American Journal of Respiratory and Critical Care Medicine

Rationale: Approximately 60 to 70% of patients with pulmonary sarcoidosis have disease that resolves spontaneously; the rest follow a chronic course with varying levels of fibrosis. It is unclear why some patients progress and if treatment affects outcome.

Objectives: To determine differential gene expression profile in lungs of patients with self-limiting sarcoidosis compared to those with progressive-fibrotic disease, and to analyze the biological relevance of these differentially expressed genes.

Methods: We examined microarray expression of 26,626 genes in transbronchial biopsies of granulomatous areas in lungs of patients with active but self-limiting (n = 8) versus those with active, progressive (± fibrotic) pulmonary disease (n = 7).

Measurements and Main Results: Three hundred thirty-four genes were differentially expressed between the two groups (P < 0.01, Bayesian moderated t test). Gene Set Enrichment Analysis showed over-representation of gene-sets (defined by Gene Ontology) related to host immune activation, proliferation, and defense, among genes up-regulated in the progressive-fibrotic group (FDR q < 0.0001 for the top 43 gene sets), and a marked enrichment of, and similarity in gene expression profiles between, progressive-fibrotic sarcoidosis and hypersensitivity pneumonitis (HP), (q < 0.001), but not idiopathic pulmonary fibrosis (IPF).

Conclusions: The findings suggest that patients with progressive/fibrotic pulmonary sarcoidosis have intense immune activity related to host defense in their lungs, with processes more similar to HP than IPF. The study also demonstrates that transbronchial lung biopsy samples can provide good-quality RNA for gene expression profiling, supporting its potential use as a prognostic classifier for pulmonary sarcoidosis.

Scientific Knowledge on the Subject

Approximately 60 to 70% of patients with pulmonary sarcoidosis have disease that resolves spontaneously; the rest follow a chronic course with varying levels of fibrosis. It is unclear why some patients progress and if treatment affects outcome.

What This Study Adds to the field

The findings suggest that the disease processes in the lungs of patients with progressive-fibrotic pulmonary sarcoidosis relate to host defense, with gene expression patterns more similar to hypersensitivity pneumonitis than to idiopathic pulmonary fibrosis.

Sarcoidosis is a disease of unknown etiology, characterized by granulomatous deposits and an intense T cell activity. Almost any organs can be affected, but the lungs are by far the most common; approximately 95% of patients have computed tomographic evidence of deposits in the lungs (1, 2). Significant advances have been made to characterize the immunological features of the condition, but the central immunological mechanisms remain unclear. There is evidence pointing to overactivity of T helper 1–biased CD4 T cells as the initial abnormality leading to activation of macrophages and downstream formation of granuloma (2, 3). It is also possible that the macrophage is the primary cellular abnormality, and that an inability to degrade an intracellular antigen causes the formation and persistence of granuloma. Overall, there is general agreement that the condition is caused by a complex genetic susceptibility to granulomagenesis, triggered by an unknown inhaled antigen. Recent genome-wide association studies and directed single nucleotide polymorphism (SNP) searches have identified BTNL-2 and ANXA11 (4, 5) as susceptibility genes, both of which are related to control of T cell activity (6, 7).

Most patients with pulmonary sarcoidosis have a good prognosis: the immune disturbances and granulomatous deposits resolve spontaneously over 2 to 5 years (2, 810). However, about 30% of patients have progressive disease leading to varying degrees of pulmonary fibrosis. Currently almost nothing is known of the cause of this progression in disease. It is also unclear whether early treatment in this group of patients could prevent fibrosis.

In this study we obtained transbronchial biopsy samples from granulomatous areas of lungs in two groups of patients with different disease course, and we report the use of a powerful analytical procedure (Gene Set Enrichment Analysis) to identify biological processes in the lungs associated with progressive, fibrotic disease. Some of the results of these studies have been previously reported in the form of an abstract (11).

Study Population

Fifteen patients with histology-proven pulmonary sarcoidosis were recruited for the study. All patients were white, had never smoked, were not on any inhaled or oral medications at the point of sampling, and did not present with Loefgren syndrome. Patients had only one other organ of involvement (eye or skin). All patients had active disease at the point of sampling as characterized by presence of nodularity on high-resolution thoracic computed tomography (HRCT) scan and at least one of the following: respiratory or constitutive symptoms, peripheral lymphopenia (1214) or raised serum angiotensin-converting enzyme (ACE). In some patients, mild fibrosis was evident on thoracic HRCT. Transbronchial biopsy was performed at the point of presentation and patients were then followed up for 2 years. Biopsy was targeted to areas of disease (nodularity or fibro-nodularity) using CT images. Half the sample from each patient was sent for histologic analysis for granuloma and the other half (two to three biopsies) immediately transferred to RNA-later. Patients were classified into nodular, self-limiting (N-SL) or progressive, fibrotic (P-F) pulmonary disease groups on the basis of persistence of symptoms and changes on chest X-ray over 2 years. In the former, patients had minor symptoms (occasional cough), normal formal pulmonary function tests (FEV1, FVC, and transfer factor of lungs for carbon monoxide or transfer coefficient for carbon monoxide) but nodularity in typical distribution in the lung interstitium on HRCT scanning; and no change in chest X-ray or symptoms in the ensuing 2 years. In the P-F group, patients presented with fibro-nodular changes on CT scans, an abnormal spirometry or KCO, and showed persistent or progressive respiratory symptoms over 2 years.

The characteristics of the patient groups from the study by Selman and coworkers are as described in their article (15) and also provided in the Supplementary Data.

The first eight patients fitting the description above were entered into the study and processed as a set (Set 1; Patient 1–4, 9–12; Table 1). The next eight patients fitting the same characteristics (Set 2) were processed as a second set. All patients were followed up for 2 years before final confirmation of groupings into N-SL or P-F.


Patient No.



HRCT Thorax

ACE (normal 18–55 U/l)

Lymphocyte Count (x109/L) (normal 1–4)

FEV1/FVC (L) (% predicted) at point of sampling

FEV1/FVC (L)(% predicted) at End of Study

KCO (% predicted) at Point of Sampling

KCO (% predicted) at End of Study

Disease Course over 2 yr

Presenting Symptoms
Nodular-self limiting group159MNodularity, subcarinal, supraclavicular, and mediastinal lymphadenopathy. One area of consolidation.520.685.0/6.7 (116/122)4.7/6.3 (109/115)90%98%Cough and lethargy. No worsening in disease or CXR.Minor cough, lethargy
236FBHL, nodularity651.603.3/3.9 (111/110)3.4/3.9 (111/111)106%91%No symptoms apart from occasional cough. Nodularity on CXR resolved.None
348MBHL and ML, nodularity691.273.4/4.9 (92/108)3.2/4.9 (90/109)105%104%Occasional cough, nodularity on CXR unchangedOccasional cough
438FBHL and ML; multifocal consolidation, nodularity690.942.4/3.1 (100/100)2.5/3.2 (97/94)98%102%Persistent cough but no worsening in CXROccasional cough
535MBHL. Multiple tiny nodules, mid and upper zone predominance; peribronchovascular distribution. No fibrosis.700.472.8/4.4 (85/100)3.0/4.6 (72/92)125%128%Cough resolved, no change in CXROccasional cough
643MBHL and ML; a few scattered nodules and some fissural nodularity431.304.3/5.5 (103/106)4.3/5.5 (103/106)102%102%No symptoms, no change in CXRNone
751MWidespread nodularity in bronchovascular and fissural distribution, most marked in upper zones; small lymphadenopathy711.703.5/4.4 (86/94)3.5/4.0 (94/86)97%97%Cough resolved, nodularity resolvedOccasional cough
851FExtensive ML, numerous small nodules740.962.6/3.6 (99/116)2.7/3.6 (100/116)92%92%No symptoms, nodules persistNone
Progressive-fibrotic group941FBHL and ML, nodularity300.592.8/3.8 (96/110)3.1/4.0 (106/118)71%71%Progressive breathlessness requiring Prednisolone but responsive; nodularity reducedCough at sampling, then progressive breathlessness
1051MExtensive BHL and ML, nodularity, conglomerate consolidation and minor fibrosis520.473.2/3.8 (92/89)2.3/3.0 (67/72)125%104%Progressive breathlessness, unprovoked pulmonary embolism; some progression in fibrosis in spite of PrednisoloneBreathless, lethargy, progressive symptoms
1162FNodularity, conglomerate masses, upper lobe fibrosis, minor ML1341.901.4/1.7 (69/72)1.9/2.4 (97/101)112%102%Symptoms worsened, Prednisolone started after sampling; stable CXR abnormalitiesBreathless
1235MNodularity, ML, fibrosis1180.902.2/2.7 (50/62)2.5/3.4 (57/65)109%97%Symptoms worsened, Prednisolone started after sampling; stable nodularity and fibrosisBreathless
1351MInnumerable nodules, predominantly mid-zone. BHL and ML.1070.462.9/4.1 (93/106)2.9/3.9 (94/102)127%112%Symptoms worsened, nodularity on CXR increased; Prednisolone started after sampling and nodularity reducedLethargy and chest pain
1455FBHL and ML with calcification, bilateral conglomerate masses and nodules, perihilar mainly upper lobes, no fibrosis1100.571.2/2.5 (41/74)1.9/3.2 (66/96)76%83%Symptoms worsened, Prednisolone started after sampling; nodularity reduced after PrednisoloneProgressive breathlessness

Fissural nodularity with areas of fibrosis in lower lobes; small BHL.
0.8/1.5 (48/71)
0.9/1.5 (54/71)
Symptoms worsened, Prednisolone started after sampling; nodularity improved, fibrosis stable
Breathlessness and cough

Definition of abbreviations: ACE = angiotensin-converting enzyme; BHL = bilateral hilar lymphadenopathy; CXR = chest X-ray; KCO = transfer coefficient for carbon monoxide; HRCT = high-resolution computed tomography; ML = mediastinal lymphadenopathy.

All patients gave informed consent, and the study was approved by the Oxfordshire Research Ethics Committee.

Gene Expression Profiling

Gene expression profiling was performed using the Affymetrix Gene ST 1.0 array (High Wycombe, UK), which provides probes for whole transcript coverage, and therefore a more complete picture of gene expression, for approximately 29,000 genes from the human genome. Total RNA from the lung biopsies was extracted within 24 hours using the Qiagen RNeasy Mini kit (Crawley, UK) according to manufacturer's instructions. RNA quality was assessed using a high-resolution electrophoresis bioanalyser system (Agilent Technologies, Wokingham, UK). Isolated total RNA was then taken through the Affymetrix preparation protocol (; see also the online supplement).

Raw data were processed using Affymetrix Power Tools ( to generate RMA-normalized gene level intensities. Further analysis was performed using the R statistical program ( and BioConductor packages (16). Quality control procedures included a comparison of boxplots of RMA normalized expression values between samples, and hierarchical clustering of all samples to identify any potential outliers. Based on these and the assessment of RNA quality, 15 of the 16 samples were deemed suitable for further analysis. Raw microarray data has been deposited with GEO (Accession number GSE19976). All data collected and analyzed in the experiments adhere to the Minimal Information About a Microarray Experiment (MIAME) guidelines.

Statistical Analysis

Before statistical analysis, the mean intensity was computed for each probeset and the lowest 20% were removed, leaving a total of 26,626 probesets (genes). A Bayesian moderated t test was applied to identify differentially expressed genes between P-F and N-SL using the “limma” (linear models for microarray analysis) package (17) from BioConductor. The Benjamini–Hochberg false discovery rate (FDR) procedure was applied to correct for multiple testing (18). The inverse log of the absolute log2 fold change (FC) generated by limma was used to determine the FC between P-F and N-SL groups, with the direction of change indicated by the sign. Further visualization of the data using heatmaps and hierarchical clustering was performed using the TMeV4 software (19).

Gene set analysis methods (GSEA) are as described by Subramaniam and colleagues (20) and by Mootha and coworkers (21), and also in the online supplement, and were performed using web-based software (

qPCR Validation

For quantitative real-time PCR, cDNA was synthesized from 200 ng DNA-free RNA using the random priming High-Capacity cDNA Archive kit (Applied Biosystems) for all patients. Quantitative rt-PCR (qPCR) was performed using the Applied Biosystems 7500 Fast Real-Time PCR system following the manufacturer's instructions. Five random genes were selected for validation using SybrGreen gene expression assays (Applied Biosystems), based on their relative expression in the groups. Two genes (TATA-binding protein [TBP] and hypoxanthine phosphoribosyl-transferase 1 [HPRT]) were chosen as endogenous controls for the normalization of all target genes, as they were consistently expressed in microarray samples and not significantly different between the groups. Samples were run in triplicate along with a standard curve, to obtain reaction efficiency values. Data were then analyzed using a method described by Pfaffl and colleagues (22).

Patient Demographics

Clinical features of patients are demonstrated in Table 1. Average age in the N-SL group was 45.1 (SD 8.5) years, and 52.0 (SD 11.6) years in P-F group; serum ACE was 64.1 (SD 10.8) and 85.1 (SD 41.5) U/l, respectively, P = 0.2 (Student t test); and circulating lymphocyte counts were 1.07 (SD 0.43) × 109/l and 0.79 (SD 0.50) × 109/l, respectively, P = 0.2 (Student t test). Although there was no statistical difference between the two groups for serum ACE levels and lymphocyte count; the trends in these parameters and the pulmonary function tests between the P-F and N-SL groups (worse in the P-F group) are in keeping with the clinical characteristic of symptomatic and progressive disease.

Histology of all samples (paired with samples that were submitted for RNA extraction) showed noncaseating granuloma, indicating that lung tissue around active lesions was obtained.

Quality Control of Samples from Transbronchial Biopsies

We obtained between 0.6 to 3 μg of RNA per patient, with no evidence of degradation (see Figure E1A in the online supplement), or significant difference in quality between the eight biopsy samples in the first set. In the second set, one patient had a RIN value of less than 6.5, and this sample was excluded. These findings are in keeping with those of one other study, which has also examined gene expression from transbronchial biopsy and shown good quality RNA extraction (23). There was a strong positive correlation between mean gene intensity in each group, as would be expected if the majority of genes are not differentially expressed (Figure E1B), and there was no systematic difference in the distribution of fluorescence intensity on the Gene Chips used (Figure E1C). There was also good concordance between qPCR and microarray data in the trend of change between N-SL and P-F groups for our randomly selected subset of genes (Figure E1D).

Since there were no systematic technical differences between the two sets of patients, analyses were performed, first on the two sets of patients independently (i.e., four N-SL versus four P-F patients in Set 1; and four N-SL versus three P-F in Set 2), then as a combined cohort of eight N-SL and seven P-F patients. Analysis of the combined dataset of eight N-SL and seven P-F patients is reported here; though highly similar results were obtained when the two sets of patients were considered independently.

Differentially Expressed Genes in Lungs from N-SL and P-F Patient Groups

We found 334 differentially expressed genes between the two groups of patients at a significance threshold of P < 0.01 (Bayesian moderated t test, unadjusted P value; Figure 1A). The vast majority of these genes (279) were up-regulated in the P-F group compared with the N-SL group. The presence of a clear set of genes showing differential expression suggests the possibility of using these genes as part of a signature profile to discriminate between these two groups of patients and eventually to predict patients that would fall into the P-F group. However, this study was not designed to address this question.

Genes Overexpressed in Lungs of P-F Compared with N-SL Pulmonary Sarcoidosis Comprise Predominantly Genes Related to Immune Activation and Host Defense

Unrestricted comparisons of very large numbers of variables between two groups introduce the problem of multiple testing, where the chance of making a Type I error (falsely identifying single genes as differentially expressed) is increased. In addition, relevant sets of genes can be lost to identification because the differences are modest relative to the noise inherent to the microarray technology. One method of overcoming these issues is to compare the groups of genes using predefined, biologically meaningful sets of genes. For this, we use Gene Set Enrichment Analysis (GSEA) to evaluate whether sets of genes, grouped together by biological processes, as defined in Gene Ontology (GO) (24, 25), show enrichment among the genes ranked according to differential expression between P-F and N-SL. This allows determination of whether a gene set is correlated with phenotypic distinction.

We found almost no enrichment of gene sets in genes up-regulated in the N-SL group (only two: histone modification gene set and covalent chromatin modification gene set), but a very large number of gene sets were highly significantly enriched among genes up-regulated in the P-F group (114 sets with FDR q value < 0.01; and top 43 sets q < 0.0001) (Table E2). In a higher level of evaluation, the genes within the gene set that contribute most highly to the enrichment are extracted, since these are the genes most likely to participate in the biological process within that gene set. This subset is known as the “leading-edge” genes. A graphical representation of GSEA results for the five most significant GO-based gene sets is shown in Figure 2A. Due to the hierarchical nature of the gene ontology, GSEA results can include several closely related categories. Based on the overlap among leading edge genes and the GO hierarchy, the 114 gene sets with FDR q values < 0.01 were grouped into 10 major categories (Figure 2B and Table E3). This analysis showed that specific genes predominantly associated with immune activity are induced in the P-F group compared with the N-SL group. This is also the case when the two sets of patients are analyzed sequentially (independent of each other) (shown in Figure E2). In particular, genes involved in leukocyte activation and differentiation, and cytokine production, were overexpressed in P-F patients compared with the N-SL group. Other major processes that were found to be up-regulated in the P-F group included intracellular signaling (NF-κB and JAK-STAT cascades) and categories related to cell life (apoptosis, cell cycle, cell proliferation, and homeostasis) (Figure 2B and Table E3). Together, these results suggest that immune activation is stronger in the P-F group relative to N-SL patients; and that the differentially regulated genes shown in Figure 1A belong to biologically relevant gene sets.

Genes Up-Regulated in the P-F Group of Patients Are Significantly Enriched for Genes Up-Regulated in Hypersensitivity Pneumonitis

Since one cause of disease progression to fibrosis in patients with pulmonary sarcoidosis is the possession of genes that predispose to pulmonary fibrosis, we hypothesized that genes up-regulated in lungs of patients with pulmonary fibrosis might be enriched in this group of patients. To test this, we created two additional gene sets based on genes identified by Selman and colleagues when they compared gene expression profile in lungs of patients with idiopathic pulmonary fibrosis (IPF) and hypersensitivity pneumonitis (HP) (15). The IPF gene set consisted of 355 genes found up-regulated in IPF compared with HP and the HP gene set consisted of 598 genes found up-regulated in HP compared with IPF. Using GSEA, we questioned whether either of these gene sets was enriched at the extreme ends of the list of all genes ranked to correlate gene intensity with phenotype (N-SL and P-F) in our patients with sarcoidosis. We found no enrichment of the IPF gene set among the genes up-regulated in either group (Normalized Enrichment Score [NES] = −1.06 q = 0.48). Unexpectedly, the HP gene set was highly significantly enriched among genes that show higher expression in the P-F group (NES = −2.47, q < 0.0001) (Figure 3A). This result shows that a similar set of genes are overexpressed in HP relative to IPF, as are overexpressed in P-F compared with N-SL. These genes could be overexpressed in HP and P-F as a result of specific induction as part of the disease process, or as a result of down-regulation in the comparison disease (IPF or N-SL). However, the fact that immune response genes are enriched among the genes up-regulated in both HP and P-F suggests that an active immune response is common to both diseases. A heatmap showing expression of the differentially regulated genes between IPF and HP in the N-SL and P-F groups (and in the patients with IPF and HP) demonstrates similarity in profile in P-F group to HP (Figure 3B, upper section). No correlation to either HP or IPF was observed for the N-SL group (Figure 3B).

In this article, we show that (1) genes up-regulated in lung samples obtained from patients with progressive-fibrotic sarcoidosis, compared with self-limiting disease, comprised predominantly genes involved in host defense and immune responses; and (2) these genes are enriched with genes up-regulated in HP in relation to IPF. Although the number of patients in the study was relatively small (though not so for microarray studies), one strength is the reproducibility of the findings in two separate and independent studies performed sequentially (Figure E2).

These findings are derived from a powerful analytical approach, which assesses the differential expression of sets of biologically relevant genes rather than individual transcripts. This is an important distinction in all studies, but particularly relevant where smaller sample sizes are involved, since the signal to noise ratio of the array data may lead to false conclusions about individual gene expression values (26). In addition, very few genes act in isolation, and complex genetic diseases like sarcoidosis are likely to be caused by groups of functionally related genes that affect immunological processes or pathways.

In our study, we compared the differential expression to data from the study by Selman and colleagues because this was a well-characterized group of patients and one of very few studies that examined gene expression in the lungs. HP can be a heterogeneous disease; however, in the Selman study, this was a well-defined group of patients where the hypersensitivity was to avian antigens (pigeons) and the diagnosis established according to criteria that include known exposure to pigeon antigens prior to the onset of the disease, and positive serum antibodies to avian antigens, as detected by ELISA (see supplementary data from the study by Selman and coworkers, included in the online supplement to this article). The median duration of disease was 18 months, making these patients likely those with active disease and persistent exposure to the antigens.

Strong enrichment of genes from the HP gene set with genes up-regulated in the P-F group suggests greater similarity in gene expression in the lungs of P-F patients to genes over-expressed in HP when compared to IPF. Since bronchoalveolar lavage of patients with HP show marked and primarily T cell response with secondary immune granuloma formation in the lungs (27), it is perhaps not surprising that patients with sarcoidosis would have a similar profile. The interesting observation is that it was only noted in the P-F disease in spite of nodularity on HRCT scans in patients from the N-SL group. One possible explanation is that patients with progressive disease are persistently exposed to an aeroallergen which stimulates a brisk and persistent T cell response. The other is that polymorphisms in immune genes responsible for control of T cell activity to an aeroallergen are responsible for this unbridled activity. The latter is supported by findings of defects in the BTNL-2 and ANXA-11 genes alluded to earlier (4, 5).

These data also suggest that patients in the P-F group are likely to have arrived at this phase of the disease via persistent immune activity rather than the presence of fibrogenic predisposition. In recent years, research in the area of pulmonary fibrogenesis has suggested that although inflammation typically precedes fibrosis, the central mechanisms at the severe end of the spectrum (e.g., in IPF) are likely to be distinct from those that drive persistent inflammation (28). These mechanisms involve those that govern myofibroblast recruitment, proliferation, differentiation, and activity, and may explain why this condition is resistant to immunosuppression (28, 29). The suggestion that P-F sarcoidosis is more similar to HP than to IPF is significant because one issue with identifying patients with P-F disease is whether early recognition will make a difference to management, since many fibrotic pulmonary lung diseases are unresponsive to treatment. In IPF, subpleural reticulation, honeycombing, and fibroblastic foci on histopathology portend a poor prognosis: most patients do not respond to treatment, and the median survival is 3 to 5 years (30, 31). On the other hand, HP (without evidence of usual interstitial pneumonitis), is very sensitive to corticosteroid treatment and the disease can be treated successfully with allergen removal or/and corticosteroids. The inference from our data is that it is possible that, if they can be identified early, patients with progressive pulmonary sarcoidosis are likely to benefit from prompt corticosteroid treatment.

One potential limitation to interpretation of the data is that, although patients from both groups were bronchoscoped at active phases of their disease (the N-SL group had nodularity on HRCT and in five of the eight patients, lymphopenia), the P-F group may have been sampled at a more active point. This is a consequence of our attempt to capture and compare two divergent groups of patients with different disease severity, and patients with more severe disease may also have presented with more active disease. Thus, the comparison may encapsulate both disease severity (intended) and disease activity (unintended). This could explain the up-regulation of immune response genes observed in the P-F group, though not the similarity of gene profile to HP rather than IPF.

In this study, we also do not have the full composition of the cellular infiltrate to the respiratory tract. Routine clinical analyses indicated that bronchoalveolar fluid from all patients showed a lymphocytosis, but comprehensive flow cytometry phenotyping of cells and their proportions were not undertaken. In retrospect, this may have been a useful adjunct to the study, since it would further characterize the patients and their lung disease.

The next challenge is to derive a gene signature profile to identify these patients. This study was small and not designed (nor statistically powered) to test the predictive value of the expression profile. However, we have shown is that it is possible to use standard clinical procedures to obtain high-quality RNA suitable for microarray analysis from lung biopsies, and these results set a precedent and platform for translation via a validation study to a clinical practice–enhancing tool.

In summary, this study provides insight into the etiopathology of P-F sarcoidosis, pointing to a persistent T cell response as the cause of chronic disease rather than genes involved in fibrogenesis in IPF. It also shows that gene array expression profiling of samples obtained using a routine clinical tool can detect a set of genes that differentiate two classes of patients with pulmonary sarcoidosis, which stands up to testing using biologically relevant categories of genes. It offers impetus to refine the treatment of sarcoidosis, particularly because the similarity of the P-F subset of patients to HP suggests that they are likely to respond to early treatment.

The authors thank Dr. Rachel Benamore (Thoracic Imaging, Churchill Hospital) for her contribution to the interpretation of HRCT scans in patients, Dr. Taane Clark and Professor C. Newbold (Wellcome Trust Sanger Institute) for discussions, and Dr. Colin Clelland (Pathology Department, John Radcliffe Hospital) for histopathology support.

1. Baughman RP, Teirstein AS, Judson MA, Rossman MD, Yeager H Jr, Bresnitz EA, DePalo L, Hunninghake G, Iannuzzi MC, Johns CJ, et al. Clinical characteristics of patients in a case control study of sarcoidosis. Am J Respir Crit Care Med 2001;164:1885–1889.
2. Iannuzzi MC, Rybicki BA, Teirstein AS. Sarcoidosis. N Engl J Med 2007;357:2153–2165.
3. Gerke AK, Hunninghake G. The immunology of sarcoidosis. Clin Chest Med 2008;29:379–390.
4. Hofmann S, Franke A, Fischer A, Jacobs G, Nothnagel M, Gaede KI, Schürmann M, Müller-Quernheim J, Krawczak M, Rosenstiel P, et al. Genome-wide association study identifies ANXA11 as a new susceptibility locus for sarcoidosis. Nat Genet 2008;40:1103–1106.
5. Valentonyte R, Hampe J, Huse K, Rosenstiel P, Albrecht M, Stenzel A, Nagy M, Gaede KI, Franke A, Haesler R, Koch A, et al. Sarcoidosis is associated with a truncating splice site mutation in BTNL2. Nat Genet 2005;37:357–364.
6. Nguyen T, Liu XK, Zhang Y, Dong C. BTNL2, a butyrophilin-like molecule that functions to inhibit T cell activation. J Immunol 2006;176:7354–7360.
7. Moss S, Morgan R. The annexins. Genome Biol 2004;5:219–227.
8. Neville E, Walker AN, James DG. Prognostic factors predicting the outcome of sarcoidosis: an analysis of 818 patients. Q J Med 1983;52:525–533.
9. Romer FK. Presentation of sarcoidosis and outcome of pulmonary changes. Dan Med Bull 1982;29:27–32.
10. Hillerdal GE, Nou K, Osterman K, Schmekel B. Sarcoidosis: epidemiology and prognosis. A 15 year European study. Am Rev Respir Dis 1984;130:29–32.
11. Sanderson S, Kulakova N, Lockstone H, Baban D, McMichael AJ, Ho LP. Gene-set analysis of lung samples from progressive, fibrotic pulmonary sarcoidosis shows pattern similar to hypersensitivity pneumonitis. Thorax 2009;64:A5–A74.
12. Morell F, Levy G, Orriols R, Ferrer J, De Gracia J, Sampol G. Delayed cutaneous hypersensitivity tests and lymphopenia as activity markers in sarcoidosis. Chest 2002;121:1239–1244.
13. Selroos O, Nikinmaa B, Koivunen E, Riska H, Tiitinen H, Weber T. Cellular immunity during follow-up of patients with sarcoidosis of varying duration. Clin Exp Immunol 1982;50:25–33.
14. Johns CJ, Scott PP, Schonfeld SA. Sarcoidosis. Annu Rev Med 1989;40:353–371.
15. Selman M, Pardo A, Barrera L, Estrada A, Watson SR, Wilson K, Aziz N, Kaminski N, Zlotnik A. Gene expression profiles distinguish idiopathic pulmonary fibrosis from hypersensitivity pneumonitis. Am J Respir Crit Care Med 2006;173:188–198.
16. Smyth GK. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 2004;3:1–28 (Article 3).
17. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, Gentry J, Hornik K, et al. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 2004;5:R80:1–16.
18. Reiner A, Yekutieli D, Benjamini Y. Identifying differentially expressed genes using false discovery rate controlling procedures. Bioinformatics 2003;19:368–375.
19. Saeed AI, Sharov V, White J, Li J, Liang W, Bhagabati N, Braisted J, Klapa M, Currier T, Thiagarajan M, Sturn A, et al. TM4: a free, open-source system for microarray data management and analysis. Biotechniques 2003;34:374–378.
20. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES, et al. Gene-set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA 2005;102:15545–15550.
21. Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, Lehar J, Puigserver P, Carlsson E, Ridderstråle M, Laurila E. PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat Genet 2003;34:244–245.
22. Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res 2001;29:e45:2002–2008.
23. Oshita F, Ikehara M, Sekiyama A, Hamanaka N, Saito H, Yamada K, Noda K, Kameda Y, Miyagi Y. et al. Genome-wide cDNA microarray screening of genes related to benefits and toxicities of platinum-based chemotherapy in patients with advanced lung cancer. Am J Clin Oncol 2005;28:367–370.
24. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2000;25:25–29.
25. Lovering RC, Camon EB, Blake JA, Diehl AD. Access to immunology through the Gene Ontology. Immunology 2008;125:154–160.
26. Simon R. Challenges of microarray data and the evaluation of gene expression profile signatures. Cancer Invest 2008;26:327–332.
27. Agostini C, Trentin L, Facco M, Semenzato G. New aspects of hypersensitivity pneumonitis. Curr Opin Pulm Med 2004;10:378–382.
28. Wynn TA. Cellular and molecular mechanisms of fibrosis. J Pathol 2008;214:199–221.
29. Wells AU, Hogaboam CM. Update in diffuse parenchymal lung disease. Am J Respir Crit Care Med 2008;177:580–584.
30. Noble PW. Idiopathic pulmonary fibrosis: natural history and prognosis. Clin Chest Med 2006;27:S11–S16.(v.).
31. King TE Jr, Tooze JA, Schwarz MI, Brown KR, Cherniack RM. Predicting survival in idiopathic pulmonary fibrosis: scoring system and survival model. Am J Respir Crit Care Med 2001;164:1171–1181.
Correspondence and requests for reprints should be addressed to Ling-Pei Ho, M.D., Ph.D., MRC Human Immunology Unit, Weatherall Institute of Molecular Medicine, Oxford University, Oxford OX3 9DS, UK. E-mail:


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record