Inhibitors of Organic Anion-Transporting Polypeptides 1B1 and 1B3: Clinical Relevance and Regulatory Perspective.

Organic anion-transporting polypeptides (OATPs) 1B1 and 1B3 are the primary hepatic transporters responsible for uptake of drugs into the liver and, as such, an area of growing research focus. Currently, evaluation of these transporters as potential mediators of drug-drug interactions (DDIs) is recommended by regulatory agencies worldwide during the drug development process. Despite the growing focus on OATP1B1/1B3 as mediators of DDIs, only 2 drugs are recommended as index inhibitors for use in clinical studies, single-dose rifampin and cyclosporine, each with limitations for the utility of the resulting data. In this study a thorough analysis of the available in vitro and clinical data was conducted to identify drugs that are clinically relevant inhibitors of OATP1B1/1B3 and, from those, to select any novel index inhibitors. A total of 13 drugs and 16 combination products were identified as clinical inhibitors of OATP1B1/1B3, showing significant changes in exposure for sensitive substrates of the transporters, with strong supporting in vitro evidence. Although none of the identified inhibitors qualified as index inhibitors, this study confirmed the utility of cyclosporine and single-dose rifampin as index inhibitors to evaluate the effect of broad, multiple-pathway inhibition and more selective OATP1B1/1B3 inhibition, respectively.

Identification and Evaluation of Clinical Substrates of Organic Anion Transporting Polypeptides 1B1 and 1B3.

Organic anion transporting polypeptides (OATPs) 1B1 and 1B3 facilitate the uptake of drugs and endogenous compounds into the liver. In recent years, the impact of these transporters on drug-drug interactions (DDIs) has become a focus of research, and the evaluation of their role in drug disposition is recommended by regulatory agencies worldwide.1-3 Although sensitive substrates of OATP1B1/1B3 have been identified in the literature and probe drugs have been proposed by regulatory agencies, there is no general consensus on the ideal in vivo substrate for clinical DDI studies as analysis may be confounded by contribution from other metabolic and/or transport pathways.1-3 A thorough analysis of the available in vitro and in vivo data regarding OATP1B1/1B3 substrates was performed using the in vitro, clinical, and pharmacogenetic modules in the University of Washington Drug Interaction Database. A total of 34 compounds were identified and further investigated as possible clinical substrates using a novel indexing system. By analyzing the compounds for in vivo characteristics, including sensitivity to inhibition by known OATP1B1/1B3 inhibitors, selectivity for OATP1B1/1B3 compared with other transport and metabolic pathways, and safety profiles, a total of six compounds were identified as potential clinical markers of OATP1B1/1B3 activity.

Applications of Clinically Relevant Dissolution Testing: Workshop Summary Report.

This publication summarizes the proceedings of day 3 of a 3-day workshop on "Dissolution and Translational Modeling Strategies Enabling Patient-Centric Product Development." Specifically, this publication discusses the current approaches in building clinical relevance into drug product development for solid oral dosage forms, along with challenges that both industry and regulatory agencies are facing in setting clinically relevant drug product specifications (CRDPS) as presented at the workshop. The concept of clinical relevance is a multidisciplinary effort which implies an understanding of the relationship between the critical quality attributes (CQAs) and their impact on predetermined clinical outcomes. Developing this level of understanding, in many cases, requires introducing deliberate but meaningful variations into the critical material attributes (CMAs) and critical process parameters (CPPs) to establish a relationship between the resulting in vitro dissolution/release profiles and in vivo PK performance, a surrogate for clinical outcomes. Alternatively, with the intention of improving the efficiency of the drug product development process by limiting the burden of conducting in vivo studies, this understanding can be either built, or at least enhanced, through in silico efforts, such as IVIVC and physiologically based pharmacokinetic (PBPK) absorption modeling and simulation (M&S). These approaches enable dissolution testing to establish safe boundaries and reject drug product batches falling outside of the established safe range (e.g., due to inadequate in vivo performance) enabling the method to become clinically relevant. Ultimately, these efforts contribute towards patient-centric drug product development and allow regulatory flexibility throughout the lifecycle of the drug product.

Commentary on the MID3 Good Practices Paper.

During the last 10 years the European Medicines Agency (EMA) organized a number of workshops on modeling and simulation, working towards greater integration of modeling and simulation (M&S) in the development and regulatory assessment of medicines. In the 2011 EMA - European Federation of Pharmaceutical Industries and Associations (EFPIA) Workshop on Modelling and Simulation, European regulators agreed to the necessity to build expertise to be able to review M&S data provided by companies in their dossier. This led to the establishment of the EMA Modelling and Simulation Working Group (MSWG). Also, there was agreement reached on the need for harmonization on good M&S practices and for continuing dialog across all parties. The MSWG acknowledges the initiative of the EFPIA Model-Informed Drug Discovery and Development (MID3) group in promoting greater consistency in practice, application, and documentation of M&S and considers the paper is an important contribution towards achieving this objective.

Towards quantitation of the effects of renal impairment and probenecid inhibition on kidney uptake and efflux transporters, using physiologically based pharmacokinetic modelling and simulations.

BACKGROUND AND OBJECTIVES: The kidney is a major drug-eliminating organ. Renal impairment or concomitant use of transporter inhibitors may decrease active secretion and increase exposure to a drug that is a substrate of kidney secretory transporters. However, prediction of the effects of patient factors on kidney transporters remains challenging because of the multiplicity of transporters and the lack of understanding of their abundance and specificity. The objective of this study was to use physiologically based pharmacokinetic (PBPK) modelling to evaluate the effects of patient factors on kidney transporters. METHODS: Models for three renally cleared drugs (oseltamivir carboxylate, cidofovir and cefuroxime) were developed using a general PBPK platform, with the contributions of net basolateral uptake transport (T up,b) and apical efflux transport (T eff,a) being specifically defined. RESULTS AND CONCLUSION: We demonstrated the practical use of PBPK models to: (1) define transporter-mediated renal secretion, using plasma and urine data; (2) inform a change in the system-dependent parameter (≥10-fold reduction in the functional 'proximal tubule cells per gram kidney') in severe renal impairment that is responsible for the decreased secretory transport activities of test drugs; (3) derive an in vivo, plasma unbound inhibition constant of T up,b by probenecid (≤1 µM), based on observed drug interaction data; and (4) suggest a plausible mechanism of probenecid preferentially inhibiting T up,b in order to alleviate cidofovir-induced nephrotoxicity.

Assessment of interaction potential of AZD2066 using in vitro metabolism tools, physiologically based pharmacokinetic modelling and in vivo cocktail data.

PURPOSE: Static and dynamic (PBPK) prediction models were applied to estimate the drug-drug interaction (DDI) risk of AZD2066. The predictions were compared to the results of an in vivo cocktail study. Various in vivo measures for tolbutamide as a probe agent for cytochrome P450 2C9 (CYP2C9) were also compared. METHODS: In vitro inhibition data for AZD2066 were obtained using human liver microsomes and CYP-specific probe substrates. DDI prediction was performed using PBPK modelling with the SimCYP simulator™ or static model. The cocktail study was an open label, baseline, controlled interaction study with 15 healthy volunteers receiving multiple doses of AD2066 for 12 days. A cocktail of single doses of 100 mg caffeine (CYP1A2 probe), 500 mg tolbutamide (CYP2C9 probe), 20 mg omeprazole (CYP2C19 probe) and 7.5 mg midazolam (CYP3A probe) was simultaneously applied at baseline and during the administration of AZD2066. Bupropion as a CYP2B6 probe (150 mg) and 100 mg metoprolol (CYP2D6 probe) were administered on separate days. The pharmacokinetic parameters for the probe drugs and their metabolites in plasma and urinary recovery were determined. RESULTS: In vitro AZD2066 inhibited CYP1A2, CYP2B6, CYP2C9, CYP2C19 and CYP2D6. The static model predicted in vivo interaction with predicted AUC ratio values of >1.1 for all CYP (except CYP3A4). The PBPK simulations predicted no risk for clinical relevant interactions. The cocktail study showed no interaction for the CYP2B6 and CYP2C19 enzymes, a possible weak inhibition of CYP1A2, CYP2C9 and CYP3A4 activities and a slight inhibition (29 %) of CYP2D6 activity. The tolbutamide phenotyping metrics indicated that there were significant correlations between CLform and AUCTOL, CL, Aemet and LnTOL24h. The MRAe in urine showed no correlation to CLform. CONCLUSIONS: DDI prediction using the static approach based on total concentration indicated that AZD20066 has a potential risk for inhibition. However, no DDI risk could be predicted when a more in vivo-like dynamic prediction method with the PBPK with SimCYP™ software based on early human PK data was used and more parameters (i.e. free fraction in plasma, no DDI risk) were taken into account. The clinical cocktail study showed no or low risks for clinical relevant DDI interactions. Our findings are in line with the hypothesis that the dynamic prediction method predicts DDI in vivo in humans better than the static model based on total plasma concentrations.

Discovery and characterization of AZD9272 and AZD6538-Two novel mGluR5 negative allosteric modulators selected for clinical development.

AZD9272 and AZD6538 are two novel mGluR5 negative allosteric modulators selected for further clinical development. An initial high-throughput screening revealed leads with promising profiles, which were further optimized by minor, yet indispensable, structural modifications to bring forth these drug candidates. Advantageously, both compounds may be synthesized in as little as one step. Both are highly potent and selective for the human as well as the rat mGluR5 where they interact at the same binding site than MPEP. They are orally available, allow for long interval administration due to a high metabolic stability and long half-lives in rats and permeate the blood brain barrier to a high extent. AZD9272 has progressed into phase I clinical studies.

A formulation comparison between micro- and nanosuspensions: the importance of particle size for absorption of a model compound, following repeated oral administration to rats during early development.

AIM: The aim of this study was to maximize the exposure of a model compound (MC) for forthcoming high-dose toxicological studies with the physical form of the original compound unaffected. METHOD: The two evaluated formulation approaches for the present poorly water-soluble compound were micro- and nanosuspensions. RESULTS: The particle size was about 280 nm for the nanosuspensions and about 4 µm for the microsuspensions. The crystallinity and the crystalline form of the ground samples were conserved. The physical and the chemical stabilities of the two kinds of suspensions were unaffected during the investigated time period. The in vivo results of the study showed that the pharmacokinetic parameters investigated were comparable at the low-dose level (6 µmol/kg) for both formulations after single administration. However, at the two higher doses (60 and 300 µmol/kg), a significant difference in exposure was observed between the two suspensions with an improved exposure for smaller particles. After Day 7 of repeated administration, a significant difference in exposure was observed at all dose levels. The overall exposures were higher on Day 7, compared to the exposures on Day 1 (most significant for nanoparticles), due to an accumulation of compound in the body. CONCLUSIONS: The nanoparticles have a larger surface, resulting in faster in vivo dissolution rate, faster absorption, and increased bioavailability, compared to microparticles. The differences in systemic exposure of model compound, following oral administration of nano- or microparticles of the drug substance, are probably caused by differences in the in vivo dissolution rate and possibly further enhanced by saturation of the systemic elimination.

Expression of CYP3A isoforms and P-glycoprotein in human stomach, jejunum and ileum.

1. CYP3A isoforms metabolise a diverse array of clinically important drugs and P-glycoprotein (P-gp), a transmembrane efflux pump, can extrude a wide variety of drugs from the cell. It has been suggested that the function of CYP3A4 is complementary to that of P-gp along the gastrointestinal (GI) tract, together forming a coordinated intestinal barrier against xenobiotics. Therefore, the expression of CYP3A4, CYP3A5, CYP3A7 and ABCB1 (P-gp) genes were quantified in five normal samples from the human stomach, seven from the jejunum and eight from the ileum by real-time reverse transcription-polymerase chain reaction and western blot analysis. 2. In the tissues examined, measurable mRNA expression of CYP3A was found in almost all samples from the stomach, jejunum and ileum. The rank order for CYP3A mRNA expression was CYP3A4 > CYP3A5 > CYP3A7 in the GI tract studied, whereas median mRNA CYP3A4 expression was highest in the small intestine and lowest in the stomach. Expression of ABCB1 mRNA was found in almost all samples and the median mRNA expression level was comparable in the jejunum and ileum, but lower in the stomach. Our data also show a significant correlation between all mRNA transcripts studied and a wide interindividual variation. 3. At the protein level, CYP3A4 was detected in all stomach and small intestine samples, the levels being substantially higher in the small intestine than in the stomach. P-Glycoprotein was detected in all GI samples, but no statistically significant difference was found along the GI tract considered. 4. Collectively, these results demonstrate that CYP3A4 is the main CYP3A expressed in the GI tract investigated, an extensive interindividual variability in the expression of the different CYP3A isoforms in all tissues examined and P-gp apoprotein levels similar in the stomach, jejunum and ileum.

Expression of cytochromes P450 3A and P-glycoprotein in human large intestine in paired tumour and normal samples.

Our objective was to investigate the expression of different cytochromes P450 3A (CYP3A4, CYP3A5, and CYP3A7) and P-glycoprotein (ABCB1) genes along the human large intestine in paired tumour and normal samples. Real-time reverse transcriptase-polymerase chain reaction was used to measure CYP3A4-, CYP3A5-, CYP3A7- and ABCB1-specific mRNA expression, and Western blot analysis was used to measure membrane protein levels of CYP3A4/7, CYP3A5 and P-glycoprotein. Levels of mRNA and membrane protein fractions in the large intestine were compared with those of normal human liver. The mRNA expressions of CYP3A4, CYP3A5, CYP3A7 and ABCB1 in the large intestine were found to be highly variable, but overall the levels were significantly lower than those measured in liver (P < 0.0001, P < 0.001, P < 0.0001 and P < 0.01, respectively). At the membrane protein level, CYP3A4/7 was detected in all large intestine samples examined and the levels were substantially higher than those of the liver (P < 0.01). Although expression of CYP3A5 was detected in all large intestine samples, in most the levels were too low to allow quantification. P-glycoprotein was readily detected at levels slightly higher than those of liver (P < 0.05). Comparison between paired samples of normal and tumour in large intestine showed no significant differences in either the mRNA or membrane protein levels of these genes. In conclusion, this work suggests a potential role of the large intestine in the absorption and metabolism of xenobiotics and nutrients and no difference in the CYP3A and P-glycoprotein membrane protein fractions and mRNA expression between normal and tumour tissues.

Association of cytochrome P450 1B1 polymorphism with first-trimester miscarriage.

OBJECTIVE: To determine whether the cytochrome P450 1B1 (CYP1B1) Val432Leu polymorphism is associated with risk of miscarriage. We also analyzed the possible interaction between this polymorphism and caffeine intake. DESIGN: The population-based case-control study included 507 women with miscarriage in the first trimester of pregnancy and 908 controls with a normal first-trimester pregnancy. The controls were frequency matched to cases. The material was analyzed taking maternal age, smoking habits, alcohol intake, caffeine intake, fetal karyotype, nausea, and vomiting into consideration. SETTING: University hospital and primary care facility. MAIN OUTCOME MEASURE(S): CYP1B1 Val432Leu genotype frequencies in cases and controls. RESULT(S): Carriers of the CYP1B1 432 Val/Val genotype were at a higher risk of miscarriage in the first trimester of pregnancy (odds ratio = 1.46; 95% confidence interval, 1.02-2.08). We also found a significant interaction between genotype and caffeine intake. CONCLUSION(S): CYP1B1 Val432Leu polymorphism is associated with first-trimester miscarriage, and it may also modify the risk among coffee drinkers.

Amodiaquine, its desethylated metabolite, or both, inhibit the metabolism of debrisoquine (CYP2D6) and losartan (CYP2C9) in vivo.

OBJECTIVE: To study the extent of in vivo inhibition by the antimalarial drug amodiaquine, its active metabolite N-desethylamodiaquine, or both, of the metabolism of four probe drugs of the enzymes CYP2D6, CYP2C19, CYP2C9 and CYP1A2. METHODS: Twelve healthy Swedish volunteers received a cocktail of four probe drugs (debrisoquine, omeprazole, losartan and caffeine) to determine their baseline metabolic capacities. After a washout period, they received a 600 mg oral dose of amodiaquine hydrochloride; and 2-3 h later the cocktail was administered again. One week after the intake of amodiaquine, the subjects received the cocktail a third time. The levels of probe drugs and their metabolites as well as amodiaquine and its metabolite were determined by HPLC. RESULTS: Plasma levels of amodiaquine and N-desethylamodiaquine could be followed in all subjects for 6 h and 28 days, respectively. Among the 12 subjects, a 3-fold variation in amodiaquine AUC and a 2-fold variation in N-desethylamodiaquine AUC, were observed. The CYP2D6 and CYP2C9 activities of the subjects were measured by debrisoquine and losartan phenotyping tests, respectively. There were significant mean increases in debrisoquine metabolic ratio (MR) between baseline and the second cocktail [MR(2 h)-MR(baseline) 1.426 (95% confidence interval 1.159, 1.755), P=0.002; ANOVA, Fisher LSD test] and in mean losartan MR between baseline and the second cocktail [MR(2 h)-MR(baseline) 1.724 (95% confidence interval 1.076, 2.762), P=0.026; ANOVA, Fisher LSD test]. The effects on CYP2D6 and CYP2C9 activities subsided within a week after intake of amodiaquine as tested by the phenotyping cocktail. The changes in omeprazole MRs and caffeine MRs were not statistically significant between any of the study phases. CONCLUSION: A single dose of amodiaquine decreased CYP2D6 and CYP2C9 activities significantly compared to baseline values. Amodiaquine has the potential to cause drug-drug interactions and should be further investigated in malarial patients treated with drug combinations containing amodiaquine.

Effects of caffeine intake on the pharmacokinetics of melatonin, a probe drug for CYP1A2 activity.

AIMS: The aim of this study was to assess the influence of concomitant caffeine intake on the pharmacokinetics of oral melatonin, a probe drug for CYP1A2 activity. METHODS: Twelve healthy subjects, six smokers and six nonsmokers, were given melatonin (6 mg) either alone or in combination with caffeine (3 x 200 mg). Blood samples for the analysis of melatonin or caffeine and paraxanthine were taken from 1 h before until 6 h after intake of melatonin. Subjects were genotyped with respect to the CYP1A2*1F (C734A) polymorphism. RESULTS: When caffeine was coadministered the Cmax and AUC of melatonin were increased on average by 142% (P = 0.001, confidence interval on the difference 44, 80%) and 120% (P < 0.001, confidence interval on the difference 63, 178%), respectively. The inhibitory effect of caffeine was more pronounced in nonsmokers and in individuals with the *1F/*1F genotype. CONCLUSION: The results of this study revealed a pronounced effect of caffeine on the bioavailability of orally given melatonin, most probably due to inhibition of CYP1A2 activity.

The Karolinska cocktail for phenotyping of five human cytochrome P450 enzymes.

OBJECTIVES: Our objectives were (1) to determine whether the drugs caffeine, losartan, omeprazole, debrisoquin (INN, debrisoquine), and quinine can be given simultaneously in low doses as a cocktail for the phenotyping of cytochrome P450 (CYP) 1A2, 2C9, 2C19, 2D6, and 3A4, respectively, and (2) to design an administration schedule to give as few sampling occasions as possible. METHODS: Twenty-four subjects were given oral doses of 100 mg caffeine, 25 mg losartan, 20 mg omeprazole, 10 mg debrisoquin, and 250 mg quinine on separate days. After a washout period of at least 4 days, all drugs were given simultaneously except for quinine, which was given 8 hours after the other drugs. Blood and urine samples were collected to determine parent drug and metabolite concentrations for assessment of phenotyping indices. Any difference between both single and cocktail doses was tested on a log-normal distribution. RESULTS: The phenotypic indices of CYP1A2 (paraxanthine/caffeine in 4-hour plasma), CYP2C9 (losartan/E-3174 [metabolite of losartan] in 0- to 8-hour urine), CYP2C19 (omeprazole/5-hydroxyomeprazole in 3-hour plasma), and CYP3A4 (quinine/3-hydroxyquinine in 16-hour plasma) were not significantly changed when probe drugs were administered alone compared with together, although a tendency toward higher concentrations of losartan was seen during simultaneous administration (95% confidence interval, 0.51-1.002; P =.051). The CYP2D6 phenotypic index (debrisoquin/4-hydroxydebrisoquin in 0- to 8-hour urine) was significantly changed when drugs were given together (95% confidence interval, 0.45-0.87; P =.007), indicating an inhibition of the debrisoquin metabolism. The within-subject coefficients of variation (8%-25%) were much lower than the between-subject coefficients of variation (34%-79%). CONCLUSIONS: The administration of drugs together suggests an inhibition of debrisoquin metabolism caused by the concurrent drugs given. By separating debrisoquin from the other cocktail drugs, this method is likely to be used as a tool to phenotype the enzymes CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 with only 2 urinary collections and 2 blood-sampling occasions.

The effect of the CYP1A2 *1F mutation on CYP1A2 inducibility in pregnant women.

AIMS: To investigate the influence of the CYP1A2*1F mutation on CYP1A2 activity in smoking and nonsmoking pregnant women. METHODS: Pregnant women (n = 904) who served as control subjects in a case-control study of early fetal loss were investigated. They were phenotyped for CYP1A2 using dietary caffeine and the urinary ratio AFMU + 1X + 1 U/1,7 U. An assay for CYP1A2*1F using 5'-nuclease assay (Taqman) was developed to genotype the population. RESULTS: The frequencies of *1 A and *1F alleles among Swedish women were 0.29 and 0.71, respectively. There was no statistically significant difference in CYP1A2 activity between the genotypes, although a trend towards enhanced activity was observed in *1F/*1F (log MRc 0.77) and *1F/*1 A (log MRc 0.82) genotypes compared with the *1 A/*1 A genotype (log MRc 0.71) (anovaP = 0.07). The mean difference between the *1 A homozygotes and the heterozygotes was 0.11 [95% confidence interval of the difference: (-0.21, -0.01)] and that between the *1 A and *1F homozygotes was 0.05 [95% confidence interval of the difference: (-0.13, 0.03)]. No significant effect (P = 0.22) of the *1F on CYP1A2 activity was observed in smokers, tested using an interaction term (smoking * genotype) in the anova model (*1F/*1F log MRc 0.79, *1F/*1 A log MRc 0.86, and *1 A/*1 A log MRc 0.73). In smokers, there was no difference in ratio between homozygotes for the *1 A and *1F alleles [mean difference -0.06; 95% confidence interval of the difference: -0.22, 0.11] or between *1 A/*1 A and *1 A/*1F genotypes [mean difference -0.13; 95% confidence interval of the difference: -0.29, 0.04]. CONCLUSIONS: The effect of the CYP1A2*1F mutation on CYP1A2 activity in smoking pregnant women could not be confirmed.

Plasma folate levels and risk of spontaneous abortion.

CONTEXT: Both folate deficiency and folic acid supplements have been reported to increase the risk of spontaneous abortion. The results are inconclusive, however, and measurements of folate have not been available in all studies. OBJECTIVE: To study the association between plasma folate levels and the risk of spontaneous abortion. DESIGN, SETTING, AND POPULATION: Population-based, matched, case-control study of case women with spontaneous abortion and control women from January 1996 through December 1998 in Uppsala County, Sweden. Plasma folate measurements were available for 468 cases and 921 controls at 6 to 12 gestational weeks. MAIN OUTCOME MEASURE: Risk of spontaneous abortion vs maternal plasma folate level. RESULTS: Compared with women with plasma folate levels between 2.20 and 3.95 ng/mL (5.0 and 8.9 nmol/L), women with low (< or =2.19 ng/mL [< or =4.9 nmol/L]) folate levels were at increased risk of spontaneous abortion (adjusted odds ratio [OR], 1.47; 95% confidence interval [CI], 1.01-2.14), whereas women with higher folate levels (3.96-6.16 ng/mL [9.0-13.9 nmol/L] and > or =6.17 ng/mL [> or =14.0 nmol/L]) showed no increased risk of spontaneous abortion (OR, 0.84; 95% CI, 0.59-1.20; and OR, 0.74; 95% CI, 0.47-1.16, respectively). Low folate levels were associated with a significantly increased risk when the fetal karyotype was abnormal (OR, 1.95; 95% CI, 1.09-3.48) but not when the fetal karyotype was normal (OR, 1.11; 95% CI, 0.55-2.24) or unknown (OR, 1.45; 95% CI, 0.90-2.33). CONCLUSION: Low plasma folate levels were associated with an increased risk of early spontaneous abortion.

Effect of caffeine exposure during pregnancy on birth weight and gestational age.

Epidemiologic studies have been unable to conclusively evaluate whether caffeine intake during pregnancy is associated with reduced birth weight and/or fetal growth restriction. The authors conducted a prospective, population-based cohort study to investigate the effect of caffeine on birth weight, gestational age, and birth weight standardized for gestational age (birth weight ratio). Of 953 women recruited in early pregnancy in Uppsala County, Sweden, from 1996 to 1998, 873 women delivering liveborn singleton infants were included in the analysis. Caffeine exposures were ascertained from in-person interviews at 6-12 and 32-34 completed gestational weeks, and maternal plasma was analyzed for cotinine levels as an indicator of smoking. Analysis of variance was used to estimate the effect of caffeine on birth weight, gestational age at delivery, and birth weight ratio after accounting for the effects of other covariates, such as maternal sociodemographic characteristics, plasma cotinine, and pregnancy symptoms. There were no associations between caffeine consumption and birth weight, gestational age, and birth weight ratio, neither when caffeine exposure was averaged from conception to the 32nd to 34th gestational weeks, nor when caffeine exposure was stratified by trimesters of pregnancy. These results do not support an association between moderate caffeine consumption and reduced birth weight, gestational age, or fetal growth.