Using serum troponins to screen for cardiac involvement and assess disease activity in the idiopathic inflammatory myopathies

Abstract Objectives Limitations in the methods available for identifying cardiac involvement and accurately quantifying disease activity in the idiopathic inflammatory myopathies (IIMs) may contribute to poor outcomes. We investigated the utility of different serum muscle damage markers [total creatine kinase (CK), cardiac troponin T (cTnT) and cardiac troponin I (cTnI)] to address these issues. Methods We assessed disease activity and cardiac involvement using the International Myositis Assessment and Clinical Studies Group core set measures in 123 participants with confirmed adult-onset IIM from the UK and Denmark. Total CK, cTnT and cTnI were measured. Associations were assessed using logistic regression and Spearman’s ranked correlation. Results Cardiac involvement (n = 18) was associated with higher cTnI levels, independent of overall disease activity [adjusted odds ratio 1.03 (95% CI 1.01, 1.05); P = 0.002]. An abnormal cTnI had the highest specificity and positive predictive value for cardiac involvement (95% and 62%, respectively). In those with a normal CK but elevated cTnT or cTnI, an association with increased disease activity scores was observed. Serum cTnT correlated with the physician (ρ = 0.39) and patient-assessed (ρ = 0.28) global visual analogue scales and HAQ (ρ = 0.41) more strongly than CK or cTnI levels. cTnT was the only marker to correlate with manual muscle testing scores (ρ = −0.24). Conclusion Serum cTnI testing may have a role in screening for cardiac involvement in IIMs. Abnormal levels of serum cTnT and cTnI are associated with increased disease activity, including in those with a normal CK.


Introduction
The idiopathic inflammatory myopathies (IIMs) are multisystem autoimmune diseases characterized primarily by skeletal muscle inflammation. Limitations in current methods for detecting early signs of organ-specific flares can delay timely commencement or escalation of treatment and may contribute to poor outcomes. Disease activity in IIMs can be quantified using the International Myositis Assessment and Clinical Studies Group (IMACS) core set measures (CSMs). These include the measurement of different serum muscle damage markers ("muscle enzymes"), total creatine kinase (CK), lactate dehydrogenase, aldolase or transaminases [1]. As total CK can be normal in patients with active IIM, the isolated use of this parameter can be misleading, but it remains common practice [2].
Cardiac involvement occurs in up to 75% of patients with IIM, is often subclinical, represents a major cause of morbidity and mortality and is sometimes only identified post-mortem [3]. Several methods are used in clinical practice to screen for cardiac involvement in IIM, including electrocardiography, echocardiography and contrastenhanced cardiac MRI [4]. So called cardiac-specific serum muscle damage markers [cardiac troponin T (cTnT) and cardiac troponin I (cTnI)] may have a role, but can lack specificity for cardiac disease in IIM, especially the cTnT, which is often elevated in those with skeletal muscle disease without cardiac involvement [5,6].
In this cross-sectional study, we evaluated the usefulness of serum cTnT and cTnI levels in screening for cardiac involvement and in quantifying disease activity in patients with IIM. Comparison is made with the use of serum total CK and the potential for using cTnT or cTnI as the second serum muscle damage marker in the IMACS disease activity CSMs assessment.

Cases
We performed an analysis of data pooled from two cohorts, one in the UK and one in Denmark (n = 123). Data regarding some of the Danish cohort has been previously published as part of separate studies [4,7]. Details of patient identification procedures at each site are contained in the supplementary data, section 'Patient identification', available at Rheumatology online.

Assessment
Each case was assessed using the IMACS disease activity CSMs. This included manual muscle testing-8 (MMT-8), HAQ disability index (HAQ-DI), physician and patient-assessed global disease activity visual analogue scales (VASs) and an extramuscular disease assessment using the Myositis Disease Activity Assessment Tool (MDAAT) [1].
Cases also supplied blood for measurement of serum total CK, cTnT and cTnI (supplementary data, section 'Laboratory assays', available at Rheumatology online). The results were not available to clinicians at the time of their performing clinical assessments. Normalization of CK and cTnI levels was performed to allow comparison of results from different assays (supplementary data, section 'Laboratory assays', available at Rheumatology online). A result was deemed abnormal if the value was above the upper limit of the laboratory reference range (for cTnT) or normalized reference range (for CK and cTnI).

Cardiac involvement
Cardiac involvement was determined as per the cardiac domain of the MDAAT, the agreed upon tool for evaluating the presence and extent of extramuscular involvement in IIM [8]. The MDAAT has been partially validated, performs well in terms of interrater reliability and is used in IIM clinical trials, including the recently published Rituximab in Myositis study [9]. In the current study, the assessment could include consideration of any recently performed cardiac investigations (e.g. electrocardiography or echocardiogram), but additional investigations were not performed. Cardiac involvement was deemed as present when the cardiac VAS score on the MDAAT was >0.

Statistical analysis
Data processing and analysis was performed using STATA for Windows version 13.0 (StataCorp, College Station, TX, USA). Normally distributed data were summarized by calculation of means and standard deviations (S.D.). Non-normally distributed data were summarized using medians and interquartile ranges (IQRs). Convergent construct validity of the serum muscle damage markers was assessed by Spearman's ranked correlation with other aspects of disease activity. A correlation coefficient (r) >0.70 was considered as representing a strong relationship, 0.250.70 a moderate relationship and <0.25 a weak relationship. Logistic regression was used to assess associations with cardiac involvement and abnormal serum muscle damage markers. All models were adjusted for cohort (i.e. UK or Denmark). The physician-completed global disease activity VAS was used to adjust for overall IIM disease activity where specified. P-values <0.05 were considered significant.

Ethics and consent
Full informed consent was obtained from all participating patients. The study was completed in full conformance with the principles of the Declaration of Helsinki. In the UK, ethical approval was obtained from the Health Research Authority (East Midlands Leicester Central Research Ethics Committee, reference 14/EM/1248). We also conformed to Good Clinical Practice and the laws and regulations of the UK. In Denmark, the study was approved by the local ethics committee (reference S-20100022).

Screening for cardiac involvement
Eighteen cases were deemed to have cardiac involvement [15% (18/121); MDAAT not completed in two cases] ( Table 1). Higher disease activity levels (physician global disease activity VAS, HAQ and extramuscular global VAS) were associated with cardiac involvement. No difference in diagnostic subtype, serotype, age, disease duration, gender or smoking status between those with or without cardiac involvement was detected.
The cTnI level correlated most strongly with the cardiac disease activity VAS (Table 2). An increased risk of cardiac involvement was associated with increasing serum levels of cTnT and cTnI, although this was independent of overall disease activity only for cTnI [adjusted odds ratio (OR) 1.03 (95% CI 1.01, 1.05), P = 0.002].
An abnormal cTnI had the highest specificity and positive predictive value (PPV) for cardiac involvement but the lowest sensitivity (Table 1). Cardiac involvement was only present in 23% (11/48) of those with an abnormal total CK and 21% (15/71) of those with an abnormal cTnT, compared with 62% (8/13) of those with an abnormal cTnI. When stratifying serum muscle damage marker levels according to the type of cardiac involvement identified (as per the MDAAT: myocardial, pericardial, arrhythmia or sinus tachycardia), only myocardial disease was associated with increasing levels of cTnT and cTnI [for cTnT: adjusted OR 1.00 (95% CI 1.00, 1.01), P = 0.006; for cTnI: adjusted OR 1.01 (95% CI 01.00, 1.02), P = 0.015] (supplementary Table S3, available at Rheumatology online).
Correlation between serum muscle damage marker levels and disease activity The cTnT levels correlated most strongly with the physician global VAS, the patient global VAS, the HAQ disability index (HAQ-DI) and the manual muscle testing score, outperforming CK and cTnI in all cases (Table 2). cTnT was the only serum marker to correlate with the MMT-8 score and cTnI was the only marker to correlate with the extramuscular global VAS. The total CK level had the weakest correlations with the physician global VAS, the patient global VAS and the HAQ-DI and showed no significant correlation with the extramuscular global VAS, the cardiac disease VAS or the MMT-8.

The added value of troponin testing
In 40% (29/73) of cases that had a normal total CK, the cTnT was elevated. In these cases, the abnormal cTnT was associated with lower MMT-8 scores [

Interpretation and implications
The cTnI levels were significantly increased in those with cardiac involvement and this was independent of overall disease activity. An abnormal cTnI had the highest PPV for cardiac involvement and was the only damage marker to correlate significantly with the extramuscular global VAS. This suggests that the cTnI level provides information about a different facet of IIM disease activity compared with CK and cTnT and is perhaps most useful as a screening tool for cardiac involvement. However, the low sensitivity for cardiac involvement implies that a normal cTnI level does not exclude it. In contrast, the low specificity of cTnT and CK for cardiac involvement may generate unnecessary concern about such involvement. Those with cardiac involvement had significantly worse overall disease activity, suggesting that clinicians should have a low threshold for considering the potential for cardiac involvement in patients with highly active IIM.
We have demonstrated that, of the muscle damage markers tested, cTnT levels correlated most strongly with IIM disease activity. We also identified significantly worse aspects of disease activity in those with normal CK but abnormal cTnT or cTnI, confirming the notion that the use of CK in isolation to assess IIM disease activity can be misleading.
Elevations of cTnT and the CK-MB fraction (both conventionally used as cardiac damage markers) are observed frequently in those with IIM in the absence of cardiac involvement [1012]. Potential explanations for this include the existence of subtle subclinical cardiac involvement, as supported by timekinetic analysis, which suggests that elevated cTnT in IIM is of cardiac origin in at least some cases [13].
Cross-reactivity of the cTnT assay with skeletal muscle troponin T (sTnT) is another possible explanation for elevations seen in IIM. However, very low levels of cross-reactivity are reported for the current generation of assays [12,14]. Finally, it has been shown that regenerating skeletal muscle fibres can express cTnT and CK-MB, thus elevations in IIM may actually originate from skeletal muscle undergoing repair rather than from myocardium [5, 14,15]. Importantly, as cTnI does not share this characteristic, elevated levels should be assumed to have arisen from the myocardium [16,17].

Strengths and limitations
This study included thorough assessment of cases using the IMACS disease activity CSMs toolkit and is the largest study to use this panel of serum muscle damage markers. Researchers were blinded to serum muscle damage marker levels during patient clinical assessments, thus minimizing the potential for bias. However, there are several limitations. The study lacked a gold standard to define cardiac involvement, relying on assessment using the partially validated MDAAT, without additional investigations being performed. However, during phase two of the MDAAT validation study, the sensitivity, specificity and PPV for active cardiac disease was 100% [8]. Despite this, our study may underestimate the presence of cardiac involvement, particularly subtle subclinical disease. It is also possible that subtle abnormalities on any investigations performed may not have been identified. Future studies may be improved by the use of skeletal muscle and cardiac MRI to understand the relative contributions to serum muscle damage marker elevations and the relationship between subclinical (e.g. purely radiologically evident) cardiac involvement and serum muscle damage marker elevations in IIM. While we did not include analysis of the intent-to-treat aspect of the MDAAT cardiac assessment, high correlation has been demonstrated between this and the severity assessment (r = 0.94) [8].
As data were pooled from separate studies, this creates the possibility of systematic bias influencing the results due to differing practices in each centre. However, similar proportions of cardiac involvement were identified in both cohorts [16% (7/44) in the UK and 14% (11/77) in Denmark]. We reported the PPV for cardiac involvement (i.e. the probability that subjects with an abnormal serum muscle damage marker truly have the cardiac involvement). Therefore, in settings where the prevalence of cardiac involvement and may differ, the PPV we defined may be less applicable.

Conclusion
We have demonstrated the potential benefits of measuring serum cTnI and cTnT levels in patients with IIM, enhancing the disease activity assessment and assisting with the identification of cardiac involvement.
Where an abnormal cTnI is encountered in IIM, this should alert one to the possibility of cardiac involvement. In such cases, and in patients with highly active IIM, clinicians should have a low threshold for arranging further investigations to evaluate this possibility. In cases with a normal CK but abnormal cTnT levels, significantly increased levels of disease activity were identified without an increased frequency of cardiac involvement.