Increased C reactive protein, cardiac troponin I and GLS are associated with myocardial inflammation in patients with non-ischemic heart failure

Inflammatory cardiomyopathy diagnosed by endomyocardial biopsy (EMB) is common in non-ischemic heart failure (HF) and might be associated with adverse outcome. We aimed to identify markers predicting myocardial inflammation in HF. We screened 517 patients with symptomatic non-ischemic HF who underwent EMB; 397 patients (median age 54 [IQR 43/64], 28.7% females) were included in this study. 230 patients were diagnosed with myocardial inflammation, defined as ≥ 7.0 CD3+ lymphocytes/mm2 and/or ≥ 35.0 Mac1 macrophages/mm2 and were compared to 167 inflammation negative patients. Patients with myocardial inflammation were more often smokers (52.4% vs. 39.8%, p = 0.013) and had higher C-reactive protein (CRP) levels (5.4 mg/dl vs. 3.7 mg/dl, p = 0.003). In logistic regression models CRP ≥ 8.15 mg/dl (OR 1.985 [95%CI 1.160–3.397]; p = 0.012) and Troponin I (TnI) ≥ 136.5 pg/ml (OR 3.011 [1.215–7.464]; p = 0.017) were independently associated with myocardial inflammation, whereas no association was found for elevated brain natriuretic peptide (OR 1.811 [0.873–3.757]; p = 0.111). In prognostic performance calculation the highest positive predictive value (90%) was detected for the combination of Global longitudinal strain (GLS) ≥ -13.95% and TnI ≥ 136.5 pg/ml (0.90 (0.74–0.96)). Elevated CRP, TnI and GLS in combination with TnI can be useful to detect myocardial inflammation. Smoking seems to predispose for myocardial inflammation.

Heart failure (HF) is a global health problem which affects approximately over 37 million people world wide 1 . According to data from the Framingham Heart Study, the lifetime risk of developing HF is estimated to be 20% for the ages between 40 and 80 years 2 . It is caused by structural, but also functional cardiac abnormalities, which result in loss of myocardial function 3 . However, in the absence of coronary artery disease, which is the leading cause of HF 4-6 , inflammatory cardiomyopathy is common, particularly in HF with reduced ejection fraction (HFrEF). Thereby, according to the report of the 1995 World Health Organization/International Society and Federation of Cardiology Task Force on the Definition and Classification of Cardiomyopathies, inflammatory cardiomyopathy is defined as cardiac dysfunction in co-prevalence with inflammatory disease of the myocardium established by immunological, histological and immunohistochemical criteria 7 . The exact incidence of inflammatory cardiomyopathy underlying HF is not known. However, according to postmortem analysis, inflammatory cardiomyopathy or myocarditis seems to account for approximately 40% of the sudden cardiac deaths in the young 8 . Regarding endomyocardial biopsy (EMB) studies myocardial inflammation can be diagnosed in 9 to 23% of the patients with non-ischemic cardiomyopathy 9,10 . According to the above mentioned definitions, inflammatory cardiomyopathy can only be diagnosed on basis of EMB results 11,12 . Although the results of studies addressing the immunosuppressive therapy of inflammatory cardiomyopathy remain controversial [13][14][15] , results of the TIMIC trial suggest that immunosuppressive therapy in patients with virus negative inflammatory cardiomyopathy may be an effective and safe option for recovery of cardiac function in addition to optimal medical therapy 16 . Inflammatory cardiomyopathy can progress rapidly and may require immediate immunosuppressive therapy to prevent adverse outcomes. Thus, we aimed to identify predictors of myocardial inflammation in HF patients in order to improve diagnostic and therapeutic management.

Methods
Study population. Patients  www.nature.com/scientificreports/ via the right or left femoral artery or via trans-radial cardiac catheterization as reported before 18 . In case of right heart biopsy, the femoral vein was used. Immediately after sampling the biopsy, the specimens were stabilized in a solution to preserve ribonucleic acid (RNA) integrity (RNAlater, Ambion Inc., Austin, Texas) and were sent for further examination to a specialized laboratory approved by Food and Drug Administration [Institut Kardiale Diagnostik und Therapie (IKDT), Berlin, Germany]. For immunohistological evaluation, heart muscle tissue probes fixated in RNAlater were embedded in Tissue Tec (SLEE, Mainz, Germany) and immediately snap-frozen in methyl butane which had been cooled in liquid nitrogen, and then stored at -80 °C until processing. Embedded specimens were cut serially into cryosections of 5 mm thickness and placed on 10% polyl-lysine-pre-coated slides. As part of a routinely performed work up in case of unclear heart failure histological and immunohistochemical examination as well as virus detection was conducted. Using specific antibodies inflammatory processes were detected by identifying immune cell infiltration and expression of cell adhesion molecules. The specimens of the patients in this study were tested for CD3-positive lymphocytes (Dako; dilution 1:25), CD11a + /LFA-1 + lymphocytes (ImmunoTools; dilution 1:250), macrophages Mac-1 macrophages (Immu-noTools; dilution 1:500), Perforin positive (cytotoxic T Cells) (clone dG9, BD Bioscience; dilution 1:150) and the expression of adhesion molecules HLA-1 (Dako; dilution 1:2000) and ICAM-1 (ImmunoTools; dilution 1:800

Statistical analysis.
Descriptive statistics for relevant baseline comparisons of symptomatic non-valvular, non-ischemic heart failure patients, who underwent EMB, stratified according inflammation (≥ 7.0 CD3+ lymphocytes/mm 2 and/or ≥ 35.0 Mac1 macrophages/mm 2 ) are provided as median and interquartile range (IQR) or absolute numbers and corresponding percentages. We tested the continuous variables of the groups (inflammation vs. no inflammation) using the Mann-Whitney-U test and categorical variables with the Fisher's exact or the chi 2 test, as appropriate.
Receiver operating characteristics (ROC) curves were calculated for TnI, BNP, CRP and GLS with regard to myocardial inflammation and the area under the curve (AUC) is presented with the corresponding 95% CI. Additionally, patient cohort-optimised cut-off values of these mentioned laboratory markers with regard to myocardial inflammation were calculated based on ROC analyses using Youden index quantification. The software SPSS (version 23.0; SPSS Inc., Chicago, Illinois) was used for computerized analysis. p values of < 0.05 (two-sided) were considered to be statistically significant.
Univariate and multivariate logistic regression models were analyzed to investigate predictors of myocardial inflammation in heart failure patients, thereby using patient-optimised cut-off values derived from the ROC analyses. Results are presented as odds ratio (OR) and 95%CI. The multivariate regression models were adjusted for (1) age and sex, obesity and (2) cardiovascular risk factors (CVRF) including history of smoking, arterial hypertension, diabetes mellitus and hyperlipoproteinaemia.

Results
Overall, 517 patients with symptomatic non-valvular, non-ischemic HF who underwent EMB between 2012 and 2018 were screened. Results from EMB with CD3+ and/or Mac1 cell count were available in 447 cases. 50 cases were excluded due to relevant virus activity (Fig. 1). Thus, 397 patients remained in this study and were analysed.   Fig. 3).

Predictors of myocardial inflammation.
ROC curves demonstrated only a moderate prognostic performance of CRP for prediction of myocardial inflammation with acceptable specificity and positive predictive values (Table 3, supplementary Figure 1). The computed best cut-off (according Youden-Index calculation) for CRP was 8.15 mg/l. In contrast, TnI, BNP and GLS were accompanied by a low prognostic performance to predict myocardial inflammation shown in the ROC curves. The computed best cut-off (according Youden-Index calculation) for TnI was 136.5 pg/ml, for BNP 1030 pg/ml and for GLS -13.95% (Table 3). Nevertheless, when      Fig. 4). However when combining laboratory parameters with echocardiographic parameters, a strong association with myocardial inflammation was found for the combined variable consisting of reduced GLS and elevated TnI (OR 9.633 [95%CI 2.027-45.769]; p = 0.004). Additionally, but to a smaller extent, the combination of TnI and CRP was also independently associated with myocardial inflammation (OR   Fig. 4).

Discussion
The key findings of the present study are that (I) elevated CRP and TnI levels are associated with inflammatory cardiomyopathy in patients with non-ischemic, non valvular heart failure; (II) TnI levels ≥ 136.5 pg/ml and CRP levels ≥ 8.15 mg/l are potential predictors of myocardial inflammation; (III) common heart failure biomarkers such as BNP and reduced LVEF were not predictive for myocardial inflammation and; (IV) GLS might add a benefit to predict myocardial inflammation when combined with TnI.
A growing body of evidence indicates that myocardial inflammation may lead to severe cardiac damage, deteriorate into dilated cardiomyopathy and be accompanied by poor prognosis. This implies that early diagnosis using predictive markers may avert poor outcome 20 , since immunosuppressive therapy can significantly improve the prognosis of HF patients with presence of myocardial inflammation 16 . Thus, readily available markers like the ones identified in our study might help to select patients for EMB and for shorter control follow-up examinations. Although cardiac MRI diagnostics has become increasingly important in the diagnosis of myocardial inflammation 21,22 , sensitivity and specificity of MRI findings are not unequivocal. Since EMB can be performed easily as part of coronary catheter examination 18 and should be carried out as early as possible if there is clinical or laboratory suspicion of inflammatory myocardial disease, our study results might help to accelerate the diagnostic work-up of HF.
The association of elevated CRP and TnI with myocardial inflammation was independent of age and sex, supporting earlier findings by Lauer et al., who reported that patients with elevated troponin T values were more likely to have EMB-proven myocarditis 23 . Because cardiac troponins are detectable in the blood for approximately one week 24 and have a half-life of 2h 25 , elevated troponins are suitable to indicate permanent myocardial cell injury 26,27 , which can be explained by inflammatory infiltration with consecutive cell damage documented in the EMB. Contrary to our results, Sramko et al. demonstrated that troponin was not able to distinguish between patients with idiopathic dilated cardiomyopathy or inflammatory dilated cardiomyopathy arguing that troponin indicates myocardial injury without indicating weather its due to myocardial inflammation or simply caused by myocardial wallstress related to heart failure 28 . However, the study sample size was small, and EMB was only performed on the right ventricle making a sampling error more likely.  www.nature.com/scientificreports/ Similar to positive troponin values, positive CRP values were associated with detection of inflammation in EMB. These results go in line with observation by Liu et al., who demonstrated that highly sensitive CRP levels were higher in patients with EMB-confirmed myocarditis than in patients without myocarditis 29 . However, it has to be pointed out that the sensitivity for highly sensitive CRP in detecting myocarditis was only 50.1% whereas the specificity was 80.7%. Since CRP is also increased in acute myocardial infarction 30 or in rheumatic valve disease 31 . CRP should not be interpreted alone, but in conjunction with cardiac troponin. Moreover, as the present study indicates, cut-off values for TnI and CRP seem to be higher with regard to myocardial inflammation compared to cut-off values used for diagnosis of acute coronary syndrom. Further studies are needed to prove those cut-off values in patients with myocardial inflammation.
Interestingly GLS was not able to distinguish between EMB proven myocardial inflammation and no inflammation. Escher et al. demonstrated before, that patients with acute myocarditis had reduced GLS values. After a follow-up period of 6.2 months patients with persistent inflammation had worse GLS values than patients without inflammation 32 . However in contrast to our study GLS values were compared when LVEF had already been improved. Eventhough it has been shown that GLS is a superior predictor of adverse cardiac events compared with LVEF 33,34 , studies demonstrate that in general and irrespectively of outcome prediction, GLS correlates well with LVEF in heart failure patients 35 . Therefore the results of our study suggest that GLS alone is not able to identify patients with myocardial inflammation in the presence of reduced LVEF. Nevertheless, the combined evaluation of GLS and Troponin values were the second strongest predictor of myocardial inflammation in multivariate regression analysis. Additionally, the best prognostic performance was found for the combination of GLS and Troponin. In particular, the high positive prognostic value of 90% to predict myocardial inflammation is promising. Thus, the combination of GLS and Troponin may help to identify patients with myocardial inflammation in the future and could add a benefit in the diagnosis and follow up treatment of those patients.
Interestingly, smoking was associated with myocardial inflammation in our study. In a very small study sample, it was shown, that in patients hospitalized for acute myocarditis smoking was the most prevalent cardiovascular risk factor 36 . Recently, Detorakis et al. reported a significant correlation of smoking habits with late gadolinium enhancement extent in cardiac MRI in patients with clinically suspected myocarditis 37 . Prolonged exposure to tobacco smoke may cause cardiovascular cell damage, suggesting that increased myocardial cell necrosis and cell death would make myocardial inflammatory infiltration more likely 38 .
Another finding of the present study was that patients with myocardial inflammation had lower hemoglobin levels. It is well known that anemia of inflammation or anemia of chronic disease is primarily a disorder of iron distribution 39 and even low grade inflammation is associated with lower hemoglobin levels 40 . As experimental models have demonstrated that low cardiac iron levels promote heart failure 41,42 the link between the observed lower hemoglobin levels could be an effect of possibly local iron distribution disorders due to myocardial inflammation. Since hemoglobin levels in the present study were generally not significantly decreased, more data are needed to support this observation.
The primary stimulus for natriuretic peptide synthesis by myocardial cells is related to increased LV wall stress induced for example by acute myocardial infarction 43 , arterial hypertension 44 , muscle hypertrophy 45 or increased pulmonary arterial pressure 46 . We expected that inflammation would also cause elevated BNP levels as reported earlier 47 , particularly in patients with myocarditis and dilated cardiomyopathy [48][49][50] . Increase of BNP has also been detected in systemic inflammation regardless of systolic function 51 . In the present study, no differences in BNP levels and LVEF were found in patients with and without myocardial inflammation, reflecting a rather compensated disease state of our study population with low rates of hydropic decompensation. Myocarditis and inflammatory cardiomyopathy are a dynamic processes; after inflammation has subsided, a reduced LVEF is due to secondary post-inflammatory fibrosis and myocardial cell death and scars 52 . As conditions with and without inflammation lead to changes of LVEF and BNP, the present study demonstrates that increase of BNP and the decrease of LVEF do not specifically reflect the presence of inflammation, but rather reflect a general loss of myocardial tissue with consecutive changes in hemodynamic conditions regardless weather inflammation can be detected or not.

Limitation
There are certain limitations of our study which need to be mentioned: Firstly, our study was conducted as a monocentric retrospective data analysis. Therefore, it has to be emphasized that the results of this study can only serve in terms of a hypothesis generating research and the results of the present study should be verified by prospective trials. Secondly, there is no generally applicable definition of myocardial inflammation. Thus, it might be difficult to adopt the results to other endomyocardial biopsy studies, but generating more information about inflammation process was one of the main aims of our study. Finally, a potential limitation regarding EMB interpretation might be the sampling error. Regarding this point, Hauck et al. demonstrated that sampling error was prevalent 45% in Ieft and up to 37% in right EMB 53 . The authors conclude that only positive results of EMB can be considered diagnostic. However, the cited study included a very small sample of only 36 patients.

Conclusion
Elevations of readily available markers like cardiac TnI and CRP as well as the combination of GLS and TnI may predict inflammatory cardiomyopathy and might be useful to select patients for EMB and for shorter control follow-up examinations. In the diagnostic approach to detect suspected myocardial inflammation, higher cutoff values concerning cardiac and inflammatory biomarkers may have to be applied. Smoking is associated with inflammatory cardiomyopathy in non-ischemic, non valvular HF patients and may indicate individuals at risk to develop inflammatory cardiomyopathy.

Ethics approval
All data were obtained from individuals enrolled between 2013 and 2018 in the retrospective monocentric Mainz Endomyocardial Biopsy in Heart Failure Study (My Biopsy-HF Study, DRKS #22178), which was approved by the Ethics Committee of Rhineland Palatinate to be in accordance with the legal regulations and the declaration of Helsinki.