4.2. Human Milk Samples
Human milk samples were provided voluntarily by breastfeeding mothers after intake of the non-prescription pharmaceutical Soledum® (Klosterfrau Healthcare Group, Cologne, Germany). Before the beginning of the study, the participants were informed verbally, as well as in written form, about the study procedure and purpose and they were told that withdrawing from the study was possible at any time without giving any reason. All volunteers gave written consent to the analysis of their milk samples and anonymous publication of the resulting data. The Ethical Committee of the Medical Faculty, University Erlangen-Nuremberg, approved the experimental procedures. Each participant was handed a copy of the Soldeum® package instructions and was advised to read it before capsule intake. They were also advised to withdraw from the study if any negative reactions such as medication allergies or interactions with other drugs they were taking were to be expected. Additionally, the volunteers were informed verbally about potential effects of the pharmaceutical preparation on breastfeeding, i.e. the risk of an altered milk flavour and rejection of the milk by the nursling. Accordingly, if this would happen, the mothers were recommended to feed their babies with stored human milk or formula milk products.
The milk donors were between the ages of 29 and 40 years (mean 34 years). During their participation in the study, their breast milk production had to be normal and in excess of their infants' needs, with no breast infections and no overall medical complaints present. Five participants were primiparous, three were multiparous and the breast milk samples were from 19 weeks to 19 months postpartum. For sampling, either electrical or mechanical breast pumps or manual expression techniques were applied, resulting in sample volumes between 10 and 100 mL. The samples were immediately subjected to sample preparation after a maximum storage time of five hours at 7 °C or were stored in a freezer at −20 °C for up to five days.
The participating volunteers were asked to ingest one Soledum® capsule, containing 100 mg 1,8-cineole, and to wait until they themselves or another person began to notice the eucalyptus-like odour on their breath. Only then a milk sample should be expressed, within the next two to four hours, because previous studies [
39] showed that the transfer of 1,8-cineole into human milk coincided with the first signs of the eucalyptus-like odour in the breath. The donors were instructed to keep notes on the dates of capsule intake, first eucalyptus-like odour on their breath and exact sampling time. Unfortunately it was difficult for some mothers to detect the smell on their breath so they expressed the samples after some time without having noticed the smell, or reported having been unsure about the smell. This resulted in several samples that did not show the typical eucalyptus-like odour. Some of the donors repeated the whole experiment and some expressed more than one sample after the capsule intake, resulting in 14 samples in total, from 8 different mothers.
4.4. Gas Chromatography-Mass Spectrometry (GC-MS)
For GC-MS analyses, an Agilent MSD quadrupole system (GC 7890A with MSD 5975C, Agilent Technologies, Waldbronn, Germany) coupled with a CIS 4C (cooled injection system) and an MPS 2 auto-sampler (Gerstel GmbH & Co. KG, Duisburg, Germany) were used. Data collection and analysis were performed with the MSD ChemStation E.02.00.493 (Agilent Technologies, Inc.) software. Quantification experiments were carried out on DB-FFAP and DB-5 capillaries (30 m length, 0.25 mm inner diameter, 0.25 µm film thickness, Agilent J&W GC Columns, USA), and for monitoring of the ratio of enantiomers the chiral capillaries Rt-bDEXsa, Rt-bDEXse, Rt-bDEXsm and Rt-yDEXsa (30 m length, 0.32 mm inner diameter, 0.25 µm film thickness, Restek GmbH, Bad Homburg, Germany) were utilised. Pre-columns of uncoated, deactivated fused silica (0.53 mm inner diameter, 0.5 to 3.5 m length) were used to ensure that no potentially present low-volatile residues could contaminate the analytical capillaries. The temperature programme of the column oven was as follows: 40 °C for 2 min, 4 °C/min up to 170 °C, 40 °C/min up to 230 °C (Rt-DEX capillaries), 240 °C (DB-FFAP) and 250 °C (DB-5), respectively, and hold time at end temperature of 5 to 10 min. As carrier gas, helium was used at a flow rate of 1.3 mL/min (velocity 41.1 cm/s). Injection of 1 or 2 µL was performed by the auto-sampler.
As transfer line into the MS, an uncoated, deactivated fused silica capillary (60 cm length, 0.25 mm inner diameter) was connected to the end of the analytical capillary and heated to 200 °C. Temperatures of the ion source and the quadrupole were 200 °C and 150 °C respectively. The CIS was programmed to the same temperatures as the column oven, so that cold-on-column injections were possible. Mass spectra were generated by electron ionisation (EI) at 70 eV ionisation energy and recorded in full scan mode (m/z range 40 to 250).
4.5. Determination of the Ratios of Enantiomers
For the separation of the pairs of enantiomers of the chiral metabolites of 1,8-cineole, four different capillaries (Rt-bDEXsa, Rt-bDEXse, Rt-bDEXsm, and Rt-yDEXsa) were evaluated for optimisation of the peak-to-peak resolution of the enantiomers by injecting a mixture of all the metabolites in a dichloromethane solution and by varying the heating rate. Besides the separation of each pair of enantiomers from each other, it was also important to confirm that no other metabolite coeluted with any one of the enantiomers, because the milk samples would most probably contain all metabolites. Moreover, the separation conditions for the mixture of reference substances were also tested on some human milk sample extracts and modified accordingly, where necessary, to avoid coelution with interfering matrix compounds.
For calculation of the ratio of enantiomers (that is, the relative abundance of each enantiomer to the not separated mixture of enantiomers), the peak area of each enantiomer was divided by the sum of the peak areas of both enantiomers and multiplied by 100. With the known concentrations of the chiral metabolites in each milk sample, the absolute concentration of each single enantiomer was also calculated by multiplying the concentration with the respective percentage.
4.6. Quantification with Internal Standard and Matrix Calibration
Human milk samples were spiked with deuterium labelled 1,8-cineole (solution in dichloromethane) as internal standard prior to SAFE distillation. The amount of internal standard added was varied for each sample according to the sample volume and the subjective perception rating of the eucalyptus-like smell as a sign of the efficiency of transfer (~0.1–0.2 µg/g for samples with detectable odour and 100-fold less for odourless samples). Sample extracts were measured on both GC capillaries DB-FFAP and DB-5; thereby, the signal ratios of internal standard and analyte on DB-5 were used for quantification of 9-hydroxy-1,8-cineole while for all other metabolites the chromatograms on DB-FFAP were utilised. The peak areas of the metabolites were from extracted ion chromatograms of the m/z values given in
Table 4 and the peak area of labelled 1,8-cineole was from the extracted ion chromatogram of m/z 157.
To set up the calibration curves, spiked matrix samples were used, taking into account the sample preparation. Therefore at first seven calibration spiking solutions in dichloromethane were prepared with descending concentrations of the following metabolites: 2,3-Dehydro-1,8-cineole, α2,3-epoxy-1,8-cineole, α2-hydroxy-1,8-cineole, β2-hydroxy-1,8-cineole, 4-hydroxy-1,8-cineole, 7-hydroxy-1,8-cineole and 9-hydroxy-1,8-cineole. 50 µL of each spiking solution containing all seven metabolites were added to ~50 mL human milk together with 35 µL of a solution of labelled 1,8-cineole before sample preparation via SAFE. This resulted in concentration ratios of metabolite to labelled 1,8-cineole of about 7:1, 3:1, 1:1, 0.6:1, 0.1:1, 0.05:1 and 0.01:1 for the respective calibration level (absolute amount of each metabolite in the milk samples between 65 and 0.02 µg; absolute amount of internal standard ~8 µg).
Table 4.
Calibration curves used for quantification of metabolites of 1,8-cineole.
Metabolite | m/z for Data Analysis | Equation of Calibration Curve | Correlation Coefficient |
---|
2,3-dehydro-1,8-cineole (low rangea) | 109 | y=6.5814x-0.0004 | 0.9988 |
a2,3-epoxy-1,8-cineole (low rangea) | 95 | y=1.8892x-0.0002 | 0.9996 |
a2-hydroxy-1,8-cineole (high rangeb) | 108 | y=2.0255x-0.0069 | 0.9938 |
a2-hydroxy-1,8-cineole (low rangea) | 108 | y=1.8600x-0.0007 | 0.9988 |
7-hydroxy-1,8-cineole (low rangea) | 111 | y=0.2487x-0.0001 | 0.9926 |
9-hydroxy-1,8-cineole (low rangea) | 139 | y=9.4586x-0.0001 | 0.9961 |
4-hydroxy-1,8-cineole (high rangeb) | 112 | y=2.1108x-0.0282 | 0.9996 |
4-hydroxy-1,8-cineole (low rangea) | 112 | y=1.9539x-0.0010 | 0.9965 |
2-oxo-1,8-cineole (low rangea) | 82 | y=4.5637x-0.0002 | 0.9995 |
3-oxo-1,8-cineole (high rangeb) | 153 | y=15.6904x-0.0002 | 0.9964 |
3-oxo-1,8-cineole (low rangea) | 153 | y=15.9449x-0.0001 | 0.9995 |
a3-hydroxy-1,8-cineole (low rangea) | 108 | y=15.2720x-0.0001 | 0.9962 |
The human milk used as calibration matrix was mixed from several smaller samples from different mothers to get a final volume of about 500 mL of which ~50 mL were taken for each calibration point. Additionally, ~50 mL were subjected to sample preparation without spiking with metabolites but only with the labelled 1,8-cineole as internal standard. This blank sample was used as a control for potential naturally occurring traces of metabolites. Altogether three mixed samples of human milk were each used for seven calibration points and the blank sample, resulting in three calibration samples for each level.
After measurement of the calibration samples, signal ratios of labelled 1,8-cineole to the respective metabolite were calculated from the respective chromatograms. Values of the blank sample were subtracted from those of the calibration samples. The signal ratios of the three calibration points of the same level were averaged and a calibration curve for each metabolite was calculated by linear regression.
For calculation of β2-hydroxy-1,8-cineole in the samples, the calibration curve of α2-hydroxy-1,8-cineole could be used because these metabolites are diastereomers and, thus, are expected to show equal behaviour during sample preparation and mass spectrometry and are differentiated only by their retention times on achiral GC-capillaries.
One sample had a very high ratio of α2-hydroxy-1,8-cineole to internal standard and thus the calibration range was extended for this sample with a calibration point at the exact ratio of 1:20.03 for metabolite to labelled 1,8-cineole (calibration sample spiked with 157.76 µg of α2-hydroxy-1,8-cineole and labelled 1,8-cineole as described above). This calibration point was also prepared in triplicate.
For α3-hydroxy-1,8-cineole and 3-oxo-1,8-cineole, reference substances were only available in very small amounts and in solutions with unknown concentration, thus no calibration could be undertaken as with the other metabolites. Nevertheless, the estimated concentrations of these substances were calculated in the analysed milk samples by using the calibration curves of the closely related metabolites α2-hydroxy-1,8-cineole and 2-oxo-1,8-cineole under the assumption of equal response in mass spectrometric detection. This assumption is only valid when comparing peak areas in total ion chromatograms because of the different fractionation patterns of each metabolite. Since the resolution was not sufficient in total ion chromatograms, all peak areas were determined from extracted ion chromatograms. Accordingly, these areas had to be mathematically converted to respective total ion chromatogram areas. This was realised by multiplication with a factor for the ratio of the total ion peak area, and the extracted ion peak area, determined from more than 20 independent measurements of the respective metabolite reference substance.
To improve linearity, the range of seven calibration points was divided into two ranges, including only the five highest and the five lowest calibration points respectively. For some metabolites, the calculated signal ratios of the samples were only in the range of the five lowest calibration points, and thus the respective calibration curve for the higher range is not reported.
Further optimisation of the calibration was carried out by varying several parameters and comparing the resulting calibration curves: Different weighting factors were tested, the area ratio was set alternatively as y-value or as x-value, the signals on DB-FFAP or on DB-5 were used and the worst matching calibration sample out of the three calibration points of the same level was excluded or left included. Finally, the best calibration equation for each metabolite was chosen based on the following criteria: With each calibration curve, concentration ratios were calculated from the measured signal ratio of the calibration samples and compared with the nominal concentration ratios. The calibration equation leading to the smallest deviation (
i.e., to the highest reovery rate) was chosen for final quantification of the human milk samples. The calibration equations are given in
Table 4.