Abstract
Objective Current smoking reduces clinical response to several disease-modifying antirheumatic drugs. It is unknown if this is also the case for prednisone. We aimed to determine whether current smoking affects the clinical response to concomitant prednisone in a methotrexate (MTX)-based treatment strategy.
Methods In the CAMERA-II trial (isrctn.com identifier: 70365169), patients with early rheumatoid arthritis (RA) initiated an MTX-based strategy and were randomized to concomitant prednisone (MTX + pred) or placebo (MTX + PBO) for 24 months. Linear mixed modeling was performed with Disease Activity Score assessing 28 joints (DAS28) as the dependent variable, and strategy group and current smoking status as independent variables, correcting for relevant covariates. The interaction between current smoking and strategy was tested to find out whether the effect of current smoking on clinical response was different between the strategy groups with prednisone or PBO.
Results Current smoking was significantly associated with higher DAS28 over time (mean difference with nonsmokers 0.57 [95% CI 0.22–0.92, P < 0.01]). This association was not different between the strategy groups with prednisone or PBO (P = 0.73). The negative effect of current smoking on DAS28 was dose dependent.
Conclusion Current smoking in patients with early RA significantly reduces the clinical effect of an MTX-based strategy, independent of whether concomitant prednisone is used. This effect is dose dependent.
Smoking is a known risk factor for the development of rheumatoid arthritis (RA),1,2 and has been negatively associated with clinical response to several disease-modifying antirheumatic drugs (DMARDs).3,4,5,6 For example, in the Swedish Farmacotherapy (SWEFOT) study and the U-Act-Early trial, current smoking negatively predicted the likelihood of response to methotrexate (MTX) in early RA.3,4 In the U-Act-Early trial, this association was dose dependent.4 Smoking was also found to be negatively associated with clinical response to rituximab (RTX) and anti–tumor necrosis factor (anti-TNF) treatment.5,6 As far as we know, the association of smoking and clinical response to glucocorticoids (GCs) has never been studied, whereas concomitant GC treatment in RA is common in daily clinical practice. Early RA patients at initiation of MTX treatment often receive concomitant GCs7 to further reduce disease activity and radiographic progression and improve functional ability.8,9 For example, in the second Computer Assisted Management in Early Rheumatoid Arthritis trial (CAMERA-II), disease activity improved faster and erosive joint damage was significantly less in the MTX strategy group with prednisone, compared to the MTX strategy group with placebo (PBO), showing benefit of concomitant prednisone.9 However, it has been found that approximately 30% of patients have a reduced or absent clinical response to GCs.10,11 This might be associated with smoking, similar to the negative effects of smoking on the efficacy of other DMARDs. The primary objective of this study was to determine if the added clinical benefit of prednisone, when used concomitantly with MTX, is reduced among smokers.
METHODS
We used data from the CAMERA-II trial (isrctn.com identifier: 70365169), in which 236 patients with early RA were randomized to a treatment strategy initiating MTX and prednisone 10 mg/day (MTX + pred) or MTX and PBO (MTX + PBO). In short, MTX treatment was started at 10 mg/week and a tight control and treat-to-target strategy was followed with patient-tailored dosing adjustments at monthly visits on the basis of predefined response criteria aiming for remission. If no remission was achieved at 4 weeks after reaching the maximum (tolerable) MTX dosage, the route of MTX administration was switched from oral to subcutaneous, and if thereafter still no remission was achieved, adalimumab was added. Prednisone or PBO was given in stable dose for 24 months. Details have been described previously.9
Ethics approval. The CAMERA-II trial was approved by the medical ethical committee of the University Medical Center Utrecht (medical ethical committee number: 02/042) and by the institutional review boards of all involved hospitals. The following Dutch hospitals were involved: University Medical Center Utrecht, Diakonessenhuis, Utrecht; St. Antonius Hospital, Nieuwegein; Meander Medical Center, Amersfoort; Tergooi Hospital, Hilversum; St. Jansdal Hospital, Harderwijk; and Flevohospital, Almere, the Netherlands. Patients gave written informed consent before entering the study.
Statistical analyses. For the present analyses, data were used on treatment strategy, current smoking status (yes/no), current smoking level (0, 1–9, 10–19, and ≥ 20 cigarettes/day), BMI (kg/m2), sex, rheumatoid factor (RF) status, MTX and biological DMARD use, and Disease Activity Score assessing 28 joints (DAS28) at baseline and monthly up to 24 months. We used linear mixed modeling with a random intercept, DAS28 over 24 months’ follow-up as the dependent variable; current smoking status and strategy group as independent variables; and DAS28 at baseline, time (trial months), time2 (nonlinear course of DAS28 over time), BMI, sex, and RF status as covariates. In Model 1, we analyzed whether the effect of current smoking status (yes/no) on clinical response was different for MTX + pred when compared to MTX + PBO. To this end, the interaction term of group allocation and current smoking status was added to the model. Model 2 was similar to Model 1, but here we used the categorical variable of the different levels of current smoking as covariate in the model instead of the binary current smoking status variable, to evaluate the presence of a dose-response effect. Also, unadjusted data are shown on the course of DAS28 in each strategy arm of the CAMERA-II trial, separately for current smokers and nonsmokers (Figure 1). We used a random intercept at patient level (i.e., for the variable patient ID) to account for the dependence of repeated measurements over time, and the other variables were added as fixed effects in the model. These were current smoking status, sex, group allocation, RF status, DAS28 based on erythrocyte sedimentation rate at baseline, time, and BMI; further, the interaction terms time*time, and current smoking status*group allocation were entered in the model. We used a log-likelihood test to assess if the model fitted the data well. All analyses were performed with SPSS version 26 (IBM Corp.). All tests were 2-sided and P ≤ 0.05 was considered statistically significant.
DAS28 over time for current smokers vs nonsmokers in each strategy arm of CAMERA-II. (A) Strategy initiating methotrexate with prednisone. (B) Strategy initiating methotrexate with placebo-prednisone. DAS28: Disease Activity Score assessing 28 joints.
RESULTS
Current smoking data was available for 213 of 236 patients of the CAMERA-II trial. The baseline characteristics of these 213 patients were similar between the MTX + pred and MTX + PBO strategy groups (Table 1). Current smoking was significantly associated with a smaller reduction of DAS28 over time compared to current nonsmoking (β 0.57, 95% CI 0.22–0.92, P < 0.01; Table 2). This association was not statistically different between the MTX + pred and MTX + PBO strategy groups (P value for interaction term of treatment strategy and current smoking status = 0.73). In line with this, Figure 1 shows a clear effect of current smoking on the course of DAS28 that is of similar magnitude in each of the 2 strategy arms. The regression coefficient for BMI in the model was 0.03 (95% CI 0.00–0.06, P = 0.04; Table 2).
Baseline characteristics for each strategy arm group*.
Smoking as predictor and DAS28 over time (24 months) as outcome: 2 models.
The association between current smoking and DAS28 over time was dose dependent: β 0.38 (95% CI –0.27 to 1.02, P = 0.25) for patients who smoked 1–9 cigarettes, β 0.59 (95% CI 0.06–1.12, P = 0.03) for patients who smoked 10–19 cigarettes, and β 0.66 (95% CI 0.18–1.13, P < 0.01) for patients who smoked ≥ 20 cigarettes per day (Table 2). The interaction terms between treatment strategy and current smoking levels were (again) not statistically different between the MTX + pred and MTX + PBO strategy groups. The model fitted the data well (P < 0.00001).
DISCUSSION
To our knowledge, this is the first study investigating the association of current smoking with the clinical effect of concomitant prednisone in RA. In this study in patients with early RA, we found a negative effect of current smoking on DAS28 for MTX-based strategies, independent of concomitant use of prednisone; this effect was dose dependent. These findings corroborate previously found negative associations between current smoking and response to MTX treatment in patients with early RA.4,12 BMI had a (small) effect additionally to current smoking.
As MTX is often used in daily clinical practice,7 our results emphasize the importance of nonsmoking in patients with early RA treated with MTX in clinical practice. Patients who smoke also have worse responses to RTX and anti-TNF treatment,5,6 which many patients require later on in the disease course.13,14 Not smoking or smoking cessation have other positive effects, such as on cardiovascular risk, which is especially important as this risk is increased in RA.15,16,17
The smaller reduction of disease activity over time induced by current smoking is not statistically significantly different between MTX + pred and MTX + PBO; this would indicate that the effect of prednisone is not statistically significantly influenced by current smoking. However, we cannot rule out (in)direct interactions between MTX and prednisone under current smoking and inflammatory conditions. If we compare our results in RA with those in other chronic diseases, such as in patients with asthma, no difference in effect of GCs on relapse rate has been shown for smokers and nonsmokers.18 Also, in a study with 18 healthy male adults, no statistically significant effect of smoking on systemic availability of prednisolone was observed.19
The strength of our analysis is the use of data from a PBO-controlled trial, enabling the comparison of the association of current smoking with clinical response to prednisone and PBO-prednisone. Further, our study examined the relatively long-term effect (i.e., over 24 months) of current smoking on treatment response to MTX as well as concomitant prednisone, whereas other studies examined the effect of current smoking on treatment effect to MTX only, and only up to a maximum of 6 months of treatment.3,6
Our study has limitations. The negative association between current smoking and clinical response might partly be explained by anticyclic citrullinated peptide status, but this status had not been assessed in CAMERA-II. However, this does not diminish the clinical relevance of our findings. Data on current smoking for almost 10% of patients were missing. However, as there is no reason to assume that these missing data were not at random, it is unlikely that they have influenced our findings. The number of participating patients was limited, so it cannot be ruled out that current smoking would have a minor negative effect on the benefit of prednisone, but our results show that a clinically relevant effect is absent.
Our finding that current smoking negatively affects the clinical response to MTX-based strategies and that this negative current smoking effect is not different if concomitant prednisone is used, warrants validation, although circumstantial evidence supports this finding.
ACKNOWLEDGMENT
The Society for Rheumatology Research Utrecht (SRU) centers that participated in the CAMERA-II trial included: University Medical Center Utrecht, Antonius Hospital Nieuwegein, Diakonessen hospital Utrecht, Gelre ziekenhuizen Apeldoorn, Meander Medical Center Amersfoort, Maartenskliniek Woerden, St. Jansdal Harderwijk, and Tergooi ziekenhuizen Hilversum.
Footnotes
JMvL received honoraria from MSD, Roche, Pfizer, BMS, and Eli Lilly. MSK was supported by a research grant from AstraZeneca; AstraZeneca was not involved in this study. MJdH is an employee of Novartis Pharma BV; Novartis was not involved in this study. JWGJ and PMJW report no competing interests.
- Accepted for publication January 19, 2021.
- © 2021 The Journal of Rheumatology
REFERENCES
DATA SHARING
Request for data access can be directed to the corresponding author.