|Home | About | Journals | Submit | Contact Us | Français|
Muscle enzyme levels are insensitive markers of disease activity in juvenile and adult dermatomyositis (DM), especially during the active treatment phase. To improve our ability to monitor DM disease activity longitudinally, especially in the presence of immune modulating agents, we prospectively evaluated whether IFN-dependent peripheral blood gene and chemokine signatures could serve as sensitive and responsive biomarkers for change in disease activity in adult and juvenile DM.
Peripheral blood and clinical data were collected from 51 juvenile and adult DM subjects prospectively over 2 study visits. Disease activity measures, whole-blood type I IFN gene and chemokine score were collected. We also measured serum levels of other pro-inflammatory cytokines, including IL-6.
Changes in juvenile and adult DM global disease activity correlated positively and significantly with changes in the type I IFN gene score before (r=0.33, p=0.023) and IFN chemokine score before and after adjustment for medication use (r=0.53, p<0.001 and r=0.50, p=<0.001). Changes in muscle and extramuscular VAS subscales positively correlated with change in IFN gene and chemokine score (p=0.002). Serum levels of IL-6, IL-8 and TNFα were positively correlated with changes in global, muscle and extra-muscular VAS before and after adjustment for medications (p<0.05).
Our findings suggest that changes in type I IFN gene and chemokine scores as well as levels of IL-6, IL-8 and TNFα may serve as sensitive and responsive longitudinal biomarkers of change in disease activity in juvenile and adult DM, even in the presence of immunosuppressant use.
Juvenile and adult dermatomyositis (DM) are autoimmune disorders characterized by proximal muscle weakness, muscle inflammation, and a characteristic skin rash. Despite advances in our understanding of the pathogenesis of DM, disease activity monitoring is heavily dependent upon the physician’s clinical assessment. Few reliable indicators of disease prognosis, disease activity, or response to treatment have been identified. Traditionally, manual muscle strength testing and serum levels of muscle enzymes have been used as markers of disease activity; however, muscle strength may be impaired by disease damage (chronic scarring, fibrosis, or atrophy) rather than ongoing disease activity (1), while muscle enzymes are insensitive markers of disease activity (2).
Accumulating data from our group and others suggest that cells from the muscle tissue and blood of patients with DM carry distinct immune “fingerprints” (3). These studies have reported upregulation of genes related to type I interferon (IFN α/β) in both muscle tissue and peripheral blood of DM patients with active disease (4). Using a subset of type I IFN upregulated genes and chemokines, we developed scores that were strongly correlated with DM disease activity based on cross-sectional study (5). In aggregate, these data suggest that type I IFN inducible genes and chemokines may be sensitive measures of disease activity in DM.
In this study we tested the hypotheses that changes in prominent type I IFN “signatures”, reflecting both transcript upregulation and elevated serum proteins, will reflect changes in DM disease activity when studied longitudinally, independent of use of immune suppressive agents.
The study protocol was approved by the Human Subjects Institutional Review Boards at the University of Minnesota and the Mayo Clinic, and informed consent was obtained from each participant. Study participants were recruited among attendees at myositis clinics at Mayo Clinic between 2005–2010; all subjects met Bohan and Peter criteria for probable or definite juvenile (n=21) or adult (n=30) DM (6, 7). Juvenile DM subjects were less than 18 years of age at enrollment and less than 16 at age of diagnosis. There were no exclusions based on disease activity or medication use. Individuals with overlapping connective tissue disease, including systemic lupus erythematosus, scleroderma, Sjogren’s syndrome, or mixed connective tissue disease, were excluded. All study participants had their disease activity assessed and blood collected at the time of enrollment and at a subsequent follow-up visit. Thirty-four of the 51 subjects (67%) were included in a cross-sectional study previously reported (5).
Disease activity was assessed by evaluating rheumatologists (SY, SA, AR, FE), blind to IFN-gene signature, IFN chemokine scores or cytokine levels, using established disease activity tools for use in myositis clinical trials described by the International Myositis Assessment and Clinical Studies Group (IMACS) (8, 9). Muscle strength was assessed using the manual muscle testing of 8 muscle groups (MMT8) (10). The Childhood Muscle Assessment Scale (CMAS) was performed on all juvenile DM subjects over the age of 4 (11). Serum levels of muscle enzymes were measured, including creatine kinase (CK), aldolase, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and lactate dehydrogenase (LDH). Disease activity measures included the Myositis Disease Activity (MYOACT) portion of the Myositis Disease Activity Assessment Tool (MDAAT) and an assessment of other organ involvement using the MYOACT which utilizes separate 100 mm visual analogue scales (VAS) to gauge the physician’s evaluation of disease activity in several discrete domains. Involvement of all non-muscle organ systems (constitutional, cardiac, pulmonary, gastrointestinal, skeletal, and cutaneous) was also evaluated using a composite score (composite extra- muscular VAS score). Physician and patient global VAS scores, rated overall disease activity (8).
Antinuclear antibodies (ANA), anti-Jo-1 antibodies and other myositis specific autoantibodies, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) were all assayed in the clinical laboratory.
Whole blood was drawn into PAXgene tubes (Qiagen/Becton Dickinson, Franklin Lakes, NJ). Total RNA was isolated according to the manufacturer’s protocol with on-column DNAse treatment. RNA yield and integrity were assessed using an Agilent Lab-on-a-Chip Bioanalyzer (Agilent Technologies, Inc., Palo Alto, CA). The whole-blood type I IFN gene expression signature was defined by expression levels of 3 IFN-regulated genes (IFIT1, G1P2, and IRF7) as measured by TaqMan quantitative real-time RT-PCR (qPCR) using ABI Prism 7900HT Sequence Detection System (Appied Biosystems, Foster City, CA). Relative quantification of expression levels was performed following manufacturer’s guidelines with normalization against GAPDH and comparison against a calibrator sample (PAXgene whole blood RNA from a healthy control subject). Calculation of IFN gene scores was performed as previously reported (5). Briefly, expression levels were truncated at the 95th percentile value for each gene to reduce the effects of outliers and then normalized so that the maximum value for each gene was 1.0. The normalized expression values for the three genes were summed for each subject, and the sums were normalized to a 100-point scale for visualization (12).
Serum was isolated from blood drawn into serum-separator vacutainer tubes (Becton Dickinson, Franklin Lakes, NJ). A protease inhibitor (aprotinin, 1μg/mL) was added to each sample, and aliquots were immediately frozen at −80°C. Multiplexed sandwich immunoassays (Meso Scale Discovery, Gaithersburg, MD) were used to quantify serum levels of IFN-regulated chemokines (13) and other pro-inflammatory cytokines (Monokine induced by gamma interferon (MIG CXCL9), Macrophage Inflammatory Protein (MIP-1α/CCL3, MIP-1β/CCL4), tumor necrosis factor- α (TNF-α), TNF-receptor 1(TNF-R1), interleukin (IL)-10, IL-6, and IL-8). Samples were run in duplicate and calibrated recombinant proteins were used to generate standard curves. A summary chemokine score based on serum levels of IFN-inducible T-cell chemoattractant (I-TAC), IFN-inducible 10-kd protein (IP-10), and monocyte chemotactic protein 1 (MCP-1) was calculated for each participant in a manner similar to the calculation of the type I IFN gene score (12).
Descriptive statistics were used to summarize the characteristics of the study subjects. Comparisons between age groups were performed using chi-square and rank sum tests. Changes in disease activity between visits 1 and 2 were compared using paired t-tests. Our primary hypothesis was that the previously defined IFN CK and gene scores would be useful for assessing disease activity longitudinally, and the remainder of chemokines/cytokines in our panel would not be correlated with changes in disease activity. Given this focused hypothesis, adjustment for multiple comparisons was not necessary and was not performed. Spearman’s rank correlation coefficient was calculated to assess the relationship between biomarkers and continuous variables (disease activity scores, blood biochemical measurements). Spearman partial correlation methods were used to adjust the correlation coefficients for medication use at both study visits. Changes were calculated as visit 2 minus visit 1, so changes <0 indicated improvement and changes >0 indicated worsening. Similarly, for cytokine values, changes <0 indicated decreases in cytokine levels between visits 1 and 2 and changes >0 indicated increases in cytokine levels. Hence, positive correlations indicated that biomarker levels and disease activity were either both decreasing from visit 1 to visit 2 or both increasing from visit 1 to visit 2. Significance levels were set at p<0.05 for all tests. While we do not believe adjustment for multiple comparisons is warranted in this study, with 22 chemokines/cytokines and 2 comparisons each (visit 1 and change from visit 1 to visit 2) for each disease activity measure, the Bonferroni adjustment would consider p<0.05/44 =0.0011 to be statistically significant. Note due to the correlated nature of chemokines and cytokines, this adjustment would be overly conservative.
Responsiveness to change was assessed by correlating the change in biomarker levels with the change in disease activity measures. Another way to assess sensitivity to change is to determine whether the measure detects meaningful change when it occurs and stays stable when no change has occurred. To assess this, we compared the change in biomarker levels between patients who improved (decrease in disease activity of ≥10 units), remained stable (increase or decrease in disease activity of <10 units) or worsened (increase in disease activity of ≥10 units) using rank-sum tests.
Finally, we performed exploratory correlations between cytokine values and medication use. Multiple definitions of medication-use were utilized to ensure comprehensive assessment of medication effects on cytokine values. For these analyses, medications were defined as yes/no at visit 1, visit 2, and ever (either visit). In addition, each medication was assessed individually, and due to small numbers for some medications, we also assessed combinations of medications, such as any disease modifying antirheumatic medication.
The study population consisted of 51 study participants with juvenile DM (N=21) or adult DM (N=30) who had 2 study visits. The mean age± SD was 8 years (range: 2–17 years) for juvenile DM and 45 years (range: 18–77 years) for adult DM; 35/51 (69%) were female (Table 1). The majority of patients had short disease duration at visit one (median <3 months). Of the 51 participants, 18 were not on immunomodulatory medications at visit 1, 16 with untreated new-onset disease and 2 who were off medications as they were in remission. Medication use at visit 1 for the remaining 33 participants included: azathioprine (AZA) (n=4), methotrexate (MTX) (n=20), mycophenolate mofetil (MMF) (n=3), hydroxychloroquine (HCQ) (n=6), and corticosteroids (n=18). Only children with juvenile DM had CMAS evaluations at visit 1 and 2. The CMAS was abnormal (<52 range 51–32) in 15/21 juvenile DM despite near normal MMT assessments.
Changes in disease activity measures between visit 1 and visit 2 are shown in Table 2. Overall these patients improved significantly on all disease activity measures between the 2 study visits. However, the median change in muscle VAS score was zero, indicating that some patients showed no changed, despite a mean decrease in the muscle VAS score of approximately 14 mm.
Changes in IFN gene scores and IFN chemokine scores, as well as some of the individual cytokine/chemokine measures, were positively correlated with changes in global VAS, muscle VAS, and extra-muscular VAS (Table 3). Positive correlations were seen between IL-6, IP-10, I-TAC and MCP1 and changes in global, muscle and extra-muscular VAS measures, before and after adjustment for medication use (Figure 1). Changes in serum IL-8 and TNFα correlated positively with the changes in global and muscle VAS regardless of medication use (Table 3).
The potential influence of medication use on the cytokine values was also examined. Participants who were not taking medications at visit 1 were somewhat younger (median age: 16 years) than patients who were on medications (median age 37 years; p=0.09). Patients without medications at visit 1 had significantly higher disease activity (median global VAS score 52 vs. 19; p=0.013). Despite these differences, the close agreement between correlations without and with adjustment for medications suggests that medication use had little influence on the cytokine values.
In order to determine whether the changes in cytokine levels and IFN scores were sensitive to changes in disease activity, we compared changes in IFN gene and chemokine scores between patients who remained clinically stable between visit 1 and 2 (n=11) and patients whose disease improved (n=33). We found that patients who had no change in disease activity generally had little or no change (defined as increase or decrease in global disease activity of <10 units) in IFN chemokine and gene scores, whereas patients whose disease improved (defined as decrease of ≥10 units) also had decreased IFN chemokine and gene scores (p=0.04 for IFN chemokine score and p=0.11 for IFN gene score; Figure 2). Despite the fact that the chemokine score reached significance in this comparison while the gene score did not, overall we found that the chemokine score is correlated with the IFN gene score (r=0.65, p<0.001 at visit 1; r=0.60, p<0.001 at visit 2 and r=0.54, p<0.001 for change in IFN chemokine scores vs. change in IFN gene scores. Furthermore, we found that differences in creatine kinase are correlated with differences in both chemokine (r=0.46, p=0.006) and IFN gene (r=0.58, p<0.001) scores and also global VAS (r=0.54, p<0.001) and muscle VAS (r=0.62, p<0.001). Results were similar using the fold change instead of the absolute difference in creatine kinase (for chemokine score, r=0.49, p=0.003; for IFN gene score, r=0.55, p<0.001; for global VAS, r=0.57, p<0.001; for muscle VAS, r=0.61, p<0.001).
Finally, we tested whether cytokine levels at the first visit were associated with impending changes in disease activity. Serum levels of several cytokines measured at visit 1 were associated with change in disease activity between visits 1 and 2, including IL-6, IP-10, I-TAC, MCP-1, IL-8, TNFα, and MCP-2, as well as the IFN gene score and IFN chemokine score (Table 4). A negative correlation indicates that participants with higher levels of cytokines at visit 1 were more likely to have improvements in disease activity from visit 1 to visit 2, while participants with lower cytokine levels at visit 1 were more likely to worsen. The values of individual cytokines and chemokines, such as IL-6, IP-10, ITAC, and MCP-1, as well as the IFN gene and chemokine scores were lower at visit 1 if medications were in use. This suggests a suppression of the type I IFN response with concurrent medication, but when these values are compared to disease activity measures they remain strongly correlated.
In addition, associations between medications and cytokine values adjusted for changes in disease activity were also examined longitudinally. Addition of MMF (n=3) was associated with decreases in serum levels of IFN-γ (r=−0.39, p=0.01) and IL-10 (r=−0.43, p=0.004), and with increases in serum MCP-1 (r=0.30, p=0.04). Use of MTX (n=20) was associated with decreases in IP-10 levels (r=−0.33, p=0.03) and IFN chemokine scores (r=−0.42, p=0.005). Use of corticosteroids (n=18) was associated with decreases in IL-1β (r=−0.46, p=0.002), and use of AZA (n=4) was associated with elevations in IL-17 (r=0.55, p=0.01) suggesting varying potential immune pathways are altered with treatment yet the IFN chemokine and gene scores remain sensitive measures of disease activity.
Several autoimmune diseases, including juvenile DM, adult DM and SLE, exhibit elevated expression of type I IFN-regulated gene transcripts, chemokines and cytokines in both peripheral blood and target tissues (14–16). Traditional biomarkers of juvenile and adult DM include serum levels of muscle-derived enzymes including CK, aldolase, AST, ALT, and LDH. Elevation of muscle-derived enzymes including CK, while only elevated in 50–70% at disease onset, becomes less sensitive as a marker of disease activity with chronic disease (2). We undertook this study to evaluate the disease monitoring utility of novel biomarkers of disease based on peripheral blood IFN gene and chemokine scores and measurement of serum cytokine levels. Our major finding was that changes in juvenile and adult DM global disease activity correlated positively and significantly with changes in the type I IFN gene score before and IFN chemokine score before and after adjustment for medication use. Assessment of non-muscle involvement has been challenging in DM. We demonstrated that changes in muscle and extramuscular VAS subscales positively correlated with change in IFN gene and chemokine scores, suggesting that these markers may be useful as a non-invasive tool to assess extramuscular disease activity.
Several soluble factors in serum have been evaluated cross-sectionally as biomarkers of disease activity juvenile DM. Neopterin, a product released primarily by macrophages and monocytes on stimulation by interferon-γ, was found in increased concentrations in plasma and urine of children with DM compared with healthy controls (17) but generally did not correlate with serum derived muscle enzymes (18). Another marker used to measure disease activity in juvenile DM is von Willebrand factor (vWF), a cleaved product of ADAMTS13 activity, present on activated endothelial cells and is elevated in a subset of patients with vasculitis including juvenile DM however no correlation was made with active skin disease or to muscle strength, CPK, or aldolase (13).
Studies of phenotyping of peripheral blood lymphocytes as markers of disease activity have demonstrated lymphopenia in both adult and juvenile DM at the onset of active disease (19). Lymphocyte enumeration has not been fully studied in relation to disease activity, and existing data show inconsistent results. In adult DM, a decrease in total lymphocyte (CD3+) numbers as well as the number of cell subsets (CD4+ and CD8+) has been reported prior to the onset of treatment, with normalization after treatment is initiated. However, CD19+ B cells have been reported to be both decreased and increased in adult DM (19). Fewer data exist in juvenile DM, however, cross-sectional studies found association between disease activity and increased CD19+ B cells and a decrease in CD3−CD16+ and/or CD56+ natural killer cells. Prospective studies are lacking and the effects of medication on lymphocyte phenotyping are nonexistent with little data to suggest usefulness in monitoring of disease activity.
Due to poor sensitivity and specificity of conventional blood-based biomarkers of disease activity in DM, we previously conducted a cross-sectional study seeking candidate DM disease activity markers (5). We found that factors such as type I IFN-regulated genes, cytokines (IL-6 and IL-1) and chemokines (I-TAC, CXCL10, also known as IP-10; CCL2, also known as MCP-1; and CCL8, also known as MCP-2) are overexpressed in peripheral blood of juvenile and adult patients with DM and that they correlate with clinical measures of disease activity (5). MCP-1 previously was shown to be upregulated in the muscle tissue of adult DM (20). Both the IFN gene and chemokine scores were associated with Global VAS. This present study was initiated to examine the longitudinal responsiveness and sensitivity of these markers and their performance under condition of treatment with immunomodulating agents. This study demonstrates that both the IFN gene and chemokine scores are responsive and sensitive measures longitudinally even when we adjusted for medication use. The IFN chemokine score correlated with clinical global, muscle specific and extramuscular (primarily including skin, lung and joint) disease activity measures, before and after adjusting for immune modulating agents in both juvenile and adult DM.
The present study suggests use of both the chemokines and gene scores would be useful for monitoring patients with DM over time, even in the context of different immune suppressive agents that might be used. Many of these are candidate biomarkers for prediction of future changes in disease activity. Assessment of disease activity at times is difficult. We found these candidate biomarkers were associated with muscle disease activity, i.e. muscle fatigue, inability to perform daily living activity and elevated muscle enzymes, even with a normal MMT score and with active lung disease, arthritis and skin disease. In the juvenile DM population we also utilized the CMAS tool, finding that many children have abnormal CMAS when the MMT is normal. The CMAS incorporates functional assessments with strength with the potential of being more sensitive in this population. The CMAS has not been validated in adults and is not used universally as a muscle assessment tool.
Evaluation of these disease activity markers prospectively is key to defining their clinical utility. We have demonstrated that changes in both the IFN gene and chemokine scores correlate well with changes in disease activity over time, as do levels of selected serum chemokines (IP-10, I-TAC, MCP-1, and MCP-2) and cytokines (IL-6, IL-8 and TNF-α). That IL-8 and TNF-α each correlate with disease activity suggests further engagement of plasmacytoid dendritic cells and Th-1 cells and each were significantly correlate with global and muscle disease activity measures in particular. Increases in IFN-γ levels correlated over time with increase muscle disease activity and may represent more of a Th-1 pathway of continued disease activity. Furthermore, our finding that the levels of several cytokines and chemokines at the first visit were correlated with changes in disease activity at the second visit suggests that these may be candidate biomarkers for prediction of future disease course. Additional controlled, prospective studies will be required to assess the utility of these markers in predicting impending changes in disease activity.
Direct detection of type I IFN (IFN α and β) is routinely not seen in the peripheral blood in juvenile or adult DM. Additional reports of serum IFNα importance in juvenile DM include a functional measure of IFNα activity by measuring the downstream products (IFN-induced protein with tetratricopeptide repeats 1, myxovirus resistance 1, and RNA-dependent protein kinase) correlate with new onset disease (21). These IFNα induced proteins are reported to be higher in untreated than treated patients. However, these IFN-inducible proteins become less sensitive markers by 36 months of treatment, are not associated with disease activity but are were weakly associated with elevation of serum muscle enzyme levels (P<0.05) prior to the introduction of therapy.
One study did directly measure IFNα in adult DM and PM, and showed it was higher in patients with anti-Jo-1 antibodies and that medications did not significantly affect the IFNα levels. A negative correlation was found between IFNα and the intensity of MRI signal in muscle (22). Additional peripheral blood studies of IFN concentrations (IFNα, IFNβ and IFNω) measured by ELISA in adults with DM found that IFNβ elevation was seen in 35% (9/26 DM) compared with 6% (3/48) of other inflammatory myopathies (IBM+PM), and 6% (2/36) of healthy volunteers (23). The highest levels were in those subjects prior to treatment or with minimal treatment (prednisone dose of ≤ 15 mg/day or treatment duration ≤ 7 days). There is a concern that direct measurement of IFNα or IFNβ does not always reflect the full effect of type I IFN since it seen in low levels and may be short lived.
The use of the type I IFN-inducible gene and chemokine signatures reflects the cellular infiltrates seen in the muscle and skin tissue in active juvenile and adult DM including the recently identified plasmacytoid dendritic cell and IL-17 producing T cells (24–26). This along with the longstanding identification of CD4+ T cells and B cells in DM tissue, the proposed gene and chemokine scores reflect the ongoing inflammatory process in the target tissue, and thus may allow a more nuanced assessment of ongoing disease activity when used in conjunction with muscle enzymes and clinical measures.
This prospective evaluation of the IFN gene and chemokine scores supports the utility of these measures of disease activity in juvenile and adult DM. Of key importance is that unlike traditional disease activity measures, the IFN-based markers reflect changing disease activity over the duration of the disease even when adjusted for medication use.
Funding: This work was funded by State of Minnesota Partnership and National Institutes of Health, NIAMS and the Arthritis Foundation