ARTHRITIS & RHEUMATISM
Vol. 65, No. 11, November 2013, pp 2803–2813
© 2013, American College of Rheumatology
Association of Low Baseline Levels of
Erythrocyte Folate With Treatment Nonresponse at
Three Months in Rheumatoid Arthritis Patients
M. C. F. J. de Rotte,1P. H. P. de Jong,1S. M. F. Pluijm,1M. Bulatovic ´ C´alasan,2
P. J. Barendregt,3D. van Zeben,4P. A. van der Lubbe,5P. B. de Sonnaville,6
J. Lindemans,1J. M. W. Hazes,1and R. de Jonge1
Objective. To investigate whether baseline con-
centrations of one-carbon metabolism biomarkers are
associated with treatment nonresponse and adverse
events in rheumatoid arthritis (RA) patients receiving
Methods. A prospective derivation cohort (n ?
285) and validation cohort (n ? 102) of RA patients
receiving MTX were studied. Concentrations of plasma
homocysteine, serum vitamin B12, serum folate, eryth-
rocyte vitamin B6, and erythrocyte folate were deter-
mined at baseline and after 3 months of treatment.
Nonresponse after 3 months was assessed using the
Disease Activity Score in 28 joints (DAS28) and the
European League Against Rheumatism (EULAR) re-
sponse criteria. Adverse events at 3 months were as-
sessed using biochemical parameters and health status
questionnaires. Analyses were corrected for baseline
DAS28, age, sex, MTX dose, comedications, and pres-
ence of the methylenetetrahydrofolate reductase 677TT
Results. In the derivation cohort, the mean
DAS28 scores at baseline and 3 months were 4.94 and
3.12, respectively, and 78% of patients experienced
adverse events. This was similar between the 2 cohorts,
despite a lower MTX dose in the validation cohort.
Patients with lower levels of erythrocyte folate at base-
line had a higher DAS28 at 3 months in both the
derivation cohort (? ? ?0.15, P ? 0.037) and the
validation cohort (? ? ?0.20, P ? 0.048). In line with
these results, lower baseline erythrocyte folate levels
were linearly associated with a 3-month DAS28 of >3.2
in both cohorts (derivation cohort, P ? 0.049; validation
cohort, P ? 0.021) and with nonresponse according
to the EULAR criteria (derivation cohort, P ? 0.066;
validation cohort, P ? 0.027). None of the other bio-
markers (levels at baseline or changes over 3 months)
were associated with the DAS28 or treatment nonre-
sponse. Baseline levels of the biomarkers and changes
in levels after 3 months were not associated with inci-
dence of adverse events.
Conclusion. A low baseline concentration of
erythrocyte folate is associated with high disease activ-
ity and nonresponse at 3 months after the start of MTX
treatment and could be used in prediction models for
MTX outcome. None of the investigated one-carbon
metabolism biomarkers were associated with incidence
of adverse events at 3 months.
Dr. de Jonge’s work was supported by the Dutch Arthritis
Association (grants 06-02-402 and 09-1-402). The Treatment in the
Rotterdam Early Arthritis Cohort (tREACH) study is supported by an
unrestricted grant from Pfizer BV (0881-102217).
1M. C. F. J. de Rotte, PharmD, P. H. P. de Jong, MD, S. M. F.
Pluijm, PhD, J. Lindemans, PhD, J. M. W. Hazes, MD, PhD, R. de
Jonge, PhD: Erasmus University Medical Center, Rotterdam, The
Netherlands;2M. Bulatovic ´ C´alasan, MD: University Medical Center
Utrecht and Wilhelmina Children’s Hospital, Utrecht, The Nether-
lands;3P. J. Barendregt, MD, PhD: Maasstad Hospital, Rotterdam,
The Netherlands;4D. van Zeben, MD, PhD: Sint Franciscus Gasthuis
Hospital, Rotterdam, The Netherlands;5P. A. van der Lubbe, MD,
PhD: Vlietland Hospital, Schiedam, The Netherlands;
Sonnaville, MD: Admiraal de Ruyter Hospital, Goes, The Nether-
Dr. van der Lubbe has received honoraria for his service on an
expert panel from Pfizer (less than $10,000). Dr. Hazes has received
education grants from Pfizer and departmental research grants from
Bristol-Myers Squibb, Pfizer, Roche, and Merck/Schering-Plough.
Address correspondence to M. C. F. J. de Rotte, PharmD,
Department of Clinical Chemistry, Erasmus University Medical Cen-
ter, ’S-Gravendijkwal 230, 3015 CE Rotterdam, The Netherlands.
Submitted for publication April 7, 2013; accepted in revised
form July 25, 2013.
6P. B. de
Methotrexate (MTX) is the cornerstone disease-
modifying antirheumatic drug (DMARD) in the treat-
ment of rheumatoid arthritis (RA). In significant num-
bers of patients, MTX fails to achieve adequate
suppression of disease activity and induces adverse
events, which impacts the ability to increase or even
continue the therapeutic dose (1). Patients who do not
respond to MTX or develop severe adverse events
within 3 months after the start of MTX treatment are
frequently treated with biologic agents, alone or in
combination with MTX (2). The ability to predict MTX
nonresponse and MTX-induced adverse events before
the initiation of this DMARD treatment is paramount,
since the first months following diagnosis represent a
window of opportunity during which outcomes can be
more effectively modulated by therapy (3). To ensure
that only those patients who are nonresponsive to MTX
receive early additional treatment with biologic agents,
and those who are responsive to MTX are spared from
treatment with costly biologic agents, it is necessary to
identify nonresponders and patients prone to experience
adverse events at baseline. In order to predict MTX
nonresponse and occurrence of adverse events, risk
factors for these outcomes should be identified (4,5).
Earlier studies have examined clinical and ge-
netic risk factors for MTX nonresponse in RA patients
(1,6). Besides clinical and genetic determinants, pheno-
typic markers (metabolites/proteins) could also be po-
tential predictors of MTX nonresponse. MTX is a folate
antagonist that uses the same transport mechanisms
as folate (7). MTX, as well as food-derived folate or
supplemented folates, are taken up intracellularly via
solute carrier 19A1 and incorporated into the folate
pathway (one-carbon metabolism). Inside cells, MTX
inhibits key enzymes involved in one-carbon metabo-
lism, and this mechanism is responsible for the thera-
peutic effects of MTX, as well as its adverse event
profile. Important phenotypic markers of one-carbon
metabolism, such as concentrations of plasma homo-
cysteine, serum vitamin B12, serum folate, erythrocyte
vitamin B6, and erythrocyte folate, may determine the
extent of MTX nonresponse and MTX-related adverse
events. Nevertheless, these biomarkers have rarely been
studied as risk factors of MTX outcome (8,9).
We therefore investigated whether these one-
carbon metabolism biomarkers, measured at baseline,
could be associated with MTX nonresponse and inci-
dence of adverse events over 3 months of followup in a
prospective cohort study of RA patients receiving MTX.
We also validated our findings in an independent vali-
dation cohort of RA patients.
PATIENTS AND METHODS
Patients. Data from 2 prospective cohorts of RA
patients, all of whom were white, were collected. The deriva-
tion cohort consisted of patients who were enrolled in the
Treatment in the Rotterdam Early Arthritis Cohort
(tREACH) study, which is a multicenter, stratified, single-
blind clinical trial of patients with early RA, as previously
described (10). The validation cohort consisted of patients
with RA from the Methotrexate in Rotterdam, Netherlands
(MTX-R) cohort. These latter patients were started on MTX
treatment between January 2006 and March 2011 in the
Department of Rheumatology of Erasmus University Medical
Center (UMC) (Rotterdam, The Netherlands). The medical
ethics committee of Erasmus UMC approved both studies, and
patients gave their written informed consent before inclusion.
The derivation cohort included patients receiving
MTX who fulfilled the American College of Rheumatology/
European League Against Rheumatism (EULAR) 2010 crite-
ria for RA (11). Patients in the validation cohort were included
when diagnosed as having RA by the physician. Patients from
the derivation cohort were started on an MTX dosage of 25
mg/week. The patients in this cohort were randomized to
receive either MTX alone or cotreatment with sulfasalazine,
hydroxychloroquine, and corticosteroids (10), whereas in the
validation cohort, the dosage of MTX and comedications were
chosen by the physician. In both cohorts, patients received folic
acid (10 mg/week) during MTX treatment. All patients were
assessed at baseline and after 3 months of treatment.
Biomarkers. Three research tubes of blood samples
were obtained during every study visit, in addition to routine
blood samples for determination of the erythrocyte sedi-
mentation rate (ESR) and levels of C-reactive protein (CRP),
alanine aminotransferase (ALT), leukocytes, and thrombo-
cytes. One serum tube was centrifuged for 10 minutes at 1,700g
at a temperature of 4°C, and the serum was divided into
aliquots and stored at ?80°C. One EDTA tube was immedi-
ately put on ice after collection and centrifuged for 10 minutes
at 1,700g at a temperature of 4°C, and plasma and aliquots of
cell pellets were stored at ?80°C. One EDTA tube was kept at
room temperature, and the whole blood was divided into
aliquots and stored at ?80°C.
The concentration of homocysteine was determined in
EDTA–plasma using isotope-dilution liquid chromatography
tandem mass spectrometry (LC-MS/MS; Waters Acquity
UPLC Quattro Premier XE), by an adapted method (12). For
chromatographic separation, a Waters Symmetry C8column
(2.1 ? 100 mm) with a precolumn (Waters) was used. Con-
centrations of vitamin B12and folate in the serum were
measured using an electrochemiluminescence immunoassay
(Modular E170; Roche). The concentration of vitamin B6was
measured in whole blood with an isotope-dilution LC-MS/MS
assay, as described previously (13). For the erythrocyte folate
assay, 100 ?l whole blood was diluted with 1,600 ?l of a 10
gm/liter ascorbic acid solution (pH 4) and incubated for 3
hours at room temperature. Tubes were centrifuged at 2,000g
and analyzed with an electrochemiluminescence immunoassay
for folate (Modular E170; Roche). The concentration of
erythrocyte folate was measured in whole blood from the
EDTA tube at room temperature within 24 hours after sample
collection. The sustained stability of erythrocyte folate at room
2804DE ROTTE ET AL
temperature for up to 24 hours has been proven in a previous
study (14). The erythrocyte folate levels were corrected for
those of serum folate and hematocrit.
Routine hematology parameters were measured using
a Sysmex XE-2100 instrument, and the ESR was measured
using a Sysmex InteRRliner. Routine chemistry parameters
were measured on a Roche Modular P analyzer. Isolation of
DNA and genotyping for the methylenetetrahydrofolate re-
ductase (MTHFR) 677T allele was done using a previously
described method (5).
Assessment of treatment nonresponse and adverse
events. The primary outcome assessed was the Disease Activity
Score in 28 joints (DAS28) with ESR (DAS28-ESR) (15),
which was assessed at baseline and after 3 months of followup.
In rheumatology practice, physicians assess disease activity
levels using a DAS28 cutoff value of ?3.2 to define active
disease, and assess treatment response according to whether
changes in disease activity meet the EULAR response criteria
for RA (16); both of these are used as the decision points for
continuing or stopping medication. Therefore, MTX nonre-
sponse was defined as a DAS28 score of ?3.2 and failure to
meet the EULAR response criteria.
Specifically, the EULAR response criteria are based
on an attained level of improvement in the DAS28 as well as
extent of change in the DAS28 over a defined followup period.
Patients are classified as either nonresponders, moderate
responders, or good responders. In this study, we dichotomized
the EULAR criteria into nonresponse versus moderate/good
response. Of note, the EULAR response criteria allow assess-
ment of only those patients whose DAS28 at baseline is ?3.3.
Adverse events were assessed with biochemical and
self-reported measures. Gastrointestinal symptoms, malaise,
psychological disorders, MTX-related hepatotoxicity, MTX-
related depression of bone marrow, and “other” adverse events
were counted as an adverse event. Dichotomized categories of
?1 adverse event (versus none) and ?3 adverse events (versus
?2) were also analyzed as outcome variables. The adverse
event categories of gastrointestinal symptoms, malaise, psycho-
logical disorders, and “other” adverse events were all accumu-
lations of different symptoms that were reported on the patient
health status questionnaires. Gastrointestinal symptoms com-
prised diarrhea, vomiting, sickness, and abdominal pain. Mal-
aise comprised fatigue, dizziness, headache, sleeplessness, and
not feeling well. Psychological disorders involved depression
and personality changes. “Other” adverse events comprised
dyspnea, alopecia, infection, mucositis, epistaxis, and skin-
related disorders. If a patient reported experiencing none
of these symptoms over the followup and also did not meet
the criteria for hepatotoxicity or bone marrow depression, the
patient was scored as having no adverse event. Hepatotoxicity
was defined as an ALT level ?3 times the upper limit of
normal. Bone marrow depression was defined as a leukocyte
count of ?3.0 ? 109/liter or thrombocyte count of ?100 ?
We did not determine the percentage of patients with
adverse events at baseline. The category of ?3 adverse events
was assigned to patients who experienced more than one
symptom over the followup. For example, if a patient reported
having the symptoms of diarrhea, vomiting, and sickness, he or
Figure 1. Flow chart of the distribution of rheumatoid arthritis (RA) patients in the derivation and validation cohorts at baseline and followup.
MTX ? methotrexate; ACR ? American College of Rheumatology; tREACH ? Treatment in the Rotterdam Early Arthritis Cohort.
LOW ERYTHROCYTE FOLATE LEVELS AND MTX NONRESPONSE IN RA 2805
she was scored as having gastrointestinal symptoms and as
having ?3 adverse events.
Statistical analysis. Statistical comparisons were made
using Student’s t-test, chi-square test, or Mann-Whitney U test,
as appropriate. Multivariate regression analysis was used to
examine the associations between biomarker levels at baseline
and the change in biomarker levels and the DAS28 at 3
months. Results are expressed as the standardized beta coef-
ficient. Logistic regression analysis was used for the dichoto-
mous outcome measures of DAS28 ?3.2 (versus DAS28
?3.2), nonresponse (versus moderate/good response) accord-
ing to the EULAR criteria, and presence (versus absence) of
adverse events. Results are expressed as the odds ratio (OR)
with 95% confidence interval (95% CI).
If necessary, biomarker levels were normalized by
transformation to their natural logarithm, to improve plots of
the residual analyses. To examine whether there was a linear or
nonlinear association, biomarkers were analyzed continuously,
and concentrations were analyzed in quintiles according to
ranges of values (from lowest levels in the first quintile to
highest levels in the fifth quintile). To test whether there was
a significant (P ? 0.05) effect modification, interaction terms,
defined as, for example, biomarker ? comedication, were
included in all multivariate models. If an interaction term was
significant, analyses were stratified. Analyses were corrected
for confounders, including age, sex, baseline DAS28, MTX
dose, presence of the MTHFR 677TT genotype, use of other
DMARDs, and use of corticosteroids (6,17).
In addition, we investigated whether the results based
on the DAS28-ESR, the routine outcome measure used in
our studies, would be comparable to results assessed with
the DAS28-CRP in the pooled cohort. We also examined
the potential effect modification of the MTHFR 677 T allele
genotype, by stratifying the significant biomarker–outcome
associations by genotype in the pooled cohort. The potential
effect modification was explored by defining the interaction
term as significant biomarker ? presence of MTHFR 677 T
All statistical analyses were performed using the SPSS
statistical package, version 220.127.116.11.
Characteristics of the patients. Flow charts for
the derivation cohort and validation cohort, with num-
bers of patients assessed for eligibility and reasons for
dropping out, are given in Figure 1. The 3-month
followup data from the derivation cohort were reported
in an earlier study (18). For the present study, 285
patients from the derivation cohort were included at
baseline, and 270 were still participating after 3 months.
For the validation cohort, 102 patients were included at
baseline, and 84 were still participating after 3 months.
Table 1 shows the baseline characteristics of both co-
horts. The mean MTX dosage was higher in the deriva-
Baseline characteristics of the cohorts*
(n ? 285)
(n ? 102)P
Plasma homocysteine, median (IQR) ?moles/liter
Serum vitamin B12, median (IQR) pmoles/liter
Serum folate, median (IQR) nmoles/liter
Erythrocyte vitamin B6, median (IQR) nmoles/liter
Erythrocyte folate, median (IQR) nmoles/liter
Rheumatoid factor positive, %
Anti–cyclic citrullinated peptide antibody positive, %
ESR, median (IQR) mm/hour
CRP, median (IQR) mg/liter
MTHFR 677T allele, %
Sex, % male
Age, mean ? SD years
Patient global assessment of general health on
100-mm VAS, mean ? SD
28–tender joint count, median (IQR)
28–swollen joint count, median (IQR)
DAS28, mean ? SD
MTX dosage, mean ? SD mg/week
Other DMARDs, %
Oral corticosteroids, %
Parenteral corticosteroids, %
Subcutaneous MTX injections, %
30 29 0.991
54 ? 14
53 ? 22
52 ? 16
54 ? 26
4.94 ? 1.15
4.26 ? 1.43
25 ? 1
15 ? 2
* IQR ? interquartile range; ESR ? erythrocyte sedimentation rate; CRP ? C-reactive protein; MTHFR ? methylenetetra-
hydrofolate reductase; VAS ? visual analog scale; DAS28 ? Disease Activity Score in 28 joints; MTX ? methotrexate;
NSAIDs ? nonsteroidal antiinflammatory drugs; DMARDs ? disease-modifying antirheumatic drugs.
2806DE ROTTE ET AL
tion cohort compared to the validation cohort. In the
validation cohort, the DAS28 was lower, more patients
were taking nonsteroidal antiinflammatory drugs, fewer
patients were taking corticosteroids, and more patients
had received MTX via subcutaneous injection when
compared to patients in the derivation cohort.
Biomarker concentrations. Baseline levels of the
one-carbon metabolism biomarkers were comparable in
both cohorts, with the exception of the baseline eryth-
rocyte folate level, which was lower in the derivation
cohort (median 844 nmoles/liter versus 1,079 nmoles/
liter in the validation cohort; P ? 0.001) (Table 1).
Figure 2 shows all biomarker concentrations at baseline
and at 3 months in both cohorts. The majority of
one-carbon metabolism biomarker concentrations re-
mained stable over time, apart from serum folate, whose
concentration increased in both cohorts (derivation
cohort, median 17 nmoles/liter at baseline versus 31
nmoles/liter at 3 months; validation cohort, median 17
nmoles/liter at baseline versus 26 nmoles/liter at 3
months) (each P ? 0.001). The erythrocyte folate level
increased only in the derivation cohort (median 844
nmoles/liter at baseline versus 940 nmoles/liter at 3
months; P ? 0.023) (Figure 2).
Treatment response. In both cohorts, disease
activity decreased over time. In the derivation cohort,
the mean ? SD DAS28 was 4.94 ? 1.15 at baseline and
decreased to 3.12 ? 1.19 after 3 months. In the valida-
tion cohort, the DAS28 decreased from 4.26 ? 1.43 at
baseline to 2.92 ? 1.23 at 3 months. When treatment
response was determined according to the EULAR
response criteria, 46% of patients were good responders,
38% were moderate responders, and 16% were nonre-
sponders after 3 months of MTX treatment in the
derivation cohort. In the validation cohort, 43% were
good responders, 39% were moderate responders, and
18% were nonresponders. In comparing the derivation
cohort with the validation cohort, the DAS28 (mean 3.12
versus 2.92; P ? 0.174) and EULAR nonresponse rate
(16% versus 18%; P ? 0.879) were comparable after 3
A lower baseline level of erythrocyte folate was
associated with a higher DAS28 after 3 months (deriva-
tion cohort, ? ? ?0.15, P ? 0.037; validation cohort,
? ? ?0.20, P ? 0.048) (Table 2). In line with these
results, the results of logistic regression analyses, in
which the erythrocyte folate level (stratified into quin-
tiles) was the independent variable and a DAS28 ?3.2
was the dependent variable, showed a linear trend
toward association in both the derivation cohort (P ?
0.049) and the validation cohort (P ? 0.021) (Figure 3).
In summary, a low concentration of erythrocyte folate
Figure 2. Concentrations of one-carbon metabolism biomarkers at baseline (circles) and after 3 months of methotrexate treatment (triangles) in
the 2 cohorts. Results are shown as the median with interquartile range.
LOW ERYTHROCYTE FOLATE LEVELS AND MTX NONRESPONSE IN RA 2807
at baseline was associated with MTX nonresponse (a
higher DAS28) at 3 months of treatment. There were no
significant interaction terms.
In the derivation cohort, there was a trend toward
a linear association of the baseline erythrocyte folate
level with nonresponse to treatment according to the
EULAR criteria (DAS28 ?3.2) at 3 months (P ? 0.066),
and in the validation cohort, this association was signif-
icant (P ? 0.027). There were no associations between
the baseline levels of homocysteine, vitamin B12, serum
folate, or vitamin B6and a EULAR-defined nonre-
sponse to MTX at 3 months.
Except for erythrocyte folate, none of the other
one-carbon metabolism biomarkers, measured at base-
line, were associated with the DAS28 at 3 months (Table
2). Moreover, changes in the levels of any of the
one-carbon metabolism biomarkers over time (from
baseline to 3 months) showed no association with either
the DAS28 or the treatment nonresponse at 3 months.
We also performed the analyses of the association
biomarker levels and the Disease Activity Score in 28 joints after 3
months of methotrexate treatment in both cohorts*
Linear regression analysis of associations between folate
Derivation cohort Validation cohort
Serum vitamin B12
Erythrocyte vitamin B6
Change from baseline to
Serum vitamin B12
Erythrocyte vitamin B6
* Analyses were corrected for age, sex, baseline Disease Activity Score
in 28 joints, methotrexate dose, use of other disease-modifying anti-
rheumatic drugs, use of corticosteroids, and presence of the methyl-
enetetrahydrofolate reductase 677TT genotype.
Figure 3. Logistic regression analyses of associations between biomarker concentrations (stratified as second to fifth quintiles, from lowest to
highest levels) and a Disease Activity Score in 28 joints (DAS28) of ?3.2 after 3 months of methotrexate (MTX) treatment in the 2 cohorts. Results
are shown as the odds ratio (OR) with 95% confidence interval, relative to the first quintile (set as an OR of 1), corrected for age, sex, baseline
DAS28, MTX dose, use of other disease-modifying antirheumatic drugs, use of corticosteroids, and presence of the methylenetetrahydrofolate
reductase 677TT genotype.
2808DE ROTTE ET AL
between erythrocyte folate levels and MTX response
using 2 different versions of the DAS28, the DAS28-
ESR and the DAS28-CRP, in the pooled cohort. Both
results were comparable (DAS28-ESR, ? ? ?0.15, P ?
0.009; DAS28-CRP, ? ? ?0.13, P ? 0.038).
In the pooled cohort, the concentration of eryth-
rocyte folate was associated with the DAS28 (? ?
?0.15). When the pooled cohort was stratified according
to genotype, the association of erythrocyte folate levels
with the DAS28 was similar between patients with the
MTHFR 677CC genotype and those with the MTHFR
CT genotype (? ? ?0.16 and ? ? ?0.15, respectively).
In the MTHFR TT stratum, the effect was smaller (? ?
?0.11). However, the interaction term (MTHFR 677CC
versus MTHFR CT versus MTHFR TT ? baseline
erythrocyte folate) was not significant (P ? 0.45). The
interaction terms with the MTHFR 677 genotype, when
divided into 2 categories (CC/CT versus TT and CC
versus CT/TT ? baseline erythrocyte folate), were also
not significant (P ? 0.31 and P ? 0.26, respectively).
Thus, there was no evidence of an effect modification of
the MTHFR 677 genotype on baseline erythrocyte folate
levels in this study.
Adverse events. Figure 4 shows the percentage of
patients with adverse events in both cohorts. The per-
centage of patients with any adverse event over 3 months
was comparable in both cohorts (78% in the derivation
cohort versus 80% in the validation cohort). Only the
percentage of patients with malaise after 3 months of
MTX treatment was significantly higher in the validation
cohort compared to the derivation cohort (49% versus
32%; P ? 0.004). No significant associations between
baseline levels of the one-carbon metabolism biomark-
ers and incidence of adverse events were found after 3
months. Furthermore, changes in the biomarkers be-
tween baseline and 3 months were also not associated
with the occurrence of any adverse events.
In RA patients, low baseline levels of erythrocyte
folate were linearly associated with nonresponse to
short-term MTX treatment in 2 independent cohorts.
None of the one-carbon metabolism biomarkers were
associated with incidence of adverse events over 3 months.
Our study is the first to demonstrate that a low
level of erythrocyte folate at baseline is associated with
nonresponse in 2 independent prospective cohorts of
RA patients receiving treatment with MTX. We showed
that the effect was similar between the derivation cohort
and the validation cohort and was independent of the
response criteria used. Erythrocyte folate levels have
been associated before with MTX outcome in 2 cross-
sectional studies in RA patients (8,9). In these studies,
a higher erythrocyte folate level was associated with
higher disease activity. However, those patients were
being treated with MTX and folic acid at the time of
blood collection. Therefore, these associations could not
be used for prediction of MTX outcome, since erythro-
cyte folate concentrations may be influenced by MTX
competition and folate supplementation. In our cohorts,
patients were not receiving folic acid or MTX at base-
line. Studies on the effects of folic acid supplementa-
tion on the MTX response have reported either no
effects (19) or a negative association (20). Taken to-
gether, these results suggest that lower concentrations of
folate during MTX treatment facilitate higher effective-
ness of MTX in the competition with folate for trans-
porter proteins, polyglutamylation proteins, and target
enzymes for MTX.
In contrast, we found that a lower baseline eryth-
rocyte folate concentration was associated with MTX
nonresponse in the 2 independent cohorts. A possible
explanation for this finding may be that in individuals
with lower concentrations of folate, the absorption,
transportation, cellular uptake, and retention of folates
may be less effective. Since MTX is structurally similar
to folate and uses the same means of transportation and
metabolism, patients with low baseline levels of intra-
Figure 4. Percentage of patients who reported experiencing no ad-
verse events, ?3 adverse events, or specific categories of adverse
events over 3 months of followup in the 2 cohorts. Percentage values
are shown over the bars.
LOW ERYTHROCYTE FOLATE LEVELS AND MTX NONRESPONSE IN RA2809
cellular folate may less easily accumulate MTX intracell-
ularly during therapy. In this sense, measurement of the
baseline erythrocyte folate level is a sort of functional
assay for the body’s capacity to accumulate and retain
cellular folate, and thereby predicts how much MTX will
be taken up and accumulated during therapy.
This hypothesis is supported in different ways.
First, to test this hypothesis, we measured the total
concentration of MTX in the erythrocytes of all patients
in our 2 cohorts, using a recently described isotope-
dilution LC-MS/MS assay (21). The median total MTX
concentration after 3 months of MTX treatment was
130 nmoles/liter packed erythrocytes (interquartile
range [IQR] 92–167) in the derivation cohort and 117
nmoles/liter packed erythrocytes (IQR 78–157) in the
validation cohort. Patients with lower baseline erythro-
cyte folate concentrations achieved lower total erythro-
cyte MTX concentrations after 3 months of MTX ther-
apy, and vice versa (? ? 0.17, P ? 0.011), when analyses
were corrected for sex, age, MTX dose, MTX adminis-
tration route, and cohort. In line with this result, others
have also shown that erythrocyte folate levels were
positively associated with erythrocyte MTX levels (22).
Second, in a recent study of patients with juvenile
arthritis, we observed that a genetic polymorphism in the
influx transporter solute carrier 19A1 was associated
with a diminished response to MTX treatment, and
efflux transporter polymorphisms were associated with
an improved response (23). This finding underscores the
need for effective uptake and cellular retention of MTX.
Levels of homocysteine and vitamin B12were not
associated with MTX outcome in this study. An earlier
study also reported no association of the baseline homo-
cysteine concentration or the 3-month change in homo-
cysteine concentration with MTX response or toxicity
in RA (24). An increased homocysteine concentration
after MTX initiation was observed in earlier studies
(24–27). A decrease in homocysteine levels was ob-
served in groups of patients who received supplementa-
tion with folic or folinic acid (24,26). In the present
study, we observed no significant change in the homo-
cysteine concentration between baseline and 3 months
of treatment. This may be explained by the fact that all
patients in our study received folic acid and MTX. The
baseline levels of folate in the serum were not associated
with MTX nonresponse in our study. We did find a
significant increase in serum folate levels after 3 months.
This can be explained by folic acid supplementation. The
skewed distribution is probably a result of the combina-
tion of the short-elimination half-life of folic acid and
the variation in time span between folic acid intake and
sample collection. The increase in the erythrocyte folate
levels in the derivation cohort could not be replicated in
the smaller validation cohort. Erythrocytes live ?3 months,
and therefore the 3-month data could be diluted.
We observed no association between a lower
erythrocyte vitamin B6concentration and MTX nonre-
sponse. However, earlier research showed that vitamin
B6levels are inversely associated with systemic markers
of inflammation (28). In addition, mild vitamin B6
deficiency characterizes a subclinical at-risk condition in
inflammation-related diseases (29). Patients with lower
vitamin B6concentrations could thus have higher levels
of inflammatory disease and, accordingly, a higher
chance of being a nonresponder. However, baseline
vitamin B6levels were not related to the baseline CRP
level (P ? 0.319) or baseline DAS28 (P ? 0.755) in our
study, and there was no significant association between
the baseline vitamin B6level and the DAS28 after 3
The derivation cohort had lower baseline eryth-
rocyte folate concentrations compared to the validation
cohort. Higher disease activity will cause higher activity
of the immune system, and this could have caused higher
usage of folate and may also have influenced the body’s
capacity to accumulate cellular folate. This might ex-
plain the lower baseline erythrocyte folate level in the
cohort with higher baseline disease activity. Erythrocyte
folate levels increased after 3 months of therapy only in
the derivation cohort. Patients in both cohorts received
10 mg/week folic acid. The patients in the derivation
cohort had a lower erythrocyte folate concentration
compared to the patients in the validation cohort. It is
plausible that the erythrocyte folate levels in the patients
in the derivation cohort increased to within the same
range as those in the validation cohort after 3 months of
treatment with 10 mg/week folic acid.
In a systematic review that included 3,463 RA
patients who had received long-term treatment with
MTX (30), the authors reported that 72.9% of patients
experienced any adverse event, 30.8% had gastrointes-
tinal adverse events, 18.5% developed liver toxicity,
5.5% developed central nervous system toxicity, and
5.2% had cytopenia. In our derivation and validation
cohorts after 3 months of MTX treatment, the percent-
ages of patients with any adverse event (78% and 80%,
respectively), gastrointestinal symptoms (43% and 42%,
respectively), psychological disorders (9% and 13%, re-
spectively), malaise (32% and 49%, respectively), and
other adverse events (35% and 24%, respectively) were
slightly higher than the values reported in the systematic
review. In contrast, the percentage of patients who
2810DE ROTTE ET AL
developed hepatotoxicity (1% and 2%, respectively) and
bone marrow depression (0% and 1%, respectively)
after MTX treatment appeared to be lower in the
There are some differences between the system-
atic review and our study. First, the adverse events
reported herein were measured after 3 months of treat-
ment, whereas in the systematic review, the adverse
events were measured after a mean of 36.5 months
(range 27–132 months). When we determined the inci-
dence of these events after 9 months of treatment in our
derivation and validation cohorts, we found that 14%
and 31% of the patients, respectively, had experienced
gastrointestinal symptoms and 4% in each cohort had
experienced psychological disorders (results not shown),
indicating that fewer patients had developed adverse
events after long-term treatment.
Second, there were differences in dosages used.
The mean MTX dosage in the systematic review was
8.8 mg/week, whereas in our cohorts, the mean dosages
were 15 mg/week and 25 mg/week.
Third, the conditions reported as gastrointestinal
symptoms in our cohorts (diarrhea, vomiting, sickness,
and abdominal pain) differed from those in the system-
atic review (stomatitis, ulcer, abdominal pain, gastro-
intestinal bleed, dyspepsia, nausea, vomiting, diarrhea,
weight loss, and appetite loss).
Fourth, the low incidence of hepatotoxicity and
bone marrow depression in our cohorts could be attrib-
uted to our use of a strict definition of these toxicities.
We used the definitions recommended by the Dutch
Association of Rheumatologists, in which hepatotoxicity
is defined as an ALT level 3 times the upper limit of
normal, and bone marrow depression is defined as a
leukocyte count of ?3.0 ? 109/liter or thrombocyte
count of ?100 ? 109/liter. In the systematic review, liver
toxicity was defined as an increase in the aspartate
aminotransferase and/or ALT level above the upper
limit of normal, and presence of cytopenia (defined as
a decrease of ?2 gm/dl in the hemoglobin level, or a
platelet count of ?150 ? 109/liter, or a white blood cell
count of ?3.5 ? 109/liter).
Most of the studies in the systematic review did
not report or insufficiently reported the use of folic acid.
However, a total coverage of folic acid use would
probably only have resulted in fewer adverse events
being reported in the systematic review, because another
systematic review reported a 79% reduction in mucosal
and gastrointestinal side effects with the use of folic acid
(OR 0.21, 95% CI 0.10–0.44) (19). In summary, we
acknowledge that there are some differences in the
incidence of adverse events between the literature and
our cohorts, but these could be attributed to the above-
mentioned differences in definitions and population
None of the investigated baseline biomarker lev-
els were associated with occurrence of adverse events
after 3 months of treatment. In contrast to this finding,
12 patients with juvenile arthritis who had a history of
intolerance to MTX treatment were shown to have
significantly lower cellular folate concentrations when
compared to 81 patients who had never been treated
with MTX (31). In addition, low-to-normal initial levels
of plasma folate and red blood cell folate have been
associated with the future toxicity of MTX in RA
patients (32). In our cohorts, all patients were treated
with folic acid. This treatment has been proven to reduce
MTX-related adverse events in RA patients (19). This
could have diluted the relationship between the investi-
gated biomarker concentrations and adverse events in
The percentages of patients with adverse events
in the 2 cohorts were similar when the groups of patients
were compared according to their different dosing
schemes. A significantly higher percentage of patients
experienced malaise in the validation cohort, although
the MTX-dosing scheme was lower (15 mg/week) than in
the derivation cohort (25 mg/week). Malaise is a sub-
jective parameter that is collected in patient self-report
questionnaires. The questionnaires in the derivation
cohort had closed questions, such as “Were you tired last
week?” In contrast, the questionnaires in the validation
cohort had a more open character, such as “Write down
all adverse events from last week.” With closed ques-
tions, one would suspect that the scores would be higher,
because patients would be more aware of possible
adverse events. However, this difference in question-
naire designs probably did not have an influence, since
we observed a lower percentage of patients with malaise
only in the group who responded to closed questions on
the health status questionnaires. Taken together, these
results indicate that a higher MTX-dosing scheme did
not lead to more adverse events in our study.
There are some limitations to the present study.
First, the hypothesis was based on the MTX working
mechanism, and therefore MTX monotherapy would be
ideal. However, more than one-half of the patients in
both cohorts received other DMARDs in addition to
MTX. These drugs can also cause a response in terms of
modulating disease activity, and can produce similar side
effects as those related to MTX. Therefore, we cor-
rected all of the analyses for the use of other DMARDs.
LOW ERYTHROCYTE FOLATE LEVELS AND MTX NONRESPONSE IN RA2811
The corrected results were not significantly different
from the uncorrected results.
Second, the difference in MTX dosage between
the 2 cohorts was considerable (25 mg/week versus 15
mg/week; P ? 0.001), but all of our analyses were done
in the 2 cohorts separately. Nevertheless, the difference
in MTX dosage within each cohort was minimal, with
standard deviations of 1 mg/week in the derivation
cohort (median 25 mg/week, range 10–25) and 2 mg/
week in the validation cohort (median 15 mg/week,
range 5–25). We corrected for the MTX dosage in all of
Third, the levels of erythrocyte folate at baseline
were not linearly associated with treatment nonresponse
according to the EULAR criteria at 3 months (P ?
0.066) in the derivation cohort, although there was a
trend toward association. This might be explained by
the smaller sample size of the derivation cohort, due to
the restriction that only patients with a baseline DAS28
of ?3.3 could be assessed when applying the EULAR
Fourth, unfortunately, we did not register infor-
mation on the time relationship between folate supple-
mentation and administration of the MTX dose and
blood sample withdrawal. Patients in both cohorts were
advised to take folate supplementation 2 days after
receiving the MTX dose. However, it would be ex-
tremely difficult to monitor in what way patients fol-
lowed this instruction.
Moreover, the time between blood sample with-
drawal and MTX dose or folate supplementation was
not registered. The timing of folate supplementation and
blood sample withdrawal could have an impact on the
3-month serum folate concentrations, but would have
less impact on the erythrocyte folate concentrations at 3
months. We are most interested in predicting the clinical
response at the start of treatment (baseline), prior to
administration of any medication. Our results showed
that the baseline erythrocyte folate level was predictive
of the clinical response to MTX after 3 months of
treatment. If the time between folate supplementation
and blood sample withdrawal were to be standardized, it
could be that there might be an effect of the 3-month
serum folate concentration on disease activity. In con-
trast, the 3-month erythrocyte folate concentration did
not show an effect on the DAS28 after 3 months
(derivation cohort, ? ? ?0.08, P ? 0.283; validation
cohort, ? ? ?0.12, P ? 0.260).
In conclusion, our study is the first prospective
study to show that a lower baseline erythrocyte folate
level was associated with nonresponse to MTX after
3 months of treatment, as measured according to the
DAS28, in 2 independent cohorts. Thus, the baseline
erythrocyte folate level may be a promising new bio-
marker for prediction models of MTX nonresponse. In
contrast, baseline levels of plasma homocysteine, serum
vitamin B12, serum folate, and erythrocyte vitamin B6
were unrelated to MTX nonresponse in RA. None of the
investigated folate biomarkers were associated with oc-
currence of adverse events after 3 months.
The authors thank all patients who are enrolled in the
tREACH and MTX-R cohorts. Without their active coopera-
tion, this study would not be possible. The tREACH trial
involves the following rheumatology centers in The Nether-
lands: Erasmus UMC (Rotterdam), Sint Franciscus Hospital
(Rotterdam), Maasstad Hospital (Rotterdam), Vlietland Hos-
pital (Schiedam), Admiraal de Ruyter Hospital (Goes and
Vlissingen), Zorgsaam Hospital (Terneuzen), and Albert
Schweitzer Hospital (Dordrecht). The authors thank the fol-
lowing people from these centers for their contributions to the
tREACH and MTX-R studies: R. Aartsen, C. Alfenaar, C.
Alves, R. Arendse, M. Baak-Dijkstra, J. Bal-overzier, N.
Basoski, S. Beer, E. den Boer, F. Bonte, R. Brouwer, H. Buijs,
N. Buijs, E. Colin, R. Dolhain, C. Fleming, F. Fodili, A.
Gerards, J. van Gorp, P. Griffioen, B. Grillet, B. Hamelink,
K. Han, S. Heil, L. van Hove, M. Huisman, M. de Jager, S.
Joziasse, P. Krijger, M. van Krugten, C. van Leeuwen, J.
Luime, J. Nijs, B. Schaeybroeck, W. Schilleman, S. Schrauwen,
T. Sutter, W. Verbree, A. van der Voordt, M. de Vroed, M.
de Waart, M. Walter, A. Weel, H. Wintjes, B. van Zelst, and L.
All authors were involved in drafting the article or revising it
critically for important intellectual content, and all authors approved
the final version to be published. Dr. de Rotte had full access to all of
the data in the study and takes responsibility for the integrity of the
data and the accuracy of the data analysis.
Study conception and design. de Rotte, Pluijm, van Zeben, Lindemans,
Hazes, de Jonge.
Acquisition of data. de Rotte, de Jong, Barendregt, van der Lubbe,
de Sonnaville, Hazes, de Jonge.
Analysis and interpretation of data. de Rotte, de Jong, Pluijm,
C´alasan, Hazes, de Jonge.
1. Stamp LK, Roberts RL. Effect of genetic polymorphisms in the
folate pathway on methotrexate therapy in rheumatic diseases.
2. Smolen JS, Aletaha D, Bijlsma JW, Breedveld FC, Boumpas D,
Burmester G, et al. Treating rheumatoid arthritis to target:
recommendations of an international task force. Ann Rheum Dis
3. Raza K. Early rheumatoid arthritis–the window narrows. Rheu-
matology (Oxford) 2010;49:406–10.
4. Wessels JA, van der Kooij SM, le Cessie S, Kievit W, Barerra P,
2812 DE ROTTE ET AL
Allaart CF, et al, for the Pharmacogenetics Collaborative Re- Download full-text
search Group. A clinical pharmacogenetic model to predict the
efficacy of methotrexate monotherapy in recent-onset rheumatoid
arthritis. Arthritis Rheum 2007;56:1765–75.
5. Bulatovic M, Heijstek MW, Van Dijkhuizen EH, Wulffraat NM,
Pluijm SM, de Jonge R. Prediction of clinical non-response to
methotrexate treatment in juvenile idiopathic arthritis. Ann
Rheum Dis 2012;71:1484–9.
6. Katchamart W, Johnson S, Lin HJ, Phumethum V, Salliot C,
Bombardier C. Predictors for remission in rheumatoid arthritis
patients: a systematic review. Arthritis Care Res (Hoboken)
7. Chan ES, Cronstein BN. Methotrexate—how does it really work?
Nat Rev Rheumatol 2010;6:175–8.
8. Dervieux T, Furst D, Lein DO, Capps R, Smith K, Caldwell J, et
al. Pharmacogenetic and metabolite measurements are associated
with clinical status in patients with rheumatoid arthritis treated
with methotrexate: results of a multicentred cross sectional obser-
vational study. Ann Rheum Dis 2005;64:1180–5.
9. Stamp LK, O’Donnell JL, Chapman PT, Zhang M, James J,
Frampton C, et al. Methotrexate polyglutamate concentrations are
not associated with disease control in rheumatoid arthritis patients
receiving long-term methotrexate therapy. Arthritis Rheum 2010;
10. Claessen SJ, Hazes JM, Huisman MA, van Zeben D, Luime JJ,
Weel AE. Use of risk stratification to target therapies in patients
with recent onset arthritis: design of a prospective randomized
multicenter controlled trial. BMC Musculoskelet Disord 2009;10:
11. Aletaha D, Neogi T, Silman AJ, Funovits J, Felson DT, Bingham
CO III, et al. 2010 rheumatoid arthritis classification criteria: an
American College of Rheumatology/European League Against
Rheumatism collaborative initiative. Arthritis Rheum 2010;62:
12. Ducros V, Belva-Besnet H, Casetta B, Favier A. A robust liquid
chromatography tandem mass spectrometry method for total
plasma homocysteine determination in clinical practice. Clin
Chem Lab Med 2006;44:987–90.
13. Van Zelst BD, de Jonge R. A stable isotope dilution LC-ESI-
MS/MS method for the quantification of pyridoxal-5?-phosphate in
whole blood. J Chromatogr B Analyt Technol Biomed Life Sci
14. Zemlin AE, Essack Y, Rensburg M, Keller T, Brinkmann T.
Stability of red blood cell folate in whole blood and haemolysate.
Clin Lab 2010;56:391–6.
15. Prevoo ML, van ’t Hof MA, Kuper HH, van Leeuwen MA, van de
Putte LB, van Riel PL. Modified disease activity scores that
include twenty-eight–joint counts: development and validation in a
prospective longitudinal study of patients with rheumatoid arthri-
tis. Arthritis Rheum 1995;38:44–8.
16. Van Gestel AM, Prevoo ML, van ’t Hof MA, van Rijswijk MH, van
de Putte LB, van Riel PL. Development and validation of the
European League Against Rheumatism response criteria for rheu-
matoid arthritis: comparison with the preliminary American Col-
lege of Rheumatology and the World Health Organization/Inter-
national League Against Rheumatism criteria. Arthritis Rheum
17. Fisher MC, Cronstein BN. Metaanalysis of methylenetetrahydro-
folate reductase (MTHFR) polymorphisms affecting methotrexate
toxicity. J Rheumatol 2009;36:539–45.
18. De Jong PH, Hazes JM, Barendregt PJ, Huisman M, van Zeben D,
van der Lubbe PA, et al. Induction therapy with a combination of
DMARDs is better than methotrexate monotherapy: first results
of the tREACH trial. Ann Rheum Dis 2013;72:72–8.
19. Ortiz Z, Shea B, Suarez Almazor M, Moher D, Wells G, Tugwell
P. Folic acid and folinic acid for reducing side effects in patients
receiving methotrexate for rheumatoid arthritis. Cochrane Data-
base Syst Rev 2000:CD000951.
20. Khanna D, Park GS, Paulus HE, Simpson KM, Elashoff D, Cohen
SB, et al. Reduction of the efficacy of methotrexate by the use of
folic acid: post hoc analysis from two randomized controlled
studies. Arthritis Rheum 2005;52:3030–8.
21. Den Boer E, Meesters RJ, van Zelst BD, Luider TM, Hazes JM,
Heil SG, et al. Measuring methotrexate polyglutamates in red
blood cells: a new LC-MS/MS-based method. Anal Bioanal Chem
22. Stamp LK, O’Donnell JL, Chapman PT, Zhang M, Frampton C,
James J, et al. Determinants of red blood cell methotrexate
polyglutamate concentrations in rheumatoid arthritis patients re-
ceiving long-term methotrexate treatment. Arthritis Rheum 2009;
23. De Rotte MC, Bulatovic M, Heijstek MW, Jansen G, Heil SG, van
Schaik RH, et al. ABCB1 and ABCC3 gene polymorphisms are
associated with first-year response to methotrexate in juvenile
idiopathic arthritis. J Rheumatol 2012;39:2032–40.
24. Van Ede AE, Laan RF, Blom HJ, Boers GH, Haagsma CJ,
Thomas CM, et al. Homocysteine and folate status in methotrex-
ate-treated patients with rheumatoid arthritis. Rheumatology (Ox-
25. Haagsma CJ, Blom HJ, van Riel PL, van’t Hof MA, Giesendorf
BA, van Oppenraaij-Emmerzaal D, et al. Influence of sulphasala-
zine, methotrexate, and the combination of both on plasma
homocysteine concentrations in patients with rheumatoid arthritis.
Ann Rheum Dis 1999;58:79–84.
26. Morgan SL, Baggott JE, Lee JY, Alarcon GS. Folic acid supple-
mentation prevents deficient blood folate levels and hyperhomo-
cysteinemia during longterm, low dose methotrexate therapy for
rheumatoid arthritis: implications for cardiovascular disease pre-
vention. J Rheumatol 1998;25:441–6.
27. Morgan SL, Baggott JE, Refsum H, Ueland PM. Homocysteine
levels in patients with rheumatoid arthritis treated with low-dose
methotrexate. Clin Pharmacol Ther 1991;50:547–56.
28. Sakakeeny L, Roubenoff R, Obin M, Fontes JD, Benjamin EJ,
Bujanover Y, et al. Plasma pyridoxal-5-phosphate is inversely
associated with systemic markers of inflammation in a population
of U.S. adults. J Nutr 2012;142:1280–5.
29. Lotto V, Choi SW, Friso S. Vitamin B6: a challenging link between
nutrition and inflammation in CVD. Br J Nutr 2011;106:183–95.
30. Salliot C, van der Heijde D. Long-term safety of methotrexate
monotherapy in patients with rheumatoid arthritis: a systematic
literature research. Ann Rheum Dis 2009;68:1100–4.
31. Becker ML, van Haandel L, Gaedigk R, Thomas B, Hoeltzel MF,
Lasky A, et al. Red blood cell folate concentrations and polyglu-
tamate distribution in juvenile arthritis: predictors of folate vari-
ability. Pharmacogenet Genomics 2012;22:236–46.
32. Morgan SL, Baggott JE, Vaughn WH, Young PK, Austin JV,
Krumdieck CL, et al. The effect of folic acid supplementation on
the toxicity of low-dose methotrexate in patients with rheumatoid
arthritis. Arthritis Rheum 1990;33:9–18.
LOW ERYTHROCYTE FOLATE LEVELS AND MTX NONRESPONSE IN RA2813