Patients with early rheumatoid arthritis who smoke are less likely to respond to treatment with methotrexate and tumor necrosis factor inhibitors: observations from the epidemiological investigation of rheumatoid arthritis and the swedish rheumatology register cohorts

2011, American College of Rheumatology Patients With Early Rheumatoid Arthritis Who Smoke Are Less Likely to Respond to Treatment With Methotrexate and Observations From the Epidemiological Investigation of Rheumatoid Arthritis and the Swedish Rheumatology Register Cohorts ´n,1 Maria Seddighzadeh,1 Camilla Bengtsson,2 Annmarie Wesley,2 Staffan Lindblad,1 Johan Askling,1 Lars Alfredsson,2 and Lars Klareskog1 Objective. To determine whether cigarette smok-
European League Against Rheumatism criteria at the
ing influences the response to treatment in patients with
3-month visit. The influence of cigarette smoking (cur-
early rheumatoid arthritis (RA).
rent or past) on the response to therapy was evaluated
Methods. We retrieved clinical information about
by logistic regression, with never smokers as the refer-
patients entering the Epidemiological Investigation of
ent group.
Rheumatoid Arthritis (EIRA) early RA cohort from
Results. Compared with never smokers, current
1996 to 2006 (n ؍ 1,998) who were also in the Swedish
smokers were less likely to achieve a good response at 3
Rheumatology Register (until 2007). Overall, 1,430 of
months following the start of MTX (27% versus 36%;
the 1,621 registered patients were followed up from the
P ؍ 0.05) and at 3 months following the start of TNF
time of inclusion in the EIRA cohort. Of these, 873
inhibitors (29% versus 43%; P ؍ 0.03). In multivariate
started methotrexate (MTX) monotherapy at inclusion,
analyses in which clinical, serologic, and genetic factors
and 535 later started treatment with a tumor necrosis
were considered, the inverse associations between cur-
factor (TNF) inhibitor as the first biologic agent. The
rent smoking and good response remained (adjusted
primary outcome was a good response according to the
odds ratio [OR] for MTX response 0.60 [95% CI
0.39–0.94]; adjusted OR for TNF inhibitor response
0.52 [95% CI 0.29–0.96]). The lower likelihood of a good
The Epidemiological Investigation of Rheumatoid Arthritis study was supported by grants from the Swedish Medical Research response remained at later followup visits. Evaluating
Council, the Stockholm County Council, the Flight Attendant Medical remission or joint counts yielded similar findings. Past
Research Institute, the Swedish Council for Working Life and Social smoking did not affect the chance of response to MTX or
Research, King Gustaf V’s 80-Year Foundation, the Swedish Rheu-matism Association, the Swedish Combine project, and the European TNF inhibitors. Evaluating the overall cohort, which
Community Sixth Framework Programme (project AutoCure).
reflects all treatments used, current smoking was simi-
1Saedis Saevarsdottir, MD, PhD, Sara Wedre´n, MD, PhD, larly associated with a lower chance of a good response
Maria Seddighzadeh, PhD, Staffan Lindblad, MD, PhD, JohanAskling, MD, PhD, Lars Klareskog, MD, PhD: Karolinska Institutet (adjusted ORs for the 3-month, 6-month, 1-year, and
and Karolinska University Hospital, Stockholm, Sweden; 2Camilla 5-year visits 0.61, 0.65, 0.78, 0.66, and 0.61, respectively).
Bengtsson, PhD, Annmarie Wesley, MPH, Lars Alfredsson, PhD:Karolinska Institutet, Stockholm, Sweden.
Conclusion. Among patients with early RA, cur-
Dr. Askling has received consulting fees, speaking fees, and/or rent cigarette smokers are less likely to respond to MTX
honoraria from Wyeth and Bristol-Myers Squibb (less than $10,000 and TNF inhibitors.
Address correspondence to Saedis Saevarsdottir, MD, PhD, Rheumatology Unit, Department of Medicine, Karolinska University In patients with rheumatoid arthritis (RA), early Hospital, 17176 Stockholm, Sweden. E-mail: saedis.saevarsdottir@ and efficient reduction of inflammatory activity is im- portant for improving long-term outcome (1–3). The Submitted for publication April 16, 2010; accepted in revised panel of effective, yet often expensive, drugs with which SMOKING AND RESPONSE OF EARLY RA TO MTX AND TNF INHIBITORS to achieve this goal is growing. Baseline predictors of PATIENTS AND METHODS
response to individual treatments would facilitate a Source population of RA patients. We evaluated RA
rational choice of therapeutic strategy. In this regard, patients from the Epidemiological Investigation of Rheuma- modifiable predictors, such as cigarette smoking habits, toid Arthritis (EIRA) study. EIRA is a population-based are of particular interest. Since methotrexate (MTX) case–control study covering the middle and southern parts of and tumor necrosis factor (TNF) inhibitors are widely Sweden, including as cases RA patients ages 18–70 years whoare within 1 year of diagnosis and, on average, within 10 used as first-and second-line treatments for recent-onset months of symptom onset. All cases of RA were diagnosed by RA, efforts to find predictors of response to these drugs a rheumatologist, fulfilled the American College of Rheuma- tology (ACR) 1987 criteria for RA (18), and were predomi- Cigarette smoking has, in most studies following nantly of Caucasian ancestry (97% of the participants). Weincluded in the present study all patients with RA who patients with early RA or inflammatory polyarthritis, participated in the EIRA study during its first decade, from been associated with more use of disease-modifying 1996 to 2006. Of 2,097 patients included from the participating antirheumatic drugs (DMARDs) and increased occur- clinics, 1,998 (95%) answered a questionnaire that included rence of extraarticular manifestations, including nod- This study was approved by the Ethical Review Board ules, while findings with regard to disease activity have at the Karolinska Institute. All participants gave informed not been conclusive (5–11). Smoking has also been associated with radiologic progression in advanced RA Capture of clinical data for EIRA patients from the
(8,12), while findings in early disease are less consistent SRR. The SRR is a web-based national surveillance system
that was started in the mid-1990s and is used optionally by
(5,9–11,13). However, only limited data are currently rheumatologists to follow incident RA cases longitudinally as a available about the influence of cigarette smoking on the part of standard care. Information about disease activity, response to individual treatments in early RA. For MTX disability, and treatment are registered at each visit, which monotherapy, a report of a randomized controlled clin- occurs at predefined time points, although the clinical practicesetting allows some flexibility around the intended dates. Also ical trial of 205 patients indicated a not significantly hosted within the SRR is the Anti-Rheumatic Therapy in reduced response among smokers (14), and no informa- Sweden (ARTIS) registry, in which patients who receive tion exists as to whether previous smoking influences treatment with biologic agents, including TNF inhibitors, are response. With respect to TNF inhibitors, current smok- followed. ARTIS covers more than 90% of patients who haveever taken a biologic agent (ref. 19 and Neovius M: unpub- ing (15,16) and the number of pack-years smoked (17) have been associated with a reduced response in patients By virtue of their early RA, patients included in EIRA with established RA. Further, the extent to which an are also invited to participate in the SRR. For the purpose of apparent effect of smoking on the risk of developing RA the present study, we linked these two data sources to captureclinical information about disease course and therapy. A may be influenced by genetic factors known to interact unique personal identification number permitted deterministic linkage between the data sources. Of the 1,998 RA patients in In the present study, we therefore linked baseline the EIRA cohort, 1,621 had been registered in the SRR; 1,430 information on cigarette smoking (current or past and of these patients started SRR followup at the time of EIRAinclusion (Figure 1) and were thus eligible for the current cumulative dose) from a large population-based cohort of patients with newly diagnosed RA to followup data on DMARD treatment at baseline. Of the 1,430 patients disease activity and treatment from the Swedish Rheu- whose cases were followed from the time of EIRA inclusion, matology Register (SRR) and investigated the influ- 873 (61%) received methotrexate monotherapy as their firstDMARD, 382 (27%) received treatment with other DMARDs ence of smoking on treatment response in the overall or combinations, and 175 (12%) received clinical care without group, as well as specifically to the subgroup of patients DMARD treatment at baseline. An overview of the DMARD who initially started MTX monotherapy and the sub- treatments or combinations initiated at baseline is shown in group who later started TNF inhibitors as the first Supplementary Table 1 (available on the Arthritis & Rheuma- tism web site at http://onlinelibrary.wiley.com/journal/10.1002/ biologic agent. Demographic and disease character- (ISSN)1529-0131). In the present study, we focused on MTX istics, serologic factors (rheumatoid factor [RF] and as the initial DMARD, since it was the most commonly used anti–cyclic citrullinated peptide [anti-CCP] antibody first-line treatment. Other treatment options were included in status), and genetic susceptibility factors (HLA– Therapy with a biologic agent. Treatment with a TNF DRB1 shared epitope and the PTPN22*620W risk al- inhibitor (infliximab, etanercept, or adalimumab) was initiated lele) were evaluated as potential confounders or effect as the first biologic agent in 535 patients at a median of 3 years after the diagnosis of RA (Figure 1). Fewer than 30 patients Figure 1. Flow chart showing the distribution of rheumatoid arthritis (RA) patients who entered the Epidemiological Investigation
of Rheumatoid Arthritis (EIRA) study during 1996–2006 and were also in the Swedish Rheumatology Register (SRR). The numbers
of RA patients who had baseline data in the SRR at EIRA inclusion, had Disease Activity Score in 28 joints (DAS28) data available
at the followup time points, and had been categorized as current, past, or never smokers are shown for the overall cohort, for patients
receiving methotrexate (MTX) at baseline, and for patients receiving tumor necrosis factor (TNF) inhibitors during the followup until
2007. * In accordance with the eligibility criteria for the SRR and the EIRA study, the baseline visit in the SRR could occur up to 12
months before, and up to 2.5 months after, inclusion in the EIRA study. ** Detailed treatment data for all patients who received MTX
monotherapy at baseline is shown in Supplementary Table 1 (available on the Arthritis & Rheumatism web site at http://
onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). *** Predefined times for followup visits were at 3 months (range 2.5–5
months), 6 months (range 5–7.5 months), 1 year (range 7.5 months to 1.5 years), 2 years (range 1.5–2.5 years), and 5 years (range
4.5–5.5 years). # Patients who continued the same treatment until the followup visit. Reasons for treatment changes were not
documented, but most patients changing treatment had no response according to the European League Against Rheumatism criteria.
had started any of the other available biologic agents (ritux- Clinical variables were captured from the SRR. RF status was imab, abatacept, anakinra, or unspecified biologic agent in determined using standard procedures, and anti-CCP antibod- clinical trials) before the end of the followup period in 2007, ies were determined by the standard enzyme-linked immu- usually after treatment with a TNF inhibitor.
nosorbent assay (Immunoscan RA Mark 2 ELISA; Euro- Definition of treatment response. In accordance with
Diagnostica). The methods for determining the HLA–DRB1 recent guidelines from the European League Against Rheu- shared epitope alleles and the PTPN22*620W (1858C/T) poly- matism (EULAR) and the American College of Rheumatol- morphism have been previously reported (22–25). At baseline, ogy (ACR) (20), we used the EULAR response criteria (21), the following data were not available: DAS28 for 0.8%, Heath which are based on the Disease Activity Score 28-joint assess- Assessment Questionnaire (HAQ) score for 3.4%, RF status ment (DAS28), to define response to treatment. The primary for 0.5%, anti-CCP antibody status for 4.3%, shared epitope outcome of interest was a good response according to the allele status for 2.0%, and PTPN22*R620W status for 3.7%.
EULAR response criteria (DAS28 Ͻ3.2 at the followup visit Statistical analysis. We used 2 approaches to analyze
and Ͼ1.2 units decrease compared with the baseline DAS28), the association between current smoking and a good response and the secondary outcome was remission according to the or remission according to the EULAR criteria. First, a good EULAR response criteria (DAS28 Ͻ2.6 units at followup). We response (DAS28 Ͻ3.2 at followup and Ͼ1.2 units decrease evaluated these outcomes at the following predefined time from baseline) as compared with no response (DAS28 Ͼ5.1 or points: the 3-month, 6-month, 1-year, 2-year, and 5-year visits.
Ͻ0.6 units decrease) or a moderate response (falls between The numbers of patients with available DAS28 and smoking good response and no response) was evaluated using univari- ate and multivariate logistic regression analyses. The results Definition of baseline variables. Smoking status. We
were expressed as univariate P values and multivariate odds retrieved information on smoking status from the EIRA ratios (ORs) with 95% confidence intervals (95% CIs). Sepa- questionnaire. Patients were classified according to their smok- rate models were constructed for each time point for the ing status at the time of RA diagnosis as never, past, current, overall EIRA cohort at the 3-month, 6-month, 1-year, 2-year, or irregular cigarette smokers, as well as other than cigarette and 5-year followup visits. For the groups treated with MTX or smokers (Table 1). The number of pack-years smoked (1 with TNF inhibitors, we restricted our analyses to the 3-month pack-year ϭ 20 cigarettes/day for 1 year) was used to measure and 6-month visits and considered only those who continued to the cumulative dose of smoking. Information about cigarette take the same treatment, since the reason for treatment smoking habits was not available for 4.6% of the patients, and changes was not registered (for example, lack of response or 8.2% lacked information about the number of pack-years.
side effects). The numbers of patients continuing each treat- Other variables. Baseline parameters that we evaluated ment are shown in Figure 1. Their smoking habits did not as potential confounders or effect modifiers are listed in Table differ significantly from the smoking habits of those who 1 and are described under the statistical analysis section below.
SMOKING AND RESPONSE OF EARLY RA TO MTX AND TNF INHIBITORS Patient characteristics at treatment baseline* Irregular smoking/other tobacco, no. (%)‡ * The study population consisted of patients from the Epidemiological Investigation of Rheumatoid Arthritis cohort who had baseline data in theSwedish Rheumatology Register (SRR). Subgroups represent patients who started monotherapy with methotrexate (MTX) or who received nodisease-modifying antirheumatic drug (DMARD) at baseline, as well as those who started a tumor necrosis factor (TNF) inhibitor as the first biologicagent during the followup period. DAS28 ϭ Disease Activity Score in 28 joints; IQR ϭ interquartile range; anti-CCP ϭ anti-cyclic citrullinatedpeptide; HAQ ϭ Health Assessment Questionnaire; NSAIDs ϭ nonsteroidal antiinflammatory drugs.
† Not all patients had available information about smoking status, disease characteristics, or genetic factors. Missing values for each parameter aresummarized in Patients and Methods.
‡ Patients in this category were excluded from further analyses of cigarette smoking habits.
The following baseline parameters were evaluated in ification, each parameter was also evaluated as a potential the multivariate regression model: age at baseline (in 10-year effect modifier. All stratum-specific ORs were similar. Thus, increments), sex, current cigarette smoking, past cigarette the final nonstratified models included only the a priori smoking, RF status (positive/negative), anti-CCP antibody status (positive/negative), baseline DAS28 (per unit increase), The distribution of DAS28 units on a continuous scale and baseline HAQ score (per unit increase), carriage of the at the 3-month visit was compared between current smokers HLA–DRB1 shared epitope (1 or 2 copies versus no copies), and never smokers and between past smokers and never and carriage of the PTPN22*R620W risk allele (1 or 2 copies smokers by use of t-tests, as well as for each of the components versus no copies), as well as disease duration for those who of the DAS28 (swollen joint count, tender joint count, eryth- received TNF inhibitor treatment. Age, sex, past smoking, and rocyte sedimentation rate (ESR), patient’s assessment of baseline DAS28 were included as potential confounders or a global health status using a 100-mm visual analog scale [VAS]) priori covariates in all models. In addition, we included by use of Wilcoxon’s rank sum test, since these parameters concurrent use of prednisolone or nonsteroidal antiinflamma- tory drugs (NSAIDs) in the models investigating the specific Statistical analysis was performed using the SAS ver- treatments. For the group starting TNF inhibitors during the sion 9.1 software (SAS Institute). All tests were 2-sided, and followup period, we also included disease duration at the start the significance level was set at 0.05.
of TNF inhibitor treatment and concurrent use of MTX orother DMARDs. Furthermore, we considered the remainingbaseline parameters as potential confounders by assessing whether their addition (one at a time) to the model containingsmoking and the a priori covariates altered the OR associated Eligibility of patients for analyses. Patients who
with smoking by more than 10%. None of these factors had available DAS28 data at baseline and were classifi- materially altered the association with smoking. Through strat- able as current, past, or never smokers constitute the Figure 2. European League Against Rheumatism (EULAR) response in the overall EIRA cohort
and in the subgroups receiving MTX or clinical care without a disease-modifying antirheumatic
drug (DMARD) at baseline, as well as in those receiving TNF inhibitors later on, according to
smoking status. Values are the proportion of patients achieving a good response according to the
EULAR criteria at the 3-month visit divided by the number in the smoking category who had
DAS28 data available at followup. Numbers across the bottom are the total number with DAS28
and smoking status data. P values were determined by univariate logistic regression. See Figure 1
for other definitions.
basis for all analyses (Figure 1). Baseline characteristics ates at the 3-month followup visit (adjusted OR 0.61 of the study patients are shown in Table 1. Baseline [95% CI 0.44–0.87]) (Figure 3, left) and was similar at characteristics did not differ between RA patients with later followup visits as compared with baseline (adjusted and those without available DAS28 data.
ORs for the 6-month, 1-year, 2-year, and 5-year visits Cigarette smoking and the chance of response.
0.65, 0.78, 0.66, and 0.61, respectively).
There was no indication that smokers were more likely The influence of smoking did not differ signifi- to be included in the SRR (data available upon request cantly between anti-CCP–positive and anti-CCP– Findings in the entire EIRA cohort irrespective
Current smokers were also less likely to achieve of treatment. Overall, 1,430 EIRA patients were fol-
remission at the 3-month followup visit (adjusted OR lowed in SRR from baseline; 1,418 of them had DAS28 0.60 [95% CI 0.40–0.88]) (Figure 3, right), and similar data available at baseline (Figure 1), and 1,199 had results were observed at all later followup visits (ad- DAS28 data available at the 3-month visit (994 with justed ORs for the 6-month, 1-year, 2-year, and 5-year smoking status). At the 3-month visit, 32% were good responders according to the EULAR criteria and 24% The cumulative dose of smoking did not further were in remission. At the 6-month visit, 1,124 had add to the results among current smokers. Thus, no DAS28 data available (952 with smoking status), and association was observed between the number of pack- 42% of them had achieved a good response and 34% years smoked and a good response according to the EULAR criteria at the 3-month followup visit (adjusted As shown in Figure 2, 26% of current smokers OR per pack-year increase 1.00 [95% CI 0.98–1.02]).
(80 of 305) in the whole cohort achieved a good response Findings in the patients starting MTX mono-
according to the EULAR criteria after 3 months, com- therapy at baseline. A total of 873 patients started
pared with 35% of never smokers (123 of 349; P ϭ 0.01).
methotrexate (MTX) as the only DMARD treatment at Past smokers had a similar chance of having a good baseline. Of these, 865 had an available DAS28 measure response (32% [110 of 340]) as never smokers (P ϭ 0.42).
at baseline (Figure 1), and 761 (626 with smoking status) This decreased chance of a good response in of those who had not changed their treatment before current smokers remained after adjustment for covari- followup had DAS28 data available at the 3-month SMOKING AND RESPONSE OF EARLY RA TO MTX AND TNF INHIBITORS Figure 3. Association between current smoking and a good response (left) or remission (right)
according to the European League Against Rheumatism criteria in the overall group of EIRA
patients. Values are the adjusted odds ratios with 95% confidence intervals (95% CIs) determined
at the 3-month, 6-month, 1-year, 2-year, and 5-year followup visits, as calculated by multivariate
logistic regression, adjusting for age, sex, past smoking, and baseline DAS28. Construction of the
models and definition of the covariables are described in Patients and Methods. See Figure 1 for
other definitions and numbers of patients in each group.
followup visit. We found that 33% were good responders current smokers as compared with never smokers re- according to the EULAR criteria, and 24% were in mained in the multivariate model and was not influ- remission. At the 6-month followup visit, 522 patients enced by any of the covariates (adjusted OR 0.60 [95% (436 with smoking data) had not changed their treat- CI 0.39–0.94]) (Figure 4, left). Similar results for a good ment and had available DAS28 data, and of that group, response at the 6-month visit were observed (adjusted 48% had achieved good response and 39% were in OR 0.58 [95% CI 0.36–0.94]). Current smokers also tended to be less likely to be in remission after 3 months After 3 months of MTX monotherapy, 27% of (OR 0.66 [95% CI 0.40–1.08]) (Figure 4, right), and at current smokers (54 of 197) had reached a EULAR the 6-month followup visit, this was significant (OR 0.41 good response as compared with 36% of never smokers (78 of 214; P ϭ 0.05) (Figure 2), whereas the frequency Among current smokers, the number of pack- of good response did not differ between past smokers years smoked was not associated with the chance of good (37% [79 of 215]) and never smokers (P ϭ 0.95).
response after 3 months (adjusted OR per pack-year The decreased chance of a good response in increase 0.99 [95% CI 0.98–1.02]).
Figure 4. Association between current smoking and a good response (left) or remission (right)
according to the European League Against Rheumatism criteria in patients starting MTX at
baseline and in patients starting a TNF inhibitor as the first biologic agent. Values are the adjusted
odds ratios with 95% confidence intervals (95% CIs) determined at the 3-month and the 6-month
visits, as calculated by multivariate logistic regression, adjusting for age, sex, past smoking, DAS28
at diagnosis, and concurrent treatment with prednisolone, nonsteroidal antiinflammatory drugs,
and in the TNF inhibitor–treated group, MTX. Construction of the models and definition of the
covariables are described in Patients and Methods. See Figure 1 for other definitions and numbers
of patients in each group.
The influence of smoking did not differ signifi- cantly between anti-CCP–positive and anti-CCP–negative patients or between patients with and withoutconcurrent prednisolone treatment (data not shown).
None of the DAS28 components differed be- tween current, past, and never smokers at baseline (datanot shown). However, consistent with the categorizedresults above for a good response and remission accord-ing to the EULAR criteria, current smokers had signif-icantly higher scores on the DAS28 as a continuousvariable than did never smokers (mean 4.10 versus 3.62;P ϭ 0.001) at the 3-month followup visit, whereas pastsmokers and never smokers had similar scores on theDAS28 (mean 3.65 versus 3.62; P ϭ 0.83). This alsoapplied to the individual components of the DAS28(Figure 5), where current smokers as compared withnever smokers had higher numbers of swollen joints(P ϭ 0.02) and tender joints (P ϭ 0.03), a higher ESR(P ϭ 0.008), and tended to have higher VAS score forthe patient’s assessment of global health (P ϭ 0.09) at3 months. These parameters were not significantly dif-ferent between past smokers and never smokers(swollen joint count P ϭ 0.47, tender joint count P ϭ0.64, erythrocyte sedimentation rate P ϭ 0.85, andVAS score for the patient’s assessment of global healthP ϭ 0.21).
Findings in patients who did not start DMARD
therapy at baseline. Eighteen percent of the patients
who did not start any DMARD treatment at baseline
(n ϭ 175) received oral prednisolone treatment, and
46% took NSAIDs regularly. At baseline, 158 of the
patients had DAS28 data available, and at the 3-month
followup visit, 129 had DAS28 data available (109 with
smoking status). At the 3-month visit, 22% of these
patients were classifiable as EULAR good responders
and 21% were in remission. At the 6-month followup
visit, 52 of those with available DAS28 data were still not
receiving DMARD treatment, and of those, 27% were
classifiable as EULAR good responders and 29% were
in remission.
After 3 months, 14% of current smokers (4 of 28) achieved a good response according to the EULARcriteria as compared with 34% of never smokers (15 of44; P ϭ 0.07), and past smokers had a significantly lowerchance of achieving a good response (14% [5 of 37])than did never smokers (P ϭ 0.04). The influence ofsmoking did not differ significantly between patients Figure 5. Influence of smoking habits on individual components of
who did and those who did not receive prednisolone in the DAS28 in patients receiving MTX or a TNF inhibitor, as deter- this non–DMARD-treated group, although those who mined at the 3-month followup visit. Data are shown as box plots. Eachbox represents the upper and lower interquartile range. Lines inside received prednisolone were more likely to reach a good the boxes represent the median. Whiskers represent the fifth percen- tiles. P values are versus never smokers, as determined by Wilcoxon’srank sum test (significant at P Ͻ 0.05). VAS-global ϭ visual analogscale (0–100 mm) for the patient’s assessment of global health.
SMOKING AND RESPONSE OF EARLY RA TO MTX AND TNF INHIBITORS Findings in the patients who later started a TNF
differ significantly between past smokers and never inhibitor as the first biologic agent. By the end of the
followup period in 2007, a total of 535 patients whoentered the EIRA cohort during 1996–2006 had started DISCUSSION
a TNF inhibitor, and 486 of these patients had DAS28data available at the start of treatment (Figure 1). The In this population-based early RA cohort receiv- median time from RA diagnosis to the start of a TNF ing real-life care, we found that current smokers were inhibitor was 3 years (interquartile range 1–5 years).
less likely to respond to MTX treatment and to TNF Of the patients who had smoking status available, 199 inhibitor treatment. The decreased chance of response received infliximab, 136 received etanercept, and 66 associated with current smoking in both treatment received adalimumab. When pooled, 301 had DAS28 groups was observed for the primary end point of a good data and smoking category available at the 3-month response according to the EULAR criteria, for remis- followup visit. Of these 301 patients, 38% were good sion according to the EULAR criteria, and for the responders and 28% were in remission. At the 6-month DAS28 scores or the joint counts at followup and was followup visit, 324 had DAS28 data available, and not explained by other factors. The influence of current 44% had achieved a good response and 36% were in smoking was similar at later followup visits and was also observed for the overall group, without regard to which Current smokers were less likely to respond well treatment was used. On the other hand, past smoking to TNF inhibitors after 3 months of therapy (29% [28 of did not influence the chance of treatment response,although it seemed to have a negative effect in the group 98]) than never smokers were (43% [49 of 113]; P ϭ that did not start DMARD treatment at baseline.
0.03), whereas past smokers (39% [35 of 90]) had a Our findings provide new information concerning similar frequency of good response as never smokers the influence of smoking on the response to MTX (P ϭ 0.52). The lower likelihood of a good response in monotherapy and extend the information concerning current smokers remained in the multivariate model response to TNF inhibitors in patients with longstanding after adjustment for covariates (adjusted OR 0.52 [95% RA (15–17). Previous findings of smoking as a predictor CI 0.29–0.96]). Findings were similar at the 6-month of later disease activity in early RA or inflammatory followup visit (adjusted OR 0.55 [95% CI 0.31–0.96]).
polyarthritis have been somewhat inconsistent, maybe Using remission as the outcome measure showed a due to the different treatments used, the outcome tendency toward an association of current smoking with measures, and the followup time points, which makes remission after 3 months (adjusted OR 0.61 [95% CI the study results difficult to compare (5,8–11,26). Thus, 0.30–1.24]), which was significant at the 6-month visit we evaluated both the likelihood of a good response and (adjusted OR 0.54 [95% CI 0.30–0.99]).
remission, which are recommended as outcome mea- Among current smokers, the number of pack- sures in the ACR and EULAR joint guidelines (20,21), years smoked was not associated with the chance of good and found similar results for current smoking up to 5 response after 3 months (adjusted OR per pack-year years later, results which were independent of other increase for current smokers 1.00 [95% CI 0.96–1.04]).
The influence of smoking did not differ signifi- The strength of our study is that it is population- cantly between anti-CCP–positive and anti-CCP– based, using information about real-life care of patients negative patients or between patients who were and with early RA, with almost complete information about those who were not receiving concurrent prednisolone smoking habits. Compared with clinical trials, the find- ings may have a higher external validity for routine care, While none of the DAS28 components differed since the only selection criteria applied were being between the groups of current, past, and never smokers between the ages of 18 and 70 years and having RA of at baseline (data not shown), at the 3-month visit, recent onset. The influence of smoking was studied in current smokers had higher DAS28 scores than did the context of the main treatment options of today, and never smokers (P ϭ 0.003). As shown in Figure 5, this most RA patients in whom these treatments were also applied to the individual DAS28 components of started should have been captured because of the swollen joint count (P ϭ 0.002) and VAS score for the population-based setting and high coverage of the EIRA patient’s assessment of global health (P ϭ 0.01), whereas and SRR registers. Other DMARDs used as part of a trend was observed for the tender joint count (P ϭ standard care during the study period (see Supplemen- 0.10) and the ESR (P ϭ 0.09). These parameters did not tary Table 1, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/ years, many were lost to followup, partly because they 10.1002/(ISSN)1529-0131) were found in the whole co- had not yet reached that time point and partly because hort, but the numbers of patients receiving each of these during the first years of the SRR and the EIRA study, treatments were too low for a meaningful analysis in clinicians were instructed to follow their RA cases for only 2 years. Further, some patients were excluded Another strength of the present study is that the because they did not start their followup in the SRR at information about smoking was detailed enough to the time of EIRA inclusion. It should be noted, however, permit analysis of both current and past smoking as well that the findings were similar if all RA patients with as of the number of pack-years. Furthermore, we had followup data were kept in the analyses.
data on genetic risk factors (HLA–DRB1 shared epitope This study did not intend to explore the reasons and the PTPN22*620W risk allele) as well as serology why smoking may have a negative effect on response to (RF and anti-CCP antibodies) for analysis of eventual treatment, but the following factors may be considered.
modifying effects of these factors, which have been First, smoking may possibly be associated with refracto- shown to interact with smoking in terms of the risk of riness to medications because of pharmacokinetic (PK) developing the seropositive form of RA and, in some but or pharmacodynamic (PD) interactions. Accordingly, a not all studies, have been reported to be associated with recent study in RA patients showed that smokers had response to MTX (27–30) or to TNF inhibitors (31,32).
lower concentrations of MTX polyglutamates, the active The association estimates for current smoking, however, metabolites retained in cells, which indicates that smok- turned out to be unaffected by all covariates tested.
ers metabolize MTX differently (33). We did not ob- The observational design of this study gives also serve any influence of initial MTX dosage on the rise to its main limitations. First, some patients stopped response frequencies (data not shown). For TNF inhib- the treatment, received alternative drugs, or received itors, we are not aware of any studies on the influence of additional drugs before the 3-month or 6-month fol- lowup visit when treatment response was evaluated. The Second, it may also be that smokers have a more reason for changes in medications before the followup persistent “natural” disease course of RA, irrespective visit (lack of response, side effects, etc.) was not re- of treatment. Since only 175 patients did not start any corded, and those patients were therefore excluded from DMARD treatment, only 52 of whom were not receiving the analyses of individual treatment groups for that time any DMARD after 6 months from inclusion, our power point (see Figure 1). Thus, the numbers of patients with to detect differences in disease course that were depen- available data for an evaluation of treatment response dent on smoking was limited. Thus, the observational are likely to be selected on response since they contin- setting of our study makes it impossible to determine ued to take the same treatment, but a more conservative whether it is the underlying disease that is made more approach of keeping in the analyses all patients who severe by smoking or whether smoking specifically af- initiated each treatment yielded similar findings (data fects response to therapy. From a clinical perspective, it is important to know the impact of smoking on disease Treatment changes also made it difficult to eval- activity in individuals who are treated with MTX or TNF uate the influence of smoking on the response to inhibitors, irrespective of whether the effect of smoking particular drugs after the 6-month followup. We did, is mainly acting on the underlying disease course (which however, evaluate the whole cohort at later followup is then not compensated for by the treatment) or visits to give an overall picture of the more long-term whether the effect of smoking is specific for the actual influence of smoking, which yielded similar findings of a detrimental effect of smoking. For the TNF inhibitor– Cigarette smoking is a well-known risk factor for treated group, smoking information was retrieved from the RA diagnosis, particularly in the subset of patients the time of study entry. This may have introduced bias, with anti-CCP antibodies (1), but the anti-CCP status but our findings are consistent with those of previous had no modifying effect on the association of current smoking with response. Thus, the mechanisms through Second, some patients were lost to followup in which smoking influences susceptibility to RA and its the SRR registry. This might affect the external validity disease course, respectively, may differ.
of the results, especially if the smoking habits of those In conclusion, our findings indicate that cigarette who were lost to followup differed from those who were smokers have a diminished chance of responding well to not, but this was not the case (data not shown). After 2 the currently first- and second-line agents of choice in SMOKING AND RESPONSE OF EARLY RA TO MTX AND TNF INHIBITORS early RA treatment today: MTX and TNF inhibitors, REFERENCES
respectively. This was also observed in the whole group 1. Klareskog L, Catrina AI, Paget S. Rheumatoid arthritis. Lancet of RA patients during the first years, indicating that smokers have a more persistent disease activity irrespec- 2. De Vries-Bouwstra JK, Goekoop-Ruiterman YP, Verpoort KN, tive of which treatment is used. These findings support Schreuder GM, Ewals JA, Terwiel JP, et al. Progression of jointdamage the idea that smoking should be taken into account when HLA–DRB1, rheumatoid factor, and anti–citrullinated protein trying to predict response to antirheumatic agents.
antibodies in relation to different treatment strategies. Arthritis Whether discontinuation of smoking prior to initiating 3. Anderson JJ, Wells G, Verhoeven AC, Felson DT. Factors treatment is beneficial, as suggested by the observation predicting response to treatment in rheumatoid arthritis: the that past smokers responded equally well to treatment as importance of disease duration. Arthritis Rheum 2000;43:22–9.
those who had never smoked, remains to be studied 4. Klarenbeek NB, Allaart CF, Kerstens PJ, Huizinga TW, Dijkmans BA. The BeSt story: on strategy trials in rheumatoid arthritis. Curr further. Meanwhile, the findings provide a strong impe- tus for clinicians to include measures against smoking as 5. Westhoff G, Rau R, Zink A. Rheumatoid arthritis patients who a fundamental part of the therapeutic armamentarium smoke have a higher need for DMARDs and feel worse, but theydo not have more joint damage than non-smokers of the same serological group. Rheumatology (Oxford) 2008;47:849–54.
6. Turesson C, O’Fallon WM, Crowson CS, Gabriel SE, Matteson EL. Extra-articular disease manifestations in rheumatoid arthritis: ACKNOWLEDGMENTS
incidence trends and risk factors over 46 years. Ann Rheum Dis2003;62:722–7.
We thank all of the participating control subjects and 7. Nyhall-Wahlin BM, Petersson IF, Nilsson JA, Jacobsson LT, RA patients, as well as the following people who recruited Turesson C, and the BARFOT study group. High disease activity patients for EIRA, along with their collaborating clinicians and disability burden and smoking predict severe extra-articular man- nurses: Ingeli Andreasson (Landvetter); Eva Baecklund (Aka- ifestations in early rheumatoid arthritis. Rheumatology (Oxford) demiska Hospital); Ann Bengtsson and Thomas Skogh (Linko- ping Hospital); Johan Bratt, Ingiald Hafstrom, and Birgitta 8. Wolfe F. The effect of smoking on clinical, laboratory, and Nordmark (Karolinska University Hospital); Kjell Huddenius radiographic status in rheumatoid arthritis. J Rheumatol 2000;27: (Rheumatology Clinic, Stockholm); Shirani Jayawardene 9. Papadopoulos NG, Alamanos Y, Voulgari PV, Epagelis EK, (Bollnas Hospital); Ann Knight (Hudiksvall Hospital and Tsifetaki N, Drosos AA. Does cigarette smoking influence disease Uppsala University Hospital); Ido Leden (Kristianstad Hospi- expression, activity and severity in early rheumatoid arthritis tal); Goran Lindahl (Danderyd Hospital); Bengt Lindell (Kal- patients? Clin Exp Rheumatol 2005;23:861–6.
mar Hospital); Christine Lindstrom and Gun Sandahl (Sophia- 10. Manfredsdottir VF, Vikingsdottir T, Jonsson T, Geirsson AJ, hemmet); Bjorn Lofstrom (Katrineholm Hospital); Ingemar Kjartansson O, Heimisdottir M, et al. The effects of tobacco Petersson (Spenshult Hospital); Christoffer Schaufelberger smoking and rheumatoid factor seropositivity on disease activity (Sahlgrenska University Hospital); Patrik Stolt (Vasteras Hos- and joint damage in early rheumatoid arthritis. Rheumatology pital); Berit Sverdrup (Eskilstuna Hospital); Olle Svernell (Vastervik Hospital); and Tomas Weitoft (Gavle Hospital).
11. Harrison BJ, Silman AJ, Wiles NJ, Scott DG, Symmons DP. The association of cigarette smoking with disease outcome in patients We thank Marie-Louise Serra and Lena Nise for excellent with early inflammatory polyarthritis. Arthritis Rheum 2001;44: collection of the data as part of EIRA, Dr. Leonid Padyukov for providing the HLA–DRB1 data, Prof. Ronald van Vollen- 12. Saag KG, Cerhan JR, Kolluri S, Ohashi K, Hunninghake GW, hoven for critical reading of the manuscript, and Drs. Jing Cui Schwartz DA. Cigarette smoking and rheumatoid arthritis sever- ¨m for methodologic discussions. We are also indebted to the Swedish Rheumatology Register and ARTIS 13. Finckh A, Dehler S, Costenbader KH, Gabay C. Cigarette smok- groups, and last, but not least, we thank all of the clinicians ing and radiographic progression in rheumatoid arthritis. Ann who have followed the cases of the registry patients over the 14. Wessels JA, van der Kooij SM, le Cessie S, Kievit W, Barerra P, Allaart CF, et al, for the Pharmacogenetics Collaborative Re-search Group. A clinical pharmacogenetic model to predict the AUTHOR CONTRIBUTIONS
efficacy of methotrexate monotherapy in recent-onset rheumatoidarthritis. Arthritis Rheum 2007;56:1765–75.
All authors were involved in drafting the article or revising it 15. Hyrich KL, Watson KD, Silman AJ, Symmons DP. Predictors of critically for important intellectual content, and all authors approved response to anti-TNF-␣ therapy among patients with rheumatoid the final version to be published. Dr. Saevarsdottir had full access to all arthritis: results from the British Society for Rheumatology Bio- of the data in the study and takes responsibility for the integrity of the logics Register. Rheumatology (Oxford) 2006;45:1558–65.
data and the accuracy of the data analysis.
16. Abhishek A, Butt S, Gadsby K, Zhang W, Deighton CM. Anti- Study conception and design. Saevarsdottir, Wedre
TNF-␣ agents are less effective for the treatment of rheumatoid arthritis in current smokers. J Clin Rheumatol 2010;16:15–8.
Acquisition of data. Saevarsdottir, Wedre
17. Mattey DL, Brownfield A, Dawes PT. Relationship between pack-year history of smoking and response to tumor necrosis Analysis and interpretation of data. Saevarsdottir, Bengtsson, Wesley,
factor antagonists in patients with rheumatoid arthritis. J Rheu- 18. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, ations with rheumatoid arthritis in Ͼ4,000 samples from North Cooper NS, et al. The American Rheumatism Association 1987 America and Sweden: association of susceptibility with PTPN22, revised criteria for the classification of rheumatoid arthritis.
CTLA4, and PADI4. Am J Hum Genet 2005;77:1044–60.
26. Forslind K, Hafstrom I, Ahlmen M, Svensson B. Sex: a major 19. Askling J, Fored CM, Geborek P, Jacobsson LT, van Vollenhoven predictor of remission in early rheumatoid arthritis? Ann Rheum R, Feltelius N, et al. Swedish registers to examine drug safety and clinical issues in RA. Ann Rheum Dis 2006;65:707–12.
27. Criswell LA, Lum RF, Turner KN, Woehl B, Zhu Y, Wang J, et al.
20. Aletaha D, Landewe R, Karonitsch T, Bathon J, Boers M, The influence of genetic variation in the HLA–DRB1 and Bombardier C, et al. Reporting disease activity in clinical trials of LTA–TNF regions on the response to treatment of early rheuma- patients with rheumatoid arthritis: EULAR/ACR collaborative toid arthritis with methotrexate or etanercept. Arthritis Rheum recommendations. Ann Rheum Dis 2008;67:1360–4.
21. Van Gestel AM, Prevoo ML, van ’t Hof MA, van Rijswijk MH, van 28. Hider SL, Silman AJ, Thomson W, Lunt M, Bunn D, Symmons de Putte LB, van Riel PL. Development and validation of the DP. Can clinical factors at presentation be used to predict European League Against Rheumatism response criteria for rheu- outcome of treatment with methotrexate in patients with early matoid arthritis: comparison with the preliminary American Col- inflammatory polyarthritis? Ann Rheum Dis 2009;68:57–62.
lege of Rheumatology and the World Health Organization/ 29. O’Dell JR, Nepom BS, Haire C, Gersuk VH, Gaur L, Moore GF, International League Against Rheumatism criteria. Arthritis et al. HLA-DRB1 typing in rheumatoid arthritis: predicting re- sponse to specific treatments. Ann Rheum Dis 1998;57:209–13.
22. Ronnelid J, Wick MC, Lampa J, Lindblad S, Nordmark B, 30. Van Dongen H, van Aken J, Lard LR, Visser K, Ronday HK, Klareskog L, et al. Longitudinal analysis of citrullinated protein/ Hulsmans HM, et al. Efficacy of methotrexate treatment in peptide antibodies (anti-CP) during 5 year follow up in earlyrheumatoid arthritis: anti-CP status predicts worse disease activity patients with probable rheumatoid arthritis: a double-blind, ran- and greater radiological progression. Ann Rheum Dis 2005;64: domized, placebo-controlled trial. Arthritis Rheum 2007;56: 23. Padyukov L, Silva C, Stolt P, Alfredsson L, Klareskog L, for the 31. Plenge RM. Recent progress in rheumatoid arthritis genetics: one Epidemiological Investigation of Rheumatoid Arthritis Study step towards improved patient care. Curr Opin Rheumatol 2009; Group. A gene–environment interaction between smoking and shared epitope genes in HLA–DR provides a high risk of seropos- 32. Potter C, Hyrich KL, Tracey A, Lunt M, Plant D, Symmons DP, et itive rheumatoid arthritis. Arthritis Rheum 2004;50:3085–92.
al. Association of rheumatoid factor and anti-cyclic citrullinated 24. Klareskog L, Stolt P, Lundberg K, Kallberg H, Bengtsson C, peptide positivity, but not carriage of shared epitope or PTPN22 Grunewald J, et al, and the Epidemiological Investigation of susceptibility variants, with anti-tumour necrosis factor response in Rheumatoid Arthritis Study Group. A new model for an etiology rheumatoid arthritis. Ann Rheum Dis 2009;68:69–74.
of rheumatoid arthritis: smoking may trigger HLA–DR (shared 33. Stamp LK, O’Donnell JL, Chapman PT, Zhang M, Frampton C, epitope)–restricted immune reactions to autoantigens modified by James J, et al. Determinants of red blood cell methotrexate citrullination. Arthritis Rheum 2006;54:38–46.
polyglutamate concentrations in rheumatoid arthritis patients re- 25. Plenge RM, Padyukov L, Remmers EF, Purcell S, Lee AT, ceiving long-term methotrexate treatment. Arthritis Rheum 2009; Karlson EW, et al. Replication of putative candidate-gene associ-

Source: http://www.eirasweden.se/Patients-With-Early-Rheumatoid-Arthritis-Who-Smoke-Are-Less-Likely-to-Respond-to-Treatment-With-Methotrexate-and-Tumor-Necrosis-Factor-Inhibitors.pdf