The metabolic effects of multi-trace elements on parenteral nutrition for pediatric patients: A Randomized Control Trial and Metabolomic Research

Qingti Tan Chengdu Women's and Children's Central Hospital: Chengdu Women and Children's Central Hospital Yu Wang Sichuan Provincial People's Hospital: Sichuan Academy of Medical Sciences and Sichuan People's Hospital Guoying Zhang Chengdu Women's and Children's Central Hospital: Chengdu Women and Children's Central Hospital Bin Lu Chengdu Women's and Children's Central hospital Tao Wang Chengdu Women's and Children's Central Hospital Tao Tao Chengdu Women's and Children's Central Hospital: Chengdu Women and Children's Central Hospital He Wang Shanghai University of Traditional Chinese medicine Hua Jiang (  cdjianghua@qq.com ) Sichuan Provincial People's Hospital: Sichuan Academy of Medical Sciences and Sichuan People's Hospital Wei Chen Chinese Academy of Medical Science & Peking Union Medical College Hospital


Introduction
The nutritional status of pediatric Subjects is closely related to disease progression and prognosis. To improve nutritional status and cure rates of associated conditions, guidelines [1] have recommended that complete or partial parenteral nutrition (PN) should supply calories, uids, and nutrients in situations where there is unavailable or insu cient intestinal nutritional uptake. Trace elements are important components of PN as they effectively utilize glucose and fat to supply energy and synthesize proteins.
These molecules are essential components for enzyme cofactors, and they play key roles in immune regulation and antioxidation. Therefore, guideline documents have provided recommendations on trace element-use and doses for pediatric PN [1][2][3] , but no global consensus exists for these measures.
Metabolomics is an emerging research area which re ects body health or disease status via the proteomic analysis of metabolites. Using this technology, we analyzed the effects of different multi-trace element injection (I) doses on the nutritional metabolism of pediatric subjects.

Subjects
In total, 40 Subjects requiring PN and hospitalized at the Chengdu Women's and Children's Central Hospital from November 2017 to March 2018 were enrolled. The research protocol is approved by the ethical committee of the hospital (No.2017 (21)). The research protocol register number is: ChiCTR-IPR-17013037. Preparation of standard solution: Dilute 10 mg/L standard mixture solution to 1 mg/L for future use.

Calibration Curve
Accurately pipette multi-element mixed standard solution, prepare standard curve solution, and the nal concentrations are 1 µg/L, 5 µg/L, 10 µg/L, 20 µg/L, 50 µg/L and 100 µg/L. The prepared standard solution were tested by the organic injection system to obtain the standard curve.
In the experiment, parallel samples and spike recovery were used as the means of sample quality control to ensure the accuracy of the results.

Preparation of plasma samples
After thawing at room temperature, plasma samples were centrifuged at 16000 rpm for 10 min. Then, 50 µl deuterated heavy water (D 2 O) was added to an NMR tube, to which 450 µL plasma was added. The sample was shaken for 2 min and incubated at room temperature for 10 min until 1 H-NMR (600 MHz) analysis.

1 H-NMR data collection
A one-dimensional hydrogen spectrum was obtained after sample processing. In this study, 1 H-NMR analysis was performed using a Bruker Avance DR × 600 MHz model, with a working frequency of 600. 13 MHz, equipped with a Bruker inverse broad band probe (rIBB). The addition of 10% D 2 O inhibited the solvent peak during sample preparation, and the pulse sequence (zgp) was used to inhibit the water peak during pre-saturation. All spectra were collected at room temperature (i.e., 300 Kelvin (K)), with a spectrum width of 20 ppm, sampling points of 32 K, and a cumulative frequency of 256 times.

1 H-NMR spectrum processing
In plasma samples, the molecular nucleus of a compound resonates in a high magnetic eld, and its frequency is gradually decreased. The original decay signal of the tested sample referred to our raw data (free induction decay; FID). The FID signal was imported into MestReNova software (MestreLab Research, Spain) for Fourier transformation, to generate one-dimensional 1 H-NMR spectra. These were processed for chemical shift and automatic baseline adjustment. The convolution technique was used to minimize changes in peaks, and to ensure that larger peaks did not cover up smaller ones. Then, for all plasma samples, 0-9 parts per million (ppm) segments of one-dimensional hydrogen spectra were divided into 0.04 ppm sections, and 223 chemical shift value segments were integrated to nally obtain corresponding integral values. The two-dimensional matrix was then exported in .CSV format for analysis.

Data preprocessing
All data matrices were preprocessed -line normalization and standardization. Due to differences in plasma sample dilution, concentration, test temperature, instrument working stability, and other factors during processing and measurement, 1 H-NMR spectra from the same types of plasma sample in different batches were not completely consistent. Therefore, line normalization of data matrices was required (it was assumed the highest peak in each 1 H-NMR spectrum referred to the same substance with very similar content). The line normalization formula was represented as:

Spectrum data analysis
We used the supervised pattern recognition method, Partial Least Squares-Discriminant Analysis (PLS-DA) to perform data dimension reduction. The variable importance in the projection (VIP) of the PLS-DA model, with corresponding chemical shifts, was calculated. Chemical shifts with VIP values > 1 and P < 0.05 were selected, and corresponding metabolites were investigated using the human metabolome database (HMDB at https://hmdb.ca/).
All clinical data were statistically analyzed using SPSS Version 21.0 software, and described by median (interquartile range) or mean values ± standard deviation (mean ± SD) according to distribution type. Measurement data were rst tested for data distribution type. Student's t-test was used for normally distributed data, and the rank-sum test was used for non-normally distributed data. The chi-square test was used for enumeration data. The statistical signi cance level was set at p < 0.05. The mean substitution method was used for missing clinical data.

Subjects status
In total, 40 subjects were enrolled into the study ranging from 29 days to 10 years old, including 18 subjects in Group A and 22 subjects in Group B. Principal diagnoses included: 1) 10 subjects had received stulation surgery, 2) Six subjects had acute upper gastrointestinal bleeding, 3) Four subjects had congenital mega-colon, 4) Three subjects had acute gangrenous appendicitis with perforation, 5) Two subjects had congenital hypertrophic pyloric stenosis, 6) Two subjects had adhesive intestinal obstruction, and 7) Two subjects had acute intussusception. The remaining diagnoses included; one patient each with Merkel diverticulitis with bleeding, necrotizing enterocolitis, acute descending colon perforation, small intestine torsion, portal hypertension syndrome, acute severe myocarditis, toxic intestinal paralysis, autotransplantation after splenectomy, congenital anal atresia and traumatic splenic rupture, and oesophageal atresia surgery Subjects information at admission is shown (Table 1). No signi cant differences were observed between groups in terms of gender, weight, pediatric critical illness score, vital signs, length of hospital stay, and hospitalization expenses. Similarly, we observed no signi cant differences in routine blood and biochemical tests, except hemoglobin.
Economic aspects were also considered in this study. After comparing the length of hospital stay and hospitalization expenses between groups, we observed that the high-dose administration of MTEI-(I) did not signi cantly prolong these factors for subjects, suggesting minimal impact and burden on subjects and their families.

General condition changes in Subjects before and after treatment
According to results of trace element detection, There was no signi cant differences observed in subjects after 1, 3 and 5 days of treatment in each group, therefore we focused on and analyzed data before treatment (T0) and after 5 days of treatment (T5). The general condition, routine blood and biochemistry data of both groups after 5 days of treatment are shown ( Table 2); No signi cant differences were observed between two groups. Routine blood and biochemistry data before and after treatments were compared between groups (Table 3). After treatment, WBC, N, Cr, TB, DB, and ALB data in both groups decreased, of which, WBC and Cr in Group B were signi cantly decreased after 5 days of treatment.

Trace element data in subjects before and after treatment
Trace element patient data before and after treatment are shown (Table 4). After 5 days of treatment, Mn and Cu decreased in different extent, whereas Zn and Se increased in the groups. For both Zn and Cu, we observed signi cant differences compared to before treatment. We also compared trace element data in both groups before and after treatment (Table 5). After treatment, Mn and Cu decreased in both two groups, whereas levels of Zn and Se increased. Zn levels in Group B increased signi cantly when compared with Group A. Cu in Group B signi cantly decreased when compared with Group A. from Group B. We used 1 H-NMR metabolic ngerprinting of patient plasma to distinguish patient metabolomics before (T0) and after treatment (T5) (Fig. 1). The VIP of chemical shift values of metabolites at T0 and T5 are shown (Fig. 5), of which those with VIP ≥ 1 and P < 0.05 were taken as characteristic metabolites (Tables 6 and 7). According to Table 6 and Table 7, after the 5 days treatment, valine, leucine, isoleucine (α-ketoisovaleric acid), taurine, hypotaurine (hypotaurine), arginine, proline (phosphocreatine), ketone (acetoacetic acid and acetone) and other metabolic processes signi cantly decreased.    Figure 3 shows PLS-DA analyses based on patient metabolic ngerprint spectra. From these data, we observed metabolic differences between subjects in both groups. VIP metabolite data are shown (Fig. 4), for which the characteristic metabolites with VIP > 1 and P < 0.05 are shown (Table 8). For Group B, βoxidation of very long chain fatty acids (hexacosanoic acid), arginine and proline metabolism (phosphocreatine), pentose phosphate metabolism (D-ribose), ketone body metabolism (acetone), citric acid cycle (succinic acid), purine metabolism (adenine), caffeine metabolism (dimethylxanthine), and pyruvate metabolism (acetyl phosphate) were all decreased when compared with Group A after T5.

Discussion
Although trace element levels in human tissue account for < 0.01% of total organism mass, these components are vital for human growth and development [4] . For enteral feeding, subjects derive adequate trace elements through diversi ed diets, enteral nutrition products, and oral drug products. For PN, due to chemical molecule stability, a variety of complex drug products containing multi-trace elements are required to meet clinical needs [5] . MTEI-(I) is a complex drug product containing multi-trace elements specially developed for children. It supplements six trace elements such as Zn, Cu, Mn, Se, uorine (Fl), and iodine (I), but not Fe or Cr, to meet guideline requirements for the addition of trace elements during PN [2,3] .
In ammatory mechanisms generated by in ammation and oxidative stress responses from free radical accumulation often cause normal proteins, lipids, and nucleic acids to attack, undermine, and destroy normal physiological functions. Trace elements are required for the regulation of substance metabolism, enzyme catalytic activity, etc., and thus affect in ammation and oxidative stress mediators. For example, during oxidative stress and in ammation, trace element distribution will be altered, thus a reasonable intake of these elements will exert positive effects towards in ammation control, and slow or reduce oxidative stress responses [6] . In previous studies, it was shown that appropriate Cu, Zn and Se levels reduced free radicals, enhanced antioxidant capacity, and regulated in ammatory reactions [7][8][9] , while Cu was positively correlated with bacterial levels and in ammatory markers [10,11] ,and Zn and Se were negatively correlated with in ammation and oxidative stress [12,13] . Supplementation with Zn improved high Cu-Zn ratios in blood, reduced oxidative stress, improved in ammatory conditions, and maintained immune functions [14] . These data were consistent with our ndings suggesting that Cu was decreased, and Zn and Se were increased after PN treatment, with trace element differences in Group B more signi cant. Equally, we observed that WBC levels in both groups were decreased after PN treatment, with levels in Group B signi cantly decreased after PN treatment (p = 0.011). This observation suggested that the appropriate high-dose administration of I was effectively controlling in ammation and antioxidation.
Hexacosanoic acid is a very long chain fatty acid, and is an important component of phospholipid molecules. In a previous study [15] , these molecules were shown to play important roles in cellular biochemical reactions, nutrient storage, and intercellular communications. Due to homeostatic imbalances between molecular transport and utilization, excessive fatty acid accumulation may cause toxicity in some tissues, which becomes manifested as oxidative stress and in ammation, potentially culminating in cell apoptosis [16] . Several studies reported that very long chain fatty acids induced the production of reactive oxygen species in the SK-N-BE neuroblastoma cell line, and enhanced oxidative stress [17] . Gursev et al. [18] observed these molecules activated nicotinamide adenine dinucleotide phosphate oxidase activity, and enhanced superoxide anion-mediated lipid peroxidation in skin broblasts. In this study, we observed that the β-oxidation of very long chain fatty acids (hexacosanoic acid) was signi cantly reduced in Group B (p 0.05), indicating subjects were less prone to oxidative damage caused by lipid peroxidation. Therefore, appropriate high-dose administration of I exerted positive antioxidation effects in this group [19] .
Stress has an important impact on various metabolic pathways. Under stress conditions, the following metabolic characteristics are often observed; high metabolic rate, increased catabolism, and reduced anabolic metabolism, resulting in a negative balance in overall metabolism. In this study, 37 children were under acute stress after surgery or disease. Chen Weiqiang et al. [20] observed that stress induced the loss of Zn from the body, whereas Zn supplementation exerted protective effects. In this study, after supplementing I, we observed that valine, leucine, isoleucine degradation, taurine and hypotaurine metabolism, arginine and proline metabolism, and other amino acid metabolism were all reduced, suggesting a bene t to disease recovery. Equally, ketone metabolism was also reduced, suggesting the high metabolic rate had been relieved. Of these components, β-oxidation of very long chain fatty acids, pentose phosphate metabolism, ketone body metabolism, citric acid cycle and pyruvate metabolism were all signi cantly reduced in Group B. These factors were related to energy metabolism [21,22] , indicating that appropriate high-dose administration of I was helpful in relieving stress induced elevated metabolism.
Hypoxia is a basic pathological process implicated in several diseases [23] . Severe hypoxia induces considerable cellular harm, and often leads to death. Kim et al. [24] observed that Zn ameliorated hypoxic neuronal death induced by deferoxamine (DFX) and sodium azide (NaN 3 ). Xinge et al. [25] reported that Zn chelating agents had protective effects towards hypoxic ischemic brain damage in zebra sh. Kun et al. [26] proposed that exogenous Zn had protective effects towards hypoxic neurons. Hypoxanthine is a naturally occurring purine derivative and is the major catabolite of adenosine triphosphate (ATP) in hypoxic or ischemic tissue [27] . In general terms, a large increase in hypoxanthine levels in bodily uids indicates ATP depletion [28] . In a trial of subjects with critical illness, burns and burn-induced sepsis [28] , the evidence suggested that elevated ATP associated degradation products i.e., adenosine, inosine, hypoxanthine, and xanthine were associated with tissue hypo-perfusion and hypoxia levels. Therefore, it was suggested that purine metabolites such as xanthine and hypoxanthine are potential markers of tissue hypoxia [29] . In our study, the administration of I in Group B signi cantly increased plasma Zn levels. In our metabolomics study, we observed that purine metabolism in Group B was signi cantly reduced, and related metabolites were similarly reduced, indicating that appropriate high-dose administration of MTEI-(I) improved hypoxic conditions in these subjects.

Conclusion
In summary, the appropriate high-dose administration of MTEI-(I) is bene cial for pediatric patients. Such administration does not increase the burden on visceral organs, and appears to exert protective effects on liver and kidney functions. Several studies have shown that Se protects the kidney from oxidative damage, and reduces oxidative damage to the kidney [30][31][32] . The metabolite also reduces serum glutamic pyruvic transaminase, total and direct bilirubin, and reduces ultra-structure liver cell damage in rats [33] .
Supplementation with Zn also delays the progression of chronic kidney disease damage, and relieves its complications [34,35] . In this study, Zn and Se plasma levels were increased by I administration. Also, liver and kidney functional analyses of our subjects indicated that Cr, TB, DB and ALB levels decreased after supplementing MTEI-(I), and Cr was signi cantly decreased in Group B, suggesting that appropriate highdose supplementation of MTEI-(I) was bene cial in improving renal function.
Additionally, E`conomic aspects and duration in hospital were also considered in this study.
Administration at high dose didn't enhance expenditure and duration dramatically. Figure 2 Partial least squares-discriminant analysis (PLA-DA) of subjects at T0 and T5 (a:1 mL/kg; b:2 mL/kg) Page 21/23 VIP chemical shift values between T5 metabolic differences in Group A and Group B