This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lampe, J. W. Articles by Potter, J. D. Search for Related Content PubMed PubMed Citation Articles by Lampe, J. W. Articles by Potter, J. D.

Signatures of Environmental Exposures Using Peripheral Leukocyte Gene Expression: Tobacco Smoke

¹ Fred Hutchinson Cancer Research Center, Seattle, WA and ² Rosetta Inpharmatics, LLC, Merck Research Laboratories, Kirkland, WA

Requests for reprints: John D. Potter, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue N, MP-900, PO Box 19024, Seattle, WA 98109. E-mail: jpotter{at}fhcrc.org

	Abstract

Top Abstract Introduction Methods Results Discussion Appendix A References

 
Functional biological markers of environmental exposures are important in epidemiological studies of disease risk. Such markers not only provide a measure of the exposure, they also reflect the degree of physiological and biochemical response to the exposure. In an observational study, using DNA microarrays, we show that it is possible to distinguish between 85 individuals exposed and unexposed to tobacco smoke on the basis of mRNA expression in peripheral leukocytes. Furthermore, we show that active exposure to tobacco smoke is associated with a biologically relevant mRNA expression signature. These findings suggest that expression patterns can be used to identify a complex environmental exposure in humans.

Key Words: microarray • gene expression • leukocytes • human • smoking • cotinine

	Introduction

Top Abstract Introduction Methods Results Discussion Appendix A References

 
One of the major preoccupations in epidemiology is the measurement of exposures such as infectious agents, foods and dietary constituents, recreational drugs, and environmental and workplace exposures and their associations with disease outcomes. Although some exposures, such as infectious exposures, have been well measured objectively, both directly and indirectly (via serological testing of antibodies), measurement of other classes of exposures has relied largely on self-report or complex environmental monitoring. Some behaviors and exposures are well measured using self-report (e.g., smoking), but many others are much more difficult (e.g., diet and workplace exposures).

Environmental exposures influence a variety of biological processes in quite distinct ways (e.g., enzyme induction, oxidation, signal transduction, etc.). Many of these responses also influence gene expression. Therefore, given a sufficiently large set of biological data to interpret, signatures or patterns of gene expression might emerge that would allow identification and even quantitation of specific exposures. These complex patterns of gene expression can be measured by DNA microarrays (1). One of the most readily obtainable biological materials from the general population is blood. Although peripheral leukocytes see only some environmental exposures and have a limited repertoire of responses, they are exposed, nonetheless, to many of the same environmental agents to which target tissues are exposed. Thus, blood is a useful and convenient biological material in which to seek exposure signatures.

We hypothesized that peripheral leukocyte mRNA expression could be used as a sensor to detect environmental exposures in observational studies. To test this hypothesis, we measured the mRNA signatures of individuals exposed and unexposed to tobacco smoke on the basis that: (a) we could verify smoking exposure by both self-report and the measurement of plasma cotinine concentrations; and (b) if tobacco smoking did not yield a detectable and repeatable signature, other more subtle environmental exposures were even less likely to do so. We also reasoned that if we could detect a signature for tobacco, other exposures, behaviors, and characteristics could also be explored for characteristic signatures.

	Methods

Top Abstract Introduction Methods Results Discussion Appendix A References

 
Study Population
The study population was drawn from individuals who responded to advertisements and fliers placed in various locations across the local community. In addition, 3000 letters were mailed to individuals selected randomly from the Department of Motor Vehicles Department of Licensure list and 1000 letters were sent to individuals over age 60. We selected participants to obtain an equal distribution of smokers and nonsmokers within age ranges and aimed to recruit 10 each of male/smokers, female/smokers, male/nonsmokers, and female/nonsmokers in the age categories 21–40 and 41–60 and 5 of each of those >60 years. Eligibility criteria included: at least 21 years of age; smoker (more than one-half pack or 10+ cigarettes/day) or nonsmoker; if female, not pregnant or lactating; no contraindication to blood draws; not on medications (including oral, intramuscular, and subdermal contraceptive methods, and over-the-counter medications taken on a regular basis). Individuals were eligible if using herbal or nutritional supplements (e.g., vitamins), and those taking specific medications for a finite period became eligible for the study, 1 week after they stopped the medication. The study activities were approved by the Fred Hutchinson Cancer Research Center Institutional Review Board and informed, written consent was obtained from all study participants.

Sample and Data Collection
Each participant was scheduled for two blood draws, 1 week apart. Blood samples were collected in the morning, between 7 and 10 a.m., after a 10-h fast. Two 10-ml blood samples were collected into heparin-containing tubes for mRNA and additional samples were collected into an EDTA-containing tubes for plasma cotinine and WBC differential count. Data on demographics, diet and exercise, and smoking history were also obtained. Blood samples collected at visit 1 were analyzed for plasma cotinine and nicotine concentrations by gas chromatography/nitrogen phosphate detection (National Medical Services, Willow Grove, PA). Cotinine, the primary metabolite of nicotine, has a longer circulating half-life than the parent compound and is used widely as a marker of tobacco use (2). WBC differential counts were determined on a Sysmex NE-8000 hematology analyzer.

RNA Isolation
Blood samples for mRNA were processed within 3 h of being drawn. Following lysis of erythrocytes and removal of cell debris, leukocytes were isolated from whole blood after dextran separation. Total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA) and extracted according to manufacturer's protocol with one adjustment: following addition of isopropanol, RNA was precipitated on ice (4°C) for at least 10 min instead of 15–30°C for 10 min. The pellet was resuspended in 100 µl Total RNase/DNase-free water.

RNA was purified with RNase-free DNase Sets and RNeasy Kits (Qiagen, Valencia, CA). All spins were performed at 8000 rpm for 1 min (Eppendorf centrifuge, model 5415C, Brinkmann Instruments, Westbury, NY) unless noted otherwise. Total RNA was mixed with 350 µl of RLT buffer with ß-mercaptoethanol (Sigma, St. Louis, MD) and 250 µl 100% ethanol (Aaper Alcohol and Chemical Co., Shelbyville, KY). This mixture was transferred to an RNeasy Mini column and centrifuged. Flowthrough was reloaded back into the column and centrifuged once again. The column was washed with 350 µl RW1 buffer. DNase I mixture (10 µl DNase I and 70 µl RDD buffer per sample) was added directly onto the membrane, and the column was left at room temperature for 15 min. Following DNase I treatment, the column was treated with additional RW1 buffer (350 µl) and centrifuged and then washed with 500 µl 80% ethanol. Following the wash, the column with bound RNA was centrifuged for 2 min at 14,000 rpm to remove any traces of ethanol. Total RNA was eluted into a collection tube with two subsequent 50-µl aliquots DEPC-treated water (centrifugation at 14,000 rpm for 1 min). Total RNA concentration was determined by A_260nm reading on an Eppendorf BioPhotometer (Brinkmann Instruments).

cRNA Labeling and Expression Profiling
cDNA was produced from 5 µg total RNA by reverse transcription (RT) using Moloney murine leukemia virus (MMLV) RTase and then transcribed into cRNA by in vitro transcription (IVT) using T7 RNA polymerase. 5-(3-Aminoallyl)uridine 5'-triphosphate (Sigma) was incorporated into cRNA in the IVT reaction. For cRNA labeling, the allylamine-derivatized cRNA products were reacted with N-hydroxysuccinimide esters of Cy3 or Cy5 dyes (Amersham Pharmacia Biotech, Piscataway, NJ) as described (3). Five micrograms of Cy-labeled cRNA from one leukocyte sample were mixed with the same amount of reverse color Cy-labeled product from a pool, which consisted of an equal amount of cRNA from each of seven individuals (men and women; smoking status unknown) unrelated to the study participants. Arrays were run on samples collected at visit 1 and on a subset of samples collected from 17 individuals at visit 2. The resulting labeled probes were hybridized to hu25k oligonucleotide microarrays. All hybridizations were done in duplicate with fluor reversal on two microarrays to compensate for potential biases due to the different chemical properties of Cy3 and Cy5 dyes. The arrays were scanned to detect the level of gene expression for 21,000 genes as described previously (4).

Data Analysis
For the analyses, we used gene expression data from 65 individuals (32 smokers and 33 nonsmokers) as a training set to select and optimize a set of reporter genes. We used the array data for 20 other participants (10 smokers and 10 nonsmokers, equally distributed by sex) as a test set (TEST1) and the follow-up visit 2 samples for 17 of these 20 individuals as a separate test set (TEST2). Thus, TEST1 and TEST2 were derived from the same individuals at two time points, chosen as such to determine whether the signature was stable over the time.

This study relied on a cRNA pool of seven unrelated individuals rather than a pool of all of the study participants. The latter approach is often used in array work and affords greater power to discriminate between groups; however, it also requires that a sufficient amount of RNA be available from all participants to contribute to the pool. In our study, the smoker/nonsmoker discrimination power has the potential to be limited due to the gene signature resulting from the unrelated pool. To reduce this undesirable effect, we preprocessed the three sets (training set, TEST1, and TEST2) by subtracting the average gene expression value from each gene across individuals in the training set (n = 65). This preprocessing is similar to centering or re-ratioing of the samples to the newly formed mathematical pool formed as a geometric mean of all individual samples within the set. This established the same baseline across the three sets.

Using data from the training set, first, we selected a set of signature genes that satisfied the following criteria: |xdev| > 2.5 (P value 0.01) and abs(log₁₀ ratio) > 0.3 in more than three individuals. The resulting 861 genes were pared down further to 857 by requiring them to have more than 95% of valid entries. These genes were ordered in descending fashion based on their absolute correlation coefficient to the plasma cotinine level. We began optimization of the reporter set by selecting the top 5 reporters and incrementing from the top of the list one reporter at a time. At each step, one profile was left out of our training set. The remaining 64 profiles were used to compute smoker and nonsmoker expression templates by averaging gene expression values across smokers and nonsmokers in the remaining set. The profile was classified as smoker or nonsmoker if it correlated more strongly to smokers or nonsmokers, respectively. This procedure was carried out for all 65 profiles in the training set and the total misclassification was computed as the sum of Type 1 and Type 2 errors. The optimization was designed to pick the set of reporters for which the total misclassification was minimal.

The reporters were then used to predict smoking status in the entire training set and test sets (TEST1 and TEST2) using the smoker and nonsmoker expression templates computed from the training set as described above. We determined the sensitivity (i.e., proportion of true positives correctly identified by the test) and specificity (i.e., proportion of true negatives that are correctly identified by the test) of the gene profile using cotinine as the gold standard to define smoking status. Because smoking is often associated with other behaviors (e.g., higher alcohol intake, lower exercise) which themselves have the capacity to influence gene expression, we attempted to determine whether these other exposures affected expression of the identified cotinine-associated genes. Given the small sample size, we used two approaches. First, we stratified individuals by sex, exercise, and aspirin use and determined whether we could discriminate between smokers and nonsmokers within the subgroups. Second, we determined the power of the reporter genes to predict differences in sex, exercise frequency, and aspirin use. (Additional data analysis details are provided in Appendix A.)

	Results

Top Abstract Introduction Methods Results Discussion Appendix A References

 
We obtained blood samples and peripheral leukocyte RNA samples without evidence of degradation on 41 self-identified smokers (16 women and 25 men) and 44 self-identified nonsmokers (21 women and 23 men). We used plasma cotinine concentrations to define smoking status. There were 6 individuals whose self-identification of smoking status was discrepant with this classification. On the basis of plasma cotinine concentrations, three of the self-identified nonsmokers who had detectable cotinine (>30 ng/ml) were reclassified as smokers, and three of the smokers who had plasma cotinine concentrations of 0 ng/ml were reclassified as nonsmokers. Participant characteristics are presented in Table 1.

View larger version (19K): [in this window] [in a new window]   Fig. 1. Optimization of the number of reporter genes that minimized Type 1 and Type 2 error resulted in 36 reporter genes.

View larger version (69K): [in this window] [in a new window]   Fig. 2. Expression data matrices across 36 optimal reporter genes for cotinine exposure and categorization of individuals in the training set (A; n = 65) and TEST1 (B; n = 20) and TEST2 (C; n = 17) test sets using the 36 reporter genes. Reporter genes are ordered in the matrix based on the expression difference between nonsmoker and smoker templates in the training set. The smoking status for each participant is shown in the right panel: white indicates participants who were classified as smokers based on plasma cotinine concentrations >30 ng/ml; black indicates participants classified as nonsmokers with plasma cotinine of 0 ng/ml. The red line indicates equal correlation of profiles to both the smoker and nonsmoker templates. The color range of log ratios is from -0.3 to 0.3 for bright green and bright red, respectively.

View larger version (13K): [in this window] [in a new window]   Fig. 3. For each profile P in the training and test sets, a cosine correlation to the TSM and TNSM, weighted by the spread of the template, is computed (see Appendix A). This generates two correlation values for each profile SM and NSM to smoker and nonsmoker templates, respectively. The classification metric is defined as d = NSM - SM. The scatter plots of this metric against plasma cotinine concentration in the three data sets, (A) training set, (B) TEST1, and (C) TEST2, are displayed. In theory, the smokers should be in the part of the plot where d < 0, while the nonsmokers where d > 0. The points that fail this classification are the false positive and false negatives. The six participants in the training set for whom self-report and plasma cotinine concentrations were discrepant are represented as triangles (self-reported nonsmokers with detectable cotinine) and squares (self-reported smokers with 0 ng/ml cotinine).

	Discussion

Top Abstract Introduction Methods Results Discussion Appendix A References

 
With this study, we have demonstrated that we can distinguish between individuals exposed and unexposed to tobacco smoke on the basis of mRNA expression in peripheral leukocytes. Furthermore, we have shown that active exposure to tobacco smoke is associated with a biologically relevant mRNA expression signature. Previously, several studies have shown that, experimentally, in vitro and in animals, single exposures [e.g., heat shock (6) and 17ß-estradiol (7)] alter gene expression in peripheral leukocytes. Our study shows that the use of expression patterns can be expanded to identify a complex exposure in human observational studies.

In this study, we examined gene expression in total leukocytes, rather than specific cellular subsets. Total numbers of peripheral leukocytes characteristically have been shown to differ by smoking status (8, 9), and differential counts have been reported to differ by number of cigarettes smoked (8) or to remain unchanged (10). Recently, interindividual variation in expression patterns also has been shown to be influenced by peripheral leukocyte composition (11). Consequently, in theory, observed differences in gene expression between smokers and nonsmokers may reflect differences in the percentage of each cell type. In our sample, there were no significant differences in the major cell type distributions by smoking status (Table 1); thus, this factor is unlikely to be a major explanation for the differences we observed in gene expression.

Many of the reporter genes that were associated with plasma cotinine concentrations in our study have not been described in relation to cigarette smoking. However, several reporter genes, such as IL-1ß and CYP1B1, are associated with the pathophysiology of smoking-induced injury. Exposure to cigarette smoke is an oxidant burden, not only at the initial site of contact (i.e., respiratory epithelium), but also in peripheral leukocytes, where oxidative damage and polycyclic aromatic hydrocarbon-adducts are readily detectable in smokers (12, 13). Few studies have examined the effects of cigarette smoking on leukocyte gene expression in circulation; however, smoking has well-established immune and inflammatory effects in bronchial tissue and bronchoalveolar macrophages. Differences in cytokine profiles in bronchoalveolar lavage fluid, including IL-1ß, have been reported for smokers and nonsmokers (14, 15). Cigarette smoke up-regulates expression of proinflammatory cytokines such as IL-1ß, which in turn induce expression of a wide variety of genes. Similarly, expression of the biotransformation enzyme CYP1B1 is increased with active cigarette smoking and polycyclic aromatic hydrocarbon exposure in bronchoalveolar macrophages (16). Although one study, using quantitative reverse transcription-PCR, was unable to detect an effect of smoking on CYP1B1 mRNA levels in blood mononuclear cells (17), our results suggest that, with the higher number of cigarettes smoked by participants in our study, CYP1B1 in total leukocytes is responsive to cigarette smoke exposure.

The physiological and biochemical effects associated with cigarette smoking and the constituents of tobacco smoke support the relevance of several other reporter genes. For example, 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase-3 (PFKFB3), positively associated with plasma cotinine concentrations, is a key enzyme that regulates glycolysis in mammalian cells. It is induced by hypoxia and IL-1ß through stabilization of hypoxia-inducible factor-1 (18). Iduronate 2-sulfatase, an enzyme responsible for degradation of dermatan sulfate and heparan sulfate, constituents of mucopolysaccharides, was inversely associated with plasma cotinine. Reduced expression of iduronate 2-sulfatase could extend the half-life of mucins in smokers.

From a biological standpoint, many of the identified genes associated with smoking in this study, especially those related to immune function and inflammation, can be attributed readily to cigarette smoke exposure. The absence of data on the association between expression of particular genes and cigarette smoking may reflect a lack of information on some of the pathways in relation to smoke exposure; it also may reflect the complexity of studying the effect of tobacco smoke on gene expression in intact humans. First, the pathways affected by tobacco smoke are numerous and interconnected, and some genes may be influenced by the expression of other genes. Second, tobacco smoke is a complex mixture of compounds, each of which is likely to have multiple targets (2). Third, tobacco use clusters with other human behaviors, which, if not acknowledged and accounted for, leads to confounding and biased estimates of association. Thus, the expression profile associated with smoking may also be associated with a pattern of other behaviors (e.g., higher alcohol intake, lower exercise) that by themselves may influence gene expression. Nonetheless, these findings show that it is reasonable to explore whether these other exposures and behaviors also have readable signatures. Proper study of the relationships between exposures and gene expression will include both observational and experimental studies in humans. Such studies, optimally, will have sufficient power to detect the whole spectrum of differences and sufficient data on other exposures to allow proper control of confounding (5). Finally, our findings raise interesting, but as yet unresolved, implications for the use of readily obtained surrogate tissues in studying the relationship between exposures and biology relevant to the progression of human cancer.

	Appendix A

Top Abstract Introduction Methods Results Discussion Appendix A References

 
Supplementary Material

1. Re-ratioing to the mean template from 65 individuals
   For a given gene, let X_I and _I denote its expression log-ratio for person I = 1,...,N, N = 65 and its error, respectively. For the expression of each gene, one could compute the mean and its error across 65 individuals as
   

   
   

   
   Equation 1
   We preprocess each gene's expression by subtracting the mean µ and propagating error as follows
   

   
   

   
   Equation 2where Y_I and _I are the expression and the error of the gene with respect to its average expression µ. The P value for the significance of expression is computed as
   

   
   

   
   Equation 3
   

2. Correlation to the cotinine profile
   For each gene, correlation to the cotinine profile was computed using Pearson correlation coefficient.
   
3. Significance of the selected predictor genes
   Thirty-six reporter genes were checked for significance by performing Monte Carlo simulations. The correlations of the 36 reporter genes to the cotinine profile were equal or above 0.34. The originally selected 861 signature genes were correlated to the randomly permuted cotinine profile using 10,000 Monte Carlo runs. For each such run, we counted how many genes out of 861 had correlation coefficients greater or equal to 0.34. We then counted the number of runs that resulted in the number of genes greater or equal to 36—the number of genes that had positive predictive value. One hundred ten of the 10,000 Monte Carlo runs resulted in 36 genes with a correlation coefficient 0.34, yielding a P value of 0.01 (figure).
   
4. Similarity metric for profile classification
   For the classification, we have used correlation-based similarity metric. Smoker and nonsmoker expression templates and their spreads have been computed as averages over smokers and nonsmokers in the training set as (example only for smokers)
   

   
   

   
   

   
   Equation 4where LR_ij are log ratios of gene i in the experiment j and we have selected 36 genes based on the optimization process.
   For each profile P in the training and test sets, a cosine correlation to the T^SM and T^NSM, weighted by the spread of the template, is computed as
   

   
   Equation 5
   Thus, we have two correlation values for each profile ^SM and ^NSM to smoker and nonsmoker templates, respectively. We can define the classification metric as
   

   
   Equation 6
   The scatter plot of this metric against cotinine level in all three data sets is displayed in the following figures. It is clear that most of the smokers should be in the part of the plot where d < 0, while most nonsmokers where d > 0. The points that fail this classification were our false positive and false negatives (see Fig. 3). Note, that 36 genes as well as smoker and nonsmoker templates are chosen from the training set and are used in the test sets.
   

5. Cross-validation
   Using the originally defined 857 signature genes, cross-validation was performed based on the "leave out one profile" approach. At each step of cross-validation, a single expression profile was left out. The remaining 64 profiles were used to re-select reporters that had correlation coefficients to the cotinine level of greater or equal to 0.34. Smoker and nonsmoker templates were computed based on the remaining 64 profiles using this new reporter set. The profile left out was classified using the similarity metrics described above. The result of such cross-validation is a more conservative estimate of classification performance in the training set because it avoids the overtraining problem, that is, classifying based on the reporters and templates chosen from the same set.
   

View larger version (17K): [in this window] [in a new window]   The originally selected 861 signature genes were correlated to the randomly permuted continine profile using 10,000 Monte Carlo runs. For each run, we determined how many genes out of 861 had a correlation coefficient 0.34 and counted the number of runs that resulted in the number of genes to 36—the number of genes that had positive predictive value.

	Acknowledgments

	Footnotes

 
Grant support: Fred Hutchinson Cancer Research Center and Rosetta Inpharmatics, LLC.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: J.W. Lampe, S.B. Stepaniants, and M. Mao contributed equally to this work.

Received 7/ 2/03; revised 10/16/03; accepted 11/12/03.

	References

Top Abstract Introduction Methods Results Discussion Appendix A References

 

1. Hughes TR, Marton MJ, Jones AR, et al. Functional discovery via a compendium of expression profiles. Cell, 2000;109–26.
2. IARC. IARC monographs on the evaluation of the carcinogenic risk of chemicals to humans. Volume 38. Tobacco smoking. Switzerland: IARC; 1986.
3. Hughes TR, Mao M, Jones AR, et al. Expression profiling using microarrays fabricated by an ink-jet oligonucleotide synthesizer. Nat Biotechnol, 2001;19:342–7.[CrossRef][Medline]
4. van't Veer LJ, Dai H, van de Vijver MJ, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature, 2002;415:530–6.[CrossRef][Medline]
5. Potter JD. At the interfaces of epidemiology, genetics and genomics. Nat Rev Genet, 2001;2:142–7.[CrossRef][Medline]
6. Sonna LA, Gaffin SL, Pratt RE, Cullivan ML, Angel KC, Lilly CM. Effect of acute heat shock on gene expression by human peripheral blood mononuclear cells. J Appl Physiol, 2002;92:2208–20.[Abstract/Free Full Text]
7. Rockett JC, Kavlock RJ, Lambright CR, et al. DNA arrays to monitor gene expression in rat blood and uterus following 17ß-estradiol exposure: biomonitoring environmental effects using surrogate tissues. Toxicol Sci, 2002;69:49–59.[Abstract/Free Full Text]
8. Schwartz J, Weiss ST. Host and environmental factors influencing the peripheral blood leukocyte count. Am J Epidemiol, 1991;134:1402–9.[Abstract/Free Full Text]
9. Jensen EJ, Pedersen B, Frederiksen R, Dahl R. Prospective study on the effect of smoking and nicotine substitution on leucocyte blood counts and relation between blood leucocytes and lung function. Thorax, 1998;53:784–9.[Abstract/Free Full Text]
10. Malech HL, Gallin JI. Neutrophils in human diseases. N Engl J Med, 1987;317:687–702.[Medline]
11. Whitney AR, Diehn M, Popper SJ, et al. Individuality and variation in gene expression patterns in human blood. Proc Natl Acad Sci USA, 2003;100:1896–901.[Abstract/Free Full Text]
12. Tang D, Santella RM, Blackwood AM, et al. A molecular epidemiological case-control study of lung cancer. Cancer Epidemiol Biomarkers & Prev, 1995;4:341–6.[Abstract]
13. Kiyosawa H, Suko M, Okudaira H, et al. Cigarette smoking induces formation of 8-hydroxydeoxyguanosine, one of the oxidative DNA damages in human peripheral leukocytes. Free Radic Res Commun, 1990;11:23–7.[Medline]
14. Kuschner WG, D'Alessandro A, Wong H, Blanc PD. Dose-dependent cigarette smoking-related inflammatory responses in healthy adults. Eur Respir J, 1996;9:1989–94.[Abstract]
15. Mikuniya T, Nagai S, Tsutsumi T, et al. Proinflammatory or regulatory cytokines released from BALF macrophages of healthy smokers. Respiration, 1999;66:419–26.[Medline]
16. Piipari R, Savela K, Nurminen T, et al. Expression of CYP1A1, CYP1B1 and CYP3A, and polycyclic aromatic hydrocarbon-DNA adduct formation in bronchoalveolar macrophages of smokers and non-smokers. Int J Cancer, 2000;86:610–6.[CrossRef][Medline]
17. Dassi C, Signorini S, Gerthoux P, Cazzaniga M, Brambilla P. Cytochrome P450 1B1 mRNA measured in blood mononuclear cells by quantitative reverse transcription-PCR. Clin Chem, 1998;44:2416–21.[Abstract/Free Full Text]
18. Minchenko A, Leshchinsky I, Opentanova I, et al. Hypoxia-inducible factor-1-mediated expression of the 6-phoshofructo-2-kinase/fructose-2,6-bisphosphatase-3 (PFKFB3) gene: its possible effect in the Warburg effect. J Biol Chem, 2002;277:6183–7.[Abstract/Free Full Text]

This article has been cited by other articles:

				I-M. Wang, S. Stepaniants, Y. Boie, J. R. Mortimer, B. Kennedy, M. Elliott, S. Hayashi, L. Loy, S. Coulter, S. Cervino, et al. Gene Expression Profiling in Patients with Chronic Obstructive Pulmonary Disease and Lung Cancer Am. J. Respir. Crit. Care Med., February 15, 2008; 177(4): 402 - 411. [Abstract] [Full Text] [PDF]

				G. Siest, E. Jeannesson, J.-B. Marteau, A. Samara, B. Marie, M. Pfister, and S. Visvikis-Siest Transcription Factor and Drug-Metabolizing Enzyme Gene Expression in Lymphocytes from Healthy Human Subjects Drug Metab. Dispos., January 1, 2008; 36(1): 182 - 189. [Abstract] [Full Text] [PDF]

				P. M. Webb, M. A. Merritt, G. M. Boyle, and A. C. Green Microarrays and Epidemiology: Not the Beginning of the End but the End of the Beginning... Cancer Epidemiol. Biomarkers Prev., April 1, 2007; 16(4): 637 - 638. [Full Text] [PDF]

				D. M.v. Leeuwen, E. v. Agen, R. W.H. Gottschalk, R. Vlietinck, M. Gielen, M. H.M.v. Herwijnen, L. M. Maas, J. C.S. Kleinjans, and J. H.M.v. Delft Cigarette smoke-induced differential gene expression in blood cells from monozygotic twin pairs Carcinogenesis, March 1, 2007; 28(3): 691 - 697. [Abstract] [Full Text] [PDF]

				A. C. Borczuk and C. A. Powell Expression Profiling and Lung Cancer Development Proceedings of the ATS, January 1, 2007; 4(1): 127 - 132. [Abstract] [Full Text] [PDF]

				F. R. Sharp, H. Xu, L. Lit, W. Walker, M. Apperson, D. L. Gilbert, T. A. Glauser, B. Wong, A. Hershey, D.-Z. Liu, et al. The Future of Genomic Profiling of Neurological Diseases Using Blood Arch Neurol, November 1, 2006; 63(11): 1529 - 1536. [Abstract] [Full Text] [PDF]

				I. Osman, D. F. Bajorin, T.-T. Sun, H. Zhong, D. Douglas, J. Scattergood, R. Zheng, M. Han, K. W. Marshall, and C.-C. Liew Novel blood biomarkers of human urinary bladder cancer. Clin. Cancer Res., June 1, 2006; 12(11): 3374 - 3380. [Abstract] [Full Text] [PDF]

				H. Chon, M. C. Verhaar, H. A. Koomans, J. A. Joles, and B. Braam Role of Circulating Karyocytes in the Initiation and Progression of Atherosclerosis Hypertension, May 1, 2006; 47(5): 803 - 810. [Abstract] [Full Text] [PDF]

				V. Dumeaux, J. Johansen, A.-L. Borresen-Dale, and E. Lund Gene expression profiling of whole-blood samples from women exposed to hormone replacement therapy. Mol. Cancer Ther., April 1, 2006; 5(4): 868 - 876. [Abstract] [Full Text] [PDF]

				C. P. Wild Complementing the Genome with an "Exposome": The Outstanding Challenge of Environmental Exposure Measurement in Molecular Epidemiology Cancer Epidemiol. Biomarkers Prev., August 1, 2005; 14(8): 1847 - 1850. [Full Text] [PDF]

				J.-B. Marteau, S. Mohr, M. Pfister, and S. Visvikis-Siest Collection and Storage of Human Blood Cells for mRNA Expression Profiling: A 15-Month Stability Study Clin. Chem., July 1, 2005; 51(7): 1250 - 1252. [Full Text] [PDF]

				D. M. van Leeuwen, R. W. H. Gottschalk, M. H. van Herwijnen, E. J. Moonen, J. C. S. Kleinjans, and J. H. M. van Delft Differential Gene Expression in Human Peripheral Blood Mononuclear Cells Induced by Cigarette Smoke and Its Constituents Toxicol. Sci., July 1, 2005; 86(1): 200 - 210. [Abstract] [Full Text] [PDF]

				R. D. Fannin, J. T. Auman, M. E. Bruno, S. O. Sieber, S. M. Ward, C. J. Tucker, B. A. Merrick, and R. S. Paules Differential gene expression profiling in whole blood during acute systemic inflammation in lipopolysaccharide-treated rats Physiol Genomics, March 21, 2005; 21(1): 92 - 104. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lampe, J. W. Articles by Potter, J. D. Search for Related Content PubMed PubMed Citation Articles by Lampe, J. W. Articles by Potter, J. D.