Phosphorus Magnetic Resonance Spectroscopy (31P MRS) and Cardiovascular Disease: The Importance of Energy

Background and Objectives: The heart is the organ with the highest metabolic demand in the body, and it relies on high ATP turnover and efficient energy substrate utilisation in order to function normally. The derangement of myocardial energetics may lead to abnormalities in cardiac metabolism, which herald the symptoms of heart failure (HF). In addition, phosphorus magnetic resonance spectroscopy (31P MRS) is the only available non-invasive method that allows clinicians and researchers to evaluate the myocardial metabolic state in vivo. This review summarises the importance of myocardial energetics and provides a systematic review of all the available research studies utilising 31P MRS to evaluate patients with a range of cardiac pathologies. Materials and Methods: We have performed a systematic review of all available studies that used 31P MRS for the investigation of myocardial energetics in cardiovascular disease. Results: A systematic search of the Medline database, the Cochrane library, and Web of Science yielded 1092 results, out of which 62 studies were included in the systematic review. The 31P MRS has been used in numerous studies and has demonstrated that impaired myocardial energetics is often the beginning of pathological processes in several cardiac pathologies. Conclusions: The 31P MRS has become a valuable tool in the understanding of myocardial metabolic changes and their impact on the diagnosis, risk stratification, and prognosis of patients with cardiovascular diseases.


Introduction
Myocardial energetics represent one of the most important biochemical processes in the human body. A heart that functions well also has a normal cardiac metabolic profile. Derangement of this metabolic profile heralds the pathophysiological processes that subsequently lead to heart failure symptoms. This "engine out of fuel," as the failing heart has previously been described [1], needs to be diagnosed appropriately and in a timely fashion, but even more importantly, the mechanism behind this failure needs to be understood.
The correlation of biochemical pathways with metabolic function and energetic state in active tissues is a challenging project that researchers have acknowledged since the late 1970s [2]. At the time, the very first studies using 31-phosphorus magnetic resonance spectroscopy ( 31 P MRS) began. These were ex vivo studies that, given the limitations of spatial localization techniques, focused on the energetics of animal hearts [2][3][4][5]. However, with the advancement of surface coils and improvements in spatial localization techniques,

Study
Year Participants Major Findings

Myocardial Infarction (MI) and Ischaemic Cardiomyopathy (ICM)
Weiss et al. [8] 1990 11 healthy controls 16 patients with CAD, of which 5 patients were re-examined after successful revascularisation 9 patients with non-ischaemic heart disease 31 P NMR spectra recorded from the anterior myocardium before, during and after isometric hand-grip exercise • PCr/ATP was significantly reduced in the HCM subjects relative to controls (by 30%) • The reduction was of a similar magnitude in all three disease-gene groups • PCr/ATP ratio was significantly reduced in HHD • PCr/ATP ratio was significantly lower in patients with systolic dysfunction compared to those without • PCr/ATP ratio was correlated linearly with LVEF • PCr/ATP ratio was significantly reduced in CAV • PCr/ATP value of 1.59 was the optimal cut-off value to predict CAV (specificity and sensitivity of 100% and 72%, respectively).

Myocardial Energetics
The heart has been described as the organ with the highest metabolic demand, consuming 8% of the total body adenosine triphosphate (ATP) [67,68]. Considering that the heart only accounts for~0.5% of the total body weight, it is estimated that it uses 20 to 30 times its weight in ATP to maintain its metabolic homeostasis [68,69]. The cardiac metabolic phenotype is not only highly demanding but also quite dynamic, as its function depends on the continuous production of ATP. If this copious and uninterrupted production of ATP is not maintained, then the heart would be deprived of energy within less than 10 s [67,70].
The healthy adult heart primarily uses free fatty acids and glucose as its main energy substrates [71][72][73], while the rest of the energy requirements are fulfilled by alternative energy substrates such as lactate, ketones, and amino acids [67,70]. The alterations in cardiac energy demands and availability of the substrates lead to shifts in energy substrate utilisation, a feature described as "metabolic flexibility" [70,71]. This reflects the ability of the heart to switch to different energy substrates in order to maintain its requirements [70,71].
In order to meet these high metabolic demands, the adult heart uses the above energy substrates to generate ample amounts of ATP, mainly through mitochondrial oxidative phosphorylation, which makes up to 95% of myocardial ATP, while a smaller amount is generated through glycolysis [70,71]. The mitochondrial oxidative phosphorylation has a key role in the generation, transfer, and utilisation of ATP by the myofibrils of the cardiomyocytes. In addition, through the creatine kinase (CK) "shuttle" system, highenergy phosphate is transferred from the ATP produced in the mitochondria to creatine (Cr), generating in this way phosphocreatine (PCr) and adenosine diphosphate (ADP). The PCr, a much smaller molecule than ATP, is then rapidly diffused from the mitochondria to the sarcomeres that comprise the myofibrils. In the myofibrils, CK catalyses the formation of ATP and Cr. The ATP is then used by the ATPases and contributes to cell contraction, while Cr is diffused back into the mitochondria [1,74] (Figure 2). substrates to generate ample amounts of ATP, mainly through mitochondrial oxidative phosphorylation, which makes up to 95% of myocardial ATP, while a smaller amount is generated through glycolysis [70,71]. The mitochondrial oxidative phosphorylation has a key role in the generation, transfer, and utilisation of ATP by the myofibrils of the cardiomyocytes. In addition, through the creatine kinase (CK) "shuttle" system, high-energy phosphate is transferred from the ATP produced in the mitochondria to creatine (Cr), generating in this way phosphocreatine (PCr) and adenosine diphosphate (ADP). The PCr, a much smaller molecule than ATP, is then rapidly diffused from the mitochondria to the sarcomeres that comprise the myofibrils. In the myofibrils, CK catalyses the formation of ATP and Cr. The ATP is then used by the ATPases and contributes to cell contraction, while Cr is diffused back into the mitochondria [1,74] (Figure 2). The PCr is the primary energy reserve metabolite in the heart, with its concentration being twice that of ATP phosphorylation [76]. However, through the CK-mediated reaction, ATP production with the use of PCr is ~10 times faster than the rate of ATP The PCr is the primary energy reserve metabolite in the heart, with its concentration being twice that of ATP phosphorylation [76]. However, through the CK-mediated reaction, ATP production with the use of PCr is~10 times faster than the rate of ATP production by oxidative phosphorylation [77]. In situations of high metabolic demand, when the rate of ATP use exceeds that of its production, the PCr/CK system therefore works as a buffer to maintain homeostasis, as the use of PCr via the CK reaction/catalysis helps in maintaining ATP at a stable normal level of phosphorylation [1,76]. Indeed, in heart failure-regardless of the aetiology-the PCr level falls in order to help maintain the ATP at a normal level [1]. While this may seem like an appropriate compensatory mechanism, it is certainly not sufficient to maintain the high metabolic demand of the cardiomyocytes. In a disease state, the failing heart is much less energy efficient, as mitochondrial dysfunction and shifts in energy substrate utilisation lead to a significant reduction in ATP production [73,78]. Additionally, PCr has been shown to decrease in pathological cardiac hypertrophy and failure as a result of the significant ATP supply-demand mismatch and partly due to a loss of creatine levels [76]. This is followed by a progressive ATP decrease [74,77]. It is therefore clear that a change in the PCr or ATP levels, or the ratio PCr/ATP, signifies a metabolic derangement that proclaims the failure of the heart. In the failing myocardium, the decrease in creatine levels occurs earlier and is faster than the reduction in ATP levels [77,79,80]. The ATP level is~30% lower than in the normal myocardium, with the respective reduction of creatine levels being up to 50-70% in severely failing myocardium [77,80]. As creatine, PCr, and ATP levels all reduce significantly, it is likely that the PCr/ATP ratio underestimates the severity of the PCr decrease [77,79]. In addition, although previous studies have shown that the PCr/ATP ratio in the healthy myocardium is maintained, there is evidence to suggest that stress may lead to its reduction even in the absence of any underlying cardiac pathology [33,81].

Assessment of Myocardial Energetics with 31 P MRS
The 31 P MRS is a non-invasive technique that provides unique insights into the abovementioned intracellular parameters through the detection of phosphorus-containing metabolites. It therefore allows an accurate evaluation of the myocardial energetic state through measurement of the myocardial PCr/ATP ratio as well as absolute levels of high-energy phosphates [82,83]. On most magnetic resonance imaging (MRI) scanners, specialised scanner software and additional hardware, including a broadband amplifier, receive system, and radiofrequency (RF) coils, are needed to permit the acquisition of 31 P MRS data [84]. In addition, for optimised data collection and correct metabolite quantification, the RF excitation pulse bandwidth must be large enough to excite all the relevant metabolites homogeneously [84].
The excitation of 31 P nuclei with RF signals is required for 31 P MRS. However, in following the excitation of the nuclei, the induced signal is recorded in the receiver coils of the scanner after the RF energy is switched off [85]. The amplitude of the received signal reflects the number of nuclei present in the interrogated tissue [85]. The resonance frequency of the nuclei being interrogated depends on their chemical environment; a phenomenon described as 'chemical shift' [75,85]. This is a result of different nuclei being surrounded by a different number and unique spatial location of the surrounding electrons (i.e., different shielding environments) [75,85]. Further, although the resonance frequency depends on the strength of the magnetic field, the chemical shift, as quantified in parts per million (ppm), does not alter.
The typical 31 P cardiac spectrum demonstrates peaks for phosphocreatine and the three phosphorus nuclei of ATP (γ-ATP, α-ATP, and β-ATP). The area under the peak is proportional to the relative concentrations of these metabolites [1] (Figure 3). The cardiac 31 P MRS is typically performed on MR scanners with a field strength of 1.5 or 3 Tesla(T), although studies using a higher field strength of 7T have been reported [86,87]. As the field strength increases, the frequencies, as reflected by the spectral peaks, are more separated and therefore better quantified [75,87] (Figure 4). The cardiac 31 P MRS is typically performed on MR scanners with a field strength of 1.5 or 3 Tesla(T), although studies using a higher field strength of 7T have been reported [86,87]. As the field strength increases, the frequencies, as reflected by the spectral peaks, are more separated and therefore better quantified [75,87] (Figure 4).
A cardiac 31 P spectrum can be obtained using a variety of techniques that help in the identification of signals from a region of interest. In brief, these include the following: (a) Depth-resolved surface coil spectroscopy (DRESS): a single slice parallel to the coil is selectively excited by an MRI gradient. (b) Image selected in-vivo spectroscopy (ISIS): data are acquired using data from inversion pulses from consecutive cycles, and voxel localisation is achieved using a slice-based intersection strategy. (c) One-, two-and threedimensional (1D, 2D, and 3D) chemical shift imaging (CSI): this technique uses phase encoding for spatial localisation, and the spectra can be acquired in a column of voxels (1D), a plane (2D), or a block (3D) of voxels [75,88,89]. The CSI has seen more widespread use in recent years, and is often applied with saturation bands to minimise 31 P contributions from skeletal muscle and the liver [75,89]. sity of East Anglia. 2,3-DPG, 2,3-Diphosphoglycerate; PDE, phosphodiesters; PCr, phosphocreatine; ATP, adenosine triphosphate The cardiac 31 P MRS is typically performed on MR scanners with a field strength of 1.5 or 3 Tesla(T), although studies using a higher field strength of 7T have been reported [86,87]. As the field strength increases, the frequencies, as reflected by the spectral peaks, are more separated and therefore better quantified [75,87] (Figure 4).  31 Phosphorus spectra acquired from the same individual at 3 and 7 Tesla field strengths real part of the spectrum from a voxel in the middle of the interventricular septum. The spectra were apodized with an exponential filter (matched to the fitted PCr linewidth), first order phase corrected, and normalized to the resulting baseline noise standard deviation. Higher magnetic field yields higher signal-to-noise ratio (SNR). The SNR is markedly higher for the  31 Phosphorus spectra acquired from the same individual at 3 and 7 Tesla field strengths real part of the spectrum from a voxel in the middle of the interventricular septum. The spectra were apodized with an exponential filter (matched to the fitted PCr linewidth), first order phase corrected, and normalized to the resulting baseline noise standard deviation. Higher magnetic field yields higher signal-to-noise ratio (SNR). The SNR is markedly higher for the 7T data. Note that the y-axes are offset for clarity. Reproduced with permission from Rodgers et al. [87] under a Creative Commons Attribution 4.0 International License. SD: standard deviation; 2,3-DPG, 2,3-Diphosphoglycerate; PDE, phosphodiesters; PCr, phosphocreatine; ATP, adenosine triphosphate; ppm, parts per million.
Using one or a combination of the aforementioned techniques, a localised cardiac 31 P spectrum can be acquired. However, through analysis of this spectrum, the PCr/ATP ratio can be determined, allowing in this way an evaluation of the cardiac energetic state. In addition, the cellular pH (from the chemical shift of inorganic phosphate (Pi) relative to PCr) can be indirectly deduced from the acquired spectra [20,85,90]. Apart from absolute measurements, cardiac 31 P MRS can also provide a unique insight into dynamic changes in the cellular metabolism through assessment of the CK flux and the rate of ATP generation [18,24,30]. Furthermore, by the measurement of the CK flux, the pseudo-firstorder unidirectional rate constant (k f ) of CK in the ATP-generating (forward) direction is first measured. This is then multiplied by the PCr/ATP ratio to give the forward CK flux, i.e., the ATP delivery rate [24,57].
The major limitation of 31 P MRS is its intrinsically low SNR (approximately 10 5 -fold lower than 1 H-MRS), which is typically compensated via coarser spatial and temporal resolutions [85,87,89]. This is primarily because of the low concentrations of the metabolites being studied compared to water [85,87,89]. This drawback is reflected in longer acquisition times that may be up to approximately 30 min [87,89]. Nevertheless, the disadvantage of the low SNR can be mitigated with the use of higher stronger magnetic fields, such as 7T, if available [31,87].

The Role of 31 P MRS in the Evaluation of Cardiovascular Diseases
Numerous studies have shown that, in a cardiovascular disease state, there is significant impairment of energy production mediated by mitochondrial dysfunction, alterations in energy substrate utilisation, and impaired ATP transfer and utilisation [1,68]. The 31 P MRS is the only technique available with the ability to study this pathological alteration in cellular metabolism in vivo, which is reflected in changes in cellular ATP and PCr concentrations, the PCr/ATP ratio, as well as the rate of ATP production as reflected by the CK flux. The research community has provided promising evidence that the role of 31 P MRS is crucial in the early detection of pathological changes of the myocardial energetics through an exponential growth of clinical research studies over the last four decades.

Myocardial Infarction (MI) and Ischaemic Cardiomyopathy (ICM)
The cardiac energy metabolism appears to be significantly impaired both in the infarcted and in the ischaemic myocardium. Following myocardial infarction, the concentrations of both PCr and ATP metabolites fall significantly [13,18]. In addition, the CK ATP supply, reflected by the CK flux, is also significantly reduced, which is likely attributable to myocyte loss [18]. The PCr/ATP ratio is very much dependent on the presence or absence of myocardial ischaemia and heart failure secondary to ischaemic cardiomyopathy. In a study that included 27 patients with severe left anterior descending artery (LAD) disease, Yabe et al. found that the PCr/ATP ratio was significantly reduced in those with reversible ischaemia compared to those with fixed defects or healthy volunteers [12]. Weiss et al. also demonstrated that the PCr/ATP ratio significantly decreased during isometric hand-grip exercise in patients with significant coronary artery disease [8]. Interestingly, the PCr/ATP ratio was not reduced in the patients that underwent successful revascularisation, suggesting the normalisation of the metabolic parameters after a timely successful clinical intervention and normalisation of the blood supply [8]. This is in keeping with another study in which 15 patients underwent 31 P MRS 3 weeks post MI and the PCr/ATP ratio was found to be normal in viable myocardium [15].
There is only one study that investigated the potential changes in myocardial energetics in patients with normal epicardial coronary arteries. Buchthal et al. recruited 35 female patients who had been admitted to the hospital with cardiac chest pain and had normal invasive coronary angiograms. A total of Seven of the 35 women (20%) had significantly reduced cardiac PCr/ATP ratios during handgrip exercise, suggesting potentially significant abnormal myocardial metabolism in this population [16].

Dilated Cardiomyopathy (DCM)
The DCM exhibits impaired myocardial energetics, with reduced PCr and ATP concentrations, and PCr/ATP ratios [10,[21][22][23]25,26,33]. The PCr/ATP ratio has attracted a lot of research interest in this population. However, with both the PCr and ATP metabolites being significantly reduced simultaneously, it has been noted that a seemingly mild reduction in the PCr/ATP ratio may underestimate the true impairment of myocardial metabolism [23]. Despite this weakness, it has been shown to have potentially strong implications for risk stratification and prognosis. More specifically, there is evidence suggesting that the PCr/ATP ratio has a strong correlation with the clinical severity of heart failure, as estimated by the New York Heart Association (NYHA) class, as well as the left ventricular ejection fraction (LVEF) [21,23,25]. In a study that included 39 patients with DCM, Neubauer et al. demonstrated that the PCr/ATP ratio is a significant predictor of cardiovascular mortality [53]. The CK flux is also not only markedly reduced in DCM, but it also carries important prognostic implications. In a study that included 58 patients, Bottomley et al. found that reduced myocardial CK flux was a significant predictor of all-cause mortality and HF outcomes, even after correction for NYHA class, LVEF, and race [27].

Hypertrophic Cardiomyopathy (HCM)
The PCr/ATP ratio is significantly reduced in patients with HCM, highlighting the abnormal myocardial metabolic changes in this population too. Early studies from almost three decades ago demonstrated unanimously that HCM is associated with impaired my-ocardial energetics [20,34,37,90]. Consequently, following those early studies, research focused on specific mutations associated with the disease. In a study that included 31 patients with HCM positive for 3 mutations (beta-myosin heavy chain, cardiac troponin T, and myosin-binding protein C), Grilley et al. found that the PCr/ATP ratio was significantly reduced, and the reduction was of similar magnitude in all three disease-gene groups [38]. Another study that included 9 patients with a point mutation (Arg403Gln) in the betamyosin heavy chain gene found significantly reduced PCr concentrations and CK flux while the PCr/ATP ratio showed a trend towards significance [41]. Furthermore, during exercise there is a pathological reduction of the PCr/ATP ratio in patients with HCM, a finding that may explain the exercise-related diastolic dysfunction in these patients [42] ( Figure 5). In addition to the above, the PCr/ATP ratio has been shown to have a significant correlation with other imaging parameters derived from cardiac magnetic resonance (CMR) imaging, including T1 values and late gadolinium enhancement [28,40]. Despite the complex and heterogeneous cohort of patients that comprise the clinical entity of HFpEF, the myocardial energetics display a uniform impairment, reflected by a significant reduction in the PCr/ATP ratio [47,50]. In addition, there is evidence suggesting that this ratio is significantly correlated with the log N-terminal pro b-type natriuretic peptide (NT-proBNP), the echocardiographic e/E', the NYHA class, and the exercise induced pulmonary congestion that occurs in these patients [50]. Significantly reduced PCr/ATP ratios are also present in diabetic cardiomyopathy [48,51] and in hypertensive heart disease associated with systolic dysfunction [26,45]. Similarly, obesity is associated with impairment of myocardial energetics with a reduced PCr/ATP ratio at rest and failure of CK flux to increase during increased workload [49]. Weight loss appears to ameliorate the dysfunction of the cellular metabolism as it leads to an increase in the PCr/ATP ratio as an well as increase in CK flux during increased workload [49].

Valvular Cardiomyopathy
Both aortic and mitral valve disease have been shown to have a significant impact on the myocardium [91]. This is depicted in the cardiac metabolic profile of these patients as Despite the complex and heterogeneous cohort of patients that comprise the clinical entity of HFpEF, the myocardial energetics display a uniform impairment, reflected by a significant reduction in the PCr/ATP ratio [47,50]. In addition, there is evidence suggesting that this ratio is significantly correlated with the log N-terminal pro b-type natriuretic peptide (NT-proBNP), the echocardiographic e/E', the NYHA class, and the exercise induced pulmonary congestion that occurs in these patients [50]. Significantly reduced PCr/ATP ratios are also present in diabetic cardiomyopathy [48,51] and in hypertensive heart disease associated with systolic dysfunction [26,45]. Similarly, obesity is associated with impairment of myocardial energetics with a reduced PCr/ATP ratio at rest and failure of CK flux to increase during increased workload [49]. Weight loss appears to ameliorate the dysfunction of the cellular metabolism as it leads to an increase in the PCr/ATP ratio as an well as increase in CK flux during increased workload [49].

Valvular Cardiomyopathy
Both aortic and mitral valve disease have been shown to have a significant impact on the myocardium [91]. This is depicted in the cardiac metabolic profile of these patients as well. More specifically, PCr/ATP is significantly decreased in symptomatic patients with aortic valve disease, and it improves post aortic valve replacement [23,52,53,55,56]. Notably, in a study that included 65 patients with aortic stenosis, Peterzan et al. found that the PCr/ATP ratio and CK flux were significantly reduced both in patients with moderate aortic stenosis, as well as those with severe aortic stenosis and normal left ventricular ejection fraction, suggesting in this way that pressure loading conditions are associated with deranged myocardial energetics even before the disease progresses to the severe stage or the presentation of symptoms [57]. In patients with mitral regurgitation, myocardial energetics are significantly impaired in symptomatic patients and in those with severe disease, while the PCr/ATP ratio is correlated with left ventricular dilatation [54].

Heart Transplantation
The 31 P MRS has also revealed valuable insights in the pathophysiology of cardiac allografts and is a promising non-invasive tool that can efficiently assess cardiac allograft vasculopathy (CAV). In a study that included 25 heart transplant recipients, Evanochko et al. used a 31 P MRS stress test and found that certain patients have abnormal cardiac energetics after transplantation, as reflected by a significant change in the PCr/ATP ratio [58]. This appeared to be unrelated to the timing of the transplantation and was present in patients with normal coronary arteries as well. The authors highlight that the 31 P MRS stress test may have an important role in the diagnosis of CAV as it is a much more sensitive means of assessing the microvasculature [58]. Similarly, Caus et al. suggest that 31 P MRS is a promising tool able to detect impairment of myocardial energetics related to CAV, with a PCr/ATP value of 1.59 being their proposed optimal cut-off value to predict CAV with specificity and sensitivity of 100% and 72%, respectively [59]. It has to be noted, however, that 31 P MRS is not standard as yet; hence, such cut-offs only apply to the particular methodology and the specific MRI scanner used in this study.

Cardiovascular Research and Future Directions
As noted above, the metabolism has a central role in the pathogenesis of a range of cardiovascular diseases. As such, medical therapies targeted at the metabolic pathways that have a key role in these diseases are extremely beneficial for the patients. This has been shown for several medications that have been proven to be not only of symptomatic but also of prognostic benefit. For example, beta blockers have been shown to increase the PCr/ATP ratio by 33% in patients with heart failure after only 3 months of treatment [63], and, similarly, perhexiline has also been shown to improve the PCr/ATP ratio by 30% and the left ventricular systolic function [64]. More recently, sodium-glucose co-transporter-2 inhibitors (SGLT2i) have shown an impact on mortality and morbidity in patients with heart failure [92]. As research focuses on their mechanism of action, there is evidence suggesting that they alter myocardial energetics and improve cellular metabolism [93], as recent evidence reveals their positive impact on the PCr/ATP ratio [51].
The cardiovascular research has now shifted to treatments that focus on the improvement of myocardial energetics and metabolism, which play a pivotal role in the pathogenesis of several cardiovascular diseases. The 31 P MRS will continue to have an instrumental part in this journey as it provides a non-invasive and accurate evaluation of the cardiac metabolic profile that helps researchers gain important insights into the complex pathophysiology of cardiac disorders. The current limiting factors in the widespread use of 31 P MRS include the requirement of both expensive hardware and clinical expertise. Nevertheless, as research emphasises the importance of myocardial energy reserve in the pathophysiology of disease and pharmacotherapy, these issues will gradually recede.

Conclusions
The myocardial energetic compromise has proved to be an important feature in the pathophysiological process of several conditions. The 31 P MRS is the only available noninvasive technique with the capacity to quantify metabolism in vivo, and it is, therefore, an excellent tool in the evaluation of the myocardial metabolic profile of patients with cardiac pathologies. Future studies assessing myocardial energetic phenotypes and how these are associated with clinical outcomes and prognosis will help further in understanding the impact of altered metabolism in clinical practice.
Author Contributions: V.T. executed the systematic review and wrote the first draft. D.C. and R.S. significantly amended the manuscript. G.B. executed a systematic review and significantly amended the manuscript. V.S.V. planned and designed the study, supervised the systematic review, and significantly amended the manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Ethical review and approval were waived for this study as this is a meta-analysis of already published data.
Informed Consent Statement: Patient consent was waived as this is a meta-analysis of already published data.