Circulating biomarker analyses in a longitudinal cohort of patients with IPF

Abstract

Idiopathic pulmonary fibrosis (IPF) is an incurable interstitial lung disease characterized by fibrosis. Two FDA-approved drugs, pirfenidone and nintedanib, only modestly prolong survival. In this study, we asked whether levels of select circulating biomarkers in patients with IPF demonstrated changes in response to treatment over time and whether treatment with pirfenidone and nintedanib led to differential biomarker expression. Serial plasma samples from 48 patients with IPF on usual treatment and six healthy volunteers were analyzed to identify differentially expressed blood protein. Hypothesis-driven potential biomarker selection was based on recent literature, internal preclinical data, and the PROLIFIC Consortium (Schafer P. 6th Annual IPF Summit. Boston, MA, 2022) proposed biomarkers of pulmonary fibrosis. We compared our findings to public databases to provide insights into relevant signaling pathways in IPF. Of the 26 proteins measured, we found that 11 (SP-D, TIMP1, MMP7, CYFRA21-1, YKL40, CA125, sICAM, IP-10, MDC, CXCL13) were significantly elevated in patients with IPF compared with healthy volunteers but their levels did not significantly change over time. In the IPF samples, seven proteins were elevated in the treatment group compared with the no-treatment group. However, protein profiles were not distinguishable between patients on pirfenidone versus nintedanib. We demonstrated that most proteins differentially detected in our samples were predicted to be secreted from the lung epithelial or interstitial compartments. However, a significant minority of the proteins are not known to be transcriptionally expressed by lung cells, suggesting an ongoing systemic response. Understanding the contributions of the systemic response in IPF may be important as new therapeutics are developed.NEW & NOTEWORTHY In this study, we confirmed protein expression differences in only a subset of predicted biomarkers from IPF and control subjects. Most differentially expressed proteins were predicted to be secreted from lung cells. However, a significant minority of the proteins are not known to be transcriptionally expressed by lung cells, suggesting an ongoing systemic response. The contributions of the systemic response in IPF may be important as new therapeutics are developed.

Keywords: biomarkers; fibrosis; idiopathic pulmonary fibrosis; prediction.

Graphical abstract

Figure 1.

Plasma protein signature in patients with IPF and controls. A: schematic of the overall study. Twenty-six proteins were measured in plasma samples from patients with IPF or control healthy volunteers both at study entry and longitudinally over multiple visits. B: overview of all proteins measured (all visits). Heatmap values are scaled (centered by the mean and divided by the standard deviation) by row and proteins are labeled by the biological group. Samples are grouped by disease status. Proteins and samples are ordered by hierarchical clustering using Pearson correlation distance and Ward linkage. [Image created with a licensed version of BioRender.com.]

Figure 2.

Proteins with statistically significant differences between control and IPF (A) or by treatment (B). A: plasma proteins with statistically significant differences between control and IPF [false discovery rate (FDR) < 0.05] at study entry (baseline). The log2-transformed values of the concentration of each protein are shown. The boxplots represent the median value and the interquartile range. Only data from the same analysis plate are shown. Adjusted P values (FDR) are shown as calculated by a mixed-effects model with the subject as a random effect and disease or no disease as the fixed effect. B: plasma proteins with statistically significant differences between no treatment and standard of care (FDR < 0.05). The log2-transformed values adjusted for the analysis plate of the concentration of each marker are shown including data from all visits. The colors represent the current treatment when the sample was collected. The proteins are organized by biological group. The boxplots represent the median value and the interquartile range.

Figure 3.

Changes in select protein levels in patients over time. Proteins measured are plotted over time relative to the first visit for each subject. Lines connect the data from the same subject. In three patients who received lung transplants, samples post-lung transplant are indicated in yellow.

Figure 4.

mRNA levels for the measured plasma proteins and their putative receptors are expressed on different cell types in patients with IPF. A: median gene expression values for the 11 elevated proteins mapped to gene symbols in 52 tissues from healthy donors. Each point represents the median log₂(TPM + 1) for all samples in that tissue. Cultured fibroblasts and lymphocytes were excluded. B: differentially expressed genes from the Habermann et al. (24) data set between IPF and controls across cell subsets that were also elevated plasma proteins in our study. The size of the dot indicates the level of statistical significance (P value by hurdle generalized linear model, aka MAST) and the color indicates cell type. C: expression of SFTPD and SFTPA1 across cell types, stratified by control or IPF samples. Cell types in bold font were statistically significant for differences between control and IPF samples (adjusted P value < 0.05) and those in italics were nominally significant (raw P value < 0.05).

Figure 5.

Potential signaling by enriched plasma IPF proteins that could potentiate disease. A: differentially expressed genes that were putative receptors of the 11 elevated proteins in our study. Single-cell data are from the Habermann et al. (24) data set between IPF and controls. The size of the dot indicates the level of statistical significance (P value by hurdle generalized linear model, aka MAST) and the color indicates cell type. B: stylized schematic of major cell types that express putative receptors for enriched plasma protein and may impact disease pathogenesis. Dotted lines indicate relationships based on 1) expression of receptors on the given cell type in the IPF scRNAseq data and 2) receptor-ligand binding relationships between the plasma protein and the receptor. Cell types in Habermann et al. were consolidated manually into higher-level cell types.