Background Cases of progressive multifocal leukoencephalopathy (PML), a rare but serious disease, have been reported in patients with rheumatoid arthritis (RA) in association with biological therapy, but little is known about the incidence of PML in patients with RA in the absence of treatment exposure.
Objective To estimate the incidence rate of PML in patients with RA compared with the general population, with and without exposure to biological agents.
Methods Patients with adult onset RA, exposure to biological agents and a diagnosis of PML from 1999 through 2009 were identified from national registries and linked using each Swedish resident's unique personal identification number. General population comparators matched on age, sex and county were also identified. Crude and age- and sex-standardised incidence rates (cases per 100 000 person-years) were calculated with 95% CI.
Results 66 278 patients with RA and 286 949 general population comparators were included in the study. The incidence rate of PML in the overall RA population was 1.0 (95% CI 0.3 to 2.5) compared with 0.3 (95% CI 0.1 to 0.6) in the general population. The difference in incidence rate was 0.7 (95% CI −0.3 to 17). Among all patients exposed to biological agents, only one patient was diagnosed with PML.
Conclusion Data from this national population-based cohort study suggest that patients with RA may have an increased rate of PML compared with the general population.
Statistics from Altmetric.com
Progressive multifocal leukoencephalopathy (PML) is a rare but serious disease caused by the reactivation of latent JC virus which typically occurs in immunosuppressed patients.1 Cases of PML have been reported in patients with rheumatoid arthritis (RA) in association with exposure to biological therapy, and specifically to rituximab.2–8 However, because of the rarity of PML, little is known about the incidence of PML in patients with RA, hampering the ability to estimate treatment-related increases in risk.
To put the signal of PML following biological treatment into context, we used population-based Swedish registers to estimate the incidence of PML in the general population and in the RA population with and without exposure to biological agents.
Based on register linkages as described elsewhere,9 ,10 we identified a national population-based RA cohort comprising virtually all patients with RA in Sweden, with RA defined as at least one visit listing RA in the Swedish National Patient Register (inpatient care 1969–2009 and outpatient non-primary care 2001–2009) and in the Swedish Rheumatology Quality Register (total n=66 278). Patients with any visit listing juvenile idiopathic arthritis, psoriatic arthritis, spondylarthritis, ankylosing spondylitis or systemic lupus erythematosus were excluded. In a sensitivity analysis, RA was defined as having ≥2 visits listing RA, with at least one at a department of Rheumatology or Internal Medicine. For each patient with RA, five general population comparators matched on age, sex and county were identified from the Swedish National Population Register (total n=286 949).
Data on exposure to biological agents were retrieved through linkage to the national population-based Swedish Rheumatology Biologics Register (ARTIS 1999–2009)11 and to the national Swedish Prescribed Drug Register (PDR; all drug dispensations except for infusion drugs and in-hospital use in Sweden 2005–2009). Individuals were considered to have been exposed to biological agents starting on the date of the first biological agent prescribed (ARTIS) or dispensed (PDR) until the end of follow-up.
Follow-up and outcome assessment
Start of follow-up began on 1 January 1999 or the first date of RA, if later. End of follow-up was the first of PML diagnosis, death, emigration or 31 December 2009.
Cases of PML (defined as a hospitalisation discharge or outpatient specialist visit listing ICD10=A81.2) were identified in the National Patient Register. The PML discharge note was evaluated by a consultant neurologist. Information from a previous review of discharge files (eight cases) from a similar sample of the general population was used to estimate whether the PML diagnoses in the general population were likely to be true cases. Confirmed cases all had a compatible clinical presentation, MRI scans and were PCR-positive for JC virus either from cerebrospinal fluid or brain biopsy. In these cases, alternative diagnoses were further sought but ruled out/not found.
Crude incidence rates (IR) of PML (cases per 100 000 person-years) in the general population, the overall RA population and in biological-naïve RA and biological-exposed RA populations were calculated. Incidence rate differences (IRD) and 95% CIs were also calculated. Age- and sex-standardised IRs were calculated using the overall RA population as the standard with CIs calculated using Fay's method for weighted sums of Poisson variables.12 Analyses were conducted using R V.2.13.0 (Epitools, V.0.5-6, www.medepi.com) and SAS V.9.2.
Tables 1 and 2 show the distribution of individuals, observed cases and person-time by exposure. Using the register-based diagnosis of PML, the estimated IR per 100 000 person-years in the overall RA population (1.0, 95% CI 0.3 to 2.5, n=4) was higher than that estimated in the general population (0.3, 95% CI 0.1 to 0.6, n=5; table 1), although these rates were not statistically significantly different from each other (IRD 0.7, 95% CI −0.3 to 17). Based on the single case of PML exposed to a biological agent, the IR was 2.3 (95% CI 0.1 to 71) in patients who were exposed to biological agents and 0.8 (95% CI 0.2 to 2.5) in those not exposed to biological agents (table 1); these were not statistically significantly different. IRs were similar in analyses restricted to patients with the more strict RA definition (data not shown).
Chart review confirmed two of the four cases with a register-based diagnosis of PML in the RA population. Of these two cases, one was exposed to rituximab and had received radiation for a malignancy several years earlier but had no other known risk factor for PML and the other had been exposed to chlorambucil in a relevant time frame. The other two RA patients (both with long-standing RA, leukoencephalopathy and methotrexate treatment but otherwise no obvious risk factors for PML) had insufficient evidence to confirm the PML diagnosis. With respect to the validity of the PML diagnosis in the general population comparator, a previous but similar validation indicated that three of eight cases could be confirmed as PML. Each of these three cases had a typical risk factor (two malignant lymphomas under treatment and one with HIV). Thus, based on the assumption that 37.5% (3/8) of the register-based cases in the general population comparator represented PML, the estimated IR in the general population would be 0.1. Including only confirmed PML cases, the IR was 0.5 in the overall RA cohort and 0.3 in those not exposed to biological agents.
In this large population-based national investigation of PML in patients with RA, the estimated rate of PML was low but was higher in the RA population than in the general population, even in the absence of biological therapy (although the difference was not significant). All cases that could be confirmed occurred in individuals with a history of immunosuppressive therapy/disease.
Even with the large number of subjects and person-time contributed to this study, there were very few cases of PML which caused uncertainty in the IR estimates as reflected by wide CIs. Age- and sex-standardised IRs allow us to compare rates between different exposure groups but, because of the small number of cases, we abstained from more sophisticated analyses accounting for potential confounders. We assume in our assessment of exposure that, once exposed, patients are always exposed until the end of follow-up. This overestimates exposure time and may not represent true exposure. Because of recent attention to PML as a complication of RA, patients with RA may be more likely to be diagnosed with PML. Three of the RA patients were diagnosed with PML before 2005, which predates awareness of this issue, and none were known to the authors who initiated the study. Therefore, we do not believe that detection bias greatly affected our study.
The rate of PML in our RA population (1 per 100 000) is of the same order of magnitude as that reported by Amend et al (0 per 100 000 person-years; 95% CI 0.0 to 2.2).13 Amend et al used a health insurance database in which participants may only contribute short periods of follow-up and patients aged ≥65 years were under-represented so cases could have been missed.13 Out of almost 300 million discharges in the US Nationwide Inpatient Sample Database, Molloy et al reported that 24 patients hospitalised for RA were diagnosed with PML from 1998 through 2005 (0.4 cases per 100 000 discharges).14 Neither study examined cases by exposure to biological agents. A recent case series description estimated a PML incidence of 1 in 25 000 rituximab-exposed patients, but this study was limited by relying on spontaneous reports and the lack of an unexposed comparison group.8 All three of these studies excluded patients with a history of PML risk factors (HIV/AIDS, cancer or organ transplantation). If we included only the cases of PML that were not explained by a known non-RA risk factor, there were two register-based cases in the RA population (IR 0.5, 95% CI 0.1 to 1.8) and none in the general non-RA population (IR 0, 95% CI 0 to 0.2). These IRs are, however, likely to be underestimates because risk factors for PML were not excluded from the denominator.
With respect to biological therapies of RA, estimates of the rate of PML have been difficult to calculate, let alone to compare different treatment exposures, owing to the small numbers. Our estimate is based on one case (rituximab) and therefore should be taken with caution. Indeed, adding a hypothetical case of PML to either of the other biological exposure groups would have resulted in crude IR estimates for these ranging from 5 to 400 per 100 000 person-years, which would be numerically but not statistically different from the observed IR for rituximab (67 per 100 000, 95% CI 1.7 to 370).
The primary strength of this study lies in its use of nationwide population-based data that include thousands of person-years of observation. Because of the low power, these rates are rough estimates and should be treated as such. Nevertheless, they provide an estimate of the order of magnitude of PML risk in RA that could be used for contextualisation of reported cases and can be compared with other known infection rates in patients with RA treated with biological agents, such as hospitalisation for any serious infection (1 to 5 per 100 person-years)15 or tuberculosis (1 per 1000 person-years).16 ,17
Dr Magnus Andersson, Department of Neurology at Karolinska University Hospital, Stockholm, Sweden, assisted in the validation of PML diagnoses.
The ARTIS Study Group E Baecklund, L Coster, C Forsblad-Elia, N Feltelius, P Geborek, L Jacobsson, L Klareskog, S Lindblad, S Rantapaa-Dahlqvist, T Saxne and R van Vollenhoven.
Funding Financial support for this study was obtained from the Swedish Research Council, the Swedish Cancer Society and from Stockholm County Council, from the Swedish Foundation for Strategic Research and from the Swedish COMBINE inflammation research consortium. The ARTIS register has agreements with Wyeth-Ayerst, Schering-Plough, Abbott Immunology, Roche, and Bristol Myer Squibb. The authors were in charge of and solely responsible for all data collection, analysis, interpretation and writing of the manuscript, without any constraints exerted from the agencies or companies that provided financial support for the study or for ARTIS. The South Swedish Anti-TNF Register has received funding from King Gustav V, Österlund and Kock Foundations, Reumatikerförbundet and from the Medical Faculty, Lund University. Financial support for the Early RA register was provided by the Swedish National Board of Health and Welfare.
Contributors JA formed the hypothesis and EVA and JA performed the analysis. EVA, RFvV and JA interpreted the results and drafted the paper. All authors contributed to editing the draft for content and approved the final version of the paper. All authors had full access to all of the data (including statistical reports and tables) in the study and can take responsibility for the integrity of the data and the accuracy of the data analysis.
Competing interests None.
Patient consent Not obtained.
Ethics approval Ethics approval was obtained from the Karolinska Institutet.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.