Caffeine Metabolism, Genetics, and Perinatal Outcomes: A Review of Exposure Assessment Considerations during Pregnancy

Purpose

To review the methodologic issues complicating caffeine exposure assessment during pregnancy; to discuss maternal and fetal caffeine metabolism, including genetic polymorphisms affecting caffeine metabolism; and to discuss the endogenous and exogenous risk factors known to influence caffeine metabolism.

Methods

A review of the relevant literature.

Results

There is wide inter-individual variation in caffeine metabolism, primarily due to variations in CYP1A2 enzyme activity. Some variability in CYP1A2 activity is due to genetic polymorphisms in the CYP1A2 gene which can cause increased or decreased inducibility of the enzyme. Considerable evidence exists that maternal caffeine metabolism is influenced by a variety of endogenous and exogenous factors and studying the genetic polymorphisms may improve understanding of the potential effects of caffeine and its metabolites on perinatal outcomes. There is substantial evidence that measurement of maternal, fetal, and neonatal caffeine metabolites may allow for a more precise measure of fetal caffeine exposure.

Conclusions

Research on the genetic polymorphisms affecting caffeine metabolism may further explain the potential effects of caffeine and its metabolites on perinatal outcomes.

Introduction

Maternal caffeine consumption during pregnancy has been studied for many years but convincing evidence for an association with poor perinatal outcomes remains elusive. Caffeine is an exposure of major public health interest because it is one of the most widely consumed drugs. Coffee makes up the largest percentage of total caffeine intake (75%), followed by tea (15%), and caffeinated sodas (10%) (1). In the United States, per capita consumption of coffee is nearly 3.5 kg of coffee per year, or more than 150 mg/day, and more than 75% of pregnant women consume caffeinated beverages 2, 3.

Epidemiologic studies of caffeine and reproductive outcomes have produced conflicting results. Some epidemiological studies have linked relatively high antenatal caffeine consumption (typically > 300 mg/day) to poor reproductive outcomes, including subfecundity 4, 5, 6, 7, 8; fetal growth retardation 2, 9, 10, 11, 12; and spontaneous abortion 13, 14, 15, 16, 17, 18. One study among smoking women who consumed > 400 mg caffeine per day noted a significant reduction in birthweight (19) and a more recent study reported a small, detrimental effect on birthweight, although it is only likely to be of clinical importance in women consuming large quantities of caffeine (20). Other studies suggest that antenatal caffeine consumption is not a reproductive hazard 21, 22, 23, 24, 25.

These equivocal findings are likely due to inconsistent definition and categorization of caffeine exposure among studies, selection and recall biases, and to varying study designs. The major limitations of the extant studies include: lack of control for confounding variables 4, 12, bias due to misclassification of caffeine exposure 4, 9, 21, and imprecise/inadequate measurement of caffeine intake 4, 21. Some studies were retrospective and the exposure information was collected long after the exposure occurred 9, 10, 21. Three prospective cohort studies that found an increase in risk for intrauterine growth retardation (IUGR) with third trimester caffeine consumption suffer from additional methodological problems including lack of a completely unexposed referent group 4, 19 and residual confounding due to inadequate control for the effects of smoking (12). Two retrospective cohort studies, one finding no effect (21) and the other finding an effect in women consuming coffee (26), inadequately assessed caffeine intake. Linn et al. (21) assessed caffeine consumption in the first trimester only and the caffeine content per cup of coffee was not estimated and McDonald et al. (26) only assessed coffee consumption.

Studies investigating the association between maternal caffeine consumption and spontaneous abortion are also fraught with methodological complexity. A comprehensive review of this literature was conducted by Signorello et al. (27). Cross-sectional and case-control studies suffer from a variety of methodologic issues including recall bias; inaccurate recall of caffeine exposure due to exposure assessment several years later; confounding due to cigarette and alcohol consumption 18, 28; and selection bias (27). Confounding due to pregnancy symptoms is another important issue complicating the relation between caffeine consumption and spontaneous abortion. Nausea may influence the amount of caffeine consumed in early pregnancy and it is also related to fetal viability. Women with nonviable pregnancies may have little or no nausea and therefore do not decrease their caffeine intake, falsely suggesting that their higher caffeine intake is related to the spontaneous abortion (29).

Change in caffeine consumption over pregnancy is another important factor complicating exposure assessment. Women frequently have an aversion to caffeinated beverages in the first trimester of pregnancy and therefore decrease caffeine intake. They may also decrease caffeine intake upon confirmation of the pregnancy 30, 31. It is therefore crucial that caffeine exposure be assessed at multiple time periods throughout pregnancy.

There is a considerable amount of heterogeneity in caffeine exposure because caffeine content of caffeinated beverages varies widely and there are large differences in caffeine content per serving of coffee, tea, and soft drinks 7, 19, 32. Methodological issues related to retrospective ascertainment of exposure include inaccurate recall by the subject due to unawareness and/or forgetfulness of consumption and biased recall. Caffeine content of caffeinated beverages varies considerably, depending on brewing method, serving size, and portion of serving consumed. Estimates of caffeine content per serving for coffee, tea, and soft drinks range from 92 to 120 mg/serving, 34 to 65 mg/serving, and 34 to 47 mg/serving, respectively, and current ranges for soft drinks may be considerably higher 7, 19, 32.

The caffeine content of coffee is quite variable and depends on brand, whether it is a blend or a pure variety (33), quantity brewed, brewing method, and type of coffee bean. Caffeine extraction efficiency varies from 75% to 100%, depending on whether coffee is boiled, filtered, percolated, or prepared as espresso (34). Serving sizes range from 5 to 32 ounces and the caffeine content per cup is reported to range as much as 19 to 160 mg depending on brewing method and cup size (34). Considerable variation in caffeine content was found, even when the same study participant brewed coffee or tea under the same conditions on the same day (35).

Variation in caffeine consumption categorization also contributes to exposure misclassification and decreased comparability among studies. Categorization for the lowest levels of consumption (often used as the referent category) varies from no intake (21) to ≤ 400 mg/day (32) and for the highest categories of consumption from > 300 mg/day (19) to > 800 mg/day (32). If caffeine consumption of approximately 300 mg/day or more is associated with poor perinatal outcomes, grouping such exposed individuals with the lesser exposed or unexposed would dilute any effect of caffeine exposure.

In most studies of caffeine consumption and perinatal outcomes, self-reported caffeine exposure is calculated using a standard measure of caffeine per unit exposure that has been obtained by laboratory analyses. In a recent study, samples of caffeinated and decaffeinated coffee and tea were collected from the study participants and analyzed for actual caffeine content (35). It was observed that for all cup sizes, the actual amounts of caffeine in both coffee and tea were much lower than the amounts predicted using widely used laboratory estimates. For example, a 10 oz. cup of drip brewed coffee is estimated to contain 300 mg caffeine, according to Bunker and McWilliams (36), but Bracken et al. (35), found that a 10 oz. drip brewed cup of coffee typically contained 100 mg caffeine. Similarly, a 10 oz. cup of tea brewed for more than 3 minutes was found to contain 42 mg caffeine compared with the predicted 94 mg (35).

Even among well designed studies with valid exposure assessment, nearly all of them relied on self-reported caffeine consumption to estimate exposure. This does not provide an accurate measure of maternal or fetal dose because it does not necessarily indicate how much caffeine or caffeine metabolites enter maternal or fetal circulation. A variety of endogenous and exogenous risk factors are known to influence caffeine metabolism and it is possible that caffeine metabolites, rather than caffeine itself, are responsible for deleterious reproductive and perinatal effects. Serum and urinary biomarkers, used in conjunction with self-reported caffeine intake, may provide a more accurate and direct measure of maternal and fetal dose than information obtained solely via questionnaire (37).

There are potential disadvantages in using a caffeine biomarker. Timing of sample collection, relative to reported intake, can greatly affect biomarker accuracy. For example, if a pregnant woman consumes all of her daily caffeine in the morning, but after urine or serum sample collection, the sample would substantially under-represent her caffeine intake since most of the previous day's caffeine would be metabolized. A more appropriate time for sampling this consumption pattern is mid-to-late afternoon, since caffeine half-life, at this point in pregnancy, is approximately 10 hours (38).

Cigarette smoking can affect accuracy of the biomarker by accelerating caffeine metabolism 38, 39, 40, resulting in decreased urinary caffeine and underestimation of caffeine consumption. Similarly, concurrent drug use affecting caffeine metabolism will alter how much caffeine is excreted in urine.

One study investigating the relation between maternal third-trimester serum paraxanthine, the primary metabolite of caffeine and fetal growth, found that maternal serum paraxanthine was only associated with growth retardation among smokers (41). These samples had been stored for 30 years and it is possible that some of the paraxanthine degraded. This would result in a differential misclassification bias because smoking and non-smoking women exposed to moderate levels of caffeine may have been subsequently categorized into the lowest paraxanthine level used as a referent category, biasing any association toward unity 42, 43.

Caffeine is absorbed rapidly and completely from the gastrointestinal tract. Peak plasma concentrations are reached within 15 (33) to 60 (44) minutes after intake, but can take as long as 120 minutes after ingestion (33). Delayed gastric emptying is thought to account for this variation in peak concentration (33). First pass metabolism occurs as oral agents are absorbed through the small intestine into the portal circulation where initial metabolism by CYP450 isoenzymes occurs in the bowel wall and liver before entering the systemic circulation. Since there is a minimal first-pass effect for caffeine (33), once it is absorbed it readily enters all body tissues and freely crosses the blood-brain, placental, and blood-testicular barriers 33, 45. Caffeine and its primary metabolites, paraxanthine, theobromine, and theophylline, are detectable in all body fluids and in umbilical cord blood 33, 46. Caffeine is eliminated from the body overnight, but some primary metabolites (theophylline and theobromine) have longer half-lives (47).

In humans, the half-life of caffeine ranges from 2 to 4.5 hours 45, 48, but can be as long as 12 hours (47). Caffeine is metabolized in the liver by the hepatic microsomal enzyme systems to dimethylxanthines. The main enzyme involved in caffeine metabolism is cytochrome P450 1A2 (CYP1A2), accounting for about 95% of caffeine clearance. The rate of caffeine metabolism is controlled by CYP1A2 and to a lesser extent by xanthine oxidase (XO) and N-acetyltransferase 2 (NAT2) (49). Only 0.5% to 2% of ingested caffeine is excreted as such in the urine due to 98% tubular reabsorption (33) and paraxanthine accounts for 72% to 80% of caffeine metabolism 33, 50.

Paraxanthine, the primary metabolite of caffeine, has a molecular structure and half-life similar to caffeine (51) and is easily measured in urine and serum. Approximately 60% of orally administered paraxanthine is excreted unchanged (52). The rate of paraxanthine degradation approximates its rate of formation and serum levels are more reliable and less variable throughout the day compared with caffeine (53), although they reflect only recent intake (16). Plasma paraxanthine concentrations decrease less rapidly than caffeine, even after accounting for inter-individual differences in metabolism, and paraxanthine concentrations become higher than caffeine within 8 to 10 hours of ingestion (33). Paraxanthine is further metabolized via two parallel but independent reactions (40). One produces 8-hydroxyparaxanthine and the other reaction is the 7-demethylation of paraxanthine which leads to three metabolites: 1-methylxanthine, 1-methylurate, and 5-acetylamino-6-formylamino-3-methyluracil (AFMU) (40). AFMU accounts for 67% of paraxanthine metabolism (33). AFMU is converted to 5-acetyl-6-amino-3-methyluracil (AAMU) which can be easily and reliably measured in urine (40). These paraxanthine metabolites appear in the urine almost as fast as they are formed due to active renal tubular secretion (40).

Theobromine comprises the largest percentage of the biologically active caffeine metabolites (54). It is rapidly absorbed and approximately 50% is excreted in urine within 8 to 12 hours (55). Its pharmacologic effects include diuresis, stimulation of the cardiovascular system, relaxation of smooth muscle, and increased glandular secretion (56). Metabolic clearance of theobromine is mediated primarily by the CYP1A2 enzyme which is estimated to account for 86% of its demethylation, and to a lesser degree by CYP2E1 (57). The half-life ranges from 7.2 (58) to 11.5 (55) hours and plasma and renal clearance is reported as 46% and 67%, respectively (59). Plasma clearance is reported to be 33% higher in smokers than in nonsmokers (60).

Theophylline is structurally similar to caffeine, lacking only one additional N-methyl group. Its pharmacological properties are very similar as well, but theophylline incurs stronger toxicological effects than caffeine and theobromine. The half-life is longer than that of caffeine (3-9 hours) but is quite variable (61). Theophylline is cleared from the body via renal and metabolic clearance. Metabolic clearance is mediated primarily by CYP1A2 with N-demethylation to monomethylxanthines and 8-hydroxylation to 1,3-dimethyl-uric acid. The urinary excretion rate is highly dependent on urinary flow and dose. Higher plasma concentrations of theophylline result in decreased metabolic clearance and increased renal clearance. Like caffeine, there is considerable inter-individual variation in theophylline clearance, metabolism, and elimination (61). Several exogenous factors influence metabolic and excretion rates including cigarette smoking, viral infections, liver and heart diseases, pregnancy, foods, and concomitant drug use. Smoking increases clearance and drugs including cimetidine, ranitidine, erythromycin, rifamycin, and troleandomycin slow metabolism. Pregnancy decreases clearance and excretion and therefore theophylline (like caffeine) can accumulate in the body (61). Fetuses and newborns lack the enzymes necessary to metabolize theophylline, so elimination is almost entirely dependent on renal excretion.

Two of the most important endogenous and exogenous factors influencing caffeine metabolism are pregnancy and cigarette smoking, respectively. Pregnancy slows caffeine metabolism while cigarette smoking accelerates it 39, 40, 62. During pregnancy, caffeine half-life remains the same during the first trimester but increases to 10 hours at 17 weeks gestation (38). By the end of pregnancy the half-life in non-smokers varies from 11.5 (63) to 18 (38) hours, leading to an accumulation of caffeine in the body. This gestational increase in caffeine half-life is likely related to a reduction in NAT2 enzyme activity in early pregnancy, and reduction in CYP1A2 activity throughout pregnancy (64). One study noted a 35%, 50%, and 52% reduction in CYP1A2 activity in early, middle, and late pregnancy, respectively, compared with 4 to 6 weeks after delivery (64). Cook et al. (65) noted that serum caffeine concentrations rose from a mean of 2.35 μg/ml in early pregnancy to 4.12 μg/ml by the third trimester, despite little change in reported consumption.

Cigarette smoking is an important exogenous factor influencing caffeine metabolism, nearly doubling the metabolism rate (38). Cigarette smoke contains polycyclic aromatic hydrocarbons known to increase liver enzyme activity, thereby increasing caffeine metabolism 39, 40. Smoking may accelerate the first and second demethylation steps of caffeine metabolism, via induction of hepatic microsomal oxidative enzymes (66). Smokers have been observed to have lower serum caffeine concentrations than non-smokers within each category of reported consumption (65).

Other exogenous factors that slow caffeine metabolism are liver disorders (33), oral contraceptive use 67, 68, and luteal phase of the menstrual cycle (69). Several drugs, including fluvoxamine (a serotonin reuptake inhibitor), mexiletine (an antiarrhythmic), clozapine (an antipsychotic), furafylline and theophylline (bronchodilators), and enoxacin (a quinolone) may slow caffeine metabolism (70). Consumption of apiaceous (Apiaceae or Umbelliferae) vegetables including dill weed, celery, parsley, parsnips and carrots, has been shown to reduce CYP1A2 activity, subsequently slowing caffeine metabolism. Consumption of brassica (Cruciferae) vegetables, particularly radish sprouts, broccoli, cauliflower, and cabbage, accelerates CYP1A2 activity, thereby accelerating caffeine metabolism (71).

There is wide inter-individual variation in caffeine metabolism, primarily due to variations in CYP1A2 enzyme activity 72, 73, 74 and there is recent interest in identifying polymorphisms which influence caffeine metabolism. Some variability in CYP1A2 activity is due to genetic polymorphisms in the CYP1A2 gene which can cause increased or decreased inducibility of the enzyme. Urinary caffeine metabolite ratios have been used extensively to phenotypically assess acetylator status 72, 75, 76, 77, 78, but recently, polymerase chain reaction (PCR) has been used to determine allelic variants of CYP1A2, specifically whether an individual is a slow (mutated allele) or fast (wild type) acetylator. To date, several recently discovered single nucleotide polymorphisms (SNPs) may help explain some of the inter-individual variation in caffeine metabolism 73, 74, 79. One SNP, CYP1A2 3858G→A (CYP1A2^∗1C) has been observed to cause a significant decrease in CYP1A2 inducibility in Japanese smokers (73). This SNP is rare in Caucasians. A common polymorphism in Caucasians, CYP1A2 164C→A (CYP1A2^∗1F) influences caffeine metabolism, and is associated with increased inducibility in smokers homozygous for the A allele (80). CYP1A2 1545T→C (CYP1A2^∗1B) is associated with three other mutations—740T→G (CYP1A2^∗1G), 951A→C (CYP1A2^∗1H), and 1042G→A (CYP1A2^∗3)—that are also frequent in Caucasians. The frequency of this SNP is 35.0% (81) and 38.2% (82) in French and British populations, respectively, however the effects of this polymorphism on CYP1A2 activity remain unclear.

Further investigation of the effects of these maternal polymorphisms on CYP1A2 enzyme activity (and subsequently caffeine metabolism) is needed to predict fetal caffeine exposure and more fully understand any effects of maternal caffeine consumption on perinatal outcomes.

Caffeine readily crosses the placenta into the fetus and amniotic fluid and maternal serum concentrations are believed to be reliable indicators of fetal serum concentration 83, 84. This equilibration occurs as early as week 7 of gestation (83). Since neither fetus nor placenta can metabolize caffeine, the fetus is exposed to caffeine and its metabolites for a prolonged period of intra-uterine life. The human placenta cannot metabolize caffeine because it contains only CYP1A1, not CYP1A2 (72). The fetus also lacks liver enzymes necessary to metabolize caffeine, which are not present until about the eighth month of age (23).

As a result of hepatic immaturity, the neonate has compensatory pathways including renal elimination for caffeine and theophylline (85). Since renal elimination is not as efficient as CYP1A2 mediated clearance, overall fetal metabolism is much less compared with adults. In neonates, 80% to 90% of caffeine is excreted in urine, compared with less than 2% in adults (85). Excretion of theophylline is less complete; nearly 50% is eliminated unchanged in neonates, compared with 10% in adults (85).

Caffeine concentrations in umbilical cord blood are higher than expected based on maternal caffeine consumption (46). Other evidence suggests that transplacentally acquired theophylline, an active alkaloid from tea, may further increase fetal and early neonatal caffeine load. Human fetal liver can methylate theophylline to caffeine as early as the 12th week of gestation (86). This reverse biotransformation of theophylline into caffeine was first reported in preterm infants treated with theophylline for apnea (33). Back-methylation, where a methyl group is added to theophylline, accounts for approximately 5% to 10% of the overall urinary excretion of caffeine in neonates (85).

Cazeneuve et al. (87) noted that formation of dimethylxanthines was significantly less in fetuses, neonates, and infants than in adults. In neonates and infants, production of total dimethylxanthine, paraxanthine, and theophylline increased significantly with postnatal age. The half-life of caffeine in the newborn is estimated to range from 50 to 103 hours, compared with 6 hours in the non-smoking adult 33, 86. Caffeine half-life decreases to 14 hours and 3 hours in 3 to 5 months and 5 to 6 months, respectively (33). Urinary caffeine and theophylline from newborn urine may accurately estimate fetal exposure to caffeine during the last month of pregnancy, since caffeine and theophylline elimination in the newborn is so immature.

One effect of caffeine ingestion is to increase release of catecholamines, particularly epinephrine, into the maternal circulation (84). The direct effects of caffeine on fetal circulation remain unknown but Kirkinen et al. (84) documented a decrease in intervillous placental blood flow after maternal caffeine ingestion of just 200 mg. Caffeine also inhibits phosphodiesterase, an enzyme responsible for the breakdown of cyclic AMP (88). An increase in cyclic AMP may interfere with cell division, or cause catecholamine-mediated uterine vasoconstriction (89). If uterine blood flow is inadequate, extraction of oxygen and nutrients increase until a critical point during pregnancy at which reductions in blood flow can profoundly affect fetal oxygenation and nutrition (90). The exact mechanisms by which caffeine may impair fetal growth remain unknown but the literature suggests that impairment of uteroplacental blood flow, fetoplacental blood flow, or villous blood flow can lead to IUGR 88, 90. In a study of very low birth weight, preterm infants with apnea who were given caffeine therapy, neonatal oxygen consumption increased and there was a reduction in weight gain (91). Caffeine exposure during the third trimester may have a similar effect on fetal growth.

In summary, there is substantial evidence that measurement of maternal, fetal, and neonatal caffeine metabolites may allow for a more precise measure of fetal caffeine exposure. There is preliminary evidence that the metabolites of caffeine may play an important role in the associations with perinatal outcomes and there is also considerable evidence that maternal caffeine metabolism is influenced by a variety of endogenous and exogenous factors. Studying the genetic polymorphisms influencing metabolism, in particular, may improve our understanding of the potential effects of caffeine and its metabolites on perinatal outcomes.