Intracellular Phosphate Dynamics in Muscle Measured by Magnetic Resonance Spectroscopy during Hemodialysis

Abstract

Of the 600–700 mg inorganic phosphate (Pi) removed during a 4-hour hemodialysis session, a maximum of 10% may be extracted from the extracellular space. The origin of the other 90% of removed phosphate is unknown. This study tested the hypothesis that the main source of phosphate removed during hemodialysis is the intracellular compartment. Six binephrectomized pigs each underwent one 3-hour hemodialysis session, during which the extracorporeal circulation blood flow was maintained between 100 and 150 ml/min. To determine in vivo phosphate metabolism, we performed phosphorous (³¹P) magnetic resonance spectroscopy using a 1.5-Tesla system and a surface coil placed over the gluteal muscle region. ³¹P magnetic resonance spectra (repetition time =10 s; echo time =0.35 ms) were acquired every 160 seconds before, during, and after dialysis. During the dialysis sessions, plasma phosphate concentrations decreased rapidly (−30.4 %; P=0.003) and then, plateaued before increasing approximately 30 minutes before the end of the sessions; 16 mmol phosphate was removed in each session. When extracellular phosphate levels plateaued, intracellular Pi content increased significantly (11%; P<0.001). Moreover, βATP decreased significantly (P<0.001); however, calcium levels remained balanced. Results of this study show that intracellular Pi is the source of Pi removed during dialysis. The intracellular Pi increase may reflect cellular stress induced by hemodialysis and/or strong intracellular phosphate regulation.

Cardiovascular risks and mortality are significantly increased by hyperphosphatemia, which is associated with elevated parathormone levels and several adverse consequences, including effects on bone formation, mineralization and remodeling, and soft tissue and vascular calcification,¹ increasing the risk for cardiovascular events and mortality.^2,3 Usually, in patients on dialysis, hyperphosphatemia can be controlled by hemodialysis (HD) associated with a low-phosphate diet and the use of phosphate chelators.^2,4 During a standard HD session, around 600–700 mg phosphate is removed from the plasma, whereas it contains only 90 mg inorganic phosphate (Pi); 85% of phosphate is stored in the bones and teeth in hydroxyapatite form, 14% is stored in the intracellular space (90% organic phosphate and 10% Pi), and 1% remains in the extracellular space.

During an HD session, phosphate kinetics are markedly different from urea.^5,6 Urea concentration decreases in function over time, assuming a simple dynamic equilibrium between intracellular and extracellular compartments, whereas plasma phosphate content decreased during the first 1 hour and then plateaued. Because phosphate concentrations have different diffusion kinetics than urea during dialysis,⁷ Spalding et al.⁶ proposed in 2002 a four-compartment model to describe phosphate transfers. On the basis of plasma phosphate kinetics observed during an HD session, this model is first composed of two compartments (intracellular and extracellular) that would interact by a diffusion mechanism, explaining the fast decrease of plasma phosphate during the first 1 hour of the session. However, to explain the plateau phase after the first 1 hour, Spalding et al.⁶ hypothesized the possibility of a third compartment, which could actively release some phosphate for maintaining plasma concentration. A fourth intracellular compartment was added for the emergency release of phosphate in the cell to maintain vital cytosolic phosphate concentration.

However, the robustness of this model was limited by the unknown intracellular phosphate concentration variations, which were assumed to decrease by passive transfer between intra- and extracellular compartments.

Currently, ³¹phosphate (³¹P) magnetic resonance imaging spectroscopy (MRS) is the only technique able to measure in vivo intracellular phosphate variations in a dynamic and noninvasive way. The interest is in the key role of phosphorylated molecules in energy metabolism, and the technique is used to investigate metabolic process and muscle energy metabolism in clinical practice. Like the proton and despite having a slightly lower sensitivity, ³¹P can be used with MRS. When ³¹P is placed in a magnetic field and excited by an appropriate radiofrequency, ³¹P restores the energy received as a signal detectable by a suitable antenna. This signal occurs after data processing in the form of different spectra corresponding to Pi, ATP, and phosphocreatine (PCr). From these spectra, intracellular pH (pHi) variations can be also measured. Moreover, ³¹P MRS is a noninvasive tool that can be used in animals and humans.

If a previous study by Durozard et al.⁸ has reported the use of ³¹P MRS to measure Pi concentration in humans, the evolution of Pi was never investigated during dialysis.⁸

Therefore, we propose in this study to follow, by ³¹P MRS, the Pi concentration during HD sessions of pigs to test the hypotheses that the intracellular compartment could be the main source of phosphate during HD and that Pi concentrations will decrease during the session, potentially disturbing energy metabolism of the cell and explaining a part of HD intolerance in some patients.

Results

Predialytic Characteristics

One HD session was performed on six pigs. Mean (±SD) weight of pigs was 28±1.4 kg. Mean weight gain between surgery and first dialysis was 1±0.2 kg. Systolic BP varied from 90 to 145 mmHg, and diastolic BP varied from 30 to 90 mmHg. Cardiac frequency varied from 50 to 110 beats per minute. Sodium, potassium, chlorine, urea, creatinine, calcium, and phosphate concentrations in blood before surgery, and before and at the end of dialysis are shown in Table 1.

Table 1.

Blood measurements before surgery and before and after dialysis

Blood Concentration	Before Surgery	Before Dialysis	After Dialysis
Sodium (mmol/L)	142±3	137±5	137±2
Potassium (mmol/L)	4.0±0.5a	7.3±1.6	4.0±0.8a
Chlorine (mmol/L)	104±5	97±6	98±3
Bicarbonates (mmol/L)	24±3a	12±3	18±5a
Urea (mmol/L)	3±1a	39±8	21±5a
Creatinine (mmol/L)	72±11a	1027±80	500±44a
Calcium (mmol/L)	2.57±0.14	2.8±0.3	2.5±0.3
Phosphate (mmol/L)	2.32±0.37	2.26±0.6	1.6±0.4a

Values are means±SDs.

^aSignificant values are compared with blood measurement before dialysis.

Plasma Urea and Bicarbonates Concentration Kinetics

Mean plasma urea concentration (39±8 mmol/L before dialysis) decreased significantly during the whole session to reach the mean value of 21±5 mmol/L (−47±3%; P<0.001) at the end of the 3-hour sessions (Figure 1).

Figure 1.

Good performance of dialysis. (A) Urea and (B) bicarbonates kinetics during dialysis. Values are means±SDs. Urea decreased over the time, and bicarbonates increased over the time.

Mean plasma bicarbonates concentration (12±3 mmol/L before dialysis) increased significantly during the whole session to reach the mean value of 18±5 mmol/L (+50%; P<0.001) at the end of the 180-minute sessions.

Calcium Balance

Mean plasma calcemia concentration was not different before dialysis (2.8±0.3 mmol/L) from after dialysis (2.5±0.3 mmol/L; P=0.30) (Figure 2). At the end, the calcium balance was at 0.72±0.6 mmol (Figure 2).

Figure 2.

Calcium in the effluent was almost identical to the calcium contained in the dialysis solution. Calcemia does not modify over the time. (A) Kinetics of calcemia during dialysis sessions. (B) Calcium balance. Calcium balance was measured with 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.

Kinetics of Changes of Extracellular Phosphate

Figure 3A shows the average of phosphate (16.56±1.43 mmol) removed during the HD session.

Figure 3.

Intra-cellular phosphate concentration increase during extracellular phosphate plateaued. Removed phosphate, extracellular Pi kinetics, and intracellular PCr-to-Pi ratio kinetics. (A) The mean removed phosphate was measured in the effluent bags throughout the sessions. (B) Extracellular phosphate decreased during the first 1 hour, plateaued, and then, rose before the end of dialysis. (C) The PCr-to-Pi ratio plateaued and then, decreased (−11%; P<0.001; i.e., increase of intracellular phosphate) when extracellular phosphate began to plateau.

The mean plasma phosphate concentration was equal to 2.26±0.5 mmol/L before dialysis, decreased rapidly during the initial part of the dialysis (−30.4%; P<0.001), and then plateaued at 1.59±0.12 mmol/L (Figure 3B). After 150 minutes, mean plasma phosphate concentration began to rise and reached a value of 1.63±0.4 mmol/L at the end of dialysis.

Kinetics of Pi

Figure 3C shows the kinetics of the PCr-to-Pi ratio. We found a significant decrease of the PCr-to-Pi ratio (−11%; P<0.001; correlation factor =0.59). Unlike the extracellular kinetics described previously, we found a significant increase of intracellular phosphate. The observed variations of intracellular Pi concentration can be described in three phases. First, there was no variation of the Pi content, despite an extracellular Pi decrease. Second, the Pi content increased significantly (P<0.001), whereas extracellular phosphate began to reach a low threshold. Third, the intracellular Pi reached a second plateau when the content of extracellular phosphate began to rise.

βATP and pHi Measurements

The PCr-to-βATP ratio increased significantly (+6%; P<0.001) (Figure 4A). The mean pHi increased significantly during the dialysis session (+1%; P<0.01).

Figure 4.

βATP decreased and intracellular pH increases during dialysis. (A) The Pcr/βATP ratio increases significantly (6%) during dialysis (meaning a decrease of βATP), whereas (B) pHi (+1%; P<0.001) increases significantly.

Discussion

In this study, we developed an original noninvasive model of intracellular phosphate measurement by in vivo ³¹P MRS during dialysis. For the first time, and surprisingly, we showed that intracellular Pi content increased during the dialysis session, whereas extracellular phosphate decreased and then plateaued.

The extracellular phosphate kinetics reported in our work are in agreement with previous observations.^5,9 Although a continuous removal of phosphate was occurring during the dialysis session (16 mmol for 3 hours), extracellular phosphate reached a plateau at a concentration of 1.6 mmol/L, in contrast to urea kinetics. Moreover, we confirmed that extracellular phosphate began to rise around 30 minutes before the end of the sessions, which was also previously described by Spalding et al.⁶ Indeed, Spalding et al.⁶ used a four-compartment model to describe the evolution of Pi transfers during dialysis: an extracellular compartment where Pi is removed, an intracellular compartment, a third pool (unknown) that releases Pi when plasma Pi reaches a threshold, and a fourth pool (intracellular), which may be the mitochondrion, considered as an “emergency pool.”⁶ This model assumed that extracellular phosphate was used as a surrogate of intracellular phosphate on the basis of a diffusive transfer between extra- and intracellular compartments and a decrease of Pi after the extracellular decrease of phosphate. Furthermore, Spalding et al.⁶ suggested that the decreased Pi concentration triggers the control mechanism. However, Pi concentration was not measured. Interestingly, we showed in this study that Pi increased when extracellular phosphate plateaued, and then stabilized when extracellular phosphate began to rise, contrary to our first hypothesis. These results suggested that a rapid production of intracellular phosphate began when extracellular phosphate reached a threshold. Pogglitsch et al.¹⁰ suggested that phosphate was released into the extracellular space proportionally to the phosphate decrease after reaching a critically low threshold. Therefore, our findings suggested that Pi may come directly from the intracellular compartment. Spalding et al.⁶ showed also that a longer dialysis removed a higher mass of phosphate and resulted in significantly higher postdialysis phosphate concentrations, which could be explained by the release of intracellular phosphate. These results were in agreement with a previous report by Maasrani et al.,¹¹ that showed an increase of intracellular Pi with a decrease of phosphorylated sugar in the red blood cells (RBCs) of patients who are uremic.

Our data provide inferences on the mechanism involved in phosphate removal during dialysis. The finding of a constant increase in the intracellular compartment seriously challenges the previously described models, the most advanced form being by Spalding et al.⁶ This study suggests a three-compartment model: intra- and extracellular compartments, with a third intracellular.

It has been suggested that the bone pool could be the origin of this Pi increase. Bones contained 86% of the total body phosphate, and phosphate regulation could be explained by efflux from hydroxyapatite contained in bones. During sessions, and the calcium in the effluent was almost identical to the calcium contained in the dialysis solution. Only a massive efflux calcium that was not observed in our study would accompany a significant efflux of phosphate. Therefore, our study is not in favor of a significant release of Pi and calcium by the bone.

Pogglitsch et al.¹⁰ suggested that plasma phosphate could be delivered to plasma at a rate that cannot be explained solely by passive movement of intracellular Pi to plasma but requires additional generation from intracellular storage forms. We hypothesized that this phosphate could come from some organelles or biochemical reaction from glycophosphate in the cytosol as previously suggested.¹²

We also suggested that it could come from the mitochondria, where polyphosphate (polyP) is produced. However, the mitochondria Pi was not detectable by MRS because of its limited degree of freedom. Intracellular Pi is sequestered in intracellular organelles, and its regulation is closely linked to cellular metabolic activity. Certainly, the inhibition of phosphate uptake impaired cellular metabolic function and reduced mitochondrial respiration, oxidative phosphorylation, and ATP content. In our study, we showed the kinetics of two metabolic parameters: the PCr-to-Pi ratio (related to phosphorylation potential) and the Pcr-to-βATP ratio. We showed a decrease of the PCr-to-Pi ratio and an increase of the PCr-to-βATP ratio, meaning an increase of Pi and a strong decrease of βATP during the dialysis sessions while pigs were at rest. These results suggested that mitochondria play a role in the regulation of intracellular Pi concentration, with an increased consumption of ATP during the HD sessions.¹³ We suggested that polyP could be involved in this increase of Pi; polyP is an energy-rich polymer consisting of up to several hundreds of phosphate residues.^14,15 There are only a few studies concerning the occurrence of polyP in animals; they are widely distributed and can be localized in the cell in a flexible, oriented, and relatively stable form, but also used in response to a wide variety of metabolic needs as a source of energy. These findings could have important consequences in clinical practice. Indeed, the important Pi transfer between the intra- and extracellular compartments strongly activated intracellular Pi release and consequently, an ATP decrease. Because ATP plays a key role in cell energy metabolism, this intracellular ATP decrease may induce a clinical intolerance during dialysis, such as asthenia and muscle weakness, which is observed in some patients. These results stress the need for additional clinical research for investigating the effect on Pi metabolism of different dialysis strategies to improve patient clinical tolerance.

Another major interest of our study consisted in its capacity to, first, measure in vivo phosphate metabolism and second, monitor its concentration during a complete dialysis session of 3 hours. Cellular phosphate dynamic transfers remained poorly studied because of the difficulty of measuring Pi variations noninvasively. We performed HD sessions on 30-kg pigs in a magnetic resonance imaging (MRI) system, a highly electromagnetic environment, and prove its feasibility. Although blood lines had to be longer, we did not experience any major alarms or hemodynamic failure. From a methodologic point of view, the magnetic resonance (MR) spectra acquisitions did not suffer from the difficult experimental conditions, including the presence of the dialysis system. The quality and the signal-to-noise ratio of the MR spectra were excellent, allowing an accurate quantification of the metabolites. The animal was always sufficiently anesthetized, and no unwanted movements disturbed the MR acquisitions. Thus, our experimental model was successful and provided precise and dynamic measurements of pH and phosphate metabolites, including Pi, during a complete dialysis session. This experimental study also proved that the clinical project can be conducted safely, because MR/MRS is a noninvasive clinical tool.

One of the limits of this study was the number of animals used (n=6), with only one HD session per pig. There are some differences between our model and patients with chronic HD. Pigs have spontaneously higher Pi blood concentration than humans (as already described in the literature).¹⁶ Moreover, we did not observe an increase of plasma phosphate concentration, because the dialysis was initiated rapidly after surgery in the context of reduced food intake. Second, BP was rather low, ranging from 90 to 145 mmHg, which is explained by the effect of anesthesia drugs. Third, pigs were severely acidotic. These points suggest that the model used did not completely mimic a stable hyperphosphatemic patient on chronic HD. However, this study shows that MRS is feasible during dialysis and could reasonably be used in patients on chronic HD.

We did not carry out a parathyroid hormone assay, because we performed only one dialysis session on each pig.

Also, we did not perform an absolute quantification of the metabolites because we did not have access to a volume selection sequence. Only relative concentrations were obtained on the basis of the first starting point of the follow-up as reference. Nevertheless, the ratios and relative concentrations were very stable throughout the follow-up period. Furthermore, the PCr-to-Pi ratio constitutes a very stable marker of the overall bioenergetics state of the cell, in particular of the mitochondrial phosphorylation.

In conclusion, we provide strong evidence that during dialysis, intracellular phosphate concentration increased, whereas extracellular phosphate decreased, contrary to previous suggestions. These unexpected findings are essential to improve our understanding of phosphate transfers during dialysis.

Concise Methods

The University Claude Bernard Lyon 1 Committee for Animal Experimentations approved the study protocol.

Animal Surgery and Monitoring Protocol

Six 3-month-old female domestic piglets (30.3±0.6 kg) were sedated before anesthesia using ketamine (5 mg/kg intramuscular injection; Imalgene 1000; Merial SAS, Lyon, France) and Rompun (2%, 1 mg/kg; Bayer Pharma Santé Animale, Puteaux, France). A 22-gauge catheter was placed in the ear vein to inject propofol (1.5 mg/kg; Diprivan; AstraZeneca Pharmaceuticals, Rueil-Malmaison, France) before a tracheal intubation. Pigs were then ventilated mechanically, and anesthesia was maintained with a constant infusion of propofol.

General anesthesia was maintained with Diprivan (7 mg/kg per hour) and Fentanyl (3 μg/kg per hour; Merck GmbH, Lyon, France).

At day 1, a catheter was inserted in the right external jugular vein for drug infusion, and a tunneled double–lumen dialysis catheter was placed into the right external jugular vein for dialysis. After that, a binephrectomy through a bilateral lobotomy was performed. Pain was managed with Perfalgan (Bristol-Myers Squibb, Rueil-Malmaison, France) administered intravenously after surgery and Doliprane (60 mg/kg per day; TheraplixGroupe Aventis Pharma, Paris, France) 4 days after surgery.

On postoperative days 2 and 3, food and beverage (500-ml/d restriction) were allowed. To prevent hyperkalemia, the animals received three spoons of Kayexalate (Sanofi Pharma, Paris, France) per day. HD was performed at day 4.

During surgery and dialysis sessions, awakening signs, BP, cardiac frequency, and venous blood oxygen saturation were monitored. The animals were weighed before surgery and before every dialysis session.

At the end of the dialysis session, the pigs were euthanized with a lethal injection of Tanax (0.3 ml/kg; Intervet Productions, Igoville, France).

HD Protocol

At day 4, the pigs were anesthetized and ventilated, and a 180-minute HD session was performed.

A Prismaflex (Hospal, Meyzieu, France) generator and an M100 Dialyzer (Hospal) were used. Hemosol B0 Dialyzing Solution (Hospal) was used with a composition of Ca²⁺ (1.75 mmol/L), Mg²⁺ (0.50 mmol/L), Na⁺ (140 mmol/L), Cl⁻ (109.5 mmol/L), acetate (3 mmol/L), and bicarbonates (32 mmol/L). When dialysis started, the blood flow pump was set at 50 ml/min and then, 100–150 ml/min after 5–10 minutes according to animal tolerance; 0- to 150-ml/h ultrafiltration was set depending on animal tolerance and weight gain.

The tolerance was estimated by BP and cardiac frequency evolution. Anticoagulation procedure consisted of a 2500-IU heparin bolus followed by a 1500-IU/h continuous perfusion.

Because of the magnetic environment of the MRI, the dialysis generator was placed outside of the MRI examination room (5 m from the animal). The dialysis lines had to be larger than usual (4.4 m for the arterial and venous lines) to connect the animals positioned on the bed of the MRI. This setup increased the extracorporeal circuit by approximately 500 ml (150 ml for the M100 set and larger lines) instead of the usual 150 ml. To prevent cardiac failure because of fast blood depletion, the generator was prepared with macromolecules (VOLUVEN; 500 ml; Fresenius Kabi France, Sèvres, France), and the dialysis began with a low blood flow (50 ml/h). A blood line heater was used to prevent hypothermia.

Plasma and Effluent Measurements

Sodium, potassium, chlorine, bicarbonate, protein, urea, creatinine, glucose, calcium, and phosphate blood concentrations were measured before surgery on day 1, before every dialysis session, and then during the HD session at every 5 minutes for a period of 40 minutes, followed by every 20 minutes until the session was concluded. At the end of the session, measurements were taken every 5 minutes for 25 minutes to study the rebound phase.

Measurements were taken on the effluent bag every 45 minutes. Sodium, potassium, chlorine, protein, urea, creatinine, glucose, calcium, and phosphate concentrations were measured.

Calcium balance was measured with 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.

Intracellular Phosphate and ATP Measurements

³¹P MRS was acquired at CERMEP, Imagerie du Vivant on a 1.5-Tesla Siemens Sonata MR System (Siemens Medical Solutions, Erlangen, Germany). The 20-cm surface coil was set on the ³¹P resonance frequency (25.6 MHz)¹¹ and placed over the gluteal muscle region, because the animal was in dorsal decubitus (Figure 5A). Appropriate positioning of the coil was assessed by the scout acquisitions of MRI images. ³¹P MR spectra (repetition time=10 s; time echo=0.35 ms) were acquired every 2 minutes and 40 seconds before, during, and after dialysis. There were means of eight acquisitions before, 67 acquisitions during, and 19 acquisitions after the session.

Figure 5.

Different peaks of molecules containing phosphate during MRS acquisition. ³¹P MRS acquisition with (A) the schematization of the surface coil sensitivity superimposed to the anatomic acquisition and (B) the acquired MR spectra showing Pi, PCr, and α-, β-, and γATP resonance peaks. L, left; R, right; I, inferior; S, superior.

A typical ³¹P nuclear magnetic resonance spectrum is represented in Figure 5B. The observed peaks were very well separated, allowing accurate measurements of different metabolite contents.

³¹P MRS data were analyzed using jMRUI Software. Five different peaks were analyzed: Pi, PCr, α-, β-, and γATP, and pHi, which was calculated using the Henderson–Hasselbach formula: pH=6.75+log(δ−3.27)/(5.69−δ), with δ being the difference (in parts per million) between Pi and PCr resonance frequencies.¹⁷

Because RBCs are circulating, atoms in RBCs excited by signal are out of the volume studied during the acquisition of the spectra. Therefore, in MRS in vivo, intracellular muscle Pi is naturally distinguished for RBC Pi, because Pi from RBCs is not recorded. Moreover, the limited blood volume in tissue would make the contribution of Pi from RBCs extremely low.

Statistical Analyses

Mean plasma and effluent measurements variations were analyzed using a t test. Metabolic variations were analyzed through a repeated ANOVA measurement using GraphPad Software. Statistical relevance was defined by P<0.05. Variations of the PCr-to-Pi ratio were fitted with a third–order polynomial cubic curve.

Disclosures

Financial support was provided by Baxter-Gambro Hospital.