Main

Parenteral nutrition is often required in the early stages of feeding immature very low birth weight infants. When compared with i.v. glucose only, high levels of hydroperoxides have been detected in nutrient solutions(1, 2) as well as in complete preparations of i.v. nutrition(3). Hydroperoxides are reactive and can form free radicals(4). In turn these oxygen species may disrupt cell membrane integrity and mediate tissue injury(4). They can also interfere with the production of vasoactive eicosanoids(5), which may facilitate, for instance, neovascularization in retinopathy of prematurity(6). Because antioxidant systems of premature infants are immature(7), the potential to cause an oxidant stress by these peroxides is high. This was suggested by Pitkänen et al.(8) who described a correlation between indices of free oxygen radical-induced lipid peroxidation and outcome in immature newborn infants.

Under the conditions met during i.v. nutrition of newborn infants, the multivitamin solution is the major source of hydroperoxides(3). It is well known that the generation of hydroperoxides is stimulated by phototherapy or even by daylight(3, 9). In an attempt to decrease the oxidant load received by immature infants, we evaluated which components of the multivitamin solutions might be involved in light-induced generation of peroxides. Rf and PS are light-sensitive molecules that could be implicated in the generation of peroxides(9–11). Most parenteral multivitamin solutions contain 5′-phosphate FMN as a source of Rf, a vitamin implicated in a large number of metabolic pathways, including respiratory chain reaction. PS are detergents added in certain multivitamin solutions to dissolve liposoluble and hydrosoluble vitamins in the same medium.

Although vitamin C (ascorbic acid/AH) is a known antioxidant, it can become prooxidant(10, 12, 13). Indeed, in the presence of oxidants, AH may give a first electron and become the radical ascorbate (A), and, with the loss of a second electron, this radical becomes the stable dehydroascorbate (A). This radical has been implicated in the generation of peroxides(10, 12, 13). Because AH is also present in multivitamin solutions, we analyzed in vitro the effect of the association of this electron donor with FMN and PS. We hypothesized that, in the presence of light, the interaction of these products is responsible for the generation of peroxides and the alteration of the quality of nutrients to be infused.

METHODS

To determine the effect of the studied components (AH, FMN, PS), we first analyzed the generation of peroxides in a solution containing each component alone or in admixture. A complete multivitamin solution served as control. We tested the consequences of adding AH and/or FMN in a dextrose solution, in an amino acid preparation, and in a lipid emulsion. All preparations containing FMN, and/or AH, and/or PS were incubated at room temperature in a polystyrene tube exposed to daylight or kept in darkness. Darkness was defined as protecting the tubes by an aluminum foil and storing them in a dark chamber during incubation. Determinations were performed in triplicate.

Protocols. The generation of peroxides was confirmed in a solution of 1% (vol/vol) MVI Pediatric® (Rhône Poulenc Rorer, Montreal, Canada) in water. This concentration is typically used to cover allowances for newborn infants. This 1% multivitamin preparation contains AH(900 μmol/L), FMN (7.4 μmol/L), and PS (PS 20, 0.1 mg/L + PS 80, 1.6 mg/L).

To determine their individual effects on the generation of peroxides, FMN(7.4 μmol/L), AH (900 μmol/L), or PS (PS 20, 0.1 mg/L + PS 80, 1.6 mg/L)(Sigma Chemical Co., St. Louis, MO) were solubilized in water at concentrations found in 1% MVI Pediatric.

To evaluate the consequences of the admixture of FMN (7.4 μmol/L), AH(900 μmol/L), and/or PS (PS 20, 0.1 mg/L + PS 80, 1.6 mg/L) in water, peroxides were measured after different incubation times. To determine the relative importance of H2O2 among the peroxides formed, catalase(50 IU) was added to an aliquot of the tested solution for 10 min, followed by a 5500 × g centrifugation, before peroxide quantification.

To evaluate the effect of PS on a complete multivitamin solution, a preparation devoid of PS was used: Cernevit® (Clintec-Baxter, Paris, France). The generation of peroxides in 1% (vol/vol) Cernevit and 1% Cernevit+ PS (PS 20, 0.1 mg/L + PS 80, 1.6 mg/L) was compared.

To characterize the effect of the concentration of the studied vitamins in a dextrose solution [solution D: dextrose 5% (wt/vol) + NaCl 0.45% (wt/vol)], a dose-effect relationship of AH on peroxides was established over 3 d, by decreasing AH from 900 μmol/L to 0.090 μmol/L in the face of a constant FMN concentration of 7.4 μmol/L. Likewise, a dose-response of FMN on peroxides was established over 3 d by decreasing FMN from 7.4 to 0.0074μmol/L with a constant AH concentration of 900 μmol/L. The effect of light was evaluated by comparing the generation of peroxides with FMN (7.4μmol/L) + AH (900 μmol/L) after 24 h of daylight or darkness in solution D.

To verify if amino acid solutions interfere with AH (900 μmol/L), FMN(7.4 μmol/L) or both, we determined the generation of peroxides in an amino acid preparation (D + AA: solution D + 13.3% (vol/vol) Primene 10%®, Clintec-Baxter, Mississauga, Ontario, Canada). This corresponds to 2 g of amino acids in a total of 150 mL of fluid. Because D + AA contains cysteine(2.1 mmol/L), the thiol functions of the preparation were measured in the presence of FMN (7.4 μmol/L), AH (900 μmol/L), or both, after exposure to daylight or darkness.

To evaluate the effect of delivering these vitamins via an alternate route, we tested the effect of diluting them in the lipid emulsion (IL, Intralipid 20%®, Kabi, Baie d'Urfé, Québec, Canada). The volume of IL infused in newborn infants represents approximately 1/15 of dextrose + amino acid solutions in which the multivitamin preparation is usually diluted. Therefore, to give the same amounts of FMN and AH as in 1% MVI Pediatric in the smaller volume of IL, the concentrations had to be adjusted accordingly. Peroxides were measured in IL alone or in IL with FMN (0.11 mmol/L) or AH(13.6 mmol/L), or both. The positive control was the admixture of FMN (0.11 mmol/L) + AH (13.6 mmol/L) in D. To determine whether other peroxides than H2O2 were present, catalase was used as described above.

Assay. To quantify hydroperoxide levels, the ferrous oxidation/xylenol orange technique was used as described previously(3, 14). One milliliter of a solution containing H2SO4 (22.5 mmol/L), xylenol orange (90 μmol/L), FeCl2(225 μmol/L), and 2,6-di-tert-butyl-4-methylphenol (3.6 mmol/L) in methanol was incubated with a 100-μL sample. The oxidation of Fe2+ in the presence of hydroperoxides forms Fe3+, which reacts with xylenol orange to produce a chromophore. After 30 min of incubation, lipids were separated by centrifugation (5500 × g, 3 min), if indicated. The absorbency of the supernatant was read (Beckman spectrophotometer DU-6) at 560 nm. The results were expressed in μmol/L equivalents of TBH. Intra- and interassay coefficients of variation were 4 and 5%, respectively. Because AH can react with Fe3+(12), we verified that the addition of various AH concentration (from 1.8 mmol/L to 0.6 mmol/L) did not change the measured level of peroxides.

We measured the thiol function by adding a 100-μL sample to 900 μL of Tris buffer (0.1 mM, pH 7.6) containing 0.6 mmol/L 5,5-dithiobis(2-nitrobenzoic acid)(15). The absorbency was read at 412 nm and compared with a standard curve. The results were expressed in mmol/L equivalents of cysteine. All reagents were purchased from Sigma Chemical Co.

Statistical analysis. Results were expressed as mean ± SEM, n = 3, and compared by t test. The level of significance was set at p < 0.05.

RESULTS

Levels of peroxides were undectable after 24 h of exposure to light when FMN or PS were solubilized, separately or together, in water. Incubation of AH alone or with PS was responsible for a slow increase in peroxides (Fig. 1). However, the admixture of FMN and AH produced a higher level of peroxides than found in 1% MVI Pediatric (862 ± 4μmol/L equivalents TBH versus 533 ± 27 μmol/L equivalents TBH after 24 h). No peroxides were detected after the addition of catalase. In the presence of FMN and AH, the peroxide concentration in darkness remained low after 24 h (8 ± 1 μmol/L versus 987± 22 μmol/L equivalents TBH). Similar to the effects of PS on AH + FMN observed in Figure 1, the addition of PS to the multivitamin solution Cernevit did not significantly modify the level of peroxides after 24 h of exposure to light (858 ± 12 μmol/L equivalents TBH versus 899 ± 12 μmol/L equivalents TBH).

Figure 1
figure 1

Peroxide generation in water. Peroxides concentrations after 4 and 24 h of incubation in solutions containing 900 μM AH and 7.4μM FMN and/or PS 20 (0.1 mg/L) + PS 80 (1.6 mg/L) (PS). AH + FMN generated large amounts of peroxides. The addition of PS did not significantly change the concentration of peroxides measured after 24 h. Data are presented as means ± SEM, n = 3.

Full size image

In the dextrose solution, AH or AH + FMN generated less peroxides than in water (after 24 h with AH, 3 ± 3 μmol/L versus 146± 3 μmol/L equivalents TBH, p < 0.05; and with AH + FMN, 791 ± 7 μmol/L versus 862 ± 4 μmol/L equivalents TBH, p < 0.05). This result could be explained by the known antioxidant effect of glucose. The dose-response relationship between AH and peroxides (represented on a logarithmic scale) in the presence of FMN (7.4μmol/L) is presented in Figure 2. Results for 0.09 and 0.9 μM of AH are not significantly different from those without AH and are therefore not presented in the figure. After subtracting the basal level of peroxides observed without AH, it appears that a change in AH concentration was associated with a similar drop in peroxides. For instance, a 50% decrease in AH concentration led to a 50% drop in peroxide concentration after 24 h. On the contrary, decreasing FMN concentrations from 7.4 to 3.7 μmol/L did not significantly change the level of peroxides after 24 h (Fig. 3). During the first 24 h, peroxide concentrations increased with FMN. After 24 h, the peroxide levels observed with 7.4 or 3.7 μmol/L FMN reached a plateau equimolar with the AH initial concentration of 900 μM. However, independently of FMN concentrations, after 24 h of incubation, peroxide concentrations did not increase even if they were below this plateau.

Figure 2
figure 2

Generation of peroxides (logarithmic scale) in dextrose as a function of AH concentration. Dose-response curves for AH (9-900 μM) in solutions containing FMN (7.4 μM) after subtracting the basal level of peroxides observed without AH at each incubation time. The results showed that a change in AH concentration was associated with an equimolar change in peroxides. Data are presented as means ± SEM, n = 3.

Full size image
Figure 3
figure 3

Generation of peroxides (logarithmic scale) in dextrose as a function of AH concentration. Dose-response curves for FMN (0.0074-7.4μM) in solutions containing AH (900 μM) after subtracting the basal level of peroxides observed without FMN at each incubation time. During the first 24 h, peroxide concentrations increased with FMN. After 24 h, the peroxide levels observed with 7.4 or 3.7 μmol/L FMN reached a plateau equimolar with the AH initial concentration. However, independently of FMN concentrations, after 24 h of incubation peroxide concentrations did not increase even if they were below this plateau. Data are presented as means± SEM, n = 3. The variations are not depicted when their size is smaller than that of the symbol.

Full size image

In the amino acid solution (D + AA), the generation of peroxides with AH + FMN was lower than in D (Fig. 4). But after 24 h, the concentration of peroxides was the same in the presence of FMN alone as with FMN + AH. In the absence of FMN, the level of peroxides in D + AA was low (Fig. 4). The disappearance of peroxides after the addition of catalase indicates that the peroxide measured in D and in the amino acid solutions was H2O2.

Figure 4
figure 4

Generation of peroxides in an amino acid preparation. Peroxides concentrations after 4 and 24 h of incubation in solutions containing amino acids + dextrose + NaCl (amino acid preparation) with or without AH (900 μM) or FMN (7.4 μM). The control solution contained dextrose + NaCl (dextrose) with AH (900 μM) and FMN (7.4 μM). In the presence of FMN and amino acids, the level of peroxides after 24 h was independent of the addition of AH. Data are presented as means ± SEM,n = 3.

Full size image

More than 95% of thiol functions in D + AA disappeared after 24 h in all solutions containing FMN (Fig. 5). After 6 h the remaining thiol functions measured in the presence of FMN alone was half that observed with AH + FMN; therefore, thiols disappeared faster in the presence of FMN alone. Darkness protected thiol functions against the reaction induced by FMN; after 28 h, 1.445 ± 0.042 versus 0.043 ± 0.001 mmol/L equivalents Cys; control (in darkness without FMN), 1.490 ± 0.001 mM equivalents Cys. In daylight, AH alone had a slight protective effect on thiol functions (1.37 ± 0.01 with AH versus 1.26 ± 0.02 mmol/L equivalents Cys without vitamin, p < 0.05, after 24 h).

Figure 5
figure 5

Effect of vitamins on thiol functions of a cysteine containing amino acid solution. Thiol functions after 1, 6, and 24 h of incubation in solutions containing amino acids + dextrose and FMN (7.4 μM) in the presence or not of AH (900 μM) or with the multivitamin solution 1% Cernevit®. The control solution was devoid of vitamins (No vitamin). In all solutions containing FMN, over 95% of the thiol function disappeared within 24 h. Data are presented as means ± SEM, n= 3.

Full size image

In the lipid emulsion, the generation of peroxides with FMN + AH was 12 times higher than in IL alone and twice that observed in D (after 5 h, 4999± 614 versus 2721 ± 97 μmol/L equivalents TBH). After 7 h, the generation of peroxides with the admixture of FMN alone was 3-fold that observed with IL (Fig. 6). The peroxides unaffected by catalase were three times higher with FMN alone, and four times higher in the presence of FMN + AH, when compared with IL alone (Fig. 6). Without FMN, AH protected the lipid emulsion against the generation of peroxides (after 4 h, 368 ± 11versus 640 ± 31 μmol/L equivalents TBH, p < 0.05) as reported by others(1).

Figure 6
figure 6

Generation of peroxides in a lipid emulsion. Peroxides concentrations after 7 h of incubation in 20% lipid emulsion (IL) with FMN (0.11 mM) and/or AH (13.6 mM). The control solution was IL. White bars represent total peroxides, whereas black bars represent peroxides measured after addition of catalase before the assay. FMN admixture increases up to three times the concentration of peroxides unaffected by catalase. Data are presented as means ± SEM, n= 3.

Full size image

DISCUSSION

FMN and AH are susceptible to reacting together and forming peroxides under concentrations and conditions close to those used in clinical practice in neonatal medicine. Because AH can induce the generation of peroxides without FMN, and because FMN accelerates this generation (Fig. 1), this suggests that AH acts as a substrate and FMN acts as a catalyst. The dose-response curve presented in Figure 3 strengthens this proposal because the maximum peroxide level was obtained faster by increasing the FMN concentration.

The generation of peroxides induced by AH and Rf in milk has already been described in the presence of oxygen and light(16, 17). The mechanism suggested by different authors(10, 13, 18) is presented in Figure 7. Rf exposed to light, produces a strongly oxidizing triplet (3Rf). In the presence of an electron donor like AH, this triplet is reduced and radicals, A as well as Rf are produced(10, 13). The reduced Rf (Rf) will then convert O2 to superoxide, regenerating Rf. Superoxide and the radical AH can react together and form A and H2O2(10). In this reaction, Rf exposed to light is the catalyst of the reduction of O2 by AH. Kim et al.(10) demonstrated that, similarly to Rf, FMN is able to generate the AH radical. Our data are consistent with these chemical reactions, because there is an equimolar drop in peroxides with the substrate AH (Fig. 2). However, in the case of the dose-response relation with FMN (Fig. 3), the catalyst effect is not detectable after 24 h.

Figure 7
figure 7

Suggested reaction for the generation of H2O2 in multivitamin preparations. In the presence of light(hv), Rf catalyzes the reduction of O2 by the AH anion AH to produce H2O2 and dehydroascorbate (A).

Full size image

In the presence of light, oxygen could also react directly with 3Rf and produce an oxygen singlet(10), which in turn reacts with AH and leads to the loss of vitamin C. The diminution over time of AH + A in solutions of parenteral nutrition exposed to light(19) suggests that this last reaction might occur. Finally, during light exposure Rf might be destroyed(10). However, because the concentration of FMN remains stable or slightly decreased over 24 h in solutions of parenteral nutrition(19, 20), this phenomenon is probably of minor quantitative significance during the first 24 h, but may be of qualitative importance because some FMN photodecomposition products are toxic(21). Alternatively, the absence of increase in peroxides observed after 24 h (Fig. 3) suggests an inactivation of FMN.

In the absence of lipids, most peroxides formed are H2O2. The toxicity of H2O2 is well documented at the cellular level(22). Indirect evidence suggests that the toxicity of H2O2 is potentially high in newborn infants(23–26). However, deleterious effects might even start in the solution of parenteral nutrition itself. Indeed, the generation of peroxides is associated with an alteration of the nutrients to be infused. Our results suggest that FMN can react with other electron donors than AH, such as amino acids or lipids. This is consistent with the observation that photosensitized Rf can be reduced by other electron donors(13).

The 40% drop in thiol function observed in the control as well as in darkness with or without FMN could be related to the formation of the compound D glucose-cysteine(27) or the spontaneous oxidation of this amino acid. Alternatively, the >95% disappearance of the thiol function, in the presence of light and FMN, suggests that an important proportion of cysteine was oxidized into cystine by the photoexcited form of riboflavin. Because cystine is insoluble in water, it raises the question about the amount of cysteine + cystine actually received by infants in the course of parenteral nutrition, when FMN is added to the solutions. Because cysteine is essential for protein synthesis(28) and glutathione metabolism(29) in premature infants, the addition of FMN in cysteine-enriched amino acid preparations might be deleterious for immature infants.

Bathia et al.(9) have already shown that tyrosine, methionine, proline, and tryptophan can be oxidized in the presence of FMN(9). They generate photoproducts, shown to be toxic in animals(30, 31). This supports an oxidation of the amino acid solution as an explanation for the lower peroxide concentrations found in D + AA compared with D (Fig. 5), as previously reported(3). Furthermore, in an amino acid solution we show that the generation of peroxides is as important with or without AH, but the residual thiol function is significantly higher after 1, 4, and 24 h in presence of AH + FMN when compared with FMN. Because AH competes as an electron donor, it may initially protect the rest of the solution against oxidation. In the absence of FMN, AH had a protective effect in the amino acid preparation as well as in the lipids, confirming the findings of others(1).

The admixture of FMN alone or of FMN + AH to IL, which contains over 50% of polyunsaturated fatty acids, increased the total amount of peroxides, but also the amount of peroxides unaffected by catalase suggesting that lipid peroxides were produced. The latter are susceptible to induce peroxyl radical formation(4). Because of their long half-life, these radicals can diffuse in the entire organism and initiate peroxidative injury in tissues distant from the site of initiation of peroxidation. The antioxidant defenses may be altered by the oxidant load represented by lipid peroxides; IL alone was reported to deplete the intracellular level of the antioxidant glutathione in isolated red blood cells of newborns infants(32). Therefore, adding FMN to lipid emulsions containing high levels of polyunsaturated fatty acids should be avoided, when exposed to light. This information may be pertinent in respect to the “three in one” preparations in which all nutrients are mixed in one bag.

To further decrease the oxidative load received by patients on parenteral nutrition, the protective effect of darkness should be enforced from the time of preparation of the solution until the infusion of the product(3). An alternate measure could be to adjust the levels of FMN intake. In parenteral regimens these levels are nine times higher than in breast milk and may be excessive(33, 34), but our results show that decreasing FMN concentrations by a factor of 2 did not decrease the concentration of peroxides after 24 h, and the levels of peroxides observed in the presence of 1/10 of the usual concentration are still above 500 μM after 24 h (Fig. 3). Because FMN can react at least with AH, with lipid emulsions or with specific amino acids, one should inject it separately from the TPN solution. Before implementing such a policy, studies on the pharmacokinetics and safety of infusing FMN directly to newborn infants need evaluation. The use of other forms of vitamin B2, such as Rf or flavin adenine dinucleotide, might be of interest. Although flavin adenine dinucleotide is a less potent photosensitizer, it can also generate A(10), therefore it is improbable that switching from FMN to another flavin will dramatically change our results. Adding ascorbyl palmitate instead of AH did not prevent the generation of the AH radical(10).

PS is susceptible to photooxidation(11); however, under the present experimental conditions this nonionic detergent was not the major contributor to peroxide generation. The admixture of PS with AH + FMN does not increase significantly the final load of peroxides after 24 h, but may slightly increase the rate of formation of peroxides (Fig. 2). In view of the fact that it is not an essential nutrient, and that it has been implicated at higher concentrations in toxic side effects in newborn infants(35), we wonder whether it should not be replaced by a nonphotosensitive product.

This study demonstrates that the admixture of FMN with electron donors such as AH, an amino acid solution, or a lipid emulsion can generate peroxides in TPN solutions. With the immature antioxidant system of premature children, the infusion of those peroxides is at least hazardous. Therefore, the infused oxidant load should be decreased. Considering that vitamins are essential nutrients, this can be achieved by protecting all solutions containing FMN from light, or by removing photosensitive products from multivitamin solutions and infusing FMN separately.