Skeletal Muscle Energetics in Heart Failure Assessed Using ³¹P Magnetic Resonance Spectroscopy—A Systematic Review and Meta-Analysis

Abstract

Introduction: Skeletal muscle (SkM) abnormalities are well-recognised in heart failure (HF). We aimed to systematically review studies of SkM energetics in patients with HF at rest and post-exercise using ³¹phosphorus magnetic resonance spectroscopy (³¹P MRS). Methods: A systematic search of cross-sectional studies used predefined search terms related to HF, SkM energetics, and ³¹P MRS (PROSPERO ID: CRD42023434698). Inclusion criteria for studies are as follows: 1. HF participants versus controls; and 2. SkM energetics assessed using ³¹P MRS reporting BOTH (i) PCr recovery time and (ii) PCr ratios (PCr/Pi and/or PCr/ATP). The primary outcome was SkM PCr recovery time following exercise, comparing patients with diagnosed HF and healthy controls and reported as standardised mean difference (SMD). Results: Of 465 identified studies, 6 met the inclusion criteria and were conducted from 1987 to 2021, comprising 162 participants (N = 86 HF, N = 76 healthy controls). HF patients (mean age 55.1 ± 4.16 years, 49 (56.9%) male) were reasonably matched to healthy controls (mean age 50 ± 8.9 years, 54 (71%) males). Two studies did not report patients’ ejection fractions (EF); the mean EF among patients from the remaining six studies was 24.8%. No studies specifically included participants with HFpEF and none characterised sarcopenia. HF patients exhibited impaired SkM energetics compared to healthy controls, which were characterised by a significantly increased PCr recovery time (SMD: −1.35, CI: −2.11, −0.59). Conclusions: PCr recovery is significantly impaired in patients with HFrEF. Females were under-represented, no HFpEF studies were identified, and no studies linking abnormal SkM energetics directly to sarcopenia were identified.

1. Introduction

Heart failure (HF) is a highly prevalent chronic disease and represents the terminal stage of heart disease of several aetiologies. Approximately 64 million people are affected worldwide [1]. The presence of multimorbidity in HF is associated with frailty and sarcopenia and can lead to skeletal muscle (SkM) myopathy and potentially impaired SkM metabolism [2]. Sarcopenia, the progressive loss of both SkM mass and function, is prevalent in ageing populations [3] and is inadequately explored in patients with HF [4]. While the relationship between HF and SkM quality is acknowledged, alterations in energetics have been less well-studied, as has the association with sarcopenia.

³¹Phosphorus magnetic resonance spectroscopy (³¹P MRS) is a non-invasive technique that allows for the quantitation of SkM energetics. High-energy metabolites (ATP, inorganic phosphate (Pi), and phosphocreatine (PCr)) are quantified as ratios. The PCr/ATP ratio is representative of the energy status within SkM. The PCr/Pi ratio represents the ability of SkM cell’s ability to control energy production. Moreover, ³¹P MRS allows for the dynamic measurement of PCr recovery time, which directly reflects the oxidative re-phosphorylation of creatine and is related to maximal mitochondrial oxidative capacity. Hence, PCr recovery time is a surrogate measure of mitochondrial function [5]. An increase in PCr recovery time may therefore be indicative of dysfunctional oxidative phosphorylation and reduced SkM quality.

Although there is an abundance of current research focussing on the diverse manifestations of HF, a critical gap exists in the place of ³¹P MRS in the investigation and understanding of SkM energetics in HF. Systematic reviews and meta-analyses in patients with HF have investigated the prevalence of sarcopenia and the impact of interventions and treatments on sarcopenia in relation to HF. Frailty in relation to sarcopenia, biomarkers, and clinical outcomes have similarly been investigated in the context of HF [6,7,8,9,10,11,12,13,14,15,16,17]. Only one review [18] has suggested the benefits of using ³¹P MRS to aid the investigation of sarcopenia in HF patients. However, their rationale for using ³¹P MRS to diagnose sarcopenia was based on two studies, neither of which included HF patients. Instead, they investigated older people with sarcopenia [19] and the pre-frail elderly [20]. To date, no review has addressed systematically the assessment of SkM energetics in patients with both HF and sarcopenia. Furthermore, there may be important differences in SkM energetics between populations with heart failure with preserved (HFpEF) and reduced (HFrEF) ejection fraction, given their different pathophysiology [21].

The aim of this review is to quantitatively summarise the available literature on ³¹P MRS performed in the SkM of patients with HF (HFrEF and HFpEF) at rest and during post-exercise recovery compared to healthy controls. We hypothesised that PCr recovery time following exercise would be longer in patients with HF.

2. Methods

This systematic review was preregistered on PROSPERO (ID: CRD42023434698) and follows PRISMA 2020 reporting guidelines (Supplementary Material S1).

The primary outcome was: PCr recovery time in skeletal muscle.

The secondary outcomes included the following:

• SkM ratios (PCr/Pi and/or PCr/ATP) measured at rest (If applicable).
• The relationship between quantified SkM energetics and sarcopenia.
• The prevalence of sarcopenia in HFpEF and HFrEF.
• Disease-specific quality of life measures (either using the Kansas City Cardiomyopathy Questionnaire [KCCQ] or the Minnesota Living with Heart Failure questionnaire [MLWHF]).

2.1. Eligibility Criteria

Inclusion criteria were as follows: original cross-sectional and observational research studies 1. Involving adult patients with HF compared to healthy controls or 2. Reporting SkM energetic quantification (both PCr ratio and PCr recovery measure) using ³¹P MRS with the 3.

Papers were excluded if they were systematic reviews/meta-analyses, cohort studies, randomised controlled trials, non-randomised controlled trials, or non-systematic reviews. We also excluded studies that were not published in full peer-reviewed journals, studies published only as abstracts, letters, commentaries, opinion pieces and studies involving animals. The full exclusion criteria are in Supplementary Material S2.

2.2. Literature Search

The following electronic bibliographical databases were searched: Science Direct, Cinahl, Medline, Web of Science, Cochrane, Scopus, and Google Scholar. Studies were also found using citation searching. All sources were last searched on 29 September 2023. The search strategy followed the population, intervention, control, and outcome criteria and used Boolean operators. The Medline search strategy can be viewed in Supplementary Material S3. EndNote^TM20 bibliographic software was used to store retrieved results. Rayyan.ai was used to detect and remove duplicates [22]. One reviewer (S.A.S.) completed the first stage of abstract screening, and a second reviewer (J.B) independently screened a random sample of 10%. The second stage of full-text screening was completed by one reviewer (S.A.S.).

2.3. Data Extraction

Extracted data were retrieved and collated into a Microsoft Excel (Version 16.86) spreadsheet by one reviewer (S.A.S.). Participant characteristics, the study design, ³¹P MRS acquisition parameters, and quantified PCr values were extracted. The effect measure for all main outcomes was mean (±standard deviation). The list of outcomes and variables retrieved is in Supplementary Material S4.

2.4. Data Quality Assessment

Papers were assessed using the Joanna Briggs Critical Appraisal Tool (JBI tool, https://jbi.global/critical-appraisal-tools, accessed on 19 October 2023). The assessment was completed independently by two reviewers (J.B. and S.A.S.). Studies for which the answer was ‘yes’ to the pre-specified questions were scored as ‘high’, ‘no’ was scored as ‘low’, ‘unclear’ was scored as ‘some concerns’, and ‘not applicable’ was scored as ‘no information’. From this scoring system, a colour map was generated using Robvis (2023) [23]. The domains are specified in Supplementary Material S5. Publication bias was visualised using funnel plots. As less than 10 studies were included, the regression-based Egger’s test was used as a measure of statistically significant small-study effects.

2.5. Synthesis Methods

Studies were included in the meta-analysis if the mean and standard deviation or standard error were reported for our primary outcome (SkM energetics). If standard errors or raw data were reported, these were converted to standard deviations and included. In the instance that specific values were not in the paper, possible estimated means were calculated by a biomedical statistician (AP). The meta-analysis and forest plot diagrams were completed in RevMan Web, Version 7.12.0. As recovery times were expressed with different units across studies, PCr values and recovery times were reported as standardised mean differences (SMD), which allowed for the comparison of effect sizes. These values were pooled into a random effects model to assess mean differences. Heterogeneity was assessed using χ², Ƭ², and I².

3. Results

3.1. Study Selection

The search generated 618 articles, with 153 identified as duplicates, leaving 465 potentially eligible articles for title and abstract screening. The full-text screen showed that 11 articles initially met the inclusion criteria for the review (Figure 1). Of these 11 articles, 2 [24,25] were excluded from the meta-analysis due to missing standard deviations or missing data that did not allow for the calculation of standard deviations from the reported standard error. One article was also excluded as there was no control group for rest or recovery PCr [26]. Two studies were excluded as they did not have healthy control values reported following recovery [27,28]. One study required calculation to estimate the mean and SD of PCr time to identify the approximate values corresponding to data points in Figure 1 of their study [29]. A total of six papers were included in the meta-analyses.

3.2. Quality Assessment

The JBI critical appraisal tool identified that there were ‘some concerns’ (i.e., unclear as to risk of bias), as depicted in Figure 2. This was affected by the ambiguity within manuscripts in discussing strategies to deal with confounding factors in four included studies. Three studies did not report whether localisation sequences were used in their methodology, and three studies did not report TR times in their methodology, contributing to a moderate risk of bias. No study had a high risk of bias in any domain. A detailed rationale for each score is in Supplementary Material S6.

3.2.1. Participant Characteristics

The six studies selected for analysis included a total of 162 participants (

Table 1. Summary table of the study and subject characteristics.

Author and Year	Sample Size	HF Age	HC Age	HF patient Sex (M/F)	Control Sex (M/F)	NYHA Class (n)	EF (%)	HF Diagnosis/ EF Measure	HF Aetiology (n)	Comorbidities (n)	Medication (n)
Massie 1987 [29]	HF: 11 HC: 7	57 ± 7	56 ± 8	11/0	7/0	II: 5 III: 4 IV:2	16 ± 3	Congestive HF NR	NR DCM: 5	CAD: 6	NR
Rajagopalan 1988 [31]	HF: 22 HC: 33	58 ± 8	58 ± 8	22/0	33/0	NR	NR	Congestive HF NR	NR	NR	NR
Mancini 1988 [32]	HF: 20 HC: 9	47 ± 6	47 + 6	NR	9/0	II	20 ± 5	Congestive HF NR	CAD: 8 DCM: 12	NR	Digoxin:20 Diuretics:20
Chati 1996 [30]	HF: 14 HC: 7	55 ± 3	58 ± 7	12/2	NR	II: 7 III: 7	29 ± 2	Congestive HF LVEF <40%: radionuclide	Ischaemic heart disease: 8 DCM: 6	NR	Diuretics:14 Nitrates:14, ACEi: 10 Calcium antagonists: 6 Digoxin: 8
Hanada 2000 [34]	HF: 13 HC: 15	58 ± 8	49 ± 11	NR	NR	II: 8 III: 7	29 ± 13	Congestive HF Radionuclide	DCM: 13	NR	Diuretics:13, BB:8 ACEi:11 Digitalis:6
Menon 2021 [33]	HF: 6 HC: 5	56 ± 7	35 ± 7	4/2	5/0	II: NR III: NR	30 ± 15	Congestive HF NR	NR	NR	NR

Participant characteristics, sample size, and age. HF: heart failure; HC: healthy controls; NR: not rfeported; NYHA: New York Heart Association; EF: ejection fraction; LVEF: left ventricular ejection fraction; CAD: coronary artery disease; DCM: dilated cardiomyopathy; ACEi: ace-inhibitors; BB: beta-blockers. The mean and standard deviation have been reported for HF, as well as healthy control age and percentage ejection fraction.

). From the total sample, 86 (53%) were patients diagnosed with HF. In the four studies that reported sex, 49 (56.9%) participants were reported as males. Two studies reported the inclusion of four female HF patients (4.6%). In two studies, sex was not reported for N = 33 (38.4%). There were 76 (47%) control participants, of whom 54 (71%) were reported as male and remaining 22 (28.9%) from two studies did not have their sex reported. No study described the number of female healthy controls. Sample sizes ranged from 6 to 22 participants in the HF group and 5 to 33 in the healthy control group. The mean age was 55.1 ± 4.1 years in HF patients and 50.5 ± 8.9 years in healthy controls. Ethnicity was not reported in any of the studies. The following five studies specified the numbers of patients in each New York Heart Association (NYHA) category: 40 patients were class II, 18 were class III, and 2 were class IV. One study did not record patients’ ejection fractions; the reported mean ejection fraction among patients from the other six studies was 24.8%, which is consistent with HFrEF (LVEF < 40%). None of the studies specified the subtype of HF (i.e., HFpEF or HFrEF).

3.2.2. ³¹P MRS Methodological Characteristics

Two studies used the forearm muscles and four used the calf muscles as the region to be scanned.

Table 2. MRS methodological parameters and muscle assessments summarised for included studies.

Author and Year	Muscle Group	MR Field Strength (T)	Surface Coil Diameter	Localisation Method	Voxel Size	Flip Angle (°)	TR (msec)	Rest Duration	Exercise Duration (Minutes)	Recovery Duration (Minutes)	Time Points of MRS (n)	PCr Ratio Investigated	Strength Assessed
Massie 1987 [29]	Flexor digitorum superficialis muscle of dominant arm	1.89	2.5 cm	NR	NR	NR	NR	256 s	7.5	10	1 spectrum per 1–2 min exercise	[PCr]/([PCr] + [Pi])	Forearm resistance
Rajagopalan 1988 [31]	Flexor digitorum superficialis muscle of dominant arm	1.9	2.5 cm	NR	NR	NR	NR	4 min and 27 s	5	10	Spectra per 1 minute’s exercise and during recovery	[PCr]/([PCr] + [Pi])	Forearm resistance
Mancini 1988 [32]	Calf muscle groups	1.9	4.5 cm	Metabolic Freeze method	NR	NR	NR	10 min	3	5	RF pulse applied every 5 s	[PCr]/([PCr] + [Pi])	Plantar flexion
Chati 1996 [30]	Left calf muscle	2.4	5 cm	Unlocalised: FID Single pulse sequence and cyclops four-phase cycling	NA	71	1000	3 min	3	NR	Exercise spectra from last 25 s of each workload. final increment: 5 spectra for recovery phase.	[PCr]/([PCr] + [Pi])	Plantar flexion
Hanada 2000 [34]	Right calf muscle	1.5	80 mm	NR	NR	NR	1000	NR	6	6	40 scans averaged per spectrum	[PCr]/([PCr] + [Pi])	Calf muscle supine plantar flexion
Menon 2021 [33]	Gastrocnemius and soleus muscles	3	31P/1H quadrature volume knee coil with inner diameter of 22 cm	Unlocalised: FID	NA	NR	6000	1–2 min	1	Remainder of 10 min scan	100 FID measurements in 10 min	PCr/ATP [PCr]/([PCr] + [Pi])	Flexion

NR: not reported; PCr: phosphocreatine; ATP: adenosine triphosphate; FID: free induction decay; RF: radiofrequency, TR: repetition time. PCr recovery time calculations.

outlines the overall methodological characteristics of each study. Only the most recent paper was conducted at 3T. Two studies used unlocalised free induction decay sequences; therefore, it was not possible to report voxel sizes.

Three studies did not report localisation sequences. One study used a single pulse sequence with quadrature reception and cyclops four-phase cycling [30]. The duration of rest ranged between 1 and 10 min, and for exercise, this was between 1 and 7.5 min or until exhaustion. The allocated time frame for recovery ranged between 5 and 10 min. One study [30] did not specify the duration of recovery but reported that five spectra were acquired after exercise.

The methods and equations with which PCr recovery was calculated for each study are outlined in Supplementary Material S8. Of the studies reported in this meta-analysis, two utilised 50% of PCr recovery (T½) [29,31]. Another study used the time to baseline Pi/PCr ratio to express recovery time [32]; one study expressed this as PCr versus time [30]. Three studies used a mono-exponential recovery curve to calculate the recovery time [30,33,34]. As a derived quantity, the expert working consensus group has acknowledged the diverse approaches to PCr recovery kinetics, with mono-exponential function modelling being a commonly used approach [35].

3.3. SkM PCr Recovery Time

Compared to healthy controls, the recovery time in the HF patients was consistently prolonged across the studies, as exhibited in

Table 3. PCr quantification at rest and recovery in HF patients and in healthy controls.

Study	Controls				Heart Failure
	Resting		Recovery		Resting		Recovery
	PCr/Pi Mean ± SD	PCr/ATP Mean ± SD	PCr Index	Recovery Mean ± SD (s)	PCr/Pi Mean ± SD	PCr/ATP Mean ± SD	PCr Index	Recovery Mean ± SD (s)
Massie 1987 [29]	0.9 ± 0.01	2.7 ± 0.1	T ½ (s)	0.86 ± 0.4	0.87 ± 0.03	2.8 ± 0.4	T ½	1.72 ± 1.5
Rajagopalan 1988 [31]	0.9 ± 0.01	NR	T ½ (s)	42 ± 18	0.89 ± 0.02	NR	T ½	75 ± 52
Mancini 1988 [32]	0.21 ± 0.06	NR	Recovery time (s)	2.1 ± 0.5	0.21 ± 0.07	NR	Recovery Time	3.3 ± 0.8
Chati 1996 [30]	0.098 ± 0.01	NR	Recovery Rate (s)	0.27 ± 0.026	0.159 ± 0.03	NR	Recovery Time	0.273 ± 0.028
Hanada 2000 [34]	0.95 ± 0.02	NR	ô PCr (s)	36.5 ± 5.8	0.91 ± 0.05	NR	ô PCr	76.3 ± 30.2
Menon 2021 [33]	0.1 ± 0.02	7.93 ± 1.7	Recovery Rate (s)	26.73 ± 4.49	0.12 ± 0.03	8.43 ± 1.39	Recovery Time	50.1 ± 8.51

The PCr index for recovery has been specified per study. Abbreviations: PCr: phosphocreatine; NR: not reported.

. In four of the studies, this difference was of statistical significance. In the remaining two studies [29,30], the difference was not statistically significant. In studies reporting T½ recovery times, this was longer in HF patients compared to controls [29,31]. Figure 3 displays the meta-analysis, which shows this difference between HF and controls to be statistically significant (SMD: −1.35, CI: −2.11, −0.59, p = 0.0005). Mean recovery time was prolonged in the HF group than in healthy controls, and this was demonstrated across studies individually. There is considerable heterogeneity; the I² is 78%, and the Ƭ² is 0.64.

Figure 3 shows a forest plot of the PCr recovery time from the included studies.

3.4. Secondary Outcomes

3.4.1. Resting Relative Phosphocreatine

As seen in Table 3, the mean Pi/PCr reported in Figure 4 at rest between the HF and control groups across the studies was not statistically different (SMD: 0.05, CI: −0.81, 0.92, p= 0.90).

Figure 4 shows a forest plot indicating the mean PCr/Pi from the included studies.

Two studies reported the PCr/ATP relative values (Figure 5). The PCr/ATP between HF and healthy controls at rest was not significantly different (−0.11, CI: −0.35, 0.14, p = 0.39) in either study. The Ƭ² was 0.00, the χ² was 0.17, and the I² was 0%, indicating little heterogeneity between the two studies based on a small sample size in each study.

Figure 5 shows a forest plot indicating the mean PCr/ATP from the included studies.

3.4.2. Sarcopenia-Related Outcomes, HF Subtyping, and Symptoms

None of the included studies investigated or screened for sarcopenia. Therefore, it was not possible to conduct an analysis to investigate our specified secondary outcomes. Additionally, as the included studies did not differentiate HFpEF and HFrEF subgroups, it was not possible to analyse the prevalence of sarcopenia between HFrEF and HFpEF cohorts. Finally, none of the studies utilised the KCCQ or MLWHF questionnaire; therefore, it was not possible to determine the quality of life in HF patients or the association with SkM eneregetics.

3.5. Publication Bias

The funnel plots are available in Supplementary Material S7. At rest, the Egger’s test showed a B1 (effect size captured with the slope of the regression line) with an estimated slope of −1.09 with a standard error of 4.99, giving a test statistic of z = −0.22 and a p-value of 0.83. This indicated no statistically significant evidence of small-study effects at rest; hence, there was a lack of evidence to indicate publication bias. For recovery, there was some evidence of small-study effects. B1 was reported as 6.17 with a standard error of 2.86, giving a test statistic of z = 2.16 and a p-value of 0.0310.

4. Discussion

In this systematic review and meta-analysis, we identified only six studies incorporating 86 patients that compared SkM energetics in HF to healthy controls that also reported PCr recovery time. Individually, studies consistently indicated a longer recovery time in HF patients. The pooled SMD was −1.35, indicating an overall large effect size. However, the magnitude of this effect varied per study, reflecting variation in the extent of delayed recovery between HF and healthy control groups. A small effect size was documented in one study (−0.11 [95% CI −1.01,0.80] [30]), and a moderate effect size was documented in one study (−0.68 [95% CI −1.66,0.30] [29]). Both studies had overlapping confidence intervals, indicating a lack of statistical significance. The remaining four studies had larger magnitude of effect sizes, suggesting a more pronounced delay in PCr recovery time.

This prolonged PCr recovery in HF patients is indicative of dysfunctional oxidative phosphorylation. Investigating PCr recovery following moderate exercise is a robust strategy, as PCr recovery is fuelled by oxidative ATP synthesis, and its kinetics reflect muscle ‘mitochondrial capacity’. Collectively, among 86 patients with HF and 76 control subjects, there was no statistically significant difference in resting SkM energetics between HF and control groups.

Two studies reported the PCr/ATP ratio at rest in HF patients and controls. Under ‘normal’ conditions, the mitochondrial sarcoplasmic reticulum is the main producer of ATP, which gives energy for fibre contraction and relaxation [35]. Previously, in healthy adults, resting PCr/ATP has been reported as 4.23 ± 0.24 in calf muscle and 4.48 ± 0.20 in thigh muscle [35]. This differs from the reported findings in our meta-analysis, as resting PCr/ATP was reported to be higher in the calf muscle and lower in the forearm muscle [29,33]. However, one study focused on the forearm muscle [29], and the other study looked at the gastrocnemius muscle in the calf [33], and no study reported resting values for the thigh muscles. Despite this, ATP, when used as an internal concentration reference standard, should be relatively constant between individuals and is relatively constant among fibre types [35]. All studies reported relative PCr/Pi ratios rather than absolute quantification. Consideration of the many confounding factors makes absolute quantification technically demanding and difficult to reproduce. Therefore, this collective strength meant that relative values were likely to be less sensitive to confounders.

4.1. Study Differences and Limitations

The JBI tool has identified some concerns, meaning that interpretation of the results requires some caution. Specifically, the tool highlighted issues related to the identification and handling of confounding factors and highlighting the specific inclusion criteria for participants. Collectively, the six studies had a paucity of female participation, which may have reduced heterogeneity. Additionally, ethnicity was not considered in any of the studies. Differences were evident across studies in the categorisation and classification of HF. The most frequently reported classification was based on the severity of symptoms using the NYHA class. Initially, we sought to compare findings between HFpEF and HFrEF participants in our meta-analysis. None of the participants in our study could be classified as HFpEF. Only one small study, which was excluded from our meta-analysis as the authors did not present the mean and standard deviation of data, examined SkM energetics in HFpEF compared to HFrEF (N = 20 HFrEF and N = 12 HFpEF, mean ejection fraction 27 ± 11 and 62 ± 5% respectively) and healthy controls (N = 11) [25]. They reported a significantly longer PCr recovery time in both HF groups compared to healthy controls. Furthermore, patients with HFpEF had the largest energetic changes of all the groups studied, such as initial PCr depletion. Only one study included NYHA class IV HF patients [29]. The inclusion of class IV may provide a better insight into metabolic deficits at the end stages of the disease [33]. Two studies assessed energetics in the forearm. Although the flexor digitorum superficialis was considered ‘relatively unimportant’ [29], it was argued in a separate study [31] that this forearm muscle would be in use despite how incapacitated patients were and that focussing on this group was therefore of clinical relevance. Other studies investigated the calf muscle [28,30,32,33,34]. This heterogeneity of muscle recruitment needs to be considered in the interpretation of exercise-induced metabolic changes. In an effort to account for this, one study used various muscle groups to measure PCr recovery rates simultaneously [33].

Differences in methodology may have impacted the validity and interpretation of findings. Variations in scanning regions, time points and field strength are likely to have contributed to result variations. Localisation has been advised in the expert consensus to avoid pitfalls in acquiring kinetics results [35]. Localisation techniques varied across studies, and three studies did not report localisation methods. Two studies utilised the smallest radio frequency (RF) coil (2.5 cm) [29,31], thereby generating spectra from a much smaller volume than a study that used coils of a larger diameter [27,28,34]. However, this was appropriate, considering the smaller size of the forearm muscles in comparison to the calf muscle groups. Additionally, it has been advised by the expert consensus to use localisation to avoid pitfalls in scanning methodology and to improve the linewidth by homogenising the magnetic field (shimming).

Each study reported the radiofrequency surface coil diameter as their initial form of localisation to the target muscle. Although three studies did not mention if they used localisation sequences, as the field strengths used ranged between 1.5 and 3 Tesla, relying on a surface coil for localisation at low field strength can therefore be considered sufficient [29,31,34].

4.2. Disparities among Heart Failure Populations

Normal ageing induces a shift in SkM metabolism in the form of compromised SkM mass and quality (i.e., sarcopenia). The elderly (aged > 65 years) represent 80% of all patients with HF [8]. Yet despite this, the overall mean of the HF group was 55 years. The combination of comorbidities combined with the effects of ageing leads to frailty, muscle atrophy, and decreased functional capacity, making the diagnosis and staging of the effects of HF in the elderly population challenging.

A confounding factor discussed as a limitation across studies [29,32] was muscle atrophy and smaller SkM mass in patients with HF. Muscle atrophy may lead to a reduction in the total number of mitochondria [32]. Rajagopalan et al. reported that patients with HF had smaller forearms and smaller lean body mass [31]. Similarly, Mancini et al. reported that calf muscle dimensions and body weight were lower in HF patients [32]. If both HF and control groups performed similar levels of work, a lower muscle mass in HF patients may require more work from each muscle fibre compared to normal participants, consequently leading to a higher PCr/Pi ratio. Despite this, Mancini et al. [32] did not account for potential differences in muscle bulk or the effect of inactivity on muscle metabolism.

4.3. Strengths and Limitations

This systematic review is a summary of currently available cross-sectional studies of ³¹P MRS investigating the SkM of patients with HF compared to healthy controls. We conducted an extensive and comprehensive reproducible literature search using different electronic databases and assessed which participants were included in the eligible studies. Two reviewers worked on independently appraising the data using the JBI critical appraisal tool. We included studies including participants who had diagnoses of HF using ³¹P MRS only and no other forms of MRS, such as near-infrared and H+ spectroscopy.

The quality of evidence identified by the JBI critical appraisal identified some concerns about how the studies addressed potential bias in their design and analysis, indicating that results from our meta-analysis should be interpreted with caution. Limitations include dated research and methodological issues. The studies included in our analysis did not differentiate HFpEF and HFrEF; thus, we could not investigate potential differences. The I² (PCr recovery time) was 76%, demonstrating considerable heterogeneity, which may in part be explained by the inclusion of relatively small studies and the fact that three studies recruited male participants only. The I² value of 83% at rest may potentially be due to the variety of muscle groups assessed in the meta-analyses. A limitation regarding study design is differences in age between populations and a lack of comorbidities being reported in the HF population across six studies. Menon et al. reported that their HF cohort was significantly older than their control cohort, with a mean age difference of 20 years [33]. The studies varied in classes of NYHA and measures of ejection fraction. Additionally, the review has considered mainly, if not entirely, HFrEF participants, and existing studies with HFpEF patients did not meet our inclusion criteria. Moreover, comorbidities and muscle weakness were seldom reported in most of the included studies.

4.4. Implications for Future Research

Future studies should explore SkM energetics in HF and their relationship with sarcopenia. Technical challenges and diversity within HF groups limit the broad applicability of ³¹P MRS in HF studies, highlighting the need to standardise existing ³¹P MRS methodologies. Understanding SkM energetics has the potential to provide clinicians with a useful tool allowing for the assessment of the effect of therapeutic outcomes on skeletal muscle in HF. Sarcopenia and HF frequently coexist and are intimately related in older adults and should be investigated together. The loss of muscle mass that occurs with sarcopenia decreases physical function and exacerbates HF symptoms. Sarcopenia is associated with an increased risk of hospitalisation and mortality in HF patients; therefore, identifying compromised SkM metabolisms is crucial for improving outcomes.

5. Conclusions

³¹P MRS has great potential in HF for investigating SkM quality and monitoring the status of peripheral abnormalities. Dynamic exercise recovery studies allow for a better understanding of in vivo kinetic control of oxidative phosphorylation in SkM [35]. The benefit of ³¹P MRS is the ability to assess the quality of muscle at various states non-invasively, unlike other methods, such as biopsies. This systematic review highlights that SkM PCr recovery time is significantly impaired in patients with HFrEF using six studies with a total of 162 included participants. Only minimal data in females exist, no HFpEF studies were identified, and no studies were identified linking abnormal SkM energetics directly to sarcopenia. We have identified that there is a critical gap in performing ³¹P MRS assessments in patients suffering from HF and sarcopenia as a multimorbidity. These are key areas for future studies.