Magnesium (Mg) is the fourth most abundant mineral in the body, and the most abundant intracellular divalent cation, and is essential for a diverse range of physiological functions. Mg deficiency, either from inadequate intake, excess excretion or altered homeostasis, is often suspected to be associated with the initiation of many symptoms and diseases(Reference Saris, Mervaala, Karppanen, Khawaja and Lewenstam1). In spite of its multiple and ubiquitous roles, Mg status can be assessed in severe deficiency and a large number of studies have investigated various markers during nutritionally induced or pathological mild and chronic deficiency. A review published in 1991 stated that the assessment of Mg status was difficult as there was no simple, rapid and accurate test to indicate total body Mg status(Reference Elin2). More recently, another review indicated that whilst functional and/or biological markers are available for many nutrients, there is still a need for indicators to specifically diagnose Mg deficiency(Reference Franz3). Another difficulty associated with Mg is related to its metabolism. It is known that the equilibrium and exchange of Mg between body compartments and tissue pools occurs slowly, so determining Mg concentration in one tissue may not provide information about Mg status in another. This review provides the most up-to-date information on the assessment of Mg status.
Metabolism
The metabolism of Mg and its body distribution have been investigated in animals and humans using the radioactive isotope 28Mg(Reference Arnaud4, Reference Danielson, Johansson and Ljunghall5) but its use is limited by the short half-life of 21 hours, and for human studies by exposure to radiation. Thus, 25Mg and 26Mg stable isotopes have been used in human to accurately assess Mg absorption, excretion, bioavailability, pool sizes, and turnover. However, the relatively high abundance of the isotopes used (10·13 % for 25Mg and 11·17 % for 26Mg) imposes the administration of doses of Mg for analytical precision that may be absorbed and eliminated dose dependently. The mean biological half-life of Mg has been estimated to be between 41 and 181 days(Reference Avioli and Berman6, Reference Watson, Hilditch, Horton, Davies and Lindsay7). These values are consistent with the observation that it takes 3 months to normalize a 20 % depletion of body Mg stores(Reference Gold, Riediger, Matthes, Kuhn, Graef, Temme, Katz, Roka, Lasserre and Durlach8).
Compartments
Observations from stable isotope studies in children and adolescents indicate that it is not possible to directly relate Mg deficiency to changes in the exchangeable pool size or pool turnover(Reference Abrams and Ellis9). In adults, compartmental analysis has shown that only 25 % of total body Mg can be studied using stable isotopes, i.e. that which rapidly exchanges between the plasma compartment and two extra-plasma pools. Consequently, the majority of total body Mg escapes this analysis and is transferred to long-term storage pools, particularly bone(Reference Sabatier, Pont, Arnaud and Turnlund10) where approximately half the body's Mg content (1 mol) is found. Less than 1 % of total body Mg is present in blood, with approximately 0·3 % in serum(Reference Elin2).
Biomarkers and tests for status assessment
Serum
Blood samples must be prepared carefully to prevent Mg contamination with anticoagulant, and generally blood concentrations are determined in serum rather than plasma. Haemolysis can increase the Mg concentration because erythrocytes contain more Mg than serum. In addition to sample preservation, a survey of the quality of the measurements of routine laboratories has been performed and it was concluded that these sources of error are significant and may reduce the diagnostic potential of serum total Mg(Reference Dewitte, Stöckl, Van de Velde and Thienpont11).
Serum Mg concentrations are dependent on dietary intake and intestinal absorption as well as kidney function. Kidney filtration and reabsorption are essential to maintain stable serum levels between 0·75 and 0·96 mmol/l, a range observed for healthy adult subjects. Serum Mg concentration is the most frequently performed analysis, and whilst some studies have found correlations between serum and tissues values(Reference Alfrey, Miller and Butkus12, Reference Arnold, Tovey, Mangat, Penny and Jacobs13) others have not(Reference Fiser, Torres, Butch and Valentine14–Reference Lukaski and Nielsen16). Bone is the primary storage site for both calcium and Mg, and provides a labile pool for the release of Mg to maintain serum concentrations. Decreases in serum Mg can be observed when drugs are taken, particularly diuretics. Otherwise, lower serum Mg values indicate deficiency and impaired metabolic control as observed in diabetes, renal tubular disorders, alcoholism and malabsorption. By comparison, higher serum values are observed when subjects take Mg medication such as Mg-rich antacids or cathartics(Reference Woodard, Shannon, Lacouture and Woolf17) or in the case of renal failure. In conclusion, serum values lower or higher than the 0·75–0·96 mmol/l range require diagnosis and treatment; but values in the normal range also do not rule out the possibility of total body deficit compensated for by the release of Mg from the bone pool.
Diabetes and obesity
Hypomagnesaemia has been reported in type 2 diabetes but after 3-months of treatment with 30 mmol/day given orally, the values increased from 0·73 ± 0·8 to 0·81 ± 0·1 mmol/l, equivalent to control group concentrations(Reference Eibl, Kopp, Nowak, Schnack, Hopmeier and Schernthaner18). A comparison of serum Mg concentrations in 109 type 2 diabetics with 156 age- and sex-matched healthy controls in Switzerland showed significantly lower values for the diabetic subjects, 0·77 ± 0·08 and 0·83 ± 0·07 mmol/l, respectively. Serum Mg concentrations were below the normal reference range in 37·6 % of the diabetic patients and 10·9 % of the control subjects (P < 0·001). This study highlights the fact that serum Mg concentrations indicate that poor Mg status is common in type 2 diabetics in Switzerland(Reference Wälti, Zimmermann, Spinas and Hurrell19).
In obese patients, a significant negative correlation was found between waist-to-hip ratios and serum Mg concentrations, whilst no such correlation was observed for either erythrocyte or platelet Mg(Reference Corica, Allegra, Ientile and Buemi20).
Heart disease
In patients with end-stage heart disease and Mg stores highly depleted, an intravenous administration of 64 mmol Mg significantly increased serum (0·68 ± 0·06 to 0·82 ± 0·3 mmol/l), erythrocyte (1·44 ± 0·3 to 1·75 ± 0·4 mmol/l) and lymphocyte Mg (1·23 ± 0·7 to 1·52 ± 0·8 μg/mg of protein) and also total urinary Mg excretion (3·4 ± 1·27 to 17·5 ± 8·1 mmol). These changes observed after the 24-hour infusion of Mg ascorbate were all highly significant(Reference Gajos, Solarska, Siembab, Nessler, Nessler, Stepniewski and Piwowarska21).
Crohn's disease
After intravenous infusion of 60 mmol Mg to subjects with Crohn's disease, a significant increase in serum and mononuclear cell Mg concentrations was observed in addition to an increased muscle and body retention(Reference Sjögren, Florén and Nilsson22).
Asthma
Mg status has been evaluated in healthy subjects and those with mild to moderate asthma. Measurements included total and ionized serum and erythrocyte Mg and the retention of an intravenous Mg load. The results showed that total serum Mg offers a useful clinical diagnostic tool and that ionized Mg is closely correlated but does not offer any diagnostic advantages(Reference Kazaks, Uriu-Adams, Albertson and Stern23).
In conclusion, serum analysis is useful for the determination of Mg status in deficient subjects with pathologies leading to increased urinary excretion. In patients seen for routine medical care at an urban family centre, there was a 20 % overall prevalence of hypomagnesaemia among this predominantly female African American population(Reference Fox, Ramsoomair, Mahoney, Carter, Young and Graham24). However, this observation did not rule out a higher prevalence of deficiency which could only be characterized by the measurement of low tissue Mg.
Blood cells
Erythrocytes
Normal erythrocytes contain a high concentration of Mg ions that are essential for ATP function and other metabolic processes(Reference Bock, Wenz and Gupta25). The Mg content of erythrocytes has been shown to decrease in humans provided with low Mg diets. These changes were only observed after several weeks of low dietary intake, the delay being attributed to the erythrocyte pool which reflects long-term rather than current nutrient status(Reference Deuster, Trostmann, Bernier and Dolev26).
Diabetes and obesity
In non-insulin type 2 diabetics, both serum and intracellular ionized erythrocyte Mg concentration was significantly lower compared with non diabetic control subjects. Oral supplementation for 8 weeks with 400 mg/d Mg restored erythrocyte concentrations to normal values without changing serum concentrations(Reference Nadler, Malayan, Luong, Shaw, Natarajan and Rude27). Lower values of Mg in erythrocytes have also been reported in type 1 diabetics compared with control subjects (1·41 ± 0·56 v. 2·94 ± 1·12 mmol/l, respectively), while serum concentrations were similar and urinary Mg excretion significantly elevated in the diabetic group(Reference Gürlek, Bayraktar and Özaltin28). A one year oral Mg supplementation study, with a daily dose of 13 mmol (300 mg) as Mg gluconate, in type 1 diabetic patients characterized by low erythrocyte Mg ( < 2·3 mmol/l) significantly increased these values from 2·03 ± 0·03 to 2·48 ± 0·12 mmol/l while serum Mg levels did not change: 0·73 and 0·77 mmol/l(Reference Engelen, Bouten, De Leeuw and De Block29). A recent study performed on diabetic children reports that erythrocyte Mg levels showed an inverse correlation with percentage of retained Mg load. Although erythrocyte and serum Mg were significantly lower in diabetic children compared with controls, serum Mg concentrations were in the normal range in both groups. The authors suggest that erythrocyte Mg measurement is preferred to serum Mg and that the load test is a reliable and sensitive method(Reference Şimşek, Karabay and Kocabay30).
In a group of normotensive obese patients, erythrocyte and platelet Mg concentrations, but not serum, were significantly lower than in healthy controls(Reference Corica, Allegra, Ientile and Buemi20).
Migraine
Total Mg levels in plasma, erythrocytes and lymphocytes were analyzed in a group of 29 migraine patients and 18 control subjects. At baseline, results showed significantly lower concentrations of total erythrocyte Mg in migraine patients compared with controls (50·7 ± 4·7 v. 52·9 ± 4·7 mg/l). Migraine patients then received a daily supplement of 1 l mineral water containing 110 mg/l Mg for 2 weeks, which resulted in a significant increase in erythrocyte Mg concentrations from 50·7 ± 4·7 to 52·9 ± 4·7 mg/l with no observed effect on plasma Mg(Reference Thomas, Millot, Sebille, Delabroise, Thomas, Manfait and Arnaud31).
Chronic fatigue syndrome
Shorter-term effects of intramuscular Mg administration given every week for 6 weeks to patients with chronic fatigue syndrome showed a significant increase in erythrocyte Mg concentrations (+0·57 ± 0·192 mmol/l). This study raises an important methodological problem. The erythrocyte Mg concentrations were different in two groups of patients with chronic fatigue syndrome, 1·29 and 1·60 mmol/l, whilst a normal range of 1·41–2·09 mmol/l have been suggested. This important difference was explained by the fact that the measurements were undertaken in different laboratories and under different conditions(Reference Cox, Campbell and Dowson32). Although it is known that Mg slowly leaches out of red cells into the plasma, the length of sample storage time could not explain the differences observed.
Critical illness
From a study in critically ill postoperative patients and healthy controls, the best Mg parameter to measure hypo- or hypermagnesaemia was ionized Mg in erythrocytes when compared with total erythrocyte Mg and total ionized serum Mg. The prevalence of hypomagnesaemia was 15·9 % from the measurement of total serum Mg, 22·2 % from ionized serum Mg and 36·5 % from ionized Mg in erythrocytes, a level almost twice as high as that observed in total or ionized serum Mg(Reference Malon, Brockmann, Fijalkowska, Rob and Maj-Zurawska33).
Asthma
Since low Mg intake has been associated with airway hyper-responsiveness, a study was undertaken in 49 asthmatic patients in which Mg in erythrocytes, serum and urine was measured. The results showed lower erythrocyte and urine Mg as compared with healthy controls whereas serum concentrations did not differ(Reference Emelyanov, Fedoseev and Barnes15).
Healthy subjects and genetic control
It has been observed that intra-individual variations in plasma and red blood cell Mg concentrations over long time periods are small when compared with inter-individual variations. The analysis of family resemblance for Mg concentrations in serum and erythrocytes, based on data from nuclear families and twins showed that while adult plasma Mg varies linearly with age, erythrocyte Mg shows a non-linear trend: quadratic for males and fifth degree polynomial for females. Univariate and bivariate model analyses of these results strongly suggested that genetic factors were primarily responsible for the observed family resemblance and that one common genetic factor alone could not explain all the correlations(Reference Darlu, Rao, Henrotte and Lalouel34). Further analysis under a mixed model yielded significant support for a major gene effect on erythrocyte Mg, but not on plasma Mg. Parameter estimates indicated that the data are compatible with a common major gene for elevated erythrocyte Mg. About 5 % of the population appeared homozygous for this gene and nonfamilial factors account for a small fraction of the total variance(Reference Lalouel, Darlu, Henrotte and Rao35). Genetic factors controlling intra- and extracellular Mg levels were shown amongst unrelated adult male blood donors to be composed of at least three components: the major histocompatibility complex (HLA and H-2)-associated genes, the non-major histocompatibility complex genes, and tissue factors modulating the respective importance of the first two sets of factors(Reference Henrotte, Pla and Dausset36).
Studies performed on mice demonstrated that animals selected for low and high Mg levels exhibited significant different total (25·1 v. 18·0 mg/l) and ionized (11·8 v. 9·1 mg/l) plasma Mg, total erythrocyte Mg (54·5 v. 40·7 mg/l) and decreased tibia (4·45 v. 3·56 mg/g) and kidney (955 v. 869 mg/kg) Mg concentrations. They also had a higher urinary excretion and changes in the size and exchange rates of other compartmental pools. The population examined were the second generation between 4 inbred strains, and pairs with the highest and the lowest erythrocyte Mg concentrations were selected for reproduction. After the same selection for 18 consecutive generations, erythrocyte Mg values diverged rapidly and regularly and remain constant. These two strains were homozygote for all the relevant alleles(Reference Feuillet-Coudray, Coudray, Wolf, Henrotte, Rayssiguier and Mazur37, Reference Feuillet-Coudray, Trzeciakiewicz, Coudray, Rambeau, Chanson, Rayssiguier, Opolski, Wolf and Mazur38). It was also shown that the genetic impact was not only observed with erythrocytes but also affects total body Mg metabolism(Reference Feuillet-Coudray, Coudray, Wolf, Henrotte, Rayssiguier and Mazur37). Today, with the discovery and progress in transient receptor potential-melastatin (TRPM) research, we anticipate that a better understanding of the genetic expression at the cellular and tissue levels of these channels controlling epithelial Mg transport will explain the differences observed in Mg homeostasis.
In spite of this genetic regulation, the Mg content of erythrocytes has been studied as an index of Mg status in 20 healthy women with erythrocyte Mg concentrations below the 15th percentile ( ≤ 1·97 mmol/l). Supplementation of 250 mg/day Mg for 3 weeks resulted in an increase in erythrocyte Mg concentration of only 1·6 % while plasma Mg level was significantly increased by 5·3 %. The authors concluded that erythrocyte Mg is not a useful measurement for monitoring the effect of Mg supplementation in individuals(Reference Basso, Ubbink and Delport39). This opinion is shared by others(Reference Borella, Facchinetti, Rocchi, Lorini, Bargellini, Golf, Dralle and Vecchiet40) although it has been reported that a 10 % lower erythrocyte Mg concentration in adults with marginal deficiency was restored to normal values after 2 weeks of supplementation with 15·6 mmol Mg (360 mg) as Mg pyrrolidone carboxylate(Reference Borrella, Bargellini and Ambrosini41).
Low Mg diet in healthy subjects
The effect of a low Mg intake (112 mg/day) for 92 days preceded and followed by control periods of 35 and 49 days with a daily supplement of 200 mg showed a significant decrease of Mg in erythrocytes and muscle and a higher Mg retention(Reference Lukaski and Nielsen16). These study protocols with induced deficiency and recovery in healthy subjects are effective to test both kinetic and quantitative changes in Mg cell and tissue pools as well as in urine. With further data, these intervention studies will allow the validation of status biomarkers by demonstrating their sensitivity and specificity for measuring Mg status.
Leukocytes
In both animal and human studies, the Mg content of white blood cells such as lymphocytes was shown to be a better index of intracellular Mg in skeletal and cardiac muscle(Reference Ryan and Ryan42, Reference Ryan, Ryan and Counihan43). It was even stated(Reference Ryan, Ryan and Counihan44) that lymphocytes have advantages over other tissues, such as erythrocytes and muscle, for assessing intracellular Mg because during experiments on Mg deficient rats, the magnitude of the Mg loss from lymphocytes was similar to that of cardiac and skeletal muscle.
Diabetes and obesity
Mg was analyzed in plasma, mononuclear cells, erythrocytes, urine and in muscle biopsies from 25 subjects with type 1 diabetes and the results were compared with those of 28 healthy controls. Mg in mononuclear cells was suggested to be an index of intracellular Mg and a significant correlation between muscle and mononuclear cells was reported in patients(Reference Sjögren, Floren and Nilsson45).
Crohn's disease
After intravenous infusion of 60 mmol Mg to 30 subjects with Crohn's disease, a significant increase in the Mg concentrations in mononuclear cells (65–94 % lymphocytes, 5–30 % monocytes, 0–3 % basophilic cells and 0–1 % granulocytes) was observed along with increases in plasma, muscle and an increased body retention of Mg. However, the authors concluded that the analysis of muscle Mg and an estimation of Mg retention during an intravenous infusion are superior markers for confirming suspected Mg deficiency(Reference Sjögren, Florén and Nilsson22).
Migraine
In migraine patients, mononuclear blood cell Mg concentrations were significantly lower than in controls subjects (8·52 ± 3·64, 7·93 ± 2·84 and 10·6 ± 3·38 mg/d DNA, respectively)(Reference Gallai, Sarchielli, Morucci and Abbritti46). Total Mg levels in plasma, lymphocytes and erythrocytes and ionized Mg in lymphocytes were analyzed in a group of 29 migraine patients and 18 control subjects. Results showed significantly lower concentrations of ionized lymphocyte Mg (12·0 ± 3·5 v. 14·2 ± 3·8 mg/l) and total Mg in erythrocytes (50·7 ± 4·7 v. 53·5 ± 2·9 mg/l) in migraine patients compared with controls. After a 2 week daily supplementation with 1 l of mineral water containing 110 mg/l Mg, a significant increase in all intracellular Mg concentrations with no effect on plasma Mg was observed in migraine patients. Among the analyzed parameters, ionized lymphocyte Mg appeared to be the most sensitive index of Mg deficiency with a 15 % decrease in migraine patients when compared with controls and a 16 % increase after 2 weeks of a Mg-rich mineral water intake(Reference Thomas, Millot, Sebille, Delabroise, Thomas, Manfait and Arnaud31).
Hypertension and heart disease
Mg content in lymphocytes and skeletal muscle biopsies from 28 subjects demonstrated no significant correlation between these values except in a group of three normal volunteers and nine patients with mild arterial hypertension(Reference Dyckner and Wester47). Another study showed that total intracellular Mg content was significantly lower in lymphocytes from hypertensive patients compared with healthy subjects (0·07 ± 0·03 v. 0·11 ± 0·04 mmol/g protein) while serum and erythrocyte Mg and ionized platelet Mg(Reference Kisters, Tepel, barenbrock, Westermann, Rahn, Zidek and Spieker48) were not significantly different.
Patients with congestive heart failure experienced cardiac arrhythmias due to digitalis toxicity. Although serum Mg concentrations were within normal ranges lymphocyte content was decreased, suggesting the existence of cellular Mg depletion. Intravenous bolus administration of Mg sulphate, followed by intramuscular Mg repletion, abolished the digitalis-toxic arrhythmia(Reference Cohen and Kitzes49).
Critical illness
Because the Mg content of mononuclear blood cells was suggested to be a better index of Mg status than serum concentrations, these measurements were performed in critically ill patients who were either moderately or severely hypomagnesaemia ( ≥ 0·4 to ≤ 0·6 mmol/l and ≤ 0·4 mmol/l, respectively) and receiving a 24-hour intravenous Mg replacement therapy (0·5 and 0·75 mmol/kg of intravenous Mg sulfate, respectively). Serum concentrations increased significantly from baseline to 48 h (0·5 ± 0·1 to 0·8 ± 0·2 mmol/l) while intracellular Mg content did not change significantly within the study period (2·6 ± 1·0 to 3·0 ± 1·3 fmol/cell). In this group of trauma patients from an Intensive Care Unit, serum Mg was a better index of Mg status than mononuclear blood cell Mg(Reference Sacks, Brown, Dickerson, Bhattacharya, Lee, Mowatt-Larssen, Ilardi and Kudsk50).
Healthy subjects
Although Mg deficiency has been diagnosed using low Mg levels in leucocytes, their concentration varies according to the pathology and the study population(Reference Borella, Facchinetti, Rocchi, Lorini, Bargellini, Golf, Dralle and Vecchiet40). Some studies do not find a correlation between mononuclear blood cells, such as monocytes and lymphocytes, and serum or erythrocyte Mg. A 2-fold larger inter-individual coefficient of variation for the Mg content of mononuclear blood cells than for serum and erythrocytes introduced by the separation step and washing of the cells may explain the lack of correlation(Reference Elin and Hosseini51).
Platelets
Diabetes, obesity and hypertension
The concentration of Mg in platelets has been measured in two groups of normotensive and hypertensive type 2 diabetic patients and in healthy subjects. Plasma and erythrocyte Mg concentrations were significantly lower in the diabetic patients. The concentrations of Mg in platelets were lower in patients but the difference was significant only for the the hypertensive diabetics(Reference Corica, Ientile, Allegra, Romano, Cangemi, Di Benedetto, Buemi, Cucinotta and Ceruso52). Later, this research group showed that both microalbuminuria and proteinuria in type 1 diabetic patients was associated with altered Mg homeostasis and a negative correlation was found between glycated haemoglobin and both plasma and platelet Mg(Reference Allegra, Corsonello, Buemi, D'Angelo, Di Benedettot, Bonanzinga, Cucinotta, Ientile and Corica53). These studies do not demonstrate that platelets are a more sensitive and specific marker of Mg status. However, in another study performed on obese normo- or hypertensive patients, platelet Mg was significantly reduced compared with controls. However, other parameters were also significantly decreased such as Mg in plasma and in erythrocytes. The conclusion of the authors(Reference Corica, Allegra, Ientile and Buemi20) was that intra platelet Mg assay was more reliable than the dosage of ionize Mg by NMR. Although the study did not investigate platelets, ionized Mg in erythrocytes showed similar but less striking changes in normotensive and hypertensive obese subjects when compared with non-insulin-dependent diabetes(Reference Resnick, Gupta, Bhargava, Gruenspan, Alderman and Laragh54).
Buccal cells
In a study investigating the effects of various Mg levels in drinking water in healthy subjects, Mg was determined in muscle biopsies and in sublingual mucous membrane cells. A negative correlation was found between muscle and sublingual cell Mg suggesting that these cells cannot be used to evaluate intracellular Mg status(Reference Rubenowitz, Landin and Rylander55). More studies are needed to evaluate whether these cells, which require non-invasive collection procedures, can provide valuable information. Contamination from Mg in saliva or the salivary microbiota and previously ingested beverages or water can affect the results and must be controlled.
Tissues
Bone and teeth
There are more experimental studies in animals than in humans on the effects of Mg deficient diets on bone. Rats fed a diet providing a surfeit of Mg and 2 others diets resulted in two degrees of Mg deficiency. While the recommended Mg dose is 40 mg/100 g dry diet, the control group of this study received 150 mg/100 g and the two groups with deficiency received either a diet without Mg for one week or 5 mg/100 g for 2 weeks. There was no consistent difference between the Mg concentrations found in liver, heart, or skeletal muscle of Mg-deficient and control rats, but bone accurately reflected the level of dietary Mg. However, there was a significant difference between the Mg concentration of the anterior and posterior halves of the ribs, indicating irregular distribution of Mg within the bone. There were also significant differences in the Mg concentration of different bones from the same animals. Therefore one entire bone, such as the sternum or the rib, should be analysed(Reference Caddell and Scheppner56). Weanling rats fed six levels of dietary Mg, ranging from 0 to 150 mg/100 g purified diet, showed a linear decrease in Mg retention with increased bone Mg. As a negative relationship was found between Mg retention following the load test and the level of dietary Mg, the load test appears to be an acceptable means of indirectly assessing Mg status provided there is normal renal and cardiovascular status and normal water balance(Reference Caddell, Heineman and Reed57). In a more recent study, it was observed that mice selected for their low Mg status had reduced total and ionized plasma Mg, and lower erythrocyte, tibia and kidney Mg levels(Reference Feuillet-Coudray, Coudray, Wolf, Henrotte, Rayssiguier and Mazur37).
A study on a 2-month-old boy with congenital hypomagnesaemia has been published in which Mg concentrations were measured at 8 and 12 years of age in milk teeth lost naturally and the values compared with those of healthy control children(Reference Guillard, Mettey, Lecron and Pineau58). A significantly lower Mg content was found in teeth, 4·80 v. 6·82 ± 0·77 mg/g dry weight and 4·12 v. 6·18 ± mg/g dry weight at 8 and 12 years old respectively compared with controls. However, a greater significant difference was observed at 12 years for serum (0·60 ± 0·02 v. 0·86 ± 0·02 mmol/l), erythrocyte (1·55 ± 0·04 v. 2·20 ± 0·15 mmol/l), and lymphocyte Mg (1·42 ± 0·14 v. 3·72 ± 0·41 fmol/cell). Similar differences were found at 8 years of age demonstrating that dental Mg is potentially interesting for the evaluation of calcified tissues(Reference Guillard, Mettey, Lecron and Pineau58). In adult subjects, Mg concentrations in serum and bone were significantly reduced in a patient with chronic hypomagnesaemia. However, as reported previously in animals, Mg in bone is not homogeneously distributed and the values measured are dependent on sampling. It was shown that 30 % of bone Mg is in a surface limited pool present either within the hydration shell or on the crystal surface. The larger fraction of bone Mg was shown not to be associated with bone matrix but rather to be an integral part of the bone crystal. The authors concluded that the major factor determining Mg concentration in bone would appear to be the serum Mg level(Reference Alfrey and Miller59). In a study of the same group where muscle, erythrocyte and bone Mg were measured in patients with reduced, normal and increased Mg levels, a highly significant correlation between serum and bone Mg was reported. The authors' conclusion was that bone Mg in man increased during Mg excess and decreased during Mg depletion(Reference Alfrey, Miller and Butkus12). The invasive sampling and the heterogeneous distribution of Mg in bone are the reasons why there is limited knowledge on the metabolism of Mg in bone.
Muscle
A quarter of total body Mg is located in muscle. This tissue is a significant compartment and seems appropriate to assess Mg status, however, the limited number of studies is explained by the need to perform invasive biopsies.
Heart disease
The Mg retention test used together with muscle biopsies in 5 patients with acute myocardial infarction (MI) and 6 healthy controls demonstrated a significantly higher mean retention of 42 % in the MI group compared with 22 % in the control group. A lower muscle content of Mg in the MI group supported the results of the retention test and indicated Mg deficiency(Reference Jeppesen60).
Crohn's disease
Mg status was evaluated in 30 subjects with Crohn's disease and 30 healthy controls. Subjects with Crohn's disease had significantly lower concentrations of Mg in muscle, mononuclear cells, and 24 h urine collections compared with controls. A significant increase in muscle, mononuclear cells, and plasma Mg concentrations was observed following intravenous infusion of 60 mmol Mg into the Crohn's disease patients. The retention of the infused Mg was significantly higher in subjects with Crohn's disease than in 11 healthy controls and was inversely correlated with muscle Mg content. The most sensitive measurements of Mg status and deficiency were the analysis of Mg in muscle and the estimation of Mg retention(Reference Sjögren, Florén and Nilsson22).
Lung disease
Muscle biopsies and serum samples have been taken in patients with chronic obstructive pulmonary disease and acute respiratory failure and Mg measurements showed that 10 % of the patients had hypomagnesaemia ( < 0·7 mmol/l) with normal muscle values (44 mmol/kg of fat-free solids), whereas low muscle values were found in 47 % of patients with normal serum Mg levels. As no significant correlation was observed between serum and muscle Mg, the authors concluded that serum Mg levels are of little value in the diagnosis of intracellular Mg deficits(Reference Fiaccadori, Del Canale, Coffrini, Melej, Vitali, Guariglia and Borghetti61).
Asthma
Skeletal muscle biopsies have been taken in asthmatics with and without oral β2-agonists and Mg concentrations compared with healthy subjects. Muscle Mg was lower in the asthmatics both with and without oral β2-agonists (3·62 ± 0·69 and 3·43 ± 0·60 v. 4·43 ± 0·74 mmol/100 g, respectively) while serum Mg was not different with the controls(Reference Gustafson, Boman, Rosenhall, Sandström and Wester62).
Low Mg diet in healthy subjects
A more recent study compared the effect of a low Mg intake (112 mg/day) for 92 days preceded by a control period of 35 days where they received a daily supplement of 200 mg and followed by the same supplementation for 49 days. During the restriction periods skeletal muscle Mg decreased significantly (53·4 ± 1·2 and 51·6 ± 1·3 v. 48·1 ± 1·3 mmol/kg dry weight), as did erythrocyte Mg (6·74 ± 0·08 and 6·68 ± 0·08 v. 5·91 ± 0·07 μmol/g haemoglobin) and Mg retention from Mg balance (+32 and +38 v. − 42 mg/day). There was a non-significant decrease of serum Mg during the restriction period confirming the lack of sensitivity of this measurement(Reference Lukaski and Nielsen16).
Healthy subjects
Measurements of muscle Mg were undertaken in biopsies taken from 49 individuals living in two Swedish cities with 5·7 and 1·7 mg Mg/litre in the local drinking water. There were significantly higher skeletal muscle Mg concentrations (4·1 ± 0·2 v. 3·9 ± 0·3 mmol/100 g fat-free dry weight) in subjects living in the area with the higher water Mg content. Dietary Mg intakes obtained from questionnaires were similar in the two groups, suggesting that muscle Mg content is a sensitive marker of Mg status(Reference Landin, Bonevik, Rylander and Sandström63). However, another study from this research group did not confirm these results(Reference Rubenowitz, Landin and Rylander55). No correlation between muscle and erythrocyte Mg was reported(Reference Ladefoged and Hagen64). However, overall these results suggest that measurements of muscle Mg may be considered to be a reliable indicator of Mg status(Reference Clague, Edwards and Jackson65).
In conclusion, more research is needed but muscle biopsies cannot provide a non-invasive method to assess Mg status.
Total, ionized Mg
Although Mg is either bound, in particular to protein, or free/ionized, only total Mg was able to be measured before ion-selective electrodes, fluorescent probes or nuclear magnetic resonance (NMR) were developed to identify and quantify this biological active form. Measurements of ionized Mg concentration were first undertaken in the 1970s on the axoplasm of squid axons(Reference Brinley and Scarpa66) and then on muscle(Reference Brinley, Scarpa and Tiffert67) and biological fluid(Reference Achilles, Scheidt, Hoppe and Cumme68). Errors resulting from changes in electrolyte composition, electrolyte interactions and interferences in the electrode response were investigated at that time(Reference Grima and Brand69). Many studies and publications only deal with the intra-method differences(Reference Niemela, Snader and Elin70, Reference Cecco, Hristova, Rehak and Elin71) in results, and errors in the analysis of ionized Mg(Reference Csako, Rehak and Elin72–Reference Huijgen, Sanders, Cecco, Rehak, Sanders and Elin75).
Diabetes
Low ionized Mg defined as serum concentrations lower that 0·46 mmol/l has been shown to be highly prevalent in diabetic subjects(Reference Corica, Corsonello, Ientile, Cucinotta, Di Benedettoet, Perticone, Dominguez and Barbagallo76).
Kidney disease
In haemodialysis patients, the values of ionized and total Mg in serum and mononuclear blood cells and total Mg in erythrocytes were shown to be significantly increased compared with a control population(Reference Huijgen, Sanders, van Olden, Klous, Gaffar and Sanders77). However, total serum Mg was not increased. In these patients, the two ionized Mg markers did not offer any advantages and total Mg concentration in serum remains the measurement of choice.
Critical illness
In critically ill children, ionized serum Mg was significantly lower than in a healthy group of children (0·37 ± 0·10 v. 0·46 ± 0·03 mmol/l). Many critically ill children exhibit ionized hypomagnesaemia with normal total serum Mg concentrations and these children would not be recognized as Mg-deficient based on routine total serum Mg tests(Reference Fiser, Torres, Butch and Valentine14).
Healthy subjects
In a study of healthy volunteers, the mean whole blood ionized Mg concentration was 0·52 mmol/l with a range of 0·44 to 0·59 mmol/l and an ionized Mg/total serum Mg ratio of 0·60 and a range of 0·50–0·69(Reference Greenway, Hindmarsh, Wang, Khodadeen and Hébert78). A positive correlation between ionized and serum total Mg was observed in 160 healthy children and the ratio was 58·3 ± 4·1 %(Reference Hoshino, Ogawa, Kitazawa, Nakamura and Uehara79). There appears to be no demonstrable advantage to measuring ionized Mg as opposed to total Mg for evaluating Mg status.
Fluorescent probes
Fluorescent probes have made it possible to measure cytosolic free Mg using a two-excitation wavelength fluorometer. The probe penetrates the plasma membrane as an ester that is hydrolyzed in the cytosol, and a microscope connected to the fluorometer is used to measure the fluorescence ratio at two excitation wavelengths in individual cells. The probe Mag-Furan-2 has been mainly used for measurements in platelets, whilst; Mag-Indo-1 has been used to assess cytosolic free Mg in mononuclear blood cells and erythrocytes. When fluorescent probes are used in ex vivo samples, the conservation of the sample is of major importance and validation of the method is necessary. Mag-Indo-1 has been used to determine the concentration of ionized Mg in lymphocytes of 29 migraine patients and 18 control subjects. The stability of lymphocytes in blood collected in either sodium citrate or heparin, during the isolation and staining process as well as the homogenous fluorescence distribution were tested to evaluate the reproducibility of the measurements. While there was no difference for total Mg in lymphocytes, significantly lower values were found for ionized Mg in migraine patients compared to controls, and after a 2 week Mg supplementation with 7·4–12 mmol Mg/d from mineral water (170–275 mg/d) the concentrations of ionized Mg in lymphocytes of migraine patients increased to those of the control subjects. In this study ionized intracellular Mg was a sensitive index to detect a deficit and the effect of a supplementation(Reference Thomas, Millot, Sebille, Delabroise, Thomas, Manfait and Arnaud31). Using the same fluorescent probe, the effect of Mg supplementation on healthy volunteers showed a significant increase in ionized Mg in lymphocytes after 2 days with unchanged total Mg in lymphocytes. After 4 days supplementation, ionized Mg returned to the initial value while total Mg concentrations in lymphocytes increased significantly(Reference Millot, Sebille, Beljebbar, Peirera, Delabroise, Sabatier, Caron, Manfait and Arnaud80). Intracellular ionized Mg concentrations have been measured in platelets using the fluorescent probe Furaptra(Reference Grima and Brand69).
Another promising technique is nuclear magnetic resonance (NMR). After blood sample measurements, the next consideration is non-invasive analysis of superficial body tissues, such as the skin and muscle. After the first determination of ionized Mg concentrations in human erythrocytes by 31P NMR spectroscopy(Reference Gupta, Benovic and Rose81), this technique was applied to measurements of intracellular erythrocyte Mg adenosine triphosphate and free Mg(Reference Bock, Wenz and Gupta25). Ionized Mg has also been analyzed in human blood plasma using 31P magnetic resonance spectroscopy with the addition of a ligand so that free and bound Mg have different resonances. The results showed that the magnetic resonance spectroscopy methods gave higher values for free ionized Mg than values obtained by ion-selective electrodes(Reference Huskens, Main, Malloy and Sherry82). With in vivo 31P NMR spectroscopy, intracellular free Mg in skeletal muscle and brain tissues was studied in 30 young volunteers after one month of daily supplementation with 12 mmol Mg (0·62 ± 0·05 v. 0·71 ± 0·03 mmol/l, respectively). Only urinary excretion was increased and the distribution of Mg in brain tissue, and muscle and also in serum and erythrocytes was unchanged in these healthy young subjects apparently without Mg deficiency(Reference Wary, Brillault-Salvat, Bloch, Leroy-Willig, Roumenov, Grognet, Leclerc and Carlier83). This non invasive technique had previously demonstrated similar intracellular free Mg concentrations in skeletal muscle and brain tissues, approximately 0·3 mmol(Reference Halvorson, Vande Linde, Helpern and Welch84). Skeletal muscle ionized Mg was measured with NMR in women over the course of one complete menstrual cycle. There was no evidence of a menstrual cycle effect on muscle ionized Mg or total Mg in serum, erythrocytes and mononuclear blood cells(Reference Rosenstein, Ryschon, Niemela, Elin, Balaban and Rubinow85). NMR brain and muscle in vivo measurement in children with migraine showed that brain intracellular ionized Mg concentrations were reduced by 25 % in patients; 0·139 ± 10 v. 0·186 ± 26 mmol, respectively(Reference Lodi, Montagna, Soriani, Iotti, Arnaldi, Cortelli, Pierangeli, Patuelli, Zaniol and Barbiroli86). Recently, a noninvasive intracellular technique for ionized Mg measurements has been developed using energy dispersive X-ray microanalysis(Reference Silver87).
In conclusion, these non-invasive methods are still in being developed and are not available to clinical laboratories, but in the future they may become the most efficient and accurate way to routinely measure intracellular ionized Mg.
Hair and nails
Hair
The potential use of hair to assess Mg status is an attractive idea as it is the least invasive sampling procedure, and samples can be taken over a long period of time and can be stored until the analysis can be performed.
Cattle and animal studies
It has been reported that cattle suffering from grass tetany have blood serum Mg concentrations below 10 mg/litre compared with a mean normal value of 21 mg/l(Reference Combs, Goodrich and Meiske88); a higher Mg content was found in the hair of cattle when diet was supplemented with Mg. However, no difference was found in hair analyzed five times during the year (to determine seasonal effects) or in the Mg content of hair from cows with grass tetany(Reference Hall, Sanders, Bell and Reynolds89). In rats fed diets containing either 82 or 17 mg/100 g, significant changes in the Mg contents of various hair fractions were observed after 2 or 3 months(Reference Brochart90).
Methods
As hair can be contaminated by the environment, it is necessary to wash it before analysis. The effects of washing hair on its Mg content have been studied. The methods selected were a detergent wash, a hexane-ethanol wash and an acetone-ether detergent wash. There was a significant difference between various washing procedures, but it was shown that the Mg content was less than half that of the unwashed hair. These results demonstrate that Mg in hair is an unreliable sample to assess Mg status(Reference Assarian and Oberleas91).
Human
In end stage heart disease patients with significantly lower plasma and erythrocyte Mg levels, a lower Mg content was also found in hair (26·7 ± 15·3 μg/g) compared with healthy control subjects (54·5 ± 19·8 μg/g). The authors explained this exceptionally low content in hair by a chronic disturbance of Mg metabolism(Reference Gajos, Solarska, Siembab, Nessler, Nessler, Stepniewski and Piwowarska21). Hair Mg concentrations in patients with Fibromyalgia were significantly higher compared with those of healthy subjects (84·7 ± 73·3 v. 46·8 ± 28·9 μg/g). As no other biomarker of Mg status was measured and no quality control procedures described, these results must be interpreted with caution(Reference Ng92). In a 2-month-old boy with congenital hypomagnesaemia, hair Mg concentrations were measured at 8 and 12 years of age and compared with control healthy children with the same brown hair colour. Surprisingly, the Mg content of hair at 8 and 12 years was higher compared with controls: 164 ± 8·08 v. 52·7 ± 21·7 and 244 ± 11·5 v. 121·6 ± 27·2 μg/g dry weight, respectively. Other trace element concentrations such as manganese, copper and zinc were also higher in hair. The authors concluded that hair acts as a sink for oligo elements, including Mg and did not reflect the severe deficiency as seen through the significantly low values for serum, erythrocyte, lymphocyte and teeth(Reference Guillard, Mettey, Lecron and Pineau58). Analysis of hair and serum Mg in neonates and their mothers showed a negative correlation between maternal hair and ionized Mg in cord serum, and male neonates had higher levels of Mg in cord blood and hair than females(Reference Kozielec, Durska, Karakiewicz and Kędzierska93). These researchers also investigated the effect of two Mg multivitamin supplements given for 3 months (each tablet contained 24 mg Mg) and 4 months (tablets containing 100 mg Mg) at a dose of 7 mg Mg per kg body weight and per day in 46 children aged 2–6 years. The results showed that hair Mg concentrations increased significantly from 7·74 ± 0·36 to 11·03 ± 0·89 μg/g dry mass. This increase was observed in 40 children out of the 46 recruited, suggesting that 3 months Mg supplementation was effective for increasing Mg hair concentrations in children(Reference Kozielec, Salacka, Radomska, Strecker and Durska94).
Nails
The mean loss of nail substance is approximately 3 g per year and the ratio between calcium and Mg is about 4·5/1 with some variability due to external adsorption(Reference Vellar95). The reproducibility over a 6-year period of the measurement in toenails of 16 trace elements, including Mg has been reported from 127 women in the United States. Toenail concentrations of some minerals can be used as biomarkers of exposure and a single sample may represent long-term exposure for toxic intakes of nutrients. For Mg, the mean value ± sd was 167 ± 130 μg/g and was fourth most abundant after sulphur, calcium and chlorine(Reference Garland, Morris, Rosner, Stampfer, Spate, Baskett, Willett and Hunter96).
In conclusion, from these studies, there is a need to validate the measurement of both hair and nail Mg and to demonstrate that the sample represents a period of either deficient or excessive Mg intake. Up to now, it is unclear how to interpret Mg values found in hair or nail.
Physiological tests
Loading test
Balance studies are time consuming, labour intensive and need well trained staff. They are often performed in a metabolic unit and require complete urine and faecal collections; therefore it is not a method that can be applied as a routine test for the evaluation of Mg status. Loading tests are simplified balance studies where absorption is supposed not to be disturbed when Mg is given orally so that body retention is calculated from urine elimination. Mg administration during a loading test can be either oral or intravenous and it is important that the subjects have normal kidney function. Urine is collected for 24 hours following administration of the Mg load as Mg excretion by the kidney has been shown to have a circadian rhythm(Reference Graham, Caesar and Burgen97). Under these conditions, the loading test is supposed to be a reliable indicator of Mg status(Reference Clague, Edwards and Jackson65).
Alcoholics
Retention of a low dose of Mg, 0·2 mEq/kg lean body weight, given intravenously has been assessed in hypomagnesaemia patients and normomagnesaemic alcoholics. It was shown that they retained significantly higher amounts of the Mg load than normal subjects. After parenteral Mg repletion, the retention was normalized showing that this test is a more sensitive index of Mg deficiency than serum concentration(Reference Ryzen, Elbaum, Singer and Rude98).
Crohn's and coeliac disease
In Crohn's disease patients, the most sensitive measurement of Mg status and deficiency has been shown to be the analysis of Mg in muscle and also the estimation of Mg retention(Reference Sjögren, Florén and Nilsson22). A 12-hour intravenous Mg loading test (30 mmol/1·73 m2) was used in children and adolescents with coeliac disease to evaluate the frequency of Mg deficiency. The cut-off level for tissue Mg deficiency was identified at the point when Mg retention was greater than 40 % of the load(Reference Rujner, Socha, Syczewska, Wojtasik, Kunachowicz and Stolarczyk99).
Critical illness
A study was performed to assess the value of intracellular Mg in erythrocytes and mononuclear blood cells in critically ill patients sub-divided into Mg depleted and non-depleted groups according to their response to a loading test. There were no significant difference between the Mg depleted and non-depleted groups (plasma 0·81 and 0·90 mmol/l and red blood cell 2·34 and 2·18 mmol/1, mononuclear blood cell 25·16 and 18·1 mmol/kg dry weight, respectively). Thus, normal values of plasma, erythrocyte or mononuclear blood cell concentrations of Mg cannot exclude Mg depletion(Reference Arnold, Tovey, Mangat, Penny and Jacobs13). Mg deficiency was identified in critically ill patients using a loading test and was validated using measurements of serum ionized Mg. There was a significant increase in both serum ionized and total Mg concentrations by 43 % and 59 %, respectively on day 1 compared with the control group. Urinary Mg excretion also increased after a load of 30 mmol from 4·8 ± 2·3 mmol/day during the 3-day study period to 22·7 ± 10·9 mmol/day. The patients with an excretion lower than 70 % of the total Mg were designated as functionally Mg-deficient retainers and patients who excreted more than 70 % were non-retainers. Interestingly, the number of retainers on day 2 was ten patients and only six on day 3, indicating a replenishment of body Mg stores. In the retainer group, only two patients had a low serum ionized Mg concentration, while two other patients had low total serum Mg values. These results show that the Mg-loading test is effective and serum ionized Mg appears to be an insensitive biochemical marker of functional hypomagnesemia(Reference Hébert, Mehta, Wang, Hindmarsh, Jones and Cardinal100).
Chronic fatigue syndrome
In a study population of 93 patients with unexplained chronic fatigue, only three subjects had plasma Mg concentrations lower than 0·6 mmol/l (0·82 ± 0·10 mmol/l) and normal erythrocyte Mg (1·97 ± 0·22 mmol/l). The Mg deficient group was identified from Mg retentions following an intravenous loading test with a four hour Mg infusion of 0·2 mEq/keg body weight. Patients with 20 % or more Mg retention were diagnosed as moderately deficient and those with 50 % or more as severely deficient. When 20 % retention was taken as the cut-off value, 47 % of the patients were classified as being Mg deficient(Reference Manuel y Keenoy, Moorkens, Vertommen, Noe, Nève and De Leeuw101).
Asthmatic
In asthmatics, a significantly increased retention of Mg was observed in 58·9 % of the patients after a loading test compared with 8·9 % in normal subjects(Reference Emelyanov, Fedoseev and Barnes15).
Renal transplant
A one-hour intramuscular or intravenous infusion of 0·1 mmol Mg per kg body weight over 1 hour followed by a 24 h urine collection was designed for outpatients. Serum and urinary Mg were analyzed and the percentage retention before and after 4 months daily supplementation containing 5 mmol Mg showed a significant decrease from 47 ± 43 in patients after renal transplantation to 16 ± 26. Thus, Mg supplementation successfully returned the percentage retention towards normal values after 4 months. In the placebo group of patients after renal transplantation, the percentage retention was 58 ± 27. Retention of 20 % of the dose or more was considered evidence of deficiency. These high levels of retention are greater than the mean ± 3sd of the control group and thus indicate Mg deficiency in spite of normal serum Mg level. The analysis of bone samples obtained from another group of patients undergoing hip replacement showed that short term Mg retention reflects femur-Mg content, the most relevant Mg store. In this study(Reference Rob, Dick, Bley, Seyfert, Brinckmann, Höllriegel, Friedrich, Dibbelt and Seelig102), dietary intake, faecal excretion and basal urinary Mg output were ignored and may explain the negative percentage retention values in healthy individuals. These data are however comparable with those obtained with 30 mmol (810 mg) Mg infused over 8 hours(Reference Gullestad, Midtvedt, Dolva, Norseth and Kjekshus103).
Low Mg diet in healthy subjects
Recently, a double blind crossover study evaluated the effect of moderate Mg deprivation in postmenopausal women receiving either 4·4 mmol (107 mg) Mg from a basal diet or 13·45 mmol (327 mg) with a Mg supplement of 9·5 mmol (220 mg) added for 72 days. Mg deprivation significantly reduced the positive Mg balance when the supplement was given, decreased red blood cell membrane Mg, increased the calcium balance, decreased the faecal excretion of phosphorus and increased its urinary excretion, and decreased the urinary excretion of potassium. The authors suggest that a non-positive Mg balance and decreased red blood cell membrane concentration may be indicators of Mg deprivation(Reference Nielsen, Milne, Gallagher, Johnson and Hoverson104).
Healthy subjects
Mg status has been measured in healthy subjects by either oral or infused Mg loading tests and deficiencies have been diagnosed with this test, while serum Mg was not affected(Reference Holm, Jepsen, Sjogaard and Hessov105–Reference Goto, Yasue, Okumura, Matsuyama, Kugiyama, Miyagi and Hijashi107). In one study urine was collected for 24 hours after an 8-hour infusion of 30 mmol Mg. The data demonstrated a Mg retention of 28 and 6 % of the dose in elderly and younger subjects respectively, suggesting a higher prevalence of Mg deficiency in the elderly. It was concluded that a significant sub-clinical Mg deficit was present in these healthy elderly subjects that was not detected by serum Mg(Reference Gullestad, Nes, Ronneberg, Midtvedt, Falch and Kjekshus108). Their study also showed that a 3-week daily oral Mg supplementation with 9 mmol Mg improved Mg status as shown by increased urinary excretion and lower body retention However, a parenteral loading test using Mg chloride (0·206 mmol/kg body weight) did not show any correlation between Mg retention and basal urinary excretion of Mg and plasma or erythrocyte Mg concentrations. This study raises the important point of the need to standardize the loading test procedure in order to produce the most sensitive and reproducible results(Reference Mazur, Felgines, Feillet, Boirie, Bellanger, Beaufrère, Gueux, Rock and Rayssiguier109). The impact on Mg status of consuming drinking water of differing Mg content (1·6 mg and 25 mg/l) for 6 weeks has been evaluated using an oral Mg loading test (575 mg of Mg) administered in tablet form. The 24-hour urinary excretion of Mg was expressed as total Mg excretion and the Mg/creatinine ratio. There was no change when the urinary excretion was expressed as total Mg excretion but a significant change from 24·8 to 39·3 Mg/creatinine ratio was found suggesting that a small increase of Mg concentration in drinking water for 6 weeks can improve body Mg status(Reference Rubenowitz, Axelsson and Rylander110).
Although there is no standardized protocol and the relative Mg deficits identified through its use may not represent the total body Mg deficit, this test has been useful and could be improved in the near future.
Isotope balance studies
The first analytical challenge when using stable isotopes to study the metabolic fate of Mg was the introduction of samples into a mass spectrometer. 26Mg was chelated with tetramethylheptanedione, extracted and recovered by sublimation and introduced by solid probe into Mass Spectrometer for analysis at enrichment levels expected to be found in plasma, urine and faecal samples from subjects who had received this isotope as a tracer(Reference Schwartz and Giesecke111). The accuracy of these analyses has continuously improved and allows a precise evaluation of intestinal Mg absorption, faecal excretion, and body retention. In addition, using kinetic data obtained from the analysis of blood samples it is now possible to undertake compartmental modelling to determine pool sizes and turnover rates. Stable isotopes are routinely used to study gastrointestinal functions but for Mg, the availability and the cost of mass spectrometry measurements, the cost of the isotope and the rather complex protocol restrict the use of this method to research and not clinical studies. Among the studies published, stable isotopes have been used to evaluate the exchangeable Mg pool size in humans and to correlate these changes with Mg status. However, an 8-week Mg supplementation study of 366 mg Mg per day in 24 year old healthy women did not show any modification in the size of the exchangeable Mg pools, but plasma ionized Mg and urinary excretion were significantly increased, while total plasma and erythrocyte levels were unchanged. In these healthy subjects, and under these experimental conditions, the study of exchangeable Mg pool size is not a sensitive biomarker of the variations of Mg status(Reference Feillet-Coudray, Coudray, Tressol, Pépin, Mazur, Abrams and Rayssiguier112). To detect marginal Mg deficiency, an 11 ng dose of 26Mg was injected into 22 healthy subjects with a wide range of plasma Mg concentrations from 0·68 to 0·95 mmol/l (mean value 0·82 ± 0·09 mmol/l) and with adequate Mg intakes (438 ± and 464 ± 138 mg for men and women, respectively). This modified version of the loading test and Mg retention showed no correlation between the excretion of the isotopic label and muscle Mg concentration. Within 24 hours only 7·9 % of the injected dose was excreted in the urine and the fraction excreted correlated with total urinary Mg excretion. In contrast to other loading tests, this dose was apparently insufficient to modify the size of the pools in subjects with marginal deficiency as defined by a range of Mg muscle concentrations from 3·50 to 4·19 mmol/100 g fat free dried solids(Reference Wälti, Walcyyk, Wimmermann, Fortunato, Weber, Spinas and Hurrell113). Using double labelled Mg in healthy adult men, 26Mg given orally and 25Mg injected intravenously, blood, urine and faeces were collected for 12 days to build a compartmental model of Mg kinetics. However, this analysis only enables the exploration of 25 % of the total body pool i.e. that which exchanges rapidly from the plasma compartment with two extra-plasma pools(Reference Sabatier, Pont, Arnaud and Turnlund10).
Physiological activities
Biomarkers such as ferritin for iron have not been found for Mg. It has been suggested that potential Mg markers could include Na/K ATPase, thromboxane B2, C-reactive protein and endothelin-1, but other biomarkers are needed(Reference Franz3). Because of the ubiquitous role of Mg and the interactions of other minerals such as calcium, the identification of a specific marker of Mg deficiency is challenging. Mg has been shown to be associated with various physiological responses such as blood pressure, but in this case known parameters, such as calcium, and additional unknown factors will also affect blood pressure. Therefore, the relationship between intracellular free Mg and diastolic blood pressure cannot be a marker of Mg status(Reference Resnick, Gupta and Laragh114). The activities of alkaline phosphatase and creatine kinase, two Mg-requiring enzymes, have been evaluated in relation to plasma and erythrocyte Mg concentration in rats to determine their usefulness as indices of Mg status(Reference Fischer and Giroux115). The results showed that plasma Mg concentration is the most useful indicator of Mg status. A very strong correlation was observed between plasma and bone concentrations. Other sensitive markers of deficiency may be developed through the study of gene expression, which was shown in animal experiments to change with Mg deficiency(Reference Petrault, Zimowska, Mathieu, Bayle, Rock, Favier, Rayssiguier and Mazur116).
Ex vivo or in vitro cellular methods
A new in vitro blood load test has been proposed to assess Mg status using stable isotopes. Blood cells were isolated and incubated with 25Mg and it was proposed that high uptake would be triggered by Mg deficiency. The uptake in human erythrocytes was low compared to rat erythrocytes and higher enrichments were obtained for human lymphocytes and platelets. Thus, these latter cells seem more appropriate to test human Mg status using this in vitro system(Reference Feillet-Coudray, Coudray, Gueux, Mazur and Rayssiguier117). In vitro erythrocyte Mg fluxes were studied using stable isotopes in mice receiving Mg deficient diet and subsequently selected for their high or low erythrocyte Mg concentrations(Reference Feuillet-Coudray, Trzeciakiewicz, Coudray, Rambeau, Chanson, Rayssiguier, Opolski, Wolf and Mazur38). Although it is not possible to extrapolate the results obtained with severe Mg restriction in animals to the human situation, these in vitro studies can help to understand the role of TRPM channels in the specific regulation of intracellular Mg concentrations in blood cells and tissues. However, the authors correctly indicate that artificial erythrocyte Mg load tests performed in vitro may induce non-physiological cellular responses so any conclusions must be drawn with caution.
TRPM (Transient Receptor Potential Melastatin) Channel
Hypomagnesaemia, first described in 1968(Reference Paunier, Radde, Kooh, Conen and Fraser118) is an autosomal-recessive disorder of early infancy resulting in convulsions, muscle spasms or tetany. Extremely low serum Mg and low calcium levels are present, but the administration of high doses of Mg prevents permanent neurological damage and death. It was originally shown that the primary defect involved intestinal Mg absorption. TRPM6 protein has been identified and exhibits homology to TRPM7, which has been characterized as a calcium- and Mg-permeable ion channel regulated by Mg-ATP. The distribution of TRPM6 along the entire small intestine and colon as well as in distal tubule cells in the kidney shows how both the absorption and the excretion of Mg are controlled by this new family of cation channels. TRPM7 exhibits significant permeation to ionized Mg and is inhibited by cytosolic Mg ions and Mg-ATP. This recent progress in epithelial Mg transport will certainly help in understanding the fine tuning of Mg homeostasis and its impact on disturbed Mg status(Reference Chubanov, Waldegger, Mederos y Schnitzler, Vitzthum, Sassen, Seyberth, Konrad and Gudermann119–Reference Voets, Nilius, Hoefs, van der Kemp, Droogmans, Bindels and Hoenderop121). The recent identification of TRPM channels and their role in hypomagnesaemia may eventually also result in chronic and mild Mg deficiency being explained by these specific transporters. It may also be possible in the future to accurately evaluate the transport capacity of Mg through these channels as a diagnostic of the risk for Mg deficiency.
Conclusions
A similar review of tests used to assess Mg status written 17 years ago concluded that there was no test that could readily be used in clinical medicine to assess the total body Mg status of a patient. Although intense research activities have been dedicated to Mg, the difficulties of accessing total body Mg, and its main two compartments, namely bone and muscle, mean that today there is still no simple, rapid, and accurate laboratory test to indicate total body Mg status in human. However, taking into account all the more recent investigations, although serum Mg < 0·75 mmol/l still remains a useful measurement for severe deficiency, for values between 0·75 and 0·85 mmol/l, a loading test must be performed to identify the deficient subjects. Loading tests appear to be the gold standard for Mg status but patients with disturbed kidney and intestinal functions should be excluded when the dose is given orally. There is also a need to reach a consensus on a standardized protocol to be used in order to compare results from different clinical units. Urinary Mg cannot replace the loading test as it does not reflect Mg status. Other cellular Mg measurements, such as total or ionized Mg are often equivocal and more research and systematic evaluations are needed. Muscle Mg appears to be a good marker but biopsies limit the use of this measurement to research. Bone, the most important Mg compartment cannot be used because of the invasiveness of the sampling and variations in measured Mg concentrations. The development of new and non invasive techniques such as NMR or ex vivo studies could in the future provide valuable tools for performing routine analyses of ionized Mg in tissues. With the development of molecular biology approaches and the recent discovery of TRPM channels, new, sensitive and fast evaluation of Mg status in humans may be developed in the near future.
