Characterizing Cardiac Involvement in Chronic Kidney Disease Using CMR—a Systematic Review

In order to provide a focused update on the developments and findings from CMR studies characterizing cardiac involvement in CKD over the last 3 years, a systematic literature review was performed. This was done in accordance with the PRISMA [13] guidelines by two researchers (KM, KMcD) (Fig. 2) who independently searched PubMed and Web-of-Science using the following keywords and variations on them: ‘Renal failure’, ‘Chronic kidney disease’, ‘Hemodialysis’, ‘Renal’, ‘Renal transplant’, ‘Cardiac magnetic resonance’, ‘cardiac imaging’.

Our search was restricted to peer-reviewed journals and human subjects. Editorials, reviews, studies with fewer than 10 patients or those not published in English were excluded.

Abstracts of all potential titles were reviewed by KM and KMcD. References of relevant reviews and all full papers were searched to retrieve any additional papers, repeating the process until no new papers were found (Table 2).

Utilizing CMR, pediatric and adult patients with CKD and incident HD have recently been reported to have LVH, when matched to healthy controls [15, 20]. In fact, it is well known that a significant proportion of patients with CKD have LV hypertrophy, which is associated with increased mortality [36, 37] and cardiac arrhythmia [36]. This has led to LV mass regression being the focus of a number of clinical trials [19, 38], as, theoretically, this should be associated with a reduction in mortality and sudden cardiac death. Surprisingly, a recent meta-analysis [39] looking at 73 trials and over 6500 patients did not identify a clear association between LV mass regression and mortality. A possible explanation for this is that most of the studies (87%) utilized echocardiography to quantify LV mass. One of the advantages of CMR is the use of a retrospectively ECG-gated axial stack of cine imaging to accurately delineate endo- and epicardial borders at end-diastole and systole [40]. This obviates the need for assumptions that the left ventricle is an ellipsoid object, such as with the bi-plane equation used in echocardiography. In the future, a meta-analysis of only CMR studies considering LV mass regression would of course be of interest, but currently there are insufficient numbers of CMR studies to make this a meaningful exercise.

Avoidance of fluid overload and targeting blood pressure control in patients on HD may prevent LV dilation and remodeling, and result in LV mass regression—a recent observational study demonstrated that a greater indexed LV mass was independently associated with greater systolic blood pressure and with greater LV indexed end diastolic volume [24].

As an alternative to conventional hemodialysis, nocturnal hemodialysis holds promise. Two recent studies have investigated whether the avoidance of significant fluid shifts by performing dialysis overnight for longer periods of time or more frequently than three times per week might result in myocardial remodeling and regression in LV hypertrophy. Both studies demonstrated LV mass regression [18, 25], this may have been related to better BP control [23]. When the authors of one of these studies [25] looked at associates of adverse cardiac remodeling in their study population over the year of their study follow up they found volume and pressure overload were the biggest factors contributing to negative myocardial changes. Following this logic, it is not surprising that more frequent, or longer nocturnal dialysis led to LV mass regression [22]. Neither of these two studies was fully randomized but their findings were in keeping with an earlier randomized study [41]. While these findings are encouraging, the next step would be to establish whether this CMR detected myocardial remodeling translates into improved clinical outcomes.

Further insights into the acute effects of HD and hemodiafiltration on myocardial function were recently described in an ambitious study where patients underwent CMR imaging during their dialysis therapy [16••]. The authors of this study must be commended for overcoming multiple logistical challenges in order to address this issue. Both modalities resulted in reduced global and regional myocardial contractile function during treatment, with the severity being proportional to ultrafiltration rate and BP reduction with partial recovery post therapy. Performing CMR during dialysis is no simple feat and must be relatively arduous for both patients and professionals involved. For this reason, it is highly likely that studies of this nature will involve very small numbers of patients only. A large-scale clinical trial involving this method of imaging is currently not practical.

However, slightly larger studies are possible in the dialysis population by making small changes to in-center dialysis regimens. For example, Odudu has demonstrated in a study of 73 incident HD patients that cooling dialysate fluid could potentially slow cardiac remodeling, with patients randomized to cooled dialysis versus normal HD regimen experiencing a reduction in LV mass and volumes at 1 year [19]. In this study, there was no improvement in ejection fraction which was the study primary outcome, however peak systolic strain, which is perhaps a more sensitive marker of cardiac function, was preserved by the intervention. Assessment of myocardial strain using CMR is a developing area of interest within the renal population and is discussed further later in this review.

Candidates for renal transplantation could potentially be risk-stratified based on CMR imaging data recently reported [21], with pre-transplantation LV hypertrophy and left atrial dilatation being independent predictors of mortality post-transplantation. A possible treatment for LV hypertrophy regression post-transplantation is the ligation of any patent arterio-venous fistulae [17].

One of the big challenges of imaging the renal population is that it is not possible to use gadolinium-based contrast agents in ESRD patients because of concerns regarding nephrogenic systemic fibrosis [42]. This has therefore stimulated interest in the nephrology community in alternative techniques to quantify tissue abnormalities such as parametric mapping. Native T1 mapping measures the longitudinal relaxation of hydrogen ions after applying inversion magnetization pulses [43, 44]. If renal function is good enough to permit the use of gadolinium-based contrast agents, post-contrast T1 maps may be acquired, and utilizing the serum hematocrit, then the extra-cellular volume fraction can be calculated [45]. If native T1 time is prolonged, then this may represent a tissue abnormality such as an increase in interstitial space (fibrosis, amyloid deposition), or edema (myocardial infarction or inflammation, e.g., myocarditis). T2 mapping measures the transverse decay time of hydrogen ions, and the main cause of raised T2 values is edema (myocardial infarction or inflammation) [46]. T2* mapping has been developed in view of the paramagnetic properties of iron (present as ferritin and hemosiderin) in the myocardium and liver [46]. Thus, iron deposition (from iron supplementation, or due to intra-myocardial hemorrhage) results in a low T2* signal on parametric mapping.

CKD is associated with myocardial interstitial fibrosis [47]. Recent research has described how native T1 values were higher than controls in patients with early stage CKD [29•], ESRD on hemodialysis for < 6 months [34•] and in patients on hemodialysis for > 12 months [31•]. In fact, looking at septal native T1 values (ms) across the three publications, native T1 values increase with a longer duration of hemodialysis supporting the fact that the degree of interstitial fibrosis is proportional to the disease duration.

A further paper recently published describes native T1 mapping as a robust technique with excellent intra-observer, inter-observer, and inter-study variability assessment [32]. T1 acquisition on T1MES phantoms [48] between two centers, as well as inter-center analysis of native T1 datasets acquired in patients with CKD-derived comparable T1 values, supporting the use of parametric mapping as an exploratory endpoint in multicenter studies. However, tissue correlation of native T1 mapping in the renal population has not yet been done and some may argue that before it becomes a standard imaging sequence in renal CMR studies this hurdle should be overcome.

T2* mapping has been applied to patients with CKD to investigate for potential iron overload. This is as these patients get erythropoietin and parenteral iron infusions for iron deficiency to due reduced iron absorption and release from tissues [49]. Interestingly, 50% of the patients in one group [35] and 80% in another recently described group [33] had reduced hepatic T2* signal in keeping with iron overload. There was, however, no evidence of myocardial iron loading in either study.

There are several techniques for assessing myocardial strain with CMR. This is of importance in the renal population, as there is emerging evidence that patients with renal disease have abnormal strain despite normal LV ejection fraction. The bespoke strain methods include phase-contrast [50], tagging [51], displacement encoding with stimulated echoes (DENSE) [52, 53], and cine-derived strain [54, 55].

Myocardial tagging derives strain estimation by imaging and tracking tissue markers (“tags”) induced by changes to the tissue magnetization [51]. Tagging is considered as the gold standard reference method for CMR strain [56]. Feature-tracking (FT) involves retrospective motion tracking of cine imaging. While vendor dependent, most cine-strain techniques derive strain by tracking the displacement of the endo- and epicardial borders, while taking into account columns of pixels in the myocardium [54]. While the derivation of strain from cine imaging has utility in keeping scanning time short in this patient group, there is a trade-off with greater measurement variability when compared to bespoke strain techniques [55, 57].

Patients with early CKD have been described to have reduced global longitudinal strain and strain rate compared with controls utilizing FT [29•]. This trend was observed in HD patients < 6 months vintage utilizing tagging [20] and FT [34•], and in HD patients > 6 months vintage [31•]. Interesting insights from a pediatric population [30] utilizing phase-contrast describe markedly reduced early diastolic ventricular function in CKD patients without LV hypertrophy on CMR. Abnormalities in myocardial strain could therefore represent some of the first detectable changes in the hearts of CKD patients and in future may even have a role in screening or identifying patients with subclinical disease, who are theoretically at greater risk of more advanced cardiac disease.

Myocardial ischemia assessment with CMR can be dichotomized into the assessment of inducible wall-motion abnormalities by infusing dobutamine or by the use of vasodilator-stress perfusion testing to identify an inducible perfusion defect. The use of dobutamine infusion results in a rise in the myocardial metabolic demand, through chronotropic stimulation. Myocardial segments subtended by a stenosed coronary artery are unable to meet the metabolic demands, and appear hypokinetic [58]. Dynamic LV outflow tract obstruction can occur, resulting in false positive wall motion “abnormalities.”

Vasodilator-stress testing makes use of vasodilator agents such as adenosine or regadenoson causing a 3-5× fold coronary vasodilatation in normal coronary arteries, while stenosed vessels are unable to vasodilate to the same extent. The infusion of gadolinium contrast at peak stress and at rest enables the clinician to identify regions of inducible ischemia. This is one of the most sensitive non-invasive techniques available for the identification of myocardial ischemia [59, 60]. This technique is however unsuitable in patients with reduced eGFR [58]. Blood oxygen level dependent (BOLD) CMR is similar to T2* mapping in making use of the paramagnetic effect of deoxyhemoglobin to observe a signal drop, in tissues with reduced tissue oxygenation, due to epicardial coronary artery stenosis or impaired microcirculation [61].

The safety and feasibility of dobutamine stress CMR was recently described in patients with ESRD [28]. 93% of the patients achieved 85% of the age-predicted heart rate. Ten percent of the completed stress scans were positive for inducible wall motion abnormalities and there were no serious adverse events reported. Adenosine-stress CMR was utilized in renal transplant recipients with no known coronary artery disease, with the myocardial perfusion reserve index (an assessment of perfusion) being reported as being lower in transplant recipients when compared with hypertensive controls [26]. CMR angiography was also acquired in this patient group, with 35% of the patients having 50% coronary artery stenosis. Adenosine-stress CMR coupled with BOLD was investigated in patients with ESRD and in transplanted patients without known coronary artery disease [27]. The mean signal intensity in both patients with CKD and in patients post-transplant were lower than that acquired in hypertensive individuals or in normal controls, with the lowest myocardial oxygen response to stress being in CKD patients on dialysis, then in CKD patients pre dialysis. Renal transplantation recipients appeared to have an improved tissue oxygen response, but still significantly different from healthy volunteers.

Arterial stiffening, through arteriosclerosis, is one of the earliest signs of subclinical cardiac involvement of CKD [62]. CMR enables the measurement of aortic pulse wave velocity by contouring the ascending and descending aorta on phase-contrast MR to derive the temporal shift and thus pulse wave velocity [63]. Aortic distensibility can be calculated by contouring a short axis view of the ascending aorta on cine imaging and obtaining an aortic pulse pressure via brachial cuff [63].

Patients with a 3 months HD vintage had reduced aortic distensibility and increased aortic pulse wave velocity when compared to healthy controls [20]. Aortic distensibility was preserved in a study group of HD patients using cooled dialysate fluid [19]. Studies suggest that increased arterial stiffness and reduced aortic distensibility are pathophysiologically associated with arterial hypertension and an increase in LV loading. This in turn leads to LV hypertrophy, which is in turn associated with adverse events. Thus, detecting reduced aortic distensibility in patients with CKD has the potential to be an early biomarker of circulatory dysfunction.

While the main focus of CMR scans is to assess myocardial changes in CKD patients, the initial coronal, longitudinal, and transverse planes of the thoracic cavity and upper abdomen are acquired by radiographers for study planning. This leads to imaging a significant volume of the patients’ bodies beyond the organ of interest, which may reveal important pathology. While this is not directly relevant to characterizing myocardial disease involvement, we felt this is an important point to discuss in view of increased use of CMR in studies in patients with CKD. Recently, a paper reported the incidence of non-cardiac findings in a CKD stage 5 population as being seven times greater than what was previously reported in an all-comers CMR study [14, 64]. While this study was limited by its retrospective nature, researchers utilizing CMR in renal populations should consider its findings seriously. It is the responsibility of researchers to ensure that there are robust reporting pathways in place for any obtained images and that all individuals are fully informed about the potential consequences of incidental findings.

Another important challenge in this patient group is the relation of timing of imaging to dialysis and the resultant fluctuant fluid changes, which could give differing hemodynamic measurements depending on loading conditions. This is especially important in multicenter studies.