Biomarkers of Myocardial Fibrosis are Associated with Diabetes but not with Coronary Microvascular Dysfunction in Women with Angina and No Obstructive Coronary Artery Disease

Background Coronary microvascular dysfunction (CMD) is highly prevalent in women with no obstructive coronary artery disease and possibly related to myocardial brosis caused by excessive extracellular matrix (ECM) remodeling. ECM turnover can be measured in blood indicating brotic activity. We hypothesized that women with DM, angina and no obstructive coronary artery disease have increased ECM turnover and that this is associated with CMD. Methods We included 344 women with angina pectoris and no obstructive coronary artery disease (187 with DM, predominantly type II) and 76 asymptomatic women without DM as controls. Biomarkers reecting formation of type IV and VI collagen (PRO-C4 and PRO-C6) and degradation of type IV, V and VI collagen (C4M, C5M, C6M), mimecan (MIM) and titin (TIM) were measured in all participants. CMD was dened as coronary ow velocity reserve (CFVR) <2.0 assessed by transthoracic Doppler echocardiography. triglycerides and


Background
Coronary microvascular dysfunction (CMD) is highly prevalent in women with angina pectoris and no obstructive coronary artery disease. CMD is a strong prognostic marker of cardiovascular morbidity and mortality, 1-3 is associated with cardiovascular risk factors, particularly diabetes mellitus (DM) type II and hypertension, 4,5 and is frequent in heart failure. 6 In CMD, transient ischemia occurs because the microvasculature cannot dilate in response to increased oxygen demand. 7 This condition may cause chronic low-grade ischemia which promotes cardiac extracellular matrix (ECM) remodeling. 8,9 The cardiac ECM consists primarily of collagens and proteoglycans preserving ventricular function and structure. 10,11 Imbalanced ECM remodeling induces accumulation of collagens, expansion of the extracellular volume and myocardial brosis. 12,13 Thus, CMD might be a precursor of myocardial brosis.
Myocardial brosis is associated with impaired ventricular function, remodeling and stiffness of the myocardium, 8,14 and is a characteristic of diabetic cardiomyopathy clinically presenting as heart failure with preserved ejection fraction (HFpEF).
Collagen and proteoglycan formation and degradation fragments can be quanti ed in blood and may be indicative of early brotic disease activity. 15,16 In a small study we have previously found that women with angina pectoris and no obstructive coronary artery disease have an imbalanced collagen turnover compared with asymptomatic controls. In this study, we investigated whether ECM turnover is associated with DM and CMD by examining seven biomarkers that have recently shown promising as markers of brotic activity.

Study population
The study-population was derived from the iPOWER (ImProve diagnOsis and treatment of Women with angina pEctoris and micRovessel disease) cohort conducted from May 2012 to December 2017. 17 In-and exclusion criteria in iPOWER have been published previously. 4 Brie y, inclusion criteria for the iPOWER study were angina pectoris and no obstructive coronary artery disease (de ned as less than 50% stenosis assessed by coronary angiography), left ventricular ejection fraction (LVEF) > 45% and no signi cant valvopathy. From 1830 patients included in iPOWER we selected all women with DM (n = 187) and a random sample of women without DM (n = 157).
A control group of 76 asymptomatic women without DM or previous cardiovascular disease were recruited from the Copenhagen City Heart Study 18 between June and September 2015. Blood biomarkers of collagen and proteoglycan turnover were successfully measured in all 420 participants.

Study assessments
We obtained demographic data from interviews. Clinical data included weight, abdominal circumference, blood pressure and heart rate measured at rest. Blood samples were analyzed for cholesterol levels (total, low-density lipoprotein [LDL] and high-density lipoprotein [HDL] cholesterol), and triglycerides and HbA1c.
Serum was collected for biomarker analysis and immediately stored at -80 ºC until analysis according to pre-de ned standard operating procedures.

Coronary microvascular function
Coronary microvascular function was assessed non-invasively by transthoracic Doppler stress echocardiography (TTDE) measuring the coronary ow velocity reserve (CFVR) of the left anterior descending artery. CFVR is the ratio of the peak diastolic ow at hyperemia to rest. We used high-dose dipyridamole (0.84 mg/kg) over 6 minutes to induce maximal hyperemia (29). All examinations were performed by the same 3 experienced echocardiographers in an unchanged setting using GE Healthcare Vivid E9 cardiovascular ultrasound system (GE Healthcare, Horten, Norway) with a 2.7-8 MHz transducer (GE Vivid 6S probe). A standard echocardiographic examination was conducted before the CFVR examination. Measurements of myocardial function at hyperemia were obtained immediately after termination of the dipyridamole infusion. After the examination, intravenous theophylline (maximum dose 220 mg) was administered. A detailed methods description of the standard echocardiography is given in Online Appendix.

Biomarkers of brotic activity
Tissue turnover was determined by neo-epitope biomarkers to assess the formation of type IV and VI collagen (PRO-C4 and PRO-C6) and degradation of type IV, V, VI collagen, mimecan and titin (C4M, C5M, C6M, MIM and TIM), respectively. (Table 1).
PRO-C4 and PRO-C6 are building stones of collagen type IV and VI and have been shown to correlate with DM and to be associated with a poor outcome in patients with HFpEF. [19][20][21] C4-6M are degradation fragments of collagen type IV-VI and have been linked to liver brosis, carotid atherosclerosis and major cardiovascular events. 22,23 Mimecan is a small proteoglycan, upregulated and released in heart diseases. 14 Titin, also known as Connectin, is a large sarcomeric protein responsible for elastic recoil and compliance. 24,25 Decreased titin content in the sarcomere leads to deposition of brotic tissue. 24,26 In the ECM remodeling process, mimecan and the cardiac isoform of titin are broken down to the neo-epitopes MIM and TIM, respectively. Both have shown promising results as blood biomarkers of brotic activity. 10,12,27 Biomarker analysis The biomarkers were quanti ed by ELISA assays produced at Nordic Bioscience, Herlev, Denmark.
Biomarkers were measured in serum samples from the symptomatic women with and without diabetes and from the asymptomatic controls. Brie y, ELISA assays were performed as following: Streptavidincoated microtiter plate was incubated with a biotinylated peptide for 30 min at 20 °C. Unbound biotinylated-peptide was washed off ve times with washing buffer (20 nM TRIS, 50 mM NaCl, pH 7.2). Subsequent, a selection peptide, ve kit control samples and patient serum samples were added to the plate and peroxidase-labelled monoclonal antibody was added and incubated for 1 h at 20 °C (PRO-C4, C4M, C6M, MIM and TIM), 3 h at 4 °C (C5M), or 20 h at 4 °C (PRO-C6). Subsequently all ELISA assays plates, after incubated with peroxidase-labelled antibody, were washed ve times with washing buffer and incubated with 3,3',5,5'-Tetramethylbenzidine (TMB) for 15 min at 20 °C in the dark. The reaction was stopped with stopping solution (1% H 2 SO 4 ) and measured on an ELISA plate reader at 450 nm absorbance with 650 nm as reference. Standard curves were generated by the selection peptide and plotted using a 4-parametric mathematical t model. Samples below the lower limit of measurement range (LLMR) were reported as the value of LLMR.

Statistics
Continuous variables with approximately normal distributions are expressed as mean ± standard deviation (SD) and continuous variables with non-normal distribution as median ± interquartile range (IQR). Pairwise comparisons of demographic, anamnestic and clinical parameters between the three groups were performed with one-way analysis of variance for continuous variables, and with χ 2 -tests for categorial variables. Age-adjusted p-value for trend across groups was calculated using linear or logistic regression analysis with the categorical variable treated as a continuous variable. Pairwise comparisons of biomarker distribution in the three groups were adjusted for age and performed by linear regression analyses with logarithmically transformed biomarker as the dependent variable. All tests were Bonferroni corrected for family-wise error rate. A two-sided p-value below 0.05 was considered statistically signi cant. Pairwise correlations between biomarkers and covariates of interest were calculated using Spearman's rho and reported both raw and Bonferroni corrected. Statistical analyses were performed using STATA/IC13.1 (StataCorp LP, College Station, Texas, USA).

Demographics and risk factor distribution
Baseline information is presented in Table 2. As expected, symptomatic women had a signi cantly higher risk factor burden than asymptomatic controls. Conversely, they received more aggressive preventive medication leading to lower LDL-and total cholesterol. Women with DM were older, had higher BMI and higher prevalence of hypertension, dyslipidemia and atheromatosis on invasive coronary angiogram ( Table 2). Sixteen women (8.6%) had DM Type I.

Coronary microvascular dysfunction and echocardiographic parameters
Coronary microvascular function was successfully assessed in 409 (98%) of the participants.
Symptomatic women with DM had lower CFVR, indicating poorer microvascular function compared with women without DM (age-adjusted p for trend < 0.001). Using a cut-off for CFVR of < = 2 to de ne CMD, the prevalence of CMD was 33.7%, 28.7% and 17.1% in symptomatic women with DM, symptomatic women without DM and controls, respectively (age-adjusted p for trend 0.016) ( Table 3).
LVEF under stress was higher in the two groups of symptomatic women than in the control group.
Parameters of diastolic dysfunction (E/e', e' and elevated lling pressure) indicated poorer function in women with diabetes than in the other groups (all p < 0.01). However, only few participants quali ed for manifest diastolic dysfunction according to international guidelines. 28 (Table 3).
Alterations of the extracellular matrix quanti ed by biomarkers ECM turnover biomarker levels across the three groups are shown in Fig. 1. For all biomarkers, the highest values were seen among symptomatic women with DM. Five biomarkers (MIM, PRO-C4, PRO-C6, C4M and C6M) were signi cantly higher in symptomatic women with DM than in women without DM (ageand Bonferroni adjusted p = 0.001-0.03, Fig. 1b-e) and four (TIM, MIM, Pro-C6 and C6M were signi cantly higher than in controls (age-and Bonferroni adjusted p = 0.001-0.009, Fig. 1a-e). None of the biomarkers differed between non-diabetes patients and controls.

Association between biomarkers, clinical parameters and CFVR
HbA1c, BMI, HDL-cholesterol and serum triglyceride levels were signi cantly correlated with all seven biomarkers, although all correlations were weak (r = 0.06-0.32, p = 0.03-0.0001) (results not shown). In addition, the biomarkers TIM, MIM, PRO-C6, C4M and C6M were also signi cantly correlated with history of hypertension. After Bonferroni correction, all pairwise spearman correlations became less signi cant (Fig. 2). There was no correlation between any of the biomarkers and CFVR, CFV at rest or CFV under hyperemia. No systolic or diastolic measurements were correlated with biomarkers after Bonferroni correction.

Discussion
An increased turnover of ECM fragments in blood may re ect remodeling and early brotic disease. 29 In a smaller sub-sample of the iPOWER cohort we have previously demonstrated imbalanced turnover of certain collagens when compared with healthy controls 30 but have failed to nd a relation between brosis on cardiac magnetic resonance imaging and CFVR assessed non-invasively, perhaps due to lack of statistical power. 31 In this larger study we aimed to verify the increased ECM activity and determine whether DM patients, who are particularly prone to developing myocardial brosis, had elevated ECM turnover as a marker of myocardial brosis and whether this was related to impaired coronary microvascular function.
We found that women with angina pectoris and DM had signi cantly higher levels of ECM biomarkers although CMD did not seem to be associated with these biomarkers. Furthermore, high levels of ECM biomarkers were associated with metabolic disturbances as re ected in higher BMI, HbA1c, triglycerides and lower HDL.
Cardiovascular risk factors other than diabetes were highly prevalent in symptomatic women with DM.
Ageing, hypertension and metabolic disturbances such as obesity and DM have previously been associated with myocardial brosis. In DM, glycation end-product deposition 25 and metabolic dysregulation have been described as triggers of broblast activation, cardiac ECM-remodeling and brosis. Adipositas and hypertension may also activate broblasts and thereby induce collagen accumulation and deposition. 8 Biomarkers of collagen type IV and VI turnover, TIM and MIM have previously been associated with brotic disease or brotic related conditions. Collagen type IV is primarily found in the basement membrane 30 and has a stabilizing function of microvessels during angiogenesis. 32 Increased levels have been correlated to endocardial hypertrophy and liver brosis. [33][34][35] Collagen type V and VI are important components of the interstitial connective tissue and contribute to the quality of the ECM by regulating the bril size of collagen type I and III. 30 PRO-C6 have been associated with diabetes 16,19 and together with PRO-C4 associated with poor prognosis in HFpEF-patients. 21 C4M and C6M have been linked to severe liver brosis. 22 Also, C4M has recently been found to predict major cardiovascular events and to be associated with carotid atherosclerosis. 23 Decreased titin in the sarcomere is thought to cause brosis, 24,26 and circulating levels of MMP-cleaved mimecan (MIM) has previously been identi ed as a marker of extracellular matrix remodeling in mice. 10 Although elevated in symptomatic women with DM compared with asymptomatic women, TIM was no longer correlated to DM and HbA1c after Bonferroni adjustment, whereas a strong correlation remained with BMI and blood cholesterol levels. Mimecan is a small proteoglycan with important functions in myo bril formation and angiogenesis. It is upregulated and released in heart disease such as after myocardial infarction, in conditions with pressure overload such as in hypertension, but is also released in in ammatory disease such as vasculitis. 10,14 In our previous study of collagen turnover in the iPOWER cohort including 71 symptomatic patients, 30 PRO-C6, C4M and C6M were increased when compared to asymptomatic controls, whereas no signi cant difference between groups was observed for PRO-C4, possibly due to lack of statistical power. C5M and C6M were found to be lower in iPOWER women than in controls. However, the previous study did not include patients with DM and are thus not directly comparable with the current results where the high values are related to the presence of DM.
To our knowledge, this is the rst study to demonstrate a consistent and signi cant overexpression of multiple biomarkers of brosis in women with angina pectoris, DM and risk factors for myocardial brosis. Although many correlations were weak, most were highly signi cant even after conservative Bonferroni adjustment. Further, all biomarkers were consistently associated with DM and metabolic risk factors: BMI, HbA1c, HDL-cholesterol and triglycerides. Also, we performed Bonferroni corrected pairwise correlations and consequently, the association with DM and HbA1c disappeared for TIM and C5M.
We found no relation between CMD and ECM biomarkers. This would indicate that non-endothelial dependent CMD, as assessed in this study by dipyridamole stress, is not causally related to the development of myocardial brosis. Other explanations, as discussed below, is that the ECM biomarker level does not only re ect cardiac remodeling but general brotic activity, making direct comparisons di cult. Also, a relation between increased ECM turnover and CMD caused by endothelial in ammation may have been missed in this study as we have only assessed non-endothelial dependent CMD. 8,12,36 Obesity, arterial hypertension and DM may induce chronic, systemic in ammation and consequently endothelial dysfunction, ECM remodeling, cardiac brosis and nally HFpEF. 37 Another explanation for the lack of relation between CMD and biomarkers is that the measured biomarker activity may re ect early stages of brotic disease that later may develop into manifest myocardial brosis, CMD and/or HFpEF. 38 Risk factors for HFpEF such as female sex, ageing, hypertension, obesity, and DM 38 are all well presented in our population but our population of women at risk did not have HFpEF. However, they did show signs of ventricular hyper-contractibility on echocardiography (higher LVEF), and of poorer diastolic function and higher left ventricular lling pressure compared with controls.

Strengths and limitations
In the iPOWER study, participants were consecutively included and systematically examined. All participants except the asymptomatic controls had a clinical invasive coronary angiography performed ruling out obstructive coronary artery disease (de ned by > 50% stenosis of coronary arteries). The prevalence of cardiovascular risk factors was high. However, if we had been able to include participants with more impaired ventricular function or more pronounced CMD, the population might have had more myocardial brosis which we might have been able to measure by increased levels of circulating biomarkers. We did not measure the endothelial dependent component of coronary microvascular function and have therefore not examined the relationship between endothelial-dependent microvascular function and biomarker turnover Further research is needed with the aim of detecting cardio-speci c biomarkers of brosis. Until then, it is possible to misinterpret brosis as myocardial when biomarkers may be increased due to brosis in other organs than the heart.

Conclusion
Women with angina pectoris, DM and a high cardiovascular risk factor burden have increased turnover of biomarkers re ecting early brotic disease compared with women without DM and few risk factors. Biomarkers were associated with BMI, HbA1c and cholesterol levels but not with non-endothelial dependent CMD. To better evaluate the relation between CMD and myocardial brosis future studies would bene t from longitudinal follow-up and include measurements of endothelial dependent CMD.  Table 1.

Ethics approval and consent to participate
This study was performed in accordance with the Declaration of Helsinki 39 and was approved by the Danish Regional Committee on Biomedical Research Ethics (H-3-2012-005). All participants have given written informed consent upon oral and written information.

Consent for publication
Not applicable.

Availability of data and materials
The dataset analyzed during the current study is available from the corresponding author on reasonable request.

Competing interests
SHN is employed at Nordic Bioscience. The other authors of this manuscript have no con icts of interest.

Funding
This work was supported by The Danish Heart Foundation (grant number: 11-10-R87-B-A3628-22678).      Supplementary Files This is a list of supplementary les associated with this preprint. Click to download.