Contemporary medical therapy, sex-specific characteristics, and outcomes of patients with non-ischemic cardiomyopathy: a prespecified interim analysis of the BIO-LIBRA study

Summary

Background

Contemporary data on characteristics, medical therapy, and outcomes in non-ischemic cardiomyopathy (NICM) with implantable cardioverter defibrillator (ICD) or cardiac resynchronization therapy defibrillator (CRT-D) are lacking, and the role of sex remains unexplored due to historical low enrollment of females in device trials. The purpose of this pre-specified interim analysis of the BIO-LIBRA study was to assess characteristics, medical therapy, and ventricular tachyarrhythmias (VT/VF) or mortality at 12 months in NICM patients with ICD or CRT-D, by sex.

Methods

In this multicenter, prospective, registry study, we recruited patients with primary prevention ICD or CRT-Ds with Home Monitoring®, aiming for 40% female enrollment, utilizing specific recruitment tools. Patients were assessed for the primary endpoint of device-treated VT or VF every 6 months. This study is registered with ClinicalTrials.gov (NCT03884608).

Findings

From May 9, 2019 to October 1, 2021, we enrolled 1000 patients, including 475 (47.5%) females, 30.4% non-white, and 9.2% Hispanic/Latino patients. Beta-blockers were prescribed in 92%, ACE-Inhibitor/ARB in 80%, ARNI in 39%, and SGLT2 in 6% at enrollment with increase in SGLT2 use over time. Diuretics were more frequently prescribed in females. Females had a 52% lower risk of VT/VF or death as compared to men through 12 months (HR = 0.49, 95% CI: 0.31–0.78, p = 0.003), driven by a 54% lower risk of VT/VF (HR = 0.46, 95% CI: 0.25–0.85, p = 0.003), even after adjustments for LVEF or QRS duration. No significant differences were reported by ICD vs. CRT-D.

Interpretation

In a contemporary cohort of patients with NICM and ICD/CRT-D, we report an increased enrollment of females and minorities, an increase in the use of novel guideline-directed medical therapy (GDMT) over time, and a lower risk of ventricular arrhythmias or death in females as compared to men at one year.

Funding

BIOTRONIK Inc.

Research in context.

Evidence before this study

We searched PubMed, from database inception to April 21, 2025, for papers published in English, using the terms “medical device trial”, “females”, and “women”. Our search yielded 9183 results.” There has been a historically low enrollment of women in device trials (15–29%) in prior studies including 15% in MADIT-II, 23% in SCD-HeFT, 25% in MADIT-CRT, and 29% in MADIT-RIT. A meta-analysis assessing the participation of females in completed cardiovascular trials registered in ClinicalTrials.gov between 2010 and 2017 showed that the overall female-to-male ratio in arrhythmia trials was ∼0.60, and the median participation prevalence ratio taking the prevalence of diseases into consideration still showed 0.78 where 1.0 would be equal representation. These rates of female participation in arrhythmia trials did not improve over time. Several white papers have been published by FDA and NIH to increase these rates without much success.

Added value of this study

The BIO-LIBRA study is a landmark prospective clinical study that succeeded to enroll 48% females utilizing a simple set of tools to address female underrepresentation and eliminate system, physician and subject barriers to study participation. The set of tools are published herein in their entirety that investigators and policymakers can utilize in the future to increase enrollment of females in device clinical trials on a global level. The current interim analysis provides information on enrollment and clinical characteristics and short-term outcomes. Long-term outcomes data are collected.

Implications of all the available evidence

The set of tools identified and successfully utilized in the BIO-LIBRA study could enable future device trials to employ these strategies and enroll a more balanced representation of both females and males in future studies. New studies could also focus on patients with ischemic cardiomyopathy as another important disease entity.

Introduction

Heart failure (HF) patients with non-ischemic cardiomyopathy (NICM) and severely reduced left ventricular ejection fraction (LVEF ≤35%) have been shown to benefit from the implantable cardioverter defibrillator (ICD) or from cardiac resynchronization therapy with a defibrillator (CRT-D) with improved survival, lower rate of heart failure events and less ventricular tachyarrhythmias.1, 2, 3, 4, 5 However, clinical trials in HF patients were conducted decades ago, and recently, the DANISH trial of patients with a NICM did not confirm survival benefit of an ICD.⁶ The introduction of CRT-D and the four pillars of guideline-directed medical therapy (GDMT) beta-blockers, angiotensin-converting-enzyme (ACE-inhibitors)/angiotensin-receptor blockers (ARB)/angiotensin receptor/neprilysin inhibitors (ARNI), diuretics, and sodium-glucose cotransporter-2 (SGLT2) inhibitors, significantly changed the landscape of treatment and outcomes. However, contemporary data on characteristics, medical therapy, and outcomes of all-cause mortality and ventricular arrhythmias in patients with NICM and an ICD or CRT-D, utilizing current GDMT, are not available.

Furthermore, data on the role of sex in ICD or CRT-D patients have been limited by insufficient enrollment of females in device studies comprising less than one third of enrollments.6, 7, 8, 9 Factors perpetuating female-underrepresentation in clinical trials include institutional factors, such as lack of awareness, underestimation of risk, lower referral for device implantations, and importantly, less guideline directed medical treatment in females.¹⁰ Clinical trial factors include study design issues, possible aversion of female participation, and patient-provider gender discordance.^11,12 However, specific recruitment tools can be developed to enhance female enrollment to allow for a more balanced analysis of characteristics, medical therapy, and outcomes of both males and females with NICM. Therefore, the Analysis of Both Sex and Device Specific Factors on Outcomes in Patients with Non-Ischemic Cardiomyopathy (BIO-LIBRA) study was designed to increase the enrollment of females via specific recruitment tools to assess clinical characteristics, medical therapy, and the rate of device-treated ventricular tachyarrhythmias or all-cause mortality by sex in prospectively followed NICM patients indicated for an ICD or CRT-D for the primary prevention of sudden cardiac death, with a particular emphasis on increasing the enrollment of females.

Methods

Study design

The BIO-LIBRA study is a multi-center, prospective, observational study aimed to assess clinical characteristics and clinical outcomes, including all-cause mortality and device-treated ventricular tachyarrhythmias in 1000 NICM patients with an ICD or CRT-D implanted for primary prevention of sudden cardiac death. In BIO-LIBRA, we aimed to promote a study-wide minimum of 40% female enrollment. Components of the clinical strategy to address root causes of decreased female enrollment including system barriers, institutional factors, clinical trial factors, and patient factors are listed in detail in Supplementary Table S1 (the BIO-LIBRA model). Specifically, we utilized 1) less stringent eligibility criteria to allow for eligibility of both males and females with primary prevention ICD/CRT-D indication without excluding subgroups; 2) a study team and PI who actively created awareness of the importance of female recruitment at each site; 3) a BIO-LIBRA trial guide; 4) physician and coordinator talking points to allow discussion of female-specific concerns related to clinical trials; 5) close tracking of female enrollment in the study and provided regular updates to sites; 6) regular sharing of consenting tips for females at investigator meetings and in newsletters; 7) purposeful inclusion of an increased number of female leaders, PIs, and investigators in the study; 8) the inclusion of a representative pool of enrolling sites across the United States; 9) a pragmatic trial design including standard of care visits and use of remote monitoring to decrease the need for in-person clinic visits. The BIO-LIBRA trial guide included multi-generational visual representation of patients from all walks of life, educational materials, and a detailed FAQ section to aid enrollment (Supplementary Figure S1 A–C). The study protocol has been approved by the institutional review board at each of the participating centers. Details of the study design have been previously published elsewhere.¹³ The study sponsor provided financial support for the conduct of the study, provided input for enrolling site selection, jointly developed the protocol with PIs, developed the electronic data capture system for data collection, and performed regular site monitoring visits to ensure data accuracy. The study Principal Investigators, Valentina Kutyifa and Jeanne Poole, had an independent role in developing the study design. All reported VT, VF, or death events were independently adjudicated by the Arrhythmia Committee and the Clinical Events Committee. Both committees were organized and managed by the University of Rochester. The ECG Core Laboratory independently reviewed all 12-lead ECGs in the study (Supplementary Figure S2). All study reports and study analyses have been independently performed at the University of Rochester of the Medical Center by statisticians Drs. Beck, McNitt, and programmer Mrs. Schleede. The Principal Investigators vouch for the design and the results presented in this manuscript. This study is registered with ClinicalTrials.gov (NCT03884608).

Inclusion and exclusion criteria

Patients in the study had to meet the following criteria for study inclusion: patient age≥18 years; non-ischemic etiology of cardiomyopathy defined as cardiomyopathy not primarily caused by coronary artery disease or myocardial infarction; current guideline-defined indication for a de novo, primary prevention ICD or CRT-D implantation^2,14; a successful implantation of a de novo BIOTRONIK ICD or CRT-D device and commercially available leads no more than 30 days prior to consent or 30 days post consent; ability to provide informed consent; availability for standard-of-care follow-up visits, and willingness to utilize BIOTRONIK Home Monitoring via CardioMessenger (BIOTRONIK SE & Co. KG, Berlin, Germany). Exclusion criteria included secondary prevention ICD indication; ischemic etiology of cardiomyopathy; enrollment in any other investigational device or drug trial that might affect the studied outcomes; expectation of a heart transplantation or ventricular assist device within 1 year; life expectancy less than 1 year; or pregnancy at the time of consent.

Recruitment and follow-up

All enrolling centers have received an IRB approval at each enrolling site. All patients provided written informed consent prior to participation. From May 9, 2019 to October 1, 2021, a total of 1000 patients were enrolled at 48 electrophysiology centers in the United States with representative distribution across all regions (Fig. 1). Baseline visit included demographics, cardiac history, medication list, echocardiogram, 12-lead ECG, as well as the collection of physician-directed standard of care echocardiographic and MRI imaging data in the study. Detailed analysis of echocardiography, ECG and MRI data will be forthcoming as secondary manuscripts in the study. Follow-up visits in the study included a 6-month Home Monitoring visit and an annual in-clinic visit. The annual in-clinic visit included reviewing the Home Monitoring report, collection of data from the most recent physical examination, documentation of changes in cardiovascular and diabetes medications, an adverse event assessment, a device interrogation, and review of all device-classified episodes with antitachycardia pacing (ATP) or shock treatment. In the present manuscript we report on clinical characteristics, medical therapy, and outcomes of 1000 enrolled patients with one-year follow-up completed, as the combination of the pre-specified first and second interim analyses. The final study analysis will be conducted after three years of follow-up.

Fig. 1.

Overview of BIO-LIBRA Site Locations across the United States.

Device implantation

Patients in the study were implanted with a BIOTRONIK ICD or CRT-D according to current clinical standards and based on current guideline-based clinical indications for an ICD or CRT-D. Lead implantation and conversion testing of induced VF in ICD and CRT-D patients were performed according to standard procedures at each enrolling center. Only implanted devices compatible with Home Monitoring via CardioMessenger were considered, to allow for remote monitoring visits and uniform assessment of the ventricular tachyarrhythmia endpoints. Uniform VT/VF programming has been utilized in the study (Supplementary Table S2), with SVT discrimination programmed ON (Supplementary Table S3).

Endpoints

The primary endpoint of the BIO-LIBRA study was the combined risk of all-cause mortality and treated ventricular tachycardia (VT) or treated ventricular fibrillation (VF) events by sex and by implanted device type (ICD vs. CRT-D). Secondary objectives included the evaluation of all-cause mortality, VT or VF, cardiac death, and sudden cardiac death (SCD), analyzed by sex and by the implanted device type, not included in this current report. Other outcomes included the evaluation of cardiovascular and diabetes medication use and dosage during the study. Device-treated arrhythmia episodes with ATP or shock treatment that were identified during follow-up at each study site were adjudicated by the Arrhythmia Committee and the Clinical Event Committee by reviewing the episode EGMs and classifying each device-treated episode as an appropriate or inappropriate device intervention per the study protocol.¹³

Statistical analysis

The BIO-LIBRA Study is designed to enroll up to 1000 patients with non-ischemic cardiomyopathy who receive an ICD or CRT-D device per standard guidelines. Patients will be followed for up to 3 years post-implant. Based upon data from MADIT-CRT (327 VT/VF events in 1820 patients with 2.4 years of follow-up), the anticipated 3-year risk of VT/VF is expected to be 15–18% while the 3-year mortality risk is anticipated to be 7–10%.¹⁵ Given the enrollment goal of 1000 patients, it is anticipated that there will be 150–180 VT/VF events and 70–100 death events. A higher event rate is possible since BIO-LIBRA will enroll patients with more advanced HF (i.e., NYHA class III), who were not present in MADIT-CRT. In this study, to ensure a sufficient sample size to evaluate sex-specific outcomes, a study-wide minimum of 40% female enrollment will be promoted. Additionally, it is estimated that about 50–60% of patients will be implanted with a CRT-D (based on the DANISH ICD trial).⁶

Baseline clinical characteristics were displayed for the total patient cohort and by sex. Continuous variables are expressed as mean ± SD, categorical data are summarized as frequencies and percentages, as appropriate. Baseline characteristics were compared between females and males using Wilcoxon rank-sum test for continuous variables and Chi-square-test test for categorical variables. Cumulative probabilities of ventricular tachyarrhythmia events or all-cause mortality by subject sex and device type are displayed according to the Kaplan–Meier method, with comparisons of cumulative event rates by the log-rank test.

Univariate Cox proportional hazards regression analyses were performed to identify candidate variables (Supplementary Tables S4 and S5). Multivariable Cox proportional hazards regression analyses were used to assess the risk of all-cause mortality or ventricular tachyarrhythmia events by sex and by device type. The clinical covariates used for adjustment in the Cox proportional hazards regression models were chosen using the best subset regression methodology which scores the models using a chi-square statistic. The “best” models for each number of covariates (i.e., model size) were fit. The final model included only those predictors which were statistically significant at p < 0.05. Additionally, because age had a statistically significant association with the endpoint of death at p < 0.05, it was also included in the multivariable models. To confirm the choice of adjustment covariates, a stepwise procedure was also used as a check.

Multivariable Cox proportional hazards regression analyses models for VT/VF or death were adjusted for blood urea nitrogen at baseline, and body mass index (BMI) > 37. Multivariable Cox proportional hazards regression analyses models for VT/VF were adjusted age greater or equal 67 years, and left ventricular ejection fraction at baseline. The proportionality assumption of the Cox regression model was tested by testing time interaction parameters.

All analyses in the present study were carried out on an Intention-to-Treat (ITT) basis, with patients analyzed according to the device at initial implant. All statistical tests were two-sided, a p-value of <0.05 was considered statistically significant. The exception to this was for interactions. For these analyses a p-value of <0.01 was used. Analyses were carried out with SAS software (version 9.5, SAS institute, Cary, North Carolina).

Role of the funding source

The study sponsor provided financial support for the conduct of the study, provided input for site selection, jointly developed the protocol with PIs, developed the electronic data capture system for data collection, and performed regular site monitoring visits to ensure data accuracy. The funder had no role in data analysis, data interpretation, or writing of the report.

Results

A total of 1000 patients were enrolled in the BIO-LIBRA study during the 2.5 years between May 2019 and October 2021 (partially conducted during the COVID pandemic), with an average of 0.69 patients enrolled per site per month. Of the 1000 enrolled patients, 475 patients were females (47.5%), far exceeding our enrollment goal of 40% female enrollment. At the time all patients had completed their one-year follow-up, a total of 74 patients (7.4%) died and there were 142 patients (14.2%) with VT or VF. Patients lost to follow-up or withdrawals in the study were infrequent (5 patients vs. 7 patients, respectively, and similar in both males and females (Supplementary Figure S3).

Clinical characteristics

Clinical characteristics of the patients enrolled in the study by sex are shown in Table 1. The mean age was 62.0 ± 12.7 years, and LVEF was 29.6 ± 11.6%. A total of 304 patients (30.4%) were non-white, and a total of 92 patients (9.2%) were Hispanic or Latino. The majority of the patients with a reported NYHA Class had either NYHA Class II HF symptoms (44.7%) or NYHA Class III (50.6%). A total of 336 (33.7%) patients had diabetes, 76.7% reported hypertension, 17.6% had chronic kidney disease, and 12.5% had paroxysmal atrial fibrillation. Females in BIO-LIBRA reported more advanced heart failure symptoms and they were less often reported to have a history of prior atrial fibrillation (Table 1). The majority of BIO-LIBRA patients have been diagnosed with NICM within three years of enrollment (66%), with 46% diagnosed within the prior one year. About 40% of the patients did not report HF hospitalizations within 12 months prior to the implant procedure. The etiology of NICM was unknown in most patients (reported as idiopathic in 84%). Other etiologies were genetic, sarcoid, or chemotherapy-induced.

Table 1.

Baseline clinical characteristics in the total cohort and by sex.

Clinical characteristics	Total	Females	Males	p-valuea
Number of patients	1000	475	525
Age at Enrollment (years) (mean ± SD)	62.0 ± 12.7	62.6 ± 12.4	61.4 ± 12.9	0.222
ICD implantation (n, %)	547 (55%)	241 (51%)	306 (58%)	0.016
White Race (n, %)	601 (60.1)	283 (59.6)	318 (60.6)	0.383
Asian Race (n, %)	11 (1.1)	7 (1.5)	4 (0.8)
African American Race (n, %)	287 (28.7)	142 (39.9)	145 (27.6)
American Indian/Alaska Native Race (n, %)	4 (0.4)	1 (0.2)	3 (0.6)
Native Hawaiian or other Pacific Islander (n, %)	2 (0.2)	2 (0.2)	0
Unknown Race	95 (9.5)	40.0 (8.4)	55.0 (10.5)	0.268
Hispanic or Latino Ethnicity (n, %)	92 (9.2)	43 (9.1)	49 (9.3)	0.878
Unknown Hispanic or Latino Ethnicity (n, %)	57 (6)	25 (5.3)	32 (6.1)	0.571
NYHA Class I (n, %)	24 (2.8)	13 (3.1)	11 (2.5)	0.026
NYHA Class II (n, %)	378 (44.7)	167 (40.3)	211 (48.8)
NYHA Class III (n, %)	428 (50.6)	229 (55.3)	199 (46.1)
NYHA Class IV (n, %)	16 (1.9)	5 (1.2)	11 (2.5)
LVEF (%) (mean ± SD)	29.6 ± 11.6	30.3 ± 12.2	28.9 ± 11.1	0.195
BMI (mean ± SD)	31.1 ± 7.9	31.8 ± 8.6	30.5 ± 7.2	0.044
Diabetes Mellitus (n, %)	336 (33.7)	159 (33.5)	177 (33.8)	0.921
Hypertension (n, %)	767 (76.7)	350 (73.7)	417 (79.4)	0.032
Paroxysmal Atrial Fibrillation (n, %)	124 (12.5)	46 (9.8)	78 (14.9)	<0.001
Persistent Atrial Fibrillation (n, %)	108 (10.9)	30 (6.4)	78 (14.9)	<0.001
Chronic kidney disease (n, %)	176 (17.6)	74 (15.6)	102 (19.4)	0.110
BUN (mg/dl)	21.3 ± 11.4	20.5 ± 11.0	22.0 ± 11.6	0.010
Beta Blocker (n, %)	917 (92.2)	436 (92.6)	481 (91.8)	0.650
ACE Inhibitor (n, %)	256 (25.6)	105 (22.1)	151 (28.8)	0.016
ARB (n, %)	175 (17.5)	97 (20.4)	78 (14.9)	0.021
ARNI (n, %)	356 (35.6)	177 (37.3)	179 (34.1)	0.296
SGLT2 Inhibitors (n, %)	57 (5.7)	24 (5.1)	33 (6.3)	0.489
MRA (n, %)	389 (38.9)	187 (39.4)	202 (38.5)	0.710
Diuretic (n, %)	744 (74.8)	370.0 (78.6)	374.0 (71.4)	0.009

Abbreviations: NYHA, New York Heart Association Class; LVEF, left ventricular ejection fraction; BMI, body mass index; ARB, Angiotensin-receptor blocker, ARNI, angiotensin receptor/neprilysin inhibitors; SGLT2, sodium-glucose transport protein 2 inhibitors; MRA, mineralocorticoid receptor blocker.

^ap-values are calculated for differences between females and males. A p-value < 0.05 was considered significant.

Guideline-directed medical therapy (GDMT) at baseline and therapy optimization

In BIO-LIBRA, beta-blockers were prescribed in 917 patients (92%), while ACE-Inhibitors (n = 256) or Angiotensin Receptor Blockers (n = 175) or angiotensin receptor/neprilysin inhibitors (ARNI) (n = 356) were prescribed in only 80%. Diuretic use was high (n = 744, 74.8%), and was more frequently prescribed in females (79% vs. 71%, p = 0.009) (Fig. 2A).

Fig. 2.

A. Sex-Difference in GDMT in BIO-LIBRA. B. Utilization and Changes of Novel GDMT, ARNI and SGLT2 in BIO-LIBRA over Time. C. Utilization and Changes of Novel GDMT, ARNI and SGLT2 in BIO-LIBRA over Time by Sex. ∗ p-values for SGLT2 use between men and women (Chi-sq cont adj): Baseline: 0.482, 12 Month: 1.000, 24 Month: 0.204. p-values for SGLT2 use between men and women (Chi-sq cont adj): Baseline: 0.328, 12 Month: 0.218, 24 Month: 0.674.

A total of 356 patients (35.6%) were prescribed with angiotensin receptor/neprilysin inhibitor (ARNI) at baseline, and 57 patients (5.7%) were prescribed with sodium-glucose transport protein 2 inhibitors (SGLT2), without differences between sexes (Table 1). The use of ARNI stayed consistent over time, with 42% use reported at 12-month and 24-month. However, the use of SGLT significantly increased during follow-up, with 11% of the patients prescribed at 12-month and 15% at 24-month (Fig. 2B). Interestingly, there were no sex-based differences in SGLT2 and ARNI use over time (Fig. 2C). Information on dosages for novel therapies of SGLT2 and ARNI are highlighted in Supplementary Table S6. Interestingly, dosages for SGLT2 remained similar over time, while dosages in sacubitril and valsartan seemed to decrease over time.

Sex, and device differences in outcomes in BIO-LIBRA at 12 months

One-year cumulative probability of device-detected VT/VF or death was 6% in females and 13% in males (log-rank p < 0.001), with a statistically significant difference (Fig. 3A). In the multivariable Cox hazards regression model, we found a 51% lower risk of VT/VF or death in females as compared to males (HR = 0.49 [95% CI: 0.31–0.78], p = 0.003), after adjustments (Table 2). Device-specific analysis showed an 11% one-year cumulative probability of VT/VF or death in patients with an ICD and 8% in those with CRT-D (log-rank p = 0.176, Fig. 3C). Multivariable Cox hazards regression model confirmed the non-significant association (Table 2).

Fig. 3.

A. Cumulative Probability of VT/VF or Death by Sex. B. Cumulative Probability of VT/VF by Sex. C. Cumulative Probability of VT/VF or Death by ICD/CRT-D. D. Cumulative Probability of VT/VF by ICD/CRT-D.

Table 2.

Sex and device differences in outcomes in BIO-LIBRA at 12 months.

	HR	95% CI	p-value
Primary Endpoint: VT or VF or death
Female vs. Male	0.49	0.31–0.78	0.003
CRT-D vs. ICD	0.76	0.49–1.20	0.239
Secondary Endpoint: VT/VF
Female vs. Male	0.46	0.25–0.85	0.012
CRT-D vs. ICD	0.83	0.47–1.48	0.531

Multivariable Cox proportional hazards regression analyses models for VT/VF or death were adjusted for blood urea nitrogen at baseline, and body mass index (BMI) > 37. Multivariable Cox proportional hazards regression analyses models for VT/VF were adjusted age greater or equal 67 years, and left ventricular ejection fraction at baseline.

Bold p-value of <0.05 was considered significant.

Regarding the key secondary endpoint of one-year probability of VT/VF, females had a 3% event rate as compared to 8% in males (p = 0.001), corresponding to a 54% lower rate in the multivariable model (HR = 0.46 [95% CI: 0.25–0.85], p = 0.012) (Fig. 3B and Table 2). There were no significant differences in VT/VF between ICD vs. CRT-D patients (Fig. 3D and Table 2). Competing risk analyses confirmed similar findings showing a 63% lower risk of VT/VF in females as compared to males (HR = 0.37, 95% CI: 0.20–0.71, p = 0.003) and no significant differences between patients with CRT-D vs. an ICD (HR = 0.89, 95% CI: 0.48–1.63, p = 0.699).

Sensitivity analyses

Sensitivity analyses were performed to examine an extensive list of covariates related to echocardiographic, clinical, and ECG measures to ascertain if sex is an independent predictor of outcomes. Multivariable Cox proportional hazards regression model with LVEF dichotomized at ≤23% had very little impact on the association of female sex with the primary outcome (HR = 0.53 when including LVEF as compared to HR = 0.49 in the primary multivariable). Likewise, the inclusion of QRS duration had a negligible change in the association of female with the primary endpoint (Supplementary Table S7).

As an additional sensitivity analysis, models using stratification for site and adjustment for clustering using a robust sandwich covariance method were estimated. Using clustering, we showed that females had a 54% lower risk of VT/VF or death as compared to males (HR = 0.46 [0.28, 0.76, p = 0.002). Using site stratification, we showed that females had a 52% lower risk of VT/VF or death as compared to males HR = 0.48 [0.31, 0.72, p = 0.001] as compared with HR = 0.48 [0.30, 0.76, p = 0.002] in the primary multivariable model [Table 2]. The results for CRT-D:ICD were likewise very similar (Supplementary Table S8).

One-year safety outcomes in BIO-LIBRA

The BIO-LIBRA study recorded information on procedure, device, and lead-related complications. Over the course of 12 months, a total of 18 (1.8%) patients had an implant-related complication and 2 patients had pulse generator related complications. Lead-related complications included 7 (0.7%) RA lead issues, 27 (2.7%) RV/ICD lead complications, and 15 (1.5%) LV-lead related complications (Supplementary Table S9).

Discussion

The BIO-LIBRA study prospectively enrolled a large contemporary cohort of non-ischemic cardiomyopathy patients with an ICD or CRT-D in order to assess sex-and device type-specific outcomes of ventricular tachyarrhythmias (VT/VF) or death. We developed a specific framework, the BIO-LIBRA model, to increase the enrollment of females to 47.5% in the study. The current manuscript provides insights into clinical trial conduct to increase the enrollment of females and describes the clinical characteristics, medical therapy, and one-year outcomes of this patient cohort. We have the following findings to report: 1) We enrolled 47.5% females in this study, exceeding our performance goal of 40%; 2) A total of 30.4% of the patients were non-white, and 9.2% reported their ethnicity to be Hispanic/Latino; 3) The BIO-LIBRA cohort was reasonably well treated with GDMT at baseline and novel GDMT use increased over time; and 4) we identified potential sex-specific differences in the combined rate of VT/VF or death, although we have been unable to adjust our multivariate models to the presence of late gadolinium enhancement, or genetic testing, as these tests were not widely available in the study.

While implantation of an ICD/CRT-D has been established as a life-saving and guideline-based therapy for NICM patients with severely reduced LVEF ≤35% and NYHA class II or III HF symptoms,1, 2, 3, 4, 5 large randomized trials of ICD/CRT-D enrolled a very low rate of females and minorities, precluding the certainty of benefit in these populations.16, 17, 18 Underrepresentation of females is widely recognized in cardiovascular and electrophysiology clinical trials. This is potentially explained by differences in disease pathophysiology between males and females,^19,20 as females present with heart disease at an older age,¹⁶ and are reported to be less often treated with guideline-directed medical therapy and less likely to be referred for interventions and device therapy.¹⁶ On the other hand, females are more likely to have NICM, a higher prevalence of left bundle branch block (thereby amenable to CRT), a lower prevalence of atrial fibrillation, and a lower rate of ventricular arrhythmias.¹⁶ Such differences could be attributed to differences in sex-specific disease development, and a potential protective role of sex hormones.¹⁶ Government agencies such as the FDA and NIH, and societies including the American College of Cardiology (ACC) and Heart Rhythm Society (HRS) urged physicians to close the enrollment gap by focusing on community outreach and educational efforts. A prior initiative launched by Boston Scientific, called WIN-HER™ developed and utilized recruiting materials customized specifically for females in the MADIT S-ICD and ASAP-TOO clinical trials (terminated early due to the COVID-19 pandemic), in an effort to increase the number of female enrollees. However, enrollment of females in device studies remain low.

In the present study, we expanded on prior knowledge, and successfully utilized a set of simple tools to increase the enrollment of females and minorities through planned clinical trial processes without introducing selection bias in the study. First, the study design was primarily pragmatic, encompassing a wide patient population without excluding subgroups by age, comorbidities, or other factors potentially excluding females who present with HF at an older age. Second, the study protocol included only standard of care study visits and the use of remote monitoring to lower time demands, which is a special concern for female participants who are often providing care to other family members. Furthermore, a few additional strategy steps were implemented, 1) we included a female physician site PI or Co-PI, as often as possible, 2) we developed special clinical trial patient and investigator materials that highlighted a representative cohort of females, males, and their families, 3) we provided scripts for coordinators and investigators to use when talking to females, and 4) encouraged sites to consider eligible females via investigator meetings and clinical trial newsletters, and shared best practices from sites with a high enrollment of females (Table 1). Our success, with nearly 50% enrollment of females and substantially higher rates of racial/ethnic minority enrollment, sets a new bar that should be met or exceeded in future cardiovascular/device trials.

In the past decades, we have seen significant improvements in the medical treatment of HF associated with a notable reduction of SCD and mortality across the spectrum of clinical cardiovascular trials, as reported previously.²¹ Likely due to this fact, the DANISH trial failed to demonstrate a statistically significant survival benefit of an ICD in NICM patients with optimal use of GDMT.⁶ These results emphasize the importance of high adherence to GDMT in eligible patients, but also raises questions about the generalizability to broader HF populations. BIO-LIBRA aimed to close this knowledge gap and assess contemporary outcomes of NICM patients implanted with an ICD or CRT-D in the United States. Importantly, we are also able to track the uptake of GDMT across the years of study follow up, especially the use of ARNI and SGLT2 medications. Use of BB and ACE/ARB were less than recommended target dose at baseline, however the uptake of ARNI and SGLT2 steadily increased over the follow up, reflecting an ongoing therapy optimization effort in patients with implanted devices and NICM.

In BIO-LIBRA, the patient cohort was broadly reflective of U.S. NICM HF patients, with a median age in the low 60's, severely depressed LVEF and frequent comorbidities of diabetes, hypertension, chronic kidney disease, and atrial fibrillation. Most patients were reported to have an idiopathic cardiomyopathy, with low percentages of patients with known sarcoidosis or amyloidosis. While genetic testing and MRI data would have strengthened our analyses, neither were being routinely performed when BIO-LIBRA began enrollment. BIO-LIBRA was a pragmatic prospective study, and while both genetic testing and MRI testing was recommended, neither could be mandated. Specific genetic variants are associated with an increased risk of VT/VF, such as LMNA.²² These additional tests are now included in updated guideline-supported recommendations for the evaluation of patients with newly diagnosed NICM. It has been demonstrated that LGE is a powerful confounder in the association between sex and VT/VF in NICM, in a cohort of 1165 patients where the association between sex and VT/VF was lost after adjustment for LGE.²³ Furthermore, in a large cohort study where both LGE and genetics were taken into account, male sex was no longer a predictor for VT/VF.²⁴ Therefore, future studies are needed with complete information available on both MRI imaging and genetic testing to fully ascertain whether sex itself is predicting VT/VF or differences in the genetic makeup and MRI scar account completely for the differences.

In this interim analysis of BIO LIBRA, we found 6% VT/VF rate through 12 months for all patients, which was 12.8% for men, and 3% for women. In this interim analysis of BIO-LIBRA, we show a 12.5% VT/VF rate for men, and 5% VT/VF rate for women. These rates are surprisingly high. For context, we can look at what the rates were in the SCD-HeFT study, now published nearly 20 years ago. The annual VT/VF rate in SCD-HeFT was 4.9% and did not differ appreciably by etiology of heart failure (ischemic or non-ischemic).²⁵ The SCD rate was ∼3% per year in the control arm of the SCD-HeFT study.²⁶ These observations occurred in patients with heart failurebefore even “triple” heart failure medications were standard, though the use of ARB or ACE was quite high at 94%. For a more contemporary comparison limited to only patients with a NICM, we can look at the DANISH study, where use of triple GDMT was very high.⁶ The SCD rate was only 1.5% per year in the control arm, and the rate of ICD treated VT/VF was 5.6% per year. Given the many differences between these clinical studies, it remains notable that BIO-LIBRA reports a high incidence of ICD treated VT/VF at one year of follow up which was over twice as high in males compared to females.

It is reassuring that procedure and lead-related complication rates were reported to be low in the study. Infections requiring complete system removal were very low (0.8%) in keeping with contemporary data.

Our study evaluated one-year sex- and device-specific differences in ventricular tachyarrhythmias or death and showed that females have a significantly lower rate of ventricular tachyarrhythmias as compared to males. Sex-specific differences have been similarly reported previously in other studies with both implanted and wearable defibrillators, suggesting a similar association.27, 28, 29 However, this has not been confirmed in recent studies utilizing high levels of targeted GDMT. The lack of differences in VT/VF or death by CRT-D vs. ICD in this interim analysis contradicts reported randomized studies which showed a significant mortality when a CRT was added to ICD therapy. Final analysis after a minimum three years of follow up may vary from this early report.

Importantly, we examined an exhaustive list of covariates related to echocardiographic, clinical, and ECG measures, and identified sex to remain an independent predictor of outcomes in this study with an appropriately large number of females enrolled.

Our study has certain limitations. First, BIO-LIBRA is a prospective registry study, and not a randomized trial, therefore, the benefit of the CRT-D or ICD cannot be firmly established without a randomized non-ICD control group. Second, this interim analysis reports only the one-year follow-up, longer follow-up is currently ongoing. Third, the use of novel GDMT is anticipated to increase in accordance with the latest guidelines,³⁰ which may alter the observed VT/VF rates and mortality in the final analyses. In addition, we now know that “NICM” is a heterogeneous group of patients with heart failure who have different risks of sudden arrhythmic death. Genetic testing and MRIs were not consistently performed in this cohort of NICM patients. If all patients had these tests the results reported may have been different. Unlike in a randomized trial where the use of specific testing (such as CMR or genetic testing) can be proscribed, BIO-LIBRA reflects US practice patterns in the management of NICM during the years of enrollment. It may not yet be consistent practice to obtain CMR or genetic testing on all newly diagnosed NICM patients outside of academic centers but should be held as the standard of care for the evaluation of new onset NICM, as genetic testing currently has a Class IIa indication to be performed in the United States in patients with HF or cardiomyopathy.³¹ Therefore, the limited use of CMR and genetic testing in this cohort has implications on the multivariate assessment of predictors of VT/VF, especially when the association between sex and VT/VF is evaluated. In addition, BIO-LIBRA was conducted entirely during the COVID pandemic, further challenging scheduling of CMRs during this time. Nonetheless, CMR data were collected if available within six months of enrollment, was available on 17% of the patients. Finally, device-detected VT/VF occurs at substantially higher rates than SCD and therefore it cannot truly represent SCD risk in this population.

In conclusion, we have successfully enrolled a large proportion of females (47.5%) in a contemporary prospective cohort of patients with non-ischemic cardiomyopathy in the United States, implementing a set of simple clinical trial tools to increase the enrollment of females in the study. The BIO-LIBRA study was so-named as the astrological sign LIBRA symbolizes balance. With an almost even distribution of both sexes, our study provides a more robust analysis than prior clinical ICD and CRT-D studies suggesting important differences in clinical characteristics, medical therapy, and VT/VF rates between females and males. However, the association of VT/VF rates and sex may have been reduced if the analysis included data on genetics and MRI studies. Further efforts are warranted to enroll a larger proportion of women and patients belonging to minority groups in future cardiovascular clinical studies.

Contributors

Valentina Kutyifa MD, PhD–study design, data analysis, data interpretation, writing and revising of the manuscript; accessed and verified the data.

Luigi Di Biase MD, PhD–data collection, reviewing the manuscript.

Karthik Venkatesh Prasad MD–data collection, reviewing the manuscript.

Vilma Torres MD–data collection, reviewing the manuscript.

Aaron Hesselson MD–data collection, reviewing the manuscript.

Craig J. McCotter MD–data collection, reviewing the manuscript.

Gregory Harris MD–data collection, reviewing the manuscript.

Karlene Cox BS–study design, data collection, data interpretation, reviewing the manuscript.

Susan Schleede MS–data analysis, data interpretation, reviewing the manuscript; accessed and verified the data.

E. Kevin Heist MD, PhD–data collection, data interpretation, reviewing the manuscript.

Scott McNitt MS–data analysis and interpretation, reviewing the manuscript; accessed and verified the data.

Mary W. Brown MS–data collection, data interpretation, reviewing the manuscript; accessed and verified the data.

Crystal Miller MS–study design, data collection, data analysis, reviewing the manuscript.

Christopher A. Beck PhD–study design, data collection, data analysis, reviewing the manuscript; accessed and verified the data.

Jeanne Poole MD–study design, data interpretation, reviewing the manuscript.

Data sharing statement

Data collected for the study, including individual participant data and a data dictionary defining each field in the set, will not be made available to others, only to study investigators participating in BIO-LIBRA, upon reasonable request.

Editor note

The Lancet Group takes a neutral position with respect to territorial claims in published maps and institutional affiliations.

Declaration of interests

Valentina Kutyifa MD, PhD: research grants from Boston Scientific, BIOTRONIK, NIH, Spire Inc., speaker fee from Medtronic, Abbott Medical, consultant fees from Medtronic, Abbott Medical, Biotronik, PIVATAL NIH trial DSMB, AHA CSSP committee.

Luigi Di Biase MD, PhD: consultant for Medtronic, Abbott Medical, Biotronik, Zoll, Boston Scientific, Baylis, Zoll, Biosense Webster, Stereotaxis, I rhythm, Zoll, Atricure and has received speaker honoraria/travel from Medtronic, Abbott Medical, Biotronik, Zoll, Boston Scientific, Baylis, Zoll, Biosense Webster, Stereotaxis, I rhythm, Zoll, Atricure and travel support from Medtronic, Abbott Medical, Biotronik, Zoll, Boston Scientific, Baylis, Zoll, Biosense Webster, Stereotaxis, I rhythm, Zoll, Atricure.

Karthik Venkatesh Prasad MD: none.

Vilma Torres MD: received consultant fees from Biotronik.

Aaron Hesselson MD: payment or honoraria from Medtronic and Abbott.

Craig J. McCotter MD: none.

Gregory Harris MD: received speaker honoraria/travel from Biotronik.

Karlene Cox BS–employee of Biotronik Inc.

Susan Schleede MS: none.

Kevin Heist MD, PhD: consultant: Biotronik, Boston Scientific, and HRS Membership Committee Chair.

Scott McNitt MS: none.

Mary W. Brown MS: none.

Crystal Miller MS–employee of Biotronik Inc.

Christopher A. Beck PhD: grants from Boston Scientific, BIOTRONIK, NIH.

Jeanne Poole MD–Research contracts direct to institution from: Biotronik, Boston Scientific, AtriCure, Kestra Medical.