Relationship between skeletal muscle intracellular ionized magnesium and measurements of blood magnesium

Relationship between skeletal muscle intracellular ionized magnesium and measurements of blood magnesium TIMOTHY W. RYSCHON, DONALD L. ROSENSTEIN, DAVID R. RUBINOW, JULIE E. NIEMELA, RONALD J. ELIN, and ROBERT S. BALABAN BETHESDA, MARYLAND

The current laboratory a p p r o a c h to assessing magnesium status is based on determining the concentration of total M g ([Mg]) in serum or plasma. This strategy is problematic in that the amount of Mg in blood is less than I% of total b o d y Mg and does not accurately reflect [Mg] in other tissues. Furthermore, the [Mg] of blood does not distinguish biologically active, ionized M g from the bound fraction. The goal of this study was to determine intracellular ionized Mg ([Mg++]i) of skeletal muscle in vivo and to c o m p a r e results with the [Mg] of blood constituents. [Mg++] i was determined in resting skeletal muscle by using phosphorus 31 magnetic resonance (31p-MR) spectroscopy. [Mg] was measured in serum (S[Mg]), serum ultrafiltrate (UF[Mg]), mononuclear blood cells (MBC[Mg]), and red blood cells (RBC[Mg]) by using atomic absorption spectroscopy or a colorimetric assay. In a sample of 60 healthy adult subjects, skeletal muscle [Mg++]~ = 557 +__97 p.mol/L (mean ± SD); S[Mg] = 0.78 ± 0.09 mmol/L; UF[Mg) = 0.60 ± 0.12 mmol/L; MBC[Mg] = 13.8 ± 2.3 mmol/L; and, RBC[Mg] = 1.92 ± 0.33 mmol/L. A significant negative correlation was found b e t w e e n [Mg++]~ and S[Mg] (r = -0.43, p < 0.05). S[Mg] was significantly lower (p < 0.05) and [Mg++]~ significantly higher (p < 0.05) in w o m e n than in men, but neither was related to age. These findings provide new insight into the relationship b e t w e e n blood Mg measures and [Mg++]i of the largest soft tissue mass of the human body. (J LAB CLIN MED 1996;127:207-13)

Abbreviations: ATP = adenosine triphosphate; CV = coefficient of variation; KD - dissociation constant; MBC[Mg] = total magnesium concentration in mononuclear blood cells; [Mg] = total magnesium concentration; [Mg++)~ = intracellular ionized magnesium concentration; MR = magnetic resonance; PCr = phosphocreatine; Pi = inorganic phosphate; 31p-MR = phosphorus 31 magnetic resonance; RBC[Mg] = total magnesium concentration in red blood cells; S[Mg] = total magnesium concentration in serum; UF[Mg] = total magnesium concentration in serum ultrafiltrate

From the Laboratory of Cardiac Energetics, National Heart, Lung and Blood Institute; the Biological PsychiatIy Branch, National Institute of Mental Health; the ClinicalPathologyDepartment, Warren Grant MagnusonClinicalCenter; and the National Institutesof Health. Supported by the Pediatric Scientist Development Program (T.W.R.). Submitted for publication Dec. 19, 1994; revision submitted Sept. 1, 1995; accepted Sept. 12, 1995. Reprint requests: Timothy W. Ryschon,MD, Laboratory of Cardiac Energetics, National Heart, Lung, and Blood Institute, National Institutes of Health, Building 1, Room B3-07, Bethesda, MD 20892. 0022-2143/96 $5.00 + 0 5/1/69865

agnesium is the second most a b u n d a n t cation in intracellular fluid and is essential in n u m e r o u s physiologic processes. 1 It serves as a cofactor for over 300 cellular enzymes, predominantly related to energy metabolism. 2 Alt h o u g h conditions of Mg excess are relatively rare, Mg deficiency occurs frequently, either as manifest h y p o m a g n e s e m i a or as latent n o r m o m a g n e s e m i c Mg deficiency. 3 Evidence of Mg deficiency has been reported in a variety of chronic illnesses, including alcoholism, diabetes mellitus, cardiovascular disease,

M

207

208

J Lab Clin Med February 1996

Ryschon et al.

and neuropsychiatric disorders. 4-n The role of Mg deficiency in the pathogenesis of these conditions is unknown. Studies of Mg metabolism are complicated by both the distribution of Mg in the body and the extent to which Mg is bound or ionized. About half of the Mg in the body is in bone and the other half in soft tissue, with less than 1% in the blood. 2 However, the majority of experimental data for this element are from studies of extracellular sources, primarily blood. 4-11 Current clinical assays for Mg determine [Mg] (protein-bound, complexed to anions, and ionized), whereas the biologically active form is ionized M g ) 2 Clinicians assume that the [Mg] in blood reflects Mg status at the tissue level. The lack of correlation between the [Mg] of skeletal muscle and blood [Mg] contradicts this assumption. ~3q9 Moreover, routine in vitro determinations of the [Mg] in blood constituents (serum, plasma, red blood ceils, and mononuclear blood cells) are inconsistently correlated with each o t h e r ) Thus the [Mg] of different tissues appears to be unrelated. The more important relationship between the [Mg] of blood components and [Mg++]i in other tissues has not been investigated. We hypothesized that skeletal muscle [Mg++]i is not correlated with the [Mg] of blood components. Our basis for this hypothesis was the aforementioned lack of correlation between [Mg] of blood components and solid tissue. To test this hypothesis, [Mg++]i was estimated from 31p-MR spectra obtained in vivo from the leg muscle of healthy male and female subjects. We chose to determine [Mg++]i in skeletal muscle because it constitutes a large accessible pool of whole body Mg. Skeletal muscle of the leg was selected for study because it is easily sampled by 31P-MR spectroscopy. Special consideration was given to the factors that influence the uncertainty of [Mg++]i estimation when using 31p-MR spectroscopy. [Mg] of blood components was determined by using standard clinical laboratory techniques.

women. Blood samples and MR scans were obtained between 8:30 AM and 12 noon to control for diurnal variations in [Mg]. 2° Subjects provided written and verbal informed consent for all procedures. 31P-MR Spectroscopy. Subjects were positioned supine on a standard patient transport bed and were offered a blanket for their personal comfort. The point of greatest circumference of muscle mass in the right lower leg (plantar flexor group) was determined by palpation and positioned over a single turn, 15 cm (diameter) circular surface coil tuned to the 31p frequency (69 MHz). A surface coil of this dimension was selected to accommodate the leg dimensions of the largest subject that we anticipated would be recruited. The leg was immobilized with Velcro straps and foam padding over and around the lower leg, and the limb and coil were positioned at the isocenter of a 4 tesla, 65 cm bore, Oxford superconducting magnet (Oxford Magnetics, Oxford, England; GE Omega Spectrometer, General Electric Medical Systems, Milwakee, Wis.). The homogeneity of the magnetic field was optimized based on the line width of the water 1H signal. In some subjects, images obtained with a 1H coil of identical geometry indicated that 31p signals were obtained from the soleus and gastrocnemius muscles. A single 31p-MR free induction decay was obtained in each subject by using the following parameters: sweep width 5 kHz, pulse width 125 microseconds, repetition delay 0.5 seconds, block size of 2048 points (acquisition time -0.48 seconds), and 128 acquisitions. Accumulation of 128 transients required - 1 minute, while orientation of the subject and positioning took another 3 to 4 minutes. Thus study participation required a total time of approximately 5 minutes. Each free induction decay was processed by baseline correction, zero filling to 4096 data points (final resolution 1.25 Hz • point-i), line broadening (exponential multiplication of 30 Hz--optimal for ATP 31p resonances), Fourier transformation, and zero and first order phase correction. The peak locations of the c~- and [3-phosphorus groups of ATP and of Pi and PCr were identified by the point of maximal amplitude. (See Fig. 1 for representative spectrum). Muscle pH was determined from the chemical shift of Pi relative to PCr (8Pi-PCr). 2~ [Mg++]i was calculated from the chemical shift difference between e~- and 13-phosphorus peaks of ATP by using a U MgATP value (50 ixmol/L at pH 7.2) adjusted for pH 22 and chemical shift limits described by Williams et al., 23 via the following equation:

METHODS Subjects. Twenty men and 40 women served as subjects

for this study. All subjects were screened by history and when necessary by radiograph to identify the presence of ferro-metallic implants that were considered a contraindication for MR scanning. All subjects were free of medical illness and taking no medications. Individuals with a history of claustrophobia or panic attacks were excluded. Serum ([3-human chorionic gonadotropin) or urine (ICON II HCG, Hybritech Inc., San Diego, Calif.) pregnancy tests before each MR scan were negative in all

[- ~ATP

Robs

= KMgATP/- ~o~13 -- ~'c~6 [Mg++]i D /Nobs[ U~x[3 __ X-@-TPuc~[ 3 NATPis the chemical shift between c~- and In this equation, ~,~ 13-phosphorus peaks in a Mg-free solution, NMgATP v~ is the chemical shift between e~- and [3-phosphorus peaks when ATP is fully complexed to Mg, and Ucq3 aobs represents the chemical shift between c~- and [3-phosphorus peaks observed in the leg 31p-MR spectrum.

J Lab Clin Med Volume 127, Number 2

Precision of 31P-MRS for [Mg++]~. The precision of this assay was tested in 5 subjects. To determine assay variability attributable to digitization, processing, and minor variations in leg position, five 3aP-MR spectra were acquired sequentially, with removal and repositioning of the subject between each acquisition. Although the leg was positioned each time according to the described method, we anticipated that slight variations in leg position would result from minor errors in leg orientation relative to the coil. [Mg++]i was determined as described. Blood Mg determinations. The determination of S[Mg], UF[Mg], RBC[Mg], and MBC[Mg] was performed by atomic absorption spectroscopy,24-26 or in some cases for serum, by colorimetric assay (Ektachem; Eastman Kodak, Rochester, N.Y.). In some subjects, platelet contamination of the mononuclear blood cell layer of the separation gradient precluded the determination of MBC[Mg]. Data analysis. All Mg measurements are expressed as mean _+ SD. The precision of the 31p-MR assay was assessed by calculating the CV for the assay [CV = (SD-mean-l.100)] in spectra acquired sequentially with repositioning. Pearson product-moment correlation coefficients were calculated between [Mg++]i and blood [Mg] values in all groups (Sigma Stat; Jandel Scientific, San Rafael, Calif.). A p value -< 0.05 was required to indicate a significant relationship. The probability of a type I statistical error was reduced by adjusting p values for multiple comparisons with the Bonferroni technique. RESULTS

The ages of subjects ranged from 20 to 62 years (20 men [37 _+ 12 years], 40 women [34 + 11 years]). The MR scanning protocol was well tolerated, and 31P-MR spectra were obtained from all subjects. There were no episodes of clinically significant anxiety, panic, or claustrophobia necessitating early removal from the magnet. Responses to a check-box questionnaire, completed by each subject after the magnet test, indicated no unusual sensations or experiences. Values for gPi-PCr, pH, sobs and [Mg÷+]i are shown in Table I for the 5 subjects that were repositioned in the magnet 5 times. The table also shows the CV for [Mg++]i, which is -<6% for all subjects. Mg measurements for all 60 subjects are summarized in Table II. No subject had a S[Mg] that exceeded the upper limit of the reference range (0.65 to 1.05 mmol/L), whereas 6 subjects were hypomagnesemic relative to our laboratory's reference interval. [Mg++]i values ranged from 307 to 901 ~xmol/L. The results from correlation analyses are shown in Table III. S[Mg] was negatively correlated with [Mg++]i (p < 0.05) (Fig. 2). Furthermore, S[Mg] was positively correlated with sobs ~=~ (p < 0.05) (Fig.

Ryschon et al,

209

PCr

ATP

//\ Pi

I 20

~

~

~

~

I 10

~

~

¥

~

~

I 0

~

a

~

~

~

13

I -10

J

~

~

J

1 -20

ppm

Fig. 1. 31P-MRspectra of resting skeletal muscle. A surface coil was positioned under the calf at the point of greatest circumference. Processingincludedbaseline correction,line broadeningvia exponential multiplication (30 Hz), Fourier transformation, and phase correction (signal-to-noiseratio of 116:1 for 13-ATP;PCr/ 13-ATPratio - 4.98:1). % c~,and 13represent the three phosphorus nuclei of ATP. 3), the inverse determinant of [Mg++]i (see Methods). S[Mg] was positively correlated with both UF[Mg] and RBC[Mg]. Women had significantly lower values of S[Mg] than men (0.76 _+ 0.10 mmol/L vs 0.81 + 0.08 mmol/E, p < 0.05) and significantly higher values of [Mg++]i (574 +_ 98 ~mol/L vs 523 _+ 87 >mol/L, p < 0.05). No significant relationship was found between age and [Mg++]i. DISCUSSION

In this study a significant negative correlation was found between skeletal muscle [Mg++]i and S[Mg]. To our knowledge this relationship has not been reported previously. Little is known about how Mg is distributed between vascular and muscle compartments and what controls [Mg ÷ ÷]i in different tissue compartments. Consequently, mechanistic explanations for this observation are quite speculative. In one possible scheme, skeletal muscle [Mg++]i is determined by the Mg concentration of the interstitial fluid surrounding the cell, which is in turn modulated by hormonal influences to maintain a gradient that favors the target intramuscular [Mg++]i. In healthy individuals, S[Mg] reflects serum [Mg++]i,27 and it is assumed for the purpose of this discussion that the latter is a close approximation of interstitial [Mg++]i. Accordingly, the system would respond to relatively low levels of muscle

210

J L a b Clin M e d F e b r u a r y 1996

R y s c h o n e t al,

Table I. pH and [Mg--+]~ in 5 subjects repositioned between each of 5 acquisitions Sub 1 2 3 4 5

~Pi-PCr

pH

~o~s

[Mg++]i

(mean __ SD)

(mean _+ SD)

(mean _ SD)

(mean -+ SD)

4,841 4.871 4.859 4.839 4.905

_+ 0.011 _+ 0,011 _+ 0.025 + 0.011 _+ 0.021

7.02 7.04 7.03 7.01 7.07

-- 0.01 _+ 0.01 _+ 0.02 ± 0.01 ± 0.02

8.476 8.494 8.470 8.495 8.510

_+ 0.012 _+ 0.0 _+ 0.008 ± 0.007 _+ 0.017

577 519 587 529 473

_+ 37 _+ 3 _+ 29 ± 19 ± 29

CV 6.3 0.6 5.0 3.6 6.1

gp~ pcr, Observed chemical shift of Pi relative to PCr peaks; pH, intramuscular pH; 8o~s, observed chemical shift of a- and I~-ATP peaks.

Table II. Magnesium measurements for the study

population Variable

n

S[Mg] UF[Mg] RBC[Mg] MBC[Mg] 8~°~~ [Mg++]i pH

60 56 49 48 60 60 60

Mean -+ SD 0.78 0.60 1.92 13.8 8.489 557 7.02

_+ 0.09 _+ 0.12 _+ 0.33 _+ 2.3 _+ 0.035 _+ 97 ± 0.04

Units mmol/L mmol/L mmol/L mmol/L ppm ixmoI/L

All [Mg] values were obtained via atomic absorption spectroscopy or colorimetric assay. Muscle [Mg++]~ and pH were measured by in viva 3~P-MR spectroscopy of calf muscle. g~l~, Observed chemical shift between a and 13-ATP peaks; pH, skeletal muscle intracellular pH.

[Mg++]i, with increases in S[Mg] and serum [Mg++]i resulting in increased interstitial [Mg++]i and a concentration gradient favoring correction of the low intramuscular [Mg++]i. The ability of extracellular [Mg] to influence intracellular [Mg] has been demonstrated by increases in intracellular [Mg] in resting amphibian muscle preparations 28 and in resting and contracting mammalian skeletal muscle29 bathed in solutions containing increased [Mg]. As with the reciprocal relationship between ionized calcium and parathyroid hormone, this hypothetical scheme of Mg regulation depends on an intermediate, possibly humoral factor that signals changes in muscle [Mg++]i. Although parathyroid hormone is a potential candidate for this role, 3°'31 there is currently no evidence linking it to muscle [Mg++]i. Regardless of the mechanistic explanation, these findings imply that in healthy people, the usual clinical approach of assessing tissue Mg status may yield erroneous interpretation, because in skeletal muscle, S[Mg] has an inverse correlation with the ionized species of this physiologically important intracellular ion. The relationship between S[Mg] and muscle [Mg++]i in disease states remains to be determined. Significant gender differences were observed for both S[Mg] and [Mg+÷]i in this study. Our finding of

lower S[Mg] for women relative to men is consistent with the findings of Lowenstein and Stanton for a sample of over 15,000 subjects. 32 Furthermore, in the current study [Mg++]i measurements were higher in women than in men, which is consistent with the observed relationship between S[Mg] and [Mg++]i for the total sample. The extent to which these gender differences are expressed in other tissues is unknown. 31p-MR for [Mg++]i measurement. The method used in this study for determining [Mg++]i from the chemical shift of the c~- and [3-phosphorus nuclei of ATP was described in 1978 by Gupta et al. 33 Factors that complicate the measurement of [Mg++]i when this approach is used include the binding state of Mg, the accuracy of chemical shift limits, the field strength, the pH, the temperature, and the ionic strength. Binding state of ATP. Although ATP is present in skeletal muscle cells in relatively high concentrations and is broadly distributed, 34 approximately 95% is bound to M g Y '36 When ATP is complexed to such a degree, ~obs u~ changes less with increases than with decreases in Mg. 37 Thus, in skeletal muscle, MR determinations of [Mg++]i are associated with less uncertainty as [Mg++]i decreasesF In this study the contribution of errors in pH and g~ obsto the uncertainty of [Mg++]i was assessed by using the MAGPAC programY This approach indicates that the uncertainty of muscle [Mg ÷ +]i can be confined to _+ 50 ~xmol/L (100 pomol/L range). The improved sensitivity of this technique at relatively low [Mg + +]i coincides with the fact that most disease states related to Mg metabolism are associated with Mg deficiency, as defined by hypomagnesemia. Thus this technique may be particularly advantageous for studying the mechanism between Mg deficiency and disease expression. Accuracy of chemical shift limits. Mosher et al. 37 have shown that large errors in the calculated [Mg++]i can occur when the value for the fully (,~MgATP,~ complexed chemical shift limit ~v~ j is derived from model solutions in which Mg is in excess (Mg/

J Lab Clin Med Volume 127, Number 2

Ryschon et al.

211

Table III. C o r r e l a t i o n matrix of m a g n e s i u m measures in t h e study p o p u l a t i o n

S[Mg] UF[Mg] RBC[Mg] MBC[Mg] [Mg++]~ obs 8~

S[Mg]

UF[Mg]

1.00 0.34* 0.51 * -0.06 -0.43* 0.44*

1.00 0.25 -0.12 -0.13 0.16

RBC(Mg]

1.00 0.01 0.18 0.15

MBC(Mg)

[Mg++]~

1.00 -0.13 0.17

1.00 -0.95*

~obs , Chemical shift between ~x and !3 phosphorus peaks observed in the leg 3~P-MR spectrum. *p < 0.05.

ATP > > 1:1 or Mg/ATP = 9). This may occur because of the formation of Mg2ATP and results in a larger chemical shift limit than found in a solution containing the minimum value of ,Robs ~ . Because the corrected "~FRMgATPassociated with the minimum Robs u~F was used in this study,37 the potential for this error was reduced. Magnetic field strength. In this study, in vivo 31P-MR spectra of resting human skeletal muscle were collected by using a 4 tesla MR system. This field strength, roughly three times greater than that found in clinical imaging systems, results in greater spectral dispersion (increased spectral resolution) of chemically shifted species and an increase in the signal-to-noise ratio of the 31p-MR spectrum relative to that found at a lower field strength. 3s Thus, compared with lower field strength magnets, this system is capable of detecting the small changes in obs 8~ that are expected to occur as Mg-to-ATP ratios approach 1:1, with greater precision. The improvement in the S/N ratio of 31p-MR spectra at this field strength permits acquisition of high quality spectra in less time than at 1.5 or 2 tesla. Tissue pH, temperature, and ionic strength. Like [Mg++]i, [H +] influences the chemical shift of ATP phosphorus peaks. 23 For this reason, the chemical shift limits (8~A~"P and RMgATP~ ,~ j used in the calculation of [Mg+÷]i should be derived in solutions having the same pH as the in vivo sample. Under the resting metabolic conditions of the current study, muscle pH ranged from 7.0 to 7.1. Although pH varied minimally, values for the chemical shift limits RATP MgATP~ ~ and Ru~ j that were used in this study were obtained from model solutions prepared across a range of pH values. 37 In addition, KI) values were propagated as a function of pH, assuming a K D of 50 lxmol/L at pH 7.2.22 Thus errors in calculated [Mg÷÷]i caused by. disregard of pH effects were minimized. It should be noted that the magnitude of the negative correlation between S[Mg] and [Mg÷÷]i was virtually the same as the positive cor-

relation between S[Mg] and Robs ,,~, suggesting that the influence of minor changes in pH on the calculation of [Mg ÷ +]i do not alter the underlying sensitivity of this method of determining [Mg+÷]i. Temperature and ionic strength can also affect the chemical shift of ATP phosphorus peaks, although to a lesser extent than found with pH and [Mg+÷]i99 Because neither was determined in this study, their influence on the current results cannot be determined. [Mg++]~ in human skeletal muscle. In the current study skeletal muscle was selected as the solid tissue for [Mg++]i determination, for two reasons. First, skeletal muscle comprises approximately 40% of body mass2 and 27% of total body Mg, 2 making muscle [Mg] a more accurate reflection of total body Mg status than blood constituent [Mg]. Second, skeletal muscle can be sampled easily by 31p-MR spectroscopy with a surface coil, because of its external location on the human body. Thus skeletal muscle Mg represents a significant pool of total body Mg and is easily assayed with this technique. The coil dimensions, leg position relative to the coil, and scanning parameters of the current study design resulted in spatial averaging of 31p-MR signal from the gastrocnemius and soleus muscles of the lower leg. The latter muscle contains a predominance (>85%) of slow-twitch fibers 4° relative to the gastrocnemius; however, Robs v ~ , pH, and hence [Mg++]i are not significantly different between soleus and gastrocnemius muscles of the cat. 41 For this reason, no effort was made to localize spectra to either muscle in this study. 31p-MR spectroscopy has been used to determine either relative or absolute values of [Mg ÷ ÷]i in skeletal muscle 42 and other tissues 43 in human subjects. Taylor et al. 42 report skeletal muscle [Mg++]i as being " . . . >1 mmol/L", corresponding to a chemical shift between [3-ATP and PCr of -15.88 to -15.98 parts per million. 42 A direct comparison of absolute values for [Mg÷+]i is not possible because the K D used in that study is

212

J Lab Clin M e d February 1996

Ryschon e t al.

1000

I

I

I

8.800

I

I

I

900 8.700 800 "-.. ++

700

-

"-.

•

"'-,.

•

00

•

8.600

•

...--"'"'""'""

oe•

~E

600

-.

..........

o•

Q •

•.,

8.500

," _.*,..~,-

. .....................

o

:~

500

o•

•

"-.o

"

/

...........

..-4""

• o•

•

8,400 400 8.300

300

200 0.4

I

I

I

0.6

0.8

1.0

1.2

S[Mg]

0.4

I

I

I

0.6

0.8

1.0

1.2

S[Mg]

Fig. 2. Plot of [Mg++]i and S[Mg] for the study population, along with the line of regression and 95% confidence limits. [Mg + +]i = 1.00213 0.000405 S[Mg]; r = -0.43; p < 0.05.

Fig. 3. Plot Ot~ %~ ~obs and S[Mg] for the study population along with the line of regression and 95% confidence limits. 8 obs ~ _9.097 + 1.163 S[Mg]; r = 0.44; p < 0.05.

different from that used in the current study. However, in the present study, the chemical shift between the PCr and #-ATP ranged from -15.9 to -16.1 parts per million, quite close to the range reported in that study. Thus the absolute chemical shift of the 13-ATP peak, at least relative to PCr, appears to be reproducible between laboratories. The relationship between [Mg] in blood constituents and [Mg++]i was not reported in that study.

for this difference include the screening procedures used in the current study, which excluded subjects on medications or with medical illness, and the time of blood collection in the current study, which was restricted as described in Methods. The finding that RBC[Mg] had no significant correlation with [Mg++]i whereas S[Mg] and [Mg++]i were significantly correlated suggests that Mg flux is specifically regulated across the RBC membrane. An additional finding of this study was the lack of correlation between MBC[Mg] and other blood constituents, which is consistent with results from other studies. 26'44 This finding further supports the existence of tissue-specific regulation of Mg movement. Conclusion. The significant negative correlation between [Mg++]i in skeletal muscle and [Mg] in serum is the primary finding of this study. This result may have significance regarding regulation of Mg concentration in the largest soft tissue compartment in the body. Correlations did not exist between mononuclear and red blood cells and may be due to the type of Mg determined (ionized vs total) or to tissue-specific differences in Mg transport or buffering. Because skeletal muscle comprises a majority of metabolically active body mass, the fact that the muscle [Mg++]i is significantly related to S[Mg] indicates that standard blood chemical analysis of [Mg] may be an inverse reflection of [Mg + +]i status in healthy subjects. This noninvasive in vivo assay of [Mg++]i via 31p MR spectroscopy appears to be sufficiently sensitive for comparative studies of Mg metabolism.

Precision end accuracy of 31p-MR spectroscopy for muscle (Mg++] i. A strategy for estimating uncertain-

ties in the calculation of [Mg++]i has been presented by Williams et al. 37 They recommend that the error in determining the chemical shift for Pi-PCr and gobsbe reduced to less than 0.08 ppm. 37 As noted in Table II, uncertainties in these variables are well below these limits. Based on the CV reported in Table I, the precision of this technique is comparable to that of S[Mg] determined by atomic absorption spectroscopy in healthy normal subjects (CV - 6 % ) 24 and in patients with congestive heart failure (CV --9%). 17 In the absence of an alternative method of determining [Mg++]i in vivo, the accuracy of this technique cannot be assessed. Total Mg in blood constituents. Significant positive correlations were found between S[Mg] and the [Mg] of RBC and UF. The high correlation between S[Mg] and UF[Mg] is a reflection of the fact that UF[Mg] is a component of S[Mg]. The positive correlation between RBC and S[Mg] contradicts the findings of other studies. 26'44 Possible explanations

J Lab Clin Med Volume 127, Number 2

Ryschon et al.

REFERENCES

1. Ryan MF. The role of magnesium in clinical biochemistry: an overview. Ann Clin Biochem 1991;28:19-26. 2. Elin RJ. Assessment of magnesium status. Clin Chem 1987; 33:1965-70. 3. Ryzen E, Elbaum N, Singer FR, Rude RK. Parenteral magnesium tolerance testing in the evaluation of magnesium deficiency. Magnesium 1985;4:137-47. 4. Cox IM, Campbell MJ, Dowson D. Red blood ceil magnesium and chronic fatigue syndrome. Lancet 1991;337:757-60. 5. Flink EB. Magnesium deficiency in alcoholism. Alcohol Clin Exp Res 1986;10:590-4. 6. Joffres MR, Reed DM, Yana K. Relationship of magnesium intake and other dietary factors to blood pressure: the Honolulu heart study. Am J Clin Nutr 1987;45:469-75. 7. Mendelson JH, Ogata M, Mello NK. Effects of alcohol ingestion and withdrawal on magnesium states of aIcoholics: clinical and experimental findings. Ann NY Acad Sci 1969; 162:918-33. 8. Sjogren A, Floren C, Nilsson A. Magnesium deficiency in IDDM related to level of glycosylated hemoglobin. Diabetes 1986;35:459-63. 9. Wester PO. Magnesium: effect on arrhythmias. Int J Cardiol 1986;12:181-3. 10. Abraham GE, Lubran MM. Serum and red cell magnesium levels in patients with premenstrual tension. Am J Clin Nutr 1981 ;34:2364-6. 11. Rosenstein DL, Elin RJ, Hosseini JM, Grover G, Rubinow DR. Magnesium measures across the menstrual cycle in premenstrual syndrome. Biol Psychiatry 1994;35:557-61. 12. Elin RJ. Magnesium metabolism in health and disease. Dis Mon 1988;34:163-218. 13. Lim P, Jacob E. Magnesium deficiency in patients on long-term diuretic therapy for heart failure. Br Med J 1972;3:620-2. 14. Martindale L, Heaton FW. The relation between skeletal and extracellular-fluid magnesium in vitro. Biochem J 1965;97:440-3. 15. Wester PO, Dyckner T. Diuretic treatment and magnesium losses. Acta Med Scand 1981;647:145-52. 16. Dyckner T, Wester PO. Skeletal muscle magnesium and potassium determinations: correlation with lymphocyte contents of magnesium and potassium. J Am Coll Nutr 1985;4:619-25. 17. Ralston MA, Murnane MR, Kelley RE, Altschuld RA, Unverferth DV, Leier CV. Magnesium content of serum, circulating mononuclear cells, skeletal muscle, and myocardium in congestive heart failure. Circulation 1989;80:573-80. 18. Fiaccadori E, Del Canale S, Coffrini E, et al. Muscle and serum magnesium in pulmonary intensive care unit patients. Crit Care Med 1988;16:751-60. 19. Reinhart R, Broste SK, Spencer S, Marx JJ Jr., Haas RG, Rae P. Relation between magnesium and potassium concentrations in myocardium, skeletal muscle, and mononuclear blood cells. Clin Chem 1992;38:2444-8. 20. Touitou Y, Touitou C, Bogdan A, Beck H, Reinberg A. Serum magnesium circadian rhythm in human adults with respect to age, sex and mental status. Clin Chim Acta 1978;87:35-41. 21. Kushmerick MJ, Meyer RA. Chemical changes in rat leg muscle by phosphorus nuclear magnetic resonance. Am J Physiol 1985;248:C542-9. 22. Bock JL, Wenz B, Gupta RK. Changes in intracellular Mg adenosine triphosphate and ionized Mg 2÷ during blood storage: detection by 3~p nuclear magnetic resonance spectroscopy. Blood 1985;65:1526-30. 23. Williams GD, Mosher TJ, Smith MB. Simultaneous determina-

24.

25.

26.

27.

28. 29. 30.

31.

32. 33.

34. 35.

36.

37.

38.

39. 40.

41.

42.

43.

44.

213

tion of intracellular magnesium and pH from the three 31P NMR chemical shifts of ATP. Anal Biochem 1993;214:458-67. Archer WH, Emerson RI, Reusch CS. Intra and extracellular fluid magnesium by atomic absorption spectroscopy. Clin Biochem 1972;5:159-61. Gochman N, Givelber H. Automated simultaneous microdetermination of calcium and magnesium by atomic absorption. Clin Chem 1970;16:229-34. Hosseini JM, Elin RJ. A method for determining the magnesium concentration of mononuclear blood cells. Trace Elem Med i988;5:47-51. Altura BT, Shirey TL, Young CC, et al. Characterization of a new ion selective electrode for ionized magnesium in whole blood, plasma, serum, and aqueous samples. Scand J Clin Lab Invest 1994;54(suppl 217):21-36. Fenn WO, Haege LF. The penetration of magnesium into frog muscle. J Cell Comp Physiol 1942;19:37-46. Rogers TA, Mahan PE. Exchange of radioactive magnesium in the rat. Proc Soc Exp Biol Med 1959;100:235-9. Allgrove J, Adami S, Fraher L, Reuben A, O'Riordan JLH. Hypomagnesaemia: studies of parathyroid hormone secretion and function. Clin Endocrinol 1984;21:435-49. Toffaletti J, Cooper DL, Lobaugh B. The response of parathyroid hormone to specific changes in either ionized calcium, ionized magnesium, or protein-bound calcium in humans. Metabolism 1991;40:814-8. Lowenstein FW, Stanton MF. Serum magnesium levels in the United States, t971-1974. J Am Coll Nutr 1986;5:399-414. Gupta RK, Benovic JL, Rose ZB. Magnetic resonance studies of the binding of ATP and cations to human hemoglobin. J Biol Chemistry 1978;253:6172-6. London RE. Methods for measurement of intracellular magnesium: NMR and fluorescence. Annu Rev Physiol 1991;53:241-58. Adam WR, Craik D J, Hall JG, Kneen MM, Wellard RM. Problems in the assessment of magnesium depletion in the rat by in vivo 31P-NMR. Magn Reson Med 1982;7:300. Kirkels JH, Van Echteld CJA, Ruigrok JJC. Intracellular magnesium during myocardial ischemia and reperfusion: possible consequences for post-ischemic recovery. J Mol Cell Cardiol 1989;21:1209. Mosher TJ, Williams GD, Doumen C, LaNoue KF, Smith MB. Error in the calibration of the Mg-ATP chemical-shift limit: effects on the determination of free magnesium by 31p NMR spectroscopy. Magn Reson Med 1992;24:163-9. Barfuss H, Fischer H, Hentschel D, et al. In vivo magnetic resonance imaging and spectroscopy of humans with a 4T whole-body magnet. NMR in Biomedicine 1990;3:31-45. Cohn M, Hughes TR Jr. Nuclear magnetic resonance spectra of adenosene di- and triphosphate. J Biol Chem 1962;237:176-81. Johnson MA, Polgar J, Weightman D, Appleton D. Data on the distribution of fibre types in thirty-six human muscles: an autopsy study. J Neurol Sci 1973;18:111-29. Meyer RA, Brown TR, Kushmerick MJ. Phosphorus nuclear magnetic resonance of fast- and stow-twitch muscle. Am J Physiol 1985;248(Cell Physiology):C279-87. Taylor JS, Vigneron DB, Murphy-Boesch J, et al. Free magnesium levels in normal human brain and brain tumors: 3~p chemical-shift imaging measurements at 1.5 T. Proc Natl Acad Sci USA 1991;88:6810-4. Ramadan NM, Halvorsen H, Vande-Linde A, Levine SR, Helpern JA, Welch KMA. Low brain magnesium in migraine. Headache 1989;29:416-9. Etin RJ, Hosseini JM. Magnesium content of mononuclear blood cells. Clin Chem 1985;31:377-80.