Baskaran Sundaram

Analysis of the Lung Microbiome in the “Healthy” Smoker and in COPD

Although culture-independent techniques have shown that the lungs are not sterile, little is known about the lung microbiome in chronic obstructive pulmonary disease (COPD). We used pyrosequencing of 16S amplicons to analyze the lung microbiome in two ways: first, using bronchoalveolar lavage (BAL) to sample the distal bronchi and air-spaces; and second, by examining multiple discrete tissue sites in the lungs of six subjects removed at the time of transplantation. We performed BAL on three never-smokers (NS) with normal spirometry, seven smokers with normal spirometry (“heathy smokers”, HS), and four subjects with COPD (CS). Bacterial 16 s sequences were found in all subjects, without significant quantitative differences between groups. Both taxonomy-based and taxonomy-independent approaches disclosed heterogeneity in the bacterial communities between HS subjects that was similar to that seen in healthy NS and two mild COPD patients. The moderate and severe COPD patients had very limited community diversity, which was also noted in 28% of the healthy subjects. Both approaches revealed extensive membership overlap between the bacterial communities of the three study groups. No genera were common within a group but unique across groups. Our data suggests the existence of a core pulmonary bacterial microbiome that includes Pseudomonas, Streptococcus, Prevotella, Fusobacterium, Haemophilus, Veillonella, and Porphyromonas. Most strikingly, there were significant micro-anatomic differences in bacterial communities within the same lung of subjects with advanced COPD. These studies are further demonstration of the pulmonary microbiome and highlight global and micro-anatomic changes in these bacterial communities in severe COPD patients.

Pulmonary function measures predict mortality differently in IPF versus combined pulmonary fibrosis and emphysema

The composite physiologic index (CPI) was derived to represent the extent of fibrosis on high-resolution computed tomography (HRCT), adjusting for emphysema in patients with idiopathic pulmonary fibrosis (IPF). We hypothesised that longitudinal change in CPI would better predict mortality than forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC) or diffusing capacity of the lung for carbon monoxide (DL,CO) in all patients with IPF, and especially in those with combined pulmonary fibrosis and emphysema (CPFE). Cox proportional hazard models were performed on pulmonary function data from IPF patients at baseline (n = 321), 6 months (n = 211) and 12 months (n = 144). Presence of CPFE was determined by HRCT. A five-point increase in CPI over 12 months predicted subsequent mortality (HR 2.1, p = 0.004). At 12 months, a 10% relative decline in FVC, a 15% relative decline in DL,CO or an absolute increase in CPI of five points all discriminated median survival by 2.1 to 2.2 yrs versus patients with lesser change. Half our cohort had CPFE. In patients with moderate/severe emphysema, only a 10% decline in FEV1 predicted mortality (HR 3.7, p = 0.046). In IPF, a five-point increase in CPI over 12 months predicts mortality similarly to relative declines of 10% in FVC or 15% in DL,CO. For CPFE patients, change in FEV1 was the best predictor of mortality.

Deep Learning at Chest Radiography: Automated Classification of Pulmonary Tuberculosis by Using Convolutional Neural Networks

Purpose To evaluate the efficacy of deep convolutional neural networks (DCNNs) for detecting tuberculosis (TB) on chest radiographs. Materials and Methods Four deidentified HIPAA-compliant datasets were used in this study that were exempted from review by the institutional review board, which consisted of 1007 posteroanterior chest radiographs. The datasets were split into training (68.0%), validation (17.1%), and test (14.9%). Two different DCNNs, AlexNet and GoogLeNet, were used to classify the images as having manifestations of pulmonary TB or as healthy. Both untrained and pretrained networks on ImageNet were used, and augmentation with multiple preprocessing techniques. Ensembles were performed on the best-performing algorithms. For cases where the classifiers were in disagreement, an independent board-certified cardiothoracic radiologist blindly interpreted the images to evaluate a potential radiologist-augmented workflow. Receiver operating characteristic curves and areas under the curve (AUCs) were used to assess model performance by using the DeLong method for statistical comparison of receiver operating characteristic curves. Results The best-performing classifier had an AUC of 0.99, which was an ensemble of the AlexNet and GoogLeNet DCNNs. The AUCs of the pretrained models were greater than that of the untrained models (P < .001). Augmenting the dataset further increased accuracy (P values for AlexNet and GoogLeNet were .03 and .02, respectively). The DCNNs had disagreement in 13 of the 150 test cases, which were blindly reviewed by a cardiothoracic radiologist, who correctly interpreted all 13 cases (100%). This radiologist-augmented approach resulted in a sensitivity of 97.3% and specificity 100%. Conclusion Deep learning with DCNNs can accurately classify TB at chest radiography with an AUC of 0.99. A radiologist-augmented approach for cases where there was disagreement among the classifiers further improved accuracy.

Type I Interferons Are Associated with Subclinical Markers of Cardiovascular Disease in a Cohort of Systemic Lupus Erythematosus Patients

Background Systemic lupus erythematosus (SLE) patients have a striking increase in cardiovascular (CV) comorbidity not fully explained by the Framingham risk score. Recent evidence from in vitro studies suggests that type I interferons (IFN) could promote premature CV disease (CVD) in SLE. We assessed the association of type I IFN signatures with functional and anatomical evidence of vascular damage, and with biomarkers of CV risk in a cohort of lupus patients without overt CVD. Methodology/Principal Findings Serum type I IFN activity (induction of five IFN-inducible genes; IFIGs) from 95 SLE patient and 38 controls was quantified by real-time PCR. Flow mediated dilatation (FMD) of the brachial artery and carotid intima media thickness (CIMT) were quantified by ultrasound, and coronary calcification by computed tomography. Serum vascular biomarkers were measured by ELISA. We evaluated the effect of type I IFNs on FMD, CIMT and coronary calcification by first applying principal components analysis to combine data from five IFIGs into summary components that could be simultaneously modeled. Three components were derived explaining 97.1% of the total IFIG variation. Multivariable linear regression was utilized to investigate the association between the three components and other covariates, with the outcomes of FMD and CIMT; zero-inflated Poisson regression was used for modeling of coronary calcification. After controlling for traditional CV risk factors, enhanced serum IFN activity was significantly associated with decreased endothelial function in SLE patients and controls (p<0.05 for component 3), increased CIMT among SLE patients (p<0.01 for components 1 and 2), and severity of coronary calcification among SLE patients (p<0.001 for component 3). Conclusions Type I IFNs are independently associated with atherosclerosis development in lupus patients without history of overt CVD and after controlling for Framingham risk factors. This study further supports the hypothesis that type I IFNs promote premature vascular damage in SLE.

Diagnosing fibrotic lung disease: When is high‐resolution computed tomography sufficient to make a diagnosis of idiopathic pulmonary fibrosis?

Idiopathic pulmonary fibrosis (IPF), a progressive and fatal diffuse parenchymal lung disease, is defined pathologically by the pattern of usual interstitial pneumonia (UIP). Unfortunately, a surgical lung biopsy cannot be performed in all patients due to comorbidities that may significantly increase the morbidity and mortality of the procedure. High‐resolution computed tomography (HRCT) has been put forth as a surrogate to recognize pathological UIP. The quality of the HRCT impacts the ability to make a diagnosis of UIP and varies based on the centre performing the study and patient factors. The evaluation of the HRCT includes assessing the distribution and predominance of key radiographical findings, such as honeycomb, septal thickening, traction bronchiectasis and ground glass attenuation lesions. The combination of the pattern and distribution is what leads to a diagnosis and associated confidence level. HRCT features of definite UIP (subpleural, basal predominant honeycomb with septal thickening, traction bronchiectasis and ground glass attenuation lesions) have a high specificity for the UIP pathological pattern. In such cases, surgical lung biopsy can be avoided. There are caveats to using the HRCT to diagnose IPF in isolation as a variety of chronic pulmonary interstitial diseases may progress to a UIP pattern. Referral centres with experience in diffuse parenchymal lung disease that have multidisciplinary teams encompassing clinicians, radiologists and pathologists have the highest level of agreement in diagnosing IPF.

Aortic Valve Area on 64-MDCT Correlates with Transesophageal Echocardiography in Aortic Stenosis

OBJECTIVE The purpose of our study was to compare aortic valve area and calcification between CT and echocardiography. MATERIALS AND METHODS We performed retrospective evaluation of 80 consecutive patients with aortic stenosis (AS) who underwent ECG-gated 64-MDCT and transesophageal echocardiography (TEE). Valve planimetry was feasible in 80 patients with CT and in 63 patients with TEE; valve area by transthoracic echocardiography was available in 46 patients. Valve calcification grade on CT was compared with TEE. One cardiologist (echocardiography) and two radiologists (CT) independently and blindly reviewed the studies. Pearsons correlations, Spearmans rank correlations, paired Students t tests, and weighted kappa tests were used. RESULTS The median valve area on TEE was 0.7 +/- 0.9 cm(2). There was excellent correlation (n = 80; r = 0.91, p < 0.001) and no difference (0.06 +/- 0.26 cm(2), p = 0.06) between CT readers. There was strong correlation (n = 63; r = 0.84, p < 0.001) and no difference (-0.06 +/- 0.48 cm(2), p = 0.33) in valve area between CT and TEE, with a strong correlation (n = 46; r = 0.83, p < 0.001) and small overestimation (0.17 +/- 0.33 cm(2), p < 0.001) in valve area with CT versus transthoracic echocardiography. The sensitivity and specificity of CT to detect severe aortic stenosis compared with TEE were 92.1% (35/38) and 89.5% (17/19), respectively. Calcification grade had fair agreement between CT readers and TEE (kappa = 0.34 and 0.37, respectively). CONCLUSION Aortic valve area on CT strongly correlates with echocardiography and has excellent sensitivity and specificity to detect severe stenosis. Valve calcification has fair agreement between studies. Valve area and calcification should be reported on CT angiography in patients with AS.

Impact of mitral isthmus anatomy on the likelihood of achieving linear block in patients undergoing catheter ablation of persistent atrial fibrillation

BACKGROUND Although prior studies have described the anatomy of the mitral isthmus in patients undergoing left atrial (LA) ablation of atrial fibrillation (AF), none has examined the impact of isthmus anatomy on the likelihood of achieving conduction block. OBJECTIVE The purpose of this study was to identify morphologic characteristics of the mitral isthmus that may influence the acute efficacy of linear ablation at the mitral isthmus. METHODS Fifty-five patients (age 61 ± 10 years, 41 [75%] men, LA 46 ± 6 mm, ejection fraction 0.55 ± 0.11, AF duration 4 ± 3 years) underwent linear ablation at the mitral isthmus during an ablation procedure for persistent AF. Computed tomographic scan was performed before the procedure. The morphology of the mitral isthmus and its anatomic relationship to the adjacent vasculature were analyzed. RESULTS Complete block along the mitral isthmus was achieved in 35 (64%) of 55 patients, 23 (66%) of whom required radiofrequency ablation in the coronary sinus (CS). Patients with incomplete block were more likely to have a pouch at the isthmus (40% vs 9%; P = .01), a greater isthmus depth (8.1 ± 4.2 mm vs 5.7 ± 3.4 mm; P = .04), and a higher prevalence of an interposed circumflex artery between the CS and the mitral isthmus (60% vs 20%; P = .003) compared to patients with isthmus block. An interposed circumflex artery was the only independent predictor of incomplete conduction block at the mitral isthmus (odds ratio 4.9, 95% confidence interval 1.3-18.2; P = .02). CONCLUSION Preprocedural computed tomographic imaging identifies patients in whom linear ablation at the mitral isthmus is unlikely to be successful. Interposition of the circumflex artery between the mitral isthmus and the CS is associated with a lower probability of achieving complete mitral isthmus block.

Imaging of coronary artery anomalies.

Coronary artery anomalies (CAA) are uncommon congenital variations in coronary anatomy, occurring in 0.2% to 1.2% of the general population, the majority of which are detected incidentally and have little clinical significance. A minority of CAA, primarily due to an interarterial course, is clinically significant, and may present with symptoms of myocardial ischemia, malignant ventricular arrhythmias, and even sudden cardiac death. Until recently, CAA were primarily detected at catheter coronary angiography. With recent advances in multidetector computed tomography (CT) technology and the use of electrocardiographic gating, coronary CT angiography provides an exquisite omnidimensional display of the anomalous coronary arteries and their relation to the adjacent structures noninvasively, and is the diagnostic test of choice. Understanding CAA morphology and clinical significance of CAA is important for establishing a diagnosis, and is essential for appropriate patient management and treatment planning.

Accuracy of high-resolution CT in the diagnosis of diffuse lung disease: effect of predominance and distribution of findings.

OBJECTIVE The purpose of this study was to determine whether the predominant findings at high-resolution CT influence the accuracy of diagnosis of diffuse lung disease. MATERIALS AND METHODS The cases of 100 patients with diffuse lung disease who underwent high-resolution CT and tissue diagnosis were studied. Three thoracic radiologists reviewed high-resolution CT images blindly and independently for patterns of abnormality, listing their three main diagnoses and level of confidence in the first choice. The effect of the findings on accuracy was analyzed. RESULTS For honeycombing, the accuracy of the main diagnosis was 96.6%, 92.2%, and 92.3% for the three readers, and that of the three main diagnoses was 96.6%, 96.1%, and 92.3%. For cysts, the accuracy of the main diagnosis was 88.9%, 80%, and 81.8% and of the three main diagnoses was 100%, 90%, and 90.9%. For bronchovascular thickening, the accuracy of the main diagnosis was 91.7%, 87.5%, and 90.9% and of the three main diagnoses was 91.7%, 100%, and 90.9%. For ground-glass opacification (GGO), the accuracy of the main diagnosis was 75.5%, 55%, and 44.2% and of the three main diagnoses was 89.8%, 75%, and 65.4%. Only combining honeycombing with GGO improved the accuracy of GGO. Anatomic craniocaudal distribution improved reader accuracy when GGO was predominantly present in the lower part of the lung. Interobserver agreement on the presence of major findings was a mean kappa value of 0.45 for honeycombing, 0.74 for lung cysts, 0.63 for bronchovascular thickening, and 0.56 for GGO. Agreement for the craniocaudal distribution of major findings was a mean kappa value of 0.48 for honeycombing, 0.52 for bronchovascular thickening, and 0.32 for GGO. CONCLUSION The predominant findings of honeycombing and bronchovascular thickening are associated with more than 90% accuracy in the first-choice diagnosis of diffuse lung disease; the finding of lung cysts has 80-89% accuracy. GGO as a predominant pattern had unreliable accuracy, but the accuracy improved when GGO was combined with either honeycombing or lower-lung distribution.

Can CT features be used to diagnose surgical adult bowel intussusceptions

OBJECTIVE The purpose of our study was to identify whether any CT characteristics can be used to diagnose surgical intussusceptions. MATERIALS AND METHODS A search of CT reports on adults revealed 118 patients with 136 intussusceptions. Two blinded readers independently reviewed the CT examinations and documented intussusception characteristics. Medical records were reviewed to determine patient outcome. Performance, interobserver agreement (A), and significance of CT characteristics to identify surgical intussusceptions were calculated. RESULTS Of 95,223 CT examinations, 0.13% (121/95,223) documented 136 intussusceptions over a 7-year period, of which 88.2% (120/136) were enteroenteric, 3.7% (5/136) were enterocolic, and 4.4% (6/136) were colocolic lesions or in other locations. Eight (5.9%) were surgical and 128 (94.12%) were nonsurgical lesions. Five of eight (63%) surgical lesions involved the colon. Only two of eight surgical lesions had malignant lead points. The mean sensitivity, specificity, positive predictive value, and negative predictive value for diagnosing surgical enteroenteric lesions using a measured lesion length of > 3.5 cm were 100%, 57.3%, 5.7%, and 100% (A = 0.68), respectively. Similar figures using the measured axial diameter > 3 cm were 100%, 32.9%, 3.7% and 100% (A= 0.65), respectively. CONCLUSION Surgical intussusceptions in adults are infrequent among the intussusceptions that are detected on CT. Most enteroenteric lesions are nonsurgical lesions, whereas lesions that affect the colon are often surgical. Many nonsurgical enteroenteric intussusceptions are longer than 3.5 cm and thicker than 3 cm, suggesting these CT features may not be useful for diagnosing surgical bowel intussusceptions in adults.