Cerebrospinal fluid immune phenotyping reveals distinct immunotypes of myalgic encephalomyelitis/chronic fatigue syndrome

Abstract

Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) is a complex heterogeneous multiorgan disease that can have severe impact on individuals' quality of life. Diagnosis of ME/CFS is based on symptom presentation, and a significant goal for the field is to establish meaningful subtypes. The heterogeneity in the literature suggests that individuals living with ME/CFS may suffer from overlapping but different underlying pathophysiological mechanisms. We enrolled 40 participants with ME/CFS and 41 matched healthy control subjects at the Bragée Clinic in Sweden. We assessed plasma samples from both ME/CFS cases and control groups and cerebrospinal fluid (CSF) samples from individuals with ME/CFS. We investigated dysregulated pathways and disease profiles through clinical questionnaires; multiplex analyses of cytokines, hormones, and matrix metalloproteinases; pathogen seroreactivity through peptide display bacteria libraries; and high-throughput microarray for autoantibodies. All samples used were from humans. We show altered interaction patterns between circulating biological factors in plasma of ME/CFS participants. Our analysis of CSF from individuals with ME/CFS revealed different immunotypes of disease. We found 2 patient clusters based on matrix metalloproteinases profiles. The subgroups had similar clinical presentation but distinct pathogen exposure and CSF inflammatory profiles. Our findings shed light on ME/CFS immune phenotypes and generate hypotheses for future research in disease pathogenesis and treatment development by exploring disease subgroups.

Keywords: ME/CFS; cerebrospinal fluid; immune phenotypes; matrix metalloproteinases; neuroimmunology.

Figure 1.

Demographics and clinical parameters of individuals with ME/CFS. (A) Schematic of study design. Numbers indicate samples collected. Diagram created using BioRender. (B) Distribution of sex assigned at birth for control subjects (left, purple) and participants with ME/CFS (right, blue). n = 41 (control subjects) and n = 39 (ME/CFS participants). (C) Reported triggers of disease onset within the ME/CFS participants. (D–F) (D) Sample collection time in minutes from midnight, (E) age in years, and (F) BMI for control subjects (left, purple) and participants with ME/CFS (right, blue). (G) Assessed Beighton scores for control subjects (left, purple) and participants with ME/CFS (right, blue). (H–M) Reported levels of (H) pain, (I) concentration issues, (J) brain fog, (K) memory issues, (L) fatigue, and (M) sleep disturbances from control subjects (left, purple) and participants with ME/CFS (right, blue). (N) Measured gait speed reported in total seconds taken to walk 10 m for control subjects (left, purple) and participants with ME/CFS (right, blue). (O–Q) Assessed scores of (O) EQ-5D, (P) Quality of Life Score (QoL), and (Q) RAND-36 for control subjects (left, purple) and participants with ME/CFS (right, blue). Each dot represents 1 participant. The central lines indicate the group median and the top and bottom lines indicate the 75th and 25th percentiles, respectively. Significance for differences between groups was assessed using 2-sided Wilcoxon rank sum test.

Figure 2.

Correlation of soluble factors in ME/CFS. (A, B) Correlations between eotaxin (A) and fractalkine (B) and 91 soluble factors including cytokines, hormones, and MMPs from plasma samples for control subjects and participants with ME/CFS (cases). Only significant correlations (P < 0.05) are represented as colored dots. Empty (white) squares represent a lack of statistically significant correlation between 2 markers within the respective group. Pearson’s correlation coefficients from comparisons of soluble factors’ measurements within the same participants are visualized by color intensity.

Figure 3.

CSF MMP signature differentiates 2 subgroups of ME/CFS patients. (A) Unsupervised hierarchical clustering of participants with ME/CFS (n = 31) based on MMP-1, MMP-2, and MMP-10 measurements from CSF samples. Individuals are arranged across rows, with each colored unit indicating normalized CSF MMP quantification. Rows are further annotated by cluster 1 (n = 9, bottom, red) and cluster 2 (n = 22, top, blue). (B–D) Distribution of (B) age in years, (C) reported time from disease onset (years) at time of sample collection, and (D) BMI for cluster 1 (left, red) and cluster 2 (right, blue). (E, F) Reported levels of (E) pain and (F) fatigue from cluster 1 (left, red) and cluster 2 (right, blue). (G) Measured gait speed reported in total seconds taken to walk 10 m for cluster 1 (left, red) and cluster 2 (right, blue). (H–K) Reported levels of (H) brain fog, (I) memory issues, (J) concentration issues, and (K) sleep disturbances for cluster 1 (left, red) and cluster 2 (right, blue). (L–N) Assessed scores of (L) EQ-5D, (M) Quality of Life Scale, and (N) RAND-36 for cluster 1 (left, red) and cluster 2 (right, blue). (O) Distribution of sex assigned at birth for cluster 1 (left, red) and cluster 2 (right, blue, nonsignificant). (P, Q) Prevalence of POTS (P) and general joint hypermobility (Q) for cluster 1 (left panel, red) and cluster 2 (right panel, blue, nonsignificant). (R, S) Assessed Beighton scores (R) and CSF opening pressures (S) for cluster 1 (left, red) and cluster 2 (right, blue). For dot plots, the central lines indicate the group median and the top and bottom lines indicate the 75th and 25th percentiles, respectively. Significance for differences between groups was assessed using 2-sided Wilcoxon rank sum test. Each dot represents 1 participant. CMV, cytomegalovirus.

Figure 4.

Pathogen reactivity and quantification of cytokines between subgroups in ME/CFS patients. (A) Proportion of each ME/CFS cluster and control subjects (ME/CFS cluster A: n = 9; ME/CFS cluster B: n = 22; control group: n = 40) seropositive for each of 12 common pathogen panels as determined by SERA, grouped by pathogen type. Statistical significance was determined by Fisher’s exact test corrected with false discovery rate (Benjamini-Hochberg). *P ≤ 0.05; **P ≤ 0.01; ****P ≤ 0.0001. (B–I) Quantification of cytokines (B) IL-8, (C) IL-15, (D) FLT-3L, (E) MCP-1, (F) IL-10, (G) IL-5, (H) M-CSF, and (I) SCF in CSF samples from cluster 1 (left, red) and cluster 2 (right, blue). Multiple comparison adjustment was performed using the false discovery rate method. Significance was calculated by 2-sample t testing accounting for variations in age, sex assigned at birth, and BMI between clusters through analysis of covariance. (J, K) Spearman’s correlation plots for measurements of fractalkine and MMP-2 in CSF samples from (J) cluster 1 (red) and (I) cluster 2 (Spearman’s ρ = 0.8158, P < 0.0001). (L) Schematic overview of main findings between clusters. Created with BioRender.