Article Text

Drug-induced systemic lupus: revisiting the ever-changing spectrum of the disease using the WHO pharmacovigilance database
  1. Laurent Arnaud1,2,
  2. Philippe Mertz1,2,
  3. Pierre-Edouard Gavand2,3,
  4. Thierry Martin2,3,
  5. François Chasset4,
  6. Martine Tebacher-Alt5,
  7. Aude Lambert5,
  8. Charlotte Muller5,
  9. Jean Sibilia1,2,
  10. Bénédicte Lebrun-Vignes6,
  11. Joe-Elie Salem6
  1. 1 Service de rhumatologie, Hôpitaux Universitaires de Strasbourg, Laboratoire d'Immuno Rhumatologie Moléculaire, INSERM UMR_S1109, Strasbourg, France
  2. 2 Centre National de Références des Maladies Systémiques et Autoimmunes Rares Est Sud-Ouest (RESO), Université de Strasbourg, Strasbourg, France
  3. 3 Service de Médecine Interne et Immunologie Clinique, Hôpitaux Universitaires de Strasbourg, Strasbourg, France
  4. 4 Assistance Publique Hôpitaux de Paris (AP-HP), Service de Dermatologie et d'Allergologie, Hôpital Tenon, Paris, France
  5. 5 Service de pharmacovigilance, Hôpitaux Universitaires de Strasbourg, Université de Strasbourg, Strasbourg, France
  6. 6 AP-HP, Pitié-Salpêtrière Hospital, Department of Pharmacology, Regional Pharmacovigilance Center, INSERM, Sorbonne Universités, Paris, France
  1. Correspondence to Professor Laurent Arnaud, Department of rheumatology, Hôpitaux Universitaires de Strasbourg & Centre National de Références des Maladies Systémiques Rares (RESO), Strasbourg, France; laurent.arnaud{at}chru-strabourg.fr

Abstract

Objective Drug-induced lupus (DIL) is an idiosyncratic side effect of treatments in which symptoms overlap with those of systemic lupus erythematosus (SLE). The spectrum of DIL constantly evolves with that of the pharmacopoeia. Here, we used VigiBase, the WHO global individual case safety reports (ICSRs) database, to identify the main drugs associated with DIL.

Methods We analysed all ICSRs classified as ‘systemic lupus erythematosus’ according to the Medical Dictionary for Drug Regulatory Activities term (preferred term level) in VigiBase. The drugs considered in the analysis were those not used to treat SLE, with a positive lower end of the 95% credibility interval for the information component (IC025) ≥0, an indicator value for disproportionate Bayesian reporting.

Results A total of 12 166 DIL ICSRs were identified using VigiBase. From those, 8163 ICSRs reporting on 118 suspected drugs with IC025 ≥0 were extracted. The median age at DIL onset was 49 years and the female to male sex ratio was 4.3. The median delay between start of suspected treatment and DIL occurrence was 172 days. DIL was reported as serious adverse event in 55.4%. Among the 118 suspected drugs, 42 had not been previously reported in association with DIL. The drugs associated with the highest number of DIL cases were infliximab, adalimumab, etanercept, procainamide and hydralazine.

Conclusion This study enables the identification of 118 drugs associated with DIL. The list of suspected drugs may prove useful to physicians when confronted with potential DIL cases.

Trial registration number NCT03480529.

  • drug-induced lupus
  • systemic lupus erythematosus
  • adverse-drug reaction
  • pharmacovigilance

Statistics from Altmetric.com

Key messages

What is already known about this subject?

  • The spectrum of drug-induced lupus constantly evolves with that of the pharmacopoeia.

What does the study add?

  • This study enables the identification of 118 drugs associated with drug-induced lupus, among which 42 have not been previously reported as linked.

How might this impact on clinical practice or future developments?

  • The list of suspected drugs may improve the recognition of this constantly evolving disease.

Introduction

Since a first report in 1945, more than 90 drugs have been associated with drug-induced lupus (DIL), an idiosyncratic side effect of treatments in which symptoms overlap with those of systemic lupus erythematosus (SLE). It has been estimated that up to 10% of systemic SLE cases are drug-induced, which approximates to 15 000–30 000 cases per year in the USA.1 2 There are no definite criteria for DIL, but most authors agree on the following definition: (1) sufficient and continuing exposure to the drug, (2) at least one symptom compatible with SLE, (3) no history suggestive of SLE before starting the drug and (4) resolution of symptoms within weeks or months after discontinuation of the putative drug.1 The pathogenesis of DIL remains largely unknown, but appears to involve both genetic predispositions3 as well as interindividual variations in drug metabolism.4 The five main classes of drugs that can be associated with DIL are (1) antiarrhythmic drugs such as procainamide and quinidine, (2) antihypertensive agents such as hydralazine, captopril or acebutolol, (3) antimicrobials agents such as minocycline or isoniazid, (4) anticonvulsants such as carbamazepine or phenytoin, and (5) immunomodulators such as interferon alpha or more recently antitumour necrosis factor (anti-TNF) agents. Previous studies have shown that the risk of DIL varies across different medications. The drugs at the highest risk5 are procainamide (with a 20% incidence per year) and hydralazine (with a 5%–8% risk per year of treatment), but these drugs are rarely used nowadays. Also, the strength of available evidence for DIL is not the same for each drug.5 DIL has similarities to spontaneous SLE, but there are some differences in clinical and immunological features as well as in the frequency of those features. Distinguishing DIL from SLE is important because the prognosis of drug-induced lupus erythematosus is usually good when the drug is withdrawn.5 Moreover, patients with DIL typically present with arthralgia or arthritis, myalgia, serositis, fever and rash, while renal and neurological involvements are less common than in SLE.6 The immune profile is also different between DIL and SLE. Approximately 90% of DIL cases have positive anti-nuclear antibodies (ANAs in a homogeneous pattern, but antihistone antibodies are positive in more than 75% of patients with DIL versus approximately 20%–83% in SLE,7–9 although the antihistone antibodies in DIL primarily target the histones H2A-H2B, while the H1 and H2B subunits are targeted in idiopathic SLE.10 Antidouble-stranded DNA antibodies are rarely observed in DIL except when induced by interferon alpha and anti-TNF.11

One striking observation is that the relative incidence of DIL from various treatments will depend on the extent to which those are prescribed. Most drugs described as strongly linked to DIL are much less prescribed nowadays. The first case of DIL was reported in 1945 with sulfadiazine, while hydralazine DIL was first reported in 1953. Since then, pharmacopoeia has strongly evolved, and one could hypothesise that so has the spectrum of drugs that can induce DIL. Here, we used VigiBase, the WHO global individual case safety reports (ICSRs) database, which contains reports of suspected adverse drug reactions (ADRs) collected by national drug authorities in over 130 countries, to describe the main drugs associated with DIL, as well as the detailed characteristics of these cases. This detailed analysis of a large case safety report database offers a unique opportunity to improve our knowledge of this ever-changing condition.-

Methods

Data source

The study is based on ADRs reported within VigiBase, the WHO global deduplicated ICSR database.12 These reports originate from more than 130 country members of the WHO Programme for International Drug Monitoring. VigiBase13 contains more than 16 million deduplicated ICSRs submitted by national pharmacovigilance centres since 1967. These reports originate from different sources such as healthcare professionals, patients and pharmaceutical companies. The data used in this study come from a variety of sources. The likelihood of a causal relationship is not the same in all reports. The information does not represent the opinion of the WHO.

Data assessment

This observational retrospective study (Monitoring the IMmUological TOXicity of Drugs (MIMUTOX)) included all ADRs reported as ‘systemic lupus erythematosus’ according to the Medical Dictionary for Drug Regulatory Activities (MedDRA) term (preferred term [PT]) level) in VigiBase, between 1967 and 19 March 2018. The drugs considered in the analysis were those notified as suspected treatments. Due to an expected coreporting (drugs used to treat SLE), the following drugs were not considered in the final drug by drug analysis: prednisone, corticosteroids, belimumab, mycophenolic acid, azathioprine, rituximab and hydroxychloroquine. Each report contains general administrative information (country of origin, date of reporting, reporter qualification), patient characteristics (sex, age), drugs (drug characterisation [coded according to the Anatomical Therapeutic Chemical classification system],14 indication for the drug, start and end dates, dosage regimen, route of administration), and reactions/events (reported terms, MedDRA classification terms, onset date, end date, seriousness, final outcome). When a given ICSR reported on several suspected or interacting drugs, those were considered to be separately declared lupus-suspected drug associations. A serious adverse event was defined as an adverse reaction that causes death, is life-threatening, requires hospitalisation (initial or prolonged), leads to persistent or significant disability, congenital anomaly, birth defect or to any other medically important conditions.

Statistics

For each drug (suspected treatment) associated with DIL in VigiBase, a case–non-case analysis was performed. This method compares the proportion of specific ADR reported for a single drug with the proportion of the same ADR for all other treatments in VigiBase. In this case–non-case study, the specific ADRs studied (DIL) were considered as cases and all other reports were considered as non-cases. For this, we extracted from VigiBase the information component (IC), an indicator value for disproportionate reporting that compares observed and expected values with drug-adverse effect combinations that have been reported more often than one would expect.15 16 The IC is computed as: IC=log2((Nobserved +0.5)/(Nexpected+0.5)), where Nexpected=(Ndrug × Neffect) / Ntotal, with Nobserved being the actual number of case reports for the drug–effect combination; Nexpected the number of case reports expected for the drug–effect combination; Ndrug the number of case reports for the drug, regardless of effects; Neffect the number of case reports for the effect, regardless of drug; and Ntotal the total number of case reports in the database. We also report the IC025, which is the lower end of the 95% credibility interval for the IC. A positive IC025 value is the threshold used in statistical signal detection in VigiBase.13 The extracted data are presented as median (25th–75th percentile, IQR) or number (percentage). Analyses were performed using the JMP V.13 software (SAS Institute, Cary, North Carolina, USA). This methodology was previously used by our group for the study of other diseases such as drug-induced capillary leak syndrome.17 For the calculation of the median delay between start of medication and DIL onset, we considered the 15th of the given month when the month but not the exact day was available. For each drug, we performed an evaluation of the association with DIL based on a Medline search using the international non-proprietary names of the drug and the medical subject headings (MeSH) term for SLE (‘Lupus Erythematosus, Systemic’). Evaluation for the risk of drug-induced DIL was based on literature data: high risk was considered when the risk of developing DIL was ≥10% per year, moderate risk was considered when the risk of developing DIL was <10% but ≥1% per year, low risk was considered when the risk of developing DIL was <1% but ≥0.1% per year, and very low risk was considered when the risk of developing a DIL was <0.1% per year.

Association of these drugs with DIL was categorised as follow: (1) definite: drugs definitely inducing DIL based on matched case–control studies18 19; (2) probable: drugs possibly inducing DIL based on reports in cohorts; and (3) possible: drugs possibly inducing DIL based on case reports (≤10 cases).

Results

A total of 12 166 ICSRs of drug-induced SLE were identified using VigiBase. From those, we were able to identify 118 suspected drugs (see table 1 and online supplementary table 1) with significant pharmacovigilance signal (IC025 ≥0), which were reported in 8163 ICSRs (in some cases, several suspected drugs resulting in several declared lupus-suspected drug associations were reported in the same ICSR).

Table 1

List of the 118 drugs associated with drug-induced lupus in VigiBase, by decreasing IC025

The 8163 ICSRs were issued from 57 different countries (see table 2), mostly from the Americas (65%) and Europe (23%) (data available for 8161/8163 ICSRs, 99.9%). DIL occurred in women in 81.3% (female to male sex ratio=4.3) (data available for 7682 ICSRs, 94.1%), and the median age at onset of DIL was 49 years (IQR: 35–61) (data available for 5681/8163 ICSRs, 69.6%). The median delay between the reported start of suspected treatment and DIL occurrence was 172 days (IQR: 35–610) (data available for 21.1% of ICSRs) (see table 2). DIL was reported as a serious adverse event in 4521 cases (55.4% of ICSRs).

Table 2

Characteristics of individual case safety reports for drug-induced lupus

Among the 118 suspected drugs (table 2), 76 (64.4%) had been previously reported in association with DIL in Medline (see online online supplementary table 1). Anti-TNF agents were the most frequent drug family reported as suspected (2793 declared lupus-suspected drug associations, 32.2%), followed by procainamide (6.0%) and hydralazine (5.1%). The drugs associated with the highest disproportional reporting were procainamide (IC025=7.48) and hydralazine (IC025=6.63). Those with the highest number of reported DIL cases were infliximab (n=1055 [12.2%], IC025=3.39), adalimumab (n=926 [10.7%]; IC025=1.66), etanercept (n=691 [8.0%], IC025=1.07), procainamide (n=518 [6.0%], IC025=7.48) and hydralazine (n=444 [5.1%], IC025=6.63). Based on the literature review, a relationship with DIL was considered definite for nine drugs (procainamide, hydralazine, minocycline, quinidine, isoniazid, terbinafine, methyldopa, dihydralazine and chlorpromazine), probable for 19 drugs and possible for 45 individual drugs.

Analysis of the median year of reported DIL onset for each suspected drug (table 3) revealed profound changes in the spectrum of DIL, with most cases associated with procainamide (n=282/518, 54.4%) and hydralazine (n=336/444, 75.7%) being reported between 1967 and 1987, while 65.9% of ICSRs involving anti-TNFs such as infliximab (n=695/1055) were reported following 2007.

Table 3

Median year (IQR) of reported DIL onset for each suspected drug, grouped by decades

Conclusions

Here, we used VigiBase, the WHO global deduplicated ICSRs database, to describe the main drugs associated with DIL, as well as the detailed characteristics of these cases. We identified a total of 12 166 ICSRs of SLE and from those 118 suspected drugs in 8163 ICSRs) with significant pharmacovigilance signals.

The epidemiology of DIL is different from that of SLE. The median age at onset of DIL was 49 years (IQR: 35–61), which is about two decades older than that of spontaneous SLE.20 However, we observed a marked female predominance, close to what is observed in spontaneous SLE, which contrasts with previously published studies as those usually report a female to male sex ratio close to 1:1 (with some divergent data such as a 9:1 sex ratio in symptomatic procainamide-induced lupus5 21). This further underlines that the epidemiology of DIL has evolved with changes in pharmacopoeia and the typology of patients receiving suspected drugs.

The median delay between the reported start of suspected treatment and DIL occurrence was 172 days (IQR: 35–610) (see table 2), suggesting that DIL mostly appears after a few months and usually within the first 2 years of treatment with the suspected drug.

Among the 118 suspected drugs (table 2), 76 (64.4%) had been previously reported in association with DIL in Medline. DIL was considered definite for nine drugs (procainamide, hydralazine, minocycline, quinidine, isoniazid, terbinafine, methyldopa, dihydralazine and chlorpromazine), probable for 19 drugs and possible for 45 individual drugs.5

The magnitude of drugs–DIL disproportionality association was the highest for procainamide (IC025=7.48) and hydralazine (IC025=6.63). This is an important finding because these are the two drugs associated with the highest risk of DIL in the literature, therefore confirming the reliability of our approach using a large pharmacovigilance database. One important fact is that the relative incidence of DIL from each agent depends on the extent to which a given drug is prescribed, which constantly changes with the evolution of pharmacopoeia. Anti-TNF agents were the most frequent drug family reported as suspected. The analysis of the median reporting years for each suspected drug shows a clear evolution of suspected drugs during the past decades. This further underlines that the constantly changing spectrum of DIL should be monitored continuously, and further validates the interest of our approach using the WHO international pharmacovigilance database, the biggest database of this kind with over 16 million deduplicated ICSRs.

The limitations of the study are largely inherent to its design. First of all, we currently lack a uniform set of criteria for the diagnosis of DIL, and the level of reported details available in VigiBase is limited. However, as mentioned earlier, several drugs identified in this analysis had been previously confirmed as associated with DIL (positive controls), which strongly strengthens the validity of our findings. Also, some cases of DIL may not be reported to the national drug authorities, and therefore not submitted to VigiBase. However, one main strength is that VigiBase aggregates ICSRs collected in over 130 countries, which enables better identification of adverse events and broader generalisation of our findings. This analysis of 12 166 ICSRs of DIL is the largest to date. Another limitation is that the use of the PT term ‘systemic lupus erythematosus’ may have not fully capture all cases of drug-induced systemic lupus for some of which other terms of the MedDRA classification may have been used. Finally, while a positive IC025 value is the threshold used in statistical signal detection in VigiBase,13 a causal relationship cannot be formally ascertained.

Altogether, this analysis of the WHO VigiBase database enables the identification of significant pharmacovigilance signals for more than 118 drugs associated with DIL. The detailed list of suspected drugs may prove useful to physicians when confronted with potential DIL cases. Altogether, these findings may help in improving the identification of this constantly evolving disease.

References

Supplementary materials

  • Lay summary

    Disclaimer : This is a summary of a scientific article written by a medical professional (“the Original Article”). The Summary is written to assist non medically trained readers to understand general points of the Original Article. It is supplied “as is” without any warranty. You should note that the Original Article (and Summary) may not be fully relevant nor accurate as medical science is constantly changing and errors can occur. It is therefore very important that readers not rely on the content in the Summary and consult their medical professionals for all aspects of their health care and only rely on the Summary if directed to do so by their medical professional. Please view our full Website Terms and Conditions.
    Copyright © 2019 BMJ Publishing Group Ltd & European League Against Rheumatism. Medical professionals may print copies for their and their patients and students non commercial use. Other individuals may print a single copy for their personal, non commercial use. For other uses please contact our Rights and Licensing Team.

Footnotes

  • Handling editor Josef S Smolen

  • Contributors 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 published. Study conception and design: LA, PM, BL-V, J-ES. Acquisition of data: LA, PM, BL-V, J-ES. Analysis and interpretation of data: LA, PM, P-EG, TM, FC, MT-A, AL, CM, JS, BL-V, J-ES.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

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.

Linked Articles