Modular Gene Analysis Reveals Distinct Molecular Signatures for Subsets of Patients With Cutaneous Lupus Erythematosus

J. L. Zhu; L. T. Tran; M. Smith; F. Zheng; L. Cai; J. A. James; J.M. Guthridge; B. F. Chong

Disclosures

The British Journal of Dermatology. 2021;185(3):563-572. 

In This Article

Abstract and Introduction

Abstract

Background: Cutaneous lupus erythematosus (CLE) is a heterogeneous autoimmune disease with clinical sequelae such as itching, dyspigmentation and scarring.

Objectives: We applied a previously described modular analysis approach to assess the molecular heterogeneity of patients with CLE.

Methods: Whole-blood transcriptomes of RNA sequencing data from a racially and ethnically diverse group of patients with CLE (n = 62) were used to calculate gene co-expression module scores. An unsupervised cluster analysis and k-means clustering based on these module scores were then performed. We used Fisher's exact tests and Kruskal–Wallis tests to compare characteristics between patient clusters.

Results:Six unique clusters of patients with CLE were identified from the cluster analysis. We observed that seven inflammation modules were elevated in two clusters of patients with CLE. Additionally, these clusters were characterized by interferon, neutrophil and cell-death signatures, suggesting that interferon-related proteins, neutrophils and cell-death processes could be driving the inflammatory response in these subgroups. Three different clusters had a predominant T-cell signature, which were supported by lymphocyte counts.

Conclusions: Our data support a diverse molecular profile in CLE that further adds to the clinical variations of this skin disease, and may affect disease course and treatment selection. Future studies with a larger and diverse cohort of patients with CLE are warranted to confirm these findings.

Introduction

Lupus erythematosus (LE) is an autoimmune disease that has a wide range of clinical manifestations affecting different organs like the skin. Skin-limited lupus, or cutaneous LE (CLE), has been estimated to occur as frequently as systemic LE (SLE).[1] Population-based studies have found that the incidence rate of CLE is approximately 4·0–4·3 per 100 000[1] and that it disproportionately affects females and African Americans.[2–6] CLE is traditionally divided into acute, subacute and chronic CLE (ACLE, SCLE and CCLE), based on disease duration.[7] CCLE has many subtypes, including discoid LE (DLE), LE tumidus and lupus panniculitis. While the clinical heterogeneity of CLE has been largely established, few studies have examined the molecular heterogeneity of the disease.

Studies have previously investigated blood transcription profiles of patients with CLE on a genome-wide scale. A microarray study investigating blood transcription profiles of patients with CCLE identified genes related to apoptosis, type 1 interferon (IFN), immune response and stress that were differentially expressed in patients with CLE compared with healthy patients.[8] Other studies profiled the transcriptome of CLE lesional skin and found a predominance of IFN-γ-producing T-helper (Th)1 cells and increased levels of M1 macrophage-related proteins.[9,10] In patients with DLE, specifically, the transforming growth factor-β1 signalling pathway was found to be overexpressed, suggesting that it plays a role in the fibrosis observed in this CLE subtype.[11] More recently, molecular profiling of skin lesions from patients with CLE identified two patient subsets that did not differ clinically – one with a lymphocytic signature and one with a monocytic signature.[12]

While studies done on a genome-wide scale have the potential to identify disease-specific molecular signatures, they can generate long lists of differentially expressed genes that make data interpretation and prioritization challenging. As a result, a novel approach in gene expression analyses has been described that groups genes into transcriptional modules to help identify relevant genes in various diseases, including SLE.[13] The modular analysis approach has been used to study SLE pathogenesis and subsequently identified that SLE is not exclusively driven by type I IFN, but that type II IFN may also play a role.[14] Furthermore, modular analysis has been used to study and characterize the heterogeneity of different autoimmune diseases. In a cohort of patients with Sjögren syndrome, transcriptional modules distinguished three clusters of patients with similar clinical features but with varying IFN modular network signatures.[15] A similar study conducted in a cohort of patients with SLE found seven molecularly distinct clusters of patients. Two clusters were distinguished by inflammation and IFN modular signatures, while others were distinguished by monocyte, neutrophil, plasmablast, B-cell and T-cell modules.[16] Characterizing the molecular heterogeneity of CLE can potentially facilitate selection of targeted therapeutics for these patients.

Together, these studies demonstrate the need for high-dimensional data in order to better understand the diverse molecular phenotypes in CLE and their implications for clinical diagnosis and management. Our cross-sectional study utilized the modular analysis technique to (i) characterize molecular phenotypes of patients with CLE from controls and (ii) group patients with CLE based on their molecular phenotypes, as well as to evaluate demographic and clinical features that distinguished each cluster. Serum levels of autoantibodies were also measured and compared between patients with CLE and controls as well as among patients with CLE. We hypothesized that subgroups of patients with CLE will have different molecular signatures emphasizing different cell types and cell processes.

processing....