Cardiac biomarkers in chronic kidney disease are independently associated with myocardial edema and diffuse fibrosis by cardiovascular magnetic resonance

Chronic kidney disease (CKD) and heart failure (HF) often coexist and worsening renal function is a strong predictor of poor cardiovascular outcome [1]. Patients with CKD have very high risk of cardiovascular disease (CVD), yet high CVD morbidity and mortality is only partially explained by the complications from atherosclerotic coronary artery disease (CAD). Phenotypically, HF in CKD is characterized by preserved left ventricular (LV) systolic function, eccentric remodeling and increased diastolic stiffness [2‐4]. Markedly raised troponin and N-terminal pro-brain natriuretic peptide (NT-pro BNP), the markers of myocardial injury and increased wall-stress, respectively, are also frequently found in CKD. Both markers are associated with poor prognosis in CKD, and especially with the HF-associated outcomes [5‐8]. Although their significance remains controversially discussed and had been related to impaired renal elimination, recent reports imply that a sustained, subclinical, non-ischemic myocardial injury is more likely to explain the raised levels of these markers in CKD [9].

Myocardial T1 and T2 mapping with cardiovascular magnetic resonance (CMR) allows quantifiable tissue characterization and provide direct measures of the pathological myocardial processes non-invasively [10]. Native T1 is a non-specific measure of abnormal myocardium; it can relate to myocardial fibrosis, edema or infiltration [10]. Native T2 is water-specific, indicating excess myocardial fluid, which co-localizes with myocardial processes, such as edema, injury and/or inflammation [10]. Several studies revealed higher native T1 values in CKD patients [11], whereas more recently, a correlation of native T1 and troponin has also been shown [12]. A single previous study revealed raised T1 and T2 markers in participants with severe CKD on hemodialysis [13]. Native T1 of non-scarred myocardium is a strong predictor of survival, CVD mortality and incidence of HF in ischemic and non-ischemic cardiac conditions [14, 15]. No study to date comprehensively examined the relationship between these imaging measures and myocardial changes in CKD, nor related these to the serological markers of myocardial injury and increased wall stress, high-sensitive cardiac troponin T (hs-cTnT) and NT-pro BNP.

This is a prospective longitudinal observational investigator-led study of T1 in adult patients undergoing a clinically indicated CMR examination, in line with cardiological practice guidelines (Consort diagram, Fig. 1). Consecutive patients with established diagnosis of CKD [16, 17] and reduced estimated glomerular filtration rate (eGFR) < 59 ml/min/1.73 m² using the Modification of Diet in Renal Disease (MDRD) formula (the CKD stages 3–5; TRUE-TypeCKD Study NCT03749551) ongoing recruitment since March 2018) were enrolled. The CKD diagnosis was established independently by nephrologists and defined as abnormalities of kidney function or structure present for more than 3 months and eGFR < 59 ml/min/1.73 m² on at least two occasions separated by a period of at least 90 days (with or without markers of kidney damage) [16, 17]. An independent cohort of controls with eGFR ≥ 60 ml/min/1.73 m² (including CKD stages 1 and 2 [16, 17]), matched for age, gender and traditional CVD risk factor profile, was sourced from the ongoing International T1 Outcome Study using propensity score matching (PSM, n = 242, recruited between 03/16 to 03/19 (NCT03749343), ongoing recruitment since March 2016). The details of both registries, inclusion/exclusion criteria, imaging protocols and sequence parameters were reported previously [14, 15, 18], and are included in Additional file 3. A further group of healthy controls (n = 38) with similar age and sex distribution was gathered from an ongoing multicenter study including healthy individuals without CVD risk factors or any regular medication (NCT04444128). Exclusion criteria were known specific cardiomyopathies, valvular heart disease or myocarditis, known allergy to gadolinium-based contrast agents (GBCA), and CMR unsafe implants or devices. All procedures were carried out in accordance with the Declaration of Helsinki (2013). All participants were informed about the possible risks of nephrogenic systemic fibrosis (NSF), as per regulatory guidelines (Food and Drug Administration, European Medicine Agency, American College of Radiology [19]). Administration of GBCA to the CKD patients was in any case strictly performed following guidelines indication [20]. Patients on hemodialysis were scheduled to receive this within 24 h from the administration of the GBCA. All participants with significant CKD were screened for symptoms of NSF at regular 6-monthly intervals. The study protocols were reviewed and approved by institutional ethics committee. Written informed consent was obtained from all participants.

Clinical meta-data, including systolic/diastolic blood pressure (BP), body mass index (BMI), presence of traditional CVD risk factors, symptoms, medication were collected for all participants. Respiratory variation of inferior vena cava (IVC) from transthoracic echocardiography performed within 30 min from the index CMR was used as an index of fluid status, obtainable in all subjects. Loop diuretics use was not considered a reliable marker of volume status in CKD patients, since hemodialysis patients with no urine production would likely have volume overload and no diuretics intake. All participants underwent a standardized CMR using a 3T clinical CMR scanner (Skyra, software version VE11, Siemens Healthineers, Erlangen, Germany) for acquisition of cardiac function, volumes, mass, myocardial T1 and T2 mapping (Fig. 2), myocardial perfusion and scar imaging. T1 mapping was performed using an in-house developed variant of the modified Look-Locker Imaging sequence (Frankfurt Main, FFM-MOLLI) [10]. For T2 mapping, a T2-FLASH sequence was employed [21]. All T1 and T2 mapping values were determined by these same sequences, where both have established normal ranges, which were previously published using identical sequence parameters [22, 23]. Myocardial perfusion imaging was performed using vasodilation (regadenoson, 400 mcg/5 ml) and administration of 0.1 mmol/kg body weight gadobutrol (Gadovist^®, Bayer Healthcare, Berlin, Germany) [24]. The presence of myocardial scar was visualized by late gadolinium enhancement (LGE) 15 min after GBCA administration.

Transthoracic echocardiography was performed by board-certified cardiologists within 30 min of completion of CMR examination with participants lying supine in left lateral position (Vivid E95, General Electric Healthcare, Chicago, Illinois, USA) for respiratory variation of IVC, as a measure of volume status, using postprocessing recommendation [25].

Analysis of cardiac volumes, function and mass was performed using semiautomated contour detection (SuiteHeart^®, Neosoft, Pewaukee, Wisconsin, USA). Interpretation of myocardial perfusion and LGE images was performed following standardized postprocessing recommendations. Myocardial LGE was visually defined by a minimum of two observers based on the presence and predominant pattern as ischemic or non-ischemic [26]. All observers were board-certified cardiologists holding established accreditations of CMR competencies. Quantitative tissue characterization and myocardial deformation analysis were performed by the core-lab staff (Goethe CVI, Frankfurt/Main, Germany), blind to the underlying participant group allocation. Myocardial T1 and T2 were measured conservatively within septal myocardium of midventricular short axis slice, using motion-corrected scanner-derived images, as per internal standardized operating procedures [10, 27]. Areas of LGE were excluded from region of interest to avoid inclusion of areas with replacement scar. Global longitudinal (GLS) and circumferential strain (GCS) were measured using CMR feature tracking (MEDIS^®, Leiden, The Netherlands) [28].

All participants underwent venous blood sampling immediately prior to CMR study. Bloods samples were spun and frozen at − 80 °C and analyzed subsequently using standardized commercially available test kits Analysis hs-cTnT and NT-pro BNP (Elecsys 2010^®, Roche, Basel, Switzerland). The cut-offs for hs-cTnT and NT-pro BNP were used to define normal/abnormal in all subjects (using a cut of value of 99 percentile of 13.9 ng/l [8] and > 300 pg/l [3], respectively).

Baseline population characteristics for groups of patients and healthy controls are summarized in Table 1. Significant differences were observed between healthy controls and both non-CKD (eGFR ≥ 60 ml/min/1.73 m²) and CKD (eGFR < 59 ml/min/1.73 m²) patients’ groups regarding native T1 and T2 which were lower in controls (Table 2, Fig. 3 p < 0.001 for all comparisons). Compared to patients with eGFR ≥ 60 ml/min/1.73 m², patients with eGFR < 59 ml/min/1.73 m² had higher hs-cTnT, NT-pro BNP and lower hematocrit (p < 0.05). The two groups were similar for New York Heart Association class, and most cardiac medications, apart from the higher rate of aldosterone and neprilysin inhibitors in controls, and loop diuretics in the eGFR < 59 ml/min/1.73 m² group (p < 0.05 for all). The eGFR < 59 ml/min/1.73 m² group also had higher LV volumes and mass, myocardial native T1 and T2 (Fig. 3), and lower LV ejection fraction (LVEF), GCS and IVC respiratory variation (Table 2, p < 0.01 for all). A third of all patients had myocardial scar by LGE with no overall difference in proportions between the patients’ groups for the presence (p = 0.15) or patterns (ischemic vs. non-ischemic, p = 0.63 vs. 0.37, respectively). Patient groups were also similar for the presence of relevant myocardial ischemia (p = 0.10); this was treated with revascularization, if amenable target vessel was identified.

In this prospective study, we demonstrate independent associations between cardiac biomarkers with imaging marker of myocardial edema and diffuse fibrosis, which are CKD-group specific. Native T1 and T2 were the only imaging markers to be independently associated with worsening CKD. Serological marker of increased LV wall stress, NT-pro BNP, and of myocardial injury, hs-cTnT, were independently associated with native T1 and T2 respectively, albeit only in patients with reduced renal function. Also, there was a significant reduction of native T2 when measured immediately post-hemodialysis with the change in T2 proportional to the removed ultrafiltration volume. Together, these results indicate a link between myocardial injury and LV wall stress with an increase of myocardial fluid, in addition to myocardial fibrosis, which is inherent to the presence of CKD.

Our observations expand the current knowledge about the pathophysiology of CVD and adverse myocardial remodeling in CKD. We employed two independent approaches of detecting myocardial abnormalities, by serology and imaging, which are both well established in terms of accuracy and prognostic significance [3, 8, 10]. Previous studies investigating the causes of raised troponin in CKD indicated that increased levels cannot be explained by the classical myocardial injury processes, such as such as infarction-like necrosis. Our findings lend support to this notion by showing that despite the higher CVD risk given by renal insufficiency, the rate of ischemic scar and myocardial ischemia or infarction was similar to that of the non-CKD group. Compared to a matched cohort of patients with eGFR ≥ 60 ml/min/1.73 m², but similar CVD risk factors by PSM of the two cohorts, we report elevated native T1 and T2, but not more prevalent LGE, reiterating that markedly raised troponin levels are unlikely fully explained by the consequences of atherosclerotic CAD [8]. Furthermore, we found that serological biomarkers hs-cTnT and NT-pro BNP and imaging markers of structural remodeling have closer correlation with native T1 and T2 as the renal function progressively declined. Together, these findings indicate that non-ischemic processes play an important role in pathophysiology of myocardial injury in CKD [29].

The independent association between native T2 and hs-cTnT in patients with severe CKD (eGFR < 29 ml/min/1.73 m²) is a central new finding, reiterating the role of increased myocardial fluid, in addition to myocardial fibrosis, as an integral part of structural LV remodeling in CKD [30]. In the literature, troponin release in CKD has been attributed to a number of mechanisms, including increased transmural pressure, small-vessel coronary obstruction, endothelial dysfunction, intracellular edema [8, 31, 32], as well as direct cellular toxicity of the uremic milieu [33, 34]. Notably, we strived to control for myocardial inflammation by excluding subjects with suspected clinical presentation or positive histology for active inflammation, which could yield a similar imaging pattern [35]. Whether edema is a cause or a consequence of the myocardial damage, cannot be fully elucidated based on the current data. Recent studies revealed the dynamic changes in native T2 measurements induced by either ultrafiltration or diuretics in CKD and HF, respectively [13, 36]. In a proof-of concept sub study, we reproduced this finding in a subgroup of participants that underwent a second scan immediately after HD. In addition to reduction of native T2 in this subgroup, we also found a trend of increasing association between native T2, hs-cTnT and NT-pro BNP with worsening of CKD stages, which may suggest that myocardial water content is increasing with worsening renal function, likely affected by the total body water content. This observation might point towards potential clinical implications. Despite discrete native T1 and T2 values overlap between non-CKD (eGFR ≥ 60 ml/min/1.73 m²) and CKD (eGFR < 59 ml/min/1.73 m²) patients, which could reduce the added value of mapping assessment in this context, T2 exhibited significant correlation with cardiac injury in patients with severe renal insufficiency (eGFR < 29 ml/min/1.73 m²) only. Further studies are required to explore whether focused myocardial water-shifts might translate into a modifiable intervention conferring cardio-protection in some patients with advanced CKD, hypothetically those with higher T2 and hs-cTnT. We also observed an interrelatedness between T1 and T2, which becomes much stronger as renal function declines, indirectly supporting that both measures are much more influenced by excess myocardial fluid in CKD [10, 27]. Accordingly, native T1 is a nonspecific measure of interstitial expansion, detecting water as well as fibrosis [30], hence, an integrated reading along with T2 values and a patient’s underlying renal function may aid the correct interpretation of this measure. Outcome data are needed to validate the prognostic relevance of our findings, as well as prospective trials of intervention algorithms based on the presented imaging readouts to ascertain whether their guidance can improve clinical care.

A few limitations of our study apply. We strived to control for inclusion bias in several ways. Respiratory variation of IVC was used as a proxy-measure of the total body fluid status, because it was obtainable in all subjects across the eGFR spectrum. Participants with reduced kidney function were recruited from tertiary centers as well as peripheral nephrology practices, thus reducing the potential referral bias. The clinical indications for CMR were uniform for both groups and approved by an independent cardiologist in line with cardiological practice guidelines. Native T1 moderately correlates with collagen volume fraction in model diseases of pressure overload (severe aortic stenosis, participants with eGFR ≥ 60 ml/min/1.73 m²), however, histological correlations with myocardial fibrosis in CKD may differ and require verification in future studies. Myocardial edema cannot be validated by classical histology, as it is complicated by dehydration through tissue fixation with formaldehyde. However, a previous experimental study used a tissue desiccation method to provide validation of T2 mapping measurements against myocardial water content [37]. CMR examinations in the present study were performed as a part of clinical service, where the much-needed time efficiency does not justify the use of rest perfusion imaging nor postcontrast T1 mapping.

In conclusion, we demonstrate independent associations between serological biomarkers with imaging markers of interstitial expansion, which are CKD-group specific. Our findings indicate the role of diffuse non-ischemic tissue processes, including excess of myocardial fluid in addition to diffuse fibrosis in CKD-related adverse remodeling.