged classification: the first is for the detection of tumors and the second is for staging. In both classification stages, five alternative classifiers, namely the Decision Tree (DT), K-nearest neighbor (KNN), Support Vector Machine (SVM), Ensemble Tree (ET), and Back Propagation Neural Network (BPNN), are applied and compared to ensure high classification performance. The average accuracy levels of 92.8% for detection and 90.6% for staging are achieved using BPNN. Experimental findings reveal that the proposed CAD method provides preferable results compared to previous methods; thus, it is applicable as a clinical diagnostic tool for lung cancer.

Lung cancer staging: a concise update.

Diagnosis and clinical staging of lung cancer are fundamental to planning therapy. The techniques for clinical staging, i.e anatomic and metabolic imaging, endoscopies and minimally invasive surgical procedures, should be performed sequentially and with an increasing degree of invasiveness. Intraoperative staging, assessing the magnitude of the primary tumour, the involved structures, and the loco-regional lymphatic spread by means of systematic nodal dissection, is essential in order to achieve a complete resection. In resected tumours, pathological staging, with the systematic study of the resected specimens, is the strongest prognostic indicator and is essential to make further decisions on therapy. In the present decade, the guidelines on lung cancer staging of the American College of Chest Physicians and the European Society of Thoracic Surgeons are based on the best available evidence and are widely followed. Recent advances in the classification of the adenocarcinoma of the lung, with the definition of adenocarcinoma in situ, minimally invasive adenocarcinoma and lepidic predominant adenocarcinoma, and the publication of the eighth edition of the tumour, node and metastasis classification of lung cancer, have to be integrated into the staging process. The present review complements the latest guidelines on lung cancer staging by providing an update of all these issues.

The ATS/ERS/JRS/ALAT statement "IPF by HRCT" could predict acute exacerbation of interstitial lung disease in non-small cell lung cancer.

Patients with non-small cell lung cancer (NSCLC) and interstitial lung disease (ILD) are at high risk of acute exacerbation of ILD (AE-ILD) when treated with systemic chemotherapy. Standard treatment for NSCLC complicated by ILD has not been established. To examine whether the type of ILD categorized by the official ATS/ERS/JRS/ALAT statement as "idiopathic pulmonary fibrosis (IPF) by high-resolution computed tomography (HRCT)" could predict chemotherapy-induced AE-ILD in NSCLC patients with ILD, we retrospectively reviewed all patients with NSCLC complicated by ILD who had received chemotherapy at our institute from January 2007 until December 2013. Patients' characteristics, pathology and clinical staging of lung cancer, chemotherapy, type of ILD and AE-ILD during chemotherapy were evaluated. ILD was classified according to the statement as follows: usual interstitial pneumonia (UIP), possible UIP, and inconsistent with a UIP pattern. A total of 46 patients had pre-existing ILD and received chemotherapy. The mean age was 73 years (range 46-83 years). Fifteen (32.6%) of 46 patients with ILD developed chemotherapy-induced AE-ILD, which was seen more frequently in patients with ILD with a UIP pattern or possible UIP pattern than in patients with a pattern inconsistent with UIP (80% versus 9.7%, p&lt;0.001). Multivariate analyses including age, sex, performance status and radiographic patterns of ILD showed that the presence of a UIP or possible UIP pattern was an independent risk factor for chemotherapy-induced AE-ILD. ILD with a UIP pattern or possible UIP pattern by the classification could be a risk factor for AE-ILD in NSCLC patients with ILD.

Analiza zmian opisywanych w niskodawkowej tomografii komputerowej klatki piersiowej w populacji mężczyzn o podwyższonym ryzyku zachorowania na raka płuca

Background. Lung cancer was a leading cause of all cancer-related deaths among men and was strongly related to long-term tobacco smoking. Clinical staging of lung cancer was one of the most important prognostic factors: the higher prevalence of early detection of lung cancer, the better results of lung cancer treatment. The purpose of the study was to present morphological characteristics and additional findings in LDCT in males with a high risk of lung cancer. Methods. We included 900 male subjects, residents in the Krakow area and its suburbs, who were current and former smokers between 50 and 74 years of age. Risk factors such as age, duration of tobacco use, exposure to asbestos, family history of non-malignant lung diseases, cancer and current respiratory symptoms were analysed for all participants. Pulmonary nodules detected in all participants in low-dose chest CT were classified into non-malignant, indeterminate and potentially malignant and all were classified into one of three groups of low (A), indeterminate (B) and high risk of lung cancer (C). Additional pulmonary findings detected in LDCT were also analysed. Results . In 831 of subjects (92.3%) 4575 pulmonary nodules were detected, among which 66.6% (n = 3048) were solid, non-calcified. 47 potentially malignant nodules were revealed (1.5% of all solid non-calcified nodules). The oldest participants, who had the longest smoking duration were males with potentially malignant nodules. Additional pulmonary findings were revealed significantly more often among the high lung cancer risk group. Among 39 participants with potentially malignant pulmonary nodules, lung cancer was detected in 17.9% (n = 7) participants (0.78% of the study group). Conclusions. Low dose chest computed tomography revealed 94.6% specificity. Statistically significant differences among participants from the high lung cancer risk group were found for median age, number of packyears, positive history of bronchitis, presence of emphysema, bronchiectasis, bronchial wall thickening, ground glass opacities and consolidations. Analyses of risk factors revealed statistically significant differences among those in a subgroup of histologically proven lung cancer, for median age over 60 years, duration of cigarettes smoking over 45 years (with number of packyears at least at the level of 37), mediastinal lymphadenopathy (especially of high grade) and GGOs (grounds glass opacities).

Marked Hematopoiesis Masquerading Multiple Bone Metastasis in a Lung Cancer Patient With Myelodysplastic Syndrome

We diagnosed a 62-year-old man with lung cancer, well-differentiated squamous cell carcinoma, on the right upper lobe and left lower lobe. He had a history of myelodysplastic syndrome and refractory anemia with ringed sideroblasts diagnosed 5 years earlier. The patient's ¹⁸F-FDG PET/CT for the clinical staging of lung cancer revealed accumulations of FDG in multiple bones including the rib bones, which strongly suggested multiple bone metastases. He underwent lobectomies and excision of the right fourth and fifth rib bones. The pathological findings of resected rib bones exhibited hypercellular bone marrow without excess of blasts.

Clinico-pathological profile of lung cancer in Uttarakhand.

Background:Lung cancer is one of the most aggressive and prevalent type of malignancy causing high morbidity and mortality. Tobacco smoking continues to be the leading cause of lung cancer worldwide. An increasing incidence of lung cancer has been observed in India.Objective:The aim of this study was to evaluate the clinico, a pathological profile of the lung cancer in hilly state of Uttrakhand.Materials and Methods:We performed a retrospective analysis of histopathologically proven cases of bronchogenic carcinoma admitted in our hospital from January 1998 to August 2005.Results:Our study included 203 patients with confirmed cases of lung cancer. Male to female ratio was 8.2:1. The common age group being 40-60 years, 9.86% of the patients were less than 40 years old age. Smoking was found to be the main risk factor in 81.77% patients. The most frequent symptom was cough (72.90%) followed by fever (58.12%). The most common radiological presentation was mess lesion (46.31%). The most common histopathological type was squamous cell carcinoma (SCC) (44.83%) followed by adenocarcinoma (19.78%) and small cell lung carcinoma (SCLC) (16.75%). The majority patients (73.29%) were diagnosed in the later stages of the disease (III B and IV).Conclusion:It was found out that SCC was the most frequent histopathological form. SCLC predominates below 40 year and SCC over 60 years of age. Smoking still remains the major risk factors in pathogenesis of lung cancer.

Supermediastinoscopies: A Step Forward in Lung Cancer Staging

Supermediastinoscopies: A Step Forward in Lung Cancer Staging

“A System for the Clinical Staging of Lung Cancer”—A Commentary

“A System for the Clinical Staging of Lung Cancer”—A Commentary

Analysis of the relationship between polymorphism of angiotensin-converting enzyme gene and lung cancer

Angiotensin-converting enzyme (ACE), the key enzyme of RAS, is expressed in abundance in lung. Recent studies show that it plays various biological roles and even contributes importantly to the carcinogenesis of malignant tumors. The aim of this study is to investigate whether the insertion (I)/deletion (D) polymorphism of ACE is related to the susceptibility, pathological type or staging of lung cancer. DNA was extracted from the peripheral veinous blood derived from 47 lung cancer patients and 54 normal persons. The ACE genotype of each sample was amplified with PCR. The frequencies of II, ID and DD genotype of ACE in lung cancer group were 0.447, 0.447 and 0.106 respectively, and the frequencies of I and D allele were 0.670 and 0.330 respectively. The frequencies of II, ID and DD geno-(type) of ACE in control group were 0.370, 0.556 and 0.074 respectively, and the frequencies of I and D allele were 0.648 and 0.352 respectively. There was no significant difference of ACE genotypes or alleles frequencies between the lung cancer and control groups. No significant difference of ACE genotypes or alleles frequencies was found between small cell lung cancer and non-small cell lung cancer groups or between the different staging groups of lung cancer patients. No association is found between the polymorphism of ACE gene and the susceptibility, pathological type or clinical staging of lung cancer.

Noninvasive clinical staging modalities for lung cancer.

Clinical staging of lung cancer helps to determine the extent of disease and stratify patients into similar therapeutic and prognostic categories. A primary goal of clinical staging is to separate patients with potentially resectable disease from those that are unresectable. Initial assessment of the patient by history and physical examination combined with laboratory values can suggest metastatic spread of the disease. When abnormal, these clinical factors may have value in terms of predicting prognosis, but their use in early stage lung cancer is limited because of the low prevalence of symptoms, physical exam findings, and laboratory abnormalities in this group. For clinical staging, patients almost always undergo a postero-anterior and lateral chest radiograph and a computed tomography (CT) scan of the chest and upper abdomen to include the liver and adrenal glands. Although CT scanning provides exquisite anatomic information, it is less than optimal for determining lymph node status. Over the last several years, CT scanning combined with positron-emission tomography (PET) using fluorodeoxyglucose (FDG) has significantly improved the accuracy of clinical staging. The use of FDG-PET continues to be defined in the non-invasive evaluation of the primary tumor, nodal involvement, and metastatic disease. Despite the recent advancements in radiologic assessment of lung cancer, invasive sampling is still often performed for pathologic confirmation.

Endoscopic Ultrasound Examination for Mediastinal Lymph Node Metastases of Lung Cancer

Among patients with primary lung cancer who were admitted to the National Cancer Center Hospital from July 1987 to April 1988 for surgical treatments, 132 underwent preoperative transesophageal endoscopic ultrasound examination (EUS) on mediastinal lymph nodes. Of the 132 patients, 101 were pathologically evaluated and studied in this article. A GF-UM2 radial scanner with 7.5-MHz (Olympus Co Ltd) was used for image examination. The lymph nodes were diagnosed as positive for metastasis when they had thickened images, clear contours, and low echoing images of fusion or lobulation. The results obtained from 509 sites were as follows: sensitivity, 53.6 percent; specificity, 97.5 percent; positive predictive accuracy, 77.1 percent; negative predictive accuracy, 93.1 percent; and overall accuracy, 91.6 percent. The sensitivity rate was 80.6 percent excluding the result of the right superior mediastinal lymph nodes that were difficult to examine for anatomic reasons. Although EUS was considered to be an excellent method in diagnosing lymph node metastases, it had a blind angle in the field. More accurate diagnoses of mediastinal lymph node metastases could be achieved by using EUS and computed tomography (CT) together.

Clinical staging of lung cancer: Mediastinoscopy, pleuroscopy or computed tomography?

The roles of mediastinoscopy and pleuroscopy in the clinical staging of lung cancer are discussed. Mediastinoscopy enables determination of metastatic spread to the mediastinal lymph nodes with great accuracy. Until recently it was indicated in the preoperative evaluation of nearly all patients with resumably resectable bronchogenic carcinoma. This indication became less common due to recent developments in the technology of computed tomography that enable non-invasive determination of mediastinal lymphadenopathy. Pleuroscopy has not been used as extensively as mediastinoscopy or computed tomography. As a simple method of pleural exploration, it enables easy determination of pleural extension of the tumor, either by direct invasion or by metastatic dissemination. Although not recommended for routine use, it is indicated in patients with suspected pleural extension of the neoplasm. These methods are complimentary rather than competitory.

The Peripheral Lymphocyte Count as an Aid in the Clinical Staging of Lung Cancer

The association between disease staging and peripheral lymphocyte count (PLC) was studied in 178 patients with histologically verified lung cancer. Mean PLC varied inversely with the stage of disease (p less than .01). The mean values were 2821, 2217, and 1594 for Stages I, II and III respectively. Thoretical normal distributions were generated about the mean peripheral lymphocyte count for each stage. Relative frequencies of lung cancer staging based on the PLC are estimated utilizing these curves.

A system for the clinical staging of lung cancer.

A SYSTEM FOR THE CLINICAL STAGING OF LUNG CANCERCLIFTON F. MOUNTAIN, M.D., DAVID T. CARR, M.D. and W. A. D. ANDERSON, M.D.Audio Available | Share