Article Text
Abstract
Background Benefits and risks of corticosteroid treatment in rheumatoid arthritis (RA) are debated. Patients with RA are at increased risk of malignant lymphomas. In a large case–control study of risk factors for lymphoma in RA, it was recently reported that steroid treatment was associated with decreased lymphoma risk.
Objective To further assess the nature of the association between steroid treatment in RA and the risk of lymphoma.
Methods In a cohort of 74 651 patients with RA, 378 patients with lymphoma and 378 matched RA controls were identified, and information on inflammatory activity and different aspects of steroid treatment (duration, therapeutic strategy and mode of administration) abstracted from their medical records. Lymphomas were reclassified (WHO classification) and examined for Epstein–Barr virus. Relative risks were assessed as adjusted odds ratios (ORs) through conditional logistic regression.
Results A total duration of oral steroid treatment of <2 years was not associated with lymphoma risk (OR=0.87; 95% CI 0.51 to 1.5), whereas total treatment >2 years was associated with a lower lymphoma risk (OR=0.43; 95% CI 0.26 to 0.72). RA duration at the initiation of oral steroids did not affect lymphoma risk. Intra-articular steroids were associated with a reduced lymphoma risk, but only when used as swift flare treatment (OR=0.22; 95% CI 0.13 to 0.37). Analyses by lymphoma subtype showed a reduced risk of diffuse large B-cell lymphoma (crude OR=0.59; 95% CI 0.37 to 0.94).
Conclusion In this RA population, use of steroids was associated with reduced lymphoma risk. Whether this association is a generic effect of steroids or specific to the studied population remains unknown.
Statistics from Altmetric.com
Introduction
Because of their potent and rapid anti-inflammatory effects, treatment with corticosteroids (later referred to as steroids) represents a well-established, and increasingly accepted, treatment in inflammatory conditions such as rheumatoid arthritis (RA).1,–,3 Several studies have tried to assess the association between steroid treatment and lymphoma risk, with contradictory results.4,–,11 In part, this is owing to the difficulty in separating the effect of treatment from the risk associated with the treated condition itself, since patients with either of several different inflammatory conditions are at increased risk of malignant lymphomas at baseline.12 13
In RA, one of the best-studied conditions in this respect, cohort studies indicate an average doubling of the lymphoma risk.14,–,17 Data from our group, moreover, indicate a strong association between disease severity and lymphoma risk,18 diffuse large B-cell lymphoma (DLBCL) in particular,19 underlining the above-mentioned potential for confounding by indication of any observed association between steroid treatment. By contrast and somewhat unexpectedly, in the same study18 we observed a markedly reduced risk of lymphoma associated with steroid treatment, which remained after adjustment for RA disease severity and use of disease-modifying antirheumatic drugs (DMARDs). In this study, we set out to further characterise and assess this strong and inverse association between lymphoma risk and steroid treatment observed in our previous study, with particular emphasis on lymphoma risks in relation to treatment strategy (flare or maintenance treatment), timing, duration and lymphoma characteristics. To do this, we used data from a case–control study of 378 patients with RA-associated lymphomas and 378 matched RA controls nested within a population-based cohort of 74 651 Swedish patients with RA.
Subjects and methods
The design, study population, validation of RA and lymphoma, and exposure assessment has been described in detail elsewhere.18
Source cohort
In the population-based, nationwide and virtually complete Swedish Hospital Discharge Register, we identified all 74 651 patients discharged from hospital with a discharge diagnosis listing RA at least once between 1 January 1964 and 31 December 1994.
Cases and controls
Through linkage of this cohort to the nationwide and nearly complete Swedish Cancer Register20 from 1 January 1964 to 31 December 1995, we identified 424 patients in the cohort who were diagnosed with a malignant lymphoma as their first primary cancer after the first discharge listing RA. Linkage through 1995 provided a minimum of 1-year follow-up from first RA discharge. From the same RA cohort (n=74 651) we randomly selected three patients with RA as potential controls, matched for sex, year of birth, year of first RA discharge and county of residence. Controls had to be alive and without any registered cancer at the time of the lymphoma diagnosis of their corresponding case. Among these potential controls we included the first of the three controls whose medical record could be identified and who fulfilled the American College of Rheumatology criteria for RA.21
Validation of RA and lymphoma diagnoses
Medical records of cases and controls were reviewed in order to confirm the RA diagnosis. Patients who did not fulfil the American College of Rheumatology criteria for RA or who did not have erosive polyarthritis with a clinical diagnosis of RA (n=27; 6%) or whose records could not be identified (n=11; 3%) were excluded, leaving 386 eligible cases. The same exclusion/inclusion criteria were used for the controls.18
The paraffin-embedded lymphoma tissues were reviewed and reclassified according to the WHO classification.22 Epstein–Barr virus (EBV) in the lymphomas was searched for using EBV-encoded RNA (EBER) in situ hybridisation.19 Of the 386 eligible RA lymphoma cases, eight were excluded as the lymphoma diagnosis could not be confirmed. In another 35 of the cases, reclassification of the specimens was not possible, although the lymphoma diagnosis could be confirmed. The study thus consisted of 378 cases (343 with reclassified lymphoma specimen) and their 378 matched controls (table 1).
Information on RA characteristics
For all cases and controls, medical records of all hospital stays and all outpatient care were collected. Detailed information on disease characteristics and treatment was abstracted by one of us (EB) from each person's medical records blinded for the case–control status.
Controls were followed up from onset of RA until the diagnosis of lymphoma of the corresponding case.
To assess RA disease activity, we abstracted information on swollen and tender joint counts, erythrocyte sedimentation rates and physicians' global assessments as recorded in the medical files. Based on these variables, RA disease activity was scored as inactive, low, medium or high. Cumulative disease activity for the whole RA period was calculated as the area under the curve (AUC) for the duration in months of the four different levels of disease activity. This score is described in detail elsewhere.18
Information on steroid treatment
Oral treatment was defined as four or more consecutive weeks on a specified drug (eg, a DMARD or oral steroid). The duration of each treatment was assessed in months. The available data did not allow exact estimation of cumulated dose. Treatment with intra-articular steroids was recorded if ever used. Treated subjects were categorised into two groups: (A) those who received intra-articular steroids within 1 month of onset of arthritis in more than 50% of documented flares and (B) the remaining patients—that is, those who, in more than 50% of documented flares, had more than 1 month between arthritis onset and any steroid injection. The time of onset of arthritis was based on the patient's information to the doctor as documented in the medical record. The presence of arthritis had to be confirmed at a physical examination. Typically the patients in group A had quick access to a clinic/doctor where intra-articular steroids were routinely used and were offered an ‘injection appointment’ often within 1–2 weeks. The patients in group B typically described ‘my knee has been swollen for the past 6 months’ at a regular visit. Some patients in that situation were given an injection but it was more common to be prescribed non-steroidal anti-inflammatory drugs or to be referred to an doctor with experience of intra-articular injections.
Information on treatment with intramuscular steroids and adrenocorticotrophic hormone at any time during the RA course was also extracted. The patients were followed up for survival until 31 December 2005.
The study was approved by the ethics committee at Uppsala University, Uppsala, Sweden.
Statistics
The relative risk of lymphoma associated with steroid treatment was assessed using odds ratios (ORs) derived from conditional logistic regression (PROC PHREG; SAS v.9), taking the matching factors into account.
The association was assessed overall, adjusted and stratified by the AUC of cumulative disease activity and DMARD use. We also assessed the duration of steroid treatment, timing of first steroid treatment, mode of administration and intra-articular treatment type (A or B, as described earlier). Duration of oral steroid treatment was categorised as more or less than 2 years. Timing of the first oral steroid treatment after RA onset was categorised as more or less than 5 years. (Alternative categorisations resulted in similar trends in risk.) Analyses were performed for lymphomas overall and separately for the different lymphoma subtypes as well as for EBV-positive lymphomas. ORs were adjusted for RA disease severity as quartiles of AUC, DMARD use (never/ever) and for the use of other steroids (intra-articular and oral, respectively). Separate analyses adjusting for smoking, diabetes and large joint involvement were also performed.
Results
Oral steroid treatment
Five of the 378 cases and none of the controls lacked information about oral steroid treatment. In one case there was no information about time of first start of treatment. A total of 183 (49%) of 373 cases and 217 (57%) of the controls had been treated with oral steroids (table 1).
Fifty-seven per cent of the steroid-treated cases were female compared with 54% of the steroid-treated controls. Eighty-one per cent of the steroid-treated cases and 85% of the steroid-treated controls were rheumatoid factor positive. Among the cases, the mean age at RA onset was similar for cases treated (50 years, range 16–83 years) or not (51 years, range 16–80 years) with oral steroids. The mean duration of RA until lymphoma diagnosis was also similar for cases treated (21 years, range 2–54 years) or not (19 years, range 1–55 years) with oral steroids. Overall, treatment with oral steroids was associated with a 30% reduced risk of lymphoma (crude OR = 0.69; 95% CI 0.51 to 0.94), which remained after adjustment for DMARD treatment and disease activity (OR = 0.58; 95% CI 0.38 to 0.90) as previously reported18 (table 2). Adjusting also for the use of intra-articular steroids did not change the result (OR = 0.63; 95% CI 0.40 to 0.98), nor did further adjustment for diabetes, smoking or large-joint involvement.
Treatment duration with oral steroids up to 2 years was not associated with any reduced lymphoma risk (adjusted OR = 0.87; 95% CI 0.51 to 1.5). By contrast, a total of more than 2 years of steroid treatment was associated with a markedly reduced lymphoma risk (adjusted OR = 0.43; 95% CI 0.26 to 0.72) (table 2).
The duration of RA disease at the start of the first steroid treatment ranged from 2 months to 35 years after RA onset. After adjustment for disease activity and DMARD treatment the relative risk of lymphoma was similarly reduced in patients starting oral steroids within the first 5 years after RA onset compared with those starting treatment later in their RA course (adjusted OR in both groups = 0.58; 95% CI 0.35 to 0.97) (table 2). The association with use of oral steroids was largely similar in subsets defined by calendar period of lymphoma diagnosis (data not shown).
Oral steroid treatment in relation to disease activity
Overall, cases had a considerably higher cumulative disease activity than controls. Of all the patients whose medical records included enough information to assess cumulative RA disease activity, 74% (274/372) belonged to the highest quartile of cumulative disease activity (as defined by the distribution among all eligible controls). These distributions were not markedly different for cases and controls treated with steroids (77% and 30%, respectively). Within each quartile of cumulative RA disease activity, steroid treatment was associated with a lower lymphoma risk. For instance, oral steroid treatment in the highest quartile of disease activity was associated with an OR of 12 (95% CI 4.9 to 28), compared with an OR of 30 (95% CI 11 to 76) for the same quartile of disease activity but in the absence of oral steroid treatment (figure 1).
Oral steroid treatment in relation to lymphoma characteristics
Analyses by lymphoma subtype suggested a significantly reduced risk of DLBCL (crude OR = 0.59; 95% CI 0.37 to 0.94) and a non-significantly reduced risk of follicular lymphoma (crude OR = 0.71; CI 95% 0.23 to 2.3) following ever treatment with oral steroids, whereas the crude OR for all other subtypes combined was 0.81 (95% CI 0.50 to 1.3).
Ann Arbor stage at diagnosis of the lymphoma was similar for cases that had or had not been treated with oral steroids for RA (table 3).At the time of lymphoma diagnosis 84/378 (22%) cases and 125 (33%) controls were treated with oral steroids. The symptoms leading to lymphoma diagnosis, as well as the occurrence of B-symptoms (fever, weight loss, night sweats) were similar for cases receiving steroids and not receiving steroids, respectively, at the time of lymphoma diagnosis. Ever treatment with oral steroids was not associated with EBV-positive lymphomas (crude OR = 1.1, 95% CI 0.41 to 3.2). In this study, survival after lymphoma diagnosis was poor, but similar for cases ever treated (or not) with oral steroids (data not shown).
Intra-articular steroids
A total of 168 (44%) cases and 240 (63%) controls received intra-articular steroids. Of these, 98 (58%) of the cases and 159 (66%) of the controls had also been treated with oral steroids. Overall, treatment with intra-articular steroids was associated with a decreased risk of lymphoma (OR = 0.35; 95% CI 0.22 to 0.56, adjusted for DMARD treatment and disease activity). Adjusting also for the use of oral steroids did not change the results (OR = 0.36; 95% CI 0.23 to 0.58).
The decreased lymphoma risk associated with intra-articular steroids was, however, restricted to those patients (‘group A’) who received intra-articular steroids within 1 month after onset of arthritis in ≥50% of flares (adjusted OR = 0.22; 95% CI 0.13 to 0.37). By contrast, the remaining patients (‘group B’) who had received intra-articular steroids after more than 4 weeks delay in more than 50% of the reported flares, were at increased lymphoma risk (adjusted OR = 2.6; 95% CI 1.1 to 6.2) (table 2).
Other types of steroid treatment
Other types of steroid treatment were uncommon. Two (0.5%) of the cases and nine (2%) of the controls had received intramuscular steroids any time during their RA course. Intramuscular steroid use entailed an adjusted OR of 0.34 (95% CI 0.07 to 1.7), adjusted for other steroid use, DMARD treatment and quartiles of AUC of disease activity. In the early phase of the study (up to ~1980), 27 cases (7%) and 26 controls (7%) had received treatment with adrenocorticotrophic hormone. This treatment was not associated with lymphoma risk (adjusted OR = 0.84; 95% CI 0.37 to 1.9).
Discussion
In this extended assessment of lymphoma risk in relation to treatment with steroids in the context of RA we made the following observations: (a) treatment with steroids, oral as well as intra-articular, was associated with reduced lymphoma risk; (b) for oral steroids, total duration of time receiving steroid treatment was important; (c) for intra-articular steroids, the strong inverse association with lymphoma risk was restricted to those patients in whom intra-articular steroids were consistently used as flare treatment; (d) for steroids overall, the strongest association was observed for DLBCL—that is, the lymphoma subtype previously shown to be most strongly associated with RA.18 19
Steroids have been widely used against RA for more than 50 years, and recently experienced somewhat of a revival, particularly in the treatment of early RA.2 3 Despite their extensive use, a comprehensive understanding of their mechanisms of action in RA is still lacking.23 24 Existing data regarding steroid treatment and lymphoma risk (reviewed by Bernatsky et al10) have been inconclusive. Some studies have reported an increased risk,7 9 25 whereas others have not.5 6 8 11 One explanation for the heterogeneous results may be the varying degrees of confounding by indication by the underlying condition treated with steroids. When the association between steroids and lymphomas has been assessed in patients with RA, the results have indicated no increased risk6 as well as increased risks,11 although neither of these studies have taken RA disease severity, treatment or accumulated steroid exposure/route of administration into account. Interestingly, in the study by Smedby et al, in which self-reported steroid intake was positively associated with lymphoma risk,11 this increased risk was confined to patients reporting taking steroids against RA rather than against other indications.
One interpretation for the reduced lymphoma risk with steroids in our study might be that steroid treatment decreases inflammatory activity (the hitherto strongest identified risk factor for RA lymphomas18) and that this reduction is more efficiently reached among the cases than among the controls. This interpretation is supported by our observation of a particularly lowered lymphoma risk among those patients in whom steroids were consistently used as swift flare treatment, but at the same time contradicted by the fact that, at the time of lymphoma diagnosis, steroid-treated cases had higher accrued inflammation than steroid-treated controls, and the fact that adjustment for cumulative disease activity did not attenuate the association.
A second possibility is that the observed association with the use of steroids, swift access to intra-articular steroids, in particular, in our study population is reflective of comparatively alert overall rheumatological management rather than specific properties of steroids.
A third possibility might be that the inverse association between the use of steroids and lymphoma risk in these patients with RA with substantial disease activity (cases and controls in our study had not received contemporary early aggressive treatment) is mediated through other mechanisms than through their anti-inflammatory properties. Steroids are known to induce apoptosis in lymphatic cells26 27 and are therefore used frequently in high doses in various lymphoma/leukaemia regimens.28 Whether this or a similar mechanism would apply also to the typically lower but longer steroid treatments used in RA is unclear, although successfully administered intra-articular steroids result in high synovial exposure to steroids that in turn might ‘control’ emerging clonal B-cell populations. This remains, however, a speculation. Importantly, and irrespective of the mechanism behind the observed association, our results indicate that steroid treatment does not alter the clinical presentation and state at diagnosis of lymphoma in subjects with RA.
Our study has some strengths and limitations. By using original medical records, data from bio-banks and national mandatory registers (hospital discharge and cancer registers), we feel confident that we have identified all lymphomas in the source cohort, and have minimised recall/information bias. Furthermore, through reclassification of the original lymphoma specimens, we had the possibility to assess risks also relating to the current classification scheme used for lymphomas.22 Contemporary RA is treated with combinations of DMARDs and/or biological agents, which makes it difficult to assess the intrinsic effects of a single drug, such as steroids. By contrast, our study population had typically not received continuous immunosuppressive treatment and thus presented a wide range of exposures, both with respect to inflammatory activity and RA treatment. On the other hand, this may limit the generalisability of our findings, as it remains an open question whether the observed reduction in lymphoma risk associated with steroids use applies also to patient populations with better inflammatory control, and perhaps a lower risk of lymphoma than the one under study.
Although we could identify dates of steroid treatment and injections through the medical files, this was insufficient information to calculate the accumulated dose, or current standard indices of disease activity such as DAS28 (disease activity score including 28 joint counts).29 As a consequence, our measure of accrued inflammation is not directly translatable to measures of accrued DAS28.
In conclusion, in a population of patients with RA up until the mid-1990s, steroid treatment was not a risk factor for malignant lymphoma but was instead associated with a decreased lymphoma risk. It is now important to understand whether the observed association is generic, or specific to the population under study.
References
Footnotes
-
Funding Financial support was obtained from the Swedish Cancer Society, the Swedish Rheumatism Society, the Lions Cancer Research Foundation of Uppsala and the King Gustav V Foundation.
-
Competing interests None.
-
Ethics approval This study was conducted with the approval of the ethics committee at Uppsala University, Uppsala, Sweden.
-
Provenance and peer review Not commissioned; externally peer reviewed.