Faculty Bibliography

Rationale: An increased ratio of pulmonary artery to aorta (PA/A) diameter, as a marker of vascular injury, measured by computed tomography (CT) predicts future exacerbations in patients with chronic obstructive pulmonary disease (COPD) Wells et. al. NEJM. World Trade Center (WTC) exposed fire fighters have developed respiratory symptoms and a subset had a decline in pulmonary function. Our group has previously shown that systemic biomarkers of inflammation and cardiovascular disease predict this decline. We hypothesize that a PA/A ratio >/= to 1 will be associated with a decline in FEV1. Methods: From a baseline cohort of never smokers with normal spirometry pre 9/11, cases and controls were selected. Cases had FEV1 fall to less than the lower limit of normal (< LLN), and controls had preserved lung function, see figure 1. Spirometry was performed according to ATS/ERS guidelines. Inspiratory series CT images, collected contemporaneously with spirometry, were retrospectively assessed using iSite PACS, (Philips iSite Enterprise, Version 3.6.114; www.healthcare.philips.com). The diameter of the main PA at the level of its bifurcation and the diameter of the ascending aorta in its maximum dimension were recorded using the same image. ES, who was blinded to group assignment, made all measurements. Statistics and data management were performed using SPSS. Results: CT images were available for 91 patients in the case/control cohort. Body mass index (BMI), age, exposure and pulmonary function data are shown in Table 1. Exposure intensity, age at exposure, time from 9/11 to spirometry and to CT were similar. BMI was increased in cases compared to controls. The mean PA diameter and PA/A ratio were increased in cases (p=0.05, 0.09), the mean A diameter was similar. Using binary logistic regression the odds ratio of having an FEV1< LLN if the PA/A ratio was >/= 1 was 3.6 (p=0.047), when corrected for exposure, age at 9/11 and BMI. Conclusions: In this preliminary study a PA/A ratio >/= 1 was associated with WTC related decline in FEV1. There are several potential confounders. Data on the presence of congestive heart failure, sleep apnea and other comorbidities are presenting lacking. Additionally, patient effort and respiratory system compliance may influence inspiratory measurements of the PA. In future studies, we plan to analyze expiratory CT images and correlate with other markers of heart disease. Increased PA/A represents another potentially useful non-invasive tool to assess for obstructive lung dysfunction and warrants further study. (Table Presented)

RATIONALE: Analysis of local in vivo inflammation is relevant to the understanding of pathogenesis and disease progression in emphysema. Bronchial reactivity is an early marker of disease in asthma but the relevance of reactivity to the natural history of emphysema is not understood. We hypothesize that bronchial reactivity is a phenotype of early emphysema that might be related to the degree of inflammation in the lung. METHODS: Normal subjects were enrolled as part of a normal volunteer protocol. Emphysema subjects were identified from the NYU Lung Cancer Biomarker Center CT-scan screening cohort. All patients underwent spirometry, plethysmography, diffusion, and oscillometry, as well as bronchoscopy with bronchoalveolar lavage (BAL). Bronchial reactivity was assessed by changes in FEV1, V50 and R5 . From the BAL fluid, cell count differential was obtained, as well as measurement of 39 cytokines in concentrated BAL fluid with Luminex using Human Cytokine Panel I (Millipore). Results amongst the groups were compared with ANOVA and post-hoc LSD comparison. RESULTS: Twenty patients were available for analysis: Six subjects in the control group, 6 emphysema subjects without bronchial reactivity (BR-), and 8 emphysema subjects with bronchial reactivity (BR+). Baseline demographics and pertinent spirometry/oscillometry are listed in Table 1. Emphysema subjects were all GOLD stage 0 or 1. Post-bronchodilator spirometric and oscillometric parameters were not significantly different between BR- and BR+ emphysema groups. There were 28/39 cytokines with reliably measurable levels. Both emphysema groups had elevated neutrophils and higher degree of inflammation as compared to controls (significant data shown Table 1). However, the BR+ emphysema group evidenced higher degree of neutrophils, IL-6, IL-8, G-CSF, Eotaxin, GRO and Fractalkine as compared with the BR- emphysema group. CONCLUSION: These data suggest that in early emphysema a phenotype of proximal and/or distal bronchial reactivity is associated with an increased degree of inflammation as assessed by neutrophils and in vivo inflammatory cytokines. In contrast with early asthma, the phenotype of bronchial reactivity in early emphysema may be characterized by neutrophilic inflammation produced by increased IL-8 in the lung. The role of IL-6, G-CSF, Eotaxin, GRO and Fractalkine in producing emphysema related bronchial reactivity requires further investigation. (Table Presented)

RATIONALE. Diagnostic decision-making for patients with pulmonary nodules balances an individual's risk for non-small cell lung cancer (NSCLC) and the likelihood of benign disease. Management guidelines include serial CT observation or diagnostic resection, respectively, and PET scans and/or invasive biopsy procedures for 8-20 mm nodules, many of which are non-malignant. Recent lung cancer biomarker research suggests the potential utility of molecular adjuncts to augment clinicians' assessment of malignancy risk based on age, smoking history and nodule size. We explored the hypothesis that a multivariate protein panel may discriminate patients with malignant and benign nodules, and used selected reaction monitoring mass spectrometry (SRM-MS) to derive a 13-protein plasma classifier optimized for identifying benign nodules. METHODS. A systems biology approach was used to nominate 388 candidate proteins for classifier development. Using a retrospective, multi-center (n=4), and paired case-control study design, K2-EDTA plasma samples from lung nodule patients with pathologic diagnoses of NSCLC or benign disease were analyzed in independent discovery (D) (n=143) and validation (V) (n=104) sets. Cancer and benign samples were matched pairwise for age, gender and nodule size. Subject and nodule characteristics for cases/controls, respectively, included: age (mean), D:65/64, V:63/62; %male, D:40%/47%, V:48%/52%; smoking pack-years (mean), D:37/20, V:40/27; and nodule size (mm, mean), D:13/13, V:16/15. 70 microliter plasma aliquots immuno-depleted of high abundance proteins were analyzed by SRM-MS using the AB Sciex QTRAP 5500 LC/MS/MS system. RESULTS. Bioinformatic methods using Monte Carlo cross-validation, identification of best performing group proteins, and logistic regression analysis yielded 13 classifier and 6 normalization proteins. To optimize sensitivity for identifying benign nodules, the method of partial receiver operating characteristics area under the curve was utilized; the classifier demonstrated 93% sensitivity and 45% specificity in discovery, and using an estimated cancer prevalence of 20%, resulted in PPV and NPV of 30% and 96%, respectively. Similarly, in validation, it yielded 90% sensitivity and 27% specificity, with PPV and NPV of 24% and 90%, respectively. There was no correlation between classifier scores and age, smoking history pack-years or nodule size. Ingenuity Pathway Analysis mapped the 13 classifier proteins to 4 nuclear proteins (AHR, FOS, MYC, NRF2) regulating lung cancer, lung inflammation and oxidative stress responses. CONCLUSIONS. The use of a systems-based, proteomic strategy yielded a 13-protein plasma classifier prioritizing a high NPV with the potential for use as a non-invasive molecular adjunct for the early diagnosis of patients with pulmonary nodules

Rationale: While bacterial colonization leads to airway inflammation and accelerated airway obstruction in advanced COPD, its role in early disease is not well understood. Major limitations have been the ability to study airway colonization in early COPD and measurements of in vivo cytokines levels in the lung. Here, we used a pyrosequence approach for 16s rDNA to study the airway microbiome and a concentration technique to achieve measurable levels of cytokines in BAL fluid from early emphysema and normal volunteers. We hypothesize that the microbiome of early emphysema will be characterized by the presence of potential pathogens that are associated with subclinical inflammation. Methods: CT-scan defined emphysema. Supraglotic and broncho-alveolar lavage (BAL) samples were obtained with two different bronchoscopes. Bacteria speciation was performed with 454 sequencing of rDNA. BAL differential was performed on diff quick stained cytospine slides. BAL was concentrated 50 fold by lyophilization and cytokes were measured using Luminex. Results: 24 subjects were available for analysis (8 healthy normal volunteers and 16 with emphysema). All emphysema subjects were ex-smokers (normal 1+/-0.5 vs. 39+/-20 pack-yr., p=0.019). Lung function of emphysema subjects was characterized by lower FEV1/FVC (78.9+/-5.1 vs. 69.6+/-6.1, p=0.001) and hyperinflation (TLC=87.7+/-10.3 vs. 104.3+/-15.2% predicted, p=0.016). Background environmental microbiome, present in "sterile" saline, is enriched with Staphylococcus and Propionobacterium rDNA (Figure 1 right panel). Supraglotic microbiome is enriched with Prevotella and Streptococcus (left panel). Individuals' BAL microbiome varied from a microbiome more similar to supraglotic with higher relative abundance of Prevotella to a background microbiome (middle panel). In this small sample there was a non-significant trend to higher relative abundance of Prevotella in emphysema vs. normal (p=0.078). There was significant positive correlation between Prevotella and BAL inflammatory cytokines (r2=0.48,p<0.001 and r2=0.22,p<0.02 for IL-1alpha and IL-8 respectively, Figure 2 panel A and B). Furthermore, Prevotella directly correlates with Neutrophils in lung (r2=0.24,p=0.02, Panel C). Conclusions: This data shows that inflammatory cytokines are produced and neutrophils recruited to the alveolar space when the lung and supraglotic microibome are similar. In patients with Prevotella, the bacteria has been present long enough to produce subclinical inflammation. It is therefore unlikely that the Prevotella microbiome is due to carryover during bronchoscopy. In individuals with a BAL microbiome similar to saline, poor sinal-to-noise ratio prevents investigation of the lung microbiome. The Prevotella microbiome more common in emphysema is associated with increased bronco-alveolar cytokines and neutrophils, suggesting an immunological response to this lung microbiome. (Figure Presented)

Rationale: The use of culture-independent techniques to evaluate resident microbial communities in the lung has opened opportunities to evaluate host response phenotype in health and disease. Background environmental microbiome found in saline and bronchoscope prior to bronchoscopy is characterized by high relative abundance of Propionibacterium. Our preliminary data from the lung microbiome project suggest that substitution of background environmental microbiome by Prevotella or Streptococcus microbiome is associated with higher inflammation. We hypothesize that patients with emphysema who have asymmetric disease on CT will have disappearance of background environmental microbiome in the more affected lung segments. We will also evaluate whether background environmental microbiome is associated with lower inflammation (neutrophil counts, and chemo-attractant cytokines). Methods: Subjects with emphysema were enrolled for research bronchoscopy from NYU/EDRN cohort and CT scans were classified as symmetrical or asymmetrical lung disease. Broncho-alveolar lavages (BAL) were obtained from the right and left lung. Sequencing of 300 bp 16S rDNA included V1-V2 region, performed with 454 pyrosequence. Propionibacterium was used as a marker of background environmental microbiome. Cytokines in BAL fluid will be assayed using Human Cytokine Panel I (Millipore). Results: To date, 15 subjects had sequence data from two segments of different lungs (5 normal volunteers, 6 symmetrical lung disease patients, and 4 asymmetrical lung disease patients). Healthy volunteers were younger than subjects with emphysema (41 +/- 11, 61 +/- 6 respectively, p = 0.003). Although no significant difference in FEV1 was observed, emphysema groups trended to have lower FEV1/FVC (p=ns). There were no differences in high relative abundant taxa (greater than 0.05) between the right and left lung of normal volunteers and emphysema subjects with symmetrical lung disease. However, despite the small n, emphysema subjects with asymmetrical lung disease trended to have higher relative abundance of background environmental microbiome in lung segments with less disease ( Propionibacterium relative abundance = 0.13 +/- 0.04 for segments with less disease as compared with 0.03 +/- 0.04 in the more disease segments, p < 0.08). We will complete sequence in 5 more emphysema subjects and compare microbiota with in-vivo BAL cytokines. Conclusions: Patients with observable asymmetrical lung disease have lower background environmental microbiome in more diseased lung when compared to the less diseased side. This difference was not observed in normal volunteers and patients with symmetrical lung disease. Disappearance of background environmental microbiome in more diseased lung segments suggests higher airway colonization, which might be associated with subclinical inflammation

Rationale: Cigarette smoke causes a field of injury and molecular changes in the airways even in histologically normal areas termed "field cancerization" which describes the site(s) of neoplasia and adjacent normal tissue with molecular abnormalities in common. MicroRNAs ( miRNAs) are small, non-coding RNAs that act as post-transcriptional regulators of gene expression by recognizing target sites in the 3' untranslated regions (3'UTRs) via incomplete base-pairing and induce mRNA degradation or translational repression. Deregulation of miRNAs has been linked to cancer initiation and progression, and miRNAs may act as tumor suppressor genes or oncogenes. We hypothesized that miRNA expression in the peripheral airways of smokers with lung cancer is distinct from that of smokers without lung cancer and therefore, miRNAs can be used as biomarkers for the early detection of lung cancer. Methods: We collected human peripheral airway epithelial cells by bronchoscopic brushing from the unaffected lung of thirteen smokers with lung adenocarcinoma and twelve control smokers. Total RNA was extracted from the peripheral airway epithelial cells by miRNAeasy and miRNA profiling was performed using the TaqMan Quantitative qRT-PCR miRNA Assay. Results: Comparison of miRNA levels in peripheral airway epithelial cells from smokers with or without lung cancer demonstrated 53 miRNAs that were significantly different (p<0.05) between the two groups. The majority of miRNAs were up-regulated (41 miRNAs) in lung cancer patients, including miR-21, miR-26a, miR-31, miR-34c, and miR-205. Down-regulated miRNAs included let-7b, let-7e, and miR-126. Several of the miRNAs with increased expression are of interest: miR-21 inhibits tumor suppressor protein PTEN, miR-26a suppresses PTEN and increases AKT phosphorylation and nuclear factor kappaB (NFkappaB) activation, miR-31 represses tumor suppressor genes LATS2 and PPP2R2A, miR-205 is associated with cancer relapses, and miR-34c, a p53 target induced by DNA damage, suggests the involvement of p53 pathway in field carcinogenesis. Down-regulated let-7b leads to higher expression of CYP2J2 and decreased miR-126 enhances adhesion, migration and invasion through increased Crk protein. Further gene expression and pathway analyses will corroborate the relationship between miRNAs and predicted pathways in real time. Conclusion: We discovered a profile of miRNAs in the contralateral lung of patients with lung cancer as biomarkers of field cancerization in smokers with lung adenocarcinoma. Further knowledge of field cancerization may lead to better understanding of tumorigenesis and development of biomarkers for early lung cancer detection

Lung cancer is the leading cause of cancer deaths worldwide largely owing to diagnosis of the disease at an advanced stage. Recent advances in blood-based biomarker research have the potential to reduce mortality by providing a means for detecting lung cancer at an earlier stage. Since the publication of the National Lung Cancer Screening Trial demonstrating reduction in mortality with computed tomography (CT) scan screening, the U.S. Preventive Services Task Force has released a draft statement recommending annual low-dose CT scan screening for high-risk patients. However, CT screening has a high false-positive rate leading to the need for additional imaging and invasive procedures. In this article, we review recent literature on blood-based lung cancer biomarkers that we believe will have a significant role in enhancing screening efficacy in the near future.

Objective: Glycomics is an emerging area for understanding carcinogenesis; studies in glycobiology have documented that aberrant glycosylation accompanies malignant transformation. We have used glycomic profiling of serum using nano-liquid chromatography-mass spectrometry and using biostatistics we report glycomics patterns distinguishing between non-small cell lung cancer cases (NSCLC) versus healthy controls. Methods: First, we obtained pre-operative sera of non-small lung cancer cases (Stages I-II) and healthy controls from the NYU biorepository. The serum sample set consisted of 50 lung cancer (adenocarcinoma) patients, 50 healthy controls and 28 COPD patients, matched on gender, smoking status, pack/year for smokers, and as best as possible for age. The samples are analyzed by extracting the N-glycans in sera and measuring the relative concentrations using nano-liquid chromatography-mass spectrometry (nano-LC-MS). Bioinformatic analysis was performed to identify glycans that are differentially present in patients with cancer compared to COPD and healthy controls using feature selection and classification algorithms. We further investigated whether a combination of multiple glycans (i.e. multiplex classifier) could improve predictive performance over individual glycans. Results: Of 330 glycans detected in the NYU serum set, twenty glycans were significantly differentiating (either over- or under-expressed) between cancer, COPD and controls at a false discovery rate < 0.05. Of these glycan features, 11 were different between cancer samples and control samples, while 14 glycans differed significantly between cancer samples and COPD samples. COPD samples differed significantly from Control samples for only one glycan in concentration, indicating large similarity in the glycosylation pattern between COPD and controls. Based on the glycomic profiles, cancer cases were well separated from both control and COPD samples, while COPD and control samples were not discriminated well from each other. Th!