Patients who met the inclusion/exclusion criteria and signed the written informed consent form were randomized (1:1:1:1) to 4 groups:18a) usual care (UC) alone; b) a home-based 12-week IMT program; c) a 12-week FES program of the lower limb muscles; or d) a 12-week combination of IMT and FES (IMT + FES). Follow-up was extended to 24 weeks with the aim of exploring the sustainability of the 12-week results. The first visit included a comprehensive medical history, physical examination, anthropometry measurements, and examination tests that were performed by 2 cardiologists blinded to the patient's allocation group. The examination tests included: electrocardiogram, echocardiography, cardiopulmonary exercise testing, 6-minute walk test, inspiratory muscle function test, QoL assessed by the Minnesota Living with Heart Failure Questionnaire, and cardiac biomarkers.

Usual Care Group. Patients allocated to this arm did not receive any physical therapy. They were checked weekly by a physiotherapist who measured their maximal inspiratory mouth pressure at each prespecified visit.

Inspiratory Muscle Training Group. These patients were instructed to train at home twice daily (20minutes each session) for 12 weeks using a threshold inspiratory muscle trainer (Threshold IMT, Respironics Inc) (Figure 1 of the supplementary material). They were instructed by a respiratory therapist to maintain diaphragmatic breathing during the training period. Participants started breathing at a resistance equal to 25% to 30% of their maximal inspiratory pressure for 1 week. The respiratory therapist examined the patients weekly by checking the diary card and measuring their maximal inspiratory pressure. The resistance was modified each session to 25% to 30% of their measured maximal inspiratory pressure.

Functional Electrical Stimulation Group. These patients were trained in a FES program for both legs by a physiotherapist for 45minutes a day, 2 days per week for a total of 12 weeks. Eight adhesive electrodes were positioned on the skin over the quadriceps (5cm below the inguinal fold and 3cm above the upper patella border) and gastrocnemius muscles (2cm below the knee joint and just over the proximal end of the Achilles tendon) of both legs (Figure 1 of the supplementary material). The stimulator (Rehab 400 Cefar Compex) was configured to deliver a biphasic current of low frequency (10 to 50Hz) for 5 s followed by rest for 5 s. The intensity of the stimulation was adjusted to achieve a visible muscle contraction without discomfort.

Inspiratory Muscle Training+Functional Electrical Stimulation Group. Patients allocated to the IMT + FES group received IMT and FES training at the same time for 12 weeks.

All patients were evaluated by a trained cardiologist from the heart failure unit at randomization and at the 12- and 24-week visits. Additional visits were permitted, based on the patient's clinical status. Prespecified protocol visits included the assessment of medical history, physical examination, anthropometry, electrocardiogram, cardiopulmonary exercise testing, 6-minute walk test, inspiratory muscle function test, Minnesota Living with Heart Failure Questionnaire, and blood tests. The cardiopulmonary exercise testing, 6-minute walk test, inspiratory muscle strength test, echocardiography, health-related QoL, and serum biomarkers are described in detail in the supplementary material.

Continuous and categorical variables are presented as the mean ± standard deviation, median [interquartile range], or percentages, as appropriate (Table 1). Continuous variables were compared among groups with 1-way ANOVA or Kruskal-Wallis test for normal and nonnormal asymmetric variable distributions, respectively. Discrete variables were compared using the chi-square test.

The null hypothesis of the study was that, among the 4 intervention arms, the mean peak VO2 absolute differences from baseline were similar at 3 and 6 months. Sample size determination for this study assumed 2-sided testing at the .05 significance α level. The effect size in peak VO2 associated with IMT was based on a pilot study conducted by our group16; in that study, IMT was associated with a significant increase of mean of peak VO2 of 3mL/kg/min (from 9 ± 2.5mL/kg/min to 12 ± 2.5mL/kg/min) at 12 weeks. With these data in mind, we assumed a mean change of 3.0mL/kg/min and a common standard deviation of 3.0. The effect size in FES was based on a recent study of heart failure with reduced ejection fraction,20 where FES was associated with a significant increase of mean of peak VO2 of 2mL/kg/min (from 9.6 ± 3.5mL/kg/min to 11.6 ± 2.8mL/kg/min) at 10 weeks. Due to the lack of any source on the effect of FES in HFpEF, we speculated a similar increase in peak VO2 for patients with HFpEF. Therefore, we assumed a mean change of 2.0mL/kg/min and a common standard deviation of 3.0. For patients allocated to both interventions (IMT + FES), and assuming a synergistic effect of both therapies, we postulated an effect size for both interventions of about 4.0mL/kg/min ± 3.0). Assuming an allocation ratio of 1:1:1:1, a total of 48 patients (12 patients per group) would provide 80% power at an α significance level < .05. In the end, 15 patients per arm (60 patients in total) were estimated, assuming 20% withdrawals or losses to follow-up. The software used for the sample size calculation was GLIMMPSE.

All efficacy endpoints (primary and secondary) were analyzed using a mixed-effects model for repeated measures to compare their Δ (absolute difference from baseline) among the different treatment modalities (between-effect) and over time (within-effect). Contrast among groups was adjusted for multiple comparisons using the Sidak method. The model included treatment, month, and a treatment × month interaction as fixed effects, and the baseline value of the endpoint as covariate. The decision to include the time variable as a random effect (random slope) in any mixed-effects model for repeated-measures was based on a likelihood test statistic. An unstructured covariance metric was always used. However, in Table 2 we have depicted the least square means of the endpoints (and not their Δ); in addition, no adjustment was made by including the baseline value of the endpoint as covariate. All analyses were based using all data available with no imputation of missing values. A 2-sided P value < .05 was considered statistically significant for all analyses. All analyses were performed with Stata 14.2

The mixed-effects model for the repeated measures model showed a significant increase in Δ-peak VO2 for IMT, FES, and IMT + FES at 12 and 24 weeks (Figure 2). When the UC group was used as a reference category, the net mean increase in Δ-peak VO2 at 12 weeks was 2.98, 2.93, and 2.47 for IMT, FES, and IMT + FES, respectively (omnibus P value < .001); furthermore, the individual contrast between each active treatment group and UC was significant (Figure 2). This significant effect persisted, but was less intense, at 24 weeks with net Δ-peak VO2 values of 1.95, 2.08, and 1.56 for IMT, FES, and IMT + FES, respectively (omnibus P value < .001). No statistically significant differences in Δ-peak VO2 were documented either at 12 or 24 weeks when the 3 active arms were compared (Figure 2).

Figure 2.

Change from baseline in peak oxygen uptake among treatments and visits. FES, functional electrical stimulation; IMT, inspiratory muscle training; UC, usual care; Δ-peak VO2, Δ-peak oxygen uptake; Tx, treatment.

(0.2MB).

Overall, there was a decrease in Δ-peak VO2 values of -0.96, -0.78, and -0.85 for IMT, FES, and IMT + FES, respectively for the 24- vs 12-week comparison (omnibus P value < .001); no significant changes were documented for the UC group [+0.06 (P = .84)].

Overall, the 3 active treatment groups showed a significant decrease in the Δ-Minnesota Living with Heart Failure Questionnaire score compared with the UC group at 12 weeks (Figure 3); while no sizable changes occurred in the UC group. The comparison with the UC group revealed a net mean decrease of 13.85, 10.91, and 16.34 for the IMT, FES, and IMT + FES groups, respectively (P ≤ .005 for all comparisons) at 12 weeks. The effect over time persisted for IMT and was borderline for IMT + FES. The effect was no longer significant for FES (Figure 3).

Figure 3.

Change from baseline in the Minnesota Living with Heart Failure Questionnaire score among treatments and visits. FES, functional electrical stimulation; IMT, inspiratory muscle training; UC, usual care; Δ-MLHF, Δ-Minnesota Living with Heart Failure Questionnaire score; Tx, treatment.

(0.21MB).

All treatment groups had a decrease in the Δ-E/E′ ratio at the 12- and 24-week visits (Figure 2A of the supplementary material). However, there were no significant differences for any of the active treatment groups compared with the UC group at 12 and 24 weeks. The Δ-left atrial volume index at 3 months showed no significant differences for any of the active treatment groups compared with UC. At the 6-month visit, a significant decrease in Δ-left atrial volume index occurred only for IMT compared with UC (-9.20; P = .029) (Figure 2B of the supplementary material).

There were no significant differences for Δ-N-terminal pro-B-type natriuretic peptide and Δ-CA125 among treatment arms at 3 or 6 months (Figure 2C, D of the supplementary material).

There were no significant differences for ventilatory efficiency among treatment arms at 3 or 6 months (Figure 3 of the supplementary material).

There was a significant improvement in the distance walked at the 3-month visit for the 3 active arms compared with UC (P < .05 for all). At 6 months, this beneficial effect persisted for the FES and IMT + FES groups (Figure 4 of the supplementary material).

HFpEF patients are usually older, predominantly female, with a high prevalence of comorbid cardiovascular1,2,21 and noncardiovascular conditions such as obesity, balance instability, and/or orthopedic limitations.21 This plethora of comorbidities plays an important role in promoting inactivity, deconditioning, and muscle wasting,22 which ultimately severely impairs the functional capacity of these patients.23 Aerobic exercise training can significantly improved the functional capacity in sexagenarian and obese patients with HFpEF.15 However, the feasibility of conventional exercise-based cardiac rehabilitation programs in older patients with advanced HFpEF is more questionable; indeed, data from current heart failure registries have shown that these patients are less likely to receive referrals to exercise-based cardiac rehabilitation programs after a heart failure admission.24

In a previous pilot study of 26 patients with symptomatic HFpEF,16 our group found that a low-intensity IMT protocol was associated with a significant improvement in exercise capacity and QoL. The results of the present study not only confirm prior findings (at 3 months the magnitude of Δ-peak VO2 was similar between the 2 studies [+2.9 and 2.7mL/min/kg]), but expand the benefits of a 6-month follow-up. Interestingly, the magnitude of the beneficial effect of IMT on peak VO2 was comparable to those found for patients with heart failure with reduced ejection fraction.25–27

To date, only 1 randomized trial has analyzed the effects of this passive intervention in HFpEF patients.17 Karavidas et al.17 included 30 patients with a mean age of 69 ± 8 years, showing that FES improved functional capacity (evaluated by distance in the 6-minute walk test), QoL, and endothelial function with no changes in levels of biomarkers or in diastolic parameters at 6 weeks. Our results are in line with those findings, but also expand the functional benefits of FES to an older and more severe subset of patients with HFpEF. As has been reported for IMT, the magnitude of functional improvement found in both trials agreed with results reported in patients with heart failure with reduced ejection fraction treated with the same FES modality (short duration program, with a total stimulation time < 30h).28

The combined protocol improved peak VO2 during the follow-up but was not superior to the IMT or FES program alone. This lack of incremental effect suggests that, in HFpEF patients, there might be a limited margin of improvement from these types of protocols. Thus, we postulate that once this threshold is reached with either intervention (IMT or FES), no additional benefit can be obtained by combining them.