Mid-wall striae fibrosis predicts heart failure admission, composite heart failure events, and life-threatening arrhythmias in dilated cardiomyopathy

Heart failure (HF) admission is a dominant contributor to morbidity and healthcare costs in dilated cardiomyopathy (DCM). Mid-wall striae (MWS) fibrosis by late gadolinium enhancement (LGE) imaging has been associated with elevated arrhythmia risk. However, its capacity to predict HF-specific outcomes is poorly defined. We investigated its role to predict HF admission and relevant secondary outcomes in a large cohort of DCM patients. 719 patients referred for LGE MRI assessment of DCM were enrolled and followed for clinical events. Standardized image analyses and interpretations were conducted inclusive of coding the presence and patterns of fibrosis observed by LGE imaging. The primary clinical outcome was hospital admission for decompensated HF. Secondary heart failure and arrhythmic composite endpoints were also studied. Median age was 57 (IQR 47–65) years and median LVEF 40% (IQR 29–47%). Any fibrosis was observed in 228 patients (32%) with MWS fibrosis pattern present in 178 (25%). At a median follow up of 1044 days, 104 (15%) patients experienced the primary outcome, and 127 (18%) the secondary outcome. MWS was associated with a 2.14-fold risk of the primary outcome, 2.15-fold risk of the secondary HF outcome, and 2.23-fold risk of the secondary arrhythmic outcome. Multivariable analysis adjusting for all relevant covariates, inclusive of LVEF, showed patients with MWS fibrosis to experience a 1.65-fold increased risk (95% CI 1.11–2.47) of HF admission and 1-year event rate of 12% versus 7% without this phenotypic marker. Similar findings were observed for the secondary outcomes. Patients with LVEF > 35% plus MWS fibrosis experienced similar event rates to those with LVEF ≤ 35%. MWS fibrosis is a powerful and independent predictor of clinical outcomes in patients with DCM, identifying patients with LVEF > 35% who experience similar event rates to those with LVEF below this conventionally employed high-risk phenotype threshold.

Heart failure (HF) is a leading cause of morbidity and mortality, affecting approximately 2% of the adult population 1 . Idiopathic non-ischemic dilated cardiomyopathy (DCM) contributes significantly to this burden 2 ; estimated to account for 8% of all HF patients 3 and up to 19% of those with an left ventricular ejection fraction (LVEF) of < 35% 4 . Despite the practical adoption of LVEF-based thresholds in clinical trials for the stratification of high-risk DCM patients 5 , substantial evidence now supports a strong incremental role for myocardial fibrosis on late gadolinium enhancement (LGE) magnetic resonance imaging (MRI) to define high-risk phenotypes in this referral population [6][7][8][9][10][11][12] .
Mid-wall striae (MWS) fibrosis of the basal septum is observed in approximately one-third of DCM patients referred for LGE-MRI 13 . Across numerous studies this marker has been associated with elevated risk of allcause death 6,7,10,14 and arrhythmic death or appropriate ICD therapy [7][8][9][10][14][15][16] ; these findings now confirmed in a multicentre setting 17 . However, the value of this unique marker to identify patients at elevated risk of HF hospitalization, an important contributor to patient morbidity and healthcare cost utilization, is poorly explored. Despite included within composite endpoints of several studies 6,[8][9][10]12 , the independent utility of MWS fibrosis to identify patients at elevated incident risk of HF hospitalization has not been previously explored due to limited population size.
In a large, prospectively recruited cohort of patients with DCM undergoing LGE-MRI we assessed the independent prognostic utility of MWS fibrosis to predict incident HF admission. Two composite secondary outcomes were also investigated, these focussed on the impact of MWS fibrosis on broader HF-related outcomes (heart transplantation, left ventricular assist device (LVAD) implantation or death) and arrhythmia-related outcomes (appropriate ICD therapy, sudden cardiac death (SCD), survived sudden cardiac arrest (SCA), or sustained VT requiring cardioversion). All analyses were stratified for patients with an LVEF above versus below conventional high-risk phenotype thresholds.

Methods
This was a prospective observational cohort study of subjects recruited between January 2015 and May 2018 and followed for a minimum of 12 months. The study was a sub-cohort analysis of the Cardiovascular Imaging Registry of Calgary (CIROC) at the Libin Cardiovascular Institute (NCT04367220). As previously described 18 , all patients in Southern Alberta referred for clinically indicated Cardiac MRI are prospectively enrolled with data collection performed using commercial software (cardioDI™, Cohesic Inc., Calgary) to deliver tablet-based patient engagement and consent, standardized patient-reported health questionnaires, indication-driven test protocolling and standardized reporting, followed by automated linkage to electronic health record (EHR) data.
Patients were considered eligible based on the following criteria: (1) referral for LGE-MRI for the evaluation of DCM; (2) confirmation of LVEF ≤ 50% by MRI-based quantification; (3) no prior clinical history of, or prior ICD-10 coded occurrence of myocardial infarction, percutaneous revascularization, or coronary artery bypass grafting (CABG); and (4) no LGE-based evidence of myocardial infarction, as defined by sub-endocardial pattern LGE in a typical coronary artery distribution. Patients with any clinically documented history of congenital heart disease, severe valvular insufficiency or stenosis, or any known cardiomyopathy etiology were excluded. The latter exclusion was inclusive of all patients receiving a final diagnosis of acute myocarditis, cardiac sarcoidosis, cardiac amyloidosis, hypertrophic cardiomyopathy, arrhythmogenic right ventricular cardiomyopathy (ARVC), restrictive cardiomyopathy (e.g.: cardiac amyloid), or constrictive pericardial disease. Only patients completing a minimum of 12-months clinical follow-up were considered eligible 18 .
The study was approved by the Conjoint Health Research Ethics Board at the University of Calgary and all subjects provided written informed consent. All research activities were performed in accordance with the Declaration of Helsinki.
CMR imaging and analysis protocol. CMR imaging was performed using 3-Tesla clinical scanners (Prisma or Skyra, Siemens Healthcare, Erlangen, Germany), as previously described 18 . Standardized CMR imaging protocols were used, inclusive of cine imaging using a steady-state free precession (SSFP) pulse sequence in sequential short-axis views from above the pulmonary valve to beyond the cardiac apex. Long-axis imaging was performed in the two-, three-and four-chamber views. LGE imaging was performed using a standard inversion recovery gradient echo pulse sequence in matched views to cine imaging. LGE imaging was performed 10-min following intravenous administration of gadolinium contrast (0.1-0.2 mmol/kg; Gadovist; Bayer, Inc).
Quantitative image analysis was performed using commercially available software (cvi 42 ; Circle Cardiovascular Imaging Inc., Calgary, Canada) with use of standardized operational procedures (SOP) adherent to published Society of Cardiovascular Magnetic Resonance (SCMR) recommendations 19 . Short-axis cine images were analyzed using semi-automated contour tracing of endocardial and epicardial borders to obtain the LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), LV ejection fraction (LVEF), LV mass, RV www.nature.com/scientificreports/ end-diastolic volume (RVEDV), RV end-systolic volume (RVESV), and RV ejection fraction (RVEF). For the LV, papillary muscles were included in the LV mass and excluded from LV blood volumes. Left atrial (LA) volumes were measured at the LV end-systolic phase prior to mitral valve opening using the bi-plane area-length method from temporally matched 4-and 2-chamber cine views. All volumetric analyses were indexed to body surface area (BSA), where appropriate, using the Mosteller formula. Standardized reporting (cardioDI™, Cohesic Inc, Calgary, Alberta) was used to collect all disease phenotype features, inclusive of regional patterns of replacement myocardial fibrosis. The presence of any myocardial fibrosis and patterns of fibrosis were coded, the latter described as: sub-endocardial, mid-wall striae (MWS), mid-wall patchy (MWP), sub-epicardial, and diffuse, as previously described 20 . All coding was performed by expert readers with a minimum of 5-years of clinical practice experience followed by blinded adjudication by a core laboratory reader (YM). For disagreement in pattern scoring the study was reviewed by a third expert reader (JW) to provide a final consensus coding. The spatial extent of fibrosis was coded for each of the 17 American Heart Association (AHA) segments using a sub-segmental model, each segment divided into 4 transmural zones, as previously validated 21 and as shown in Fig. 1. MWS fibrosis was defined as the visual presence of a linear hyper-intensity of greatest intensity at the mid-myocardial transmural region of the interventricular septum seen in at least 2 contiguous or perpendicular imaging planes.
Collection of clinical outcomes. The primary outcome was time to first incident HF admission, defined as occurrence of an ICD-10 coded heart failure admission (I50.X) followed by confirmation of hospitalization ≥ 24 h duration. Two composite secondary outcomes were selected: a HF-related outcome of HF admission, left ventricular assist device (LVAD) implantation, cardiac transplantation, or all-cause mortality; and an arrhythmia-related composite secondary outcome of appropriate ICD therapy, sudden cardiac death (SCD), survived sudden cardiac arrest (SCA), or sustained VT requiring cardioversion. Heart transplantation or LVAD occurrence was tracked by procedural ICD-10 coding. Device interrogations were blindly reviewed for confirmation of, and appropriateness of therapy delivered for ventricular fibrillation or fast VT (R-R interval < 320 ms). All-cause mortality was determined through the combined consideration of data from Vital Statistics Alberta and in-hospital coded death. All events were referenced to the date of index Cardiac MRI study.
Statistical analysis. All descriptive statistics were expressed as mean ± standard deviation (SD) or median (interquartile range [IQR]). Categorical variables were expressed in counts with percentages. Missing laboratory or ECG data was accounted for by multiple imputation by chained reactions (MICE) with no more than 15% of missing values permitted for any variable entered into multivariate models. Comparison between 2 groups was performed with the independent Student t test or Mann-Whitney U test. Fisher exact or Chi-squared tests were used to compare categorical data. Univariable associations between demographic or CMR characteristics and www.nature.com/scientificreports/ clinical outcomes were performed using Cox proportional hazards regression. Variables that were statistically significant (p < 0.05) were considered eligible for inclusion into multivariate models. All eligible variables were tested for collinearity prior to being included in final multivariable models using stepwise Cox regression. This was restricted to eleven variables to avoid over-fitting and priority given to those with high univariable hazards or high perceived clinical value. Kaplan-Meier curves were generated for LVEF as a binary category with a cut off > 35%. Kaplan-Meier curves were also generated for strata consisting of combinations of LVEF and presence/ absence of mid-wall striae (MWS). A competing risk analysis was performed using a sub-distribution hazards model (Fine-Gray model) to account for the effect of death on the primary outcome, as previously described 22,23 . Statistical analyses were performed using SPSS for Macintosh Version 26.0 (IBM Corp, Armonk NY) and R version 3.5.3.

Ethical approval and consent to participate. The study was approved by the Conjoint Health Research
Ethics Board at the University of Calgary and all subjects provided written informed consent. All research activities were performed in accordance with the Declaration of Helsinki.

Results
Baseline clinical characteristics. A total of 719 patients meeting study eligibility were prospectively recruited; their baseline characteristics described in Table 1. The median age was 57 (IQR 47-65) years old with a higher proportion of males. Thirty-five percent of patients had hypertension, 39% hyperlipidemia, and 15% were diabetic. Approximately one-third of patients had NYHA III/IV symptoms at time of enrolment.

Population differences according to LVEF-based stratification. Stratification by conventional
LVEF-based risk thresholds for "severe" LV dysfunction resulted in 289 patients (40%) with LVEF ≤ 35% and 430 patients (60%) with LVEF 35-49%. Significant differences between these cohorts are shown in Supplemental Table 1. Patients with LVEF ≤ 35% were older, more likely diabetic, and described worse symptom burden by NYHA status. Corresponding MRI measurements showed higher biventricular volumes with a higher prevalence of MWS fibrosis (37% vs 16%, p < 0.001). ACE-inhibitors/ARB, beta-blocker and loop diuretic use was also higher among patients with LVEF ≤ 35%.
Population differences according to fibrosis phenotype stratification. Stratification by fibrosis phenotype resulted in 178 patients (25%) with MWS and 541 patients (75%) without. Significant differences between these cohorts are shown in Supplemental Table 2. Patients with MWS were older and described a worse symptom burden by NYHA status. Corresponding MRI measurements showed higher left ventricular volumes and higher prevalence of LVEF ≤ 35% among patients with MWS (61% vs 34%, p < 0.001). Patients with MWS had significantly greater use of ACE-inhibitors/ARB and loop diuretics compared to patients without.
Associations with the primary outcome. Over a median follow up of 1044 (IQR 721-136) days, 104 patients (15%) experienced the primary outcome of incident HF admission. Univariable analysis identified a number of baseline clinical variables associated with this primary outcome, inclusive of age, BMI, heart rate, systolic BP, diabetes, hypertension, hyperlipidemia, NYHA class, hemoglobin, and serum sodium (Table 1). Similarly, numerous MRI-based characteristics were associated with the primary outcome, including LVEF, LVEDVi, LVESVi, LV indexed mass, RVEF, indexed LA volume and MWS fibrosis. As shown in Table 1, the unadjusted hazard associated with LVEF ≤ 35% was 2.87 (95% CI 1.92-4.27). The respective hazard observed for MWS was 2.14 (95% CI 1.44-3.16).
Stepwise multivariable models were constructed to identify the independent predictive value of MWS fibrosis for the primary outcome. Following adjustment for age, sex, diabetes, hypertension, NYHA III/IV, ACE-inhibitor/ ARB use, beta-blocker use, LVEF, RVEF and indexed LV mass, the presence of MWS fibrosis remained an independent predictor of incident HF admission, providing a hazard ratio of 1.65 (95% CI 1.11-2.47) ( Table 2). In this model, diabetes remained a strong predictor, providing a 2.45-fold risk of HF admission (95% CI 1.59-3.77). Other independent predictors were female gender, NYHA III/IV symptom status, beta-blocker use, worsening LVEF and indexed LV mass. A competing risk analysis was performed to adjust for the interval occurrence of death and its influence on the primary outcome. In this model, MWS fibrosis remained an independent predictor of the primary outcome with a HR of 1.57 (95% CI 1.04-2.37). www.nature.com/scientificreports/ Kaplan-Meier analysis was performed to evaluate event-free survival in patients with and without MWS (Fig. 2a). The cumulative risk of HF admission at 1-year was 12% in those with MWS versus 7% in those without (p < 0.001). Similar analysis stratified by LVEF ≤ 35% showed 1-year cumulative event rates of 14% and 5%, respectively (p < 0.001), as shown in Fig. 2b. Given independent utility demonstrated in multivariable analysis we then assessed their combined value, as shown in Fig. 3. This demonstrated patients with LVEF > 35% plus MWS experienced risk equivalent to those with an LVEF ≤ 35%. Respective hazards for each of the combined phenotype categories is provided in Fig. 4.
Finally, separate multivariable models were constructed to assess the prognostic value of MWS in patients with and without LVEF ≤ 35% (66 and 38 events, respectively). Among those with LVEF > 35%, MWS provided Table 1. Baseline clinical, electrocardiographic and laboratory characteristics for the study population, and for those patients with and without the primary outcome of heart failure admission. Results of univariable regression analysis is shown for associations with the primary outcome. BMI body mass index, BSA body surface area, NYHA New York heart association, ACE-I angiotensin converting enzyme inhibitor, ARB angiotensin II receptor antagonist, LVEF left ventricular ejection fraction, LVEDV left ventricular end diastolic volume, LVEDVi indexed left ventricular diastolic volume, LVESV left ventricular end systolic volume, LVESVi indexed left ventricular end systolic volume, RVEF right ventricular ejection fraction, RVEDV right ventricular end diastolic volume, RVEDVi indexed right ventricular end diastolic volume, RVESV right ventricular end systolic volume, RVESVi indexed right ventricular end systolic volume, LA vol left atrial volume, LGE late gadolinium enhancement. a NTproBNP was clinically performed in 208 subjects at time of CMR imaging. The "bold, asterisk" means that those hazard ratios are statistically significant with a p value of < 0.05.      Stepwise multivariable analysis demonstrated MWS to be independently associated with the composite secondary outcome following adjustment for age, sex, diabetes, hypertension, NYHA III/IV, ACE-inhibitor/ARB use, beta-blocker use, LVEF, RVEF and indexed LV mass, providing a 1.80-fold increased risk (95% CI 1.26-2.58). LVEF was not an independent predictor of the secondary outcome in multivariate analysis.
Kaplan-Meier analysis for occurrence of the secondary outcome in patients with versus without MWS fibrosis is shown in Fig. 5a. Respective cumulative risks of the secondary outcome at 1 year were 15% and 8% (p < 0.001).
One-year cumulative event rates in patients with versus without LVEF ≤ 35% were 16% and 6%, respectively (p < 0.001), as shown in Fig. 5b. The combined phenotype of LVEF > 35% plus MWS was associated with a 1-year cumulative event rate equivalent to those with LVEF ≤ 35% (Fig. 6). Table 3. Multivariable analysis performed for the prediction of the primary outcome in (1) sub-group of patients with LVEF ≤ 35%, (2) sub-group of patients with LVEF > 35%. Variables included in the stepwise multivariate model include age, sex, body mass index, diabetes, hypertension, NYHA III/IV, ACE-inhibitor/ ARB use, beta-blocker use, LVEF, RVEF, indexed LV mass and MWS fibrosis. NYHA New York heart association, RVEF right ventricular ejection fraction, MWS mid-wall fibrosis. The "bold" means that the p-value is statistically significant at < 0.05.

LVEF strata
Hazard ratio (95% CI) P  Stepwise multivariable analysis demonstrated MWS to be independently associated with the composite secondary arrhythmic outcome following adjustment for age, sex, NYHA III/IV and LVEF, providing a 2.23-fold  Kaplan-Meier analysis for occurrence of the secondary arrhythmic outcome in patients with versus without MWS fibrosis is shown in Fig. 7. Respective cumulative risks of the secondary arrhythmic outcome at 1 year were 1.7% and 8% (p < 0.001). One-year cumulative event rates in patients with versus without LVEF ≤ 35% were 5% and 2%, respectively (p = 0.025). The combined phenotype of LVEF > 35% plus MWS was associated with a 1-year cumulative event rate equivalent to those with LVEF ≤ 35% (p = 0.732) (Fig. 8).

Discussion
This study was dedicated to assessing the influence of mid-wall striae fibrosis on incident HF admission rates in patients referred for the CMR evaluation of DCM. Our findings demonstrate MWS pattern fibrosis to be a strong predictor of future HF admission independent of confounding variables, inclusive of LVEF. Among those with an LVEF above 35%, MWS was observed in 16% of patients and was associated with event risks equivalent to those with LVEF ≤ 35%.
Global estimates of LV function (i.e. LVEF) remain the dominate phenotypic arbitrator of therapeutic decisions in the management of systolic heart failure 5,24-27 . However, solitary use of this crude phenotypic marker is increasingly being challenged by the expanding availability of advanced phenotypic markers predictive of major cardiovascular outcomes in this population. Although commonly managed as a singular disease entity, DCM is increasingly recognized to have complex pathophysiology with a broad range of phenotypic expression (Prasad review paper). Cardiac MRI has become increasingly engaged for characterizing unique DCM phenotypes, both for the purposes of excluding known causative states (e.g. cardiac sarcoidosis) and for identifying unique patterns of adverse remodelling. The latter has led to the identification of DCM sub-cohorts demonstrating advanced degrees of replacement fibrosis on LGE imaging in a septal striae patter, this finding consistently associated with greater risk of sudden cardiac arrest (SCA) or appropriate implantable cardioverter-defibrillator (ICD) therapy [6][7][8][9][10][14][15][16] . However, a number of studies have similarly shown prognostic value for non-arrhythmic clinical outcomes, including cardiovascular mortality or transplantation, and broader composite endpoints inclusive, but not focussed on, HF hospitalization 6,8,10,12,28,29 . In a study of 472 DCM patients Gulati, et al. demonstrated that the presence of MWS pattern fibrosis was associated a 2.43-fold (95% CI 1.50-3.92) increased risk of allcause mortality, and a borderline increased risk (HR 1.62, 95% CI 1.0-2.61) of the composite endpoint of HF death, hospitalization or transplantation 6 . The same group subsequently studied a composite endpoint inclusive of all-cause mortality, heart failure hospitalization and aborted SCA among 120 patients with new-onset DCM, describing a 2.97-fold (95% CI 1.37-6.45) increased risk in those with MWS 29 . Combined consideration of all studies including HF admission within composite clinical endpoints has allowed for meta-analyses to explore   25 . The current study provided a sufficient population size with cumulative clinical events to appropriately examine this association in the context of multivariable adjustment, demonstrating that MWS presence is associated with independent risk of HF admission in this referral cohort. A critical observation from this study was the capacity of LGE-MRI to identify DCM patients with mildmoderate LV dysfunction (LVEF 36-50%) who experience equivalent HF event rates to those with severe LV dysfunction (LVEF ≤ 35%). Approximately 1 of 6 patients (16%) with mild-moderate LV dysfunction showed MWS fibrosis and experienced incident event rates similar to those with LVEF ≤ 35%. This was similarly observed for the secondary arrhythmic outcome: those patients with LVEF > 35% and MWS have an identical event curve to those with LVEF < 35% without MWS. This demonstrates the potential for fibrosis-based phenotyping to significantly expand DCM population eligibility for intensive HF therapeutic strategies and primary prevention ICD.
Our study's unique focus on the primary outcome of HF admission recognizes the strong influence of this event has on patient morbidity and healthcare resource consumption. Each HF admission is estimated to incur a median cost of $14,621 USD and associated with repeat admission at 60 days in one-third of patients 35 . Further, occurrence of HF admission has been associated with elevated future risk of mortality, this highlighted by Blackledge et al. 36 who described a 1-year mortality rate of 43% among patients admitted for decompensated HF; climbing to 73% at 5-years. Similar findings were confirmed in two other large cohort studies 37,38 . Accordingly, validation of diagnostic markers with capacity to identify patients at high risk of this clinical outcome delivers expanded opportunity for personalized cardiovascular care strategies.
Limitations. This study was prospectively conducted at two associated hospitals within a single tertiary care healthcare system. Accordingly, our study population may suffer from regional practice bias and would benefit from external validation. Invasive coronary angiography was not mandated in this study and was conducted in accordance with justifiable clinical need. Patients not undergoing invasive angiography were considered to have a non-ischemic etiology on the basis of composite clinical history (i.e. lack of prior myocardial infarction or revascularization) in combination with absence of subendocardial pattern injury on LGE imaging. This approach introduces the potential for an exclusion of DCM patients with incidental ischemic injury patterns (i.e. embolic infarction), however is considered a conservative approach for the exclusion of CAD-related cardiomyopathy justified by prior cohort studies 7,15 . Signal threshold-based quantification of replacement fibrosis volume was not performed. Therefore, a comparison of fibrosis extent by this technique versus the binary classification of fibrosis patterns was not feasible. Finally, as an imaging service-based Registry, serum BNP levels were not consistently ordered and collected within close temporal association to CMR imaging: being captured within 3-months in only 208 subjects. Accordingly, we were unable to include this serum-based heart risk marker in multi-variable analyses.

Conclusions
In a large cohort of patients referred to CMR for evaluation of DCM we demonstrated MWS fibrosis to be a powerful and independent predictor of HF admission, identifying patients at approximately a two-fold elevated risk. This risk was independent of LVEF and permitted the identification of patients with intermediate range LV dysfunction (LVEF 36-50%) who experience equivalent event rates to those with severe LV dysfunction (LVEF ≤ 35%). Similar value was observed for composite endpoints related to both HF and arrhythmia-focussed outcomes. Future randomized controlled trials aimed at expanded the use of intensive heart failure therapies for DCM populations with intermediate range LVEF and MWS fibrosis are warranted.

Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Author contributions
Y.P. analysed the data and was a major contributor in writing of the manuscript. L.L., A.C. and S.D. acquired data and prepared the dataset. A.S. and R.H. analyzed the data and revised the manuscript. J.F., M.S., S.R., R.S. coordinated the registry and the data acquisition. C.L., A.H., B.H., N.M., M.B., L.K. and N.F. contributed the study design and revised the manuscript. J.W. conceived of and designed the study, supervised the work, revised and finalized the manuscript, and is a guarantor of this work. All authors read and approved the final manuscript.

Funding
Funding was provided by the Calgary Health Trust.