This large retrospective CMR study has examined the prevalence, patterns and clinical significance of LGE in patients with pre-dialysis CKD. Over a third of subjects had evidence of LGE with normal LVEF and low rates of LVH. However, LGE, did not predict adverse cardiac events and did not offer incremental risk stratification in this population. This finding may reflect that over half of the LGE was focal at the right ventricular insertion points, a pattern shown to be associated with a better prognosis and correlated on histology with myocardial disarray, increased collagen and fat between fibre bundles but not replacement fibrosis.[16] Only 18/159 (11%) patients had a mid-wall / diffuse pattern in keeping with myocardial interstitial fibrosis from a non-coronary aetiology which is far lower than the reported ~30% seen in non-ischaemic dilated cardiomyopathy (NI-DCM) where LGE has prognostic stratification. [17] CMR plays an increasingly important role in characterizing heart disease in CKD and an appreciation of the myocardial changes frequently present are important for clinicians. The observed low cardiac event rate over intermediate follow up might also offer a degree of reassurance to clinicians when assessing CKD patients without high risk CV markers such as diabetes or proteinuria.
RVIP-LGE has been reported not only in NI-DCM, pulmonary hypertension and hypertrophic cardiomyopathy but also in athletes [18-20]. It is thought to represent plexiform fibrosis related to interdigitation of RV and LV fibers as these form the septum rather than myocardial fibrosis and represents only 1-3% of overall LV mass [21]. The natural history of RVIP-LGE in CKD remains unknown. Our finding of isolated RVIP-LGE and a more benign phenotype is supported in a recent prospective observational study examining patterns of LGE in NI-DCM with severe LV impairment and a mean LVEF of 25% over 46 months follow up. RVIP-LGE was present in 14% of the cohort and was associated with higher systolic and diastolic blood pressure but did not increase the risk of adverse cardiac event or arrhythmic events compared to patients with no LGE. Furthermore, there was a lower incidence of heart failure than observed with myocardial LGE [16]. In pulmonary hypertension, observational studies have also shown that RVIP-LGE does not predict mortality in contrast to more extensive LGE extending into the interventricular septum [20].
In contrast, there are extensive data to support the role of myocardial LGE as a marker of risk in ischaemic and non-ischaemic cardiomyopathies even when the LVEF is normal.[1, 17]. Myocardial LGE correlates on histology with expansion of the extracellular space, increased collage deposition and ultimately irreversible fibrosis.[22] The latter serves as a substrate for ventricular arrhythmias and promotes adverse ventricular remodelling.[22] In CKD, histological data showing myocardial interstitial fibrosis and disarray on endomyocardial biopsy was first reported in the 1990s and the subsequent use of LGE-CMR demonstrated 14% of patients with ESKD had evidence of mid-wall reparative fibrosis [7, 8]. Unlike other disease cohorts, large outcome studies demonstrating the prognostic role of LGE in CKD have not followed and further gadolinium LGE-CMR studies have been restricted since 2006 with regulatory changes to reflect the reported association between gadolinium-based contrast agents used for magnetic resonance imaging (MRI) procedures and nephrogenic systemic fibrosis (NSF).
The largest study to date in CKD was a retrospective analysis of 966 “all comers” undergoing LGE-CMR between 2006-2008 with a serum creatinine measured within 30 days of CMR. Patients were stratified according to eGFR greater or less than 70ml/min/1.73m2. [23] The authors demonstrated lower eGFR to be associated with more LGE even after adjustment for demographics, coronary disease and diabetes. Stratification of survival based on eGFR demonstrated higher mortality with LGEpos and eGFR<70ml/min1.73m2 with a hazard ratio of 1.8 (1.07-3.04, p=0.03) compared to LGEneg subjects. [23] This cohort was a very different population to our study. Firstly, subjects were ‘all comers’ who had all undergone CMR for clinical reasons and most subjects had coronary artery disease (54%) or diabetes (16%).[23] This contrasts with those recruited to our study, the majority of whom were identified a priori based on the presence of renal disease confirmed on biopsy/imaging and without cardiovascular disease. Secondly, the predominant pattern of LGE found in 88% was sub-endocardial ‘coronary’ infarct, suggesting that the study by Dandamudi et al was of a cohort of clinical patients with ischaemic heart disease and mild secondary renal impairment (mean eGFR 73ml/min/1.73m2).[23] The ratio of sub-endocardial LGE in their study is disproportionate not only to the amount in non-coronary pattern seen in our study but also that in the landmark study by Mark et al. [7]
There are several potential reasons why LGE did not predict adverse outcomes in our study of CKD contrary to the near universal finding in non-ischaemic cardiomyopathies where “LGE is bad”. Firstly, our cohort was selected as part of research trials to investigate the impact of CKD on changes in myocardial structure and function, and hence specifically excluded subjects with diagnosed cardiovascular disease and diabetes. Secondly, although the majority of our cohort were male, hypertensive and many were prescribed lipid lowering therapy, most also had only early stage CKD and the overall CV adverse event rate was low. It is possible that the follow-up period was too short to identify adverse outcomes related to myocardial LGE but it may also be that the impact of myocardial fibrosis in CKD may be most adverse in those with an incremental number of risks, perhaps including sub-clinical ischaemia. The latter hypothesis is consistent with the recent scientific statement from the American Heart Association, concluding that multiple mechanisms contribute to the progression of uremic cardiomyopathy rather than this being driven solely by a decline in eGFR (26). The impact of multiplicative CV risk factors including hypertension, calcium-phosphate metabolism and diabetes on LGE prevalence remains to be established in higher risk CKD cohorts.
Limitations and clinical implications
There are several limitations of our study; the data were retrospectively acquired and are observational from pre-defined and selected research cohorts. The small sample size, long follow up and small number of events makes this study vulnerable to type two statistical error. The low number of adverse clinical outcomes is partly due to a highly selected patient group and single centre data which might have precluded identification of an independent association of adverse events with LGE. It is also possible that clinical events have been missed in patients presenting to other hospitals. We acknowledge a selection bias in our cohort. Patients recruited into our studies are a highly selected group with controlled blood pressure and without diabetes and many comorbidities. The aetiology is also most commonly primary glomerulonephritis or adult polycystic kidney disease. We also acknowledge that the duration of CKD would be expected to affect the prevalence of myocardial disease. However, it was not possible to reliably established this vintage in pre-dialysis patients many of whom were only referred from primary care at a stage of more rapid decline or falling below eGFR referral cut offs. Further studies will require the enrolment of unselected subjects with CKD and longer follow up times to assess the prognostic implications of LGE in CKD. It is noteworthy that we have no reported cases of NSF in over 10 years of CKD based research. However, precautions are taken to minimise dose (0.15-0.2mmol/kg), newer low risk gadolinium agents are used (Gadovist®) and contrast is avoided in patients with acute kidney injury and when eGFR <30ml/min/1.73m2. The absence of any cases of NSF is reassuring but, our sample size is too small and NSF is too rare to draw conclusions.
Recruitment and data collection commenced before widespread availability of newer non-contrast T1 mapping CMR sequences. These sequences have undoubtedly allowed better detection of diffuse fibrosis which can be “missed” with LGE inversion recovery sequences due to limited spatial resolution.[24] Observational data in CKD cohorts have been consistent in demonstrating elevated native T1 times in early stage CKD and ESKD on dialysis [9, 25, 26]. However, to date there is not histological correlation with T1 in CKD. Indeed a recent proof of concept study correlating the extent of myocardial fibrosis on LGE with absolute native T1 signal failed to identify “cut off” thresholds in a validation model [27]. To date, the most comprehensive assessment of myocardial fibrosis and its clinical impact appears to be through combining multi-parametric CMR biomarkers of LGE, T1 and ECV. This approach was recently used in a cohort with aortic stenosis and successfully identified patients with worse LV remodelling, adverse blood biomarkers such as NT pro-BNP and worse functional capacity. [28] Continued restrictions on use of gadolinium in CKD means T1 mapping endpoints will potentially have increasing importance for longitudinal follow up and outcome prediction.