Development and validation of a decision tree for distinguishing pulmonary adenocarcinomas with mucinous features and metastatic colorectal adenocarcinoma.

BACKGROUND: Diagnosis of mucinous carcinomas in the lung on transbronchial biopsy or fine-needle aspiration (FNA) samples can be difficult for the pathologist, because primary and metastatic tumors can have similar morphological, immunohistochemical, and molecular characteristics. Correct diagnosis is key to determine appropriate therapy and to distinguish primary from metastatic disease. This distinction often falls to the pathologist in patients with a history of mucinous adenocarcinoma of the colon. Despite its drawbacks, immunohistochemistry is often employed to help assign a primary site for mucinous adenocarcinomas in the lung. However, the published data in this regard is limited to studies that use only a handful of markers. METHODS: The authors examined the staining characteristics and heterogeneity of CK7, TTF-1, NapsinA, CK20, CDX2, and SATB2 in resection specimens of pulmonary adenocarcinomas with mucinous features and metastatic colorectal adenocarcinoma. RESULTS: Based on the heterogeneity, sensitivity, and specificity in this cohort, the authors developed a decision tree based on TTF-1, SATB2, CDX2, and CK7 to categorize tumors as primary or metastatic lesions. Validation of the decision tree in FNA specimens from the lungs and lung-draining lymph nodes showed 84% concurrence in cases from the lung and 100% concurrence in cases from the lymph node. In cases where the algorithm assigned a primary site, it was 95% accurate compared to the multidisciplinary diagnosis. CONCLUSIONS: This method holds promise in distinguishing primary versus metastatic lesions in resection, biopsy, and FNA samples from the lungs.

Nucleic Acid Quality Assessment is Critical to the Success of the Oncomine Dx Target Test for Lung Cancer.

BACKGROUND AND OBJECTIVE: The Oncomine Dx Target Test (ODxTT) has been used as a companion diagnostic test for lung cancer. Here, we evaluated whether the amount of nucleic acid and the degree of RNA degradation are related to the success of the ODxTT. METHODS: This study included 223 samples from 218 patients with lung cancer. For all samples, DNA and RNA concentrations were quantified using Qubit, and the degree of RNA degradation was evaluated using the Bioanalyzer. RESULTS: Of the 223 samples, 219 samples were successfully analyzed in the ODxTT and four were not. DNA analysis failed in two samples, which were attributed to low DNA concentrations and both were cytology specimens. Meanwhile, RNA analysis failed in the other two samples. These samples had sufficient amounts of RNA, but it was highly degraded with DV200 (the percentage of RNA fragments > 200 base pairs) less than 30. Compared with RNA samples with DV200 ≥ 30, analysis of RNA with DV200 < 30 yielded significantly fewer reads for the internal control genes. This test showed actionable mutations were identified in 38% (83/218) of all patients and in 46.6% (76/163) of patients with lung adenocarcinoma. CONCLUSIONS: DNA concentration and degree of RNA degradation are key factors determining the success of diagnostic testing by the ODxTT.

Economic Analysis of Exclusionary EGFR Test Versus Up-Front NGS for Lung Adenocarcinoma in High EGFR Mutation Prevalence Areas.

BACKGROUND: This study sought to determine whether exclusionary EGFR mutation testing followed by next-generation sequencing (NGS) is a cost-efficient and timely strategy in areas with high prevalence rates of EGFR mutation. METHODS: We developed a decision tree model to compare exclusionary EGFR testing followed by NGS and up-front NGS. Patients entered the model upon diagnosis of metastatic lung adenocarcinoma. Gene alterations with FDA-approved targeted therapies included EGFR, ALK, ROS1, BRAF, RET, MET, NTRK, and KRAS. Model outcomes were testing-related costs; time-to-test results; monetary loss, taking both costs and time into consideration; and percentage of patients who could be treated by FDA-approved therapies. Stacked 1-way and 3-way sensitivity analyses were performed. RESULTS: Exclusionary EGFR testing incurred testing-related costs of US $1,387 per patient, a savings of US $1,091 compared with the costs of up-front NGS. The time-to-test results for exclusionary EGFR testing and up-front NGS were 13.0 and 13.6 days, respectively. Exclusionary EGFR testing resulted in a savings of US $1,116 in terms of net monetary loss, without a reduction of patients identified with FDA-approved therapies. The EGFR mutation rate and NGS cost had the greatest impact on minimizing monetary loss. Given that the tissue-based NGS turnaround time was shortened to 7 days, up-front NGS testing would become the best strategy if its price could be reduced to US $568 in Taiwan. CONCLUSIONS: In areas with high prevalence rates of EGFR mutation, exclusionary EGFR testing followed by NGS, rather than up-front NGS, is currently a cost-efficient strategy for metastatic lung adenocarcinoma.

A novel immune-related long non-coding RNAs risk model for prognosis assessment of lung adenocarcinoma.

BACKGROUND: The abundant immune-related long non-coding RNA (IRLNRs) in immune cells and immune microenvironment have the potential to forecast prognosis and evaluate the effect of immunotherapy. IRLNRs analysis will provide a new perspective for LUAC research. METHODS: We calculated the immune score of each sample according to the expression levels of immune-related genes (IRGs) and screened the survival-related IRLNRs (sIRLNRs) by Cox regression analysis. The expression levels of AC068338.3 and AL691432.2 in tissues and cell lines were confirmed by RT-qPCR. RESULTS: 36 IRLNRs were selected by Pearson correlation analysis. Ten sIRLNRs were significantly correlated with the clinical outcomes of LUAC patients. Five sIRLNRs were identified by multivariate COX regression analysis to establish the immune-related risk score model (IRRS). The overall survival (OS) in the high-risk group was shorter than that in the low-risk group. IRRS could be an independent prognostic factor with significant survival correlation The distributions of immune gene concentrations were different between high-risk group and low-risk group. Furthermore, we further verified that the expression levels of AC068338.3 and AL691432.2 in different LUAC cell lines and tumor tissues were lower than that in Human bronchial epithelial cell (HBE) and adjacent tissues respectively. The lower expression levels of AC068338.3 and AL691432.2 were detected with the more advance T-stages. CONCLUSIONS: Our results highlighted some sIRLNRs with significant clinical correlations and demonstrated their monitored and prognostic values for LUAC patients. The results of this study may provide a new perspective for immunological research and immunotherapy strategies.

Using DNA sequencing data to quantify T cell fraction and therapy response.

The immune microenvironment influences tumour evolution and can be both prognostic and predict response to immunotherapy1,2. However, measurements of tumour infiltrating lymphocytes (TILs) are limited by a shortage of appropriate data. Whole-exome sequencing (WES) of DNA is frequently performed to calculate tumour mutational burden and identify actionable mutations. Here we develop T cell exome TREC tool (T cell ExTRECT), a method for estimation of T cell fraction from WES samples using a signal from T cell receptor excision circle (TREC) loss during V(D)J recombination of the T cell receptor-α gene (TCRA (also known as TRA)). TCRA T cell fraction correlates with orthogonal TIL estimates and is agnostic to sample type. Blood TCRA T cell fraction is higher in females than in males and correlates with both tumour immune infiltrate and presence of bacterial sequencing reads. Tumour TCRA T cell fraction is prognostic in lung adenocarcinoma. Using a meta-analysis of tumours treated with immunotherapy, we show that tumour TCRA T cell fraction predicts immunotherapy response, providing value beyond measuring tumour mutational burden. Applying T cell ExTRECT to a multi-sample pan-cancer cohort reveals a high diversity of the degree of immune infiltration within tumours. Subclonal loss of 12q24.31-32, encompassing SPPL3, is associated with reduced TCRA T cell fraction. T cell ExTRECT provides a cost-effective technique to characterize immune infiltrate alongside somatic changes.

Identification and assessment of PLK1/2/3/4 in lung adenocarcinoma and lung squamous cell carcinoma: Evidence from methylation profile.

Lung cancer is a very aggressive cancer characterized with molecular heterogeneities in different subtypes, including lung adenocarcinoma and lung squamous cell carcinoma. However, few related molecular signatures have been established for the treatment of lung cancer subtypes. Polo-like kinase (PLK) family is a crucial regulator during cell division. Aberrant genetic and epigenetic alteration of PLK members plays a controversial role among different cancers. In this study, we performed an analysis of transcriptional and protein expression to identify overexpressed PLK1/4 and under-expressed PLK2/3 in lung cancer subtypes. We then analysed biological function of PLKs and related genes. Besides, we estimated a correlation of PLKs with patient's genders and TP53 mutation in lung cancer. Higher PLK1/4 expression was significantly associated with male patient and TP53 mutant status, separately. Moreover, we carried out a methylation profile analysis including methylation level, DNA methyltransferases correlation and survival analysis of global methylation. Global methylation survival analysis showed that prognostic value of PLK1/2/4 methylation remained the same significant trend between two lung cancer subtypes, whereas prognostic value of PLK3 methylation lacked consistency. Taken together, these results provided instructive insights into a comprehensive evaluation for advanced therapeutic strategy based on epigenetic evidences.

Evaluation of Computer-Aided Nodule Assessment and Risk Yield (CANARY) in Korean patients for prediction of invasiveness of ground-glass opacity nodule.

Differentiating the invasiveness of ground-glass nodules (GGN) is clinically important, and several institutions have attempted to develop their own solutions by using computed tomography images. The purpose of this study is to evaluate Computer-Aided Analysis of Risk Yield (CANARY), a validated virtual biopsy and risk-stratification machine-learning tool for lung adenocarcinomas, in a Korean patient population. To this end, a total of 380 GGNs from 360 patients who underwent pulmonary resection in a single institution were reviewed. Based on the Score Indicative of Lung Cancer Aggression (SILA), a quantitative indicator of CANARY analysis results, all of the GGNs were classified as "indolent" (atypical adenomatous hyperplasia, adenocarcinomas in situ, or minimally invasive adenocarcinoma) or "invasive" (invasive adenocarcinoma) and compared with the pathology reports. By considering the possibility of uneven class distribution, statistical analysis was performed on the 1) entire cohort and 2) randomly extracted six sets of class-balanced samples. For each trial, the optimal cutoff SILA was obtained from the receiver operating characteristic curve. The classification results were evaluated using several binary classification metrics. Of a total of 380 GGNs, the mean SILA for 65 (17.1%) indolent and 315 (82.9%) invasive lesions were 0.195±0.124 and 0.391±0.208 (p < 0.0001). The area under the curve (AUC) of each trial was 0.814 and 0.809, with an optimal threshold SILA of 0.229 for both. The macro F1-score and geometric mean were found to be 0.675 and 0.745 for the entire cohort, while both scored 0.741 in the class-equalized dataset. From these results, CANARY could be confirmed acceptable in classifying GGN for Korean patients after the cutoff SILA was calibrated. We found that adjusting the cutoff SILA is needed to use CANARY in other countries or races, and geometric mean could be more objective than F1-score or AUC in the binary classification of imbalanced data.

Hepatoid adenocarcinoma of lung: A diagnostic challenge - Series of six cases with histopathological, predictive molecular and PD.L1 assessment.

Hepatoid adenocarcinoma of lung is a rare entity, accounting for 5% of all hepatoid adenocarcinoma. Distinguishing it from metastatic hepatocellular carcinoma is essential, but occasionally can be very challenging, especially with concurrent liver mass. A judicious immunohistochemical panel is warranted for accurate diagnosis and subsequent preservation of tissue for molecular testing. There is limited data on the mutational status, behavior and management strategies of this type of lung adenocarcinoma. We report largest series of six cases of hepatoid adenocarcinoma of lung citing the clinical, histopathological, immunohistochemical and molecular parameters including PD-L1 immunoexpression as a predictive biomarker for immunotherapy. None of the evaluated cases showed targetable mutation; however, four out of six cases showed significant PD-L1 expression. All the cases presented with advanced stage and received chemotherapy, however overall prognosis was dismal. In view of significant PD-L1 expression in these tumors and poor response to conventional chemotherapy, these cases might be considered for upfront immunotherapy.

Cost-effectiveness analysis comparing companion diagnostic tests for EGFR, ALK, and ROS1 versus next-generation sequencing (NGS) in advanced adenocarcinoma lung cancer patients.

BACKGROUND: The treatment of choice for advanced non-small cell lung cancer is selected according to the presence of specific alterations. Patients should undergo molecular testing for relevant modifications and the mutational status of EGFR and translocation of ALK and ROS1 are commonly tested to offer the best intervention. In addition, the tests costs should also be taken in consideration. Therefore, this work was performed in order to evaluate the cost-effectiveness of a unique exam using NGS (next generation sequencing) versus other routinely used tests which involve RT-PCR and FISH. METHODS: The target population was NSCLC, adenocarcinoma, and candidates to first-line therapy. Two strategies were undertaken, strategy 1 corresponded to sequential tests with EGFR RT-PCR, then FISH for ALK and ROS1. Strategy 2 differed from 1 in that ALK and ROS1 translocation testing were performed simultaneously by FISH. Strategy 3 considered single test next-generation sequencing, a platform that includes EGFR, ALK and ROS1 genes. A decision tree analysis was used to model genetic testing options. From the test results, a microsimulation model was nested to estimate survival outcomes and costs of therapeutic options. RESULTS: The use of NGS added 24% extra true cases as well as extra costs attributed to the molecular testing. The ICER comparing NGS with sequential tests was US$ 3479.11/correct case detected. The NGS improved a slight gain in life years and QALYs. CONCLUSION: Our results indicated that, although precise, the molecular diagnosis by NGS of patients with advanced stage NSCLC adenocarcinoma histology was not cost-effective in terms of quality-adjusted life years from the perspective of the Brazilian supplementary health system.

A ten-gene signature-based risk assessment model predicts the prognosis of lung adenocarcinoma.

BACKGROUND: Lung adenocarcinoma (LUAD) is a major cause of cancer death. Therefore, identifying potential prognostic risk factors is critical to improve the survival of patients with LUAD. METHODS: Here, relevant datasets were downloaded from TCGA and GEO databases to screen the differentially expressed genes (DEGs). Univariate Cox analysis, LASSO regression analysis and multivariate Cox analysis were conducted on the DEGs combined with TCGA clinical data, and finally a risk assessment model based on 10 feature genes was constructed. RESULTS: The prognosis of patients was evaluated after the patients were grouped based on the median risk score and the results showed that the survival time of patients in the high-risk group was significantly shorter than that in the low-risk group. ROC analysis showed that the AUC values of the 1, 3, 5-year survival were 0.753, 0.724, and 0.73, respectively, indicating that the model was precise in predicting the prognosis, which was also verified in the external dataset GSE72094. In addition, a significant correlation was found between the risk score and the clinical stages of LUAD, that is, a later stage always corresponded to a higher risk score. Then, we performed survival analysis on the 10 feature genes independently in the TCGA-LUAD dataset through the GEPIA database, finding that the high expression of 6 genes (COL5A2, PLEK2, BAIAP2L2, S100P, ZIC2, SFXN1) was associated with the poor prognosis of LUAD patients. CONCLUSION: To sum, this study established a 10-gene risk assessment model and further evaluated its value in predicting LUAD prognosis, which provided a new method for the prognosis prediction of LUAD.

Molecular testing on bronchial washings for the diagnosis and predictive assessment of lung cancer.

Cytopathological analyses of bronchial washings (BWs) collected during fibre-optic bronchoscopy are often inconclusive for lung cancer diagnosis. To address this issue, we assessed the suitability of conducting molecular analyses on BWs, with the aim to improve the diagnosis and outcome prediction of lung cancer. The methylation status of RASSF1A, CDH1, DLC1 and PRPH was analysed in BW samples from 91 lung cancer patients and 31 controls, using a novel two-colour droplet digital methylation-specific PCR (ddMSP) technique. Mutations in ALK, BRAF, EGFR, ERBB2, KRAS, MAP2K1, MET, NRAS, PIK3CA, ROS1 and TP53 and gene fusions of ALK, RET and ROS1 were also investigated, using next-generation sequencing on 73 lung cancer patients and 14 tumour-free individuals. Our four-gene methylation panel had significant diagnostic power, with 97% sensitivity and 74% specificity (relative risk, 7.3; odds ratio, 6.1; 95% confidence interval, 12.7-127). In contrast, gene mutation analysis had a remarkable value for predictive, but not for diagnostic, purposes. Actionable mutations in EGFR, HER2 and ROS1 as well as in other cancer genes (KRAS, PIK3CA and TP53) were detected. Concordance with gene mutations uncovered in tumour biopsies was higher than 90%. In addition, bronchial-washing analyses permitted complete patient coverage and the detection of additional actionable mutations. In conclusion, BWs are a useful material on which to perform molecular tests based on gene panels: aberrant gene methylation and mutation analyses could be performed as approaches accompanying current diagnostic and predictive assays during the initial workup phase. This study establishes the grounds for further prospective investigation.

Analysis of gene expression profiles of lung cancer subtypes with machine learning algorithms.

Lung cancer is one of the most common cancer types worldwide and causes more than one million deaths annually. Lung adenocarcinoma (AC) and lung squamous cell cancer (SCC) are two major lung cancer subtypes and have different characteristics in several aspects. Identifying their differentially expressed genes and different gene expression patterns can deepen our understanding of these two subtypes at the transcriptomic level. In this work, we used several machine learning algorithms to investigate the gene expression profiles of lung AC and lung SCC samples retrieved from Gene Expression Omnibus. First, the profiles were analyzed by using a powerful feature selection method, namely, Monte Carlo feature selection. A feature list, ranking all features according to their importance, and some informative features were obtained. Then, the feature list was used in the incremental feature selection method to extract optimal features, which can allow the support vector machine (SVM) to yield the best performance for classifying lung AC and lung SCC samples. Some top genes (CSTA, TP63, SERPINB13, CLCA2, BICD2, PERP, FAT2, BNC1, ATP11B, FAM83B, KRT5, PARD6G, PKP1) were extensively analyzed to prove that they can be differentially expressed genes between lung AC and lung SCC. Meanwhile, a rule learning procedure was applied on informative features to construct the classification rules. These rules provide a clear procedure of classification and show some different gene expression patterns between lung AC and lung SCC.

Establishment and Evaluation of a 6-Gene Survival Risk Assessment Model Related to Lung Adenocarcinoma Microenvironment.

OBJECTIVE: A survival risk assessment model associated with a lung adenocarcinoma (LUAD) microenvironment was established and evaluated to identify effective independent prognostic factors for LUAD. METHODS: The public data were downloaded from the TCGA database, and ESTIMATE prediction software was used to score immune cells and stromal cells for tumor purity prediction. The samples were divided into the high-score group and the low-score group by the median value of the immune score (or stromal score). The Wilcoxon test was used for differential analysis. GO and KEGG enrichment analysis of differentially expressed genes (DEGs) was performed using "clusterProfiler" of R package. Meanwhile, univariate and multivariate regression analysis was performed on DEGs to construct a multivariate Cox risk regression model with variable gene expression levels as independent prognostic factors affecting a tumor microenvironment (TME) and tumor immunity. RESULTS: This study found that LUAD patients with high immune cell (stromal cell) infiltration had better prognosis and were in earlier staging. Functional enrichment analysis revealed that most DEGs were related to the proliferation and activation of immune cells or stromal cells. A survival prediction model composed of 6 TME-related genes (CLEC17A, TAGAP, ABCC8, BCAN, FLT3, and CCR2) was established, and finally, the 6 feature genes closely related to the prognosis of LUAD were proved. The AUC value of the ROC curve in this model was 0.7, indicating that the model was reliable. CONCLUSION: Six genes related to the LUAD microenvironment have a predictive prognostic value in LUAD.

Toward improving prognosis prediction in patients undergoing small lung adenocarcinoma resection: Radiological and pathological assessment of diversity and intratumor heterogeneity.

OBJECTIVES: Prediction of prognosis based on the ground-glass opacity (GGO) ratio for small (≤2 cm) lung adenocarcinomas is not completely accurate. The aim of this study was to clarify the diversity of small adenocarcinomas and to identify ways to more accurately predict prognosis. MATERIALS AND METHODS: We retrospectively reviewed 62 patients (64 lesions) that underwent lobectomy for small (≤2 cm) lung adenocarcinoma. Proportions of histological components were measured and the presence of tumor spread through air spaces (STAS) was assessed. The correlations between GGO and histological components were examined. Furthermore, histological components of pure GGO lesions were analyzed using CT values. The intratumor heterogeneity of programmed death ligand 1 (PD-L1) expression was analyzed in 40 lesions. Furthermore, the relationship between CT/histological findings and prognoses was analyzed. RESULTS: In 13 pure GGO lesions, 7 (53.8%) lesions contained invasive components such as papillary, acinar, solid, and colloid. Tumor spread through air spaces (STAS) was also found in pure GGO lesions. Pure GGO lesions containing invasive components (p =  0.002) and STAS-positive lesions (p =  0.011) demonstrated strongly higher CT value. Differences in expression of PD-L1 among histological subtypes were observed in four of six (66.7%) PD-L1 positive lesions. Patients with papillary component, positivity for STAS, or CT value ≥ -140.6 Hounsfield units (HU) had significant poorer prognoses than patients without those in disease-free survival analyses (p =  0.007, p =  0.048, p = 0.012). Patients with the CT value < -383.4 HU and GGO ≥ 50% did not have recurrence. CONCLUSIONS: Invasive component and STAS can be present even in small GGO lesions, and patients with papillary components or STAS showed significantly poorer prognoses. STAS-positive lesions were strongly associated with a high CT value, and combined use of GGO ratio and CT value may be able to predict recurrences of lung cancer more accurately.

TP53 and DNA-repair gene polymorphisms genotyping as a low-cost lung adenocarcinoma screening tool.

AIM: TP53 and DNA repair polymorphisms have been proposed as cancer risk factors. This study evaluated the usability of TP53 Arg72Pro single-nucleotide polymorphism, X RCC1 Arg399Gln and RAD51 G135C as a low-cost lung adenocarcinoma screening tool. PATIENTS AND METHODS: This case-control study included 78 atients with lung adenocarcinoma and 79 healthy matched controls. TP53, XRCC1 and RAD51 genotyping was done by PCR followed by restriction length polymorphism. Descriptive analyses included genotype and allelic frequencies and deviations of the frequencies from those expected under Hardy-Weinberg equilibrium were assessed using the χ2 test. The OR and 95% CIs were calculated as an estimate of relative risk, with significance set at p value <0.05. RESULTS: The TP53 codon 72 Pro allele and the XRCC1 codon 399 Arg allele in a homozygous state were associated with lung adenocarcinoma (p=0.037; OR (95% CI) 2.42 (1.10 to 5.31)), that is, p=0.037; OR (95% CI) 2.16 (1.08 to 4.33), respectively. Also, carriers of the TP53 codon 72 Pro allele and the XRCC1 codon 399 ArgArg genotype older than 50 showed an even higher risk of developing lung adenocarcinoma (p=0.03 in both cases). CONCLUSIONS: The TP53 codon 72 Arg allele and XRCC1 codon 399 Gln allele are likely to have a protective effect against lung adenocarcinoma, especially in individuals older than 50 years of age. XRCC1 and TP53 genotyping might be a useful low-cost tool for evaluating individual lung cancer risk, leading to earlier detection and management of this disease.

Clonality assessment of multifocal lung adenocarcinoma by pathology evaluation and molecular analysis.

The aim of this study was to explore morphologic and molecular features distinguishing between multifocal lung adenocarcinoma (MLA) and intrapulmonary metastases (IMs). Sixteen patients with MLAs, a total of 34 tumors, were reviewed. Four approaches were used: (1) array-comparative genomic hybridization (CGH) as a standard clonality assessment; (2) EGFR and KRAS mutational profiles as a supplementary method; (3) comprehensive histologic assessment (CHA) was method I in pathology evaluation; and (4) CHA combined with lepidic component analysis was method II. The lepidic component was divided into low grade and high grade according to extent of atypia; tumors with low-grade lepidic component were defined as primary. Eight patients were found to have IMs and 8 to have multiple primaries (MPs) by array-CGH; 7 had MPs and 9 had IMs by method I; 5 had MPs and 11 had IMs by method II. Compared with array-CGH, method I had a lower coincidence rate (65%) than method II (85%). Univariate analysis revealed that patients with MP had a better clinical outcome than those with IM only if the MPs were diagnosed by array-CGH (P = .034) or method II (P = .027) but not EGFR/KRAS mutation (P = .843) or method I (P = .493). Our results suggest that a low-grade lepidic component is a sign of a primary tumor. CHA combined with a low-grade lepidic component (method II) is more accurate clinically and more cost-effective in distinguishing MLAs from IMs. Also, EGFR mutation is not an appropriate molecular marker for clonality assessment.

Assessing the inter-observer variability of Computer-Aided Nodule Assessment and Risk Yield (CANARY) to characterize lung adenocarcinomas.

Lung adenocarcinoma (ADC), the most common lung cancer type, is recognized increasingly as a disease spectrum. To guide individualized patient care, a non-invasive means of distinguishing indolent from aggressive ADC subtypes is needed urgently. Computer-Aided Nodule Assessment and Risk Yield (CANARY) is a novel computed tomography (CT) tool that characterizes early ADCs by detecting nine distinct CT voxel classes, representing a spectrum of lepidic to invasive growth, within an ADC. CANARY characterization has been shown to correlate with ADC histology and patient outcomes. This study evaluated the inter-observer variability of CANARY analysis. Three novice observers segmented and analyzed independently 95 biopsy-confirmed lung ADCs from Vanderbilt University Medical Center/Nashville Veterans Administration Tennessee Valley Healthcare system (VUMC/TVHS) and the Mayo Clinic (Mayo). Inter-observer variability was measured using intra-class correlation coefficient (ICC). The average ICC for all CANARY classes was 0.828 (95% CI 0.76, 0.895) for the VUMC/TVHS cohort, and 0.852 (95% CI 0.804, 0.901) for the Mayo cohort. The most invasive voxel classes had the highest ICC values. To determine whether nodule size influenced inter-observer variability, an additional cohort of 49 sub-centimeter nodules from Mayo were also segmented by three observers, with similar ICC results. Our study demonstrates that CANARY ADC classification between novice CANARY users has an acceptably low degree of variability, and supports the further development of CANARY for clinical application.

Semiquantative Visual Assessment of Sub-solid Pulmonary Nodules ≦3 cm in Differentiation of Lung Adenocarcinoma Spectrum.

We aimed to analyze CT features of persistent subsolid nodules (SSN) ≦3 cm diagnosed pathologically as adenocarcinoma spectrum to investigate whether parameters enable distinction between invasive pulmonary adenocarcinomas (IPAs) and pre-invasive lesions. A total of 129 patients with 141 SSNs confirmed with surgically pathologic proof were retrospectively reviewed. Of 141 SSNs, there were 57 pure ground-glass nodules (GGNs), 22 heterogeneous GGNs, and 62 part-solid nodules. SSN subclassification showed a significant linear trend with invasive degree of the adenocarcinoma spectrum (pure GGNs 7%; heterogeneous GGNs 36.4%; part-solid nodules 85.5%, P for trend <0.0001). For IPA detection in 141 SSNs, a solid part of ≧3 mm was the most specificity (sensitivity, 76.9%; specificity, 94.7%), followed by air-bronchogram sign (sensitivity, 53.8%; specificity, 89.5%), SSN subclassification (sensitivity, 81.5%; specificity, 88.2%), and a lesion size ≧12 mm (sensitivity, 84.6%; specificity, 76.3%). For IPA detection in 79 pure or heterogeneous GGNs, the heterogeneous GGN sign was the most useful finding, with most specificity (sensitivity, 66.7%; specificity, 79.1%), followed by CT attenuation (HU) of ≧-493 (sensitivity, 75%; specificity, 74.6%) and a lesion size ≧10 mm (sensitivity, 83.3%; specificity, 70.1%). In conclusion, this simple combined visual and semiquantitative analysis of CT features helps distinguish IPAs from pre-invasive lesions.