Drug-induced lupus

Introduction and historical perspective

Typical drug-induced lupus (DIL) was first reported by Morrow and coworkers in which a late onset “collagen disease” resembling systemic lupus erythematosus (SLE) developed in 17 out of 211 hydralazine-treated patients and by Dustan and coworkers in which 13 out of 139 patients on hydralazine developed constitutional and rheumatologic symptoms. Eight years later Ladd reported a patient who developed lupus-like features after six months of procainamide therapy, followed by another report in 1967 of 11 cases of procainamide-induced lupus. By then a scattering of cases of lupus-like disease appeared as a side effect of therapy with isoniazid, diphenylhydantoin, sulphamethoxypyridazine, primidone, and tetracycline. As the pharmacopeia has expanded over the subsequent 50 years, so has the number of drugs that have been established or potentially implicated in DIL.

Diagnosis of drug-induced lupus

There is considerable heterogeneity in symptoms and signs of DIL, and its onset can be rapid or gradual in severity. Because of this varied disease expression and its commonly obscure relationship to one or more medications a patient might be using, rigid diagnostic criteria are inappropriate for DIL. The following guidelines can suggest a patient is experiencing lupus-like symptoms as a side effect of an ingested medication.

• Continuous treatment with a known lupus-inducing drug for at least one month and usually much longer

• Presenting symptoms

  • Common: arthralgias, myalgias, malaise, fever, serositis (especially procainamide and statins), polyarthritis (quinidine, minocycline).

  • Rare: vasculitic rash or other cutaneous problems

• Laboratory profile

  • Common: antinuclear antibodies (ANA), which are due to histone- or nucleo-histone-reactive antibodies especially IgG anti[(H2A-H2B)-DNA], antisingle-stranded (denatured) DNA antibodies, increased erythrocyte sedimentation rate, antineutrophil cytoplasmic antibodies [perinuclear antineutrophil cytoplasmic antibodies (pANCA) especially with hydralazine, propylthiouracil, and minocycline]

  • Uncommon: leukopenia, thrombocytopenia and mild anemia (e.g., with hydralazine, isoniazid, statins), hemolytic anemia (methyldopa), glomerulonephritis (hydralazine), and hepatitis (minocycline, atorvastatin, and sulfasalazine).

  • Rare: IgG antibodies to native DNA, Sm/RNP, SS-A/Ro, SS-B/La, hypocomplementemia.

• Improvement and permanent resolution of symptoms after discontinuation of the implicated drug and without antiinflammatory therapy.

In most patients with DIL symptoms are mild although indistinguishable from early-onset SLE and usually include fever, malaise, weight loss, polyarticular arthralgia, and symmetric myalgias. Some drug-specific characteristics have been reported, but lupus induced by any one drug cannot generally be identified by clinical or laboratory features. Central nervous system disease, while not a feature of lupus induced by most drugs with the possible exception of quinidine, should not be an exclusion criterion because of possible independent neurotoxic effects of drugs or the occurrence of stroke, convulsions or dementia syndromes in the elderly commonly treated with these drugs. Similarly, although serious kidney disease is rare for most lupus-inducing drugs, glomerulonephritis has been associated with hydralazine-induced lupus so it should also not be a formal exclusion criterion. In the latter cases, it could be difficult to distinguish hydralazine-induced glomerulonephritis from that related to the underlying hypertension for which hydralazine was administered although the focal nature of drug-induced glomerulonephritis is consistent with an immune complex-mediated process . Mucocutaneous manifestations are also rare in DIL. Some patients with DIL have mild leukopenia, thrombocytopenia and/or anemia and elevated sedimentation rate but rarely hypocomplementemia. Finally, although a history of rheumatologic disease independent of the suspected drug tends to negate a diagnosis of DIL, obviously a patient can have two diseases. This situation is characteristic of patients with various forms of arthritis who also develop DIL from penicillamine, sulfasalazine, minocycline, and therapy with certain biologics. In these difficult cases serologic findings are especially informative.

IgG anti[(H2A-H2B)-DNA] antibodies (antichromatin/nucleosome antibodies) or antihistone antibodies can help to confirm a diagnosis of DIL. Although antihistone and anti[(H2A-H2B)-DNA] antibodies are also common in SLE, SLE patients are rarely monospecific for this autoantibody. Therefore when a diagnosis of SLE or DIL cannot be clearly distinguished on clinical grounds or history, the presence of SLE-associated antibodies such as to native DNA, Sm, RNP, SS-A/Ro, SS-B/La, or other nuclear antigens should be considered as evidence against a diagnosis of DIL.

DIL usually occurs after several months or years of continuous therapy and should not be confused with other side effects that are often suffered by patients treated with pharmaceuticals. It is not uncommon for procainamide- or hydralazine- treated patients to require more than 3 years before symptoms become manifested. In some patients, symptoms gradually appear and worsen over the course of many months of treatment with the implicated drug, whereas in others, symptom onset is rapid. Variation in time of onset of DIL is due in part to differences in the steady-state drug concentrations employed in order to maintain therapeutic control, but genetic factors may also be involved (see as follows). Genetic or acquired differences in metabolism of pharmaceuticals are likely to underlie much of the variation in risk for developing DIL in general.

Lupus-inducing drugs with specific clinical features

Some suggestion of drug-specific symptoms in addition to the usual musculoskeletal and constitutional symptoms is suggested by the literature, although patient heterogeneity in symptomatology and kinetics of disease onset and evolution complicate interpretation. Serositis is distinctly common in procainamide-induced lupus, polyarthritis in quinidine- and minocycline-induced lupus, and glomerulonephritis and rash has been frequently reported in hydralazine-induced lupus . Lung involvement in procainamide-induced lupus occurs in approximately 50% of patients and consists of pleuritis, pleural effusions and/or pulmonary infiltrates; pericardial effusions are also common. Quinidine is sometimes associated with a mild drug reaction characterized only by polyarthalgias within seven days to three months after initiation of therapy . Lupus-like disease associated with minocycline is also atypical—patients frequently present with symmetrical polyarthritis and may have evidence of hepatitis (elevated liver transaminases) and pneumonitis (due to pulmonary lymphocytic infiltrates), which are assumed to be autoimmune in nature. These patients may not have ANA but frequently have perinuclear antineutrophil cytoplasmic antibodies (pANCA) due to antimyeloperoxidase (MPO) . Autoimmune hepatitis in addition to systemic autoimmunity suggestive of DIL has also been associated with atorvastatin and with mesalamine (5-aminosalicylic acid) .

Distinguishing DIL from idiopathic SLE

Although some of the criteria for the diagnosis of SLE are applicable to DIL, the requirement for four manifestations as established by the American College of Rheumatology is overly rigid for DIL and obviously not useful for distinguishing between drug-induced and idiopathic SLE. In particular symptoms common to SLE such as malar or discoid rash, photosensitivity, oral ulcers, alopecia, renal, or neurological disorders are very unusual in DIL. Such symptoms would be suggestive of true SLE, a much more serious disease. DIL patients frequently present with mild or few lupus-like symptoms, which typically worsen the longer the patient is maintained on the causative drug. As a result, early symptoms of DIL are commonly ignored, delaying a correct diagnosis.

Treatment and management of DIL

Resolution of symptoms and laboratory abnormalities by withdrawing the offending drug is a defining feature of DIL. Symptoms usually resolve within a few weeks after discontinuing the offending drug and, therefore, this action provides a key (although retrospective) diagnostic tool. Patients with severe DIL may require up to six months for full recovery. Although autoantibodies and related laboratory abnormalities should also resolve after discontinuation of therapy, autoantibody disappearance often takes much longer and can still be present one to two years after withdrawal of therapy. However, quantitative measurements of autoantibodies should show a systematic decline in activity once the causative agent is withdrawn.

Although there is inducement to treat patients suspected of DIL with antiinflammatory agents, this action may confound the diagnosis and should not be required for recovery from DIL. Treatment with immunosuppressive agents, including corticosteroids, may be indicated for those with severe manifestations of the disease such as pericarditis with tamponade, inflammatory pleural effusions, debilitating polyarthritis, or glomerulonephritis. The prolonged use of high dose corticosteroids, or the use of chloroquine, hydroxychloroquine, or immunosuppressive agents is not indicated in the treatment of DIL. Although some DIL patients have been restarted on procainamide or hydralazine without incident, this is not advisable if insufficient time (which may be as long as one year) has elapsed since the initial episode of DIL.

Clinicians are often consulted to consider the safety of drugs associated with DIL in the treatment of patients with idiopathic SLE. The use of hydralazine to treat hypertension in SLE patients has not been associated with exacerbations of the disease , and procainamide has been used without incident in SLE . Patients with established SLE have been treated with procainamide for long duration without exacerbation of symptoms . The use of anticonvulsants to treat seizure disorders in SLE patients has not been associated with flares or acceleration of disease activity. Isoniazid has been given to SLE patients on corticosteroids without aggravating lupus . Despite their relative safety in the setting of SLE, it would be judicious to document the clinical and serologic status of the patient being considered for treatment with a known lupus-inducing drug so that new-onset symptoms and signs due to DIL could be more readily detected.

Lupus-inducing drugs

Table 56.1 lists drugs reported to be associated with a lupus-like syndrome, organized by therapeutic classes and indicating their approximate risk for developing DIL. Thus some six dozen drugs currently in use have a propensity for inducing autoantibodies and a lupus-like syndrome. While listed separately, some drugs, particularly those in the same therapeutic class, are essentially the same chemical structure and could be considered a single agent. By far the highest risk drugs are procainamide and hydralazine, with approximately 20% incidence for procainamide and 5%–8% for hydralazine during one year of therapy at historically used doses. Quinidine can be considered moderate risk while sulfasalazine, chlorpromazine, penicillamine, methyldopa, carbamazepine, acebutalol, isoniazid, captopril, propylthiouracil, and minocycline are relatively low risk. The risk for developing lupus-like disease from the remainder of the drugs is much lower, probably less than 0.1% of treated patients, although there are few if any formal assessments of DIL propensity. With some drugs on this list there is only one case reported in the literature. Excluded from this list are drugs implicated only in the exacerbation of SLE or drugs that induce nonmultisystem cell or tissue abnormalities such as cytopenias or cutaneous-only manifestations. Some drugs may be falsely implicated or are currently of negligible risk because customary treatment doses have decreased, but most reports on DIL are convincing because cessation of therapy usually results in prompt resolution of symptoms and eventually autoantibodies.

Tumor necrosis factor alpha (TNFα)-blocking biologics have replaced the cardiovascular (CV) drugs procainamide, hydralazine, and quinidine as the most commonly reported agents associated with DIL: 0.19% of patients on infliximab, 0.18% on etanercept, and 0.41% on adalimumab . This change is due in part to the decrease use and/or dose of the CV agents and to the surge in use of macromolecular biologics in the treatment of rheumatic diseases. However, because patients being treated with TNFα-blockers usually have a preexisting autoimmune disease, they may be at increased risk for spontaneously evolving into another one such as SLE. Many of these patients present with cutaneous manifestations typical of SLE-like malar rash as opposed to classical DIL . Using the French Pharmacovigilance 2000–2012 Database, Moulis and coworkers also reported a high incidence (65%) of cutaneous involvement and the presence of antidsDNA in 64% of 39 patients treated with TNFα-blocking agents. It should be noted that, where studied, antiDNA antibodies were predominately of the IgM isotype and, less frequently, IgA antidsDNA ; IgG antidsDNA in significant titer, which is specifically associated with SLE, was distinctly uncommon. Although organ involvement is generally rare, renal disease has been reported in some patients .

The expanding breadth of lupus-inducing drugs

Recently, several large-population studies were undertaken to reevaluate and update information on possible associations between currently used medications and ANA or DIL. In the Dinse et al.’ study sera from 3608 US patients using a broad range of prescription medications were tested for ANA, and this data were analyzed for statistical association with medication usage. No medications previously implicated in ANA induction in individual case reports or small patient series were found as risk factors for ANA in this population-representative study. This is not surprising because <1% of commonly used drugs are associated with DIL and this generally occurs in at most 1% of treated patients, which would be undetectable in the Dinse et al.’ study , although the propensity for drug-induced ANA without symptoms is largely unknown for most drugs. While quinidine, hydralazine and especially procainamide were previously reported to induce ANA in up to one-half of treated patients, these drugs are now uncommonly used for long periods or are used at lower dose. Nevertheless, these findings are consistent with the concept that the idiosyncratic drug reaction underlying induction of ANA and lupus are dependent on a coincidence of genetically dependent pharmacologic and immunologic events not typically occurring in the general population. Interestingly, several medications not previously implicated in DIL had significantly increased risk for ANA, especially the benzodiazepine class of drugs [odds ratio (OR) = 2.1], as well as furosemide (OR = 1.8), omeprazole (OR = 2.1), triamterene (OR = 3.1), salmeterol (OR = 3.8), ciprofloxacin (OR = 4.2), and tolterodine (OR = 6.4). However, as exemplified by procainamide-induced ANA versus DIL , a significant risk for ANA induction does not necessarily translate into increased risk for DIL.

Two recent data-mining projects evaluated the extent of association between DIL and currently used medications. The Schoonen and coworkers’ study collected clinical information from 4507 primary care patients from the United Kingdom General Practice research database for up to 14 years, some of whom were treated with drugs previously reported to be at high, moderate, or low risk for causing DIL . During this period patients who were treated with one of the selected drugs for at least one year and received a diagnosis of SLE or DIL were matched in time with control subjects who had no autoimmune disease. Among the 875 cases of a lupus-like diagnosis, only six subjects were treated with the high or moderate risk drugs (hydralazine or quinidine), and 50 subjects were treated with minocycline, considered a low risk drug. After correcting for confounders, the OR for a lupus diagnosis was statistically significant but only ∼4-fold for hydralazine and minocycline and 1.9-fold for carbamazepine. In addition, it was not determined by follow-up studies whether these patients actually had DIL. Overall, this analysis is consistent with the notion that DIL is rare for medications commonly used in the general population.

In another data-mining study Arnaud and coworkers searched for possible DIL in >16 million reports of suspected adverse drug reactions that were entered into the phamacovigilance database of the World Health Organization International Drug Monitoring Programme from >130 countries over the past 51 years. Of these, 8163 reports in the category of new onset SLE were associated with 118 medications, and statistically significant disproportionate cases of SLE-like symptoms were calculated. As expected, drugs known to be at high risk for DIL (procainamide and hydralazine) occurred with particularly high prevalence of new onset lupus-like symptoms. However, the TNFα-blocking biologics infliximab, adalimumab, and etanercept were the most commonly reported agents associated with SLE-like symptoms, reflecting the frequent use of these macromolecular biologics over the past three decades. TNFα-blocking agents also showed significant association with lupus diagnosis in the report of Moulis and coworkers , with infliximab and adalimumab having a higher OR than etanercept. However, as with the Schoonen and coworkers’ study , there was no follow-up data verifying the diagnosis of DIL after discontinuation of the suspected lupus-inducing drug.

Overall, while an unknown number of suspected cases of drug-induced ANA or DIL may be false reports, these recent data-mining, large population studies provide a valuable resource for implicating frequently used medications that might be at risk for causing DIL. The Arnaud and coworkers’ study identified 42 drugs that have not previously been known to cause DIL, and Dinse and coworkers added seven new drugs suspected of inducing ANA. These findings should sensitize the health care community to increased vigilance for DIL with patients taking these drugs and motivate the publication of novel, convincing case reports.