The formulation studied was composed by: 3 g of Pediamino TAU amino acid (Baxter; São Paulo; Brazil), 8,64 g of glucose 50% (Fresenius; Campinas; Brazil), 4 mEq sodium chloride 20% (Darrow; Rio de Janeiro; Brazil), 4,5 mEq calcium gluconate 10% (Halex Istar; Goiânia; Brazil), 1.1 mmoL sodium glycerophosfate 21.6% (Fresenius), 0.25 mEq magnesium sulfate 0.8%, 0.1 mL Ped element (Darrow) and 10 mL MVI 12 opoplex for adult (ICN Farmacêutica; São Paulo; Brazil). The vitamins concentration in PN: 2 mg/mL of vitamin C, 8 μg/mL of folic acid, 1.2 μg/mL of biotin, 0,01 μg/mL of cianocobalamin, 0.03 mg/mL of pantotenic acid, 0.072 mg/mL of B2, 0,8 mg/mL of nicotinamin, 0.08 mg/mL of B6, 0.06 mg/mL of B1, 0.0364 mg/mL of vitamin A, 0.1 μg/mL of vitamin D, 0.2 mg/mL of vitamin E. The final volume of 100 mL was completed with water for injection and packaged in a trilaminate bag with capacity of 300 mL (Halex Istar).
Reagents and chemical substance of reference
The reagents hydrogen peroxide PA, hydrochloric acid, sodium 1-hexanesulfonate, methanol and glacial acetic acid spectrophotometric grad were purchased from Merck (Darmstadt, Germany). The chemical substance of reference of thiamine hydrochloride (purity: 99.0%) and pyridoxine hydrochloride (purity: 99.9%) were purchased from Roche (Basel, Switzerland). The chemical substance of reference of ascorbic acid (purity: 99.0%) was purchased from Spectrum Chemical (New Brunswick, USA) and the riboflavin 5-fosfate sodium dehydrated (purity: 99.2%) from Merck. Other reagents were obtained from local suppliers.
The PN admixture formulas for neonatal use were prepared aseptically in a 300-mL 3-layered bag composed by polyester, polypropylene and polyethylene (HalexIstar, Goiânia, Brazil) under a laminar-flow hood in accordance with the National Health Department specifications Nr. 272 (1988), designed for infusion through central access. The PN admixture formulas were prepared with market products from pharmaceutical industries and based on official regulations [25–29].
The choice of composition was based in the literature recommendations of preterms . For micronutrients, today, there are controversies regarding the optimal dose of vitamins for preterms. A standard supplementation is often not enough to compensate this patient. It is necessary to use pharmacological doses that exceed the recommendations of the literature . Because of this evidence, the concentrations of the vitamins were overestimated.
Each admixture was prepared separately in triplicate or quadruplicate (referred to as 3 or 4 lots) divided into two at different temperatures: 4°C ± 2°C, in refrigerator; and in 25°C ± 3°C, simulating room temperature. A different number of samples were tested from each of the lots depending on the technique applied. The experiments were performed on the day of the admixtures preparation and also 24 h, 48 h, and 72 h after preparation, time periods indicated as D0, D1, D2 and D3 respectively.
Samples were aseptically collected from each formulation at appropriate intervals using a plastic syringe for the physicochemical tests.
For the evaluation of pH, a Mettler Toledo potentiometer calibrated with pH 4 and pH 7 buffers was used. For each measurement, a 10 mL sample was collected and placed in an amber glass flask. The pH was measured by dipping the electrode directly into the solution, at room temperature. The pH determination was carried out each day in quadruplicate for each formulation in all the conditions studied.
Development of analytical methodologies
The initial analytical conditions for the development of the chromatographic method for vitamins B1 and B6 were based in methods described in the literature for other purposes [1, 2, 21, 23]. The wavelength of maximal absorption for each vitamin was determined using individual standard solutions of vitamins B1 and B6 (8 μg/mL) for obtaining an absorption spectrum in the UV region. Later, the retention time of each vitamin in the chromatographic system was determined. For the routine analysis, a single standard solution containing the B1 and B6 was prepared, using the mobile phase as diluent. The HPLC analysis was realized using diode array detection, by the area under the peak in the wavelength of maximal absorption, 250 and 295 nm for B1 and B6, respectively. For the mobile phase selection, different proportions of methanol: water were tested, 50:50, 40:50, 27:73, 25:75 and 20:80 (v/v). The mobile phase was kept at pH 3.0 using glacial acetic acid. A Shimadzu HPLC system was used equipped with a SPD-M10APDA detector. Data was acquired with Class-VP 6.1 software. The ideal chromatographic conditions obtained for the assay of the vitamins was Bondapack C18 column size 300 mm × 13 mm, 10 μm (Waters, Milford, USA); mobile phase consisting of a mixture of methanol: water (27:73; v/v) and 1.4% of sodium 1-hexanesulfonate for the ionic par formation; flux rate of 0.35 mL/min with detection at 250 and 295 nm and 30 μL of injection volume.
The fluorescence is a common phenomenon in aromatic molecules, as vitamin B2 . Based in the method applied to the raw material vitamin B2 of the USP a selective and sensitive method to assay B2 in the PN  was developed. Initially, a preliminary study was realized to set the emission and excitation wavelength to detect the vitamin B2 in the PN. The emission range was between 400-700 nm, with intervals of 5.0 nm; with excitation at 360 nm, with 2.5 nm of interval; and scan velocity of 1000 nm/min in a Jasco fluorimeter.
Based on the reduction potential of vitamin C, the iodometric titration was applied to assay this vitamin in the PN based in the pharmacopeical methodology for raw material and tablets . The titrant used was iodine 0.05 M SV. The sample was diluted in sulfuric acid 10% (w/v; 25 mL) and a solution of starch 1% (3 mL) and distilled water in sufficient quantity for completing 100 mL were used as indicators. The end point was determined by the formation of a blue coloration. Each mL of iodine 0.05 M VS (volumetric solution) corresponds to 8.806 mg of ascorbic acid .
Validation of analysis methodologies for assaying vitamins B1, B2, B6 and C in the PN
The methodologies used for the assay of vitamins B1, B2, B6 and C in the PN were validated by the selectivity of the vitamins and their degradation products in the formulation, linearity, precision and accuracy in accordance with recommended procedures [8, 32, 33].
The specificity of the method for assay of the B1 and B6 in the presence of other formulation components of the PN was evaluated by the comparison of the chromatograms obtained from a PN containing the standard vitamins in study (B1 and B6) with PN without the vitamins (placebo). The purity determination of the chromatographic peaks was also used with the software of diode array detector. The specificity of the assay of vitamin B2 was determined by the comparison between the spectrum of PN with the standard B2, placebo and the PN, verifying that the peak observed in the spectrum is attributed to one component alone. For the vitamin C the titration of the placebo was carried out. The results obtained in the placebo were discounted later from the volume obtained with the PN containing the vitamin C.
The linearity was evaluated in three different days, by three vitamins standard with five concentration levels, in the ranges of 20-60 and 30-90 μg/mL (B1 and B6, respectively), 1-5 μg/mL (B2) and 50-150 mg (C). The linearity of the method was determined by linear regression analysis of the values obtained experimentally with the software Excel® (Microsoft, 2002).
The precision of the injection was evaluated in five concentration levels of vitamins B1 and B6: 20-60 μg/mL and 30-90 μg/mL, respectively, in three runs each. The standard deviation (SD) and the relative standard deviation (RSD) were calculated for each point, from the obtained area.
Precision was considered at two levels: repeatability and intermediate precision. It was determined by intra and inter-day assays. Stock solutions of these vitamins were prepared and aliquots were taken to prepare solutions at three levels of concentration: 80%, 100% and 120% of the sample work concentration. For B1 (30, 40 and 50 μg/mL); B6 (45, 60 and 75 μg/mL); B2 (3, 3.5 and 4.5 μg/mL); and C (80, 100 and 130 mg). The precision of the method was assessed by the SD and RSD of the values obtained experimentally in three days consecutively. The accuracy of the method was verified by determining the known recovery amount of standard vitamins in the spiked PN placebo.
Stress stability study of vitamins B1, B6, and B2 in the PN
In order to test the selectivity of the developed methodologies for assaying the vitamins studied in PN in the presence of possible degradation products, an accelerated degradation of vitamins, under forced conditions  was performed. The assay was carried out separately for each vitamin using the respective analytical method developed. For this, standard solutions were prepared, one with 40 μg/mL of vitamin B1 and 60 μg/mL of B6, and another with 3.5 μg/mL of B2. Each vitamin standard solution was stressed using 3% and 10% of hydrogen peroxide for 24 h, at room temperature. After 0, 6 and 24 h of reaction the samples were withdrawn and analysed using the respective methods.
Samples from PN were withdrawn volumetrically to carry out the vitamins assay. For B1 and B6 were used aliquots of 7 mL and for B2 0.5 mL. Each sample was transferred to a volumetric vessel of 10 mL and the volume completed with mobile phase for B1 and B6 and with PN without vitamins for the B2.
Vitamin C assay was carried out on a volumetric aliquot of 50 mL. The sample of PN was transferred to an Erlenmeyer containing 25 mL of sulfuric acid 10%, 3 mL of starch solution 1% and distilled water in sufficient quantity for 100 mL. This mixture was immediately titrated with iodine volumetric solution 0.05 M.
The experimental results obtained were presented as mean and standard deviation (SD). Comparisons between the results were evaluated by unpaired Student's t test with a 95% of confidence limit. Values of p < 0.05 were considered statistically significantly.