Genetic Insights into Circulating Complement Proteins in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: A Potential Inflammatory Subgroup

Abstract

Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) is a debilitating multi-system illness with heterogeneity that complicates identifying the pathophysiology, biomarkers, and therapeutic targets. Evidence indicates the importance of immune dysregulation, including the complement system, in ME/CFS. This study investigates the contribution of genetic drivers to potential dysregulation of the complement pathway in ME/CFS. We used protein quantitative trait loci (pQTL) analyses, adjusted for covariates using linear and logistic regression, to identify genetic variants significantly associated with plasma complement protein levels in a study sample identified from the general population (50 ME/CFS and 121 non-fatigued). ME/CFS patients carrying certain pQTLs exhibited dysregulation of the alternative complement pathway, which defined an inflammatory subgroup with a high C3/low Bb profile and established a genetic link to dysregulation of the alternative complement pathway. Six of the significant pQTLs were also associated with fatigue-related phenotypes in the UK Biobank, four of which were complement-associated, providing some validation in an independent population. Our findings highlight a mechanism by which risk alleles contribute to ME/CFS heterogeneity, providing evidence of a genetic basis for complement dysregulation in a subset of patients. This approach could identify pathway-focused subgroups in ME/CFS and related illnesses to inform personalized approaches to diagnosis and treatment.

Keywords: Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS); complement system; genetics (pQTLs); heterogeneity; subgroups (genotype-stratified analysis).

Figure 1

The complement system and associated component levels in ME/CFS compared with non-fatigued (NF) control subjects. (A) Schematic of the complement system and associated components, depicting the three activation pathways (color-coded: yellow—(1) classical; orange—(2) lectin; brown—(3) alternative; and maroon—C3 and subsequent products). The same color coding is used in other figures to facilitate linkage back to the complement pathway. (B) Demographic comparisons of sex, age (p = 0.96), and BMI (p = 0.007) between NF (red) and ME/CFS (blue) subjects (n = 171 total; 50 ME/CFS, 121 NF). Statistical comparisons for age and BMI were performed using a Wilcoxon rank-sum test. (C) Radar plot showing mean scores on Multidimensional Fatigue Inventory (MFI-20) subscales between NF and ME/CFS subjects, where lower MFI-20 scores indicate better health scores. GF = general fatigue; PF-MFI = physical fatigue; RA = reduced activity; RM = reduced motivation; MF = mental fatigue. (D) Radar plot showing mean T-scores on Short Form 36 survey (SF-36) subscales between NF and ME/CFS subjects, where a T-score of 50 refers to the norm of the US general population and lower SF-36 T scores signify a worse functional score: PF-SF36 = physical functioning; RP = role: physical; BP = bodily pain; V = vitality; GH = general health; RE = role: emotional; SF = social functioning; MH = mental health. (E–K) Log2-transformed plasma levels of complement proteins between ME/CFS and NF subjects after adjustment for covariates (Table 1): (E) C-Reactive protein (CRP, mg/L; p_lin = 0.15, p_log = 0.14), (F) C3 (mg/mL; p_lin = 0.002, p_log = 0.002), (G) Bb (mg/mL; p_lin = 0.78, p_log = 0.77), (H) Bb/C3 (p_lin = 0.038, p_log = 0.038), (I) Factor B (µg/mL; p_lin = 0.71, p_log = 0.69), (J) Factor D (µg/mL; p_lin = 0.16, p_log = 0.14), and (K) Factor H (µg/mL; p_lin = 0.07, p_log = 0.07). Boxplots display the five-number summary: minimum (smallest value in the dataset), first quartile (Q1, 25th percentile), median (Q2, second quartile), third quartile (Q3, 75th percentile), and maximum (largest value). The central rectangle spans from the first quartile to the third quartile (the interquartile range (IQR)), a segment inside the rectangle shows the median, the diamond shows the mean, the vertical lines (sometimes referred to as whiskers 1.5xIQR) are extended to the extrema of the distribution in the data set, and the outliers are values outside the whisker range. Statistical comparisons of circulating complement protein levels were performed using a covariate-adjusted (Table 1) linear (p_lin) and logistic (p_log) regression analyses. * p ≤ 0.05, ** p ≤ 0.01. (L) Scatterplots depicting the extent of correlation (Pearson) of log2-transformed Factor H with (left to right): CRP (p = 6.1 × 10⁻¹²), C3 (p = 9.9 × 10⁻²⁸), C3a (p = 4.3 × 10⁻⁹), Factor B (p = 1.4 × 10⁻²⁰), and Factor D (p = 6.9 × 10⁻⁴) in all subjects (NF = red, ME/CFS = blue). Linear regression lines with 95% confidence intervals are overlaid.

Figure 2

Association of circulating complement protein levels with participants’ function and symptom scores. Heatmap depicting covariate-adjusted associations between plasma complement protein levels and scores from three questionnaires: Symptom Inventory (CDC-SI, top), Multidimensional Fatigue Inventory (MFI-20, middle), and Short Form 36 survey (SF-36 T scores, bottom). Lower SF-36 T scores indicate worse functional scores, whereas lower MFI-20 and CDC-SI scores indicate better health scores. Rows represent individual domains or subscales for each instrument, and columns represent complement-related proteins. Color represents the coefficient, or magnitude, of each association, with blue indicating a positive association, while red/maroon indicates a negative association. Statistical comparisons were performed using covariate-adjusted linear regression analysis. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001. (SI score = CDC Symptom Inventory; MFI: MF = mental fatigue, RM = reduced motivation, RA = reduced activity, PF = physical fatigue, GF = general fatigue; SF-36: MH = mental health, SF = social functioning, RE = role: emotional, GH = general health, V = vitality, BP = bodily pain, RP = role: physical, PF = physical functioning). NOTE: Complement protein labels are color-coded to match components of the complement system illustrated in Figure 1A.

Figure 3

Functional group distribution of complement protein-associated SNPs/genes and complement protein levels by genotype and illness status. (A) Donut plot summarizing the distribution across 7 functional groups of single-nucleotide polymorphisms (SNPs—dark gray outer ring) and corresponding genes (light gray inner ring) associated with complement proteins (p ≤ 0.01 after covariate-adjustment). Functional group assignments were based on annotated gene functions for genes and associated SNPs (Tables S2–S12). (B) Lollipop plot showing functional group enrichment among SNPs and genes significantly associated with C3 plasma levels, following the pQTL analysis between complement protein levels and SNPs (776 SNPs (359 genes) were identified as significantly associated with at least 1 complement protein). Dot size reflects the proportion of significant functional group genes (light gray) and SNPs (dark gray). Enrichment score > 1 (dotted red line) indicates overrepresentation. (C–H) Boxplots of plasma protein levels (log2, mg/mL) by genotype. Dot color indicates disease status (NF = red; ME/CFS = blue); x-axis = percentage of ME/CFS subjects within each genotype group; (C) C5a levels for rs17611/C5 genotypes (p = 1.61 × 10⁻²⁷, log2, ng/mL), a pQTL that has been validated in other cohorts, serving as an internal control in our dataset; (D) Bb for rs800292/CFH genotypes (p = 0.003); (E) C3 for rs800292/CFH genotypes (p = 0.61); (F) C3 for rs17759529/DPP4 genotypes (p = 0.0002); (G) Bb for rs1061170/CFH genotypes (p = 0.025); (H) left: Bb for rs4151667/CFB genotypes (p = 0.002), right: Bb for rs9332739/C2 genotypes (p = 0.002). The percentage of ME/CFS subjects within each genotype group is indicated on the x-axis, while the (CFS/Total) row indicates the number of ME/CFS subjects vs. the total number of subjects per genotype. Boxplots represent the median ± 25th and 75th quartiles. Whiskers represent 1.5× the interquartile ranges. Outliers are values outside the whisker range. Asterisks indicate a significant overall genotype effect based on a covariate-adjusted full-versus-reduced (FvR) linear regression model comparing models with and without the genotype term. Significance (p-value) reflects improvement in model fit attributable to genotype and does not represent pairwise comparisons between individual genotype groups. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001.

Figure 4

Directionality of pQTL impact on both disease (odds ratio (OR) for ME/CFS) and molecular trait. (A) Overlay boxplots of Bb (brown, primary y-axis, log2, mg/mL) and C3 (maroon, secondary y-axis, log2, mg/mL) plasma levels by genotype combinations of rs9332739/C2 and rs1061170/CFH (NF = empty dot, ME/CFS = filled dot). The percentage of ME/CFS subjects within each genotype group is indicated on the x-axis, while the (CFS/Total) row indicates the number of ME/CFS subjects vs. the total number of subjects per genotype. Boxplots represent the median ± 25th and 75th quartiles. Whiskers represent 1.5× the interquartile ranges. Outliers are values outside the whisker range. The red or orange line links the mean plasma levels of C3 and Bb, respectively, across the five genotype combinations. (B,C) Linear regression plot showing the relationship between odds ratios (OR) for disease risk and beta coefficients (β) for plasma protein levels, with each point representing one SNP. All associations are done with respecte to the minor allele frequency in cases and controls: (B) C3 (27 SNPs included, X-axis—OR, Y-axis—β); (C) Bb (30 SNPs included, X-axis—OR, Y-axis—β); (D) Bb (6 SNPs associated with compliment pathway included, X-axis—OR, Y-axis—β). NOTE: Complement protein labels and data points/lines are color-coded to match components of the complement system illustrated in Figure 1A.

Figure 5

Stratification of ME/CFS and non-fatigued control subjects using complement protein-associated pQTL genotypes. (A) Subgroup distribution and demographics (age, sex, BMI) based on genotype status of rs9332739 (C2) and rs800292 (CFH). Group A (CFShet, dark blue) = ME/CFS heterozygous at one or both loci; Group B (CFSrem, light blue) = Remainder of ME/CFS subjects; Group C (NFhet, dark red) = NF controls heterozygous at one or both loci; Group D (NFrem, pink) = remainder of NF controls. (B) Complement protein ROC curves for ME/CFS (yellow = CRP, light orange = C3, dark orange = Factor H, brown = C3a), restricted to heterozygous groups (Group A (CFS het) and Group C (NFhet). (C) Heatmap based on the log2-transformed mean range for each complement protein, low (blue) to high (yellow). Mean values for genotype-defined subgroups (CFShet, CFSrem, NFhet, NFrem) are shown. Asterisks indicate statistical significance, adjusted for covariates (gray = likelihood ratio test with p ≤ 0.05, red = subgroup-specific multinomial logistic regression with reference = NFhet and p ≤ 0.05). (D–I) Boxplots showing distributions for plasma levels of complement protein (log2, mg/mL) by genotype group, CFShet = dark blue, CFSrem = light blue, NFhet = dark red, NFrem = pink. (D) Bb (NFhet vs. NFrem p = 0.002), (E) Bb/C3 (CFShet vs. NFrem p = 0.003; NFhet vs. NFrem p = 0.03), (F) C3 (CFShet vs. NFhet p = 0.005, CFShet vs. NFrem p = 0.02), (G) C3a, (H) CRP, and (I) Factor H (CFShet vs. NFhet p = 0.05). Boxplots display the five-number summary: minimum (smallest value in the dataset), first quartile (Q1, 25th percentile), median (Q2, second quartile), third quartile (Q3, 75th percentile), and maximum (largest value). The central rectangle spans from the first quartile to the third quartile (the interquartile range (IQR)), a segment inside the rectangle shows the median, the diamond shows the mean, the vertical lines (sometimes referred to as whiskers 1.5xIQR) are extended to the extrema of the distribution in the data set, and the outliers are values outside the whisker range. Statistical comparisons of circulating complement protein levels (black asterisk) were performed using a covariate-adjusted (Table 1) pairwise linear regression analysis. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001.

Figure 6

Associations between complement protein levels and symptom/function scores for genotype-restricted subgroup (CFShet, NFhet, n = 71). Heatmap of covariate-adjusted associations between complement protein levels and scores: Symptom Inventory (SI, top), Multidimensional Fatigue Inventory (MFI, middle), and Short Form 36 survey (SF-36 T scores, bottom). Rows represent individual domains or subscales for each survey, and columns represent complement-related proteins. Color represents the coefficient, or magnitude, of each association, with blue indicating a positive association and red indicating a negative association. Statistical comparisons were performed using covariate-adjusted linear regression analysis. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001. (SI = CDC Symptom Inventory; MFI: MF = mental fatigue; RM = reduced motivation; RA = reduced activity; PF = physical fatigue; GF = general fatigue; SF-36: MH = mental health; SF = social functioning; RE = role: emotional; GH = general health; V = vitality; BP = bodily pain; RP = role: physical; PF = physical functioning). NOTE: Complement protein labels are color-coded to match components of the complement system illustrated in Figure 1A.