CLINICAL COURSE AND REMISSION RATE IN PATIENTS WITH EARLY RHEUMATOID ARTHRITIS: RELATIONSHIP TO OUTCOME AFTER 5 YEARS

British Journal of Rheumatology 1998;37:1324 1329 CLINICAL COURSE AND REMISSION RATE IN PATIENTS WITH EARLY RHEUMATOID ARTHRITIS: RELATIONSHIP TO OUTCOME AFTER 5 YEARS K. EBERHARDT and E. FEX* Department of Rheumatology, Lund University Hospital, Lund, Sweden SUMMARY Objective. To investigate the clinical course in early rheumatoid arthritis (RA) patients followed prospectively, to relate course to outcome after 5 yr, and to try to identify prognostic features. Methods. A total of 183 patients with definite RA and a mean disease duration of 11 months were included. Of these, 75% were rheumatoid factor ( RF) positive; 85% carried the shared epitope, 32% on both alleles. Most patients were assessed every 6 months. Disability was evaluated with the Health Assessment Questionnaire ( HAQ) and radiographic findings according to Larsen. Remission was defined in two ways: with the American Rheumatism Association (ARA) criteria and as no arthritis at least at one follow-up visit. Results. Twenty per cent achieved ARA-defined remission periods of at least 6 months duration; 21 were spontaneous and 18 drug induced. Average length of remission was 20.5 months. The remission periods constituted 7% of follow-up for all patients. Another 36% achieved remission according to the second definition. All 56% were considered to have a relapsingremitting disease pattern, in contrast to the remaining 44% with a persistent disease pattern. More patients with persistent disease were treated with disease-modifying anti-rheumatic drugs (DMARDs) and had also received a larger number of different drugs. Outcome after 5 yr regarding disability, joint inflammation and joint damage was worse for patients with persistent disease. Neither ARA-defined remission nor disease pattern could be accurately predicted. Conclusions. Long-term ARA-defined remission was rare, constituting 7% of follow-up for the entire cohort. For those 20% achieving remission, this period represented 34% of their follow-up. A total of 56% had a relapsing-remitting disease pattern and 44% had a persistent disease pattern. This classification had prognostic implications with persistency being a bad prognostic sign. KEY WORDS: Rheumatoid arthritis, Remission, Disease course, Prognosis, Outcome. THE prognosis for patients with rheumatoid arthritis in remission is rarely reported. However, in a retrospective ( RA) varies considerably. Early prediction of outcome study of established RA patients, the median has, therefore, been the subject of numerous investi- length of remission was 10 months [7]. gations focusing on, for example, sociodemographic, Different patterns of disease have been described. clinical or laboratory variables. No features of early The two main patterns are chronic persistent and the disease have so far shown sufficient power to allow relapsing-remitting disease course. Twenty-six per cent clinically meaningful prediction [1, 2]. In most studies, of patients in the Middlesex early RA study were prognosis of RA was related to the state of the disease chronic persistent over 3 yr [8], while in a Canadian at a certain defined end point. However, prognostic study of a 1985 inception cohort 40% had a persistent information may also be gained by evaluation of the disease course during 6 7 yr. The disease pattern may disease course up to that point in time. have prognostic implications. In the Canadian study, The most desirable patient outcome is probably patients with chronic persistent disease had worse achievement of stable remission. The length of remis- outcome at study finish [9]. sion periods during the course of the disease is consequently In this study, we examined the disease course in our a relevant prognostic measure. Remission cohort of early RA patients especially regarding ARA- can be defined according to the rigorous American defined remission and different disease patterns. The Rheumatism Association (ARA) criteria shown in disease course was related to outcome after 5 yr followup Table I [3]. More clinical definitions like being symptom and the prognostic value of different features, free or no arthritis on examination have also including rheumatoid factor ( RF) and shared epitope, been used. Remission criteria can either be examined was investigated. at a single point in time or during a longer time period. Reported cross-sectional remission rates in early RA TABLE I patients vary from 32% [4] to 7% after ~6 yr [5, 6 ]. Preliminary ARA remission criteria Various methods to define remission and different Five or more of the following criteria must be fulfilled patient selection are factors that contribute to the wide for at least two consecutive months: variation of remission rate. The length of time spent Duration of morning stiffness not exceeding 15 min No fatigue Submitted 30 March 1998; revised version accepted 24 August No joint pain (by history) 1998. No joint tenderness or pain on motion No soft-tissue swelling in joints or tendon sheaths Correspondence: K. Eberhardt, Department of Rheumatology, Erythrocyte sedimentation rate ( Westergren method) <30 mm Lund University Hospital, S-221 85 Lund, Sweden. after 1 h for a female or <20 mm after 1 h for a male *Deceased. 1324 1998 British Society for Rheumatology

EBERHARDT AND FEX: CLINICAL COURSE IN EARLY RA 1325 PATIENTS AND METHODS ation at least at one follow-up visit, and persistent if Patients the patient had active joints at all evaluation points. The patients were taking part in an ongoing prospective study of course and outcome of RA at the Statistical analyses Lund University Hospital in Southern Sweden [10]. Differences between groups regarding continuous The inclusion criteria were definite RA with disease variables were estimated by Mann Whitney s test for duration of <24 months and age >18 yr. Disease independent samples. The x2 test was used for discrete duration was defined as time from onset of symptoms. variables. Logistic multiple regression analyses were Most patients were referred from primary care as a performed in an attempt to predict ARA remission result of a special campaign to recruit cases of recent and clinical course, respectively. Demographics, RF onset. One hundred and eighty-three patients were status, genotype, and clinical and laboratory variables enrolled during the years 1985 1989. at baseline, were used as independent variables. All The study group consisted of all the 183 patients, tests were two-tailed and the level of significance was 67 men and 116 women, with a mean (S.D.) age of 51.4 set at P = 0.05. (12.4) yr at onset. Mean (S.D.) duration of joint symptoms at inclusion in the study was 11.1 (6.1) months. RESULTS Clinical features Table II shows some clinical and laboratory charac- Clinical assessment teristics at study start. Fifty-two per cent of the patients The patients were assessed at least every 6 months had no erosions. Onset of the disease was acute (within with the exception of a few patients in sustained 1 day) for 33 patients, intermediate (within 1 week) remission who were seen once a year. These patients for 32, insidious (within weeks to months) for 99, and were told to make contact with the clinic earlier if they palindromic for 19. One hundred and twenty-eight got joint symptoms. Joint tenderness was estimated (75%) were RF positive. One hundred and seventy according to Ritchie et al. [11]. Joint inflammation patients were genotyped, 145 (85%) of these carried was assessed by an active joint count (defined as the shared epitope. Fifty-four (32%) carried the epitope swollen and either tender or painful on motion). The on both alleles. 50 joints evaluated included all but two from the Patients with active disease were offered treatment Ritchie index, namely the neck and subtalar joints. with DMARDs and 114 (62%) received this treatment Disability was evaluated with a Swedish version of the some time during follow-up. Sixty-seven received one Stanford Health Assessment Questionnaire ( HAQ) drug, 27 two drugs and 18 received 3 5 drugs. disability index [12]. Radiographs of hands and feet Most commonly used was chloroquine (68 patients), were taken at study start and annually thereafter. We followed by D-penicillamine ( 51), gold compounds used standard film and an anterior posterior projec- (25), sulphasalazine (21) and methotrexate (6). tion. The radiographic findings were scored by the Twenty-nine (16%) were treated with oral cortico- Larsen method, grading changes from 0 (normal ) to 5 steroids. Active disease was defined as the presence of (maximal damage) [13]. Thirty-two joints were evalusix or more swollen joints and at least two of following ated. The wrist score was multiplied by five, then all features: (i) 9 tender joints; (ii) morning stiffness scores were added to give a joint damage score with a lasting at least 45 min; (iii) an ESR of 28 mm/h. theoretical range of 0 200. Clinical and laboratory During a follow-up time of 5 yr, seven patients were information was not available at the time of the lost. Five died (three had malignant disease; two radiological evaluation. The erythrocyte sedimentation carcinoma in the lungs and one in the uterus; one had rate ( ESR) was measured according to Westergren. cardiovascular disease and one committed suicide). Rheumatoid factor of IgM class was analysed with an One patient moved to another part of the country and enzyme-linked immunosorbent assay ( ELISA) [ 14]. one stopped attending due to psychiatric disease. HLA-DRB alleles were typed by restriction fragment length polymorphism analysis with sequence-specific Remission primers as previously described [15]. Twenty per cent (37 patients: 16 men/21 women) had 39 remission periods of at least 6 months. Mean Clinical course (S.D.) remission time was 20.5 (12.9) months with a Remission was defined according to the ARA criteria range of 6 48 months. The patients were in remission [3] as shown in Table I. Fatigue was not measured in this study, and remission was therefore considered TABLE II present if a patient fulfilled four of the remaining five Some clinical and laboratory features at the start of the study in criteria. Concomitant use of disease-modifying antias the median (interquartile range) 183 RA patients with ~1 yr disease duration. All values are given rheumatic drugs (DMARDs) and/or oral corticosteroids was allowed. The patients had to fulfil Active joint count (0 50) 6 (4 10) remission criteria at least at two consecutive follow-up HAQ disability index (0 3) 0.8 (0.5 1.3) visits (6 months apart). Joint damage score (0 200) 6 (3 11) (available for 151 patients) The course of the disease was regarded as relapsing ESR (mm/h) 29 (14.5 52) remitting if the patient had no active joints on examin-

1326 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 37 NO. 12 34% of their follow-up time. Mean (S.D.) disease dura- free episode within the first study year, and another tion at achievement of the first remission period was 20% within 2 yr. 36.2 (14.3) months with a range of 9 64 months. Most patients had during the course of the 5 yr Twenty-one remissions were spontaneous and 18 drug both large and small joint involvement (82% in the induced (chloroquine 8, D-penicillamine 6, aurothio- relapsing-remitting group and 86% in the persistent malate 3, and sulphasalazine 1). group). Twenty-nine patients ( 16%) had involvement Figure 1 illustrates the remission rate over time. The of hands and feet only. Nine of them had a limited number of patients moving in and out of remission disease with involvement of four or less small joints. during each 6 month period in the study is displayed. Five of the nine with limited disease, compared to 32 Figure 2 shows the starting point and length of remission of 142 with more widespread disease, had remission for each individual patient during the course of periods of 6 months or more (P < 0.02). follow-up. Table III shows a comparison of demographics and The time spent in remission constituted 7.0% of disease characteristics between patients with persistent follow-up for all patients. and relapsing-remitting disease. The only significant Cross-tabulation comparing patients with (n = 37) finding was that patients with persistent disease were and without remission periods of at least 6 months more often RF positive (P = 0.04). No clinical or (n = 146) showed that remitting patients were signifi- laboratory variable at baseline differed between the cantly less often RF positive (P = 0.01) and less often two groups. We obtained no logistic regression model carried two shared epitopes (P = 0.05). Age, gender that could predict the disease course with any accuracy. and mode of onset did not differ between the two Seventy-two per cent of the patients with persistent groups. Comparison between the clinical and labora- disease had received DMARD therapy compared to tory variables at onset displayed in Table II showed 58% of the patients with relapsing-remitting disease. that patients with remission periods had a significantly Patients with persistent disease had also received a lower active joint count (P = 0.05). Prediction of larger number of different drugs (P = 0.01). Ten remission by means of logistic regression analysis did patients with relapsing-remitting disease and 19 not yield any clinically meaningful model, only 26% of patients with persistent disease had oral corticosteroid the patients with remission periods could be correctly treatment some time during the 5 yr. This difference classified. was significant (P = 0.02). Table IV shows some clinical and laboratory features Clinical course for the 176 patients who remained at study finish, One hundred and two patients (56%) had a subdivided according to clinical course. Patients with relapsing-remitting disease course and 81 (44%) had a persistent disease fared significantly worse in all persistent disease course. Forty per cent of patients respects. They were more disabled, had more joint with relapsing-remitting disease had their first arthritis- inflammation and more erosive radiographic changes. Table V displays the relationship between remission and DMARD treatment. More non-treated patients went into remission according to either definition. DISCUSSION In this early RA cohort, the patients were followed prospectively with standardized measures. The followup period was the same for all the patients, and the drop-out rate was only 4%. Remission was defined according to the rigorous ARA criteria. However, the required observation period of 2 months was prolonged to 6 months due to our study design. Twenty per cent of our patients achieved remission some time during the 5 yr follow-up time. Remission was mostly long lasting with an average length of almost 2 yr. Onethird of the follow-up time for these patients was spent in remission. However, the remission periods constituted only 7% of the entire follow-up time for the cohort. Other authors applying similar methods to calculate remission rates have reported findings in accord with ours both in early and established RA [7, 16]. In other studies of early RA patients, remission criteria were applied at a single point in time. The highest cross-sectional rates were found in a Finnish study [4] where 27% of the patients were in ARA- defined remission at the 2 yr follow-up and 32% at FIG. 1. ARA-defined remission rates over time in originally 183 early-ra patients with a mean disease duration of ~1 yr at study start. Altogether, 37 patients achieved 39 remission periods during follow-up. The numbers of patients moving in and out of remission during each 6 month period in the study are shown. The number of patients left in the study in each period is displayed below.

EBERHARDT AND FEX: CLINICAL COURSE IN EARLY RA 1327 FIG. 2. Remission periods (n = 39) for each individual patient (n = 37) during the course of follow-up. Each line denotes the starting point and length of remission. Mean (S.D.) remission time was 20.5 (12.9) months. TABLE III TABLE IV Demographic data and disease characteristics subdivided according Clinical and laboratory features subdivided according to clinical to clinical course for all the 183 patients course for 176 patients who remained in the study at the 5 yr follow- up. Median (interquartile range) Persistent Relapsing-remitting disease disease Relapsing-remitting (n = 81) (n = 102) Persistent disease disease (n = 78) (n = 98) Age at onset 50.7 (11.7) 52 (13) [yr; mean (S.D.)] Active joint count 5 (2 8) 1 (0 4)** Gender (% women) 67 61 HAQ disability index 1.2 (0.9 1.5) 0.8 (0.3 1.2)** Single epitope 86 84 Joint damage score 53 (32 74) 27 (12 53)** (% positive) Double epitope 28 36 ESR 31 (15 55) 21 (9 31)* (% positive) *P = 0.01, **P < 0.001; Mann Whitney s test. Rheumatoid factor 78* 64 (% positive) TABLE V Disease onset 14/20/59/7 21/16/50/13 Relationship between remission and DMARD treatment during (% acute/intermediate/ follow-up insidious/palindrome) Treated Non-treated *P = 0.04 (x2 test). (n = 114) (n = 69) ARA remission n (%) 18 (16) 21 (30) the 6 yr follow-up. Harrison et al. [17] studied a No arthritis on examination n (%) 60 (53) 42 (61) community-based cohort and found that 19% of their RA cases were in remission after 2 yr. Remission was defined as no arthritis on examination and no sion was often a temporary state with new patients DMARD or steroid treatment within the previous 3 achieving remission or relapsing at any point in time. months. Two other studies reported a cross-sectional Cross-sectional remission rates do not measure this remission rate of ~7% after 3 and 7 yr, respectively longitudinal aspect and are therefore less reliable, as [5, 6]. Both studies used being symptom free as a also pointed out by Harrison et al. [17]. Comparisons definition for remission. As illustrated in Fig. 1, remis- between different studies are thus not only hampered

1328 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 37 NO. 12 by differences in defining remission and different ACKNOWLEDGEMENTS patient referral and selection, the different methods to Grants were obtained from the Medical Faculty of estimate remission rates are also rendering difficulties. Lund, the Österlund and Kock Foundations, the King Remission was more likely in seronegative patients Gustaf V 80-year Fund, and Reumatikerförbundet. having no or only a single shared epitope and fewer active joints at baseline. However, even in combination, REFERENCES these factors were not strong enough to allow accurate prediction of remission. Wolfe and Hawley [7] found 1. Hochberg MC. Predicting the prognosis of patients with rheumatoid arthritis: Is there a crystal ball? J Rheumatol that female sex, onset before the age of 60 yr and 1993;20:1265 7. early development of erosions were associated with 2. Wollheim FA. Predictors of joint damage in rheumatoid decreased proportions of remission periods. Harrison arthritis. APMIS 1996;104:81 93. et al. [17] could confirm their finding of association 3. Pinals RS, Masi AT, Larsen RA. Preliminary criteria for with gender, but not with age. Like us, they found clinical remission in rheumatoid arthritis. Arthritis association to less widespread joint involvement at Rheum 1981;24:1308 15. baseline. 4. Möttönen T, Paimela L, Ahonen J, Helve T, Hannonen In this study, we also applied a less rigorous definrheumatoid P, Leirisalo-Repo M. Outcome in patients with early arthritis treated according to the Sawtooth ition of remission, namely having no arthritis on examination at least at one follow-up visit. Defined in strategy. Arthritis Rheum 1996;39:996 1005. 5. Nissilä M, Isomäki H, Kaarela K, Kiviniemi P, Martio this way, 56% of our patients had achieved remission J, Sarna S. Prognosis of inflammatory joint diseases. at least once during the 5 yr. They were considered to A three-year follow-up study. Scand J Rheumatol have a relapsing-remitting disease pattern, in contrast 1983;12:33 8. to the 44% with chronic persistent joint inflammation. 6. Wolfe F, Ross K, Hawley D, Roberts K, Cathey MA. This classification turned out to be clinically useful, The prognosis of rheumatoid arthritis and undifferentiated identifying two groups of patients with different polyarthritis syndrome in the clinic: A study of prognosis. Our findings confirm a previous study of 1141 patients. J Rheumatol 1993;20:2005 9. a 1985 inception cohort [9]. Young et al. [ 8] have 7. Wolfe F, Hawley D. Remission in rheumatoid arthritis. reported that only 26% of their patients had a persistent J Rheumatol 1985;12:245 52. 8. Young A, Bielawska C, Corbett M, Roitt I. A prospective disease course. The lower figure is probably due to the study of early onset rheumatoid arthritis over fifteen fact that probable RA cases were also included in years: prognostic features and outcome. Clin Rheumatol their study. 1987;6(suppl. 2):12 9. It was not possible to predict the disease pattern 9. Suarez-Almazor MW, Soskolne CL, Saunders LD, with any accuracy. The only significant variable was Russel AS. Outcome in rheumatoid arthritis. A 1985 positive RF, which was weakly associated with persist- inception cohort study. J Rheumatol 1994;21:1438 46. ence. It is noteworthy that genotype did not influence 10. Eberhardt K, Rydgren LC, Petersson H, Wollheim FA. disease pattern. Only 40% of the patients in the relapsover 2 years. Rheumatol Int 1990;10:135 42. Early rheumatoid arthritis onset, course and prognosis ing-remitting group had their first evaluation without 11. Ritchie DM, Boyle JA, McInnes JM et al. Clinical signs of arthritis during the first study year and another studies with an articular index for the assessment of joint 20% during the second year. This feature therefore had tenderness in patients with rheumatoid arthritis. Q J Med small predictive power. 1968;37:393 406. Mode of onset had no prognostic value in this study. 12. Ekdahl C, Eberhardt K, Andersson SI, Svensson B. Acute onset was associated with better prognosis in Assessing disability in patients with rheumatoid the Middlesex study [18], but others [9, 19] have not arthritis. Use of a Swedish version of the Stanford Health confirmed this. Assessment Questionnaire. Scand J Rheumatol 1988; The influence of DMARD therapy on the course of 17:263 71. the disease was fairly limited, inducing ARA-defined 13. Larsen A, Dale K, Eek M. Radiographic evaluation of rheumatoid arthritis and related conditions by standard remission in only 16% and short-term arthritis-free reference films. Acta Radiol Diagn (Stockh) 1977; periods in another 37%. The frequency of drug-induced 18:481 91. ARA-defined remission is in accord with previous 14. Truedsson L, Sjöholm AG, Sturfelt G. Complement findings [ 7]. Patients not treated with DMARDs had activating rheumatoid factors in rheumatoid arthritis higher remission rates, probably reflecting a less severe studied by haemolysis in gel: relation to antibody class disease. Any conclusions regarding effects of treatment and response to treatment with podophyllotoxin deriv- vs non-treatment can therefore not be drawn. ates. Clin Exp Rheumatol 1985;3:29 37. To conclude, in this early RA cohort, 20% of 15. Eberhardt K, Fex E, Johnson U, Wollheim FA. the patients achieved ARA-defined remission periods Associations of HLA-DRB and DQB genes with two and five year outcome in rheumatoid arthritis. Ann of 6 months or more. Average length of remission Rheum Dis 1996;55:34 9. was almost 2 yr. If a wider definition of remission was 16. Prevoo MLL, van Gestel AM, van T Hof MA, van applied, almost 60% had a relapsing-remitting disease Rijswijk MH, van de Putte LBA, van Riel PLCM. course. These patients had better outcome after 5 yr Remission in a prospective study of patients with rheumatoid than patients with chronic persistent disease. arthritis. American Rheumatism Association prelim-

EBERHARDT AND FEX: CLINICAL COURSE IN EARLY RA 1329 inary remission criteria in relation to the disease activity 18. Fleming A, Crown JM, Corbett M. Prognostic value of score. Br J Rheumatol 1996;35:1101 5. early features in rheumatoid arthritis. Br Med J 17. Harrison BJ, Symmons DPM, Brennan P, Barrett EM, 1976;1:1243 5. Silman AJ. Natural remission in inflammatory polyarthritis: 19. Luukkainen R, Isomäki H, Kajander A. Prognostic value issues of definition and prediction. Br J Rheumatol 1996;35:1096 100. of type of onset of rheumatoid arthritis. Ann Rheum Dis 1983;42:274 5.