Arginine flux and intravascular nitric oxide synthesis in severe childhood undernutrition

¹ From the US Department of Agriculture/Agricultural Research Service, Children's Nutrition Research Center, Department of Pediatrics, Baylor College of Medicine, Houston, TX (FJ and SV), and the Tropical Metabolism Research Unit, Tropical Medicine Research Institute, University of the West Indies, Mona, Kingston, Jamaica (AB, MR, and TF)

² Supported by NIH grant 2RO1 DK 056689 and by federal funds from the US Department of Agriculture/Agricultural Research Service under Cooperative Agreement Number 58-6250-6001.

³ Address reprint requests to F Jahoor, Children's Nutrition Research Center, Department of Pediatrics, Baylor College of Medicine, 1100 Bates Street, Houston, TX 77030-2600. E-mail: fjahoor{at}bcm.tmc.edu.

ABSTRACT  
Background: Although nutritionally dispensable amino acids are not essential in the diet, adequate synthesis is necessary for maintenance of good health. Whereas children with edematous severe childhood undernutrition (SCU) can maintain production rates of glycine and serine despite a slower body protein breakdown rate, it is unknown whether the same is true for the semidispensable amino acid arginine.

Objective: We aimed to measure arginine flux and intravascular nitric oxide synthesis in children with SCU.

Design: Arginine flux and the fractional and absolute synthesis rates of plasma nitrite plus nitrate were measured postabsorptively by using a 6-h infusion of [¹⁵N₂]-arginine in 2 groups of children with edematous (n = 14) or nonedematous (n = 7) SCU when they were infected and malnourished (postadmission day 3; clinical phase 1), when they were no longer infected (postadmission day 15; clinical phase 2), and when they were recovered (postadmission day 55; clinical phase 3).

Results: Arginine flux was slower (P < 0.01) and plasma arginine concentrations were lower in the edematous group than in the nonedematous group at clinical phase 1. At clinical phase 2, flux doubled to a value that was not significantly different from the value at clinical phase 3. There were no significant differences in the plasma concentration or fractional or absolute synthesis rate of plasma nitrite plus nitrate between the groups at any clinical phase and among clinical phases within each group.

Conclusion: Whereas children with nonedematous SCU can maintain arginine flux at the same rate as when recovered, children with edematous SCU cannot. The slower arginine flux was not, however, associated with slower nitric oxide synthesis.

Key Words: Arginine kinetics • nitric oxide • edematous severe childhood undernutrition • nonedematous severe childhood undernutrition • marasmus • kwashiorkor

INTRODUCTION  
In severe childhood undernutrition (SCU), less amino acids of dietary origin are available for metabolic purposes because of the reduced protein intake associated with chronic food deprivation. Because breakdown of body proteins is the major contributor to the overall flux of amino acids (1), our finding that the whole-body protein breakdown rate is slower in children with the edematous form of SCU than in children with the nonedematous form of SCU (2) suggests that a more severe shortage in the availability of dietary essential amino acids will exist in children with edematous SCU. In the case of the dietary nonessential amino acids, it is possible for de novo synthesis to fill the gap created by any reduction in the amount released from protein breakdown. Whereas we found this to be true for glycine and serine (3, 4), this was not the case for cysteine, which requires sulfur from the essential amino acid methionine for its synthesis (5). These findings suggest that in children with edematous SCU, the availability of a particular nonessential amino acid may depend on whether it is derived from an essential amino acid or not.

An amino acid that occupies a unique position is arginine, a semi-essential amino acid, because it is produced in adequate quantities to meet the requirements of healthy adults but not of growing children or persons stressed by severe trauma or sepsis (6). Because the production of arginine depends on the availability of citrulline, a nonprotein amino acid synthesized mostly in the gut from glutamine and proline, it is highly likely that arginine will be in short supply in children with SCU, especially those with edematous SCU because of decreased release from a slower protein breakdown (2).

In addition to its role as a substrate for protein synthesis, arginine serves several important physiologic functions related to ammonia detoxification to urea, wound healing, and immune function. As a precursor for the synthesis of creatine and nitric oxide (NO), arginine serves important roles in energy metabolism and regulation of blood pressure plus the numerous other regulatory roles attributed to NO (6). Low plasma arginine has been shown to be correlated with a worse prognosis in septic patients (6), which suggests that there may be an overall increase in the requirement for arginine that is not met by endogenous production in stressed states. The situation will be worse in severely malnourished children stressed by concurrent infections. In a recent study, we reported a marked reduction in plasma arginine concentrations in children with edematous SCU compared with the value of their nonedematous counterparts and the value at recovery (3), which suggests that its availability as a precursor molecule in anabolic reactions may be limiting. This does not seem to be the case with respect to NO synthesis, however, because Fechner et al (7) reported that children with edematous SCU have plasma nitrite plus nitrate (NOx) concentrations, products of NO metabolism, that are twice those of children with nonedematous SCU. Together, these observations suggest an increased conversion of arginine to NO in the face of decreased arginine availability in children with edematous SCU. Hence, we propose to test the hypothesis that at the time they are admitted to the hospital for treatment, children with edematous SCU will have a slower arginine flux but faster NO synthesis than do children with nonedematous SCU. We also hypothesized that arginine flux would be slower in the malnourished state than in the recovered state in the edematous group but not in the nonedematous group.

SUBJECTS AND METHODS  
Subjects
Twenty-one children who were admitted to the Tropical Metabolism Research Unit (TMRU), University of the West Indies, for treatment of SCU participated in the study. During their hospitalization, the children were managed according to a standard treatment protocol as previously described by us (2-5). As shown in Table 1, each subject had a deficit in body weight–for-age of >20%, which indicated severe undernutrition. Seven of the children had nonedematous and 14 had edematous SCU; 7 had kwashiorkor and 7 had marasmic kwashiorkor (Table 1 and Table 2). The diagnosis of type of SCU, ie, marasmus, kwashiorkor, or marasmic kwashiorkor, was based on the Wellcome Classification (8). All except one of the children had evidence of one or more infections at admission.

View this table:
TABLE 1. Age and physical characteristics of the subjects¹

 

View this table:
TABLE 2. Clinical characteristics of the subjects at admission¹

 
This study was approved by the Medical Ethics Committee of the University Hospital of the West Indies and the Baylor Affiliates Review Board for Human Subject Research of Baylor College of Medicine. Written informed consent was obtained from at least one parent of each child before enrollment.

Study design
Arginine and NO kinetics were measured 3 times during the children's hospitalization by using a constant intravenous infusion of [¹⁵N₂]-guanidino arginine at 3 d after admission when the subjects were both infected and undernourished but clinically stable as indicated by blood pressure, pulse, and respiration rates (clinical phase 1); at 15 d after admission when the subjects were still severely undernourished (anthropometrically) but no longer infected (ie, all clinical features of the infective episode had resolved), they had lost edema, and had improved affect and appetite (clinical phase 2); and at 55 d after admission when the rate of catch-up growth had reached a plateau and weight-for-length was 90% of expected (clinical phase 3).

Treatment and diets
During their hospitalization, the children were managed according to a standard protocol that divided their treatment into phases. The acute resuscitation and maintenance phases of treatment extended from admission until the child's appetite returned, infection was cleared, and edema was lost in those children with the edematous forms of severe undernutrition. The mean duration of this phase was 15.4 d. During this period, fluid and electrolyte imbalances were first corrected, and infections were treated with broad-spectrum antibiotics, usually parenteral penicillin and gentamicin, plus oral metronidazole. The children were fed a resuscitative diet made by using a commercial milk powder (61 g Nan; Nestlé SA, Vevey, Switzerland), 81 g glucose, and 858 g water. The energy content of the feed was 2633 kJ/kg with a macronutrient composition per kg feed of 7.6 g protein, 14.6 g lipid, and 116.7 g carbohydrate. The energy distribution of the feed was 74% from carbohydrate, 21% from fat, and 5% from protein. The amount offered aimed to provide 418 kJ · kg^–1 · d^–1 and 1.2 g protein · kg^–1 · d^–1. The feed was given as boluses every 3 h throughout the day or as smaller 2-hourly boluses when the child was having problems tolerating the feed.

The next phase in the clinical care of the children was the rapid catch-up growth phase. In this phase of treatment, the children were fed an energy-dense, milk-based formula until their growth rate plateaued and weight-for-length was 90% of expected. The high-energy feed given during rapid catch-up growth was made from the same commercial milk powder (NAN; Nestlé SA) plus additional carbohydrate. The energy content was 6071 kJ/kg, and the macronutrient composition per kg feed was 64.8 g lipid, 183 g carbohydrate, and 33.75 g protein. The energy distribution of the feed was 40.2% from fat, 50.5% from carbohydrate, and 9.3% from protein. The children were fed every 4 h ad libitum. During this phase, energy intake may have been as high as 626–750 kJ · kg^–1 · d^–1 and protein as high as 3.5–4 g · kg^–1 · d^–1.

In addition, both diets were supplemented with vitamins (Tropivite; Federated Pharmaceuticals, Kingston, Jamaica) and a mineral mix prepared in the TMRU metabolic kitchen. Each child received 2 mL/d of the vitamin solution, which contained 6000 IU vitamin A (palmitate), 1600 IU vitamin D (calciferol), 2 mg thiamine HCL, 3.2 mg riboflavin, 120 mg vitamin C (ascorbic acid), 4 mg vitamin B-6 (B-6 HCL), and 28 mg nicotinamide. They also received 5 mg folic acid/d and 2 mL of the mineral mix · kg^–1 · d^–1. The mineral mix consisted of 37.28 g KCl + 50.84 mg MgCl₂O.6H₂O + 3.36 g (CH₃COO)₂Zn.2H₂O/L H₂O (BDH Chemicals, Poole, United Kingdom). During the rapid catch-up growth phase but not in the maintenance phase, the children also received 60 mg FeSO₄. Weight and length were monitored throughout the hospitalization, the former daily with an electronic balance (model F150S; Sartorius, Göttingen, Germany) and the latter weekly with a horizontally mounted stadiometer (Holtain Ltd, Crymych, United Kingdom).

Experimental protocol
The subjects were studied in the postabsorptive state. The isotope infusion started 3 h after the subject's last bolus of feed, and the first blood sample used to measure arginine and NO kinetics was taken 4 h after the infusion started. Hence, arginine kinetics were calculated by using isotopic data obtained from blood samples taken during the 7–9-h postabsorptive period. To avoid possible hypoglycemia during the experimental period, a 0.278-mol/L glucose solution was infused intravenously at 3 mg · kg^–1 · min^–1 starting 1 h after the last bolus feed, that is, 2 h before the isotope infusion started.

The rate of synthesis of NO from arginine in the plasma compartment was estimated by determining the fractional rate of synthesis of plasma NOx from arginine and the concentration of NOx in plasma. This approach is feasible on the basis of the fact that ¹⁵N -labeled nitric oxide is a product of the reaction in which [¹⁵N₂]-guanidino arginine is converted to [¹⁵N-ureido]citrulline (9). Hence, [¹⁵N₂]-guanidino arginine is the only possible precursor of ¹⁵N-labeled NO. Also, ¹⁵N-labeled NOx can only be made from ¹⁵N-labeled NO. Therefore, ¹⁵NOx is an excellent surrogate measurement of ¹⁵NO. By administering [¹⁵N₂]-guanidino arginine and measuring the isotopic enrichment of plasma arginine at steady state (precursor pool) and the plasma isotopic enrichment of NOx (product), one can calculate the fractional rate of synthesis of nitrite and nitrate, and hence NO, by using the standard precursor-product equation.

Infusion protocol
Sterile solutions of [¹⁵N₂]-guanidino arginine (98%; Cambridge Isotope Laboratories, Woburn, MA) were prepared in 9 g NaCl/L. Two intravenous access sites were established in opposite arms by the insertion of 24 G catheters after preparation of the access sites with a topical anesthetic (EMLA cream; Astra Pharmaceuticals Ltd, Langley, United Kingdom). One intravenous catheter was used for infusion of the labeled arginine and the other for blood sampling.

After 2 h of continuous glucose infusion, a 3-mL blood sample was drawn for baseline measurements, and a bolus injection of 100 mg of ²H₂O/kg (99.9%; Cambridge Isotope Laboratories, Woburn, MA) was given intravenously. This was immediately followed by administration of a priming dose of 5 µmol/kg of [¹⁵N₂]-guanidino arginine, which was followed immediately by a continuous infusion at 5 µmol · kg^–1 · h^–1 for 6 h. One-milliliter blood samples were drawn hourly for the first 3 h, and 3-mL blood samples were drawn hourly from hours 3 to 6 of the infusion. The infusion and blood sampling protocols were the same for the 2 subsequent experiments performed at clinical phases 2 and 3.

Sample analyses
The blood samples were centrifuged immediately at 1000 x g for 15 min at 4 °C, and the plasma was removed and stored immediately at –70 °C for later analyses. The isotopic enrichment of plasma arginine was measured by negative chemical ionization gas chromatography–mass spectrometry (NCI GC-MS) with a Hewlett-Packard HP 5989B quadrupole mass spectrometer (Palo Alto, CA). Plasma arginine was extracted by cation-exchange chromatography, and the trifluoroacetyl ester derivative was prepared by adding 0.4 mL of a 4:1 mixture of dichloroethane:trifluoroacetic anhydride to the dried eluant containing arginine and heating at 100 °C for 2 h. Isotope ratios were measured by selectively monitoring ions at mass-to-charge (m/z) ratios of 444 to 446.

The isotopic enrichment of plasma NOx was also determined by NCI GC-MS with the method described by Tsikas (10). Briefly, the nitrate in 0.2 mL of plasma was reduced to nitrite by adding 25 mg cadmium. The mixture was acidified with 0.1 mL of 20% acetic acid and shaken at room temperature for 15 min to reduce nitrate to nitrite. After centrifugation at 1000 x g for 5 min, the supernatant fluid was removed, further extracted with acetone, and the nitrite converted to its pentafluorobenzyl derivative by adding 25 µL of 2,3,4,5,6-pentafluorobenzyl bromide and heating at 50 °C for 1 h. The isotope ratio was measured by selectively monitoring ions at m/z ratios 46 to 47. The ²H₂ content of plasma water was determined by reducing water extracted from 10 µL of plasma with zinc in quartz vessels and determining the ²H₂ abundance of the resulting hydrogen gas by gas isotope-ratio mass spectrometry (-E; Finnigan MAT, San Jose, CA).

Plasma arginine concentrations were measured by standard ion-exchange chromatography. Plasma NOx concentrations were measured by in vitro isotope dilution as described by Tsikas (10). Briefly, 0.2 mL of the baseline plasma sample was spiked with a known quantity of Na¹⁵NO₃, the internal standard, the nitrate was reduced to nitrite, and the isotopic enrichment of the nitrite was measured as described above.

Calculations
As shown in Figure 1, the isotopic enrichment of arginine reached a plateau in plasma in all 3 groups of subjects during the final 3 h of the infusion, which permitted use of the steady state equation described below to calculate flux.

View larger version (20K):
FIGURE 1.. Mean (±SEM) isotopic enrichment (IE) of plasma arginine and nitrite plus nitrate (NOx) during a primed-constant intravenous infusion of [¹⁵N₂]-guanidino arginine in 7 children with marasmus (•), 7 with marasmic kwashiorkor (), and 7 with kwashiorkor () at clinical phase 1, 2 d after admission when the subjects were infected and malnourished.

 
Arginine flux was calculated from the equation

RESULTS  
There were significant differences in the subjects' mean ages at the time they participated in the clinical phase 1 experiment, with the mean age of the marasmic kwashiorkor group being the lowest. The children had a mean weight-for-age between 46% of expected in the marasmic kwashiorkor subjects to 71% in the kwashiorkor subjects (Table 1). As expected in the undernourished state, there were significant differences in weight-for-age and weight-for-length by diagnosis. At clinical phase 1, the marasmic and marasmic kwashiorkor children were lighter than the children with kwashiorkor. This difference persisted at clinical phase 2 between the marasmic kwashiorkor and kwashiorkor groups. When the children had recovered at clinical phase 3, all anthropometric measurements except for length increased significantly compared with the values at clinical phase 1.

The clinical characteristics of the subjects at admission are shown in Table 2. All subjects but one were anemic. Twenty of the 21 subjects had one or more infections, but white blood cells were elevated in only 14 subjects. Mean hemoglobin and white blood cell concentrations were significantly higher (P < 0.05) in the nonedematous group than in the edematous group.

There was a significant interaction between clinical phase (nutritional status) and diagnosis. Thus, the mean arginine flux of all subjects was significantly slower (P < 0.001) at clinical phase 1 than at clinical phases 2 and 3 (Table 3). This slower arginine flux at clinical phase 1 was associated with a markedly lower plasma arginine concentration (P < 0.001) compared with the values at clinical phases 2 and 3. At clinical phase 2, arginine flux increased by 62% to a value that was almost identical to the value at clinical phase 3. Concurrently, plasma concentrations increased by 77% but remained significantly lower (P < 0.05) than the clinical phase 3 value. When the subjects were separated into different groups according to diagnosis, arginine flux remained significantly slower (P < 0.001) at clinical phase 1 than at clinical phases 2 and 3 in the marasmic kwashiorkor and kwashiorkor groups but not in the nonedematous group. Between groups, at clinical phase 1, flux was significantly slower (P < 0.001) in the marasmic kwashiorkor and kwashiorkor groups than in the nonedematous group. These slower fluxes in the marasmic kwashiorkor and kwashiorkor groups at clinical phase 1 were associated with lower plasma arginine concentrations than at clinical phases 2 and 3. In the nonedematous group, although there was no significant difference in arginine flux among clinical phases, the plasma concentration was lower (P < 0.01) at clinical phase 1 than at clinical phases 2 and 3. When the marasmic kwashiorkor and kwashiorkor subjects were grouped together, arginine flux was significantly slower (P < 0.001) in the edematous group at clinical phase 1 than at clinical phases 2 and 3 and slower than at the corresponding clinical phase 1 value of the nonedematous group. In both groups, arginine flux did not change significantly from clinical phase 2 to 3.

View this table:
TABLE 3. Endogenous arginine production and plasma concentration in children with edematous and nonedematous severe undernutrition¹

 
There was no statistically significant effect of clinical phase or diagnosis on nitric oxide outcome measures. When all subjects were grouped together, there was no significant difference in plasma NOx concentration or fractional or absolute synthesis rates among clinical phases (Table 4). Similarly, when the subjects were separated into different groups according to diagnosis, there were no significant differences in plasma NOx concentrations or fractional or absolute synthesis rates between the different groups. None of these parameters changed significantly in response to treatment.

View this table:
TABLE 4. Plasma nitrate plus nitrite (NOx) concentration and fractional and absolute synthesis rates in children with edematous and nonedematous severe undernutrition¹

 

DISCUSSION  
In this study, we aimed to test the hypothesis that at the time they are admitted to the hospital for treatment, children with edematous SCU will have a slower arginine flux but faster NO synthesis rate than do children with nonedematous SCU, and that arginine flux will be slower in the malnourished state than in the recovered state in the edematous group but not in the nonedematous group. Our results show that arginine flux was slower in children with edematous SCU than in children with nonedematous SCU at clinical phase 1, when the children were both severely undernourished and infected. At clinical phase 1, arginine flux was slower in the children with edematous SCU than at clinical phase 2, when the children were still anthropometrically undernourished but their infections were cured, and at clinical phase 3, when they were recovered. This difference in arginine flux was not present, however, in children with nonedematous SCU. In both the edematous and the nonedematous SCU groups, there were no significant differences in plasma NOx concentration or fractional or absolute synthesis rates among clinical phases. These results suggest that children with edematous SCU cannot maintain arginine production in the undernourished and infected state but those with nonedematous SCU can. Finally, all children with SCU can maintain NO synthesis rates despite the slower arginine production of those with edematous SCU when they are infected and undernourished.

In humans, arginine is regarded as a semi-essential amino acid, meaning that it is produced in adequate quantities to meet the requirements of healthy adults but not of growing children or persons stressed by severe trauma or sepsis (6). Because its total flux depends on dietary intake, release from protein breakdown, and de novo synthesis, it is highly likely that arginine will be in short supply in children with SCU. Our data, however, show that this is only true for children with edematous SCU. In children with nonedematous SCU, arginine flux was not slower in the undernourished state than in the recovered state. However, despite maintaining arginine flux at a rate not significantly different from the rate at recovery, the plasma arginine concentration was lower than the value at recovery, which suggests that arginine utilization was greater than production in the infected undernourished state. In past studies, we reported that whole-body protein breakdown, the major contributor to overall amino acid flux, was not diminished in children with marasmus (2, 5), which indicates that they may still be able to maintain arginine flux in the infected and undernourished state. There is evidence, however, that arginine utilization and catabolism are increased in patients with systemic infections and that this is associated with lower plasma concentrations, which suggests that flux cannot keep up with utilization in this condition (6, 13). Because 6 of the 7 marasmic children had clinical evidence of infections at clinical phase 1, our finding of a lower plasma arginine concentration suggests that these children could not increase de novo arginine synthesis to compensate for inadequate dietary intake and also to meet any increased demands due to their infections.

Whereas the children with marasmus did not have a slower arginine flux when undernourished and infected, those with edematous SCU did. In the latter group, arginine flux was 50% slower than at clinical phases 2 and 3. On the basis of our finding that the whole-body protein breakdown rate was markedly slower in children with edematous SCU when they were undernourished than when they were recovered (2, 5), it is expected that these children will have slower fluxes of most amino acids at clinical phase 1. In the case of the dietary nonessential amino acids, however, it is possible for de novo synthesis to fill the gap created by any reduction in the amount released from protein breakdown. We have found this to be true for glycine and serine, but not for cysteine, a nonessential amino acid that depends on the availability of the essential amino acid methionine for de novo synthesis (3-5). Our present findings indicate that in the case of the semi-essential amino acid arginine, de novo synthesis was not sufficient to fill the shortage created by inadequate dietary intake plus decreased release from protein breakdown when the children with edematous SCU were severely undernourished and infected. Furthermore, the lower plasma arginine concentration at clinical phase 1 suggests that flux of arginine was not sufficient to meet requirements in the infected undernourished state. Similarly, at clinical phase 2, although arginine flux doubled to a rate almost identical to the rate at recovery, the plasma pool was still not fully replenished. That is, despite a concurrent increase of 107% in the plasma arginine concentration from clinical phase 1 to 2, it was still lower than the value at recovery. Hence, flux was still not enough to meet requirements and to replenish the plasma pool. In the case of subjects with kwashiorkor, at recovery, plasma concentrations were still lower than the corresponding values of the subjects with marasmus, which suggests that the plasma pool was still not fully replenished or that a systemic difference exists between the 2 types of SCU. Finally, it is very likely that increased de novo synthesis contributed to the increased arginine flux at clinical phase 2, because in a previous study we reported that the protein breakdown rate is still slower at clinical phase 2 than at recovery in children with edematous SCU. If true, this would suggest that in the children with edematous SCU, de novo arginine synthesis was impaired in the infected undernourished state.

Besides its importance for protein synthesis, ammonia detoxification, wound healing, and immune function, arginine as a precursor for NO synthesis serves important roles in the regulation of blood pressure plus the numerous other regulatory roles attributed to NO (6). Low plasma arginine and increased NO production have been reported in septic patients (6), which suggests an increased conversion of arginine to NO despite reduced arginine availability. The findings of Fechner et al (7) of a 100% higher plasma NOx concentration in children with kwashiorkor than in healthy controls suggest that the same phenomenon may exist in children with edematous SCU and infections. The findings of the present study do not corroborate those of Fechner et al (7), because we found no differences in plasma NOx concentration or fractional or absolute synthesis rates from the severely undernourished and infected state to the recovered state in the children with edematous SCU. The same was true for the children with nonedematous SCU. Similarly, although arginine flux was 50% slower in the children with edematous SCU than in the children with nonedematous SCU in the severely undernourished and infected state, there were no significant differences in plasma NOx concentration or fractional or absolute synthesis rates, which suggests that there was still sufficient arginine to support NO synthesis in the children with edematous SCU. For example, NOx synthesis in the kwashiorkor group at clinical phase 1 was 0.84 µmol · L plasma^–1 · h^–1, which, assuming 70 mL plasma/kg at this stage, translates to 0.059 µmol · kg^–1 · h^–1. This amount of NO will require the metabolism of just 0.16% of the 35 µmol · kg^–1 · h^–1of arginine being produced by these subjects. If anything, these results in the kwashiorkor group are opposite those reported by Fechner et al (7), which suggests that increased NO production may not be contributing to the oxidative stress of children with kwashiorkor.

ACKNOWLEDGMENTS  
We are grateful to the physicians and nursing staff of the TMRU for their care of the children and to Hyacinth Gallimore, Lorraine Wilson, O'Neil Brown, Bentley Chambers, Sharon Howell, Margaret Frazer, and Melanie Del Rosario for their excellent work and support in the conduct of the studies and analysis of the samples.

All 5 authors contributed to all aspects of the production of this manuscript, including the design of the study, data collection, analysis and interpretation, and writing of the manuscript. None of the authors had any conflict of interest with the funding agencies.

REFERENCES