Intracellular Phosphate and ATP Depletion Measured by Magnetic Resonance Spectroscopy in Patients Receiving Maintenance Hemodialysis

Abstract

Visual Abstract

Background 

The precise origin of phosphate that is removed during hemodialysis remains unclear; only a minority comes from the extracellular space. One possibility is that the remaining phosphate originates from the intracellular compartment, but there have been no available data from direct assessment of intracellular phosphate in patients undergoing hemodialysis.

Methods 

We used phosphorus magnetic resonance spectroscopy to quantify intracellular inorganic phosphate (Pi), phosphocreatine (PCr), and βATP. In our pilot, single-center, prospective study, 11 patients with ESKD underwent phosphorus (³¹P) magnetic resonance spectroscopy examination during a 4-hour hemodialysis treatment. Spectra were acquired every 152 seconds during the hemodialysis session. The primary outcome was a change in the PCr-Pi ratio during the session.

Results 

During the first hour of hemodialysis, mean phosphatemia decreased significantly (−41%; P<0.001); thereafter, it decreased more slowly until the end of the session. We found a significant increase in the PCr-Pi ratio (+23%; P=0.001) during dialysis, indicating a reduction in intracellular Pi concentration. The PCr-βATP ratio increased significantly (+31%; P=0.001) over a similar time period, indicating a reduction in βATP. The change of the PCr-βATP ratio was significantly correlated to the change of depurated Pi.

Conclusions 

Phosphorus magnetic resonance spectroscopy examination of patients with ESKD during hemodialysis treatment confirmed that depurated Pi originates from the intracellular compartment. This finding raises the possibility that excessive dialytic depuration of phosphate might adversely affect the intracellular availability of high-energy phosphates and ultimately, cellular metabolism. Further studies are needed to investigate the relationship between objective and subjective effects of hemodialysis and decreases of intracellular Pi and βATP content.

Clinical Trial registry name and registration number  

Intracellular Phosphate Concentration Evolution During Hemodialysis by MR Spectroscopy (CIPHEMO), NCT03119818

Hyperphosphatemia has been associated with increased cardiovascular mortality in ESKD.^1–3 In order to reduce hyperphosphatemia, international recommendations suggest dietary restriction of phosphate intake, use of oral phosphate binders, and increased dialytic phosphate (inorganic phosphate [Pi]) removal.⁴ However, current understanding of Pi removal is imperfect. Pi kinetics do not display the same bicompartmental kinetic behavior as urea. Around 700 mg to 1 g of Pi can be removed during a hemodialysis (HD) session. Study of available circulating Pi compared with the total mass transfer (700–1000 mg) during HD strongly suggests that serum phosphate cannot be the sole source. The plateau phase of serum Pi concentration, observed 1 hour after the start of HD, suggests a replenishment from one or more additional compartments. Everything we currently know about such additional compartments is derived from mathematical modeling on the basis of patient’s serum Pi changes during the HD.^5–8 All of these mathematical models make the same central assumption, namely that removed Pi originated exclusively from the intracellular compartment. There have been no available data concerning direct assessment of intracellular Pi in patients on HD. This remains an important question as consequences of intracellular Pi depletion on patients could be deleterious. Pi is indeed crucial for cell metabolism, especially because Pi allows ATP synthesis⁹ and reduction of intracellular Pi (as high-energy phosphate compounds) might lead to respiratory failure, depressed myocardial performance, and skeletal muscular weakness.¹⁰ Challenged cellular metabolism could be involved in a range of commonly observed objective and subjective consequences of HD.

To address this question of the decreased Pi origin, we previously set up an experimental study in a swine model of HD.¹¹ We measured intracellular Pi and βATP during the HD session using phosphorus magnetic resonance spectroscopy (³¹P-MRS) 2 days after binephrectomy. ³¹P-MRS recorded real-time changes in the relative concentrations of metabolites that are involved in high-energy phosphate metabolism, including ATP, phosphocreatine (PCr), and Pi, as well as changes in muscle intracellular pH. Intracellular Pi and βATP could be assessed noninvasively, continuously, and repeatably by ³¹P-MRS.¹² We reported a surprising increase in intracellular Pi content and a decrease of βATP during an HD session. These unexpected findings, relating to Pi concentration, were contrary to current mathematic models on the basis of data from established patients on HD. Hence, we postulated that the observed changes might be the result of studying an AKI model that could be subjected to severe acidosis, a situation not fully comparable with maintenance HD. We hypothesized that intracellular Pi dynamics should be different (with an appreciable amount of depurated Pi originating from within the cell) in patients on maintenance HD. We, therefore, applied similar magnetic resonance (MR)–based methods developed in the initial preclinical study to directly investigate the effect of conventional 4-hour HD session on intracellular Pi and βATP.

Methods

Study Design

The study was a pilot, single-center (Edouard Herriot Hospital, Lyon, France), prospective trial. It was approved by an ethics committee (2017-A00020–53; CPP Ouest 6, Brest, France) and by the French National Medicine Agency in April 2017. Written informed consent was obtained for all participants (ClinicalTrials.gov: NCT03119818).

Inclusion criteria were patients with ESKD of 18–80 years of age on maintenance HD >6 months with predialysis serum Pi between 1.5 and 3 mmol/L (4.6–9.3 mg/dl). Major exclusion criteria were protein-energy wasting, obesity, severe secondary hyperparathyroidism (parathyroid hormone ≥1000 pg/ml), significant anemia (hemoglobin ≤100 g/L), temporary vascular access, and contraindication to magnetic resonance imaging (MRI).

Outcomes

The primary outcome was the change of the PCr-Pi ratio over the period of a single HD treatment session. Principal secondary outcome was the change of the PCr-βATP ratio over the same period.

Sample Size Considerations

Justification of the clinical significance margin was challenging for this pilot study. However, a change in intracellular Pi (measured by MR spectroscopy) of around 20% has been reported to be associated with respiratory failure.¹³ Using data derived from a previous preclinical study,¹¹ we were able to derive the sample size needed to detect a similar magnitude of change. Assuming a reduction of the PCr-Pi ratio of 15% during the HD session, 12 patients would be required to deliver 90% power to this change at a 5% significance level.

HD Protocol

All patients underwent ³¹P-MRS during a standard 4-hour HD session. HD sessions were performed using a 5008 dialysis monitor (Fresenius, Bad Homburg, Germany) and a TS-2.1SL dialyzer (Toray Medical Co., Ltd, Tokyo, Japan). Dialysate composition was Na⁺=138 mmol/L, Cl⁻=107 mmol/L, K⁺=2 mmol/L, Ca²⁺=1.5 mmol/L, HCO³⁻=35 mmol/L, Mg²⁺=0.50 mmol/L, and acetate =3 mmol/L. The blood flow pump was set at 300 ml/min, and dialysate flow rate was set at 500 ml/min. Ultrafiltration was individually determined by interdialytic weight gain. Hemodynamic tolerance was assessed by the change in BP and heart rate. Because of the magnetic environment of MRI, the dialysis monitor was placed outside of the MRI examination room (4.2 m from the patient). The dialysis lines passed through a radiofrequency wave guide and had to be longer than usual in order to connect patients positioned on the bed of the MRI system. This setup increased the extracorporeal circuit by approximately 150 ml (18 ml/m of extended lines). Dialysate temperature was set to 37°C to prevent hypothermia given the low ambient temperature in the scanning room and extended extracorporeal circuit.

Plasma and Effluent Measurements

Sodium, potassium, chlorine, bicarbonate, protein, urea, creatinine, calcium, and Pi blood concentrations were measured before each dialysis session, as was blood gas for pH measurement. During the HD session, measurements were performed every 15 minutes for a period of 60 minutes and then every 60 minutes until the session was concluded. Parathyroid hormone (PTH) and 25-OH vitamin D were measured before and at the end of the session. During the session, effluent was harvested. Effluent phosphate concentrations were measured every 40 minutes, and effluent volumes were noted to measure phosphate mass transfer.

Intracellular Pi and βATP Measurements

³¹P-MRS was performed using a 3-Tesla MRI system (GE Healthcare, Boston, MA). The subjects were studied in the supine position, with the calf muscle of the right leg lying over a 20-cm-diameter surface coil. A sliding bed was used to move the patient’s leg in the center of the magnet. The surface coil was set to the phosphorus resonance frequency and positioned under the calf muscle to acquire the signal from a volume of interest of 4×4 cm, as previously described.¹¹ Phosphorus MR spectra (repetition time =2.6 seconds; echo time ≤1 ms) were acquired at baseline (before the beginning of the HD session) with patients at rest and every 152 seconds during dialysis. A typical baseline spectrum is represented in Figure 1. Spectra were processed and analyzed using jMRUI software (v. 5.2, http://www.jmrui.eu). The content of phosphorus metabolites was estimated by measuring the peaks area of the spectra, and particularly, the resonance peaks of Pi, PCr, and βATP were analyzed. The use of βATP peak was preferred to α- and γATP peaks because they are known to be contaminated by NADP and ADP signals, respectively. In contrast, the βATP peak contains only ATP signal, allowing an accurate estimation of intracellular ATP concentration.

Figure 1.: 31P-MRS MR spectrum acquisition from calf's muscle. (A) Axial T1-weighted MRI of the calf muscle showing the localization of the ³¹P-MRS acquisition from the volume of interest (red box) and (B) the acquired MR spectrum showing the different phosphorus containing compounds (Pi; PCr; and α-, β-, and γATP resonance peaks).

As the signal-noise ratio of the spectrum is dependent on technical parameters (such as surface coil tuning and sensitivity, magnetic field homogeneity, and radio-frequency power), we used a relative quantification method on the basis of metabolite ratios to follow the evolution of metabolite contents during time and between individuals.¹⁴ Metabolite ratios are thus proportional to the content variations of each compound, and the effect of technical variations is cancelled out. As PCr concentration is known to be stable in muscle at rest and in physiologic conditions, its content can be considered as an internal reference. Thus, we choose to use PCr-Pi and PCr-βATP ratios to describe the evolution of intracellular Pi and ATP concentrations during HD, consistently with our previous publication on this topic.¹¹ As a result, a decrease of PCr-Pi meant an increase of Pi and vice versa.

Reported intracellular concentrations of Pi and ATP are considered to reflect only the intracellular cytosolic part of these two metabolites because metabolites signal coming from circulating red blood cells is cancelled out of the acquisition volume of interest.

The intracellular pH (pHi) was calculated using the Henderson–Hasselbach formula: pH=6.75+log(δ−3.27)/(5.69−δ), δ being the difference (in parts per million) between the chemical shifts of Pi and PCr peaks.¹⁵ During the HD sessions, acquisitions were stopped hourly for 5 minutes to allow BP and heart rate measurements.

Calcium Balance Measurement

Calcium balance was calculated using the formula (Ca_e−Ca_b)(V_e−UF)+(Ca_e×UF), where Ca_e is the calcium in the effluent, Ca_b is the calcium in the dialysis solution, V_e is the volume of effluent, and UF is the ultrafiltration.

Statistical Analyses

Statistical analyses was performed using GraphPad Prism version 8.0.0 for Mac OS X (GraphPad Software, San Diego, CA) and R (R for Windows, version 3.4.4 [R-Cran project, http://cran.r-project.org/]). Continuous variables were presented as mean ± SD. Categorical variables were expressed as percentages. Metabolic ratios of PCr-Pi and PCr-βATP were averaged by 20-minute intervals for better readability. Comparisons of continuous variables between the beginning Time 0 (T0 minutes) and the end of dialysis (T240 minutes) were assessed with a nonparametric paired test (Wilcoxon test). Correlations were assessed by calculating Spearman correlation coefficient.

A linear mixed effects regression model was applied separately to each metabolic ratio (PCr-Pi and PCr-βATP) to quantify the rate of concentration change during the dialysis. A deviance analysis was applied to this model to test the significance of the fixed effects (type 2 Wald F tests with Kenward–Roger estimation for the degrees of freedom). This analysis was conducted using the “lme4” and “lmertest” packages in R software. P values of 0.05 were considered statistically significant.

Results

Characteristics of Patients and HD Parameters

Characteristics of the 11 patients are described in Table 1. The mean (± SD) age was 54.5 (±16.0) years, 91% of participants were men, and mean (± SD) dry weight was 70.8 (±10.8) kg. The mean (± SD) dialysis vintage was equal to 3.4 (±2.1) years, the mean (± SD) ultrafiltration volume was 1.8±0.7 L, and the mean (± SD) Kt/V was 1.31 (±0.2) (Table 1). As shown in Figure 2, the mean (± SD) systolic BP at T0 minutes (139±14 mm Hg) was not significantly different from that at T240 minutes (130±13 mm Hg, P=0.28).

Table 1. - Patient characteristics at screening visit

Patient Population, n=11	Mean ± SD
Mean age ± SD, yr	54.5±16.0
Men, n (%)	10 (91)
Mean dry weight ± SD, kg	70.8±10.8
Mean height ± SD, cm	172±7
Mean BMI ± SD, kg/m2	23.8±3.6
Mean systolic BP ± SD, mm Hg	139±14
Mean diastolic BP ± SD, mm Hg	80±8
Diabetes, n (%)	3 (27.3)
Mean HD vintage ± SD, yr	3.4±2.1
Mean ultrafiltration ± SD, L	1.8±0.7
Mean Kt/V ± SD	1.31±0.2
Blood volume processed ± SD, L	68.5±9.8
Treatment, n (%)
Calcium-phosphate binders	0
Noncalcium-phosphate binders	5 (45.4)
Natural vitamin D	5 (45.4)
1-OH vitamin D	3 (27.3)
Calcium	4 (36.4)
Cinacalcet	0

BMI, body mass index.

Figure 2.: No hemodynamic instability occurred during HD session. Systolic and diastolic BPs (millimeters of mercury) measured every hour until T240 minutes.

Before dialysis (T0 minutes), the mean (± SD) calcium level was 2.20 (±0.16) mmol/L, and the mean (± SD) phosphate level was 1.65 (±0.34) mmol/L (Table 2). Urea concentration decreased significantly (P<0.001) by 75% between T0 minutes (19.9±7.8 mmol/L) and T240 minutes (4.9±1.9 mmol/L), as reported in Figure 3.

Table 2. - Blood measurements before and after dialysis

Serum Concentration	Before (T0 min)	After (T240 min)
Mean sodium ± SD, mmol/L	138±2	138±0.7
Mean potassium ± SD, mmol/L	4.9±0.3	3.3±0.3 a
Mean chlorine ± SD, mmol/L	102±2	98±2 a
Mean bicarbonates ± SD, mmol/L	21±3	26±2 a
Mean urea ± SD, mmol/L	19.9±3	4.9±1.9 a
Mean creatinine ± SD, μmol/L	852±151	274±88 a
Mean calcium ± SD, mmol/L	2.20±0.16	2.45±0.16 a
Mean phosphate ± SD, mmol/L	1.65±0.34	0.71±0.16 a
Mean 25-OH vitamin D ± SD, mmol/L	80±49	77±46
Mean PTH ± SD, ng/L	164±161	94±98 a
Mean pH ± SD	7.37±0.03	7.48±0.04 a
Mean albumin ± SD, g/L	37.4±4.7	—
Mean hemoglobin ± SD, g/L	112.2±7.7	—

—, blood measurements before and after dialysis.

^aSignificantly different from the value before dialysis (P=0.05).

Figure 3.: Decrease of urea during HD session. Urea (millimoles per liter) measured every 15 minutes the first hour and then every hour until T240 minutes.

Phosphate Kinetics and Intracellular PCr-Pi Ratio

We were able to measure phosphate in the effluent in only five patients. The mean (± SD) cumulative amount of Pi removed at the end of dialysis was 38.5 (±4.7) mmol (Figure 4A). The plasmatic Pi concentration (1.65±0.35 mmol/L at T0 minutes) decreased significantly compared with its concentration at T60 minutes (0.97±0.38 mmol/L; −41%, P<0.001), which continues to decrease slowly to reach a plateau at T240 minutes (0.71±0.16 mmol/L). Between T0 and T240 minutes, we observed a significant (P=0.002) decrease of Pi content (−57%), as shown in Figure 4B. The PCr-Pi ratio increased significantly (+23%, P=0.001) from T0 to T240 minutes, meaning that intracellular Pi decreased during the HD session (Figure 4C).

Figure 4.: Decrease of intracellular Pi regarding decrease of extracellular Pi during a HD session. (A) Cumulative concentration of mean phosphate removal (in millimoles) during the HD session. (B) Phosphatemia level changes over the time. (C) PCr-Pi ratio changes over HD session. Dashed lines note the first hours of treatment when Pi starts to decrease.

Kinetics of PCr-βATP

The PCr-βATP ratio increased significantly (+31%, P=0.001) from T0 to T240 minutes, meaning that βATP decreased during the HD session (Figure 5A). The change in PCr-βATP ratio is significantly correlated to the amount of depurated Pi (P=0.002) (Figure 5B) and negatively to serum phosphate content (P=0.01) (Figure 5C).

Figure 5.: Decrease of ATP during a HD session. (A) PCr-βATP over the time. The rate of PCR-Pi changes during the HD session increased significantly using a mixed effects regression model (+31%, P=0.001). (B) Correlation between changes of the PCr-βATP ratios and the depurated phosphate content using the Spearman test (P=0.002). (C) Correlation between the PCr-βATP ratios and the phosphatemia-depurated Pi using the Spearman test (P=0.01).

Calcium Balance and Factors Important in Bone and Mineral Metabolism

The mean (± SD) calcium concentration increased significantly from 2.2 (±0.2) at T0 minutes to 2.4 (±0.2) mmol/L at T240 minutes (+9%, P=0.01). The mean (± SD) PTH concentration significantly decreased from 164 (±161) at T0 minutes to 94 (±98) ng/L at T240 minutes (P=0.002). There was no significant difference in 25-OH vitamin D concentration between T0 minutes (80±49 nmol/L) and T240 minutes (77±46 nmol/L, P=0.15) (Table 2). Mass balance of calcium was calculated at +17.1 mmol.

Bicarbonate Concentrations, Blood pH, and pHi Kinetics during HD

The mean (± SD) bicarbonate concentration increased significantly from 21 (±3) at T0 minutes to 26 (±2) mmol/L at T240 minutes (+24%, P<0.001) (Figure 6A, Table 2). Blood pH increased significantly from 7.37 (±0.02) at T0 minutes to 7.48 (±0.03) at T240 minutes (P<0.001) (Figure 6B), and there was no significant difference in pHi between T0 minutes (6.90±0.05) and T240 minutes (6.94±0.02, P=0.80) (Figure 6C).

Figure 6.: Normalization of the acid base status during HD session. Measurements (mean ± SD) during the HD session of (A) bicarbonates concentration, (B) blood pH, and (C) intracellular pH.

Discussion

We have conducted the first clinical study to report a significant reduction of intracellular Pi during HD, with an associated reduction in intracellular high-energy phosphates. Furthermore, we have demonstrated that MR spectroscopy is feasible to use as a metabolic probe in humans during HD itself and is worth further consideration as a research tool.

Our results confirmed previous mathematical assumptions supposing that Pi content belongs to the intracellular compartment for the most part. We report a biphasic kinetic of intracellular Pi concentration, whereby after a threshold (calculated herein to be 0.97 mmol/L), intracellular Pi content started to decrease. Interestingly, Daugirdas⁸ described the same phosphate threshold in a previous modified two-pool phosphate kinetic model, showing that intercompartmental phosphate clearance significantly increased after serum Pi concentration fell below 0.97 mmol/L. Previous modeling also hypothesized that Pi could be provided by the bone compartment.^5,16 To assess this question, we calculated the calcium balance that remained positive during the HD session as previously reported.^17,18 This positive calcium balance and concomitant decrease in PTH might not be in favor of massive phosphate release by the bone compartment, as these two factors are known to limit bone remodeling.

Results of this study differ from those obtained from our previous swine study showing an increase in intracellular Pi content during HD. First, this discrepancy could be explained by the experimental model used. Our swine model was an AKI model with acute ionic perturbation without the adaptive/maladaptive changes that accompany CKD. Pi pathophysiology in AKI has different pathophysiology and probably does not reflect existing mathematical models on the basis of data from patients on maintenance HD. Second, pigs had relative hypophosphatemia compared with their baseline serum phosphate, suggesting that animals had an acute negative phosphate balance. Hence, it could be hypothesized that this increase of intracellular Pi could result from an intracellular compensation mechanism. Lastly, the animals, when subjected to this very severe model of AKI (bilateral nephrectomy), exhibited particularly severe metabolic acidosis, leading to a low intracellular pH. The increase of Pi might be due to this acidosis, suggesting that free Pi was mobilized from polyphosphate or another Pi bond to proteins to buffer intracellular compartment.

Pi depletion can result in multiple organ system dysfunction, such as respiratory¹⁹ or cardiac²⁰ failure. Muscle fatigue has been shown to be related to the decrease in pHi and to Pi depletion.²¹ We did not report change in pHi suggesting that intracellular Pi depletion could lead to muscle weakness. The 23% decrease of Pi in our study corresponds to what has been described in patients with acute respiratory failure in chronic obstuctive pulmonary disease, emphasizing the potential harmful effect of intracellular Pi drop resulting from precipitous Pi depletion.¹³ Compared with a severe model of hypophosphatemia, patients with X-linked hypophosphatemia exhibit a 55% intracellular Pi decrease compared with controls at baseline, and these patients also experienced muscle weakness and fatigue.²²

These results questioned, therefore, the benefit of increasing phosphate removal during HD (at least within standard relatively short time delivery of 4 hours) as it seems such practices could alter cell metabolism. It is known that daily HD sessions²³ or longer nocturnal session lengths improve phosphate control. It is possible that daily short HD treatment would be more favorable for cell metabolism because the Pi depletion would be lower. Equilibrium between pools has been demonstrated to be very fast during exercise and probably exhibits the same kinetics in patients on HD. In a nocturnal long HD session, reduction in serum Pi is so marked that it is often necessary to introduce phosphorus into the dialysate to compensate and prevent clinically significant hypophosphatemia.²⁴ Further studies are needed to directly investigate the difference in Pi and ATP depletion between daily, classic, and long HD sessions.

Because intracellular Pi is an essential substrate for the formation of high-energy phosphate compounds and modulation of several key enzymes,²⁵ we hypothesized that the immediate implication of intracellular Pi drop would be reduced ATP synthesis. Pesta and coworkers²⁶ confirmed that muscle Pi depletion correlated with decreased ATP synthetic flux and thus, could lead to organ injury/dysfunction. The 15% reduction of ATP measured in our study is comparable with the ATP reduction reported after intense exercise reaching muscle exhaustion in patients on HD.²⁷ This decrease is also equivalent to those described in hypertrophy cardiomyopathy²⁸ and in dilated cardiomyopathy,²⁹ where PCr-ATP has been shown to be correlated with severity of heart failure. The magnitude of ATP reduction observed herein may, therefore, be implicated in both objective hemodynamic and subjective symptomatic intolerance of HD, commonly encountered in clinical practice.

Change in PCr-βATP ratio seemed to begin early, during the initial hour of dialysis, when serum P was >1.0 mmol/L. Even though the change in ATP correlated with phosphorus removal, we could also speculate that decrease of ATP could be explained by other mechanisms. Indeed, although patients with X-linked hypophosphatemia exhibit a decreased Pi-PCr ratio compared with the control group, Clarke and coworkers³⁰ did not report any difference in PCr-ATP compared with controls, suggesting that Pi depletion is not always associated with ATP depletion at rest. It is also likely that decrease of ATP could be the consequence of several mechanisms including: 1) dialysis induced circulatory stress inducing muscle vascular beds damage³¹ or 2) collapse of muscle mitochondrial membrane potential, as previously described in lymphocytes of patients on HD, although this observation remains contentious.³²

However, muscle is supposed to tolerate a reduction in resting cell Pi concentration of around 30%–40% without exhibiting bioenergetics. Mean intracellular Pi decrease was 22% in our study, suggesting that the depletion of ATP is not only a simple consequence of reduced intracellular Pi. Moreover, patients with X-linked hypophosphatemia exhibiting a decreased Pi-PCr ratio compared with the control group did not show any difference in PCr-ATP compared with controls. However, because previous report showed that HD leads to the collapse of mitochondrial membrane potential of lymphocytes, we could hypothesize that decrease of ATP concentration could be a consequences of this collapse of mitochondrial potential membrane. Finally, it is also well described that HD stress leads to ischemic episodes induced by dialysis itself. Although data are currently lacking, it is clearly possible that other vascular beds might also be damaged during conventional HD. Possible candidates include muscle with recurrent ischemic injury (including, perhaps, wasting as a cumulative effect) as well as the kidney itself.

This study has several limitations. First of all, we hypothesized that Pi and/or βATP depletions could be associated with fatigue and muscle weakness but did not include direct assessment of objective weakness or associated patient symptomatology. This study was constructed to explore kinetics of Pi and βATP and was not constructed to study consequences of these changes. It would certainly have been of interest to have included assessments, such as grip strength and sit-to-stand testing, as objective markers or time to recovery as a subjective marker. However, the experimental design and associated recruitment issues were already challenging, and this militated against increasing protocol complexity/burden further. We were not able to correlate βATP decrease and hemodynamic stress, but we can also suppose that patients with intradialytic hypotension could have a greater βATP decrease, and these patients were systematically excluded on the basis of patient safety in the setting of remote dialysis and limited monitoring possible within the MR scanner. These tests could be of interest to further explore in dedicated studies. We used as an inclusion criterion a serum Pi concentration between 1.5 and 3.0 mmol/L. This criterion was chosen to increase total quantity of depurated phosphate, aiming to increase phosphate transfers between compartments during HD sessions and our ability to measure them. However, it would also be interesting to test patients with lower or higher range of Pi to increase the generalizability of our findings. We did not quite attain the number of subjects defined a priori by power calculation, but a significant difference was found for the primary outcome. However, we have probably chosen a high threshold regarding the clinically significant margin by using observed Pi depletion in acute respiratory failure (a severe condition). Finally, providing that ³¹P-MRS measures intracellular Pi, we were not able to differentiate the Pi coming from the mitochondria, whereas it would be interesting to know as suggested previously by Spalding et al.⁵

Nevertheless, this study has strengths. We were able to safely measure intracellular βATP and Pi concentrations during HD for this technically challenging study in order to provide a better understanding of phosphate transfers between compartments during HD. Furthermore, no hemodynamic instability occurred, even after the lengthening of extracorporeal circuit lines (increasing extracorporeal blood volume to around 150 ml).

In conclusion, we confirmed by a direct in vivo measurement that depurated Pi comes from the intracellular compartment. These results question the benefit of increasing phosphate removal during HD as it could alter cell metabolism. Further studies are needed to investigate the relationship between HD symptoms and decreases of intracellular Pi and βATP contents. ³¹P-MRS may be an effective tool to monitor phosphate removal but also, to monitor energy consumption in patients on HD to optimize dialysis prescription.

Disclosures

L. Juillard reports consultancy agreements with Amgen, Hemotech, Baxter, Vifor, Otsuka, Astellas, Fresenius, Novartis, Leo, and Sanofi; research funding from Amgen, Baxter, and Sanofi; honoraria from Amgen, Hemotech, Baxter, Vifor, Otsuka, Astellas, Fresenius, Novartis, Leo, and Sanofi; scientific advisor to or membership with Amgen, Hemotech, Baxter, Vifor, Astellas, Fresenius, Novartis, and Leo. S. Lemoine reports honoraria from Chiesei and Anylam. All remaining authors have nothing to disclose.

Funding

This work was supported by Hospices Civils de Lyon.