Clinical Parameters
RA patients were predominantly female (71.0%), with a mean age of 50.0 years, and mostly ACPA (92.3%) and RF (95.2%) positive. Disease activity before MTX initiation was moderate to high (mean DAS28-CRP: 4.7 ± 1.4) (Table 1). Of the 31 initial RA patients, 25 were treated with MTX for at least 3 months. After a 3-month MTX treatment period, 12 patients were classified as responders and 13 as non-responders. The two groups before treatment did not significantly differ in baseline DAS28-CRP (4.6 ± 1.1 vs. 3.8 ± 1.5, p = 0.17 for responders and non-responders, respectively) or in tender and swollen joint counts (Supplementary Table S1).
Table 1
Baseline RA patient characteristics (M0)
|
RA (n = 31)
|
Women, n(%)
|
22 (71.0)
|
Age (years)
|
50.0 (18.8)
|
Disease duration (years)
|
4.9 (7.0)
|
ACPA+, n(%)
|
24 (92.3)
|
RF+, n(%)
|
20 (95.2)
|
History or actual MTX or other DMARDs
|
0 (0)
|
History or actual bDMARDs
|
0 (0)
|
Oral steroid*
|
13 (39.4)
|
TJC (/28)
|
8.3 (7.0)
|
SJC (/28)
|
6.3 (4.3)
|
VAS pain (/100)
|
58.4 (29.3)
|
ESR (mm)
|
36.7 (33.8)
|
CRP (mg/L)
|
22.1 (26.6)
|
DAS28-CRP
|
4.7 (1.4)
|
Type II diabetes
|
2 (6.7)
|
Dyslipidaemia
|
5 (16.7)
|
High blood pressure
|
2 (6.7)
|
* ≥ 5mg/d
Abbreviation: ACPA, anti-citrullinated protein antibodies; RF, rheumatoid factor; TJC, tender joint count; SJC, swollen joint count; VAS, visual analogue scale; ESR, erythrocyte sedimentation rate; CRP, C-reactive protein
RBC Metabolites associated with RA (vs. healthy controls)
Under our conditions, 33 metabolite features from the C18(+) and 74 from the HILIC(-) analysis matched the accurate masses and retention times of the metabolites included in our chemical database, yielding 105 unique annotated metabolites (See Appendix C). Univariate analysis of the differences in mean expression between the two groups showed a significant difference (p-value < 0.05) for 14 metabolites (Figure 1A, Table 2) and all of them remained significant after correction for multiple testing (FDR q-value <0.05) (Supplementary Figure S1). Three of them – histidine, ergothioneine (EGT), and citrulline – were found to be less expressed in RBCs from RA patients (RA/HC ratio: 0.86, 0.66 and 0.83, respectively, with all three FDR qvalue<0.02). Meanwhile, eleven – including 4 metabolites in the carbohydrate subclass and 2 amino acids (creatine and valine) – showed accumulation. Although the results obtained under C18(+) conditions were statistically less robust (Figure 1B, Supplementary Table S2), the same trend was observed for EGT (RA/HC ratio: 0.70, p=0.040) and citrulline (RA/HC ratio: 0.83, p=0.027), showing lower concentrations in the RA group compared to HCs. These differences did not reach FDR-adjusted significance. In this dataset, 5 metabolites involved in the arginine and proline pathway showed accumulation in RA RBCs compared to HCs (aminovaleric acid, creatine, spermidine, spermine and stachydrine). Metabolites were also studied in plasma samples from RA patients and HCs (See Appendix C, supplementary tables S3 and S4). We found lower concentrations of several amino acids in the plasma of RA patients compared to HCs (DHEA, alanine, threonine, ornithine, theophylline) and higher hydroxyproline.
Table 2
RBC metabolites associated with RA (HILIC(-) dataset), ranked lowest to highest RA/HC ratio
Metabolite
name
|
KEGG ID¤
|
Class¤
|
Subclass¤
|
Pathway¤
|
Pvalue*
|
FDR-Qvalue*
|
RA/HC ratio
|
Ergothioneine
|
C05570
|
Carboxylic acids and derivatives
|
Amino acids, peptides, and analogues
|
Histidine metabolism
|
0.016
|
0.024
|
0.66
|
Citrulline
|
C00327
|
Carboxylic acids and derivatives
|
Amino acids, peptides, and analogues
|
Arginine and proline metabolism
|
0.018
|
0.025
|
0.83
|
Histidine
|
C00768
|
Carboxylic acids and derivatives
|
Amino acids, peptides, and analogues
|
Histidine metabolism
|
0.005
|
0.014
|
0.86
|
NADP
|
C00006
|
Phenols
|
Benzenediols
|
Histidine metabolism
|
0.046
|
0.051
|
1.14
|
Hexoses-bisphosphate
|
C01231
|
Organooxygen compounds
|
Carbohydrates and carbohydrate conjugates
|
Amino sugar and nucleotide sugar metabolism
|
0.011
|
0.020
|
1.16
|
Creatine
|
C00300
|
Carboxylic acids and derivatives
|
Amino acids, peptides, and analogues
|
Arginine and proline metabolism
|
0.019
|
0.025
|
1.17
|
Uridine 5’-diphosphogalactose
|
C00052
|
Pyrimidine nucleotides
|
Pyrimidine nucleotide sugars
|
Pyrimidine metabolism
|
0.005
|
0.014
|
1.19
|
Gluconic acid
|
C00257
|
Organooxygen compounds
|
Carbohydrates and carbohydrate conjugates
|
Ketogluconate metabolim
|
< 0.001
|
0.002
|
1.22
|
Hydroxyglutaric acid
|
C02630
|
Hydroxy acids and derivatives
|
Short-chain hydroxy acids and derivatives
|
Butanoate metabolism
|
0.011
|
0.020
|
1.22
|
Phosphoenolpyruvic acid
|
C00074
|
Organic phosphoric acids and derivatives
|
Phosphate esters
|
Pyruvate metabolism
|
0.007
|
0.016
|
1.23
|
Diphosphoglyceric acid
|
C01159
|
Organooxygen compounds
|
Carbohydrates and carbohydrate conjugates
|
Gluconeogenesis
|
< 0.001
|
0.002
|
1.29
|
Glyceraldehyde -phosphate
|
C00118
|
Organooxygen compounds
|
Carbohydrates and carbohydrate conjugates
|
Gluconeogenesis
|
< 0.001
|
0.002
|
1.29
|
Valine or betaine
|
C00719
|
Carboxylic acids and derivatives
|
Amino acids, peptides, and analogues
|
Glycolysis
|
0.040
|
0.048
|
1.30
|
Guanosine - diphosphate
|
C00035
|
Purine nucleotides
|
Purine ribonucleotides
|
Purine Metabolism
|
< 0.001
|
< 0.001
|
1.86
|
* Significant changes in metabolite levels were tested using the t-test, and False Discovery Rate (FDR)-q value were determined according to Storey and Tibshirani |
¤ Metabolite ID, class, subclass and pathway were provided by the Human Metabolome Database (HMDB) |
High level of RBC ergothioneine is associated with favorable methotrexate (MTX) response
Since RBC analysis identified EGT as a differentiating metabolite between RA patients and HCs, we investigated whether the concentration of EGT differed between responders and non-responders before (M0) and/or after (M3) MTX initiation. We found significantly lower EGT values in non-responders at M0 (p=0.036) when using the HILIC(-) dataset (Figure 2). A similar trend was observed using C18(+) (p=0.07, Wilcoxon test).