Observed Variability in Gene Expression in Individuals with Sickle Cell Disease during Baseline Health and Acute Pain

Background: Acute pain episodes are the most common complication of sickle cell disease (SCD). There is wide variability in frequency and presentation of pain between patients despite inheritance of the same genetic defect. The underlying biology that accounts for this variability is still unknown. Prior data support cytokine profiles change in individuals with SCD during acute pain as opposed to baseline health and may contribute to SCD pain biology. We therefore hypothesized that there would be differences in gene expression profiles within a patient between baseline health and acute pain (intraindividual variability). Further, due to the variability in pain expression between individuals with SCD, we also hypothesized that gene expression profiles would show interindividual variability between patients with SCD.

Methods: Our study included a convenience sample of individuals with SCD and healthy Black controls recruited from tertiary care facilities. Paired plasma samples were collected from individuals with SCD during baseline health and acute pain. We conducted a cross-sectional analysis in 2 independent datasets (i.e., one single center, one multicenter – “The Magnesium for Children in Crisis trial”) of these paired samples. We used a novel plasma-based transcription bioassay to assess gene transcription in a peripheral blood mononuclear cell (PBMC) that served as a sensor to plasma-borne factors. Patient plasma was co-cultured with cryopreserved PBMCs from a healthy donor to induce transcription. We identified informative transcripts that differentiated individuals with SCD from healthy controls thereby defining the disease-specific plasma-induced signature and retained 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%. For further data reduction, we also retained the top 25% of transcripts with the highest intensity signals and retained the top 5000 genes with the highest mean absolute deviation (MAD) in both analyses. The intersection of genes identified by these two analyses (i.e., top 25% and MAD) was used as an input gene list for Weighted Gene Correlation Network Analysis (WGCNA). WGCNA was used to extract the genes that were most closely associated with pain. To this end, we retained all genes that had a correlation of ≥0.3 and p-value of <0.05 between the gene and the number of acute care visits for pain in the prior 3 years. This gene list was designated as the final SCD pain-specific list. Heatmaps of gene expression of those in the SCD pain-specific list were created using Genesis.

Results: We analyzed paired samples from 149 individuals with SCD (median age 13.6, IQR 9.3-17.1) during baseline health and acute pain (27 from single center cohort, 122 from multicenter cohort). The data reduction filtering analyses identified 1177 genes of interest. Figure 1 shows both mean gene expression in the single center cohort and paired gene expression for each individual. As depicted, overall mean gene expression does not appear to differ between baseline health and acute pain. When evaluating each individual as a paired set we observe that in the majority of individuals gene expression does not change substantially between the two states of health, suggesting minimal intraindividual variability. However, a minority of individuals do exhibit more substantial changes between states of health. When considering the same state of health, wide interindividual variability among patients is observed. These findings were reproducible in a larger, multicenter cohort, as shown in Figure 2.

Conclusion: Our findings suggest low intraindividual variability in most individuals with SCD, signifying fixed gene expression that may not change within an individual between baseline health and acute pain. Wide interindividual variability in gene expression was observed which might be responsible for the differences in phenotypic pain expression despite inheritance of the same hemoglobin S mutation. These data suggest pain endotypes within SCD may exist that need further investigation. If identified, these SCD pain endotypes could be leveraged for tailored analgesia or early intervention with higher risk treatments such as curative therapy. Further investigation of groups of genes that show the most variability could be investigated as potential targets for pain treatment.