A 44-year-old woman with a history of cocaine and heroin abuse presented with a painful, necrotic rash.
History of the present illness
Five weeks before presentation, the patient had experienced a cough and mild dyspnea. She was diagnosed with pneumonia, completed a 7-day course of ciprofloxacin and doxycycline, and her respiratory symptoms resolved. Three weeks before presentation, the patient had reported fevers, chills, diaphoresis, arthralgias, and myalgias, and developed a small erythematous lesion on her left breast. The lesion subsequently enlarged and became hemorrhagic. She simultaneously developed additional lesions on her chest, abdomen, arms, and back. The largest of these, located on her left arm, was 15 cm in diameter. All of the lesions were exquisitely tender. She was hospitalized at an outside hospital for 5 days and given a diagnosis of cocaine-induced vasculitis. Upon discharge, she was instructed to abstain from cocaine and was prescribed trimethoprim/sulfamethoxazole for a possible cellulitis involving the skin lesions.
Ten days after discharge from the outside hospital, the patient presented to our institution for pain management and further evaluation. The rash had progressed since her earlier discharge. She had continued to use cocaine, heroin, and prescription opioids. She had last smoked cocaine 3 days before presentation and last used intranasal heroin 1 week before admission. She reported acquiring cocaine from a new supplier 2 weeks before her rash had begun.
Past medical history
The patient had a history of nephrolithiasis and had undergone a cholecystectomy for gallstones. She had had 5 normal pregnancies.
Medications and allergies
The patient had received ciprofloxacin, doxycycline, and trimethoprim/sulfamethoxazole in the weeks before her presentation. She also took combined oxycodone and acetaminophen for pain. She had no known drug allergies.
Family and social history
Her youngest son had developed idiopathic thrombocytopenia at age 3 years. A paternal aunt had breast cancer and a nephew had cystinosis. She was unmarried, unemployed, and homeless. Three of her children were grown; the youngest two were in state custody. The patient had smoked 1 pack of cigarettes/day since age 15 years. She denied injection drug use and did not drink alcohol, but smoked cocaine and used intranasal heroin.
Review of systems
The patient reported headache and fatigue secondary to poor pain control and disrupted sleep. She denied visual changes, sicca symptoms, jaw pain, oral ulcers, and dysphagia. She denied shortness of breath, cough, chest pain, palpitations, gastrointestinal symptoms, and abdominal pain. She had no genitourinary symptoms, arthritic pain, or musculoskeletal tenderness. She reported new upper extremity edema in her hands and arms bilaterally, more significant on the left than on the right, and equal bilateral lower extremity swelling. She had no sick contacts or recent travel.
The patient was an obese woman in significant pain from her skin lesions. She was afebrile but had a pulse of 108/minute. Her blood pressure was 163/86 mm Hg, her respiratory rate was 22/minute, and her oxygen saturation was 97% on room air. She had a rash distributed in a symmetric pattern overlying fatty areas on the breasts, abdomen, upper arms, back, and flanks. The lesions consisted of stellate, purpuric plaques ranging in size from 5 cm to 15 cm (Figure 1A). The largest lesion was located on the left upper extremity (Figure 1B). The lesional borders were well demarcated and contained a reticular pattern surrounded by a 5-mm rim of erythema (Figure 1C). Small 1–2-cm lesions were also located on the earlobes bilaterally (Figure 1D). Many of the lesions were associated with flaccid bullae and exquisite tenderness to palpation. An additional dermatologic finding was a livedo pattern on her back and extremities. There was pitting edema in the upper and lower extremities bilaterally.
Examination of the head and neck was unremarkable, and there was no lymphadenopathy. The cardiovascular examination demonstrated a regular rate and rhythm, normal heart sounds, and no murmurs. The lungs were clear to auscultation. The abdomen was soft, nondistended, nontender, and without palpable masses. A neurologic examination showed intact cranial nerves and normal sensation and strength. Cerebellar signs were absent, and the gait was normal.
The complete blood cell count, coagulation times, electrolytes, bilirubin levels, liver enzymes, cardiac biomarkers, and urinalysis were within normal limits. The patient had a mild normochromic anemia with a hematocrit of 35% and a mean corpuscular volume of 82 femtoliters (normal range 80–100). The serum iron saturation level was 4.5% and the ferritin was mildly elevated at 217 ng/ml (normal range 10–200). The erythrocyte sedimentation rate was 33 mm/hour (normal range 0–17). The patient's D-dimers were significantly elevated at 5,541 ng/ml (normal value <500), but the fibrinogen concentration was normal. Blood and urine cultures showed no growth. Assays for syphilis, hepatitis B and C, and the human immunodeficiency virus were nonreactive.
An antinuclear antibody assay was positive at a titer of 1:160 (speckled). Assays for antibodies to both double-stranded DNA and histones were positive at titers of 1:20 (normal titer <1:10) and 2.2 units (normal value <1.5), respectively. Assays for antibodies to the Ro, La, Sm, and RNP antigens were negative. The serum C3 and C4 levels were 137 mg/dl (normal range 86–184) and 10 mg/dl (normal range 20–58), respectively; cryoglobulins were not detected. An IgM anticardiolipin antibody (aCL) assay was strongly positive at 150 IgM phospholipid units (normal range 0–15), but the IgG aCL assay was negative at 4.2 IgG phospholipid units (normal range 0–15). Assays for a lupus anticoagulant and antibodies to β2-glycoprotein I were negative. The serum antithrombin IIIA level was 66% (range 80–130%). Because of the resemblance of the skin lesions to warfarin necrosis, a serum warfarin level was assayed and found to be normal.
An immunofluorescence assay for antineutrophil cytoplasmic antibodies (ANCAs) was positive in a perinuclear ANCA (pANCA) pattern of staining. Enzyme-linked immunosorbent assays (ELISAs) confirmed the presence of antibodies to myeloperoxidase (MPO ANCA). Antibody activity was measured at 1,050 units (normal value ≤2.8). An ELISA for antibodies to proteinase 3 (PR3) was negative.
A serum parathyroid hormone (PTH) was normal, and assays for PTH-related peptide and plasma activated inhibitor 1 were negative. A serum protein electrophoresis with immunofixation studies was normal. Cocaine, oxycodone, and opiates were identified in a urine toxicology screen. A chest radiograph was normal, and ultrasound studies of the lower extremities and left arm showed no deep venous thromboses.
A diagnostic test was performed.
The patient is a 44-year-old woman with active cocaine and heroin use who presented with a painful, bullous, necrotic rash that developed after receiving a short course of antibiotics. The skin lesions have persisted despite conservative management for 5 weeks. The physical examination is otherwise remarkable for bilateral upper and lower extremity swelling. Laboratory studies are significant for an IgM aCL and MPO ANCA.
Here we discuss the differential diagnosis of the patient's necrotic rash in the context of her laboratory abnormalities.
A primary concern on admission was a hypercoagulable state. Heritable disorders of coagulation typically are considered in patients with a history of venous or arterial thrombi or pregnancy loss, or a family history of such events. Our patient had no such history. Laboratory testing confirmed the absence of mutations in the prothrombin, antithrombin III, and factor V Leiden genes. Protein C and S activities and the homocysteine level were normal.
Acquired hypercoagulable states were considered early in the management of this patient. The patient had no history of recent surgery, immobilization, cancer, pregnancy, or hormonal therapy. All of these risk factors are associated more strongly with venous thrombi than with arterial clots, and ultrasonography excluded evidence of venous thrombosis. In any event, the distribution and necrotic appearance of the rash were more consistent with an arterial process.
One common hypercoagulable state that leads to arterial or venous thrombi (or both) is the antiphospholipid syndrome (APS). The APS can develop at any age, is more common among women, and has a diverse range of clinical manifestations (1). The APS is characterized by the detection within serum of antiphospholipid antibodies (aPL); specifically, aCL, antibodies to β2-glycoprotein I, or a lupus anticoagulant. By the Sapporo criteria, the diagnosis of APS is entertained in the setting of a typical rash (e.g., livedo racemosa) and the presence of aPL within the serum (2, 3). To fulfill the laboratory criteria, aPL must be detected on at least two occasions separated by more than 12 weeks. Overall, we had a high suspicion of APS, given the clinical manifestations and laboratory findings of IgM aCL.
Small-or medium-vessel vasculitis
Polyarteritis nodosa (PAN) is a systemic necrotizing vasculitis of medium-sized arteries that can affect any organ system. Skin involvement is common. The characteristic skin findings include livedo reticularis (racemosa), tender erythematous nodules, bullae or vesicular eruptions, and ulcers (4). Palpable purpura can be present to a minor degree and when found, represents concomitant small-vessel involvement. Skin lesions can result in infarction and gangrene of the distal extremities or deep ulcerations that extend into the subcutaneous tissue. In PAN, the cutaneous involvement is often accompanied by renal disease (e.g., renin-mediated hypertension, kidney infarctions), an axonal sensorimotor mononeuritis multiplex, and mesenteric vasculitis. None of these other features was present in our patient. A form of PAN limited to the skin exists, but that condition is more subacute in its presentation and seldom leads to such dramatic bulla formation (5).
PAN is a “seronegative” condition, i.e., one that is not associated with known autoantibodies. This fact poses a significant challenge in diagnosis compared with, for example, systemic lupus erythematosus and the ANCA-associated vasculitides (AAVs). A minority of PAN patients have been infected recently by the hepatitis B virus and have serologies consistent with that condition. Our patient had no history of hepatitis B exposure, and the titers of aCL and MPO ANCAs suggested alternative diagnoses.
Cryoglobulinemia was considered in the differential diagnosis because the patient had palpable purpura and bilateral earlobe lesions. Both of these physical findings suggest small vessel involvement. Nonetheless, the absence of lymphadenopathy, hepatosplenomegaly, peripheral neuropathy, and renal insufficiency made cryoglobulinemia less likely. Secondary causes of type I cryoglobulinemia were considered. Electrophoresis studies of the serum and urine were negative, appearing to exclude a paraproteinemia. Mixed cryoglobulinemia (types II and III) can occur secondary to hepatitis C or human immunodeficiency virus infections, but testing for our patient was negative. Most importantly, careful attempts to isolate cryoglobulins from the serum were unsuccessful.
The necrotic rash and positive ANCA assay are consistent with an AAV. There are two major ANCA immunofluorescence patterns: cytoplasmic ANCA (cANCA) describes diffuse staining throughout the cytoplasm and pANCA refers to perinuclear staining. In patients with systemic vasculitis, the preponderance of cANCA staining is caused by antibodies directed against serine PR3. In contrast, in a patient who has a primary vasculitic syndrome, the pANCA pattern is generally caused by antibodies to MPO. In our patient, the high-titer MPO ANCA suggested several possibilities: Wegener's granulomatosis (WG), microscopic polyangiitis (MPA), Churg-Strauss syndrome (CSS), and drug-induced AAV.
The clinical presentations of WG and MPA are difficult to distinguish in some cases. In WG, skin lesions can range from mild erythematous purpura to severe ulcerations (6). Skin manifestations are reported in 40% of patients with WG and generally correspond to the presence of generalized disease (7). However, vasculitic involvement of the skin in WG is unlikely to involve the trunk and upper extremities in the absence of lower extremity disease. In addition, our patient had no sign of the classic extracutaneous disease manifestations that typify WG: specifically, upper respiratory tract dysfunction, pulmonary disease, and glomerulonephritis.
The MPO ANCA detected in our patient is more typical of either MPA or CSS than of WG. In MPA, skin findings usually consist of purpura or livedo racemosa rather than bullae and extensive necrosis (8). Approximately 50% of CSS patients are ANCA positive, and in such cases, the antibody specificity is directed against MPO. However, the diagnosis of CSS is unlikely in the absence of renal or pulmonary involvement and eosinophilia. In summary, the diagnosis of an idiopathic form of AAV is tenuous without some sign of extracutaneous disease.
Drug-induced AAV and drug-induced lupus.
The constitutional symptoms, necrotic rash, and high MPO ANCA titer are consistent with both drug-induced AAV and drug-induced lupus. The two entities, which have been described both together and separately in the literature, may represent a spectrum of the same disease process. Both syndromes are associated with arthralgia, myalgia, and rash at disease onset (9).
Dramatically elevated MPO ANCA titers are more strongly associated with drug-induced AAV than with idiopathic AAV (10). In addition, antihistone and aPL antibodies (both present in this case) are observed more commonly in drug-induced processes than in idiopathic AAV (11, 12). In both drug-induced ANCA vasculitis and lupus, the syndromes occur after exposure to a drug over a prolonged period of time. This contrasts with the short exposures to medications usually linked to hypersensitivity vasculitis syndromes, of which antibiotics are the most common offenders (13).
Numerous medications have been implicated in both drug-induced AAV and drug-induced lupus (Table 1). However, the strongest correlations are with medications used to treat hyperthyroidism (particularly propylthiouracil) and with hydralazine, D-penicillamine, and minocycline (10, 14, 15). Our patient had no exposure to any of these agents. She had been treated previously with a short course of doxycycline, but tetracyclines other than minocycline have not been associated with drug-induced AAV (15).
Table 1. List of medications implicated in drug-induced antineutrophil cytoplasmic antibody–associated vasculitis (AAV) and the antiphospholipid syndrome (APS)
Ciprofloxacin has been implicated in vasculitis case reports (16, 17). In the 5 most recent reports, ciprofloxacin exposure ranged from 3–14 days, similar to our patient's 7-day course. However, the skin findings in ciprofloxacin-induced cutaneous vasculitis are generally mild in comparison to our patient's lesions, and those patients do not develop ANCAs or antinuclear antibodies (17).
In both drug-induced AAV and drug-induced lupus, early withdrawal of the offending agent usually leads to resolution of the clinical syndrome (9). Prolonged cases of AAV can occur if exposure to the medication has been lengthy, as in some cases of propylthiouracil-associated disease. In our patient's case, her doxycycline and ciprofloxacin had been stopped even before the onset of the rash. In summary, the likelihood that the antibiotics to which she had been exposed had caused her severe skin rash appeared low.
Case reports correlating cocaine use with skin lesions emerged first in the 1980s (18, 19). Since that time, palpable purpura; violaceous papules, bullae, and blisters; livedo racemosa with cutaneous necrosis; and multifocal necrotic lesions of the skin and muscle have been associated with cocaine use (4, 20–22). Skin biopsy samples in such patients often demonstrated vasculopathic patterns rather than frank vasculitis (21, 22).
Cocaine use is well known to trigger a disorder known as midline destructive lesion that mimics limited WG closely, even including the finding of cANCA positivity on immunofluorescence testing (21–23). In such cases, ELISA testing demonstrates ANCA directed against PR3 and/or human neutrophil elastase (HNE). It is not clear if ANCA contributes to the pathogenesis of these lesions. HNE and PR3 are both serine proteinases that share gene localization as well as structural and functional characteristics. It has been speculated that the true antigen in cases of cocaine-induced midline destructive lesion is HNE, and that the PR3 ANCA results from cross-reactivity with HNE (21, 24). However, investigators recently demonstrated the coexistence of distinct antigen-specific antibodies that recognize unique epitopes (25).
Cocaine use is also associated with the APS. One prospective study showed a higher percentage of aCL positivity in cocaine users compared with controls (26). Whether this was an incidental finding or has broader clinical implications for the occurrence of APS in the setting of cocaine use remains unknown. Some cocaine users who have experienced acute strokes or myocardial infarctions have been reported to have aPL, suggesting the possibility of a synergistic effect between cocaine and aPL that propagates thromboembolic phenomena (27, 28).
Based on the patient's clinic findings and serologic abnormalities, the two most likely causes of the patient's presentation appeared to be either drug-induced AAV or drug-induced APS, either of which might have been induced by cocaine. Skin biopsy was essential to distinguish between these two conditions, which require different approaches to treatment (in addition to cocaine cessation). Skin biopsy samples were obtained from the right proximal and distal forearm. Histologic examination revealed fibrin thrombi within small and medium-sized blood vessels in the superficial and deep dermis and the subcutaneous tissue (Figures 2A and B). There was no evidence of endothelial swelling, leukocytoclasis, or fibrinoid necrosis within blood vessel walls. The histopathology was thought to be highly consistent with APS.
The most striking features of this case were the rapid appearance of the patient's bullous, necrotic rash and the high titers of both aPL and MPO ANCA in her blood. Although we believe that both autoantibodies were induced by her cocaine use, the histopathology of her skin lesions suggests that aPL were responsible for her skin rash. This conclusion is consistent with the current knowledge regarding drug-induced ANCA positivity. In several cross-sectional and prospective studies of propylthiouracil, between 20% and 60% of patients treated with this medication on a long-term basis developed ANCA, usually in high titers (29, 30). However, only a minority of patients who developed ANCA demonstrate clinical manifestations of drug-induced AAV. In one study (31), 8 (26.7%) of 30 patients receiving long-term antithyroid medication became ANCA positive. Among those, none developed frank AAV but 3 experienced myalgia and arthralgias after the appearance of ANCAs.
Patients with idiopathic AAV are known to have a predisposition to venous thrombotic events (32, 33). Variable degrees of thrombosis can also be observed on lesional biopsy samples from patients with AAV, but such findings are accompanied by more classic patterns of injury to the blood vessel wall: endothelial swelling, leukocytoclasis, and fibrinoid necrosis. Although it is impossible to exclude an interaction between the patient's aPL and MPO ANCA that heightened the intensity of her cutaneous reaction, the major histopathologic findings implicate aPL as the primary etiology of the patient's skin disease.
One theory of APS pathogenesis proposes that aPL develop in genetically susceptible individuals after exposure to an unknown infectious agent. Clinical manifestations subsequently occur after a “second hit” such as smoking, prolonged immobilization, pregnancy, hormone use, or cancer. Various medications are suspected triggers of APS (Table 1). IgM aCL in particular have been linked to drug-induced APS cases. Cocaine has also been implicated in the pathogenesis of APS. Although our patient was a chronic cocaine user, it is of interest that prior to the onset of her rash she had found a new cocaine supplier, perhaps leading to exposure to new contaminants in the cocaine.
One commonly recognized contaminant in cocaine is levamisole. Levamisole, an alkaline phosphatase inhibitor, appears to potentiate the action of cocaine on dopaminergic pathways. Levamisole is widely available because it is used in agriculture worldwide as an antihelminthic drug. In humans, levamisole has been studied in various clinical settings for the treatment of autoimmune and malignant diseases (34). The compound has established efficacy in relapsing nephrotic syndrome in children. Most recently, levamisole has been reported to cause agranulocytosis in cocaine users (35, 36).
Approximately 70% of seized cocaine entering the US contains levamisole (35). We conducted a literature search on levamisole-induced vasculitis and vasculopathies. Levamisole-induced vasculitis has only been reported in 4 adult cases (37–39). The histopathology reported in these publications describes leukocytoclastic vasculitis, but the results of ANCA or aPL testing were not included. In the pediatric literature, there are 6 case reports of cutaneous vasculitis secondary to levamisole (40, 41). Five cases were characterized by earlobe purpura and necrosis, as seen in our case (41). In 4 of the patients, skin biopsy samples demonstrated a thrombotic vasculopathy. Positive ANCA serologies were reported in 4 patients: pANCA patterns in 3 cases and a cANCA pattern in 1 case. aPL antibodies were detected in 3 patients.
We suspect that levamisole is the active compound in many cases of cocaine-induced vasculitis or cocaine-induced vasculopathy. This potential relationship warrants further investigation.
Cocaine-induced vasculopathy associated with antiphospholipid antibodies.
During her hospitalization, anticoagulation was started with a continuous heparin drip and was bridged to warfarin prior to discharge. Lifelong anticoagulation was advised for the diagnosis of APS. Given the high ANCA titer, there was discussion among the medical team whether the patient would benefit from immunosuppressants. Given her acute presentation and the severity of the skin lesions, the patient was started on cyclophosphamide and prednisone. On this treatment, her skin lesions stabilized and her immunosuppression was tapered after 3 months. At 6 months after her admission, she had abstained from cocaine and had no recurrence of her rash. She has developed atrophic scars of her necrotic skin rash.
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 submitted for publication. Dr. Stone had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Lee, Stone, Khosroshahi.
Acquisition of data. Lee, Stone, Khosroshahi.
Analysis and interpretation of data. Lee, Stone, Gimbel, Khosroshahi.