Drug Therapy in Undifferentiated Arthritis
After screening 3608 titles and abstracts, 67 articles were screened full text.
In total 30 articles were selected: 11 on 10 clinical trials of which eight were RCTs and five placebo controlled, seven on three cohort studies, and 10 review/opinion articles and two recommendations which were disregarded for this analysis. (figure 1)
(Enlarge Image)
Figure 1.
Flow diagram of systematic literature search for publications on drug treatment of patients with undifferentiated arthritis (UA).
(Enlarge Image)
Figure 2.
(A) Remission percentages after 1 year and 30 months16 follow-up of the four completed placebo controlled trials on temporary treatment of patients with UA. (B) Percentages of patients who progressed to rheumatoid arthritis after 1 year and 30 months16 follow-up of the four completed placebo controlled trials on temporary treatment of patients with UA * p=0.048; CS, corticosteroids; IM, intra-muscular; RA, rheumatoid arthritis.
Characteristics of clinical trials are shown in Table 1. Three trials included patients with UA and RA. Drug therapies studied were intramuscular (IM) or intra-articular corticosteroids (three RCTs, two placebo controlled, one open-label trial), DMARDs with or without oral GCs (two RCTs, one placebo controlled, one open-label trial), biological agents (two placebo controlled RCTs) and one RCT comparing tight control (TC) with 'routine DMARD' treatment. On one placebo controlled RCT two articles with different follow-up duration were published.
Characteristics of observational studies are shown in Table 2. Five of seven publications were based on the Norfolk Arthritis Register (NOAR), a primary care-based cohort in adults with ≥two swollen joints for at least 4 weeks, of which 46% at baseline fulfilled the 1987 criteria for RA. Four studies compared treated and untreated patients and/or early versus delayed start of treatment. Adjustments for differences in disease severity or time dependent confounders between groups were made using propensity scores or marginal structural models. One publication comparing anticitrullinated protein antibodies positive and negative patients and two publications without comparisons between treated and/or untreated patients are not mentioned further.
Main outcomes of all included clinical trials are shown in Table 3. Seven clinical trials investigated temporary treatment and three trials continuous tight controlled treatment. Follow-up varied between 3 months and 5 years. Many different outcome measures were used for assessing response to treatment in terms of disease activity state, achieving remission, joint damage or progression to RA.
Van Dongen et al and van Aken et al compared a 12 months course of methotrexate (MTX) with placebo after 30 months and 5 years follow-up. After 30 months 22 (40%) in the MTX group and 29 (53%) in the placebo group had progressed to RA (1987 criteria) (p value not published), after 5 years 25 (45%) and 29 (53%) did (p=0.45). Remission was achieved in comparable numbers after 30 months and 5 years (after 30 months 15 (27%) and 13 (24%) and after 5 years 20 (36%) and 15 (27%) in the MTX and placebo groups, respectively (p values not published)). All patients in the placebo group who progressed to RA did so within 1 year compared with half of the patients in the MTX group (p=0.04), suggesting that progression to RA was at least postponed by 1 year of MTX treatment. After 30 months, more patients showed radiological progression in the placebo group (14 vs 6, p=0.046), but after 5 years median Sharp- van der Heijde (SHS) progression did not differ between groups (p=0.78). Up to 30 months, fewer adverse events (AEs) were reported in the placebo group, serious AEs (SAEs) were reported similarly in both groups.
de Jong et al compared MTX monotherapy with MTX+sulphasalazine (SSZ)+hydroxychloroquine (HCQ) in patients with early arthritis at high risk for developing persistent arthritis according to the prediction model of Visser et al All patients received GC bridging therapy (either a tapering scheme or IM injection). After 3 months, the combination therapy group had a lower mean disease activity score (DAS) than the monotherapy group (difference (95% CI) 0.39 (0.67 to 0.11)). No significant difference was seen between oral and IM GC bridging therapy. AE and SAE were reported in 67 (75%) and 50 (56%) in the combination therapy and the monotherapy groups. Fewer medication changes were made in the monotherapy group (14 (16%) vs 18 (20%), p=0.006).
In an open-label trial in patients with UA or recent onset RA (2010 criteria) MTX was combined with a tapered high dose of prednisone for 4 months. Remission after 4 months was achieved in 79 (65%) patients with UA and 291 (61%) patients with RA (p=0.5). Median (IQR) SHS progression was 0 (0–0) in patients with UA and RA (p=0.9). AEs were reported in 341 (56%) and SAEs in 16 (3%) of all patients.
These studies indicate that synthetic DMARDs suppress disease activity in patients with UA. MTX monotherapy may postpone but not prevent the development of RA and may slow down radiological progression. It appears that initial combination therapy with MTX and multiple DMARDs or corticosteroids (oral or parenteral) results in better short-term clinical outcomes. No long-term data are available.
Two trials have investigated biological agents in patients with UA. Saleem et al compared a 14 week course of infliximab with placebo in patients with UA who had relapsed after a single corticosteroid injection. If clinical inflammation was persistent after week 14, MTX was started. Independent safety monitors halted recruitment 'because of poor outcomes in all subjects' before inclusion was completed, after inclusion of 17 patients (10 randomised to infliximab, 7 to placebo). Clinical remission at 26 weeks was achieved in one patient and two patients in the placebo and infliximab groups, respectively. After 1 year, all patients in the infliximab group had progressed to RA (1987 criteria) compared with 5/7 in the placebo group, in which this occurred earlier (after a median of 14 weeks compared with 26 weeks, respectively). Data on (S)AE were not reported.
Emery et al compared a 6 months course of abatacept with placebo. After 1 year, respectively 12 (46%) and 16 (67%) of the abatacept and placebo groups progressed to RA (1987 criteria) (difference (95% CI) −21% (−47% to 8%)). Radiological progression (Genant-modified Sharp score) after 1 year was significantly less in the abatacept group (difference in total score −1.10 (95% CI −2.05 to −0.15)). Remission (DAS28 definition) after 1 year was achieved in nine (47%) and five (39%) patients in the abatacept and placebo groups, respectively. Numbers of reported AEs and SAEs were similar.
These trials suggest that a biological agent may slow down progression to RA in patients with UA. Early treatment with abatacept appears to suppress radiological damage progression. Long-term benefits remain uncertain.
Green et al performed an open-label pilot with intra-articular GC injections in all arthritic joints in 51 patients. Clinical synovitis was absent in 23 (45%) and 26 (51%) after 12 weeks 1 year, respectively.
Marzo-Ortega et al injected all inflamed joints with GC (early intervention (EI) group) and compared this with 'conservative treatment' (CT group) with non-steroidal anti-inflammatory drugs. In case of progression to polyarthritis SSZ was started. Clinical synovitis was absent in 25 (81%) and 16 (57%) patients after 52 weeks in the EI and CT groups, respectively (p=0.05), but more patients in the EI group started DMARD treatment (14 (45%) versus 4 (14%), p=0.012). The EI group reported a significantly lower mean visual analogue scale pain after 4 weeks than the CT group, but not after 12 weeks and 52 weeks. Data on (S)AE were not reported.
Machold et al compared a single IM injection of GC with placebo. Respectively, 32 (16%) and 33 (18%) in the GC and placebo groups achieved persistent remission without additional treatment after 1 year (p=0.68). Initiation of a DMARD and core set variables were comparable. AEs generally were mild and comparable between groups.
Verstappen et al compared a 3-week course of IM GC injections with placebo. When patients met ≥two of four predefined criteria (≥three swollen joints, ≥six painful joints, morning stiffness ≥45 min or erythrocyte sedimentation rate (ESR) ≥28 mm/h), they were referred for DMARD treatment. After 6 months, patients in the placebo group were more often referred for DMARD treatment than the GC group (96 (76%) vs 77 (61%), adjusted OR (95% CI) 2.11 (1.16 to 3.85), p=0.015). After 1 year, remission without DMARD use was less often achieved in the placebo group (11 (10%) vs 22 (20%), adjusted OR (95% CI) 0.42 (0.18 to 0.99), p=0.048). Sixty-seven (60%) and 54 (49%) in the placebo and GC groups were classified as RA (1987 criteria) (adjusted OR (95% CI) 1.58 (0.85 to 2.93), p=0.15). AEs were comparable between groups.
These studies indicate that a single corticosteroid injection probably has no long-term benefit. Repeated IM corticosteroid injections may postpone the need to start DMARDs but not prevent progression to RA, and in one study possibly encourage remission. No long term follow-up data exist.
van Eijk et al compared TC treatment (TC group) with conventional care (CC group) in patients with early arthritis. The TC group (n=42) started with MTX monotherapy, medication was intensified in case of no remission (19 patients changed to adalimumab, 15 increased adalimumab, 11 switched to multiple DMARDs, 3 added prednisone and 1 switched to leflunomide). The CC group (n=40) used conventional DMARDs without treatment target (24, 14 and 2 patients started HCQ MTX and SSZ, respectively). No prednisone or biologicals were allowed. After 2 years, respectively 66% and 49% in the TC and CC groups were in remission (numbers and p values not published). Median SHS progression was 0 (0–1.0) and 0.25 (0–2.5) in the TC and CC groups (p=0.17). Over 2 years, no significant differences in DAS and Health Assessment Questionnaire (HAQ) levels were seen. The number of reported AEs was higher in the TC group (62 vs 35, p=0.03). The number of SAEs was comparable.
In conclusion, this trial shows no benefit to patients with UA of TC treatment over CC in terms of radiological and clinical outcomes and achieving remission.
No RCTs compared early versus delayed start of treatment in patients with UA. In an open label study no difference in proportions remission was found between patients with UA and RA (2010 criteria) after 4 months of MTX and a tapered high dose of prednisone. But although patients with UA had a lower baseline DAS, baseline symptom duration was similar between patients with UA and RA.
In an observational study in the ESPOIR cohort, Lukas et al compared early versus delayed treatment, adjusting for selection bias using propensity scores. The estimated marginal mean (SE) SHS progression was 0.8 (0.37) and 1.7 (0.19) in patients who respectively started DMARD therapy within and after 3 months (p=0.03). Stratification in propensity quintiles showed that only patients with high baseline disease activity starting DMARD therapy after 3 months showed more progression than patients starting within 3 months.
Bukhari et al found a similar result in the NOAR cohort, also using propensity scores. Starting treatment within 6 months after symptom onset was associated with less radiological damage after 5 years than starting after 6–12 months and >12 months (OR (95% CI) 1.5 (0.9 to 2.3) vs 2.3 (1.4 to 3.9) and 2.2 (1.4 to 3.5), respectively, with untreated patients as reference (1.0)).
Wiles et al compared early versus delayed treatment in the NOAR cohort using propensity scores, with functional ability after 5 years as outcome. Starting treatment early (within 6 months of symptom onset) was not associated with a HAQ score ≥1.0 (OR 0.71 (0.34 to 1.44), but starting treatment after 6–12 months and >12 months was (OR (95% CI) 1.98 (0.86 to 4.54) and 2.03 (1.10 to 3.75), respectively, with untreated patients as reference (1.0)). Farragher et al used functional ability after 10 years as outcome and adjusted for time-dependent confounders using marginal structural models. Patients treated within 6 months after symptom onset improved more in functional ability than untreated patients, although not significantly (difference (95% CI) in change from baseline HAQ −0.24 (−0.58 to 0.09)). In patients treated after 6–12 months and >12 months functional ability improved less than in untreated patients (difference (95% CI) in change from baseline of respectively 0.12 (−0.13 to 0.37) and 0.18 (−0.06 to 0.41)). For each month that treatment was started earlier within 6 months, a significant additional benefit was found (difference (95% CI) in change from baseline HAQ −0.10 (−0.19 to −0.02) per month).
In conclusion, results from these observational cohort studies may indicate that disease outcomes improve if treatment is started within at the most 6 months after symptom onset, and starting sooner may even be better.
Results
After screening 3608 titles and abstracts, 67 articles were screened full text.
In total 30 articles were selected: 11 on 10 clinical trials of which eight were RCTs and five placebo controlled, seven on three cohort studies, and 10 review/opinion articles and two recommendations which were disregarded for this analysis. (figure 1)
(Enlarge Image)
Figure 1.
Flow diagram of systematic literature search for publications on drug treatment of patients with undifferentiated arthritis (UA).
(Enlarge Image)
Figure 2.
(A) Remission percentages after 1 year and 30 months16 follow-up of the four completed placebo controlled trials on temporary treatment of patients with UA. (B) Percentages of patients who progressed to rheumatoid arthritis after 1 year and 30 months16 follow-up of the four completed placebo controlled trials on temporary treatment of patients with UA * p=0.048; CS, corticosteroids; IM, intra-muscular; RA, rheumatoid arthritis.
Characteristics of clinical trials are shown in Table 1. Three trials included patients with UA and RA. Drug therapies studied were intramuscular (IM) or intra-articular corticosteroids (three RCTs, two placebo controlled, one open-label trial), DMARDs with or without oral GCs (two RCTs, one placebo controlled, one open-label trial), biological agents (two placebo controlled RCTs) and one RCT comparing tight control (TC) with 'routine DMARD' treatment. On one placebo controlled RCT two articles with different follow-up duration were published.
Characteristics of observational studies are shown in Table 2. Five of seven publications were based on the Norfolk Arthritis Register (NOAR), a primary care-based cohort in adults with ≥two swollen joints for at least 4 weeks, of which 46% at baseline fulfilled the 1987 criteria for RA. Four studies compared treated and untreated patients and/or early versus delayed start of treatment. Adjustments for differences in disease severity or time dependent confounders between groups were made using propensity scores or marginal structural models. One publication comparing anticitrullinated protein antibodies positive and negative patients and two publications without comparisons between treated and/or untreated patients are not mentioned further.
Main outcomes of all included clinical trials are shown in Table 3. Seven clinical trials investigated temporary treatment and three trials continuous tight controlled treatment. Follow-up varied between 3 months and 5 years. Many different outcome measures were used for assessing response to treatment in terms of disease activity state, achieving remission, joint damage or progression to RA.
Results of Synthetic DMARDs
Van Dongen et al and van Aken et al compared a 12 months course of methotrexate (MTX) with placebo after 30 months and 5 years follow-up. After 30 months 22 (40%) in the MTX group and 29 (53%) in the placebo group had progressed to RA (1987 criteria) (p value not published), after 5 years 25 (45%) and 29 (53%) did (p=0.45). Remission was achieved in comparable numbers after 30 months and 5 years (after 30 months 15 (27%) and 13 (24%) and after 5 years 20 (36%) and 15 (27%) in the MTX and placebo groups, respectively (p values not published)). All patients in the placebo group who progressed to RA did so within 1 year compared with half of the patients in the MTX group (p=0.04), suggesting that progression to RA was at least postponed by 1 year of MTX treatment. After 30 months, more patients showed radiological progression in the placebo group (14 vs 6, p=0.046), but after 5 years median Sharp- van der Heijde (SHS) progression did not differ between groups (p=0.78). Up to 30 months, fewer adverse events (AEs) were reported in the placebo group, serious AEs (SAEs) were reported similarly in both groups.
de Jong et al compared MTX monotherapy with MTX+sulphasalazine (SSZ)+hydroxychloroquine (HCQ) in patients with early arthritis at high risk for developing persistent arthritis according to the prediction model of Visser et al All patients received GC bridging therapy (either a tapering scheme or IM injection). After 3 months, the combination therapy group had a lower mean disease activity score (DAS) than the monotherapy group (difference (95% CI) 0.39 (0.67 to 0.11)). No significant difference was seen between oral and IM GC bridging therapy. AE and SAE were reported in 67 (75%) and 50 (56%) in the combination therapy and the monotherapy groups. Fewer medication changes were made in the monotherapy group (14 (16%) vs 18 (20%), p=0.006).
In an open-label trial in patients with UA or recent onset RA (2010 criteria) MTX was combined with a tapered high dose of prednisone for 4 months. Remission after 4 months was achieved in 79 (65%) patients with UA and 291 (61%) patients with RA (p=0.5). Median (IQR) SHS progression was 0 (0–0) in patients with UA and RA (p=0.9). AEs were reported in 341 (56%) and SAEs in 16 (3%) of all patients.
These studies indicate that synthetic DMARDs suppress disease activity in patients with UA. MTX monotherapy may postpone but not prevent the development of RA and may slow down radiological progression. It appears that initial combination therapy with MTX and multiple DMARDs or corticosteroids (oral or parenteral) results in better short-term clinical outcomes. No long-term data are available.
Results of Biological DMARDs
Two trials have investigated biological agents in patients with UA. Saleem et al compared a 14 week course of infliximab with placebo in patients with UA who had relapsed after a single corticosteroid injection. If clinical inflammation was persistent after week 14, MTX was started. Independent safety monitors halted recruitment 'because of poor outcomes in all subjects' before inclusion was completed, after inclusion of 17 patients (10 randomised to infliximab, 7 to placebo). Clinical remission at 26 weeks was achieved in one patient and two patients in the placebo and infliximab groups, respectively. After 1 year, all patients in the infliximab group had progressed to RA (1987 criteria) compared with 5/7 in the placebo group, in which this occurred earlier (after a median of 14 weeks compared with 26 weeks, respectively). Data on (S)AE were not reported.
Emery et al compared a 6 months course of abatacept with placebo. After 1 year, respectively 12 (46%) and 16 (67%) of the abatacept and placebo groups progressed to RA (1987 criteria) (difference (95% CI) −21% (−47% to 8%)). Radiological progression (Genant-modified Sharp score) after 1 year was significantly less in the abatacept group (difference in total score −1.10 (95% CI −2.05 to −0.15)). Remission (DAS28 definition) after 1 year was achieved in nine (47%) and five (39%) patients in the abatacept and placebo groups, respectively. Numbers of reported AEs and SAEs were similar.
These trials suggest that a biological agent may slow down progression to RA in patients with UA. Early treatment with abatacept appears to suppress radiological damage progression. Long-term benefits remain uncertain.
Results of Corticosteroid Injections
Green et al performed an open-label pilot with intra-articular GC injections in all arthritic joints in 51 patients. Clinical synovitis was absent in 23 (45%) and 26 (51%) after 12 weeks 1 year, respectively.
Marzo-Ortega et al injected all inflamed joints with GC (early intervention (EI) group) and compared this with 'conservative treatment' (CT group) with non-steroidal anti-inflammatory drugs. In case of progression to polyarthritis SSZ was started. Clinical synovitis was absent in 25 (81%) and 16 (57%) patients after 52 weeks in the EI and CT groups, respectively (p=0.05), but more patients in the EI group started DMARD treatment (14 (45%) versus 4 (14%), p=0.012). The EI group reported a significantly lower mean visual analogue scale pain after 4 weeks than the CT group, but not after 12 weeks and 52 weeks. Data on (S)AE were not reported.
Machold et al compared a single IM injection of GC with placebo. Respectively, 32 (16%) and 33 (18%) in the GC and placebo groups achieved persistent remission without additional treatment after 1 year (p=0.68). Initiation of a DMARD and core set variables were comparable. AEs generally were mild and comparable between groups.
Verstappen et al compared a 3-week course of IM GC injections with placebo. When patients met ≥two of four predefined criteria (≥three swollen joints, ≥six painful joints, morning stiffness ≥45 min or erythrocyte sedimentation rate (ESR) ≥28 mm/h), they were referred for DMARD treatment. After 6 months, patients in the placebo group were more often referred for DMARD treatment than the GC group (96 (76%) vs 77 (61%), adjusted OR (95% CI) 2.11 (1.16 to 3.85), p=0.015). After 1 year, remission without DMARD use was less often achieved in the placebo group (11 (10%) vs 22 (20%), adjusted OR (95% CI) 0.42 (0.18 to 0.99), p=0.048). Sixty-seven (60%) and 54 (49%) in the placebo and GC groups were classified as RA (1987 criteria) (adjusted OR (95% CI) 1.58 (0.85 to 2.93), p=0.15). AEs were comparable between groups.
These studies indicate that a single corticosteroid injection probably has no long-term benefit. Repeated IM corticosteroid injections may postpone the need to start DMARDs but not prevent progression to RA, and in one study possibly encourage remission. No long term follow-up data exist.
Results of Tight Control and Treat-to-Target Strategies
van Eijk et al compared TC treatment (TC group) with conventional care (CC group) in patients with early arthritis. The TC group (n=42) started with MTX monotherapy, medication was intensified in case of no remission (19 patients changed to adalimumab, 15 increased adalimumab, 11 switched to multiple DMARDs, 3 added prednisone and 1 switched to leflunomide). The CC group (n=40) used conventional DMARDs without treatment target (24, 14 and 2 patients started HCQ MTX and SSZ, respectively). No prednisone or biologicals were allowed. After 2 years, respectively 66% and 49% in the TC and CC groups were in remission (numbers and p values not published). Median SHS progression was 0 (0–1.0) and 0.25 (0–2.5) in the TC and CC groups (p=0.17). Over 2 years, no significant differences in DAS and Health Assessment Questionnaire (HAQ) levels were seen. The number of reported AEs was higher in the TC group (62 vs 35, p=0.03). The number of SAEs was comparable.
In conclusion, this trial shows no benefit to patients with UA of TC treatment over CC in terms of radiological and clinical outcomes and achieving remission.
Early versus Delayed Treatment
No RCTs compared early versus delayed start of treatment in patients with UA. In an open label study no difference in proportions remission was found between patients with UA and RA (2010 criteria) after 4 months of MTX and a tapered high dose of prednisone. But although patients with UA had a lower baseline DAS, baseline symptom duration was similar between patients with UA and RA.
In an observational study in the ESPOIR cohort, Lukas et al compared early versus delayed treatment, adjusting for selection bias using propensity scores. The estimated marginal mean (SE) SHS progression was 0.8 (0.37) and 1.7 (0.19) in patients who respectively started DMARD therapy within and after 3 months (p=0.03). Stratification in propensity quintiles showed that only patients with high baseline disease activity starting DMARD therapy after 3 months showed more progression than patients starting within 3 months.
Bukhari et al found a similar result in the NOAR cohort, also using propensity scores. Starting treatment within 6 months after symptom onset was associated with less radiological damage after 5 years than starting after 6–12 months and >12 months (OR (95% CI) 1.5 (0.9 to 2.3) vs 2.3 (1.4 to 3.9) and 2.2 (1.4 to 3.5), respectively, with untreated patients as reference (1.0)).
Wiles et al compared early versus delayed treatment in the NOAR cohort using propensity scores, with functional ability after 5 years as outcome. Starting treatment early (within 6 months of symptom onset) was not associated with a HAQ score ≥1.0 (OR 0.71 (0.34 to 1.44), but starting treatment after 6–12 months and >12 months was (OR (95% CI) 1.98 (0.86 to 4.54) and 2.03 (1.10 to 3.75), respectively, with untreated patients as reference (1.0)). Farragher et al used functional ability after 10 years as outcome and adjusted for time-dependent confounders using marginal structural models. Patients treated within 6 months after symptom onset improved more in functional ability than untreated patients, although not significantly (difference (95% CI) in change from baseline HAQ −0.24 (−0.58 to 0.09)). In patients treated after 6–12 months and >12 months functional ability improved less than in untreated patients (difference (95% CI) in change from baseline of respectively 0.12 (−0.13 to 0.37) and 0.18 (−0.06 to 0.41)). For each month that treatment was started earlier within 6 months, a significant additional benefit was found (difference (95% CI) in change from baseline HAQ −0.10 (−0.19 to −0.02) per month).
In conclusion, results from these observational cohort studies may indicate that disease outcomes improve if treatment is started within at the most 6 months after symptom onset, and starting sooner may even be better.