Declining Pulmonary Function in Interstitial Lung Disease Linked to Lymphocyte Dysfunction

To the Editor:

Sarcoidosis and idiopathic pulmonary fibrosis (IPF) may differ in etiology but they share a common outcome: end-stage fibrotic lung disease. Patients with sarcoidosis and patients with IPF who exhibit end-stage fibrotic changes requiring lung transplant have similarly high mortality rates (1). Innate and adaptive immunity are known drivers of both diseases, but the mechanisms underpinning the fibrotic changes remain poorly understood. We previously observed that a decrease in peripheral blood microRNAs in patients with sarcoidosis was linked to lymphopenia and declining pulmonary function, suggesting a role for peripheral immune regulation in disease progression (2). Recently, T-helper cell type 17 (Th17) lymphocytes expressing the negative costimulatory molecule PD-1 were identified in the peripheral blood of patients with sarcoidosis and patients with IPF (3). These cells produced TGF-β (transforming growth factor β) and induced collagen deposition from fibroblasts in a coculture system. These findings suggest an integral role for the immune response in the maintenance, resolution, and fibrotic progression of both sarcoidosis and IPF. We hypothesized that common immune-regulatory mechanisms govern systemic responses and contribute to fibrotic progression in sarcoidosis and IPF.

We performed a meta-analysis to identify common lymphocyte expression patterns in patients with IPF or sarcoidosis compared with healthy control subjects. A systematic search of publicly available peripheral blood mononuclear cell expression data from the Gene Expression Omnibus (National Center for Biotechnology Information) identified two case–control studies with severity metrics: GSE37912 (sarcoidosis cohort) and GSE38958 (IPF cohort). Severe disease in GSE37912 was defined as “FVC% predicted < 50% OR cardiac OR neurological involvement” (4). GSE38958 did not define severity, but a cutoff of “FVC% predicted < 55% AND diffusion capacity of the lung for carbon monoxide (Dl_CO)% predicted < 40%” was used (5). To identify modules of gene expression in our datasets, we performed a consensus weighted gene correlation network analysis (6). Among the eight modules constructed, we identified the lymphocyte immunity module (LIM), which had significant associations with immune response and lymphocyte activation genes (Figure 1A). Significantly reduced expression of LIM was identified in patients with IPF and patients with sarcoidosis compared with control subjects (Figures 1B and 1C). Of the 442 member genes that made up this module, 263 had significantly decreased expression in both patient cohorts compared with control subjects. The most significant genes within the LIM included the costimulatory molecules CD40LG (cluster of differentiation 40 ligand) and ICOS (inducible T cell costimulator, and CD28). Other LIM members included lymphocyte surface markers, the chemokine receptors CCR4 (C-C chemokine receptor type 4), CCR6, CCR7, and lymphocyte differentiation factors RORA (RAR-related orphan receptor α), GATA3 (GATA binding protein 3), EOMES (eomesodermin), and IKZF3 (ikaros family zinc finger protein 3). These findings suggest that pan-lymphocyte depression of effector-related factors plays a role in the systemic immune response of patients with sarcoidosis or IPF. Interestingly, the Th1 transcription factor TBX21 (T-box transcription factor 21) was not assigned to the LIM but was significantly decreased in patients with sarcoidosis.

Figure 1.

The lymphocyte immunity module (LIM) is significantly associated with disease. (A) A weighted gene correlation network analysis identified eight modules of gene expression. A module of 442 genes was identified and member genes were submitted to the MSigDB database (Broad Institute) to determine gene ontology (GO) pathways associated with the module. Significantly differentiated genes in both sarcoidosis and idiopathic pulmonary fibrosis (IPF) with z-scores > 2 are plotted along with associated top enriched pathways. These pathways related primarily to lymphocyte activity, and the module was named the LIM. LIM expression data were then measured using module eigengenes, a surrogate value for module expression derived from principal component analysis. (B and C) LIM expression was significantly decreased in patients with sarcoidosis and patients with IPF compared with control subjects. (D and E) We validated LIM expression using ACCESS (A Case Control Etiologic Study of Sarcoidosis) sequencing and GSE28042 microarray data. Expression data were normalized and subset to fit commonly sampled genes across all datasets. The modular framework from the training dataset was then applied to validation cohorts and module eigengenes were obtained. LIM was differentially expressed in both the sarcoidosis and IPF validation cohorts. *P < 0.05 and ***P < 0.001; unpaired t-test.

We validated LIM expression using expression data from additional IPF and sarcoidosis datasets (microarray data from GSE28042 and RNA sequencing data from ACCESS [A Case Control Etiologic Study of Sarcoidosis]; ClinicalTrials.gov identifier: NCT00005276; approved by the University of Illinois at Chicago Institutional Review Board, protocol #2016-0275) (7, 8). ACCESS research materials were obtained from the NHLBI. As expected, LIM had decreased expression in both the sarcoidosis and IPF validation cohorts (Figures 1D and 1E). No significant differences in LIM expression were noted across all four datasets among control subjects. This analysis not only validated decreased LIM expression in both diseases but also showed that the findings were independent of the methodology used. Interestingly, whereas all other cohorts varied in time to diagnosis, all patients from the ACCESS cohort were sampled within 6 months of diagnosis. This suggests that, at least in sarcoidosis, these changes in expression may occur early in the disease course.

To examine how features associated with systemic lymphocyte dysfunction might contribute to disease manifestations, we compared LIM expression in patients with severe or nonsevere sarcoidosis or IPF and found significantly reduced expression in patients with the severe form of both diseases (Figure 2A). Given this relationship, we examined the relationship between LIM expression and pulmonary function in the cohorts for which pulmonary function test data were available (GSE37912 and GSE28042 did not have pulmonary function test data available). In the GSE38958 IPF cohort, LIM expression correlated with Dl_CO (Spearman’s rank correlation coefficient = 0.458; 95% confidence interval, 0.224–0.642; Figure 2B). LIM expression correlated with FEV₁ in subjects with sarcoidosis from the ACCESS cohort (Spearman’s rank correlation coefficient = 0.476; 95% confidence interval, −0.003 to 0.777), but few patients had spirometry values below the lower limit of normal, likely owing to their early inclusion after diagnosis. These data indicate that the systemic immune response processes associated with decreased LIM expression in sarcoidosis and IPF may contribute to a decline in pulmonary function.

Figure 2.

Progressive declines in lymphocyte immunity module (LIM) expression are associated with more severe disease. (A) Decreased expression of LIM was observed in patients with more severe sarcoidosis and idiopathic pulmonary fibrosis (IPF). *P < 0.05; unpaired t-test. (B) Expression of LIM was directly correlated with Dl_CO (% predicted) in patients with IPF from GSE38958. Increased expression was correlated with improved Dl_CO (Spearman’s rank correlation coefficient). (C) Patients with IPF from GSE28042 were split into tertiles based on LIM module eigengenes (MEs), and a survival analysis was performed. Patients with intermediate or low LIM expression had significantly reduced transplant-free survival compared with high-LIM-expressing patients (log-rank test). Tick marks represent censored cases.

An analysis of LIM expression in the GSE28042 IPF cohort revealed that relatively high LIM expression was significantly associated with transplant-free survival in patients with IPF. We performed a Cox regression analysis on the dataset and found that reduced LIM expression and male sex significantly reduced survival in the cohort (P < 0.001), with a stronger association with LIM expression alone than in combination with sex (P = 0.00031 vs. P = 0.0005, likelihood ratio test). To test our postulate that LIM expression portends worse outcomes, we divided the cohort into tertiles based on LIM expression levels (Figure 2C). We identified significant differences in transplant-free survival between high-LIM-expressing subjects and lower-LIM-expressing subjects (P < 0.05, log-rank test). These findings indicate that systemic decreases in lymphocyte effector function may promote progression to end-stage pulmonary disease.

Our analyses were limited by potential confounders across our datasets. Reduced lymphocyte counts, which were observed in both diseases, may explain decreased LIM expression, but LIM expression was independent of the lymphocyte percentage in the ACCESS sarcoidosis cohort (not shown). In addition, we were unable to determine the contribution of immunosuppressive treatment regimens to outcomes or LIM expression owing to a lack of treatment data across all cohorts. Future studies will be necessary to consider these confounders.

In this study, we identified a common pattern of gene expression that is suggestive of deficient lymphocyte function in both sarcoidosis and IPF. Consistent with prior reports, we noted further abrogated expression of LIM-associated genes and processes in more severe manifestations of sarcoidosis and IPF (4, 5, 8). Mounting evidence from multiple studies continues to suggest that a functional systemic lymphocyte response facilitates better outcomes for patients (3, 9–11). Immunosuppressive therapies have been used to treat sarcoidosis (with improvements in FVC but no impact on fibrosis) and IPF (without success) (12–15). By targeting the mechanisms that drive systemic immune deficiencies in sarcoidosis and IPF, it may be possible to limit profibrotic processes and improve patient outcomes.