In this medical research, where we used identifiable human sera and data, we obtained agreement from the French Ministry of Research (AC 2008-87) for the collection, analysis, storage and reuse. All samples were obtained for diagnostic purposes only with no extra sampling and, in this situation of retrospective study, patients’ consent was not required. The research protocol was approved by research ethics committee of Pitié-Salpêtrière University Hospital.

Serum or blood samples (n = 150 patients) that were sent to the Laboratory of Immunology of Rouen University Hospital for analysis of anti-HMGCR aAbs because of suspicion of NAM were analyzed. These samples were obtained between August 2012 and January 2013 from 24 French centers and were all negative for anti-Jo-1 and anti-SRP aAbs. All patients who were found positive for anti-HMGCR aAbs were negative for anti-PL-7 and anti-PL-12 aAbs. For these anti-HMGCR-positive patients, the following parameters were recorded from their medical reports: age, sex, past medical history, statin exposure, characteristics of myopathy including clinical manifestation and muscular deficit evaluation by muscle manual testing using the Medical Research Council scale on five points, and creatine kinase (CK) level at the date of the anti-HMGCR assay. Special emphasis was performed by the clinician to check statin exposure during follow-up medical consultations. For correlation experiments, we additionally analyzed 25 anti-SRP-positive sera that were described previously 16 . The term SRP in text and figures refers to the 54 kDa subunit of the multimeric SRP complex.

Control sera were collected from 100 healthy blood donors and from 142 patients with different inflammatory/autoimmune diseases, according to established classification criteria: American College of Rheumatology (ACR) revised criteria for systemic lupus erythematosus (SLE) 18 with anti-double-stranded DNA (dsDNA) aAbs (n = 19), American Rheumatism Association (ARA) criteria for rheumatoid arthritis (RA) 19 with anti-cyclic citrullinated peptide (CCP) antibodies and/or rheumatoid factor (n = 18), revised European criteria for primary Sjögren’s syndrome 20 with anti-SSA and/or anti-SSB aAbs (n = 15), Bohan and Peter criteria 21 for dermatomyositis (DM) (n = 8), Troyanov criteria for overlap myositis 10 with anti-tRNA-synthetase Ab (anti-Jo-1, n = 27; anti-PL-7, n = 1; anti-PL-12, n = 2) and Griggs criteria for inclusion body myositis (IBM) 22 (n = 30). Another control consisted in sera from patients with polyclonal hypergammaglobulinemia (serum immunoglobulin (Ig)G: 23.1 ± 6.9 g/L, mean ± standard deviation (SD)) (n = 24). All serum samples were stored at -80°C until use.

The catalytic C-domain (amino acids 441 to 880) of human HMGCR protein fused to a glutathione S-transferase (GST) tag was obtained from Sigma-Aldrich (St Louis, MO, USA). Purity of this recombinant HMGCR protein was first determined by 4 to 10% gradient sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) under nonreducing conditions, followed by Coomassie Blue staining. Western blot analysis was further performed by transfer of proteins separated by nonreducing SDS-PAGE to a nitrocellulose membrane followed by incubation with rabbit anti-HMGCR Abs (Abcam, Cambridge, MA, USA) or a human anti-HMGCR-positive serum and revelation with corresponding secondary antibody coupled to Alexa Fluor 680 (Invitrogen, Cergy Pontoise, France).

Immunoprecipitation was performed by adding 20 μl of human serum sample or 1 μg of positive control rabbit anti-HMGCR Ab to 10 μg of recombinant HMGCR C-domain and bringing the volume to 0.5 ml of IP buffer (1% Nonidet P40, 20 mM Tris pH 7.4, 150 mM NaCl, 1 mM EDTA) and rotating the mix for 2 h at 4°C. Protein G agarose beads (Pierce, Rockford, IL, USA) were added for an additional hour. Then, immunoprecipitates were boiled in the presence of Laemmli buffer and subsequently electrophoresed on 4 to 10% gradient SDS-PAGE under reducing conditions. Immunoprecipitated HMGCR was revealed with a rabbit anti-HMGCR Ab coupled to Alexa Fluor 680 and read on an Odyssey Li-Cor (Li-Cor Biosciences, Lincoln, NE, USA).

We previously described an ALBIA for anti-SRP antibodies (ALBIA-SRP) 16 . Here, to develop immunoassay for anti-HMGCR antibodies (ALBIA-HMGCR), a similar strategy was applied with some modifications. Five to 12 μg of recombinant HMGCR C-domain were coupled to 1.25 × 10⁶ fluorescent Bio-Plex^™ COOH-microspheres (Bio-Rad, Hercules, CA, USA) with the Bio-Plex^™ amine coupling kit (Bio-Rad) according to manufacturer’s protocol. After optimization to maximize the signal/noise ratio, the quantity of HMGCR protein per coupling reaction was set at 10 μg. After coupling, coated beads were either used immediately or stored at -20°C in the dark. When indicated, a human recombinant SRP or intrinsic factor protein (Diarect AG, Freiburg, Germany) was used in place of HMGCR for controls.

Immediately prior to use, HMGCR-coated beads were vigorously agitated for 30 s. Then, a 10 μl volume containing 1,250 beads was added to 100 μl of serum from patients or controls (diluted in Dulbecco's phosphate-buffered saline (DPBS) plus 1% fetal bovine serum) in Multiscreen 96-well plates (Millipore, Bedford, MA, USA). Plates were incubated for 2 h at room temperature in the dark on a plate shaker at 650 rpm. Blank (no serum, secondary antibody only), negative control (anti-HMGCR negative serum) and positive controls (highly positive human anti-HMGCR Ab serum or rabbit anti-HMGCR Abs with appropriate secondary antibody) were included in every assay. Beads were collected by filtration under vacuum and washed twice with 150 μl DPBS containing 0.1% Tween-20. Biotinylated mouse anti-human IgG- or isotype (IgG1, IgG2, IgG3 or IgG4)-specific secondary Ab (Southern Biotech, Birmingham, AL, USA) were added at 1/2,000 dilution and incubated for 1 h at room temperature under shaking. After washing, beads were incubated with 50 μl of streptavidin-R-phycoerythrin (Qiagen, Venlo, The Netherlands) at 1/1,000 dilution for 15 min. Finally, beads were resuspended in 100 μl of DPBS and mean fluorescence intensity (MFI) was determined on a Bio-Plex^™ apparatus using the Bio-Plex^™ Manager software 4.0 (Bio-Rad).

The analytical sensitivity of detection of ALBIA-HMGCR was determined by triplicate dosage of serial dilutions of rabbit anti-HMGCR antibodies. For each well, after subtraction of negative control MFI value, mean MFI and standard deviation (SD) were calculated.

The analytical specificity of detection of ALBIA-HMGCR was determined by using high-level human anti-HMGCR, anti-SRP or anti-intrinsic factor positive sera at 1/1,000 dilution. Sera were incubated with purified HMGCR, SRP or intrinsic factor beads for 1 h at room temperature and MFI was determined as described above.

Specific inhibition of ALBIA-HMGCR was performed using increasing concentrations of recombinant HMGCR protein, with 1/2,000 dilution of the anti-HMGCR serum used as 100% reference. Percent inhibition was calculated as (1 - (MFI pre-adsorbed serum/MFI serum)) × 100. Homologous and heterologous inhibition of ALBIA-HMGCR was further performed by pre-absorption of three anti-HMGCR, three anti-SRP or three anti-intrinsic factor positive sera with 100 μg/ml of recombinant HMGCR, SRP or intrinsic factor protein.

All patients' sera were initially assayed at a 1/500 dilution. The anti-HMGCR Ab levels were determined at a 1/D dilution using the following formula: (MFI^serum/MFI^calibrator) × (level of calibrator) × D/500. The calibrator is a human highly positive anti-HMGCR serum (the same throughout the study) whose level was arbitrarily set to 100 arbitrary units (AU)/mL). When sample MFI at 1/500 dilution was higher than 80% of calibrator MFI, further dilutions were performed and the first dilution yielding a MFI inferior to 80% of calibrator MFI was retained for calculation of level.

Reproducibility of ALBIA-HMGCR was determined using three independent anti-HMGCR Ab positive sera (low, medium and high level). Intra-assay variation was determined by measurements of the same samples within the same run. Inter-assay variation (assay to assay) was determined by measurements of the same samples in separate runs. Inter-assay variation (batch to batch) was determined by measurement of the same samples in the same run using different batches of coupled beads. Coefficients of variation (CV) were computed as SD/mean.

To detect simultaneously anti-SRP or anti-HMGCR aAbs from a single sample, we used two different beads harboring a special mixture of red/infrared dyes so that they are endowed with a specific spectral signature. Color codes of HMGCR- and SRP-coupled beads were number 55 and 28 (Bio-Rad), respectively. After coupling with HMGCR or SRP, 10 μl of each bead preparation were mixed and incubated with serum samples in the same conditions than described previously. Analytical specificity of ALBIA-NAM was further determined by analyzing sera from patients with anti-HMGCR or anti-SRP aAbs or from healthy controls or patients with different inflammatory/autoimmune conditions including RA, systemic sclerosis, SLE, DM, anti-tRNA synthetase aAb positive myositis and IBM, or polyclonal hypergammaglobulinemia. Levels of anti-SRP and anti-HMGCR sera were compared with those obtained with the corresponding monoplex ALBIA-HMGCR or ALBIA-SRP.

The Myositis Euroline 7 (IgG) line immunoassay was kindly provided by Euroimmun (Lübeck, Germany). It consists in a membrane strip coated with different antigens, including two different recombinant HMGCR proteins. The first consists of the HMGCR C-domain fused to GST and the second a human cDNA expressed in Escherichia coli. The assay was used according to the manufacturer’s instruction and the intensity of coloration was graded from one to four.

Indirect immunofluorescence was performed on HEp-2000™ cells (ImmunoConcepts, Sacramento, CA, USA) and Kallestad HEp-2 cells (Bio-Rad). Sera were tested at 1/80 screening dilution in PBS buffer, using a FITC-coupled antibody against human IgG.

The normality of data distribution was analyzed using the D'Agostino and Pearson test, and the receiver operating characteristic (ROC) curve was determined using Prism software (GraphPad Software, La Jolla, CA, USA).

To allow quantitative analysis of anti-HMGCR aAbs in patients, we developed an addressable laser bead immunoassay (ALBIA-HMGCR) 16 23 . For this, we used as antigen the recombinant catalytic C-domain of the human HMGCR protein fused to GST. Purity of this protein was confirmed by Coomassie blue staining after SDS-PAGE, revealing a unique band (Figure 1A) that was specifically recognized by rabbit anti-HMGCR Abs or a human anti-HMGCR positive serum in Western blot (Figure 1B).

HMGCR was coupled through a covalent link to fluorescent beads and used to measure the levels of anti-HMGCR Abs. The development of ALBIA-HMGCR first required optimizing concentration of recombinant HMGCR protein used to coat beads, incubation times, washing steps and revelation with the secondary antibody (not shown, see Material and Methods for details). This yielded a method allowing detection of as low as 5 ng/mL anti-HMGCR Abs (Figure 1C), indicating a very good analytical sensitivity of the assay.

The specificity of HMGCR-coated beads was next evaluated. For this, HMGCR, SRP and intrinsic factor-specific beads were prepared and analyzed using a human anti-HMGCR, anti-SRP or anti-intrinsic factor positive serum. Only anti-HMGCR but not irrelevant anti-SRP or anti-intrinsic factor Ab-positive sera reacted with HMGCR beads (Figure 2A). Similarly, beads coated with SRP or intrinsic factor recombinant proteins that were produced and purified using the same process that for HMGCR protein reacted only with the corresponding aAb-positive serum.

Addition of free homologous protein, that is HMGCR, resulted in a dose-dependent inhibition of the HMGCR signal (Figure 2B) while no inhibition was observed when using a heterologous protein, that is SRP or intrinsic factor, at the concentration of 100 μg/mL (Figure 2C). To further confirm the antigenic specificity of the assay, five sera that scored positive for anti-HMGCR aAbs using ALBIA-HMGCR were used to immunoprecipitate the recombinant HMGCR protein. This yielded a positive result in 5/5 cases but not when using an irrelevant serum (Figure 2D).

The method used for calculating anti-HMGCR levels is illustrated in an additional figure (see Additional file 1). The serum used in this example showed a saturating signal at the 1/500 screening dilution. A further 1/4,000 dilution was retained to calculate the level by reference to a highly positive anti-HMGCR serum used as calibrator whose level was arbitrarily set to 100 AU/mL.

Reproducibility and repeatability of the test were determined by calculating both intra- and inter-assay variation for sera with high, medium and low anti-HMGCR levels. The intra-assay coefficients of variation (CV) ranged from 5% for high/medium levels to 13% for low levels, as indicated in an additional table (see Additional file 2). Similarly, the mean CV for inter-assay variation, as determined from plate-to-plate or batch-to-batch variations ranged from 9 to 11%, and 2.8 to 5.6%, respectively. Together, these data indicate a good reproducibility and repeatability of the ALBIA-HMGCR assay.

We used ALBIA-HMGCR to screen serum samples for the presence of anti-HMGCR aAbs. First, a threshold of positivity was calculated using 100 control sera from healthy blood donors. The values obtained with control sera did not follow a normal distribution (P <0.0001), their median was 0 AU/mL and the 99^th percentile was 14 AU/mL. At a cutoff of 14 AU/mL, 99/100 control sera scored negative (Figure 3A).

Next, we analyzed sera from 150 patients with suspicion of NAM. Of these, 37 patients (24%) were anti-HMGCR positive, with levels ranging from 24 to 2.656 AU/mL. ROC curve analysis revealed a specificity of 100% and a sensitivity of 100% when the cutoff was set at 20 AU/ml (Figure 3B). Patients with different inflammatory/autoimmune conditions including RA, systemic sclerosis, SLE, DM, anti-tRNA synthetase Ab positive myositis or IBM, as well as patients with polyclonal hypergammaglobulinemia all scored negative (Figure 3A). Anti-HMGCR positive and 16 negative sera from healthy donors were further tested by a recently developed line immunoassay. All 37 ALBIA-HMGCR positive sera were positive using this assay while 16/16 controls were negative (not shown). Yet, there was no correlation between band intensity and ALBIA-HMGCR level.

Sixty-one percent (20/33) of ALBIA-HMGCR positive sera displayed a finely granular cytoplasmic pattern on a minority (3% or less) of HEp-2000® cells (Figure 4). This immunoreactivity could be inhibited by previous incubation with free HMGCR. Using Kallestad HEp-2 cells, only 30% of the ALBIA-HMGCR positive sera yielded this aspect (not shown).

Fifteen ALBIA-HMGCR positive sera were tested for anti-HMGCR IgG subclasses. Only IgG1 were represented among anti-HMGCR aAbs in this series (not shown). IgG2, IgG3 or IgG4 subclasses were never found, neither as a single isotype nor in association with IgG1.

In order to simultaneously quantify anti-HMGCR and anti-SRP aAbs, we next developed a multiplex assay (ALBIA-NAM) and compared its results to that of monoplex ALBIA-HMGCR and ALBIA-SRP 16 . For this, we assayed 26 anti-HMGCR positive patients from the present series plus 25 anti-SRP positive sera that were described previously 16 . Results from both assays were compared by linear regression and showed excellent correlation, with R values of 0.9942 and 0.9937 for ALBIA-HMGCR versus ALBIA-NAM (Figure 5A) and ALBIA-SRP versus ALBIA-NAM (Figure 5B), respectively.

ALBIA-NAM revealed able to perfectly discriminate the two populations of NAM patients, that is those with anti-HMGCR from those with anti-SRP aAbs (Figure 5C) and was negative for all other tested conditions, that is patients with different inflammatory/autoimmune conditions, DM, anti-tRNA synthetase Ab positive myositis or IBM, as well as patients with polyclonal hypergammaglobulinemia (see Additional file 3). The sensitivity of monoplex and multiplex assays was equivalent. No NAM patient was positive for both aAbs.

Characteristics of anti-HMGCR positive patients are summarized in Table 1. These patients presented with proximal muscle weakness (92%), had elevated (mean >6,000 IU/L) CK levels and 60% had not taken statins. Muscle biopsies always showed regenerating and necrotic muscle fibers, with occasionally (28%) perivascular inflammatory infiltrates.

HMGCR, 3-hydroxy-3-methylglutaryl coenzyme A reductase; SD, standard deviation.