Caffeine (1,3,7-trimethylxanthine) is the most widely used pharmacologic stimulant in the world, with an estimated 75–90% of pregnant women and fetuses exposed through dietary intake over the course of gestation.1 Given its prevalence and neuroexcitatory profile, the potential effect of antenatal exposure on adverse pregnancy outcomes, including early pregnancy failure, impaired fetal growth, and maternal insulin resistance, has been studied extensively with conflicting results.2–6
Paraxanthine (1,7-dimethylxanthine) is the primary metabolite of dietary caffeine, formed by 3-demethylation of the parent structure and representing almost 80% of caffeine byproduct, and it has been reported to be the single marker that best reflects steady-state intake.7,8
The apparent effect of caffeine consumption on health outcomes may be a surrogate marker for interindividual variations of metabolism through the cytochrome P-450 1A2 system (CYP1A2), through which almost all 3-demethylation of caffeine to paraxanthine takes place.9,10 Neither the fetus nor the placenta demonstrates CYP1A2 activity, rendering maternal CYP1A2 the primary pathway for all caffeine clearance during pregnancy.9–11 The serum paraxanthine/caffeine ratio is an established, validated marker of CYP1A2 enzymatic activity with fast metabolizers of caffeine marked by relatively higher ratios and slower metabolizers marked by lower ratios.12–16
To define the role of caffeine in preeclampsia risk, we sought to measure the association of both antenatal caffeine intake (through maternal serum paraxanthine) and caffeine metabolism (through maternal serum paraxanthine/caffeine ratio) on the risk of subsequent severe preeclampsia. Based on the work of Bakker and Khoury, we hypothesized that higher maternal serum paraxanthine concentrations (reflecting higher consumption) and higher paraxanthine/caffeine ratios (reflecting faster metabolism) would be associated with a lower risk of severe preeclampsia.
MATERIALS AND METHODS
We conducted a nested case–control study comparing second-trimester serum paraxanthine concentrations and paraxanthine/caffeine ratios among women with and without severe preeclampsia within a cohort of women who underwent serum testing for fetal aneuploidy and subsequently delivered at the University of North Carolina, Chapel Hill from January 2004 through November 2008. These nonfasting blood samples were previously collected for routine genetic multiple marker screening between 15 and 20 weeks of gestation, and serum aliquots were barcoded and frozen at −70°C. Maternal demographic and pregnancy outcome data were chart-abstracted. Approval from the University of North Carolina, Chapel Hill institutional review board was obtained before data collection.
Cases of severe preeclampsia were defined by the following criteria: systolic blood pressure of at least 160 mm Hg or a diastolic blood pressure of at least 110 mm Hg, recorded on at least two occasions 6 hours apart plus proteinuria (300 mg or greater in a 24-hour collection or 1+ on a urine dipstick) or a systolic blood pressure of at least 140 mm Hg or a diastolic blood pressure of at least 90 mm Hg recorded on at least two occasions 6 hours apart plus 5 g of proteinuria in a 24-hour period. We further classified cases of preeclampsia as severe in the setting of pulmonary edema, seizures, oliguria (less than 500 mL per 24 hours), elevated liver enzymes accompanied by right upper quadrant pain, thrombocytopenia (less than 100,000/mm3), or persistent cerebral symptoms including headache or blurry vision.17
Controls were healthy women who delivered at term (37 weeks of gestation or greater) deliveries and who did not have preeclampsia. We excluded all patients with significant medical comorbidities (including chronic hypertension), multiple gestations, or fetal anomalies. We also excluded patients with preterm deliveries, either iatrogenic or spontaneous, because the underlying process prompting the preterm delivery may represent an unmeasured but significant confounder. A single investigator reviewed all medical charts retrospectively for case and control eligibility.
Maternal serum paraxanthine and caffeine concentrations were measured using high-performance liquid chromatography at the University of Utah's Center for Human Toxicology with a lower limit of quantification of 10 ng/mL. Maternal serum was thawed at room temperature and a 0.1-mL aliquot was immediately transferred for extraction. Internal standards (final concentration, 400 ng/mL each) were then added to each sample, control, and calibrator. Tubes were vortexed for 5 seconds, the pH of the sample was adjusted with the addition of 1 mL of 0.1 M sodium acetate buffer, and 3 mL of ethyl acetate was then added. Samples were extracted for 30 minutes on a tube mixer, centrifuged for 15 minutes, and the organic layer transferred to a clean tube. Samples were evaporated to dryness under nitrogen and reconstituted in mobile phase before analysis. The extracted analytes and their internal standards were analyzed by liquid chromatography with electrospray ionization tandem mass spectrometry. Chromatographic separation was achieved on an Agilent Zorbax SB-C18, 3.5-micron, 2.1×100-mm high-performance liquid chromatography column and a thermostated column compartment (40°C). The mass spectrometer was operated in selected reaction monitoring mode. The concentration of each analyte was determined from the ratio of the peak area of the drug to the peak area of its internal standard and comparison of this ratio with the calibration curve that was generated from the analysis of human oral fluid fortified with known concentrations of caffeine, its metabolite, and their internal standards. Interassay coefficients of variation ranged from 3.5% to 11.7%. The laboratory was masked to case or control status.
The case and control samples used in this analysis have been previously analyzed in other case–control studies of second-trimester serum analytes and severe preeclampsia risk.18,19 In the original case–control study by Baker and colleagues, 51 patients (from a population of 3,992 participants; 1.3% disease prevalence) were frequency matched by race–ethnicity in a one-to-four ratio to 204 healthy women. Our participants represent the 33 remaining samples with sufficient serum for paraxanthine assay and 99 women as a control group selected at random from the original study's control population. Of note there were no significant demographic differences between the 33 patients with sufficient serum for assay and the 18 without. With this one-to-three matching, we estimated that we would have 80% power to detect a 0.61 standard deviation (SD) difference in serum paraxanthine concentrations between women in the case group and those in the control group with an α of 0.05.
Statistical analyses were performed using SAS 9.2. For values below the lower limit of quantification (10 ng/mL) for either paraxanthine or caffeine, we imputed values of 5 ng/mL. We further calculated paraxanthine/caffeine ratios as an index of caffeine metabolism for all patients with both paraxanthine and caffeine concentrations above the lower limit of quantification (25 women in the case group and 87 in the control group). This was achieved by dividing the absolute paraxanthine level by the absolute caffeine level, both in ng/mL, for each patient.
Distributions of paraxanthine/caffeine ratios, paraxanthine, and caffeine were assessed for normality; if nonnormal, we tested log, square root, and inverse transformations. We selected the transformation that resulted in the most normal distribution, assessed by visual inspection and the Shapiro-Wilk test.
We used t tests, Wilcoxon-Mann Whitney, χ2, and Fisher's exact test where appropriate to compare values between women in the case group and those in the control group. We used logistic regression to quantify the strength of associations between measured analytes and severe preeclampsia. We further used multivariable logistic regression to test whether race–ethnicity, body mass index (BMI, calculated as weight (kg)/[height (m)]2), age, smoking status, and parity modified the association between serum paraxanthine or paraxanthine/caffeine ratios and severe preeclampsia. Because our sample size was small, we constructed a parsimonious model by entering potential confounders into the model sequentially, ranked by strength of association with the primary outcome in bivariate analyses. Covariates that were independent predictors of the outcome (Partial F P<.05) or that altered the parameter estimate for paraxanthine<assay or SD of log serum caffeine, log paraxanthine or log paraxanthine/caffeine ratio by more than 10% were retained in the model. To test whether paraxanthine/caffeine ratio was associated with timing of onset of preeclampsia, we used Pearson correlation to test whether paraxanthine/caffeine ratio was associated with gestational age at birth among women in the case group. In sensitivity analyses, we tested whether exclusion of smokers or inclusion of participants with paraxanthine or caffeine less than the lower limit of quantification altered our results. All P values were two-tailed with P<.05 considered statistically significant.
There were no significant demographic differences between women in the case group and those in the control group except for an earlier gestational age at delivery among the women in the case group, as was expected based on the diagnosis of severe preeclampsia (Table 1). In a review of medical records, 4% of the cohort reported having smoked during the index pregnancy. Median gestational age at time of serum collection was 17 weeks (interquartile range 16–19).
We identified serum paraxanthine concentrations above the lower limit of quantification (10 ng/mL) in 25 of 33 women in the case group and 87 of 99 women in the control group (76% compared with 88%, P=.08). Median paraxanthine for all patients was 91 ng/mL (interquartile range 22–206). Median second-trimester paraxanthine was not statistically different between women in the case group and those in the control group (96.4 ng/mL compared with 38.0 ng/mL, P=.12; Table 2).
Paraxanthine, caffeine, and paraxanthine/caffeine ratios were skewed, and log transformation achieved the most normal distribution for all three measures. Log paraxanthine/caffeine ratio was normally distributed (Shapiro-Wilk P=.58), whereas log paraxanthine and log caffeine were near normal (Shapiro-Wilk P=.01 and .05, respectively). We did not identify a significant association between higher paraxanthine and lower risk of severe preeclampsia (odds ratio [OR] 0.72, 95% confidence interval [CI] 0.48–1.08 per log paraxanthine SD; Table 3). When we sequentially entered age, parity, race–ethnicity, BMI, and smoking status into our model, none of these variables were independent predictors of the outcome or significantly altered the effect size, and therefore our final model was unadjusted.
With regard to CYP1A2 enzymatic activity, the median paraxanthine/caffeine ratio for patients with caffeine and paraxanthine concentrations above the lower limit of quantification (n=112) was 0.29 (interquartile range 0.19–0.55). Women in the control group had a significantly higher median paraxanthine/caffeine ratio than women in the case group (0.37 compared with 0.23, Wilcoxon P=.02; Table 2). To determine whether caffeine metabolism through CYP1A2 was related to severe preeclampsia risk, we divided the cohort into two groups, those above and those below the median paraxanthine/caffeine ratio. We found a significant difference between the frequency of severe preeclampsia among patients with lower paraxanthine/caffeine ratios compared with those with higher ratios (32.1% compared with 12.5%, Fisher exact P=.02).
In logistic regression analysis, we found a decreased risk of severe preeclampsia with increasing paraxanthine/caffeine ratio (OR 0.53, 95% CI 0.31–0.90 per one log SD; Table 3; Fig. 1). When we sequentially entered each covariate into our model, none significantly adjusted the OR; thus, our final model was unadjusted. We found no correlation between gestational age at birth and log paraxanthine/caffeine ratio among women in the case group (R=−0.16, P=.44).
We performed two sensitivity analyses. Although smoking is known to affect CYP1A2 metabolism, we found that excluding all smokers did not materially affect our results. Because our primary analysis of paraxanthine/caffeine ratios included only those patients with both paraxanthine and caffeine concentrations above the lower limit of quantification (25 women in the case group, 87 women in the control group), we performed a sensitivity analysis including the entire cohort (33 women in the case group, 99 women in the control group). Because we imputed all values below the lower limit of quantification (10 ng/mL) to be 5 ng/mL, participants with both paraxanthine and caffeine measures below the lower limit of quantification had an imputed paraxanthine/caffeine ratio of 1. Again, we found a significant association with a decreased risk of severe preeclampsia per log SD increase in paraxanthine/caffeine ratio (OR 0.65, 95% CI 0.43–0.99).
In this study, we found that a higher second-trimester maternal serum paraxanthine/caffeine ratio, a marker for CYP1A2 enzymatic activity, was associated with a significantly lower risk of developing severe preeclampsia, suggesting that faster metabolism of dietary caffeine, rather than absolute caffeine consumption, is associated with a lower risk of disease.
Our results further refine earlier work linking higher caffeine consumption with reduced risk of preeclampsia. Khoury and colleagues20 examined the association of smoking and caffeine intake on pregnancy outcomes among a historic cohort of 191 women with type 1 diabetes mellitus followed from 1978 to 1993 and found a 70% reduction in risk of preeclampsia among those reporting ingestion of at least one 8-ounce coffee (or caffeine equivalent) per day as compared with those with no reported caffeine intake (OR 0.3, 95% CI 0.1–1.0). More recently, Bakker and colleagues21 conducted a prospective cohort study examining the effect of caffeine consumption on blood pressure metrics in pregnancy and found a reduction in the risk of preeclampsia among women consuming 180–360 mg of caffeine per day as compared with those consuming less than 180 mg per day (OR 0.63, 95% CI 0.40, 0.96). Neither of these studies quantified caffeine metabolism.
Other studies examining the relationship between caffeine consumption and risk of subsequent preeclampsia using serum markers as opposed to self-report involve theobromine, the primary methylxanthine metabolite in chocolate. In 2008, Triche and colleagues22 reported a significant reduction in preeclampsia risk with higher umbilical cord serum theobromine concentrations (adjusted OR 0.31, 95% CI 0.11–087 for lowest to highest quartiles). However, in 2009, Klebanoff and colleagues23 reported no association between maternal serum theobromine and risk of preeclampsia in a historic cohort of 2,769 women who gave birth from 1959 to 1966. Again, variations in caffeine metabolism through CYP1A2 activity were not considered in either analysis.
Our study has several strengths. All samples were collected between 15 and 20 weeks of gestation, therefore bypassing the potential confounding effect of first-trimester nausea and vomiting on a change in otherwise routine dietary habits.24,25 Furthermore, the samples were all collected at a time when there was no clinical manifestation of the preeclampsia disease process, thereby reducing the chance that subclinical disease affected caffeine consumption patterns. We chose severe preeclampsia as our outcome variable because it is more rare, more likely to cause maternal and neonatal morbidity and mortality, and associated with less diagnostic and clinical heterogeneity when compared with mild disease.
We used a case–control study design given the rarity of our outcome of interest, severe preeclampsia (1.3% prevalence in our larger cohort), as well as the expense of high-performance liquid chromatography for paraxanthine and caffeine assays. However, this study design has inherent limitations, including the potential for selection bias. Although it is possible that unmeasured differences exist between our women in the case group and those in the control group, all of our patients were nested in a larger cohort of women who were offered and consented for second-trimester maternal screening for fetal aneuploidy at our institution, thereby decreasing that likelihood. One limitation of our study is that smoking significantly alters CYP1A2 activity level,15 and our cohort was limited by very few self-reported smokers. We were also unable to correlate our serum paraxanthine concentrations with self-reported caffeine intake for comparison. Furthermore, although we assessed major known confounders, including race, BMI, and smoking status, there exists the possibility that unmeasured confounders affect the relationship between serum paraxanthine and severe preeclampsia.
In conclusion, our results suggest that absolute caffeine intake, as measured by isolated serum paraxanthine concentrations, is not associated with an altered risk for severe preeclampsia; however, faster caffeine metabolism is associated with a reduced risk of subsequent severe preeclampsia. The mechanism by which higher CYP1A2 activity, as measured by elevated paraxanthine/caffeine ratios, may be associated with a decreased risk of severe preeclampsia remains unknown. Potential explanations include genetic polymorphisms in the coding gene on chromosome 1526; that higher CYP1A2 activity itself has beneficial effects; or that higher CYP1A2 activity may be a surrogate marker for epigenetic modification through environmental exposures that also alter preeclampsia risk.
1. Adén U. Methylxanthines during pregnancy and early postnatal life. Handbook Exp Pharmacol 2011:373–89.
2. Klebanoff MA, Levine RJ, DerSimonian R, Clemens JD, Wilkins DG. Maternal serum paraxanthine, a caffeine metabolite, and the risk of spontaneous abortion. N Engl J Med 1999;341:1639–44.
3. Laughon SK, Powers RW, Roberts JM, Parana S, Catov J. Caffeine and insulin resistance in pregnancy. Am J Perinatol 2011;28:571–8.
4. Grosso LM, Rosenberg KD, Belanger K, Saftlas AF, Leaderer B, Bracken MB. Maternal caffeine intake and intrauterine growth retardation. Epidemiology 2001;12:447–55.
5. Weng X, Odouli R, Li DK. Maternal caffeine consumption during pregnancy and the risk of miscarriage: a prospective cohort study. Am J Obstet Gynecol 2008;198:279.e1–8.
6. Cook DG, Peacock JL, Feyerabend C, Carey IM, Jarvis MJ, Anderson HR, et al.. Relation of caffeine intake and blood caffeine concentrations during pregnancy to fetal growth: prospective population based study. BMJ 1996;313:1358–62.
7. Bracken MB, Triche E, Grosso L, Hellenbrand K, Belanger K, Leaderer BP. Heterogeneity in assessing self-reports of caffeine exposure: implications for studies of health effects. Epidemiology 2002;13:165–71.
8. Klebanoff MA, Levine RJ, Dersimonian R, Clemens JD, Wilkins DG. Serum caffeine and paraxanthine as markers for reported caffeine intake in pregnancy. Ann Epidemiol 1998;8:107–11.
9. Grosso LM, Bracken MB. Caffeine metabolism, genetics, and perinatal outcomes: a review of exposure assessment considerations during pregnancy. Ann Epidemiol 2005;15:460–6.
10. Kalow W, Tang BK. Use of caffeine metabolite ratios to explore CYP1A2 and xanthine oxidase activities. Clin Pharmacol Ther 1991;50:508–19.
11. Tracy TS, Venkataramanan R, Glover DD, Caritis SN; National Institute of Child Health and Human Development Network of Maternal-Fetal-Medicine Units. Temporal changes in drug metabolism (CYP1A2, CYP2D6 and CYP3A Activity) during pregnancy. Am J Obstet Gynecol 2005;192:633–9.
12. Fuhr U, Rost KL. Simple and reliable CYP1A2 phenotyping by the paraxanthine/caffeine ratio in plasma and in saliva. Pharmacogenetics 1994;4:109–16.
13. Grosso LM, Triche EW, Belanger K, Benowitz NL, Holford TR, Bracken MB. Caffeine metabolites in umbilical cord blood, cytochrome P-450 1A2 activity, and intrauterine growth restriction. Am J Epidemiol 2006;163:1035–41.
14. Fuhr U, Rost KL, Engelhardt R, Sachs M, Liermann D, Belloc C, et al.. Evaluation of caffeine as a test drug for CYP1A2, NAT2 and CYP2E1 phenotyping in man by in vivo versus in vitro correlations. Pharmacogenetics 1996;6:159–76.
15. Dobrinas M, Cornuz J, Eap CB. Pharmacogenetics of CYP1A2 activity and inducibility in smokers and exsmokers. Pharmacogenet Genomics 2013;23:286–92.
16. Cornelis MC, El-Sohemy A, Kabagambe EK, Campos H. Coffee, CYP1A2 genotype, and risk of myocardial infarction. JAMA 2006;295:1135–41.
17. Diagnosis and management of preeclampsia and eclampsia. ACOG Practice Bulletin No. 33. American College of Obstetricians and Gynecologists. Obstet Gynecol 2002;99:159–67.
18. Baker AM, Haeri S, Camargo CA Jr, Espinola JA, Stuebe AM. A nested case-control study of midgestation vitamin D deficiency and risk of severe preeclampsia. J Clin Endocrinol Metab 2010;95:5105–9.
19. Woodham PC, Brittain JE, Baker AM, Long DL, Haeri S, Camargo CA Jr, et al.. Midgestation maternal serum 25-hydroxyvitamin D level and soluble fms-like tyrosine kinase 1/placental growth factor ratio as predictors of severe preeclampsia. Hypertension 2011;58:1120–5.
20. Khoury JC, Miodovnik M, Buncher CR, Kalkwarf H, McElvy S, Khoury PR, et al.. Consequences of smoking and caffeine consumption during pregnancy in women with type 1 diabetes. J Matern Fetal Neonatal Med 2004;15:44–50.
21. Bakker R, Steegers EA, Raat H, Hofman A, Jaddoe VW. Maternal caffeine intake, blood pressure, and the risk of hypertensive complications during pregnancy. The Generation R Study. Am J Hypertens 2011;24:421–8.
22. Triche EW, Grosso LM, Belanger K, Darefsky AS, Benowitz NL, Bracken MB. Chocolate consumption in pregnancy and reduced likelihood of preeclampsia. Epidemiology 2008;19:459–64.
23. Klebanoff MA, Zhang J, Zhang C, Levine RJ. Maternal serum theobromine and the development of preeclampsia. Epidemiology 2009;20:727–32.
24. Furneaux EC, Langley-Evans AJ, Langley-Evans SC. Nausea and vomiting of pregnancy: endocrine basis and contribution to pregnancy outcome. Obstet Gynecol Surv 2001;56:775–82.
25. Rifas-Shiman SL, Rich-Edwards JW, Willett WC, Kleinman KP, Oken E, Gillman MW. Changes in dietary intake from the first to the second trimester of pregnancy. Paediatr Perinat Epidemiol 2006;20:35–42.
26. Ou-Yang DS, Huang SL, Wang W, Xie HG, Xu ZH, Shu Y, et al.. Phenotypic polymorphism and gender-related differences of CYP1A2 activity in a Chinese population. Br J Clin Pharmacol 2000;49:145–51.