Ultrasensitive Interferons quantification in idiopathic inflammatory myopathies serve as biomarkers of activity in dermatomyositis and anti-synthetase syndrome

Objectives Inflammatory idiopathic myopathies (IIM) are a heterogeneous group of disorders, ranging from a muscle-specific autoimmune disease to a systemic one that are difficult to assess. Recent insights into IIM pathogenesis highlighted the role of interferon (IFN) in the pathophysiology. The aim of this study was to test if IFN serum levels can a use as a biomarker of disease activity in IIM. Methods IFN type I and II were measured using an ultrasensitive detection technology and assess the potential of IFN. Results One hundred and fifty-two patients (dermatomyositis (DM); n=50, anti-synthetase syndrome (ASyS); n=46, immune-mediated necrotizing myopathy (IMNM); n=32, inclusion body myositis (IBM); n=24) and 33 age-matched healthy donors were included. IFN-α levels were higher only in DM (0.07 pg/ml [0.03-0.23], p<0.005) and ASyS groups (0.07 [0.02-0.16], p<0.05) compared with controls (0.02 [0.01-0.05]). IFN-β was increased only in DM and IFN-γ among all IIM. IFN-α levels were correlated with disease activity in DM (r=0.76, p<0.0001). The predictive accuracy of IFN-α level to discriminate active and non-active disease was excellent as reflected by an area under the ROC-curve of 0.88. Using an IFN-α level cut-off above 0.11 pg/ml, the sensitivity was 75% and the specificity was 96% in DM patient. IFN-α and IFN-γ were correlated with disease activity in ASyS groups (r=0.55 and r=0.46 p<0.05)).


Introduction 4
Idiopathic inflammatory myopathies (IIM) are a heterogenous group of autoimmune diseases including four main groups: dermatomyositis (DM), anti-synthetase syndrome (ASyS), immune-mediated necrotizing myopathies (IMNM), and inclusion body myositis (IBM) (1,2). IIM can either be a musclespecific autoimmune disease (IBM and IMNM) or systemic involving mainly the skin, the joints and/or the lungs (DM and ASyS). The development of tools assessing disease activity is crucial in daily clinical practice to improve patients' care, as well as improving the design of randomized clinical trials. International Myositis Assessment and Clinical Studies Group developed a disease activity core set measures permitting to define an improvement using a total improvement score calculated with two timepoints measures (3). Nevertheless, reliable biomarkers are needed to assess the disease activity at one given timepoint. While creatine kinase (CK) level (one core set measure) is well correlated with IMNM disease activity (4), it may lack sensitivity in ASyS or DM patients (5).
Interferons (IFN) are involved in the pathophysiology of IIM in which an overexpression of IFNstimulated genes is observed (6,7). There are three main types of IFN. The type I IFN (IFN-α and β) blood levels are increased in DM (8, 9) whereas the role of type I IFN in ASyS has not been yet clarified. In IBM, large amounts of type II IFN (IFN-γ) are produced by activated CD8 + T cells that play key role in muscle fibers damages (10,11). Serum level of type II IFN have not been evaluated in IIM yet. In addition, a new ultrasensitive technology has recently been developed and permits to detect very low level of proteins suitable for cytokine detection at femtomolar concentration such as IFN-α (12).
The aim of this study was to assess the potential of type I and II IFN, using an ultrasensitive digital ELISA technology, as a new blood biomarker of activity for IIM especially in DM and ASyS.
Sera were collected at diagnosis and/or during the follow-up and were rapidly (< 3 h) frozen after one centrifugation. All the sera were thawed only once to avoid potential freeze/thaw effects. Patients who had increased dose of corticosteroids (> 0.5 mg/kg and/or pulses) the week before the sampling were excluded as it may rapidly abrogate the IFN levels (16). Moreover, patients with active infectious diseases (e.g. flew or viral B hepatitis) were excluded. Thirty-three age-matched healthy donors (HD) from a French blood bank were used as negative controls.
The limit of detection (LOD) was 0.0035 pg/ml for IFN-α and 0.026 pg/ml for IFN-γ. The positivity 6 threshold was defined as the mean plus three times the standard deviation of the 33 healthy donors (HD) and was 0.22 pg/ml for IFN-α and 1.97 pg/ml for IFN-γ.
For IFN-β serum quantification, we developed a homebrew assay using Simoa® reagents and mAbs from (PBL Assay Science, Piscataway, NJ, USA), as recently described (Llibre et al, In Press). The LOD was calculated by the mean value of the blank plus two times the standard deviation (positivity at 95% confidence) calculated on logarythmic values and was 1.24 pg/ml and the positivity threshold was defined by the mean plus three times the standard deviation of the HD and was 2.50 pg/ml.

Statistical analysis
Quantitative variables were expressed as median with inter-quartile range, and numbers with proportions for categorical variables. Multiple comparisons were performed using Kruskal-Wallis test then Dunn's post-hoc test for quantitative data. To analyse the correlation between IFN and disease activity assessed by the PGA, we performed Spearman's rank correlation tests using Graphpad prism 5. Positive threshold to discriminate active from inactive patient was assessed by ROC curve analysis.
We used the point of the curve nearest the top of the left-hand corner to minimise both the number of false-positive and false-negative results. CK and IFN values were transformed through a base-10 logarithm for analysis. After verifying the absence of multicolinearity, we included IFN-α, IFN-γ and CK levels in binary multivariate logistic regression to determine the association with disease activity (binary outcome using PGA > 5 to define active patients). P < 0.05 was considered statistically significant. Multivariate statistical analyses were realized with R program version R-3.6.0.

Patients' characteristics
One hundred fifty-two IIM patients (DM, n = 50; ASyS, n = 46; IMNM, n = 32 and IBM, n = 24) and 33 healthy donors were included. Main patients' characteristics are shown in Table 1.  No difference was observed in the therapeutic profile, including the use of corticosteroids and immunosuppressors, between IMNM, DM, and ASyS while IBM patients did not receive any treatment.  (Fig. 1c).

Longitudinal analysis
Focusing on DM and ASyS treatment-naive patients at diagnosis, we performed a serial longitudinal analysis (DM, n = 6/11 and ASyS, n = 4/11) (supplementary Fig. 1 and Fig. 3). The majority of DM 9 patients had a high type I IFN level at diagnosis and a decrease parallel to the clinical improvement ( Fig. 3a, Fig. 3b and supplementary Fig. 1). IFN-γ level wasn't associated with the disease activity in follow-up of DM (Fig. 3c). In most ASyS patients increased levels of IFN-α and IFN-γ but not IFN-β were observed at the diagnosis ( Supplementary Fig. 2). Similarly, these levels decreased following the clinical improvement.

Discussion
In this study, we showed that type I IFN is a reliable biomarker of disease activity in DM and ASyS patients. While IFN-α is increased in both conditions, IFN-β is only increased in DM patients. Type II IFN (i.e IFN-γ) is increased in all myositis subgroups but ASyS patients showed the best correlation with the disease activity.
Previous studies have described an IFN signature in muscle, skin and blood samples of DM patients (20)(21)(22). The IFN-stimulated genes levels can be assessed by an IFN score (6) combining a set of at least five genes, but it is not standardized and performed routinely in clinical practice, and requires RNA extraction. The IFN score in DM was only correlated with the cutaneous disease activity (9).
Digital ELISA for type I IFN is more sensitive than ELISA and was very well correlated with IFN-gene signature (9,17). For the first time, we were able to show a very good correlation with DM disease activity parameters and IFN-I levels. This technology was recently used in lupus with interesting data (19). Only one previous study used Digital ELISA to measure IFN-α blood level in adult DM (9). It did not show a significant correlation with disease activity but in this study, only the skin disease activity was assessed whereas we included all domains of disease activity (e.g skin, muscle and joints). In this study, IFN-β only was correlated with the skin disease activity (9) in line with a previous study showing that IFN-β was increased in DM (8). In our study, we included a large cohort of myositis patients and we confirmed that IFN-β is a DM specific IFN cytokine and showed for the first time that both IFN-α and β are reliable biomarkers. Nevertheless, we have to underline that the limit of detection of IFN-α was 300 times lower than for IFN-β it might be possible that interest of IFN-β level monitoring will increase when the limit of detection will be improved.
In addition, for the first time, we demonstrated that IFN-α is also associated with disease activity in ASyS. Along that line, it was previously shown that anti-Jo1 positive patients harbored an IFN signature (23). If both DM and ASyS have an IFN signature, we observed that only DM have increased level of IFN-β and that IFN-γ was a good biomarker in ASyS but not in DM. It was shown that only DM 11 patients expressed in the muscles IFN-related proteins (24,25) while muscle fibers of ASyS overexpressed MHC-II (26), a type II IFN inducible protein (27). Along that line, a recent study showed that different IFN signature is found in muscle biopsy of IIM (28). Altogether these results highlight that pathways in both conditions are different: namely IFN-α and -β in DM and IFN-α and -γ in ASyS.
In IMNM, only IFN-γ was associated with disease activity but CK level was a better biomarker according to a previous study showing a strong correlation between the percentage of necrotic muscle fibers and CK level (4). Increased type II IFN levels may be due to Th-1 immune responses and CD8 + T cells or NK cells, two types of immune cells involved in pathophysiology of IBM and ASyS respectively (10,29), but also by macrophages (17)

Availability of data and materials
All data relevant to the study are included in the article or uploaded as supplementary information.
Data are available upon reasonable request.

Competing interests
All the authors declare no disclosure.

Funding
Authors declare no funding.