Caffeine is recognized as the first-line therapeutic agent for apnea of prematurity. The dosage regimen is 10 mg/kg loading dose and 2.5 mg/kg maintenance dose. However, the plasma concentration achieved, not always, is therapeutically useful. It makes necessary to increase the doses to reach plasma concentration up to 30 or 35
Caffeine (1,3,7-trimethylpurine-2,6-dione) (Figure
Caffeine and its major metabolites and internal standard structures.
This drug has psychoactive properties. It is a slightly dissociative and stimulant drug because of its nonselective antagonist action against adenosine receptors [
Clinically, caffeine is recognized as the first-line therapeutic agent for apnea of prematurity [
The presence of multiple cytochrome P-450 isoforms (CYP450) may explain the species, strain, age, tissue, and sex differences as well as the effect of inducers, nutritional status, and human drug metabolism. In humans, caffeine is metabolized through the liver microsomal drug-metabolizing enzyme system CYP450. In neonates, the expression of this enzymatic system is dependent on their prenatal and postnatal age [
The analytical methods used for the quantification of caffeine and its main metabolites are immunoassay with monoclonal antibodies, which presents as main disadvantage the selectivity and sensitivity of the method [
Another analytical problem is to obtain caffeine-free human plasma from volunteers. Some authors have tried to use substitutes, human plasma treated with activated charcoal to remove caffeine, or the manufacture of synthetic plasma devoid of caffeine [
Recent studies have reported that intake of caffeine during pregnancy (plasma concentration estimated by self-reported consumption) is associated with the development of sickness such as obesity and hypertension, besides others [
Our group is interested in quantifying plasma levels of caffeine and its metabolites to determine what is the prenatal exposure to caffeine in our population and if this exposure has an impact on therapeutic response (apnea) or if it increases risk or predisposition to other diseases [
For this reason, initially we had to develop an analytical method to quantify caffeine and its metabolites in plasma at our laboratory.
In this study, we present the results on the use of fetal bovine serum (FBS) for the development and validation of an analytical method by HPLC for the quantification of caffeine and three of its main metabolites in samples of neonates and their mothers to perform the previously mentioned studies and results of its clinical application.
Caffeine (1,3,7-trimethylxanthine) (C0750), paraxanthine (1,7-dimethylxanthine) (D5385), theobromine (3,7-dimethylxanthine) (T4500, >98%), and 7-(
HPLC system consisted in a quaternary pump with a degasser, and it was coupled to an autosampler and DAD–UV detector (Agilent 1200 series) (Agilent Technologies Mexico, S. de R.L. de C.V.). Separation was performed on a reverse-phase column Zorbax® SB-Aq narrow bore RR (2.1 × 100 mm, 3.5
For optimization of chromatographic conditions, the effects of various method parameters such as mobile phase, column, flow rate and solvent ratio, and detection system were evaluated, and the chromatographic parameters such as asymmetric factor, resolution, and column efficiency were calculated. The best results were obtained with a mixture of 10 mM phosphate buffer, pH 6.8, and acetonitrile, in a gradient phase mobile composition obtained using a gradient program (Table
Final gradient program for HPLC sample analysis.
Time (min) |
|
|
Flow rate (ml/min) |
---|---|---|---|
0 | 97 | 3 | 0.7 |
8.5 | 97 | 3 | 0.7 |
12 | 92 | 8 | 0.7 |
13 | 97 | 3 | 0.7 |
Mobile phase A: 10 mM phosphate buffer, pH 6.8. Mobile phase B: acetonitrile 100%.
At the end of each work day, the column was washed with acetonitrile : water (90 : 10 v/v) during 30 min. Chromatographic data were processed using Chemstation for LC systems software (Agilent Technologies).
Stock solutions of Par, Theo, and Caf (4 mg/mL) were prepared separately, dissolving an appropriate amount of each drug in diluent (Milli-Q water). Theb solution was prepared at half concentration than the others (2 mg/mL) due to theobromine having the lowest aqueous solubility compared to the rest of caffeine alkaloids. This solution was prepared adding Theb to diluent (Milli-Q water) and heating and mixing the solution prior to obtain its total volume. The solution was cooled to room temperature before it was adjusted to its final volume.
Working solutions for the different points in the calibration curve or quality control samples were prepared simultaneously for all interest drugs in water. Those solutions were obtained mixing the necessary reagent volume to attain different concentrations; 1, 3, 5, 10, 25, 50, 100, 200, 300, and 400
IS working solution (7-(
Seven level calibration curves were constructed by spiking water or drug-free FBS with known amounts of Caf, Theo, Par, and Theb to reach concentrations of 0.1, 0.3, 1, 2.5, 10, 20, and 40
The analytical method was developed, validated, and challenged in a clinical trial. The clinical study was focused to quantify prenatal caffeine exposure, and later the analytical method will be applied to study clinical pharmacokinetics of caffeine in preterm neonates. This study reports only the results of the pilot trial of prenatal caffeine. The clinical protocol was reviewed and approved by our Institutional Research and Ethics Board registered according to Mexican law to authorize and oversee the conduction of clinical trials (13CI19039138 and CONBIOETICA-19-CEI-011-20161017). The protocol was conducted in accordance with the Declaration of Helsinki. All participants (fathers, mothers, or legal responsible) agreed to be included in the study, by signing an informed consent form.
Inclusion criteria are as follows: women 18 years old or older, pregnant with single or multiple products, gestational age less than 34 completed weeks (based on dates and confirmed with an ultrasound examination), and without a history of preeclampsia or any neonatal abnormality.
We excluded women with known fetal genetic or major malformations, very critical condition, or fetal demise.
Blood samples (1 mL) were collected during the 15 minutes immediately after delivery from umbilical cord (neonates) or from venous puncture (mothers). Samples were collected in heparinized Vacutainer® tubes and centrifuged for 10 min at 3500 rpm under a controlled temperature of 10°C. The plasma supernatant was carefully transferred to two polypropylene tubes and frozen at −60°C until they were analyzed.
One hundred microliters of the IS working solution was added to 150
The sample pretreatment procedure was carried out on plasma by means of solid-phase extraction (SPE) on polymeric 96-well plates Strata-X™ (30 mg, 1 mL) (Phenomenex®, Torrance, CA). The plate was placed in a 96-well plate vacuum manifold system (Phenomenex, Torrance, CA). The vacuum pressure was adjusted between −15 and −20 mmHg, and each well was activated by washing with 1 mL methanol followed by 1 mL water.
One milliliter of each sample supernatant (previous step) was placed on its corresponding well and allowed to pass through low vacuum (−5 to −6 mmHg). The wells were then washed with 1 mL 5% methanol at vacuum pressure between −15 and −20 mmHg. Caffeine and its metabolites were eluted with 1 mL methanol-2% acetic acid (70 : 30) solution. The eluent was dried under N2 gas at 40°C. The residue was dissolved in 50
Validation was carried out following the criteria established in the Mexican regulatory guidelines [
Results of caffeine and metabolites for prenatal exposure pilot study were expressed as plasma concentration for mothers and newborns. Plasma concentration differences by each mother/child pair (M/C) were calculated and used to construct a Bland–Altman plot for each compound. The differences between each M/C pair plasma concentrations were analyzed for means difference with Student’s
We tested different analytical methods reported for caffeine and metabolite quantification [
It is well known that fluorescence is more sensitive than UV for analysis of various compounds. However, fluorescence sensitivity to detect caffeine in biological samples remains a challenge.
At the beginning of our experimental process, we found a report that described fluorescence parameters for caffeine analysis. We obtained the same values of fluorescence spectrum by scanning a caffeine aqueous solution at 40
We believe that another problem could be related with the caffeine fluorescence in biological samples. Some researchers have described that caffeine and its metabolites (theophylline and theobromine) can quench fluorescence signals for proteins, amino acids, and hemoglobin [
Recently, a “light traffic” detector has been developed to improve caffeine fluorescence detection. This device uses a compound called “caffeine orange” to conjugate caffeine and increase about 250-fold the fluorescence signal. However, the main setback is that it is not selective and can increase fluorescence for caffeine, theobromine, theophylline, and other methylxanthines. This result suggests that fluorescence intensity signal by caffeine is too low for direct detection and needs derivatization to increase response [
Conversely, the HPLC/UV method had enough sensitivity for 0.1
Representative chromatograms of FBS blank (a), FBS + IS (b), FBS spiked with caffeine and metabolites at all calibration curve levels lower (0.1
The calibration curves were linear over the concentration range of 0.1–40
Typical calibration curves for quantification of theobromine (a), paraxanthine (b), theophylline (c), and caffeine (d) in human plasma. Theobromine (-○-), paraxanthine (-♦-), theophylline (-◊-), and caffeine (-▲-).
For all analytes, LLOQ in BFS was 0.1
The optimal extraction recovery was obtained with SPE cartridges as it was described previously. To calculate absolute recovery of caffeine, metabolites, and IS, 5 sets of samples (0.5, 5, and 30
Results of intra- and interday variability during validation of the HPLC method. Accuracy is expressed as percentage of the nominal value. Precision is expressed in terms of the percentage of the coefficient of variation.
Concentration ( |
Interday, |
Intraday, |
||||||
---|---|---|---|---|---|---|---|---|
Theb | Par | Theo | Caf | Theb | Par | Theo | Caf | |
0.5 | 103.7 (8.8) | 96.6 (4.0) | 96.5 (2.3) | 100.73 (8.0) | 101.2 (8.7) | 97.1 (3.5) | 97.5 (4.1) | 98.4 (6.3) |
5 | 100.9 (5.7) | 103.2 (4.9) | 99.0 (4.6) | 102.2 (6.6) | 100.6 (6.8) | 102.1 (5.0) | 99.9 (5.0) | 103.4 (5.6) |
30 | 101.9 (7.5) | 102.1 (6.4) | 103.2 (5.5) | 105.2 (5.7) | 102.9 (8.7) | 101.0 (5.5) | 103.0 (5.4) | 106.2 (4.8) |
Interday results are the accumulation of determination of 3 quality control samples in three different days for each concentration level.
No differences were observed between relative recovery of caffeine and its metabolites from FBS samples compared with human plasma assessed by Bland–Altman analysis and Student’s
Differences between concentrations of caffeine and metabolites in human plasma or fetal bovine serum spiked sample.
Concentration ( |
Human plasma mean (CV) | Fetal bovine serum mean (CV) |
| ||||
---|---|---|---|---|---|---|---|
0.5 | 5.0 | 30.0 | 0.5 | 5.0 | 30.0 | ||
Theb | 0.50 (3.4) | 5.21 (3.4) | 31.49 (4.1) | 0.52 (8.8) | 5.04 (5.7) | 30.57 (7.5) | NS |
Par | 0.50 (2.6) | 5.01 (4.4) | 30.92 (7.4) | 0.48 (4.0) | 5.16 (4.9) | 30.62 (6.3) | NS |
Theo | 0.49 (4.2) | 4.98 (4.6) | 31.10 (5.2) | 0.48 (2.3) | 4.95 (4.6) | 30.95 (5.5) | NS |
Caf | 0.52 (6.7) | 5.03 (7.0) | 32.28 (4.9) | 0.50 (8.0) | 5.11 (6.6) | 31.57 (5.7) | NS |
NS: no differences besides groups for each analyte at the same concentration.
Caf, Theo, Theb, Par, and IS proved to be stable in biological samples for at least 6 months at −60°C (final mean recovery of 101.7, CV 5.7%); samples were also stable for at least two freeze and thaw cycles (98%; CV 4.3%) and for at least 8 h on the worktable at room temperature (25°C; 96.5%; CV 4.5%). This was enough time according to our preparation sample method. Analytes were also stable for at least 13 h in the autosampler (101.5%; CV 6.0%), which made it possible to analyze about 50 samples in a row.
There were no endogenous compounds that interfered either with caffeine or its metabolites or with IS, even in hemolyzed or jaundiced samples. Other results of selectivity evaluation demonstrated that there were no interfering peaks from any of the following drugs: acetaminophen, ibuprofen, erythromycin, furosemide, or heparin, which are commonly used during neonatal pharmacotherapy.
The trial was conducted from July 2013 through December 2013 in three medical centers located in Northeastern Mexico: the Hospital Regional Materno Infantil (Center 1), the Hospital Metropolitano Dr. Bernardo Sepúlveda (Center 2), and the Hospital Zambrano-Hellion (Center 3). The first two are public hospitals that harbor level II-III neonatal intensive care units and belong to the Servicios de Salud Network in Nuevo León, Mexico. The last is a private hospital with a level III neonatal intensive care unit and belongs to the academic and research health branch of Tecnológico de Monterrey in Monterrey, Nuevo León, Mexico. Seventy-three pregnant females that were admitted to the obstetric ward during labor were eligible to participate. Forty-six were excluded for different reasons: fetal death (19), malformation (1), or transfer to another unit (4), and 22 eligible mothers who declined the invitation to participate. Thus, only 27 pregnant volunteers participated in this study.
In our study, we were able to demonstrate the maternal consumption of caffeine in 85.2% of the volunteers (
Analyte | Mean difference | SE | 95% CI difference | Diff (mother-neonate) versus zero | |
---|---|---|---|---|---|
Theobromine | −0.1462 | 0.087 | (−0.327 to 0.035) |
|
NS |
Paraxanthine | −0.0352 | 0.037 | (−0.113 to 0.042) |
|
NS |
Theophylline | −0.0044 | 0.028 | (−0.064 to 0.055) |
|
NS |
Caffeine | −0.0194 | 0.097 | (−0.221 to 0.182) |
|
NS |
Caffeine and its metabolites were measured by our HPLC method; CI: confidence interval; Diff: difference; SE: standard error.
Bland-Altman plot for theobromine (a), paraxanthine (b), theophylline (c), and caffeine (d) plasma concentrations for mother versus newborn samples. Differences between mother-child pairs. Continuous line represents mean of difference between maternal concentration and neonatal concentration for each analyte. Dotted lines represent lower and upper limits of 95% confidence interval. Theb: Theobromine, Par: paraxanthine, Theo: theophylline, and Caf: caffeine.
For Caf in maternal and newborn plasma concentration, there was a positive linear regression (Figure
Linear regression between plasma caffeine concentrations in mother versus newborn samples.
We developed a rapid and sensitive analytical method for the simultaneous determination of caffeine and its three principal metabolites in adult and newborn human plasma using HPLC technique after solid-phase extraction with Strata-X™ plates. This method had an acceptable accuracy (96.5–105.2% for intraday and 97.1–106.2% for interday) and precision with a CV less than 10% in all concentration levels used as quality control samples for all analytes. This method is suitable for use in individuals in which blood extraction volumes are limited (150
The authors report no conflicts of interest.
This work was supported by Tecnologico de Monterrey Fund, GEE Investigación en Cáncer (Cátedra Hematología y Cáncer) and GEE Genética Humana (Cátedra Crecimiento y Desarrollo del Ser Humano). The authors would like to thank Juan Carlos Carrazco Arroyo for his technical assistance in improving the quality of chromatogram images.