Autoantibodies against 3-hydroxy-3-methylglutaryl-coenzyme A reductase in patients with statin-associated autoimmune myopathy

Authors

  • Andrew L. Mammen,

    Corresponding author
    1. Johns Hopkins University School of Medicine, Baltimore, Maryland
    • Johns Hopkins Bayview Medical Center, Johns Hopkins Myositis Center, 5200 Eastern Avenue, Mason F. Lord Building Center Tower, Suite 4100, Baltimore, MD 21224
    Search for more papers by this author
    • Drs. Mammen, Christopher-Stine, Rosen, and Casciola-Rosen have filed a report of invention concerning the use of HMGCR autoantibodies as a diagnostic marker.

  • Tae Chung,

    1. Johns Hopkins University School of Medicine, Baltimore, Maryland
    Search for more papers by this author
  • Lisa Christopher-Stine,

    1. Johns Hopkins University School of Medicine, Baltimore, Maryland
    Search for more papers by this author
    • Drs. Mammen, Christopher-Stine, Rosen, and Casciola-Rosen have filed a report of invention concerning the use of HMGCR autoantibodies as a diagnostic marker.

  • Paul Rosen,

    1. Johns Hopkins University School of Medicine, Baltimore, Maryland
    Search for more papers by this author
    • Drs. Mammen, Christopher-Stine, Rosen, and Casciola-Rosen have filed a report of invention concerning the use of HMGCR autoantibodies as a diagnostic marker.

  • Antony Rosen,

    1. Johns Hopkins University School of Medicine, Baltimore, Maryland
    Search for more papers by this author
  • Kimberly R. Doering,

    1. Johns Hopkins University School of Medicine, Baltimore, Maryland
    Search for more papers by this author
  • Livia A. Casciola-Rosen

    1. Johns Hopkins University School of Medicine, Baltimore, Maryland
    Search for more papers by this author
    • Drs. Mammen, Christopher-Stine, Rosen, and Casciola-Rosen have filed a report of invention concerning the use of HMGCR autoantibodies as a diagnostic marker.


Abstract

Objective

In addition to inducing a self-limited myopathy, statin use is associated with an immune-mediated necrotizing myopathy (IMNM), with autoantibodies that recognize ∼200-kd and ∼100-kd autoantigens. The purpose of this study was to identify these molecules to help clarify the disease mechanism and facilitate diagnosis.

Methods

The effect of statin treatment on autoantigen expression was addressed by immunoprecipitation using sera from patients. The identity of the ∼100-kd autoantigen was confirmed by immunoprecipitation of in vitro–translated 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR) protein. HMGCR expression in muscle was analyzed by immunofluorescence. A cohort of myopathy patients was screened for anti-HMGCR autoantibodies by enzyme-linked immunosorbent assay and genotyped for the rs4149056 C allele, a predictor of self-limited statin myopathy.

Results

Statin exposure induced expression of the ∼200-kd/∼100-kd autoantigens in cultured cells. HMGCR was identified as the ∼100-kd autoantigen. Competition experiments demonstrated no distinct autoantibodies recognizing the ∼200-kd protein. In muscle biopsy tissues from anti-HMGCR–positive patients, HMGCR expression was up-regulated in cells expressing neural cell adhesion molecule, a marker of muscle regeneration. Anti-HMGCR autoantibodies were found in 45 of 750 patients presenting to the Johns Hopkins Myositis Center (6%). Among patients ages 50 years and older, 92.3% had taken statins. The prevalence of the rs4149056 C allele was not increased in patients with anti-HMGCR.

Conclusion

Statins up-regulate the expression of HMGCR, the major target of autoantibodies in statin-associated IMNM. Regenerating muscle cells express high levels of HMGCR, which may sustain the immune response even after statins are discontinued. These studies demonstrate a mechanistic link between an environmental trigger and the development of sustained autoimmunity. Detection of anti-HMGCR autoantibodies may facilitate diagnosis and direct therapy.

Statins lower cholesterol levels by specifically inhibiting 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR), a key enzyme in the cholesterol biosynthesis pathway. These drugs significantly reduce cardiovascular end points and are among the most commonly prescribed medications, with almost 30 million people in the US prescribed a statin in 2005 (1). Musculoskeletal symptoms are a well-known complication of statin use and range from myalgias and cramps, which occur in 9–20% of statin users (2–4), to life-threatening rhabdomyolysis, a rare event occurring at a rate of ∼0.4 per 10,000 patient years (5).

In most cases, statin-induced myopathic events are self-limited, with complete recovery in the weeks or months after the statin is discontinued (6). However, 2 recent studies have described 33 patients who developed an autoimmune myopathy following statin exposure, which did not abate after discontinuing the statins (7, 8). Taking a different approach, we recently identified 16 patients with a necrotizing myopathy who had a novel autoantibody that recognized ∼200-kd and ∼100-kd proteins (9). Given that these patients had proximal muscle weakness, elevated creatine kinase (CK) levels, class I major histocompatibility complex (MHC) molecule expression on nonnecrotic muscle fibers, autoantibodies, and responded to immunosuppression, we concluded that these patients had an immune-mediated necrotizing myopathy (IMNM). Of the 12 anti–200/100-kd antibody–positive patients who were over the age of 50 years, 10 had previously been exposed to statins (83%). This was significantly higher than the frequency of statin exposure in age-matched controls with polymyositis (PM), dermatomyositis (DM), and inclusion body myositis (IBM).

We reasoned that statin-associated autoimmune myopathy with anti–200/100-kd autoantibodies provides a model system for defining the mechanistic relationship between drug exposure and the development of a specific autoimmune response. Identification of the autoantigen(s) targeted by the immune response is a critical first step.

In the present study, we demonstrated that statin exposure up-regulates the expression of the ∼200-kd and ∼100-kd autoantigens. Given that statin exposure also increases the expression of the ∼97-kd HMGCR (10, 11), we investigated whether this enzyme might be the ∼100-kd autoantigen. Serum from anti–200/100-kd–positive patients specifically recognized the intracellular catalytic domain of HMGCR. Based on this finding, we developed an enzyme-linked immunosorbent assay (ELISA) to rapidly screen patient sera for anti-HMGCR autoantibodies. We found that of 750 patients with symptoms of muscle disease, 45 (6%) were anti-HMGCR positive. Anti-HMGCR–positive patients had an IMNM phenotype and >90% of those over the age of 50 years had previously been exposed to statins. Interestingly, HMGCR expression was up-regulated in regenerating muscle fibers from anti-HMGCR–positive patients. These findings suggest that statins trigger an autoimmune response against HMGCR by up-regulating the expression of this autoantigen. Even after discontinuing statins, the presence of high levels of HMGCR in regenerating muscle fibers may perpetuate the immune response.

PATIENTS AND METHODS

Patients and genotyping.

Between May 2002 and April 2010, 750 patients in whom myopathy was suspected, as defined by proximal muscle weakness, elevated CK levels, myopathic findings on electromyography (EMG), muscle edema on magnetic resonance imaging (MRI), and/or myopathic features on muscle biopsy, were enrolled in a longitudinal study. Patients were defined as having PM or DM if they had probable or definite disease according to the Bohan and Peter criteria (12, 13) and as having IBM if they met the Griggs et al criteria for possible disease (14). Serum was available from each subject, and DNA samples were available from 260 subjects. Serum samples from 20 healthy control subjects without prior statin exposure were also obtained. All subjects were enrolled in protocols approved by the Johns Hopkins Institutional Review Board. Genotyping of the rs4149056 C allele was performed using the appropriate verified TaqMan Drug Metabolism Genotyping Assay (Applied Biosystems) on all 17 anti-HMGCR–positive patients for whom DNA samples were available (see Table 1 for detailed clinical information).

Table 1. Clinical features of the 45 patients who were positive for anti-HMGCR by ELISA*
SerumStatin useHMGCR ELISAAge at onset, yearsSexRaceHighest CKProximal weaknessEMGMuscle biopsyrs4149056 genotype
  • *

    The cutoff value for a positive result on the enzyme-linked immunosorbent assay (ELISA) for 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR) antibodies was 0.215 absorbance units, representing 3 SD above the mean in 20 healthy subjects who had never taken statins. Statin use represents the period prior to serum testing. Creatine kinase (CK) values are expressed as IU/liter. Electromyography (EMG) findings were categorized as normal, irritable myopathy (IM) or nonirritable myopathy (NIM). Muscle biopsy findings were categorized as necrosis plus inflammation (N + I), necrotizing myopathy (NM), or necrosis plus rimmed vacuoles (N + RV). Genotyping for rs4149046 was performed on 17 anti-HMGCR–positive patients for whom DNA samples were available. NA = not applicable.

07039No0.96949MB20,000YesNot doneN + I 
07056No0.749<40FW6,323YesIMNM 
07090No1.30457MW10,310YesIMNM 
08024No1.12332FB7,225YesNIMNM 
08038No0.34736MW4,071YesNIMN + I 
08050No1.26021FB17,967YesIMNM 
08109No0.84968MW3,275YesIMNM 
08126No1.37840FA13,506YesIMN + ITT
08196No1.52442FB35,000YesIMNot doneTT
08209No0.94745FW8,500YesIMNMCT
09029No0.7654FB16,000YesNIMNMTT
09063No0.98220FW2,000YesNAN + I 
09088No0.62947FB22,733YesIMNA 
10029No0.92416FA16,000NoNormalNM 
09184No1.75938MW17,976YesIMN + I 
03004Yes1.25958MB24,714YesIMNMTT
05017Yes1.22854MW13,600YesNot doneNM 
06031Yes0.71871MW3,052YesIMNM 
06061Yes0.54754FW15,000YesIMNM 
07054Yes0.35543MW11,427YesIMN + I 
07094Yes0.94848FW200YesNANot done 
07109Yes0.94244FA11,200YesNIMNM 
08001Yes0.24275FW8,602YesIMNM 
08040Yes1.15957FB3,993YesIMNM 
08076Yes1.25970MW8,800YesIMNMCC
08089Yes0.76847FB17,000YesIMNMTT
08100Yes0.37857FW8,000YesIMNM 
08130Yes0.75162MW16,500YesIMNM 
08144Yes0.28765MW254NoNot doneNot done 
08145Yes1.41154FW17,000YesIMNM 
08148Yes0.60865MW5,800YesNAN + I 
08176Yes1.14266FW6,000YesIMNMTT
08227Yes0.96649MW7,000YesNIMNM 
09125Yes0.51756FW1,876YesNIMN + I 
09135Yes0.74658FW3,000YesNIMNMTT
09153Yes1.27365MW4,197YesIMNMTT
09170Yes0.55680FW1,200YesNIMNM 
09172Yes1.49553FW6,840YesIMNMTT
09176Yes1.00070MW8,800YesIMNMTT
09188Yes1.99665MW4,065YesIMNMCT
09190Yes1.48649FW3,700YesNANMTT
10009Yes0.73666MW5,000YesNIMNMTT
10044Yes1.81062MW11,600YesIMN + RVTT
10062Yes0.29260FW4,000YesNANATT
10072Yes1.16954FW4,000YesIMNM 

Immunoprecipitations from radiolabeled cell lysates.

HeLa cells were cultured in the absence or presence of 10 μM mevinolin (Sigma) for 22 hours and were then radiolabeled with 100 μCi/ml of 35S-methionine/cysteine (MP Biomedicals), lysed, and immunoprecipitated with patient sera (9). Immunoprecipitates were reduced, boiled, subjected to electrophoresis 10% sodium dodecyl sulfate–polyacrylamide gels, and visualized by fluorography.

Immunoprecipitations using 35S-methionine–labeled in vitro transcription/translated (IVTT) proteins.

DNA encoding full-length human HMGCR was purchased from Invitrogen. DNA encoding the N-terminal piece (amino acids [aa] 1–377) was generated by mutating R377 to a stop codon. DNA encoding the C-terminus of HMGCR (aa 340–888) was prepared by polymerase chain reaction using the full-length DNA as a template. Constructs were sequence verified and used in IVTT reactions (Promega), generating 35S-methionine–labeled proteins. Immunoprecipitations using these products were performed, with detection of the immunoprecipitates as described above.

Competition experiments.

One microliter of each patient serum was preincubated (30 minutes at 4°C in 50 μl) with the catalytic domain of human HMGCR (aa 426–888) expressed as a fusion protein with glutathione S-transferase (hereinafter referred to as “C-terminal HMGCR”; Sigma). Preincubated antibodies were subsequently used for immunoprecipitations with full-length IVTT HMGCR or radiolabeled lysates made from mevinolin-treated HeLa cells.

Anti-HMGCR ELISA.

ELISA plates (96-well) were coated overnight at 4°C with 100 ng of C-terminal HMGCR (Sigma) diluted in phosphate buffered saline (PBS). Replicate wells were incubated with PBS alone. After washing the plates, human serum samples diluted 1:400 in PBS with 0.05% Tween were added to the wells for 1 hour at 37°C. After washing, horseradish peroxidase–labeled goat anti-human antibody (1:10,000; Pierce) was added to each well for 30 minutes at 37°C. Color development was performed using SureBlue peroxidase reagent (KPL) and the absorbance at 450 nm was determined. For each sample, the background absorbance from the PBS-coated wells was subtracted from that of the corresponding C-terminal HMGCR–coated well. Test sample absorbance was expressed as a proportion of the absorbance in an arbitrary positive control sample (sample 9176), a reference serum included in every ELISA.

Immunohistochemistry.

The collection and use of human biopsy specimens was approved by the Johns Hopkins Institutional Review Board. Muscle biopsy specimens from 6 patients with anti-HMGCR antibody and 3 normal control subjects were studied. All biopsy specimens were obtained from patients who had not taken statins for >3 months. Staining of paraffin sections was performed as described previously (9). Antibody incubations comprised mixtures of rabbit anti-HMGCR (Millipore) and mouse anti–neural cell adhesion molecule (anti-NCAM; Santa Cruz Biotechnology) primary antibodies, followed by donkey anti-rabbit IgG Alexa Fluor 594 (to detect HMGCR) and donkey anti-mouse IgG Alexa Fluor 488 (to detect NCAM) secondary antibodies (Invitrogen).

RESULTS

Up-regulation of 200-kd and 100-kd autoantigen expression by statins.

We previously demonstrated that sera from a group of patients with IMNM immunoprecipitate ∼200-kd and ∼100-kd proteins from radiolabeled HeLa extracts (9). Given the strong association of statin use with the development of these anti–200/100-kd autoantibodies, we labeled HeLa cells with 35S-methionine/cysteine after pretreatment for 24 hours with either 10 μM mevinolin or vehicle (DMSO) alone. To validate the protein equivalence of these lysates, immunoprecipitations were performed using antibodies against Mi-2 or PM-Scl. As anticipated, equal amounts of Mi-2 and the 5 protein components of the PM-Scl complex were detected in each lysate type (results not shown). In contrast, 3-fold–increased levels of both the 200-kd and the 100-kd protein were immunoprecipitated from the mevinolin-treated cells, demonstrating that levels of these autoantigens are up-regulated by statins (Figure 1A).

Figure 1.

Up-regulated expression of the 200-kd and 100-kd autoantigens by statins and identification of the 100-kd autoantigen as 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR). A, Radiolabeled lysates generated from HeLa cells treated for 24 hours in the absence (lane 1) or presence (lane 2) of 10 μM mevinolin were immunoprecipitated with patient serum 9190, as described in Patients and Methods. B, 35S-methionine–labeled full-length in vitro transcription/translated (IVTT) HMGCR was immunoprecipitated using sera from anti–200/100-kd–positive patients (lanes 3-7; representative of 16 anti–200/100-kd–positive serum samples tested), anti–200/100-kd–negative patients with dermatomyositis (lanes 8–10), or healthy controls (lanes 11–13). The input IVTT product is shown in lane 14. Results in A and B are representative of at least 3 separate experiments. Molecular weight markers are shown at the left.

Identification of the 100-kd autoantigen as HMGCR.

Goldstein and Brown (10) originally demonstrated that the expression of HMGCR is up-regulated by statin treatment. Morikawa and colleagues (11) extended these findings to muscle cells. They used DNA microarray analysis to demonstrate that statins induce the expression of 19 genes in a human skeletal muscle cell line, most of which are related to cholesterol biosynthesis (11). Among these, we selected HMGCR as a candidate for the ∼100-kd autoantigen because of its ∼97-kd molecular weight.

35S-methionine–labeled HMGCR was generated by IVTT and used in an immunoprecipitation assay with serum from 16 patients with anti–200/100-kd autoantibodies, as well as serum from 6 negative control subjects, consisting of 3 DM patients and 3 normal individuals without statin exposure. We found that serum samples from anti–200/100-kd–positive patients immunoprecipitated HMGCR, whereas serum samples from the control groups did not (Figure 1B).

Human anti-HMGCR autoantibody recognition of the C-terminus of HMGCR.

HMGCR is a membrane protein with a small extracellular domain, 7 membrane-spanning domains, and an intracellular catalytic domain. To define the region(s) of the protein recognized by sera from patients with anti-HMGCR antibodies, we synthesized 35S-methionine–labeled full-length HMGCR protein, an N-terminal fragment including the extracellular and membrane-spanning domains (aa 1–377), and a C-terminal fragment including the intracellular portion of the molecule (aa 340–888). Serum from anti-HMGCR–positive patients consistently immunoprecipitated full-length HMGCR and the C-terminal fragment, but not the N-terminal fragment (Figure 2). When anti-HMGCR–positive sera were preincubated with increasing concentrations of unlabeled C-terminal HMGCR prior to immunoprecipitation of 35S-methionine–labeled full-length HMGCR protein, immunoprecipitation was abolished (Figure 3A). Taken together, these findings demonstrate that anti-HMGCR autoantibodies recognize the intracellular C-terminal portion of this enzyme.

Figure 2.

Immunoprecipitation (IP) of full-length 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR) and a piece corresponding to the C-terminus (amino acids 340–888) by human anti-HMGCR antibodies. Immunoprecipitations were performed using 3 different 35S-methionine–labeled HMGCR products: full-length (FL; lanes 4–8), C-terminus (C-term; lanes 9–13), and N-terminus (N-term; lanes 14–18). Serum samples 10009, 9190, and 8050 are from anti–200/100-kd–positive patients; samples 488 and 495 are from normal control subjects. Input in vitro transcription/translated (IVTT) products are shown in lanes 1–3; in each case, 0.4 times the amount used for the immunoprecipitation was used. Results are representative of 2–8 separate experiments. Molecular weight markers are shown at the left.

Figure 3.

Competition immunoprecipitation (IP) experiments, confirming that human anti–3-hydroxy-3-methylglutaryl-coenzyme A reductase (anti-HMGCR) antibodies detect the C-terminus and that the 200-kd protein is not recognized by a unique autoantibody. A, Serum samples 10009 and 9190 were preincubated with the indicated amounts of unlabeled C-terminal HMGCR and then used to immunoprecipitate full-length 35S-methionine–labeled HMGCR. IVTT = in vitro transcription/translated. B, Serum samples 9190 and 9176 were preincubated in the absence or presence of 300 ng of unlabeled C-terminal HMGCR and were subsequently added to radiolabeled lysates generated from HeLa cells treated with 10 μM mevinolin for 24 hours. The resulting immunoprecipitates were processed as described in Patients and Methods. Identical data were obtained in 2 separate experiments using 4 (A) or 6 (B) different patient sera. Molecular weight markers are shown at the left.

No recognition of the 200-kd protein by a unique autoantibody.

To determine whether serum from anti-HMGCR–positive patients includes distinct autoantibodies that recognize the 200-kd protein, we performed immunoprecipitations from 35S-methionine-labeled, mevinolin-treated HeLa cell extracts, again preincubating with purified C-terminal HMGCR protein (Figure 3B). This procedure inhibited the immunoprecipitation of both HMGCR and the ∼200-kd protein, suggesting that the ∼200-kd protein is either coimmunoprecipitated with HMGCR or is an HMGCR dimer.

Validation of a new ELISA for the detection of anti-HMGCR autoantibodies in patient sera.

To screen patients rapidly for anti-HMGCR autoantibodies, we developed an ELISA. We defined a serum sample as being positive for anti-HMGCR if the relative absorbance value was 3 standard deviations or higher than the mean value in 20 healthy control subjects who had never taken statins. Using this method, we found that all 16 of the anti–200/100-kd–positive serum samples previously identified by immunoprecipitation from HeLa cell extracts were anti-HMGCR positive. In contrast, none of 33 patients with DM (including 5 who had previously taken statins) and none of 31 patients with IBM (including 11 who had previously taken statins) were anti-HMGCR positive (data not shown).

Next, we used the HMGCR ELISA to screen serum samples from all 750 patients enrolled in our longitudinal study of patients at the Johns Hopkins Myositis Center between May 2002 and April 2010. Of these, 45 patients (6%) were anti-HMGCR positive by ELISA (Table 1). To validate the ELISA, we compared ELISA and IVTT immunoprecipitation data obtained using a subset of sera from this cohort that were collected from 307 consecutive unique patients between January 2009 and April 2010. In this subgroup, 17 anti-HMGCR–positive patients were identified by both methods. The ELISA identified 1 additional anti-HMGCR–positive serum that was negative by immunoprecipitation (serum 10029). Since this patient had a necrotizing myopathy with elevated CK levels, we concluded that this was a true anti-HMGCR–positive serum and not a false-positive serum. These results demonstrate a very high correlation between these 2 methods and validate the ELISA test as a reliable, efficient screen for detecting anti-HMGCR autoantibodies.

Clinical features of anti-HMGCR–positive patients.

Of the 45 anti-HMGCR–positive patients, 30 (66.7%) had previously taken statins (Table 1). Among the 26 patients who presented to our clinic at age 50 years or older, 24 had taken statins (92.3%). Thus, the prevalence of statin use in patients with anti-HMGCR autoantibodies is significantly higher than what we and others have previously reported in age-matched patients with other myopathies (ages ≥50 years), including DM (25%), PM (36.8%), and IBM (33.3%) (8, 9).

Anti-HMGCR–positive patients were characterized by proximal muscle weakness (95.6%), elevated CK levels (mean ± SD 9,718 ± 7,383 IU/liter), and myopathic findings on EMG (97.3%) (Table 2). All of the 40 available muscle biopsy samples (100%) were reported to have prominent degenerating, regenerating, and/or necrotic fibers. Significant inflammatory infiltrates were noted in 8 of 40 muscle biopsy samples (20%) and rimmed vacuoles were visualized in 1 of 40 biopsy specimens (2.5%); this patient had predominantly proximal muscle weakness and did not have clinical features typical of IBM. Patients who had not taken statins were clinically indistinguishable from those who had, except for their younger age (mean ± SD 37 ± 17 years versus 59 ± 9 years), higher CK levels (13,392 ± 8,839 versus 7,881 ± 5,875 IU/liter), and race (46.7% versus 86.7% white) (Table 2).

Table 2. Clinical features of the 45 patients with anti–3-hydroxy-3-methylglutaryl-coenzyme A reductase antibodies*
CharacteristicAll patientsStatin-naive patientsStatin-exposed patientsP
No. (%) of patientsTotal no. assessedNo. (%) of patientsTotal no. assessedNo. (%) of patientsTotal no. assessed
  • *

    NS = not significant; EMG = electromyography.

  • Age and creatine kinase (CK) are reported as the mean ± SD.

  • Statin-exposed versus statin-naive patients.

White33 (73.3)457 (46.7)1526 (86.7)300.012
Male19 (42.2)455 (33.3)1514 (46.7)30NS
Age, years52 ± 164537 ± 171459 ± 930<0.0001
CK, IU/liter9,718 ± 7,3834513,392 ± 8,839157,881 ± 5,875300.0164
Myopathy on EMG36 (97.3)3712 (92.3)1324 (100)24NS
 Irritable myopathy27 (72.9)379 (69.2)1318 (75)24NS
 Nonirritable myopathy9 (24.3)373 (23.1)136 (25)24NS
Proximal weakness43 (95.6)4514 (93.3)1529 (96.7)30NS
Necrosis on biopsy40 (100)4013 (100)1327 (100)27NS
Inflammation on biopsy8 (20)405 (38.5)133 (11.1)270.11

While 43 of 45 anti-HMGCR–positive patients had no other systemic autoimmune disease (95.6%), patient 8196 had Jo-1 antibodies and interstitial lung disease. Another patient (patient 8038) had scleroderma, anti–PM-Scl antibody, and interstitial lung disease. Neither of these patients had taken statins prior to developing muscle symptoms.

The vast majority of anti-HMGCR–positive patients had clinical features consistent with an immune-mediated myopathy. However, a single patient (patient 8144) presented with only persistent myalgias after statin use, normal subjective and objective muscle strength, unremarkable findings on MRI of both thighs, normal findings on EMG, and a CK level of only 254 IU/liter.

No increased prevalence of the single-nucleotide polymorphism associated with statin myopathy in anti-HMGCR–positive patients.

A recent study published by the Study of the Effectiveness of Additional Reductions in Cholesterol and Homocysteine (SEARCH) Collective demonstrated that the presence of a specific polymorphism in the SLCO1B1 gene (i.e., the rs4149056 C allele) is strongly associated with the development of statin myopathy (15). This gene encodes the organic anion-transporting polypeptide OATP-1B1, which regulates the hepatic uptake of statins. While the prevalence of the C allele in their population of ∼12,000 participants (mostly of European ancestry) was 0.15, its prevalence in those who developed a statin myopathy within 1 year of starting simvastatin at a dosage of 80 mg/day was 0.54.

DNA samples were available from 17 of our anti-HMGCR–positive patients, and the frequency of the rs4149056 C allele in this population was 0.12. When the 6 patients who had not taken statins and/or were of non-European ancestry were excluded, the prevalence of the C allele in the remaining 11 patients was 0.14. Although the number of subjects genotyped was small, the prevalence of the rs4149056 C allele in these anti-HMGCR–positive patients is consistent with the range of 0.14–0.22 previously reported among those of European ancestry (15).

Up-regulation of HMGCR in regenerating muscle fibers.

To directly examine HMGCR expression in vivo, we stained muscle biopsy sections with polyclonal HMGCR antibodies. Because other myositis-associated autoantigens are expressed at high levels in muscle cells with features of regeneration (16, 17), we costained sections with NCAM, an established marker of muscle regeneration. In muscle biopsy specimens showing normal features, HMGCR and NCAM were expressed at relatively low levels (Figures 4D–F). In contrast, NCAM positive fibers were prominent in muscle biopsy samples obtained from anti-HMGCR–positive patients (who had not taken statins for months to years). Interestingly, most of these NCAM-positive fibers also expressed high levels of HMGCR (Figures 4A–C). These findings provide in vivo confirmation that regenerating muscle fibers from anti-HMGCR–positive patients express high levels of HMGCR.

Figure 4.

Up-regulation of 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR) expression in regenerating myofibers expressing neural cell adhesion molecule (NCAM). Muscle biopsy samples from anti-HMGCR–positive patients (A–C) and control subjects (D–F) were costained with anti-NCAM antibodies (green) (A and D), anti-HMGCR antibodies (red) (B and E), and DAPI (blue) to stain nuclei. Overlay images (C and F) demonstrate that HMGCR and NCAM are frequently coexpressed at high levels in the same myofibers in anti-HMGCR–positive muscle biopsy tissues (arrows) but not control muscle biopsy tissues. To ensure comparability, images A–C and D–F were obtained using identical exposure settings for each channel. Results are representative of the staining seen in 6 anti-HMGCR–positive and 3 normal muscle biopsy samples. Original magnification × 20.

DISCUSSION

Statins are a widely prescribed class of medications with known adverse effects on muscles, usually mild. We recently described novel autoantibodies that recognize ∼200-kd and ∼100-kd proteins associated with autoimmune myopathy and statin use (9). In the present study, we demonstrated a plausible causal link between statin exposure and this distinct form of IMNM through identification of the autoantigen as HMGCR. Immunoprecipitation assays demonstrated the specificity of the autoantibodies for the carboxy-terminus of this enzyme, while competition experiments confirmed that anti-HMGCR autoantibodies immunoprecipitated both HMGCR and the ∼200-kd protein. The larger protein may be an associated protein or a multimer of HMGCR. The latter possibility is supported by other studies showing that HMGCR can be immunoprecipitated as a 97-kd monomer and as a ∼200-kd dimer (18).

Having identified HMGCR as the relevant autoantigen, we developed an ELISA to rapidly screen patient sera. Using this ELISA, we found the prevalence of anti-HMGCR autoantibodies to be 6% among patients with suspected myopathy who presented to the Johns Hopkins Myositis Center. Extending our previous studies, we found that anti-HMGCR autoantibodies are preferentially found in patients with a necrotizing myopathy on muscle biopsy and were not found in patients with IBM, DM, or normal controls (9). Thus, anti-HMGCR autoantibodies are one of the most frequent autoantibodies in our cohort, second only to anti–Jo-1 (19). Since necrotizing myopathy is not always immune mediated, the detection of anti-HMGCR by ELISA may be diagnostically helpful for the identification of patients with this form of IMNM, the majority of whom respond to immunosuppressive therapy (9).

Among the 45 anti-HMGCR–positive patients, one had Jo-1–positive antisynthetase syndrome (2.2%), and another had scleroderma with anti–PM-Scl autoantibodies (2.2%). Therefore, as with other forms of autoimmune muscle disease, patients with anti-HMGCR autoantibodies may, in rare cases, have an overlap syndrome with another connective tissue disease.

Importantly, we have demonstrated that muscle expression of HMGCR is increased with statin exposure (11), as well as in regenerating muscle cells marked by NCAM expression. This suggests that immune-mediated muscle damage initiated in the presence of statins and associated with anti-HMGCR autoantibodies may be sustained even after the statin is discontinued, through persistently increased HMGCR expression associated with muscle repair.

Since most patients taking statins do not develop an immune-mediated myopathy, other factors, including genetic susceptibility, must also play a role. The most common genetic factor predisposing patients to self-limited statin myopathy is the presence of the rs4149056 C allele, which accounts for up to 60% of statin myopathies in patients taking 80 mg of simvastatin daily (15). This polymorphism most likely increases the risk of myopathy by decreasing the hepatic uptake of statins by the OATP-1B1 transporter. However, this genetic alteration was not overrepresented in anti-HMGCR–positive patients, suggesting that other genetic susceptibilities or environmental coexposures are required for the development of the autoimmune response.

Interestingly, 33% of the anti-HMGCR–positive patients had not previously taken statins. Although these patients were younger at the time of disease onset and had higher CK levels, they also had an apparently immune-mediated myopathy and were otherwise indistinguishable from those with statin exposure. We hypothesize that other genetic and/or environmental factors may cause high levels of HMGCR expression in these patients.

Because our clinic patients are referred with weakness and other prominent features of myopathy, this study does not address how prevalent anti-HMGCR autoantibodies are among patients taking statins who have milder symptoms. However, we did identify one anti-HMGCR–positive patient with persistent statin-induced myalgias who had no other compelling clinical evidence of myopathy. This suggests that an autoimmune response may also be associated with low-grade myopathic symptoms in some patients. Future studies will determine how frequently patients with self-limited statin myopathy and statin-induced myalgias develop anti-HMGCR autoantibodies.

AUTHOR CONTRIBUTIONS

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. Dr. Mammen 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. Mammen, Chung, Christopher-Stine, A. Rosen, Casciola-Rosen.

Acquisition of data. Mammen, Chung, Christopher-Stine, P. Rosen, Casciola-Rosen.

Analysis and interpretation of data. Mammen, Chung, P. Rosen, A. Rosen, Doering, Casciola-Rosen.

Acknowledgements

We thank Dr. Jennifer Mammen for editing the manuscript, Dr. Katherine Pak for technical assistance, and Drs. Thomas Lloyd and Sonye Danoff for enrolling patients.

Ancillary