Significance & Innovations
- Almost one-third of new-onset rheumatoid arthritis patients with insufficient response to methotrexate experienced unacceptable pain (visual analog scale for pain score >40 mm) after 21 months despite early active combination treatment.
- Pain indicative of a noninflammatory mechanism accounted for >80% of all unacceptable pain 21 months after randomization.
- Adding infliximab, as compared to sulfasalazine plus hydroxychloroquine, to methotrexate reduced both cumulative pain over 21 months and unacceptable pain at 21 months.
- The results display insufficient effects of current treatment strategies on inflammation-independent pain components, warranting alternative approaches in affected patients.
During the last 2 decades, a growing armamentarium of biologics in combination with adequate doses of methotrexate (MTX) and aggressive treat-to-target regimens have markedly improved the possibilities of reaching inflammation control in RA patients. Many reports have focused on the effects on inflammation, often comparing addition of a biologic disease-modifying antirheumatic drug (DMARD) to MTX with MTX in monotherapy, showing better results for the combination therapy (12). Three randomized controlled trials (of which the Swedish Pharmacotherapy [SWEFOT] trial, studied here, is one ), have also directly compared MTX plus an anti–tumor necrosis factor (anti-TNF) agent with so-called triple therapy (adding sulfasalazine [SSZ] and hydroxychloroquine [HCQ] to MTX), showing no difference between these approaches in the capacity of reducing standard measures of disease activity over 2 years or longer (13–15). However, none of these trials have compared the course of different aspects of pain between these 2 clinically relevant combination therapy options in new-onset RA.
The aim of the current study was to investigate the pain course over 21 months of follow-up between MTX-refractory, early RA patients randomized to addition of infliximab (IFX) as compared to addition of SSZ plus HCQ in the SWEFOT trial, including pain patterns indicative of both inflammatory and noninflammatory mechanisms.
In total, 487 patients were enrolled and started on MTX monotherapy. After the 3-month run-in phase, 258 patients did not reach low disease activity (DAS28 score ≤3.2) and were randomized to addition of IFX (n = 128) or SSZ plus HCQ (n = 130). As reported earlier, there was higher dropout during the trial in the SSZ plus HCQ group than in the IFX group (43% versus 30%; P = 0.01) (13). In total, 90 patients in the IFX group and 74 in the SSZ plus HCQ group continued the protocol until the 21-month follow-up. The dropout difference was mainly explained by therapy discontinuation due to lack of efficacy (for flowchart, see Supplementary Figure 1, available on the Arthritis Care & Research website at http://onlinelibrary.wiley.com/doi/10.1002/acr.24264/abstract).
At inclusion, patients assigned to either IFX or SSZ plus HCQ had overall high disease activity, with a mean DAS28 score of almost 6.0, and a severe pain load, with an average VAS pain score of nearly 60 mm (range 0–100) in both groups. Baseline characteristics, both at inclusion and randomization, were well balanced between the treatment arms, although with a trend toward higher CRP level (P = 0.074) at inclusion and a higher VAS pain score (P = 0.11) at randomization in the SSZ plus HCQ group (Table 1).
(n = 128)
SSZ + HCQ
(n = 130)
|Age, years||52 ± 13||53 ± 14|
|Women, no. (%)||97 (76)||101 (78)|
|Education level, no. (%)|
|≤9 years||19 (15)||26 (20)|
|10–12 years||71 (56)||63 (49)|
|>12 years||30 (23)||30 (23)|
|Symptom duration, months||6.2 ± 3.5||6.2 ± 3.5|
|RF positive, no. (%)||89 (69)||85 (65)|
|VAS pain, mm||59 ± 20||59 ± 21|
|CRP level, median (range) mg/liter||17 (0–150)||24 (0–203)|
|Swollen joint count (of 28 joints)||11 ± 5||11 ± 5|
|Low-dose glucocorticoid use, no. (%)||8 (6)||10 (8)|
|DAS28 score||5.89 ± 0.93||5.98 ± 0.96|
|HAQ score||1.27 ± 0.60||1.32 ± 0.59|
|EQ-5D score||0.35 ± 0.33||0.36 ± 0.33|
|VAS pain, mm||45 ± 21||41 ± 24|
|CRP level, median (range) mg/liter||9 (0–149)||9 (0–155)|
|Swollen joint count (of 28 joints)||7 ± 5||6 ± 5|
|DAS28 score||4.91 ± 0.98||4.79 ± 1.05|
|HAQ score||0.94 ± 0.54||0.99 ± 0.57|
|EQ-5D score||0.51 ± 0.29||0.55 ± 0.27|
Values are the mean ± SD unless indicated otherwise. No. of missing values at inclusion for the IFX group: education level 8; VAS pain 1; HAQ 2; and EQ-5D 18; for the SSZ plus HCQ group: education level 11; VAS pain 1; HAQ 3; and EQ-5D 17. No. of missing values at randomization for the IFX group: VAS pain 4; CRP 3; HAQ 4; and EQ-5D 21; for the SSZ plus HCQ group: VAS pain 1; CRP 1; HAQ 2; and EQ-5D 13. CRP = C-reactive protein; DAS28 = Disease Activity Score in 28 joints; EQ-5D = EuroQol 5-domain questionnaire (UK preference set); HAQ = Health Assessment Questionnaire; RF = rheumatoid factor; VAS = visual analog scale.
Three months after inclusion.
Pain composition in all patients
In the observational, nonimputed setting, 81% of all randomized patients (n = 258) reported unacceptable pain at inclusion, with a mean ± SD VAS pain score of 59 ± 20 (Figure 1). The proportion with unacceptable pain decreased to 50% at randomization, with a mean ± SD VAS pain score of 43 ± 23 (both P < 0.001), and a further decline was seen until the 3-month follow-up. Thereafter, both measures remained largely stable throughout the study period; at 21 months, 29% of the patients reported unacceptable pain, and the mean ± SD VAS pain score was 31 ± 24 (both P < 0.001 for the change from randomization). When unacceptable pain was stratified by concomitant inflammation, the following patterns were displayed: inflammatory pain (see Methods section) dominated at inclusion but declined from initially 53% to 19% at randomization (P < 0.001), further decreasing to 5% at 21 months (P < 0.001 for the change from randomization). In contrast, unacceptable pain despite inflammation control (refractory pain) remained fairly stable, at ~25–30% during the study period, and accounted for 82% of all unacceptable pain at 21 months.
Comparison of pain composition between treatment arms
AUC for VAS pain
Among patients assigned to IFX, the AUC for VAS pain from randomization to 21-month follow-up was significantly lower than for patients randomized to SSZ plus HCQ in the main analysis (P = 0.01), with the period from 9 to 21 months accounting for most of that difference (Table 2 and Figure 2A). In the sensitivity analyses, a similar between-group difference regarding AUC for VAS pain from randomization to 21 months was demonstrated in the ITT (MI) approach (P = 0.008), while no significant difference was found in the completer analysis (Table 2 and Figures 2B and C).
|Characteristic||No.||IFX†||SSZ + HCQ†||
(IFX vs. SSZ + HCQ)‡
|Randomization to 21 months|
|AUC for VAS pain|
|ITT (LOCF)||258||705 ± 421||761 ± 465||−111 (−195, −27)||0.010|
|ITT (MI)||258||706 ± 358||773 ± 404||–112 (−195, −29)||0.008|
|Completers||149||649 ± 377||608 ± 419||−12 (−113, 88)||0.81|
|At 21-month follow-up|
|ITT (LOCF)||258||41 (32)||59 (45)||0.68 (0.51, 0.90)||0.008|
|ITT (MI)||258||44 (34)||62 (48)||0.69 (0.52, 0.91)||0.030|
|Completers||152||22 (26)||23 (35)||0.72 (0.45, 1.14)||0.16|
|ITT (LOCF)||258||30 (23)||37 (28)||0.80 (0.54, 1.18)||0.27|
|ITT (MI)||258||24 (18)||32 (25)||0.72 (0.46, 1.14)||0.21|
|Completers||151||17 (20)||19 (29)||0.74 (0.42, 1.31)||0.30|
|Refractory pain (strict)#|
|ITT (LOCF)||258||17 (13)||15 (12)||1.13 (0.61, 2.09)||0.70|
|ITT (MI)||258||11 (9)||12 (9)||0.93 (0.45, 1.96)||0.86|
|Completers||151||11 (13)||12 (18)||0.75 (0.37, 1.51)||0.42|
AUC = area under the curve; CRP = C-reactive protein; ITT = intent-to-treat; LOCF = last observation carried forward; MI = multiple imputation; VAS = visual analog scale.
Values for randomization to 21 months are the mean ± SD. Values for 21-month follow-up are the no. (%).
Values for randomization to 21 months are the mean difference (95% confidence interval [95% CI]), estimated by analysis of covariance with adjustment for age, sex, and VAS pain at randomization. Values for 21-month follow-up are the relative risk (95% CI), estimated by modified Poisson regression analysis with robust variance with adjustment for age, sex, and VAS pain at randomization.
VAS pain score >40 mm (range 0–100).
VAS pain score >40 mm plus CRP level <10 mg/liter.
VAS pain score >40 mm plus CRP level <10 mg/liter plus ≤1 swollen joint (of 28 joints).
In patients randomized to IFX, the fraction with unacceptable pain declined from 57% at randomization to 32% at 21 months (P < 0.001), while for this period, no significant difference was seen for the SSZ plus HCQ group (ITT [LOCF] analysis) (Figure 3A). Comparing randomized arms, unacceptable pain started diverging at the 9-month follow-up, and the adjusted RR of having unacceptable pain after 21 months was 0.68 (95% confidence interval [95% CI] 0.51, 0.90; P = 0.008) for patients randomized to IFX versus those assigned to SSZ plus HCQ (Table 2). These results are also in line with the between-group result for unacceptable pain at 21 months when using the ITT (IM) analysis (for IFX versus SSZ plus HCQ, adjusted RR 0.69 [95% CI 0.52, 0.91]; P = 0.03) (Table 2 and Supplementary Figure 2A, available on the Arthritis Care & Research website at http://onlinelibrary.wiley.com/doi/10.1002/acr.24264/abstract). In the completer analysis, a numerical trend toward less unacceptable pain in the IFX group at 21 months was observed (26% versus 35%) but did not reach statistical significance (adjusted RR 0.72 [95% CI 0.45, 1.14]; P = 0.16) (Table 2 and Supplementary Figure 3A, available on the Arthritis Care & Research website at http://onlinelibrary.wiley.com/doi/10.1002/acr.24264/abstract).
As for unacceptable pain without inflammation control (inflammatory pain) at 21 months, the adjusted RR was 0.48 (95% CI 0.24, 0.93; P = 0.031) comparing the IFX group with the SSZ plus HCQ group (ITT [LOCF] analysis) (Figure 3B). Neither of the sensitivity analyses showed any significant difference in inflammatory pain at 21 months between the randomized groups, although in the ITT (MI) analysis, a numerical trend toward less inflammatory pain in the IFX group was displayed (16% versus 24%; adjusted RR 0.67 [95% CI 0.39, 1.16]; P = 0.19) (see Supplementary Figures 2B and 3B, available at http://onlinelibrary.wiley.com/doi/10.1002/acr.24264/abstract).
None of the randomized treatment arms showed a within-group difference in unacceptable pain despite inflammation control (refractory pain) from randomization to the 21-month follow-up. When comparing the randomized arms at 21 months, there was a nonsignificant, numerically lower proportion with refractory pain in the IFX group (23% versus 28% in the SSZ plus HCQ group; P = 0.27) (ITT [LOCF] analysis; see Table 2 and Figure 3C). As for refractory pain with a strict definition, a within-group increase of patients with this feature was seen from randomization to 21 months in both treatment arms, but no significant between-group difference was detected at the 21-month follow-up (Table 2 and Figure 3D). Neither in the sensitivity analyses were any significant between-group differences at 21 months for refractory pain or refractory pain with a strict definition displayed (ITT [MI] and completer analyses, respectively) (Table 2, Supplementary Figures 2C and D, and Supplementary Figures 3C and D, available at http://onlinelibrary.wiley.com/doi/10.1002/acr.24264/abstract).
In this study of new-onset RA patients who responded insufficiently to initial MTX, we show that almost one-third still report unacceptable pain (VAS pain score >40 mm) 21 months after addition of either IFX or SSZ plus HCQ. At this time point, unacceptable pain despite inflammation control (refractory pain; VAS pain >40 mm plus CRP level <10 mg/liter) constituted more than 4/5 of all unacceptable pain. Notably, this pattern (with a domination of refractory pain) was evident already 3 months after starting combination therapy. Furthermore, we demonstrate that addition of IFX as compared to addition of SSZ plus HCQ resulted in significantly less cumulative pain over 21 months (as measured by AUC for VAS pain). Addition of IFX also reduced the risk of unacceptable pain at 21 months by >30%.
Inflammation is a well-established cause of pain in patients with inflammatory arthritides (5). Although inflammation control clearly reduces pain (12), there is mounting evidence of an uncoupling between pain and inflammation during the disease course (6, 8, 19, 21). In support of this, the present findings show that, in early RA, 24% of MTX nonresponders who were randomized to additional therapy experienced refractory pain indicative of a noninflammatory mechanism after 21 months. Moreover, the results of the current study are consistent with previous observational reports from our group, which demonstrated that significant pain is common despite a 3-month EULAR good response to MTX (8), and that after 3 months with MTX, 16% of patients had refractory pain (19). In addition, our present results are congruent with the findings of Koop et al, who reported significant pain with a pattern indicative of a noninflammatory mechanism in 17% of RA patients in a well-treated, established cohort (21).
The higher prevalence of refractory pain shown in the current study compared to the findings mentioned above is likely explained both by the measure used to capture refractory pain (which differed from the pain DETECT questionnaire used by Koop et al) (21) and by the fact that patients reaching EULAR-defined low disease activity after 3 months of receiving MTX were not included here. The latter group is known to do very well also in the longer term (22), and the present analysis (not encompassing these MTX responders) was hence performed on a more therapy-resistant cohort compared to an unselected early RA setting, implying that the occurrence of refractory pain seen here might be higher than had otherwise been the case. The current results displaying a substantial subgroup with refractory pain despite aggressive treatment are also in line with a report by Lee et al, who demonstrated the persistence of significant pain in patients with DAS28 remission (7). Hence, whereas nociceptive pain in RA patients (elicited by a normal somatosensory nervous system as a response to inflammation) generally responds well to immunosuppressive agents (12), our results strengthen the evidence that current therapy regimens are insufficient for preventing the occurrence of noninflammatory pain components often arising from central sensitization (increased responsiveness to stimuli in the central nervous system) (23).
To the best of our knowledge, unacceptable pain and its inflammation-derived and noninflammatory subcomponents have not been compared before between early RA patients randomized to the addition of a biologic therapy versus the addition of SSZ plus HCQ in MTX nonresponders in the setting of a randomized controlled trial. O’Dell et al, however, analyzed the level of VAS pain in a similar trial setting (the Rheumatoid Arthritis: Comparison of Active Therapies [RACAT] trial) (14); but in contrast to our findings of less pain when adding IFX, they did not find any significant difference between addition of etanercept versus addition of SSZ plus HCQ to MTX. Apart from a more limited pain analysis, their follow-up was also longer (48 weeks versus 21 weeks in the current study), and it could be argued that this might have attenuated gains from a potentially faster initial disease activity decline in the biologics group, although our results rather indicate that between-group differences increased over time.
Furthermore, the significantly lower pain load over 21 months (as measured by the AUC for VAS pain), as well as less unacceptable pain at 21 months, here demonstrated for IFX, as compared to SSZ plus HCQ, interestingly contrasts to some earlier SWEFOT findings. In those, no significant differences between the randomized arms were reported for the proportion of patients reaching a 21-month EULAR-defined good response (13). Neither were any significant differences detected regarding EuroQol 5-domain questionnaire (UK preference set) utilities, gains in quality-adjusted life years (24), or sick leave/disability pension (25), measures previously shown to associate with pain in RA patients (1, 2). The between-group difference in unacceptable pain in the current study seems to be driven mainly by a difference in unacceptable pain without inflammation control. This suggests a more pronounced effect of the biologics combination strategy on pain aspects of inflammation as compared to the composite EULAR-defined good response rather than a difference between randomized arms regarding uncoupling between pain and inflammation. Furthermore, the within-group increase in refractory pain with a strict definition from randomization to 21-month follow-up in both treatment arms was not unexpected because the definition of this measure included a combination of CRP level <10 plus ≤1 SJC (of 28), which was hardly seen at randomization (when patients per definition needed to have a DAS28 score >3.2).
Concerning sensitivity analyses, the statistically significant 21-month difference regarding both AUC for VAS pain and unacceptable pain between the IFX group and the SSZ plus HCQ group was practically unaltered when MI was used instead of LOCF to replace missing data (due to protocol breach or other reasons) in the ITT analysis. However, in the analysis conducted solely on patients completing 21 months of follow-up with the randomized treatment (completer analysis), no significant between-group differences were shown, which is likely explained by a significantly higher dropout rate due to medication inefficacy in the SSZ plus HCQ group, adequately compensated for in the ITT (LOCF) and ITT (MI) analyses.
The clinical question addressed in the SWEFOT trial, whether adding an anti-TNF agent is superior to addition of further conventional DMARDs in early MTX refractory disease, is still of central interest in treatment guidelines (26). Pain, including occurrence of noninflammatory pain, is a major component of remaining unmet needs for RA patients, and here, early recurrent measurements of VAS pain, CRP level, and SJC allow for thorough assessment of pain patterns.
This study also has limitations. An important issue is how to capture unacceptable pain. Here, we used the validated measure of PASS (VAS pain score >40 mm) described by Tubach et al (18), but a lower cutoff for significant pain (VAS pain score >20 mm) has been proposed by Wolfe and Michaud (27), rendering higher prevalence estimates when applied (8). Furthermore, although a VAS pain score >40 mm combined with a CRP level <10 mg/liter is an established measure of pain despite inflammation control (refractory pain), there are also other indices described to capture noninflammatory pain, such as the continuous DAS28-P (a modified DAS28 based on patient global assessment and TJC) (28). To enable a more straightforward clinical interpretation, we preferred a dichotomous cutoff reflecting pain deemed acceptable or unacceptable by patients and therefore used PASS combined with inflammation control, as outlined above. To further strengthen the assessment of inflammation control, we added an analysis with the criterion ≤1 SJC to a CRP level <10, but still no between-group differences in refractory pain were displayed. However, it cannot be ruled out that even with this stricter definition some patients might experience unacceptable pain due to low-grade inflammation.
Regarding the trial, a main concern is the open-label design. The absence of masking could subconsciously have biased both patients’ pain reports and physicians’ assessments and also provided patients with an incentive to abandon conventional treatment in favor of biologic therapy. Furthermore, IFX given at the clinic led to more total visits, potentially providing a sense of being better taken care of. In addition, it is known that more intensive interventions, such as an infusion, may have higher placebo effects (29). This might have affected reports of pain, presumably toward less pain in the IFX group. Other limitations include the relatively short follow-up (central pain sensitization might turn up after more longstanding inflammation) and that the trial was designed and powered mainly to detect a difference in EULAR-defined treatment response rates rather than pain and was thus possibly underpowered for the current analyses.
Finally, it should be noted that the current study comprises several secondary analyses that have an exploratory character. This warrants caution in the interpretation of the results, which ought to be viewed as suggestive rather than confirmatory.
In conclusion, almost one-third of new-onset RA patients with insufficient response to MTX experienced unacceptable pain after 21 months despite early active additional treatment, and refractory pain indicative of a noninflammatory mechanism accounted for more than 4/5 of this pain load. This pattern was observed already 3 months after the start of combination therapy. Adding IFX, as compared to addition of SSZ plus HCQ, reduced both cumulative pain over 21 months and unacceptable pain at 21 months. This lends some support to a better effect on long-term pain for the biologic treatment compared with triple therapy. However, our findings are also in line with insufficient effects of current treatment strategies to prevent development of inflammation-independent pain components, warranting early alternative treatment approaches in affected patients.
We are indebted to all patients, colleagues, and staff who made the SWEFOT trial possible.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Olofsson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design
Olofsson, Wallman, Jöud, Schelin, Ernestam, van Vollenhoven, Saevarsdottir, Lampa.
Acquisition of data
Olofsson, Wallman, Ernestam, van Vollenhoven, Saevarsdottir, Lampa.
Analysis and interpretation of data
Olofsson, Wallman, Jöud, Schelin, Ernestam, van Vollenhoven, Saevarsdottir, Lampa.
Author Saevarsdottir is an employee of deCODE genetics.