Cardiac magnetic resonance findings and prognosis in type 1 myotonic dystrophy

Background Cardiac involvement is a major determinant of prognosis in type 1 myotonic dystrophy (DM1), but limited information is available about myocardial remodeling and tissue changes. The aim of the study was to investigate cardiac magnetic resonance (CMR) findings and their prognostic significance in DM1. Methods We retrospectively identified all DM1 patients referred from a neurology unit to our CMR laboratory from 2009 to 2020. Results Thirty-four patients were included (aged 45 ± 12, 62% male individuals) and compared with 68 age-matched and gender-matched healthy volunteers (43 male individuals, age 48 ± 15 years). At CMR, biventricular and biatrial volumes were significantly smaller (all P < 0.05), as was left ventricular mass (P < 0.001); left ventricular ejection fraction (LVEF) and right ventricular ejection fraction (RVEF) were significantly lower (all P < 0.01). Five (15%) patients had a LVEF less than 50% and four (12%) a RVEF less than 50%. Nine patients (26%) showed mid-wall late gadolinium enhancement (LGE; 5 ± 2% of LVM), and 14 (41%) fatty infiltration. Native T1 in the interventricular septum (1041 ± 53 ms) was higher than for healthy controls (1017 ± 28 ms) and approached the upper reference limit (1089 ms); the extracellular volume was slightly increased (33 ± 2%, reference <30%). Over 3.7 years (2.0–5.0), 6 (18%) patients died of extracardiac causes, 5 (15%) underwent device implantation; 5 of 21 (24%) developed repetitive ventricular ectopic beats (VEBs) on Holter monitoring. LGE mass was associated with the occurrence of repetitive VEBs (P = 0.002). Lower LV stroke volume (P = 0.017), lower RVEF (P = 0.016), a higher LVMi/LVEDVI ratio (P = 0.016), fatty infiltration (P = 0.04), and LGE extent (P < 0.001) were associated with death. Conclusion DM1 patients display structural and functional cardiac abnormalities, with variable degrees of cardiac muscle hypotrophy, fibrosis, and fatty infiltration. Such changes, as evaluated by CMR, seem to be associated with the development of ventricular arrhythmias and a worse outcome.

Background Cardiac involvement is a major determinant of prognosis in type 1 myotonic dystrophy (DM1), but limited information is available about myocardial remodeling and tissue changes. The aim of the study was to investigate cardiac magnetic resonance (CMR) findings and their prognostic significance in DM1.
Methods We retrospectively identified all DM1 patients referred from a neurology unit to our CMR laboratory from 2009 to 2020.
Conclusion DM1 patients display structural and functional cardiac abnormalities, with variable degrees of cardiac muscle hypotrophy, fibrosis, and fatty infiltration. Such changes, as evaluated by CMR, seem to be associated with the development of ventricular arrhythmias and a worse outcome.
In this study, we investigated cardiac involvement in DM1 focusing on ventricular mass and volumes and tissue changes at CMR, and their association with clinical outcome.

Patient population
We reviewed electronic health records (EHRs) of the Fondazione Toscana Gabriele Monasterio (Pisa, Italy) to retrieve all patients with a genetic diagnosis of DM1, referred for cardiological assessment between 2009 and 2020, and with sufficiently detailed clinical information at the Neurology Unit of the Azienda Ospedaliera Universitaria Pisana (Pisa, Italy). Patients with a history of myocardial infarction, coronary revascularization or cardiac surgery were excluded, because we were interested in studying the nonischemic cardiomyopathy specifically associated with DM1.
Seventy-three patients (43 male individuals, age 48 AE 15 years) with DM1 were identified. One patient was excluded because of the lack of follow-up data, and two because of a prior myocardial infarction. The other 36 patients had not undergone a CMR scan because of a PM/ICD (n ¼ 14) or refusal to undergo the examination (n ¼ 22). The final study population included 34 patients with DM1. Moreover, a control population of 68 healthy volunteers (43 male individuals, age 48 AE 15 years; agematched and gender-matched 2 : 1 with DM1 patients) who had undergone a noncontrast CMR scan at the Fondazione Toscana Gabriele Monasterio was retrospectively retrieved, as a reference control group.
All clinical, laboratory, electrocardiographic, and echocardiographic data at the time of CMR were recorded. Neuromuscular disability was assessed by the Muscular Impairment Rating Scale (MIRS), a five-point scale evaluating extent and severity of muscular impairment (from score 0 -asymptomatic to score 5 -severe proximal weakness). 35 The study complied with the Declaration of Helsinki; all patients provided written informed consent.
All CMR studies were analyzed offline on the Advantage Workstation (GE Healthcare) with dedicated software (MASS 6.1, Medis, Leiden, Netherlands) by one experienced CMR reader blinded to clinical data (GDA, AB, CG, GT); all CMR analysis were revised by a second CMR reader and, in case of discordance, final agreement was found with a third reader. LV and right ventricular (RV) volumes, mass, and global function were calculated on SA cine images and indexed on body surface area. Biventricular fatty infiltration was assessed as banding artifacts in the cine SSFP sequences, confirmed by the presence of hyperintense intramyocardial areas in PDweighted FSE imaging. LGE presence and extent were determined on SA images as areas with signal intensity at least 6 standard deviations above remote, nonenhanced myocardium. 34,36 Native and postcontrast T1-mapping were analyzed by drawing a region of interest in the septum (segments 2, 3,8,9,14). Native T1-mapping was available for a subset of 13 (38%) patients, while both native and postcontrast T1 were available for 9 (27%) of them. In these nine patients, myocardial extracellular volume (ECV) was calculated as (DR1myocardium/ DR1blood) Ã (1-hematocrit), where DR1 ¼ (1/T1postcontrast À 1/T1precontrast). 37 Total LV matrix and cell volumes were calculated from the product of LV myocardial volume [LV mass (LVM) divided by the specific gravity of myocardium (1.05 g/ml)] and ECV or (1 À ECV), respectively. 38 Mass/volume (M/V) ratio was calculated as the ratio between the LVM index (LVMi) and the LV enddiastolic volume index (LVEDVi). The mass/thickness index 39 was calculated as the ratio between the LVM and the maximal end-diastolic thickness (the thickest of the two standard measurements at the anteroseptal and inferolateral basal wall).

Follow-up
Follow-up data were retrieved in August 2022 from EHRs, patients, cardiologists, or general practitioners. All available ECG, Holter recordings and device interrogations performed after CMR examination were searched for evidence of atrioventricular blocks (AVB), intraventricular conduction disturbances (IVCD), atrial fibrillation or flutter (AF/Fl), and repetitive (Lown class 4) 40 ventricular ectopic beats (VEBs). Furthermore, all echocardiographic and CMR reports were checked for the presence of LVSD. Overall, AVB, IVCD, AF/Fl, evidence of Lown class 4 VEBs at Holter monitoring and LVSD were considered as separate surrogate end points. When available, CMR examinations after the first one were analyzed.

Statistical analysis
The R software (version 4.0.2, 2020) was used. Normality was assessed through the Shapiro-Wilk test. Categorical variables are reported as count (percentage). Normal continuous variables are presented as mean AE SD, while nonnormal continuous variables were presented as median (interquartile interval). Paired-sample and unpairedsample Wilcoxon or t-tests were used as appropriate to compare continuous variables; chi-square tests were used for proportions. Logistic regression analysis was performed to find clinical predictors of fatty infiltration or LGE at CMR. Univariate Cox regression models were fitted to the data; survival curves were compared through the likelihood ratio test. Schoenfeld residuals were tested for each model; when data significantly deviated from proportional hazards, time-dependent weights were applied. Accordingly, an average hazard ratio (AHR) is shown instead of the hazard ratio in such cases; these are marked by italic type in Tables S1 and S2, http://links. lww.com/JCM/A531. We computed 95% confidence intervals (CI) for hazard ratio and AHR. A two-tailed P-value of less than 0.05 was considered to indicate statistical significance in all tests.
At the time of CMR, 13 (38%) patients had a history of AVB, 30 (88%) an intraventricular conduction disturbance, and 4 (12%) an atrial fibrillation or flutter. No patient had severe valvular heart disease at echocardiography. The main baseline characteristics of our cohort are reported in Table 1.
Nine patients (26%) presented with mid-wall LGE (mean extent 5 AE 2% of LVM). The distribution of LGE is shown in Fig. 1: the inferoseptal and inferolateral segments were more frequently involved. Fourteen patients (41%) had some areas of fatty infiltration (n ¼ 9 involving the LV, n ¼ 13 the RV). Among clinical parameters, there was only a weak association between dyslipidemia and the presence of fatty infiltration (P ¼ 0.04), whereas there were no significant clinical predictors of LGE. Some exemplar cases with LGE and/or myocardial fatty infiltration are shown in Figures  S1-S3, http://links.lww.com/JCM/A531.
In 13 (38%) patients with T1 mapping sequences, native T1 in the interventricular (1041 AE 53 ms) was higher than healthy controls (1017 AE 28 ms) and approached the upper reference limit (1089 ms) for our CMR laboratory. 41 The extracellular volume could be measured in nine patients (27%), and was slightly increased (33 AE 2%, reference values <30%). 41 Eleven (32%) patients underwent at least another CMR scan [after 1.6 (1.2-2.3) years from the baseline scan]. Table S5, http://links.lww.com/JCM/A531 compares the first and the last examinations. Overall, LVEF increased, though none of the three patients with LVSD and serial CMR evaluations achieved a normal LVEF during follow-up. Eight out of 11 patients (73%) had an increase in LVEF. Of these eight patients, four were not on therapy, one was on loop diuretics, one on b-blockers, one on ACE inhibitors and one on both b-blockers and ACE-inhibitors. Three of them were on thyroid hormone replacement therapy. The anteroseptal wall thickness also increased, whereas the inferolateral wall thickness did not vary significantly over time. The number of patients with LGE or fatty infiltrations did not change. However, patients with LGE showed a nearly significant expansion of LGE mass, both in absolute terms and relative to cardiac mass.

Follow-up
After a median follow-up of 3.7 years (2.0-5.0) after baseline CMR, six (18%) patients died -four of CMR findings and prognosis in DM1 Leali et al. 343 infectious and respiratory complications, two for unknown reasons. Five (15%) underwent device implantation -four (12%) permanent pacemakers and one (3%) cardioverter/defibrillator (ICD). Three pacemakers were indicated for progression of conduction disturbances, one for bradyarrhythmias including sinoatrial pauses up to 3.5 s; the ICD was implanted because of trifascicular block and family history of sudden cardiac death. No device was used for cardiac resynchronization therapy. Data on the other end points are shown in Table S6, http://links.lww.com/JCM/A531. Only one patient developed LVSD during follow-up, while most patients had already developed an IVCD before CMR, so we did not consider these two end points for further analyses.
The influence on our end points of some major potential confounding factors was tested; results are shown in Table S1, http://links.lww.com/JCM/A531. Except for an effect of age on all-cause death, of female gender on atrial fibrillation, device implantation, and death and of high cholesterol on atrial fibrillation, no other significant relationship was noted. A paradoxical weak association between smoking and lower mortality was found that might be explained either by a casual relationship or by a more frequent smoking habit in patients with better clinical conditions.
The association between CMR findings and our end points is presented in Table S2, http://links.lww.com/ JCM/A531. No significant predictor was found for the occurrence of AVB and atrial fibrillation/flutter. A smaller RV volume, a thicker anteroseptal wall and a lower mass/ thickness ratio were associated with device (PM/ICD) implantation.
LGE extent was significantly correlated with the appearance of Lown class 4 VEBs. A lower LV stroke volume, a lower RVEF, a higher M/V ratio, fatty infiltration, and LGE extent were all univariate predictors of all-cause mortality.

Discussion
We investigated cardiac remodeling and tissue changes occurring in DM1 patients. Compared with healthy controls, we found lower cardiac volumes and mass, together with a lower mass/thickness ratio and LV or RV systolic dysfunction in a minority of patients. Tissue characterization showed LGE in 26% and fatty infiltration in 41% of patients. Over 3.7 (2.0-5.0) years, 6 (18%) patients died of extracardiac causes, 5 (15%) underwent device implantation, and 5 of 21 (24%) developed repetitive VEBs on Holter monitoring.
LGE extent was associated with the occurrence of repetitive VEBs; a lower LV stroke volume, a lower RVEF, a higher M/V ratio, fatty infiltration, and LGE extent were predictors of death.  CMR studies on DM1 have been highly heterogeneous as to the prevalence and patterns of cardiac involvement. 16,17,18,33,34,[25][26][27][28][29][30][31] To some extent, this is also because of the rarity of the disease and the limited availability of the technique -as compared with echocardiography. Previous reports seem to confirm echocardiographic findings of left ventricular dilation, 29 hypertrophy, 16,26,29 biventricular systolic dysfunction, [16][17][18]26,27,29,30,31 and wall motion abnormalities. 18,29,35 In addition, CMR could detect a trend towards reduced ventricular volumes 16,17 and mass. [16][17][18] Our study provides additional evidence of a reduction of cardiac cavities and mass in DM1 patients.
Such a trend seems to have been overlooked by echocardiography. 27,34 This is probably because of the higher accuracy of CMR. Echocardiography is known to overestimate mass [43][44][45] when compared with CMR. However, it is also known to underestimate volumes, 46 but the frequently poor acoustic window accompanying muscular dystrophies 4 may account for this apparent inconsistency. It has also been supposed that CMR studies may exclude more severely compromised patients, 27,29,33 thereby ignoring some cases of ventricular dilation. In the latter case, CMR studies might depict an early stage of the disease, before more severe systolic dysfunction occurs.
Various phenomena may be held responsible for the small cardiac size and mass in DM1. Above all, a reduced stroke volume has been noted 17,18 and must be taken into account: this is likely to be the result of the decreased metabolic demands of dystrophic muscles. Indeed, an inverse correlation has been observed between LV enddiastolic volume and a disease-specific scale for myotonic dystrophy. 35 Primary cardiac disease should be considered as well. At pathology, cardiomyocyte atrophy has been observed in DM1, 19,22,39 so cardiac muscle hypotrophy is another likely explanation for a decrease in mass and volume indices, which would parallel the reduced volume of skeletal muscles, similarly to ageing in healthy people. 47 The reduction of mass and volume appears to be proportional, with a M/V ratio similar to healthy people, confirming the results of a previous CMR study on DM1 by Turkbey et al. 17 Another characteristic of DM1 patients was a reduced LV mass/thickness ratio compared with healthy controls. The anteroseptal wall is known to be thicker than the inferolateral wall in healthy people, but the interventricular septum is also a preferential location for LGE 16,18,26,28,29,31 and, seemingly, for fibrosis at pathology 22 in DM1, suggesting a possible relationship with AVB. Indeed, despite a short followup and a limited sample size, we found an increase of ASW over time, which hints at some progressive process. Moreover, a thicker anteroseptal wall and a lower mass/ thickness ratio were associated with device (PM/ICD) implantation during follow-up, even though this result may have been strongly influenced by clinicians' and patients' choices.
Left ventricular dysfunction was quite prevalent (%15%) in our cohort, and a lower LV stroke volume was associated with increased mortality. This is consistent with existing literature. [16][17][18]25,27 Nonetheless, we observed a significant increase in left ventricular ejection fraction (LVEF) over time, which was evident in 8 out of 11 patients with serial CMR examinations, though none of the 3 patients with LVSD and serial CMR evaluations normalized their LV function at follow-up. Therapy does not seem to justify such changes, as half of the patients who apparently improved their function were not on therapy. The increased accuracy of CMR in determining volumes might partly explain the increase in LVEF, which might be influenced by the reduction in LV end-diastolic volumes. Another potentially confounding factor is the link between DM1 and mitral valve prolapse. 48 Further studies are needed to confirm our findings, which might prove relevant to the correct evaluation of systolic function in diabetes mellitus.
Our data confirm that LGE is rather common (%26%) in DM1 and that it is mainly found in the mid-wall layer of septal and inferolateral segments. 16 -18,25 -30 We lacked statistical power to demonstrate an increase of LGE mass over time, but such a trend could be observed. Despite the existence of considerable clues to a link between myocardial tissue alterations at CMR and conduction disturbances and arrhythmias, 16 -18,25 -30,33,34 no definitive evidence has been obtained so far. We could add another piece to the puzzle by observing that LGE extent was associated with the occurrent or repetitive VEBs at follow-up. Moreover, LGE extent showed a significant association with all-cause mortality, even though a clear relationship with cardiac and arrhythmic death could not be assessed because of the limited number of events.
We report that intramyocardial fat is common in both ventricles of myotonic dystrophy type 1 patients and that it may be encountered in as many as 40% of patients. Adipose tissue is a distinctive feature of a number of diseases, where it is likely to play some pathogenetic role. The extent and patterns of infiltration, which we observed seemed to go beyond what we would normally expect in an otherwise healthy patient. Fatty infiltration of the RV in DM1 has been associated with the induction of ventricular arrhythmias, 39 though they were mainly nonsustained. Our study further suggests that there might be a link between fatty infiltration and outcome, as adipose tissue was associated with increased mortality, even though a clear relationship with ventricular arrhythmias could not be demonstrated.
Much of the relevance of our work lies in the inclusion of serial CMR evaluations and the assessment of the prognostic value of CMR in DM1. Indeed, except for one analysis on PR and QRS prolongation over time, 17 more generally purposed longitudinal studies and serial CMR evaluations in these patients are currently lacking or have provided insufficient follow-up data. 29 Evidence exists of right cardiac involvement in DM1 both from mechanical and electrical standpoints. 18 Some link with Brugada syndrome, which appears to arise from the RV outflow tract, has also been suggested. 49 In our cohort, smaller RV volumes seemed to anticipate the need for PPM/ICD implantation; moreover, a lower RVEF was associated with increased mortality.
Several limitations must be acknowledged. The small sample size and the limited number of events did not allow a multivariate Cox regression analysis to be performed. Although AVB, IVCD, AF/Fl, and LVSD are widely accepted predictors of adverse prognosis in diabetes mellitus, 15,16,50 Lown class 4 VEBs need validation as a surrogate end point. Device (PM/ICD) implantation during follow-up was considered a clinically significant event but could not be considered a proper end point, because it is strongly influenced by clinicians' and patients' discretion. Although CMR is offered as part of our routine assessment in DM1 patients, many of them refuse it. Such a low compliance may well be related to the cognitive impairment observed in some of these patients. 1 Although the retrospective observational design and enrolment by chart review are less than ideal, they are common devices for dealing with the rarity of DM1.

Conclusion
Patients with DM1 display several structural and functional cardiac abnormalities, with variable degrees of cardiac muscle hypotrophy, fibrosis, and fatty infiltration. Such changes, as evaluated by CMR, may anticipate the occurrence of electrocardiographic and/or clinical end points, and might help the clinician to guide the management of these cases.

Conflicts of interest
There are no conflicts of interest.