Gene Transcription Data Are Correlated with Pain-Related Phenotypic Traits in Individuals with Sickle Cell Disease

Background: Sickle cell disease (SCD) pain is acute and chronic and associated with significant morbidity across the lifespan. Interindividual variability in SCD pain exists, despite the same hemoglobin gene defect. The biology that explains this variability is not well understood, which is a barrier to novel analgesics and disease modifying therapy. Our prior work used plasma-based gene transcription via a peripheral blood mononuclear cell (PBMC) reporter assay to identify differentially expressed transcripts between individuals with SCD and healthy Black controls. In this study, we sought to provide clinical meaning to gene transcription data in the context of SCD pain. We hypothesized that we could identify gene modules using Weighted Gene Co-expression Network Analysis (WGCNA) that are significantly associated with pain-related phenotypic traits in individuals with SCD.

Methods: We leveraged our prior data from 149 individuals with SCD (median age 13.6, IQR 9.3-17.1) during baseline health and 45 healthy Black controls to discover differentially expressed transcripts between the two groups. Using our well-developed assay, PBMCs from a healthy blood donor functioned as reporter cells to plasma-born factors present in individuals with SCD during baseline health and controls. Gene expression in PBMCs was measured with Affymetrix array. Using our pre-defined filtering criteria of retaining transcripts differentially expressed between individuals with SCD and controls that exhibit a fold change >1.4, ANOVA p-value of <0.05 and an FDR <10%, we identified 3028 genes differentially expressed between individuals with SCD and controls. We then applied this gene list to an independent cohort of newly recruited individuals with SCD during baseline health. Plasma-based transcription using our PBMC reporter cell assay was conducted in the new independent cohort using identical methods to our derivation cohort. The gene list (n=3028) generated from our derivation cohort was used as the input list for WGCNA to investigate the association of gene modules with pain-related phenotypic traits in our new cohort. Clinical pain-related data included: patient-reported outcomes (PROs) (i.e., PROMIS Pain Interference, Pain Behavior, Pain Quality-Neuropathic, PedsQL Pain Impact), acute care pain visits, chronic pain phenotyping. PROs were scored as per developers’ guidelines; means and final PRO scores were used as WGCNA traits. Descriptive statistics summarized clinical variables. WGCNA was used to assess the correlation of co-expressed gene modules with SCD pain-related phenotypic data. The correlation coefficient generated by WGCNA is between the module eigengene (i.e., first principal component of a principal component analysis which is representative of expression profiles of all genes within that module) and the phenotypic trait. We retained modules with significant correlations of ≥0.3 between these modules and SCD pain-related data. Database for Annotation, Visualization and Integrated Discovery (DAVID) was used to identify biological processes represented by the differentially expressed transcripts within each module.

Results: A total of 87 individuals with SCD at baseline health were included in our current cohort. Mean age was 14.5 (SD 14.2) years, 51.7% were female and 70.1% had HbSS/HbSβ⁰thal. Our WGCNA analyses identified 11 modules (Figure 1). Of these, 10 exhibited significant correlations with pain-related phenotypic traits that met our module inclusion criteria of ≥0.3. Modules with the highest number of significant pain-related trait correlations included: 1) pink, black, blue (4 each) and 2) lime, brown, red, turquoise (3 each). DAVID analyses of the biological processes contained within the modules significantly associated with pain-related traits are displayed in Figure 1. A strong immune-mediated component was found across the modules significantly associated with pain-related traits.

Conclusions: We identified gene modules significantly associated with pain-related phenotypic data. These analyses give clinical meaning to gene expression data which is imperative for translational SCD pain research. A strong immune-mediated component was identified in the modules significantly associated with pain. The transcriptional data and pathways identified can be leveraged to investigate SCD pain biology and discover targets for novel pain therapeutics.