Prophylactic Use of Cardiac Medications and Survival in Duchenne Muscular Dystrophy

ABSTRACT

Introduction/Aims

Prophylactic treatment of left ventricular dysfunction (LVD) in Duchenne muscular dystrophy (DMD) delays onset of LVD, but there is limited data showing impact on survival. Our aim was to describe survival among treated and untreated individuals with DMD.

Methods

Retrospective, population‐based surveillance data from the Muscular Dystrophy Surveillance, Tracking and Research Network (MD STARnet) were used. We analyzed 327 males with DMD born between 1982 and 2009 who were at least 6 years old at the last visit and who initiated cardiac prophylactic medication before age 14 years. Death status was ascertained through vital record linkages and medical record review. Prophylaxis was defined as cardiac medication use at least 1 year before LVD onset (ejection fraction < 55% or shortening fraction < 28%). Age at first visit, corticosteroid use, scoliosis surgery, initiation of noninvasive ventilation, and loss of ambulation were also coded. Cox Proportional Hazard modeling with time‐varying covariates describes associations.

Results

Prophylactic cardiac treatment was documented for 27.7% (n = 90); corticosteroids were used by 60.9% (n = 157). Adjusting for age at first visit and MD STARnet site, prophylactic treatment was associated with a 54% lower hazard of death (HR = 0.46, 95% CI = 0.22–0.93) compared to no prophylaxis. Adjusting for selected clinical covariates did not appreciably change the estimate (HR = 0.46, 95% CI = 0.22–0.99).

Discussion

Initiation of cardiac medication when left ventricular function is normal was associated with prolonged survival in this study of males with DMD. Only one‐quarter of individuals received this treatment, however, indicating a topic of focus for improving care.

1. Introduction

Current care, including corticosteroids and management of respiratory compromise, has contributed to longer survival among those diagnosed with Duchenne muscular dystrophy (DMD); however, left ventricular dysfunction (LVD) continues to be a significant contributor to mortality [1, 2, 3, 4, 5, 6, 7]. Although prophylactic treatment with cardiac medications has been shown to delay the onset of LVD [6, 8, 9, 10], epidemiological investigation into the association between prophylaxis and prolonged survival is limited.

Cardiomyopathy is seen in most individuals with DMD who are over 18 years of age [11]. The exact pathophysiological mechanisms underlying dystrophic cardiomyopathy are unknown; however, DMD‐related cardiomyopathy likely begins with fibrosis [12]. Echocardiograms are routinely used as a screening tool to detect LVD (defined as ejection fraction [EF] < 50%); however, myocardial damage is present well before LVD is observed on echocardiogram. Cardiac MRI (cMRI) is increasingly used to supplement or replace echocardiograms due to greater sensitivity to detect myocardial damage and fibrosis. Additional parameters, such as cardiac strain as measured by either echocardiogram or cMRI, might also be early predictors of low EF [13, 14]. Using advanced imaging, early signs of cardiomyopathy have been seen in children as young as 6 years old [12].

Current DMD treatment recommendations include initiation of angiotensin‐converting enzyme inhibitors (ACEis) or angiotensin receptor blockers (ARBs) by age 10 years with the intent of delaying onset or slowing the progression of cardiomyopathy [15, 16]. The cumulative epidemiological evidence across variable study designs supports this recommendation [9, 10, 17, 18, 19, 20]. Using data from the US population‐based Muscular Dystrophy Surveillance, Tracking and Research Network (MD STARnet), we reported 52.5% of individuals treated prophylactically with cardiac medication had survived without LVD, as measured by echocardiogram, to age 27 years compared to 8.5% of untreated individuals (adjusted hazard ratio [aHR] = 0.39, 95% confidence limit [CL] = 0.22–0.65) [8].

Preservation of cardiac function resulting from early initiation of cardiac medication is expected to result in prolonged survival, but this predicted effect has not been systematically studied [6, 10, 21]. The available data from small cohorts or registries suggest prolonged survival among individuals treated prophylactically with cardiac medication. Here, we use MD STARnet data from well‐characterized and longitudinally followed individuals with DMD to examine the effect of prophylactic cardiac medication on survival.

2. Methods

2.1. Sample

MD STARnet is a US‐based, multisite population‐based cohort for surveillance and research of selected muscular dystrophies. There have been several phases of MD STARnet data collection with differing surveillance periods and contributing sites. To maximize follow‐up and standardize data sources across all data sets, this study analyzed only surveillance data for individuals diagnosed with a childhood‐onset dystrophinopathy and followed within the Colorado (CO), Iowa (IA), and western 21 counties of New York State (wNY) surveillance regions, which are the longest participating programs in MD STARnet [22, 23, 24]. Individuals who were born between January 1, 1982 and December 31, 1999 were followed through December 31, 2016 and those born between January 1, 2000 and December 31, 2015 were followed through December 31, 2015.

2.2. Medical Record Abstraction

Health and vital status information were systematically collected on all eligible individuals until the end of the surveillance period, their date of death, or when they moved out of the surveillance area. Trained abstractors identified potential cases and completed medical record abstraction using a standardized protocol and data collection tool. Key health information associated with diagnosis, clinical tests, and medical treatments were collected.

Diagnostic data were reviewed by a clinical review committee composed of neurologists with experience in treating individuals with dystrophinopathy and assigned a case status (see Table S1): definite, probable, possible, asymptomatic, or manifesting female [22]. In addition to the assigned case status, phenotype (Duchenne, Becker; see Table S2) was classified using a multivariable algorithm as described previously [25].

Public health authority for birth defects surveillance in CO, IA, and wNY was expanded to permit active case finding and record abstraction of childhood dystrophinopathy.

2.3. Vital Status

Following abstraction, data managers linked eligible individuals to state vital records to identify deceased individuals. For those ascertained before 2014, the National Death Index was also used to identify individuals who may have died outside of an MD STARnet surveillance site. Medical records were also used, where available. All‐cause mortality and the age at death were identified for each verified death.

2.4. LVD

The month and calendar year, as well as clinical results, from each available echocardiogram were collected as documented in the medical record. Clinical results recorded included EF and shortening fraction (SF), as well as the interpretation of function (normal, abnormal, indeterminate, and results not available). An echocardiogram result was defined as abnormal if the left ventricular EF was below 55% or the SF was below 28% if EF was missing [26].

2.5. Prophylactic Cardiac Treatment

All medications in the medical chart were abstracted annually. Cardiac medications were extracted from the abstracted medications and assigned a medication class (see Table S3). To calculate the age at first cardiac medication, the medication year was linked with clinical visits during the same calendar year, and the earliest visit date within a calendar year was used to assign the month of cardiac medication use. Prophylactic cardiac medication was defined as any cardiac medication with age at first use that was at least 1 year before the age at first abnormal echocardiogram or in the absence of any abnormal echocardiogram during follow‐up [27]. The 1‐year threshold was selected to be conservative due to lack of precision about medication start dates and to be consistent with published work [6, 8].

2.6. Corticosteroid Medication

Information about corticosteroid use included complete dates of use (start and stop), from which periods of use were derived (maximum of three periods during which corticosteroids were stopped or started), type of corticosteroid prescribed, and reason for stopping use. Individuals with periods of use of less than 6 months were coded as nonusers. Periods of discontinued corticosteroid use less than 1 year were coded as continued use.

2.7. Additional Clinical Covariates

Mobility status (independent ambulation, ambulation ceased, part‐time, or full‐time wheelchair use) was reviewed from the medical record annually. Changes in mobility status and partial dates (month, year) or ages (year, month) were abstracted at each stage of transition from independent ambulation until full‐time wheelchair use. Age at loss of ambulation was coded as the age at ceased ambulation or full‐time wheelchair use. Completion of scoliosis surgery was recorded along with the month and year of completion. The month and year of bilevel positive airway pressure or continuous positive airway pressure and reason for use were recorded, and documentation of either was used to indicate noninvasive ventilation (NIV). We calculated the first age at which chronic NIV use began.

2.8. Study Inclusions and Exclusions

We included people classified as having definite DMD. To be consistent with inclusion criteria used for the propensity analysis by Porcher et al. [6], we included people who were at least 6 years of age at last follow‐up, did not have a co‐existing condition that could impact survival, and, among people treated prophylactically, initiated cardiac medication before age 14 years. We excluded people who initiated cardiac medication within 1 year of LVD, instead of considering them as not receiving prophylaxis, to minimize the uncertainty of prophylaxis status due to imprecise start dates described in Section 2.5.

2.9. Statistical Analysis

Descriptive statistics are reported (means and standard deviation for parametric continuous data, interval data and counts, and percentages for categorical data). Kaplan–Meier curve (K–M) estimation was used to compare time to death using prophylaxis treatment. HRs and 95% CLs were estimated from Cox proportional hazards modeling with time‐varying covariates. To code the time‐varying covariates, start and stop ages were used for cardiac prophylaxis and corticosteroid use, whereas only start ages were used for scoliosis surgery, NIV, and loss of ambulation. Models were adjusted for ages at the first clinic visit and MD STARnet site as a random effect (limited adjusted models). Additional covariates were selected a priori to approximate adjustments made by Porcher et al. [6] (fully adjusted models) and included ages at loss of independent ambulation (LOA), scoliosis surgery, LVD onset, and NIV as time‐varying covariates. Because we relied on a historical cohort, we conducted supplemental analyses that included examining potential biases introduced by temporal changes in clinical care, analysis of all‐cause mortality, and medication class prescribed. Analyses were conducted using Statistical Analysis Software (SAS) version 9.4 (SAS Institute, Cary, NC, USA).

3. Results

3.1. Sample and Clinical Characteristics

After exclusions, our analytical sample comprised 325 individuals, of whom 94.5% were classified as definite DMD (Figure 1). Prophylactic cardiac medication was documented for 27.7% (90 out of 325) overall. On average, people treated prophylactically had a younger age at first encounter and were followed longer (Table 1). ACEis were the most commonly prescribed medication class among most people (86.7%), with angiotensin II receptor blockers (ARBs) and beta‐blockers (BBs) first prescribed less frequently (7% for each). Multiple cardiac medication classes were prescribed at some point during the follow‐up period for 26.7% of people treated prophylactically (n = 24; ACEi + ARB: n = 5, ACEi + BB: n = 18, ARB + BB: n = 1).

FIGURE 1.

Flowchart of exclusions. ^aComorbid conditions affecting disease progression or survival included: Seizure disorder, spina bifida, cerebral palsy, encephalitis, Xp21.3 deletion, cerebral hemorrhage (near birth). LVD = left ventricular dysfunction.

TABLE 1.

Sample characteristics and baseline echocardiogram results among individuals diagnosed with Duchenne muscular dystrophy by prophylactic treatment with cardiac medications, Muscular Dystrophy Surveillance, Tracking and Research Network (MD STARnet), 1982–2009.

	Cardiac prophylaxis
	No (n = 235)		Yes (n = 90)		All (n = 325)
Characteristic	n	Statistic	n	Statistic	n	Statistic
Age (years) at first clinic visit age, M (SD)	235	4.6 (4.2–5.0)	90	3.9 (3.4–4.5)	325	4.4 (4.1–4.8)
Age (years) at last follow‐up, M (SD)	235	15.1 (14.2–15.9)	90	16.0 (14.9–17.0)	325	15.3 (14.6–16.0)
Duration (years) of follow‐up, M (SD)	235	10.4 (9.5–11.3)	90	12.0 (10.9–13.2)	325	10.9 (10.1–11.6)
MD STARnet site, n (%)
Colorado	127	54.0	31	34.4	158	48.6
Iowa	37	15.7	47	52.2	84	25.8
Western New York State	71	30.2	12	13.3	83	25.6
Year of birth, n (%)
1982–1988	62	26.4	2	2.2	64	19.7
1989–1997	61	26.0	33	36.7	94	28.9
1998–2003	50	21.3	36	40.0	86	26.5
2004–2009	62	26.4	19	21.1	81	24.9
LVD, n (%)	91	38.7	33	36.7	124	38.1
Age at LVD, M (SD)	91	14.2 (13.3–15.1)	33	16.1 (15.0–17.2)	124	14.7 (14.0–15.4)
Corticosteroids, n (%)	128	54.5	70	77.8	198	60.9
Age began corticosteroids, M (SD)	128	6.8 (6.4–7.1)	70	6.5 (5.9–7.0)	198	6.7 (6.4–7.0)
Loss of ambulation, n (%)	130	55.3	55	61.1	185	56.9
Age ambulation ceased, M (SD)	130	11.0 (10.6–11.4)	55	11.4 (10.9–12.0)	185	11.1 (10.8–11.4)
Spinal surgery, n (%)	59	25.1	23	25.6	82	25.2
Age at scoliosis surgery, M (SD)	59	14.8 (14.2–15.3)	23	14.8 (13.9–15.7)	82	14.8 (14.3–15.2)
Initiation of NIV, n (%)	49	20.9	25	27.8	74	22.8
Age at initiation of NIV, M (SD)	49	18.3 (16.4–20.2)	25	17.3 (16.1–18.6)	74	18.0 (16.8–19.3)

Abbreviations: CI = confidence interval; LVD = left ventricular dysfunction; M = mean; NIV = noninvasive ventilation.

The lowest percentage of people prescribed prophylactic cardiac treatment were those born during the earliest birth years (1982–1988) (Table 1). The percentages of people treated ranged from 21.1% to 40.0% among people born between 1989 and 2003, with roughly equivalent percentages from 1989–1997 to 1998–2003. A slightly lower percentage with LVD, and later onset of LVD, was found among people treated prophylactically vs. those untreated. Corticosteroids were initiated at similar ages in both groups, but a higher percentage of corticosteroid use was seen in those treated prophylactically. A slightly higher percentage of people lost independent ambulation in those prophylactically treated vs. untreated, but at equivalent ages. Similar percentages of people underwent spinal surgery and at similar ages among treated and untreated people. Finally, a higher percentage of people treated prophylactically initiated NIV, and at a slightly younger age, than untreated people.

3.2. Time‐to‐Event Analyses

Adjusting for the MD STARnet site as a random intercept and continuous birth year, the aHR for the association between prophylactic treatment and LVD onset was 0.35 (95% confidence interval [CI] = 0.21–0.59) in the current data set. The K–M curve estimation for cardiac prophylactic treatment and survival shows delayed death among people treated prophylactically with cardiac medications compared to untreated (Figure 2). Median survival age was 26.4 years (95% CI = 22.6–NC) among people treated prophylactically and 23.3 years (95% CI = 21.5–24.2) among people untreated. Additionally, 53.9% (SE = 0.13) of treated vs. 36.2% (SE = 0.05) of untreated people survived until 25 years of age.

FIGURE 2.

Kaplan–Meier curve estimation for time from birth to death by cardiac prophylaxis.

Cox proportional hazards modeling with time‐varying covariates showed lower hazard of death among individuals receiving cardiac prophylaxis compared to untreated (Table 2). Adjusting for age at first visit and MD STARnet site did not diminish (move toward the null [1.0]) the association between cardiac prophylaxis and survival. After adding corticosteroid use to the limited adjusted model, the aHR for the association between prophylaxis and survival was slightly diminished, but not substantively (< 10% change in aHR). Further adjustment for additional clinical variables also did not diminish the association but did widen the CI slightly (Table 2, Model 3). Parameter estimates of the covariates are presented in Table S4.

TABLE 2.

Cox proportional hazard modeling predicting survival among individuals with Duchenne muscular dystrophy by cardiac prophylaxis and corticosteroid treatment, Muscular Dystrophy Surveillance, Tracking and Research Network (MD STARnet), 1982–2009 (n = 325).

Model	HR, 95% CL a
Crude, prophylaxis	0.53 (0.27–1.03)
Limited adjustment a	0.46 (0.22–0.93)
Limited adjustment, corticosteroids b	0.49 (0.24–1.00)
Fully adjusted c	0.45 (0.21–0.96)

Abbreviation: 95% CL = 95% profile likelihood confidence limits.

^a Limited model adjusted for first visit age and MD STARnet site as random effect.

^b Corticosteroids added to limited model as time‐varying covariate.

^c Fully adjusted model added ages at corticosteroid use, scoliosis surgery, noninvasive ventilation, left ventricular dysfunction, and loss of ambulation as time‐varying covariates.

3.3. Supplemental Analyses

Our supplemental analyses suggest our primary findings were unbiased by temporal changes in clinical care, analysis of all‐cause mortality, and medication class prescribed (Table S4). The aHR showed an increased association between cardiac prophylaxis and prolonged survival after removing people born during the earliest birth period, prophylactic cardiac treatment was associated with all‐cause mortality among those who later developed LVD, and the magnitude of the association between prophylactic treatment and survival was unchanged when ACEi was the first‐line prophylactic medication.

4. Discussion

Our study shows an association between prophylactic cardiac treatment and prolonged survival among individuals with DMD in a population‐based cohort. The findings support delay of cardiomyopathy as a significant contributor to prolonged survival and provide additional strong support for current recommendations to initiate prophylactic cardiac medication before LVD.

Improved management has resulted in prolonged survival for individuals with DMD. Median survival for those born before 1970 was 18.3 years (95% CI = 18.0–18.9) compared to 28.1 years (95% CI = 25.1–30.3) in a cohort born after 1990 [28, 29]. Poor LV function is associated with death from any cause for those with DMD, suggesting that delay of cardiomyopathy could delay mortality [1].

Numerous studies support the benefit of prophylaxis on LV function but do not directly demonstrate survival benefit, in part because of the need to follow patients through their lifespan [6, 9, 10, 14, 15, 19, 20, 30, 31, 32, 33, 34, 35]. Our demonstration of prolonged survival associated with prophylactic cardiac medication is consistent with the few previous studies. In one of the first such studies, steroid‐naïve males with DMD were randomized to either an ACEi (perindopril) or placebo for 3 years, after which all participants took the ACEi. During 10‐year follow‐up, the early treatment group had significantly fewer deaths than the group initially treated with placebo, suggesting both a survival benefit from prophylactic treatment and earlier treatment was more effective than later treatment [21]. The generalizability of this early study is limited due to the small number of individuals enrolled, the use of radionuclide ventriculography to assess LV function, and inclusion of steroid‐naïve patients. Analysis of French DMD Registry data for all‐cause mortality risk in 576 individuals showed that cardiac prophylaxis was associated with decreased risk of death, with an HR of 0.47 (95% CI = 0.31–0.71) in a fully adjusted model [6], very similar to our HR of 0.45 (95% CI = 0.21–0.96) in the fully adjusted model. In a subset of registry participants, time‐based propensity matching was used to simulate a clinical trial of prophylactic treatment for heart failure. Prophylaxis was associated with prolonged survival; the HR for death was 0.38 (95% CI = 0.17–0.92) in the treated group. This analysis also observed a lower risk of hospitalization for heart failure in the treated group, with similar risks of hospitalization for respiratory failure in both groups, lending further support to the survival advantage being related to preserved cardiac function [6].

Treatment guidelines regarding cardiac prophylaxis have evolved. The initial 2010 DMD care considerations stated that no recommendation could be made about prophylactic use of cardiac medication [36]. The updated care considerations (2018) continue to view the use of cardiac prophylaxis as controversial, with use optional after discussion of risks and benefits [16]. In contrast, the National Heart, Lung, and Blood Institute (NHLBI) working group consensus statement recommended initiation of ACEi or ARB by age 10 years [37]. A European Neuromuscular Centre (ENMC) workshop on cardiac management held in 2018 concluded that cardioprotective medications should be started earlier than the guidelines at the time recommended. However, the optimum age is not known [38]. Thus, although some guidelines strongly suggest early cardiac treatment, the widely used 2018 care considerations are often interpreted to mean that an ACEi or ARB should be considered at around age 10 years and not earlier, which was reflected in our analysis with prophylactic treatment started at a median age of 10.3 (95% CI = 9.9–10.8).

Despite the compelling evidence of the benefit of prophylaxis in delaying cardiomyopathy, prophylactic use of cardiac medications in the United States has not been widely adopted. A survey of 31 cardiologists providing care for people with DMD reported that 68% started an ACEi at age 10 years in the absence of clinical cardiomyopathy [39]. In contrast, the US‐based Muscular Dystrophy Association registry data show that among boys who turned 10 years old after 2016, approximately 35% were on an ACEi or ARB [40]. Consistent with these registry data, we observed that 27.7% of the MD STARnet cohort reported here were started on cardiac medication before evidence of LVD at the mean age of 10 years. Although the observation period for our cohort ended in 2015, before the 2018 recommendations, available data from the literature suggest most boys with DMD in the United States are not receiving cardiac care that is consistent with guidelines and available data.

Studies to date have focused on early initiation of ACEi or ARBs. It is possible that early or prophylactic use of a mineralocorticoid receptor antagonist (MRA) to modify cardiac fibrosis could further improve outcomes and some guidelines suggest this is a reasonable treatment [41]. A newly FDA‐approved corticosteroid (vamorolone) has MRA activity in vitro and is cardioprotective in an aldosterone‐exposed mdx mouse. Thus, it is possible that early use of vamorolone might be cardioprotective [42].

Our data focus on prophylaxis, not treatment of documented cardiac dysfunction, and newer heart failure treatments (e.g., sacubitril/valsartan and sodium‐glucose cotransporter‐2 (SGLT2) antagonists) are expected to further delay death from cardiac failure in DMD [43, 44].

4.1. Limitations

We report surveillance data from medical records in three MD STARnet sites to maximize the length of follow‐up. Medications were only entered annually with stop dates presumed when the medication no longer appeared in the medical record. Dosage information was not collected so standard practice was presumed. Echocardiograms were used to determine the absence of LVD and for our definition of prophylaxis (together with age at initiation of < 14 years). Advanced imaging (cMRI) might have been abnormal despite normal echocardiogram. We did not control for cardiac medications initiated after the onset of LVD, but such treatment would tend to bias our results to the null due to the reduced risk of cardiac‐related death among those treated.

4.2. Conclusions

The MD STARnet data show that treating individuals with DMD with first‐line cardiac medications (ACEi or ARB) during the period when LV function is normal by echocardiogram is associated with delayed onset of LVD and prolonged survival in a large, population‐based sample [8]. During the time period of this study, only one‐quarter of individuals received this treatment, suggesting a focus on improving care.

Author Contributions

Kristin M. Conway: conceptualization, methodology, supervision, funding acquisition, writing – original draft, formal analysis. Shiny Thomas: writing – review and editing, validation, formal analysis. Tahereh Neyaz: writing – review and editing, validation, formal analysis. Emma Ciafaloni: writing – review and editing. Joshua R. Mann: writing – review and editing. Michelle Staron‐Ehlinger: writing – review and editing. Gary S. Beasley: writing – review and editing. Paul A. Romitti: conceptualization, funding acquisition, writing – original draft, methodology, supervision. Katherine D. Mathews: conceptualization, methodology, writing – original draft.

Ethics Statement

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Conflicts of Interest

Katherine D. Mathews serves as an advisory board member for MDA and the FSH Society; is a board member for the Friedreich Ataxia Research Alliance (FARA); receives or has recently received clinical trial funding from PTC Therapeutics, Sarepta Therapeutics, Pfizer, Reata, Italfarmaco, Fibrogen, Italfarmaco, CSL Behring, AMO, and Reata. Emma Ciafaloni has received personal compensation for serving on advisory boards and/or as a consultant for Alexion, Argenx, Biogen, Amicus, Momenta, Medscape, Pfizer, PTC Therapeutics, Sanofi/Genzyme, Sarepta, Janssen, NS Pharma, Wave, and Strongbridge Biopharma; has received research and/or grant support from the Centers for Disease Control and Prevention, CureSMA, Muscular Dystrophy Association, National Institutes of Health, Orphazyme, the Patient‐Centered Outcomes Research Institute, Parent Project Muscular Dystrophy, PTC Therapeutics, Santhera, Sarepta Therapeutics, Orphazyme, and the US Food and Drug Administration; and has received royalties from Oxford University Press and compensation from Medlink for editorial duties. The other authors declare no conflicts of interest.

Supporting information

Table S1. MD STARnet case classification for Duchenne or Becker muscular dystrophy [1].

Table S2. MD STARnet phenotype classification for Duchenne or Becker muscular dystrophy [2].

Table S3. Cardiac medications (generic) classes from MD STARnet surveillance.

Table S4. Cox proportional hazard modeling predicting survival among individuals with Duchenne muscular dystrophy by cardiac prophylaxis and corticosteroid treatment, Muscular Dystrophy Surveillance, Tracking and Research Network (MD STARnet), 1982–2009 (n = 325).

Data Availability Statement

Due to privacy concerns, data from the MD STARnet are not publicly available. Researchers interested in MD STARnet data can contact mdstarnet@cdc.gov. Data used for this analysis are maintained at the University of Iowa MD STARnet site.