Genetic SLC variants in MTX efficacy and toxicity in Caucasian RA patients
Keywords:methotrexate; rheumatoid arthritis; pharmacogenetics; SLC genes; SLC22A2; SLC28A2; SLCO4A1; SNP
Currently, low-dose methotrexate is the first-line therapy for the treatment of rheumatoid arthritis. Despite of the high efficacy of methotrexate, 10 % of the patients discontinues the treatment due to toxic side effects. Using pharmacogenetics both toxic side effects and efficacy of methotrexate treatment can be estimated.
In our study the influence of five earlier investigated single nucleotide peptides (rs2236553, rs624249, rs316019, rs10519020 and rs1060896), encoding for solute carrier genes, were examined on the toxicity and efficacy of methotrexate in patients gathered from the BeST study. The efficacy was determined by the European League Against Rheumatism response and the toxicity was determined by laboratory measures (liver enzymes such as ALT) and by interrogations performed by physicians. The isolated DNA samples were amplified by quantitative polymerization chain reaction (Sensoquest or Thermocycler) and genotyped by pyrosequencing (PyroMark Q96: Qiagen) and High Resolution Melting (Lightscanner). For statistical analysis multiple logistic regression was done.
According to our results, all investigated solute carrier genes (SLCO4A1, SLC22A2 and SLC28A2) showed no significant association (p > 0.05) with both efficacy and toxicity of methotrexate in Caucasians. These findings were in disagreement with earlier findings of Aslibekyan et al.. Further research on these solute carrier genes in other populations is warranted. Besides, future investigations on other potential biomarkers is preferred in Caucasian patients to realize personalized therapeutic approaches for early aggressive rheumatoid arthritis patients.
Rheumatoid arthritis (RA) is an autoimmune disease that affects approximately 1.5 % of the Dutch
population. Hereby, the body mistakenly attacks the joints, that results in inflammation that causes the tissue that lines the inside of joints (so called synovium) to thicken, resulting in swelling and pain in and around the joints. The synovium makes a fluid (synovial fluid or synovia) that lubricates joints and hereby reduces friction during movement.,, Because of this, the quality of life reduces in patients with RA. There is seen that most of the patients with RA (40 %) stop or reduce working within ten years.
Currently, there are drugs that reduce the synovial damage in joints. These drugs can be distinguished by three different groups: non-steroidal anti-inflammatory drugs (NSAIDs), glucocorticosteroids such as prednisone and disease-modifying anti-rheumatic drugs (DMARDs). NSAIDs include aspirin, ibuprofen and diclofenac, by which aspirin and ibuprofen are weakly COX-1 inhibitors and diclofenac is a weakly COX-2 inhibitor. There are two types of DMARDs: biological DMARDs such as abatacept and immunosuppressive small molecules DMARDs such as methotrexate (MTX). Decades ago, RA patients started their treatment with aspirin alone. Despite NSAIDs have an anti-inflammatory effect, treatment with aspirin alone is not sufficient enough. Besides, treatment with DMARDs is more effective than non-DMARDs. Therefore, an initial treatment of DMARDs combined with NSAIDs leads to a better suppression of the disease activity nowadays.
At the moment, MTX is the first line in treatment of RA, because of its known efficacy and experience (MTX has been prescribed since 1988)., However, a certain percentage of RA patients failed on MTX due to side effects or low efficacy which was observed within 1 year of treatment (15-30 % side effects; 40-60 low efficacy).,,, Those toxic side effects include hepatotoxicity, gastrointestinal problems, stomatitis or soreness of the mouth, abnormal liver chemistries (2-3 upper limit of normal (ULN) of the hepatic transaminases), central nervous system symptoms (including headache, fatigue), alopecia, fever (drug-related, although fever can also occur due to infection), hematologic abnormalities, particularly macrocytosis, and myelosuppression.,
However the mechanism of action of MTX is not clearly known, previous studies found that these toxic side effects depend on genetic factors.,, Small differences in genetic material can have a lot of impact on genetic outcomes. For example, Single Nucleotide Polymorphisms (SNPs) may have a huge impact on fenotypes of the specific genes. A SNP is a difference in one nucleotide on specific locations in genes. When these genetic factors can be examined, treatment with MTX can be improved. With the aid of pharmacogenetics the efficacy and toxicity of MTX can be estimated in patients with RA. In pharmacogenetic studies different SNPs in encoding proteins can be detected by several methods such as high resolution melting (HRM) and pyrosequencing (PSQ). Different SNPs can lead to different outcomes of efficacy and toxicity of MTX. Hereby, toxicities can be prevented and efficacy can be improved. However, pharmacogenetics is preferred in investigations, the ethics also play a major role. Sometimes, patients are not willing to participate the study because of privacy issues.
Reduced Folate Carriers (RFC)
Previous studies show correlations between the reduced folate carrier-1 and the efficacy and toxicity of MTX. MTX reduces folate acid levels by inhibiting folate-dependent enzymes like thymidylate synthase, dihydrofolate reductase and 5-aminoamidazole-4-carboxamide ribonucleotide transformylase. Lower folate acid levels can probably result in toxic side effects. MTX enters the cell via reduced folate carrier-1 (RFC-1). The efficacy and toxicity can depend on this transporter. The higher the affinity of MTX is on those transporters the higher the efficacy will be. Although, the toxicity can also rise because of lower folate acid levels.
Solute Cariers (SLC)
One well known type of RFC-1 transporters is the solute carrier (SLC). SLCs can be located around the joints, but they can also be located anywhere else such as the liver, kidneys, intestines and skin. When MTX binds to one of the receptors in the liver, kidneys or intestines toxicities may arise. Previous studies found some associations between these SLC genes (Table 1). One of these SLC receptors is the SLC receptor that encodes for the SLC22A2 gene, also known as organic cation transporter 2 (Oct2). These receptors are located in the liver and kidneys which can induce toxicities. Sook Wah Yee et al. examined the identification and characterization of the SLC28A2 transporter. This study showed that SLC28A2 play a major role in the absorption, disposition and biological effects in the intestine and the liver. Ryoichi Fujiwara et al. concluded that SLCO4A1 receptors are 30-fold more located in the skin than in the liver which means that toxicity in the liver may be 30-fold lower than on the skin. Aslibekyan et al. investigated the influence of three different SLC genes on the efficacy and toxicity of MTX in patients with RA. These genes include: SLCO4A1 (rs2236553), SLC22A2 (rs316019 and rs624249) and SLC28A2 (rs1060896 and rs10519020). They concluded that there is an association between these genes and both efficacy and toxicity of MTX. They expected that SNP rs2236553 (SLCO4A1) will show the highest correlation with the efficacy of MTX. On the contrary, SNP rs316019 (SLC22A2) and rs624249 (SLC22A2) will probably show the highest association with toxicities according to Aslibekyan et al., but further research is required. In another study (Goekoop-Ruiterman et al.), the efficacy in a population in the western part of the Netherlands (BeST-study) was examined. In this study efficacy was chosen as primary endpoint. A Disease Activity Score in 44 joints (DAS44) was determined to measure the efficacy. The study population was randomized into four different treatment arms. In concluding, the initial combination treatment group consisting of whether prednisone or infliximab, show better efficacy compared with monotherapy of MTX and combination step-up therapy of MTX.
Table 1. Earlier found associations between SLC genes (SLCO4A1, SLC22A2 and SLC28A2) and MTX efficacy and toxicity.
Studies Efficacy Toxiticy
Aslibekyan et al. SLCO4A1 (rs2236554) SLC22A2 (rs316019, rs624249)
SLC28A2 (rs10519020, rs1060896)
Jonker, J. W. et al.; Fujiwara, Ryoichi et al. SLC22A2 (rs316019, rs624249)
Sook Wah Yee et al. SLC28A2 (rs10519020 ,rs1060896)
The CHIMERA study
During this study the association between five earlier investigated SLC SNPs and the efficacy and toxicity of MTX were examined in 350 patients gathered from the BeST study (Goekoop-Ruiterman et al.). The single nucleotide polymorphisms (SNPs) were gathered from the study of Aslibekyan et al. which were the following: rs2236553 (SLCO4A1), rs624249 (SLC22A2), rs316019 (SLC22A2), rs10519020 (SLC28A2) and rs1060896 (SLC28A2). All BeST-DNA samples were genotyped by pyrosequencing (PSQ) or High Resolution Melt (HRM). PSQ was carried out by Pyromark Q96 ID (Qiagen) and HRM was done by the Lightscanner (Idaho Technologie Inc.). The efficacy of MTX was determined by the European League Against Rheumatism (EULAR) response. The toxicity was derived from laboratory measures (liver enzymes such as ALT *ULN) and a self-report of adverse events.
In conclusion, our aim was to replicate the findings of Aslibekyan et al. and to bring them in a higher level, so that a priori prediction of MTX for individual patient could be made for the response (efficacy and toxicity) in RA. If these SLC-genes are again associated with MTX response, it is useful as a biomarker and potentially cost-saving in the clinical setting. Our hypothesis was that there would be an association between the SNPs encoding for the genes SLCO4A1 (rs2236553), SLC22A2 (rs624249, rs316019) and SLC28A2 (rs10519020, rs1060896) and the toxicity and/or efficacy of MTX.
Materials and Methods
From the BeST-study, 352 patients were included. The inclusion and exclusion criteria were described in the BeST study. Another criteria was that patients had to be treated for at least six months with MTX and had an DAS on baseline and 6 months.
Patients were allocated into four different treatment arms. The first group was treated with monotherapy of MTX which consisted of 15 mg/week MTX. The dose was increased to 25 ' 30 mg/week MTX, when a DAS of > 2.4 was found. When patients gave still bad response to MTX, MTX treatment was replaced with sulfasalazine (SSZ) or leflunomide monotherapy. The second group was treated with a step-up combination therapy of MTX (15 mg/week). The dose was increased to 25 ' 30 mg/week MTX, when a DAS of > 2.4 was found. SSZ, hydroxychloroquine (HCQ) and prednisone were added to the treatment when patients gave still bad response. The third group was treated with an initial combination therapy consisting of MTX and prednisone which consisted of 7.5 mg/week MTX, 2,000 mg/day SSZ, and 60 mg/day prednisone. The dose of MTX was increased to 25 ' 30 mg/week when DAS > 2.4. Combination was replaced by MTX with CSA and prednisone or MTX with infliximab, when patients still showed insufficient improvement in DAS. Finally, treatment was tapered to zero. The fourth group was treated with an initial combination therapy consisting of MTX and infliximab with a dose of 25 ' 30 mg/week MTX, 3 mg/kg infliximab. The dose of infliximab was increased to 6 mg/kg/every 8 weeks when the DAS was > 2.4.
The following patient information was provided: gender, age of disease onset, age at treatment onset, methotrexate dosage, smoking, other concomitant drugs such as NSAIDs and corticosteroids, rheumatoid factor and anti-CCP antibody seropositivity, disease activity score 44 (at baseline, on 3 and 6 months). These patient information was gathered before treatment and after every six months of treatment to determine the efficacy and toxicity.
Assessment of endpoints
Every three months assessment of the endpoints was performed. The efficacy of MTX was determined by the European League Against Rheumatism (EULAR) response. Patients were categorized as good responders (according the EULAR criteria) when the DAS on 6 months was decreased by at least 1.2 points compared with baseline and with. A DAS had to be ' 2.4 on 6 months treatment.
The toxicity was measured by laboratory measures for liver enzymes such as ALT *ULN and a self-report existing of questions about adverse events. This was gathered at the hospital every three months. The adverse events of one year follow-up were obtained and included into the study. Adverse events were distinguished in serious adverse events like hepatotoxicity and adverse events like headache.
Firstly, 2.5 ng/''l (4 ''l per well) of isolated DNA samples were added to a 96 well plates. The systems Sensoquest or Thermocycler were used for polymerization chain reaction (PCR) to amplify the preferred SNPs. During this study two different techniques were used for genotyping: Pyrosequencing (PSQ) and High Resolution Melting (HRM).
rs10519020 (SLC28A2) was genotyped by PSQ. The other four SNPs were genotyped by HRM. For this SNP another genotype method was required such as PSQ, because the different SNP nucleotides show approximately the same melting points. Right before genotyping the SNP was amplified by Sensoquest system for PCR. Q-solution was added to each well to facilitate the amplification. Q-solution changes the melting behavior of DNA and will improve a PCR caused by templates that have a high degree of secondary structure or in our case a high GC-contents. Genotyping was performed by using the PyroMark Q96 ID (Qiagen) system.
PSQ is based on the DNA 'sequencing by synthesis' principle. For DNA polymerase a primer is needed to start making a new DNA strand. The primer is a short single-stranded RNA which binds at the beginning of the DNA-replication via base-pairing to the DNA template strand. Our primer was Biotin labeled. After PCR one of the DNA strand will be removed by Vacuum Prep Tool (VPT). Streptavidin Sepharose beads were added to the PCR product and were bound to the Biotin label. The diameter of the filter probes on VPT were smaller than the added beads. Because of this, the Biotin labeled strand remained on the filter probes and the other non-Biotin labeled DNA strand was sucked by VPT. The template strand remained and after some wash steps consisting of ethanol, NaOH and washing buffer a primer plus annealing buffer was added to this template strand. The primer bound specifically to the ssDNA under heating. Thereafter, the four different nucleotides (A, T, C, G) were added one by one.
When a nucleotide is built in the sequence, pyrophosphate was released. This resulted in adenosine triphosphate (ATP). ATP converted luciferin into oxiluciferin by using luciferase and light was omitted. The omitted light was detected by the system. In figure 1 three pyrograms are given for rs1051920. The found genotypes for the specific SNP is indicated by a green arrow. The amount of nucleotides in a row in the sequence could be determined by peak ratios. In this case, three T nucleotides followed after the SNP. In the pyrogram the T nucleotide gave approximately three times more light intensity than the SNP nucleotide. In figure 1B a heterozygote is given by which the peak intensities is divided for two SNPs.
High Resolution Melting (HRM)
The other four SNPs (rs624249, rs1060896, rs316019 and rs223655) were genotyped by the Lightscanner (Idaho Technologie Inc.). The primers were validated at first to determine the PCR conditions. For all PCRs 45 cycli were performed. The obtained genotypes were confirmed by Sanger sequencing.
HRM is a method for genotyping based on melting temperatures. Differences in melting temperatures depend on several variables: fragment length, base pair composition and GC content. For example, C and G nucleotides have higher melting temperatures than A and T nucleotides, because of the different amount of hydrogen bonds between the base pairs.
The system differentiates the different melting temperatures by fluorescence. In this study LCgreen was chosen as fluorescence. The fluorescence dye (LCgreen) bound on the nucleotides during PCR. During HRM the temperature rose and at a given moment the DNA strands denaturized. Because of this the fluorescence dye lost its connection with the nucleotides. This resulted in a decrease in fluorescence. These changes in fluorescence were registered by the system. The arisen graphs of rs2236553 are given in figure 2. The found genotypes were confirmed by Sanger sequencing of validated DNA samples gathered from healthy persons from the Netherlands.
We determined the association between the SNPs and the primary endpoints toxicity or efficacy by logistic regression. All regression analyses were analyzed for covariates. If the covariate was statistical significant, it was added to the model. The covariates were DAS28 at baseline, concomitant drug treatment (treatment arms), smoking (yes/no), Rf-status (positive/negative), patients age and gender. Since, we tested for the same gene, we should not adjust for multiple testing (e.g. with Bonferroni, Sidak or FDR). Statistical analyses were performed using SPSS 20.0 software (SPSS Inc., Chicago, IL, USA) and (G)Plink software (http://pngu.mgh.harvard.edu/purcell/plink/). Genetics were
analyzed with the additive model. Additive models imposed a structure in which each additional copy of a variant allele increases the response or toxicity, measured by the (log) odds ratio, by the same amount.
The most common adverse events on MTX is gastrointestinal.
The general characteristics of the 324 included RA patients are summarized in Table 2. Most participants were female (~ 68 %) and the mean age at treatment onset was approximately 54 years with a standard deviation (SD) of 13.4. 35.5 % of all participants were active smokers and approximately 65 % of the patients had a positive Rheumatic factor (Rf). The mean dose of MTX after 6 months of treatment was in all treatment groups 19.9 mg/week (SD '' 6.9). After six months of treatment 48.8 % of the participants showed good response (EULAR Response) to the given treatment.
During this study 28 patients were excluded for analysis, because of the following reasons: no or less than six months MTX use (n=24), patients by which a DAS was not registered for more than six months (n=3) and patients whose DNA was not available for genotyping (n=1). 158 patients were assigned as cases for EULAR response and 138 patients were assigned as cases for adverse or serious adverse events during 1 year follow up (Table 2). 57 out of 324 patients were classified as cases for both efficacy and toxicity.
Table 2. General characteristics of the study population (n=324)
Age, years 54.3 '' 13.4
Female, n (%) 220 (67.9)
Methotrexate doses @6 months, mg/week 19.9 ''6.9
Rf-positive, n (%)
Smoking, n (%)
Disease duration, weeks 211 (65.1)
4.6 '' 9
Sequential monotherapy, n (%) 93 (28.7)
Step-up combination therapy, n (%) 90 (27.8)
Initial combination therapy with infliximab, n (%) 68 (21.0)
Initial combination therapy with tapered high-dose prednisone, n (%) 73 (22.5)
DAS at baseline, points 4.4 '' 0.8
EULAR Response, 6 months, n (%) 158 (48.8)
Experience any adverse or serious adverse events during 1 yr follow-up, n (%) 138 (42.6)
Abbreviations: DAS = Disease Activity Score, Rf=Rheumatoid factor
Table 3 showed the adverse events or serious adverse events during the first year of treatment. Total, 42.6 % of the patients showed (serious) adverse events. Most adverse events were observed in participants who were allocated in treatment group 2, subsequently step-up combination therapy (53.3 %). Gastrointestinal adverse events such as liver toxicities, nausea, abdominal pains and diarrhea were found as the most common adverse events in all four treatment arms.
Table 3. Any adverse events or serious adverse events during 1 year follow-up
Adverse events Sequential monotherapy,
n (%) Step-up combination therapy, n (%) Initial combination therapy with tapered high-dose prednisone Initial combination therapy with infliximab Total
Central nerve system
Trauma / fractures 17 (18.3)
0 (0.0) 19 (21.1)
0 (0.0) 5 (6.8)
0 (0.0) 10 (14.7)
1 (1.5) 51 (15.7)
Total 37 (39.8) 48 (53.3) 26 (35.6) 26 (38.2) 138 (42.6)
The allele frequencies in cases and controls are close together.
For five SNPs that encode for SLCO4A1, SLC22A2 and SLC28A2 the association with the efficacy and toxicity of MTX was examined. The association was determined by multiple logistic regression with gender, DAS at baseline, randomization groups, age, Rf and smoking as covariates. HW was used to determine the SNP distribution in the examined population (Table 4 and Table 5). rs1060896 and rs10519020 showed high HW values. A HW of 7.47 for rs1060896 and a HW of 19.82 for rs10519020 respectively. The minor allele frequencies (MAF) for cases and controls were close together in all five SNPs which means that a given SNP allele was found just as often in both cases and controls (Table 4 and Table 5).
SLCO4A1, SLC22A2 and SLC28A2 genes have no association with MTX efficacy in patients with early progressive RA after 6 months of treatment.
The association between the investigated SLC genes and the efficacy of MTX is given in Table 4. All univariated p-values were not significant. A p-value of 0.05 was seen as significant. When covariates were added to the model, p-values, that represented the association of rs316019 (SLC22A2) and rs1060896 (SLC28A2), showed better values than the univariated p-values. In this case a p-value of < 0.008 was significant. However, p-values for other SNPs gave worse p-values after correction with covariates. Remarkably, in all ORs value the 95 % confidence intervals of both univariate and multivariate analysis contained value 1.
Table 4. SLCO4A1, SLC22A2 and SLC28A2 association with efficacy of MTX. Minor allele frequencies of cases and controls, p-values univariate and multivariate are given. P < 0.008 was seen as significant for multiple linear regression.. Gender, DAS at baseline, randomization groups, age, rheumatoid factor and smoking were covariates in multiple linear regression.
SNP MA MAF CASE/CONTROL HW P UNI OR UNI '' 95 % CI P MULTI OR MULTI '' 95 % CI
rs223655 C 0.2957 / 0.2355 2.88 0.1557 1.274 (0.912-1.779) 0.2811 1.221 (0.8493-1.755)
rs624249 A 0.3897 / 0.3669 0.00 0.3748 1.157 (0.8381-1.598) 0.5787 1.105 (0.7763-1.574)
rs316019 T 0.1184 / 0.1277 0.35 0.7491 0.9245 (0.5713-1.496) 0.5685 0.8544 (0.4974-1.467)
rs1060896 C 0.3447 / 0.3796 7.47 0.6801 0.9383 (0.693-1.27) 0.4811 0.8861 (0.6329-1.241)
rs10519020 C 0.009756 / 0.01481 19.82 0.4021 2.079 (0.3753-11.52) 0.4702 1.965 (0.3142-12.29)
Abbreviations: SNP = Single Nucleotide Polymorphism; MA = Minor Allele; MAF = Minor Allele Frequency; OR = Odd Ratio; CI = Confidence Interval; UNI = Univariate; MULTI = Multivariate.
SLCO4A1, SLC22A2 and SLC28A2 genes have no association with MTX toxicities in patients with early progressive RA after 6 months of treatment.
In Table 5 the correlation between the investigated SLC genes and toxicities of MTX is presented. All p-values were not significant despite of correction for covariates. In univariated analysis, values were significant when p < 0.05. Correction for covariates was done twice in which one multivariate analysis was done with all covariates included (significance when p < 0.008) and the other multivariate analysis was done with only Rf and smoking as covariates included (significance when p < 0.025). Like the efficacy, all toxicity values for 95 % CI of ORs contained 1.
Table 5. SLCO4A1, SLC22A2 and SLC28A2 association with toxicity of MTX. Minor allele frequencies of cases and controls, p-values univariate and multivariate are given. Gender, DAS at baseline, randomization groups, age, rheumatoid factor and smoking were covariates in multiple linear regression. MULTI 1 is multiple linear regression corrected by all covariates (p < 0.008). MULTI 2 = correction by two covariates: rheumatoid factor and smoking (p < 0.025).
SNP A1 MAF CASE/CONTROL HW P UNI OR UNI '' 95 % CI p MULTI 1 OR MULTI 1 '' 95 % CI p MULTI 2 OR MULTI 2 '' 95 % CI
rs223655 C 0.2957 / 0.2355 2.88 0.3164 1187 (0.849-1.658) 0.2967 1.201 (0.8517-1.692) 0.3189 1189 (0.8456- 1.673)
rs624249 A 0.3897 / 0.3669 0.00 0.6228 0.9215 (0.6652-1.277) 0.5382 0.8999 (0.6432-1.259) 0.4767 0.8861 (0.6351-1.236)
rs316019 T 0.1184 / 0.1277 0.35 0.1394 0.6844 (0.4138-1.132) 0.1807 0.7051 (0.4227-1.176) 0.1751 0.7044 (0.4245-1.169)
rs1060896 C 0.3447 / 0.3796 7.47 0.8482 0.9706 (0.7146-1.318) 0.9397 0.9878 (0.7186-0.1.358) 0.9029 0.9806 (0.7162-1.343)
rs10519020 C 0.009756 / 0.01481 19.82 0.07898 6.923 (0.7993-59.96) 0.07819 7.358 (0.7984-1.133) 0.0981 6.348 (0.7106-56.7)
Abbreviations: SNP = Single Nucleotide Polymorphism; MA = Minor Allele; MAF = Minor Allele Frequency; OR = Odd Ratio; CI = Confidence Interval; UNI = Univariate; MULTI = Multivariate.
Discussion and Conclusion
This study demonstrated that the efficacy and toxicities of MTX were not significant associated with the investigated polymorphism that encodes for the SLC genes (SLCO4A1, SLC22A2 and SLC28A2). The results were not in agreement with our hypothesis which was based on previous studies. Besides, the not significant P-values showed calculated OR values of approximately 1, which means that the appearance of efficacy and/or toxicity in cases and controls was equal. Because of that, there could not be any association between the studied SNPs and both efficacy and toxicity of MTX.
Despite these conflicting findings, the found (serious) adverse events in this study agreed findings from previous studies. These adverse and serious adverse events could be related to MTX usage. The most common adverse events were gastrointestinal. The two initial combination therapies showed the least (serious) adverse events. These findings were in consistent with the outcomes in the BeST-study.
In contrast with the study of Aslibekyan et al., our population was homogenous which existed mostly out of Caucasians, thus limiting our findings to applicate it to other patient populations and to directly compare with the findings from Aslibekyan et al.. The study population of Aslibekyan et al. consisted of Caucasians, African American and other populations, whereas in our population mainly Caucasians were included. Other possible outcome differences can be explained due to our study disabilities. The statistics were done by multiple logistic regression, whereas Aslibekyan et al. used pure lasso technique as statistical method. The LASSO-technique involved both variable selection and regularization in order to form an prediction. However, the technique does not give quantitative data, for example it does not give p-values, Odds or Hazard Ratio's and thereby is the technique not directly comparable with regression method. Despite we corrected for multiple testing, another potential statistical error might be caused by false correction in multiple logistic regression. Another difference in statistics in comparison with Aslibekyan et al. is that they used a dominant and recessive model. In our study an additive model was used for analyzing of genetics. In our opinion it is better to analyze the genetics with an additive model, because the genetic model was not well known. All variants of genotypes can be analyzed with an additive model, whereas with a dominant or a recessive model only the dominant or the recessive alleles are analyzed. When a underlying genetic model was not well known, the outcomes can be therefore less accurate., Besides, correction by covariates was done differently in our study compared with Aslibekyan's study. Aslibekyan et al. included baseline DAS28, treatment arms, race, sex, age and smoking status in both efficacy model and toxicity model. In our study not all these covariates were included.
Besides, the population, which Aslibekyan et al. examined, BeST-patients received other kind of treatment than the TEAR study. Instead of initial combination therapy of MTX with infliximab or prednisone, they used etanercept or SSZ plus HCQ. The step-up combination therapy consisted of MTX and etanercept when efficacy was too low, whereas in our population MTX was combined with SSZ and HCQ when efficacy was too low. MTX monotherapy almost agreed our treatment of MTX monotherapy. Differences in treatment groups could also affect the different study outcomes. Therewithal, all different treatment arms were merged during statistical analysis in Aslibekyan's study. The found efficacy and toxicity values can also be caused by other DMARDs or anti-rheumatic drugs. Moreover, our findings cannot be compared with Aslibekyan et al. well, because they did not give any p-values nor OR values.
In our study the efficacy and toxicity were determined according to other criteria which could also have an influence on the main study outcomes due to false case control ratios. During this study the efficacy was determined by the EULAR response, whereas the efficacy in Aslibekyan's study was based on DAS28. The EULAR response is more precise than the DAS28, so we can assume that our findings were better than the findings in Aslibekyan's study. The toxicity in this study was gathered from adverse events recorded by questions from the treating physician, whereas in the study of Aslibekyan et al. the toxicity was derived from self-reports that patients filled in at home by themselves. Because physicians have an objective view on patients' toxicities, the results gathered from physicians are more reliable than results from self-reports.
Previous studies such as Aslibekyan et al. suggested that SLC genes (SLCO4A1, SLC22A2 and SLC28A2) could be used as biomarkers for efficacy and toxicity of MTX in RA patients. The aim of our study was to confirm these outcomes by replicating the study of Aslibekyan et al. With our findings personalized therapeutic approaches could be realized. Thereby, efficacy of MTX could be improved and toxicities of MTX could be prevented in patients with early aggressive RA. According to our study SLCO4A1, SLC22A2 and SLC28A2 cannot be used as biomarkers for MTX efficacy and toxicity in Caucasians. Therefore, future investigations to other potential biomarkers such as ALDH2 and several CYP enzymes are needed to realize personalized therapeutic approaches for early aggressive RA.
Further studies in different treatment populations is required to determine the relations between the SLC genes and the toxicity and efficacy of MTX. Besides, the determination of the efficacy and toxicity must be done precisely and objectively to prevent false outcomes.
In conclusion, according to our findings SLCO4A1, SLC22A2 and SLC28A2 genes were not associated with efficacy nor toxicity of MTX in Caucasians. Further research for these SLC genes in other populations is warranted. Also, further investigation with other potential biomarkers in Caucasian patients is preferred. Another conclusion is that the earlier found (serious) adverse events in the BeST study agreed our findings. Initial combination therapy provide better progression of RA in early progressive RA patients which was also in consistent with the findings from the BeST study.
 'Reumafonds - Home.' .
 H. P. Rang, M. Dale, J. M. Ritter, R. J. Flower, and G. Henderson, General Anaesthetic Agents. 2012.
 D. M. Lee and M. E. Weinblatt, 'Rheumatoid arthritis.,' Lancet, vol. 358, no. 9285, pp. 903'911, 2001.
 D. L. Scott, F. Wolfe, and T. W. J. Huizinga, 'Rheumatoid arthritis.,' Lancet, vol. 376, no. 9746, pp. 1094'108, 2010.
 R. B. Altman, D. Flockhart, and D. B. Goldstein, Principles of pharmacogenetics and pharmacogenomics. 2012.
 P. L. Mottram, 'Past, present and future drug treatment for rheumatoid arthritis and systemic lupus erythematosus,' Immunology and Cell Biology, vol. 81, no. 5. pp. 350'353, 2003.
 R. R. Brinker and P. Ranganathan, 'Methotrexate pharmacogenetics in rheumatoid arthritis,' Clin. Exp. Rheumatol., vol. 28, no. 5 SUPPL. 61, 2010.
 W. M. Kooloos, T. W. J. Huizinga, H.-J. Guchelaar, and J. a M. Wessels, 'Pharmacogenetics in treatment of rheumatoid arthritis.,' Curr. Pharm. Des., vol. 16, no. 2, pp. 164'75, 2010.
 W. M. Kooloos, J. A. Wessels, T. van der Straaten, C. F. Allaart, T. W. Huizinga, and H.-J. Guchelaar, 'Functional polymorphisms and methotrexate treatment outcome in recent-onset rheumatoid arthritis.,' Pharmacogenomics, vol. 11, pp. 163'175, 2010.
 L. Klareskog, D. van der Heijde, J. P. de Jager, A. Gough, J. Kalden, M. Malaise, E. Martin Mola, K. Pavelka, J. Sany, L. Settas, J. Wajdula, R. Pedersen, S. Fatenejad, and M. Sanda, 'Therapeutic effect of the combination of etanercept and methotrexate compared with each treatment alone in patients with rheumatoid arthritis: double-blind randomised controlled trial,' Lancet, vol. 363, no. 9410, pp. 675'681, 2004.
 T. Mottonen, P. Hannonen, M. Leirisalo-Repo, M. Nissila, H. Kautiainen, M. Korpela, L. Laasonen, H. Julkunen, R. Luukkainen, K. Vuori, L. Paimela, H. Blafield, M. Hakala, K. Ilva, U. Yli-Kerttula, K. Puolakka, P. Jarvinen, M. Hakola, H. Piirainen, J. Ahonen, I. Palvimaki, S. Forsberg, K. Koota, and C. Friman, 'Comparison of combination therapy with single-drug therapy in early rheumatoid arthritis: a randomised trial. FIN-RACo trial group,' Lancet, vol. 353, no. 9164, pp. 1568'1573, 1999.
 G. Ferraccioli, M. De Santis, and B. Tolusso, 'Pharmacogenetics/pharmacogenomics and antirheumatic drugs in rheumatology.,' Pharmacogenomics, vol. 5, pp. 1107'1116, 2004.
 'trexall-drug/patient-images-side-effects.htm.' [Online]. Available: http://www.rxlist.com/trexall-drug/patient-images-side-effects.htm. [Accessed: 26-May-2016].
 S. Aslibekyan, E. E. Brown, R. J. Reynolds, D. T. Redden, S. Morgan, J. E. Baggott, J. Sha, L. W. Moreland, J. R. O'Dell, J. R. Curtis, T. R. Mikuls, S. L. Bridges, and D. K. Arnett, 'Genetic variants associated with methotrexate efficacy and toxicity in early rheumatoid arthritis: results from the treatment of early aggressive rheumatoid arthritis trial.,' Pharmacogenomics J., vol. 14, no. April 2013, pp. 48'53, 2014.
 Y. P. M. Goekoop-Ruiterman, J. K. De Vries-Bouwstra, C. F. Allaart, D. Van Zeben, P. J. S. M. Kerstens, J. M. W. Hazes, A. H. Zwinderman, H. K. Ronday, K. H. Han, M. L. Westedt, A. H. Gerards, J. H. L. M. Van Groenendael, W. F. Lems, M. V. Van Krugten, F. C. Breedveld, and B. A. C. Dijkmans, 'Clinical and radiographic outcomes of four different treatment strategies in patients with early rheumatoid arthritis (the best study): A randomized, controlled trial,' Arthritis Rheum., vol. 52, no. 11, pp. 3381'3390, 2005.
 B. Alberts, A. Johnson, J. Lewis, M. Raff, K. Roberts, and P. And Walter, Molecular Biology of the Cell, vol. 54. 2008.
 P. Ranganathan, R. Culverhouse, S. Marsh, A. Mody, T. J. Scott-Horton, R. Brasington, A. Joseph, V. Reddy, S. Eisen, and H. L. McLeod, 'Methotrexate (MTX) pathway gene polymorphisms and their effects on MTX toxicity in caucasian and african american patients with rheumatoid arthritis,' J. Rheumatol., vol. 35, no. 4, pp. 572'579, 2008.
 J. W. Jonker, E. Wagenaar, S. Van Eijl, and A. H. Schinkel, 'Deficiency in the organic cation transporters 1 and 2 (Oct1/Oct2 [Slc22a1/Slc22a2]) in mice abolishes renal secretion of organic cations.,' Mol. Cell. Biol., vol. 23, no. 21, pp. 7902'8, 2003.
 R. Fujiwara, S. Takenaka, M. Hashimoto, T. Narawa, and T. Itoh, 'Expression of human solute carrier family transporters in skin: possible contributor to drug-induced skin disorders.,' Sci. Rep., vol. 4, p. 5251, 2014.
 S. W. Yee, J. E. Shima, S. Hesselson, L. Nguyen, S. De Val, R. J. Lafond, M. Kawamoto, S. J. Johns, D. Stryke, P.-Y. Kwok, T. E. Ferrin, B. L. Black, D. Gurwitz, N. Ahituv, and K. M. Giacomini, 'Identification and characterization of proximal promoter polymorphisms in the human concentrative nucleoside transporter 2 (SLC28A2).,' J. Pharmacol. Exp. Ther., vol. 328, no. 3, pp. 699'707, 2009.
 M. Fakruddin, A. Chowdhury, M. N. Hossain, K. S. Bin Mannan, and R. M. Mazumdar, 'Pyrosequencing-Principles and Applications,' Int. J. Life Sci. Pharma Res., vol. 2, no. 2, pp. L65'76, 2012.
 KAPABiosystems, 'Introduction to High Resolution Melt Analysis,' p. 17, 2013.
 'Genotype Association Tests ' SNP & Variation Suite v8.' .
 G. M. Clarke, C. a Anderson, F. H. Pettersson, L. R. Cardon, and P. Andrew, 'Basic statistical analysis in genetic case-control studies,' Nat. Protoc., vol. 6, no. 2, pp. 121'133, 2011.
...(download the rest of the essay above)