Secondary Logo

Journal Logo

Reproduction

Maternal Serum Paraxanthine During Pregnancy and Offspring Body Mass Index at Ages 4 and 7 Years

Klebanoff, Mark A.; Keim, Sarah A.

Author Information
doi: 10.1097/EDE.0000000000000220

Abstract

Caffeine is among the most widely consumed pharmacologically active food substances; 80% of adults in the United States consume caffeine daily,1 and caffeine is commonly consumed by pregnant women.2–4 Caffeine consumption during pregnancy has been variably associated with increased risks of spontaneous abortion5,6 and birth of a small-for-gestational age (SGA) newborn.5,7 In spite of an extensive literature on the associations between caffeine consumption and pregnancy-related outcomes, there have been few studies on the association between caffeine use during pregnancy and outcomes in children beyond the immediate neonatal period.

In a recent report, maternal caffeine consumption during pregnancy was associated with a substantially increased risk of obesity among the offspring.8 Specifically, maternal consumption of ≥150 mg of caffeine per day during pregnancy was associated with an odds ratio for obesity in the child of 2.1 (95% confidence interval = 1.22–3.50) as compared with no caffeine intake; the association was stronger for what the authors referred to as “persistent,” rather than “transitory,” obesity. Since 18% of U.S. children were obese in 20129 and caffeine is commonly consumed during pregnancy, if this association was causal, then a considerable fraction of childhood obesity might readily be prevented by reduction of caffeine intake during pregnancy. However, the mechanism for the association is obscure—although, as noted, maternal caffeine use during pregnancy has been associated with SGA birth, and SGA infants have been observed in some studies to have reduced lean mass compared with non-SGA infants of comparable size.10 A rodent model also documented 23% increased offspring weight at 3 weeks of age associated with maternal caffeine consumption and 10% increased weight at 8 to 10 weeks of age, corresponding to young adulthood.11 Nevertheless, this unexpected finding requires replication. The purpose of this report is to describe the association between maternal serum concentration of paraxanthine, caffeine’s primary metabolite in humans, at 2 points in pregnancy, and the child’s body mass index (BMI) at 4 and 7 years of age in a large multisite U.S. cohort, the Collaborative Perinatal Project.

METHODS

This secondary analysis was conducted among women originally selected as controls for a case–control study, nested within the prospective Collaborative Perinatal Project, of serum caffeine metabolites and spontaneous abortion.12 This project was a prospective cohort study that recruited pregnant women at registration for prenatal care at 1 of 12 sites in the United States from 1959 to 1965 and followed them through pregnancy, labor, and delivery.13 Serum was collected at registration, approximately every 8 weeks during pregnancy, at delivery, and at 6 weeks postpartum and was stored in glass vials at −20°C with no recorded thaws until analysis in 1997–1999. As controls, the women gave birth to liveborn infants of at least 28 completed weeks’ gestation (considered to be the limit of viability in the Collaborative Perinatal Project era) and had a serum sample drawn at <140 days’ gestation, on the same gestational day as a women from the same study site who subsequently experienced a spontaneous abortion. In addition, the first serum sample obtained after 26 weeks was analyzed, originally to assess the association between caffeine metabolites and size-for-gestational age.14 Therefore, serum paraxanthine was assessed at <20 weeks’ and at ≥26 weeks’ gestation. Paraxanthine was chosen as the marker of exposure because it was shown to correlate more strongly than serum caffeine with concurrently obtained 24-hour recalled caffeine intake among pregnant women,15 and because in persons displaying usual caffeine consumption patterns, serum paraxanthine varies less than serum caffeine in response to acute caffeine ingestion.16

The Collaborative Perinatal Project children were followed up to age 7 or 8 years.17,18 There were study visits at 48 and 84 months (as well as at other times) when children were weighed and measured. Height was measured to the nearest 0.5 cm using a standard backboard, and weight was measured in pounds to the nearest ¼ pound or grams to the nearest 100 g on scales calibrated semiannually.19 Study outcomes were assessed at the 48-month (range = 46–50) and 84-month (82–86) visits, as the strong clustering of measurements at these times and small number at other times precluded modeling individual growth trajectories20; if a child had 2 measurements in a window, the first was used. Sex-specific BMI-for-age z-scores and percentiles were calculated from the Centers for Disease Control (CDC) year 2000 growth charts.21 Study outcomes included mean z-score, percent with BMI at least the 85th percentile, and percent with BMI at least the 95th percentile for age and sex. In addition, among the subset of children whose BMI was measured at both 48 and 84 months, we evaluated the association of paraxanthine with persistent obesity, defined as BMI at least the 95th percentile at both the 48- and 84-month visits as compared with children who were not obese at either visit.

Serum concentrations of caffeine, paraxanthine, theophylline, and theobromine were assayed by high-performance liquid chromatography15; the limits of detection and quantitation for paraxanthine were 25 and 50 μg/liter, respectively. Actual machine assay values were not available, and so we assumed values of 0 and 37 for samples below the limits of detection (14% of early-pregnancy and 9% of third-trimester samples) and quantitation (3% of early-pregnancy and 3% of third-trimester samples). When we assumed a value of 12 for values below the detection limit, the results were essentially identical.

The association between paraxanthine and BMI z-score was evaluated by linear regression, and associations with BMI ≥85th and ≥95th percentile were evaluated by log-binomial22 or modified Poisson23 regression. In all cases, serum paraxanthine was considered as a restricted cubic spline with 3 knots; significance was assessed by the Wald test. When the P value for the spline term for paraxanthine in an adjusted model was >0.05, that analysis was re-examined with paraxanthine entered as a simple continuous term for parsimonious evaluation of a linear trend.

When building adjusted models, characteristics that were continuous were entered as such although these variables were categorized for ease of presentation when describing the population in Table 1. All models were adjusted for potential confounders selected a priori, including race and diabetes before or during pregnancy as categorical variables and maternal age, education, cigarettes smoked per day at enrollment, prepregnancy weight, and gestation at blood draw as continuous variables. The Collaborative Perinatal Project collected data neither on maternal or childhood diet or physical activity nor on breastfeeding beyond the immediate hospital nursery stay. Models including terms for the study sites failed to converge for several of the analyses of BMI ≥95th percentile, but results for the remaining outcomes were not substantially different when the terms were included. All continuous variables except maternal age and gestation at blood draw were entered as restricted cubic splines with 3 knots. Maternal prepregnant weight was chosen in preference to BMI because the use of BMI resulted in the loss of up to 88 more observations than when we used weight; results were not substantially different among women with data available when BMI was substituted for weight. Approximately 3% of observations were subsequent pregnancies to the same woman; therefore, generalized estimating equations were used to provide robust standard errors.24

TABLE 1
TABLE 1:
Serum Paraxanthine Concentration (Microgram/Liter) at <20 and ≥26 Weeks’ Gestation and Child Body Mass Index at 48 and 84 Months in Relation to Maternal Characteristics, Collaborative Perinatal Project, 1959–1974

A sensitivity analysis evaluated the potential impact of loss to follow-up. The probability of successful follow-up was assessed in 4 logistic models (for each combination of blood draw and examination) that included terms for the above-noted confounders plus serum paraxanthine, child birthweight and gestational age, study site, and whether follow-up was successful at age 7 years (for models of 4-year follow-up) or 4 years (for 7-year follow-up). Adjusted analyses were redone weighting each individual by the inverse of their probability of successful follow-up. Analyses were conducted by SAS (SAS Institute, Inc., Cary, NC), version 9.3, and a macro for splines25 that was modified to yield relative risks rather than odds ratios and to incorporate weights. Graphics were prepared using STATA (StataCorp LP, College Station, TX), version 13. This analysis of previously collected data without personal identifiers was deemed exempt from review by the Nationwide Children’s Hospital Institutional Review Board.

RESULTS

The original case–control study included 2816 control pregnancies of at least 28 weeks’ gestation, 2808 of which had at least 1 sample analyzed for methylxathines. Exclusion of 35 twin pregnancies, plus 8 samples with missing paraxanthine, resulted in 2765 pregnancies, of which 2726 had early-pregnancy and 2515 had third-trimester samples assayed. Among the pregnancies with early-pregnancy samples, 53 children died before age 4 and 4 more died before age 7; 1723 (64%) and 1815 (68%) children had valid BMI measures at 48 and 84 months, respectively. Among the 2515 pregnancies with third-trimester samples, 48 children died before age 4 and 4 more died before age 7; 1622 (66%) and 1698 (69%) had valid BMI measures at 48 and 84 months, respectively. Exclusion of children with missing covariates resulted in a final study sample of 1651 and 1743 children with 48- and 84-month measures for those with <20 weeks samples, and 1559 and 1635 with 48- and 84-month measures for those with ≥26 weeks samples.

The associations between participant characteristics and maternal serum paraxanthine at <20 weeks’ and ≥26 weeks’ gestation are presented in Table 1. In general, paraxanthine increased with increasing maternal age, heavier smoking at study registration, and fewer years of maternal education; it was also higher among white women compared with other women. Third-trimester paraxanthine concentration declined with increasing maternal prepregnant weight, and early-pregnancy concentration was elevated in diabetic women.

Table 1 also presents associations between maternal characteristics and child’s mean BMI z-score and the percentage of children with BMI ≥85th percentile and BMI ≥95th percentile at ages 48 and 84 months. On average, the children in this 1960s and 1970s cohort had higher BMI values than U.S. children from 2000 at both 48 and 84 months; the corresponding probabilities of having an age- and sex-specific BMI at least the 85th and 95th percentiles were similarly elevated. BMI at both ages tended to increase with increasing maternal age, with increasing amount of maternal smoking, and with increasing maternal prepregnant weight; in addition, it was higher among offspring of white versus nonwhite women and among offspring of diabetic versus nondiabetic women.

The unadjusted associations between maternal serum paraxanthine concentrations and mean BMI z-score and BMI ≥85th and ≥95th percentiles for age and sex at 4 and 7 years of age are presented in eFigures 1 and 2 (http://links.lww.com/EDE/A859). P values for the unadjusted and adjusted overall associations, the nonlinear component of the association, and (when the adjusted nonlinear component was not significant) the adjusted simple linear trend are presented in Table 2. In the unadjusted analyses, mean BMI and the risk ratio for being ≥85th and ≥95th percentiles tended to increase with increasing serum paraxanthine, up to approximately 1,000 μg/liter, which is approximately 75th to 80th percentile at both time points and then either to remain constant or decrease. Most of the overall associations exhibited notable nonlinear components.

TABLE 2
TABLE 2:
Wald P Values for Association Between Maternal Serum Paraxanthine Obtained at 2 Points in Pregnancy and Child Body Mass Index at 4 and 7 Years of Age, Collaborative Perinatal Project, 1959–1974

The corresponding associations, adjusted for maternal prepregnant weight, education, smoking, age, race, diabetes, and days of gestation when blood was drawn, are presented in Figures 1 and 2. Although the shapes of the curves were generally similar to the unadjusted values, adjustment substantially reduced the magnitude of the associations (Table 2). There were, however, nonlinear associations between paraxanthine at <20 weeks and child BMI ≥95th percentile at age 4 years and between paraxanthine at ≥26 weeks and child BMI ≥95th percentile at age 7 years. In addition, there was a modest simple linear association between paraxanthine at ≥26 weeks and child mean BMI z-score at 4 years. However, the regression coefficient for paraxanthine had a negative sign, such that higher paraxanthine was associated with reduced mean child BMI.

FIGURE 1
FIGURE 1:
Mean difference in BMI z-score at ages 4 and 7 years by maternal serum paraxanthine (micrograms/liter) at <20 and 26+ weeks, adjusted for maternal age, prepregnant weight, smoking, education, gestation at blood draw, and diabetes. Lines indicate estimate (dark line) and 95% CI (lighter line).
FIGURE 2
FIGURE 2:
Adjusted risk ratios for elevated BMI at ages 4 and 7 years by maternal serum paraxanthine (micrograms/liter) at <20 and 26+ weeks, adjusted for maternal age, prepregnant weight, smoking, education, gestation at blood draw, and diabetes. Lines indicate estimate (dark line) and 95% CI (lighter line). Y-axis values truncated at 3.0 and 0.33.

The analysis of persistent obesity was based on children who had BMI measured at both 48 and 84 months. There were 1406 such children whose mothers had paraxanthine measured at <20 weeks and 1324 children whose mothers had paraxanthine measured at third trimester. For both paraxanthine measurements, there were 53 children with persistent obesity, which precluded adjustment for confounding factors. However, in unadjusted analyses, persistent obesity was not associated with paraxanthine at <20 weeks and during the third trimester (overall P = 0.72 and P for nonlinearity at <20 weeks = 0.45; corresponding values for third trimester, 0.49 and 0.98). Although values were imprecise, the greatest risk ratio for persistent obesity was approximately 1.25, for paraxanthine <20 weeks of approximately 700 μg/liter.

Mothers of children evaluated at both ages 4 and 7 years had higher serum paraxanthine values at both blood draws than that in mothers of children not evaluated. Other characteristics of mothers and children by follow-up status are presented in eTable 1 (http://links.lww.com/EDE/A859). However, study results were not substantially different when children were weighted by the inverse of their probability of being examined at each time point (data not shown).

DISCUSSION

This study did not, generally, support the association of an increased risk of childhood obesity with increasing maternal serum paraxanthine concentration during pregnancy. In most comparisons, we observed an increase in child BMI or risk of overweight/obesity with increasing paraxanthine up to approximately 1,000 μg/liter, with leveling off or reduction at higher concentrations. However, adjustment for confounders substantially attenuated all associations. The maximum modeled adjusted risk ratio was <1.5, and over most of the range of paraxanthine, the modeled adjusted mean child BMI z-score was less for women with detectable paraxanthine compared with those who had undetectable concentrations.

These results differ from those of the only previous report on this general topic, which found obesity to be approximately twice as common among offspring of women consuming ≥150 mg/day of caffeine during pregnancy as compared with offspring of women not consuming any caffeine.8 One possible explanation for the divergent results is differences in follow-up. Li et al8 followed children up to age 15 years, whereas we assessed BMI up to age 7 years. A second possible explanation may be the differences in the amount of caffeine consumed during pregnancy between the 1960s, when our data were collected, and the 1990s, when the previous study was conducted. Per capita coffee consumption in the United States was greater in the 1960s than in the 1990s.26 The Collaborative Perinatal Project did not collect data on maternal caffeine or coffee use, but it is plausible that, compared with our study population, the population studied by Li et al8 had relatively fewer women who consumed sufficient caffeine to cause a serum paraxanthine concentration of >1,000 μg/liter. Because this is the concentration above which our dose–response became nonlinear, Li et al may have had limited ability to detect a reduction in the risk ratio of child obesity with very high maternal caffeine consumption.

Our study has several strengths. It was conducted in an era when per capita coffee use peaked in the United States and when there was little concern regarding the safety of caffeine use during pregnancy. Therefore, our population is likely to include relatively more women who consumed large amounts of caffeine than would a more modern cohort. Maternal serum was collected prospectively, and this large cohort of children was followed prospectively. Children were weighed and measured at standardized times according to a standardized research protocol. We used a biomarker for caffeine use, thereby bypassing possible errors in recall and reporting of caffeine intake.

However, there are also limitations. Compared with the present time,27 maternal prepregnant obesity was less common, and pregnancy weight gain was lower.28 The Collaborative Perinatal Project predated nutritional supplementation programs for children and pregnant women, such as WIC (The Special Supplemental Nutrition Program for Women, Infants and Children). However, the impact of these differences on the applicability of our results is not clear. The prevalence of maternal smoking during pregnancy was much greater in the Collaborative Perinatal Project than in contemporary populations.29 Because smokers metabolize caffeine and paraxanthine more rapidly than nonsmokers,30 a given serum paraxanthine value may correspond to a higher caffeine intake in the Collaborative Perinatal Project than in a modern population. We had no data on maternal or childhood diets or on physical activity. If women who consumed large amounts of caffeine differed substantially in these behaviors from women consuming more typical amounts of caffeine, this unmeasured confounding might explain our nonlinear associations. Although paraxanthine correlates well with very recent caffeine intake,31 we had measures from only 2 points during pregnancy. Finally, paraxanthine is a marker for both caffeine intake and metabolism. In response to hormonal changes, CYP1A2 (the enzyme metabolizing caffeine to paraxanthine) activity decreases as pregnancy progresses.32 Failure of CYP1A2 to decline results in higher paraxanthine concentrations for a given caffeine intake.33 Therefore, an abnormal pregnancy per se may impact paraxanthine concentration, complicating interpretation of our results. Therefore, our use of a caffeine metabolite, rather than reported caffeine intake, as a measure of exposure may have accounted for the difference between our results and those of Li et al.

In summary, maternal serum concentration of paraxanthine, caffeine’s primary metabolite, was not associated with child BMI or with elevated BMI at ages 4 and 7 years. Future contemporary large prospective studies may help resolve the conflicting findings to date, and mechanistic studies could evaluate the biological plausibility of any observed associations.

REFERENCES

1. FDA. . Medicines in My Home: Caffeine and your Body. Available at: http://www.fda.gov/downloads/drugs/resourcesforyou/consumers/buyingusingmedicinesafely/understandingover-the-countermedicines/ucm205286.pdf. Accessed 28 May 2014
2. Frary CD, Johnson RK, Wang MQ. Food sources and intakes of caffeine in the diets of persons in the United States. J Am Diet Assoc. 2005;105:110–113
3. Hinkle S, Laughon S, Catov J, Olsen J, Bech B. First trimester coffee and tea intake and risk of gestational diabetes mellitus: a study within a national birth cohort. BJOG. June 20, 2014 doi:10.111/1471-0528.12930. [Epub ahead of print]
4. Hoyt AT, Browne M, Richardson S, Romitti P, Druschel CThe National Birth Defects Prevention S. . Maternal caffeine consumption and small for gestational age births: results from a population-based case-control study. Matern Child Health J. 2014;18:1540–1551
5. Peck JD, Leviton A, Cowan LD. A review of the epidemiologic evidence concerning the reproductive health effects of caffeine consumption: a 2000–2009 update. Food Chem Toxicol. 2010;48:2549–2576
6. Signorello LB, McLaughlin JK. Maternal caffeine consumption and spontaneous abortion: a review of the epidemiologic evidence. Epidemiology. 2004;15:229–239
7. Leviton A, Cowan L. A review of the literature relating caffeine consumption by women to their risk of reproductive hazards. Food Chem Toxicol. 2002;40:1271–1310
8. Li D-K, Ferber J, Odouli R. Maternal caffeine consumption during pregnancy and the risk of obesity in offspring: a prospective cohort study with 15 years of follow-up. 2014 Seattle, Washington Society for Pediatric and Perinatal Epidemiologic Research
9. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of childhood and adult obesity in the United States, 2011–2012. JAMA. 2014;311:806–814
10. Hediger ML, Overpeck MD, Kuczmarski RJ, McGlynn A, Maurer KR, Davis WW. Muscularity and fatness of infants and young children born small- or large-for-gestational-age. Pediatrics. 1998;102:E60
11. Buscariollo DL, Fang X, Greenwood V, Xue H, Rivkees SA, Wendler CC. Embryonic caffeine exposure acts via A1 adenosine receptors to alter adult cardiac function and DNA methylation in mice. PLoS One. 2014;9:e87547
12. 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;341:1639–1644
13. Niswander KR, Gordon MJNational Institute of Neurological Diseases and Stroke. The Women and Their Pregnancies; the Collaborative Perinatal Study of the National Institute of Neurological Diseases and Stroke. DHEW Publication no (NIH) 73–379. 1972 Washington National Institutes of Health
14. Klebanoff MA, Levine RJ, Clemens JD, Wilkins DG. Maternal serum caffeine metabolites and small-for-gestational age birth. Am J Epidemiol. 2002;155:32–37
15. 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;8:107–111
16. Lelo A, Miners JO, Robson R, Birkett DJ. Assessment of caffeine exposure: caffeine content of beverages, caffeine intake, and plasma concentrations of methylxanthines. Clin Pharmacol Ther. 1986;39:54–59
17. Myrianthopoulos NCCollaborative Perinatal Project (U.S.). Malformations in Children from One to Seven Years: A Report from the Collaborative Perinatal Project. 1985 New York, NY Liss
18. Lassmann FM, LaBenz PJ, LaBenz ES Early Correlates of Speech, Language, and Hearing; the Collaborative Perinatal Project of the National Institute of Neurological and Communicative Disorders and Stroke. 1980 Littleton, MA PSG Pub. Co.
19. Branum AM, Parker JD, Keim SA, Schempf AH. Prepregnancy body mass index and gestational weight gain in relation to child body mass index among siblings. Am J Epidemiol. 2011;174:1159–1165
20. Ribas-Fitó N, Gladen BC, Brock JW, Klebanoff MA, Longnecker MP. Prenatal exposure to 1,1-dichloro-2,2-bis (p-chlorophenyl)ethylene (p,p’-DDE) in relation to child growth. Int J Epidemiol. 2006;35:853–858
21. (U.S.) Centers for Disease Control and Prevention. . A SAS Program for the 2000 CDC Growth Charts (ages 0 to <20 y). http://www.cdc.gov/nccdphp/dnpao/growthcharts/resources/sas.htm. Accessed 7 November 2014
22. McNutt LA, Wu C, Xue X, Hafner JP. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol. 2003;157:940–943
23. Zou G. A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159:702–706
24. Diggle P, Liang K-Y, Zeger SL. Oxford statistical science series. Repr. 1994, 1995 (with corrections) ed. Analysis of Longitudinal Data. 1995 Oxford, NY: Oxford University Press
25. Desquilbet L, Mariotti F. Dose-response analyses using restricted cubic spline functions in public health research. Stat Med. 2010;29:1037–1057
26. Barone JJ, Roberts HR. Caffeine consumption. Food Chem Toxicol. 1996;34:119–129
27. Fisher SC, Kim SY, Sharma AJ, Rochat R, Morrow B. Is obesity still increasing among pregnant women? Prepregnancy obesity trends in 20 states, 2003–2009. Prev Med. 2013;56:372–378
28. Martin JA, Hamilton BE, Sutton PD, Ventura SJ, Menacker F, Kirmeyer S, Munson MLCenters for Disease C, Prevention National Center for Health Statistics National Vital Statistics S. . Births: final data for 2005. Natl Vital Stat Rep. 2007;56:1–103
29. Tong VT, Dietz PM, Morrow B, et al.Centers for Disease C, Prevention. Trends in smoking before, during, and after pregnancy—Pregnancy Risk Assessment Monitoring System, United States, 40 sites, 2000–2010. MMWR Surveill Summ. 2013;62:1–19
30. Parsons WD, Neims AH. Effect of smoking on caffeine clearance. Clin Pharmacol Ther. 1978;24:40–45
31. Grosso LM, Triche E, Benowitz NL, Bracken MB. Prenatal caffeine assessment: fetal and maternal biomarkers or self-reported intake? Ann Epidemiol. 2008;18:172–178
32. Aldridge A, Bailey J, Neims AH. The disposition of caffeine during and after pregnancy. Semin Perinatol. 1981;5:310–314
33. 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;163:1035–1041

Supplemental Digital Content

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.