Main

The AIDS epidemic has been with us for 15 y and, although major breakthroughs in our understanding of the disease and modalities of treatment have occurred, it is likely to be with us for many more years to come. A major emphasis has been placed on prevention, and nowhere has this been more successful than in the pediatric population(1,2). AZT treatment pre- and postnatally reduced the transmission rate from a mother to her child by two-thirds in a relatively low risk populations. Many questions remain with respect to the effectiveness of AZT and perhaps just as many as to why failures occurred.

Two strategies proposed to prevent mother to child transmission of HIV infection are reduction of maternal viral load and provision of prophylactic therapy for the fetus(2,3). Risk of transmission is clearly associated with maternal viral load; however, the mechanism of action of reverse transcriptase inhibitors might suggest that AZT is more likely to reduce perinatal transmission by fetal prophylaxis(4,5). This is supported by the absence of significant reduction in maternal viral load and the subsequent lack of correlation between transmission and viral load in HIV-positive women treated with AZT(4,5). Fetal prophylaxis requires delivery of effective concentrations of AZT to the fetus at the time of exposure. Therefore, to optimize therapeutic regimens, an understanding of the factors that determine fetal plasma concentration of AZT is essential.

Previous studies from this laboratory and others evaluating the fetal to maternal ratio of AZT, under steady-state conditions in primates, showed the fetal concentration of AZT was always less than that in the mother(6,7) (our unpublished observation). This finding could not be explained on the basis of protein binding, ionization, or active transport of AZT. Thus, it was hypothesized that the fetal concentration was less than that in the mother due to clearance of AZT by the fetus(8). The presence of AZT-glu in the fetus suggested that fetal metabolism of AZT was a clearance pathway. Alternatively, the presence of AZT-glu in the fetus could be explained by placental transfer from the mother. Both of these explanations are in conflict with current beliefs that the fetal capacity for glucuronidation is minimal, and the placental transfer of hydrophilic glucuronide conjugates is very restricted(9–12). The present study was designed to determine whether AZT is metabolized in the fetal compartment and to quantitate specific maternal, placental, and fetal clearance pathways of AZT. The alternative hypotheses were that, during fetal infusion of AZT, the concentration of AZT-glu in the fetus 1) would be greater than that in the mother if it were formed in the fetus, or 2) would be the same as or less than that in the mother if it were formed only in the mother. Finally, amniotic fluid was evaluated as a potential modulator of fetal concentration.

METHODS

Study population. Pharmacokinetic data from eight baboons(Papio species) were used. Gestational age was calculated from the estimated day of conception (±3 d) and confirmed by ultrasound measurements of head circumference and femur length made at 70-100 d of gestation, with term ∼175 d. Animals were maintained in accordance with all National Institutes of Health, U.S. Department of Agriculture, and American Association for the Accreditation of Laboratory Animal Science regulations for the care and use of laboratory animals. Research protocols were approved by the Institutional Animal Care and Use Committee of the Columbia University College of Physicians and Surgeons.

Surgical procedures and tethering system. The animals were studied using an individualized backpack and tether system. This system, along with the methods for maintenance, breeding, preconditioning, anesthesia, surgery, and postoperative care, were described in detail in a previous report and are only summarized here(13). In brief, surgery was performed using general anesthesia and sterile surgical techniques at a mean (±SD) gestational age of 130 ± 3 d. Maternal catheters were placed in the femoral artery and vein and fetal catheters in the carotid artery and internal jugular vein. An additional catheter was placed in the amniotic fluid cavity. Electroencephalogram, electro-oculogram, and electrocardiogram electrodes and a tracheal catheter were placed in three fetuses to monitor neurophysiologic parameters. Catheters and electrodes were tunneled s.c. to the mid-scapula region of the mother then passed into the backpack. This backpack was secured in place with a shoulder harness and served the dual purpose of protecting the catheters, transducers, and leads while allowing attachment of a flexible, stainless steel tether cable. Access to the catheters for drug infusion and sampling was available at the top of the cage. Postoperative analgesia was supplied with a continuous infusion of morphine sulfate to the mother, which was reduced stepwise as animals resumed normal activity in the 2-4 d after surgery. No tocolytic agents were used. The mothers were maintained in their home cages near other familiar animals. Light cycling (0700 h on, 1900 h off) and feeding times (0800 and 1600 h) remained constant. The maternal and fetal vascular catheters were continuously infused with normal saline containing heparin (10 U/mL) at rates of 5 and 2 mL/h, respectively, using peristaltic infusion pumps to maintained catheter patency. With this system, all pharmacokinetic studies were carried out without the need for sedation, anesthesia, or undue restraint. A minimum of 5 d was allowed postoperatively for stabilization. During individual studies, the mothers were monitored for signs of labor, assessed by maternal behavior and the pressure recording from the catheter in the amniotic fluid cavity. Physiologic stability of the fetus was assessed by fetal heart rate, blood pressure, and arterial pH and blood gases values.

Study protocols. AZT was infused to the mother at a dose of 226-300 µg/h and to the fetus at 77-104 µg/h for a minimum of 40 h before collecting samples. This duration of infusion was 4-5 times the elimination half-life of AZT-glu from the amniotic fluid, and thus it would be reasonable to assume that subsequent samples obtained were at steady state(our unpublished observation). Infusions were separated by a minimum of 3 d to ensure clearance of drug from the amniotic fluid compartment. The order of infusions was pseudo-randomized to control for gestational age effects and prior exposure. The initial precision pump (ambulatory infusion pump, model ML-6-4; Cormed, Medina, NY) used to deliver the drug in the first five animals had a fixed flow rate of 1 mL/h. An aqueous solution of AZT was prepared containing 20 mg/mL for maternal infusion and 6 mg/mL for the fetal infusion. Infusion rate was confirmed from pre- and postinfusion solution weights and the total time infused. It became apparent there was considerable difference in infusion rates between studies. However, the values taken at steady state implied that the variability in infusion rates within a study was acceptable. Although between-study variability in infusion rates will affect individual plasma concentrations, it will not influence the fetal to maternal ratios or calculated clearances as analyses were done within animals. Nonetheless, to reduce between-study variability in drug delivery variable speed precision pumps (Instech 720, World Precision Instruments, Sarasota, FL) were used for the final three animals. An AZT solution was prepared with a concentration of 3 mg/mL and was infused at 6 mL/h to the mother and 2 mL/h to the fetus. Three maternal and fetal plasma samples and three amniotic fluid samples were obtained at approximately 2-h intervals over the final hours of the infusion. Samples were collected from the catheter that had not been infused with drug to avoid contamination, in most cases the arterial catheter. Blood samples (0.6 mL) were collected into heparinized microcontainers and separated by centrifugation, and plasma was frozen within 30 min of collection. Amniotic fluid samples were treated in a similar manner to remove cell and other debris. All samples were stored at-18°C until batch analysis.

Drug analysis. Plasma concentrations of AZT and AZT-glu were determined by RIA with a commercially available kit (ZDV-Trac, IncStar Corporation, Stillwater, MN). The aliquot for AZT-glu determination was first incubated for 90 min with β-glucuronidase (type X-A, G-7896, Sigma Chemical Co., St. Louis, MO). The concentration of AZT was determined from the unhydrolyzed aliquot and the concentration of AZT-glu from the difference between the total (hydrolyzed) and unhydrolyzed aliquots. Interassay variability over a period of 4 y (30 assays) was determined from the standard controls included in each run. Mean and CV of measured values were 87.2 ng/mL(CV = 17.0%) for the 85 ng/mL and 999.0 ng/mL (CV = 12.9%) for the 962 ng/mL AZT controls. An additional AZT control, 50 ng/mL, was 49.6 ng/mL (CV = 13.1%). For AZT-glu, these were 507.2 ng/mL (CV = 13.4%) for the 500 ng/mL and 959.6 ng/mL (CV = 10.7%) for the 1000 ng/mL controls. Intraassay variation was assessed by using the same quality control concentrations except for the 50 ng/mL level. AZT at 85 and 962 ng/mL gave means of 88.4 (CV= 4.8%) and 965.5 (CV = 5.2%), respectively. AZT-glu at 500 and 1000 ng/mL gave means of 496.6 (CV = 3.3%) and 971.0 (CV = 5.2%) ng/mL, respectively(n = 16 for each concentration). Of note, all samples from the individual paired experiments were assayed on the same day to eliminate interassay variation.

Protein binding. To determine whether, as is the case in humans and rhesus monkeys, protein binding of AZT is low, protein binding was measured in the plasma obtained from three additional animals and two of their fetuses(7,14). A nonpregnant human control was also included. AZT (0, 100, or 1000 ng) in 10 µL of phosphate buffer (pH 7.4) was added to 600 µL of plasma. Phosphate buffer was added to 50 µCi of 3H-labeled AZT (Sigma Chemical Co.) to make up 1 mL, and 10 µL of labeled AZT solution were then added to each sample. All samples were vortexed and incubated for 30 min at room temperature. Two 40-µL aliquots were placed in scintillation vials, and the remainder of the sample was placed in Centrifree micropartition devices(Amicon, Inc., Beverly, MA) for ultrafiltration. After centrifugation at 2000× g for 30 min, two 40-µL aliquots of the ultrafiltrates were placed in scintillation vials. Ten milliliters of Atomlight scintillation mixture (Du Pont, Boston, MA) was added, and the samples were vortexed. Samples were counted for 10 min in a 1211 Minibeta liquid scintillation counter (Wallac Inc., Gaithersburg, MD). The mean difference in duplicates was 2.1%.

Pharmacokinetic analyses. Steady-state concentrations during maternal and fetal infusions of AZT and AZT-glu were determined for each compartment in individual animals as the mean plasma and amniotic fluid concentration obtained from the three samples taken at the end of each infusion. The mean plasma concentrations at steady-state were used to calculate the within animal fetal to maternal plasma concentration ratios and metabolite to drug ratios for both maternal and fetal infusion. The maternal(M) and fetal (F), total (t), placental(p) and nonplacental (np) clearances (CL) of AZT were calculated from the maternal and fetal steady-state concentrations of AZT during separate, but within-animal, maternal and fetal infusions(c and c′, respectively), and the infusion rates to the respective compartments (RM and RF)(8). Equation 1, Equation 2, Equation 3, Equation 4,Equation 5, Equation 6

Statistical analysis. The three final samples from each compartment for the two infusions were evaluated by repeated measures ANOVA to determine whether differences existed between the presumed steady-state values. Mean steady-state concentrations, ratios, and the maternal and fetal clearances were compared using paired t tests. F tests were used to evaluate the hypotheses that the fetal to maternal ratio of AZT and AZT-glu was not different from 1 with 95% confidence intervals reported.α was set at 0.05 except when repeated t tests were performed, then α was set at 0.018 (Bonferroni adjustment for three tests). All results are presented as mean ± SE unless otherwise stated. Statistical analyses were performed using the SYSTAT statistical package (SPSS Inc., Chicago, IL).

RESULTS

Eight baboons were studied (Table 1). The time from surgery to study was a minimum of 6 d. In all studies the fetus remained stable until the onset of labor (or in one case embolic death) with animals delivering between 148 and 171 d. The mean duration of infusion to first and last samples, respectively, was 46.2 h (range, 42.8-53.8 h) and 50.1 h(range, 46.8-53.8 h) for maternal infusions and 45.0 h (range, 41.5-47.6 h) and 49.7 h (range, 43.8-52.8 h) for fetal infusions. The mean sample interval was 2.7 h (range, 0.6-4.5 h). Paired maternal and fetal data were obtained in six animals. Amniotic fluid samples were obtained in six animals with complete data in three.

Table 1 Demographic data on study animals
Full size table

For both AZT and AZT-glu, no significant differences were seen between the concentration values of the three consecutive samples from the mother, fetus, or amniotic fluid. This was in keeping with samples being obtained under steady-state conditions. The mean AZT and AZT-glu concentrations for each compartment and the fetal to maternal concentration ratios for individual animals during maternal infusion are presented in Table 2. The fetal to maternal AZT ratio during maternal infusion was significantly less than one (95% confidence interval, 0.80-0.93), whereas the fetal to maternal AZT-glu ratio was not different from one (95% confidence interval, 0.87-1.13). These data confirm previous findings of 1) lower levels of AZT in the fetus compared with those in the mother and 2) significant concentrations of AZT-glu in the fetus (our unpublished observation). In addition, concentrations of drug and metabolite in the amniotic fluid were similar to those found in the previous study after only 24 h of infusion, suggesting that a reasonable approximation of steady state was achieved in the amniotic compartment.

Table 2 Steady-state concentrations of AZT and AZT-glu in maternal (M) and fetal (F) plasma and amniotic fluid (AF) and their various concentration ratios during fetal infusion
Full size table

The mean concentrations and the fetal to maternal ratios for individual animals during fetal infusion are presented in Table 3. During fetal infusion, the concentration of AZT was on average 20 times higher in the fetus than in the mother, reflecting the high maternal clearance rate compared with placental clearance. The concentration of AZT-glu in the fetus was 7.2 ± 0.8 times that in the mother. The AZT-glu to AZT ratio in the mother is the same as that seen during maternal infusion of AZT, whereas the ratio in the fetus during fetal infusion is about one-third that seen during maternal infusion. These findings are evidence that glucuronidation occurs within the fetal compartment, although it may not account for all the glucuronide present in the fetus during maternal infusion.

Table 3 Steady-state concentrations of AZT and AZT-glu in maternal (M) and fetal (F) plasma and amniotic fluid (AF) and their various concentration ratios during fetal infusion
Full size table

The maternal and fetal placental and nonplacental clearances are presented in Table 4. Maternal nonplacental clearance is well above all other clearances. Although of a similar order of magnitude (∼5% of total maternal clearance), the maternal placental clearance was consistently less than the fetal placental clearance. The calculated values for fetal nonplacental clearance were too small to be meaningful. This was not consistent with the fetal to maternal ratio of AZT during maternal infusion being less than 1 and the direct evidence that fetal glucuronidation occurred during fetal infusions.

Table 4 Maternal (M), fetal (F), total (t), placental (p), and nonplacental (np) clearances calculated from the equations of a two-compartment model at steady-state
Full size table

Protein binding was 21.3, 18.4, and 22.9% in the three maternal baboons. Values in their respective fetuses were 17.4 and 15.2% with no fetal sample in the third animal. The human control was 19.8%. The slightly lower protein binding in the two fetuses compared with their mothers would account for F/M ratios of 0.95 and 0.96, respectively. The F/M ratio from the study animals (0.87, see Table 2) is less than the lower of these two values (p < 0.001).

DISCUSSION

This study has shown that the fetus is capable of glucuronidating AZT to an extent that leads to significant plasma concentrations of AZT-glu in the fetus. Thus, fetal metabolism, might be expected to contribute to fetal nonplacental clearance. In addition, even accounting for differences attributed to fetal and maternal protein binding, the steady-state fetal levels of AZT during maternal infusion were less than those in the mother. This is in keeping with significant fetal nonplacental clearance(8). Furthermore, the amniotic fluid compartment appears to be at steady state, suggesting no net loss of drug or metabolite to the amniotic fluid. This would make fetal excretory losses unlikely to account for significant fetal clearance.

However, contrary to this evidence for fetal metabolism, the calculated clearances implied that the differences in fetal and maternal plasma concentrations were due to a higher placental clearance from the fetus to the mother than that from the mother to the fetus. That fetal placental clearance should be greater than maternal placental clearance seems highly unlikely. This would imply that an active placental mechanism exists for transporting AZT from the fetus to the mother. Several investigators have failed to find evidence for active transport of AZT or other drugs using the human placental perfusion model, and it is very unlikely one would exist in the non-human primate(15,16). It also seems unlikely that these apparent differences in placental clearances are due to changes in baseline conditions between the maternal and fetal infusions as the study design randomized the order of infusions with respect to gestational age and previous exposure to AZT. Other influences would be anticipated to have a random effect rather than result in a consistent finding in all animals. A more likely explanation for the inconsistencies lies in a key underlying assumption of the steady-state model used to calculate clearances, namely that all processes are first order. This is true over the dose range studied for maternal and placental clearances but has not been demonstrated with respect to fetal nonplacental clearance(15–18). Saturation of nonplacental clearance in the fetus would cause the fetal and maternal concentrations of AZT to be higher than otherwise expected with increasing dose while conserving the fetal to maternal AZT ratio during fetal infusion. This will increase the absolute magnitude of the difference between the fetal and maternal concentrations. The overall effect would be to underestimate total fetal clearance, and subsequently, underestimate maternal placental clearance and fetal nonplacental clearance (see Eqs. 1-6). In the present study, these calculations used fetal AZT concentrations for one set of measurements (fetal infusion) that were almost 8 times greater than those for the other set of measurements (maternal infusion). Several drugs have been studied in pregnant sheep using this method, and in almost all instances a similar discrepancy was found in the placental clearances(19–23). In all of these studies, fetal concentrations were much higher during fetal administration. The only exception to date is with acetaminophen, where bidirectional clearances were calculated to be the same. In this case, fetal concentrations were very similar during maternal and fetal administration(23). Thus, there is a very real possibility that saturation of fetal glucuronidation could have led to the inconsistency between maternal and fetal placental clearance, which in turn, would lead to underestimation of fetal nonplacental clearance. Furthermore, a saturable process would lend support to fetal metabolism being a significant component of fetal nonplacental clearance. Some observations from the present data shed light on this question.

The calculation of fetal placental clearance would not be affected by saturation of fetal nonplacental clearance (Eq. 4). Therefore, this value can be used to approximate the dose delivered to the fetus during maternal infusion, which in the present study, is ∼14µg/min. At this dose the mean glucuronide concentration in the fetus was 645 ± 84 ng/mL. When AZT was administered directly to the fetus at a dose of 80-100 µg/min, the AZT-glu concentration was 1588 ± 215 ng/mL or more than 2 times higher. This would suggest that, although not saturated at the lower dose, fetal glucuronyl-transferase capacity becomes saturated at the higher dose as a 6-fold increase in substrate would result in a 6-fold increase in product if the process was operating under first-order kinetics. One problem with this argument is that during maternal infusion the fetal glucuronide concentration may be the result of placental transfer of maternally formed glucuronide. However, these explanations are not mutually exclusive. Future studies to clarify this issue include direct evaluation of the kinetics of fetal metabolism with a series of fetal infusions over the dose range in question and infuson of AZT-glu into the mother to determine the maternal contribution to fetal AZT-glu.

The estimate of fetal placental clearance can be used to approximate fetal nonplacental clearance by an alternative method derived from the rate equation for the fetal compartment during maternal infusion at steady state(Eq. 7).

The amount of fetal nonplacental clearance that occurred during maternal infusion can be calculated from the fetal to maternal ratio, assuming that fetal and maternal placental clearances are equal. From this, fetal nonplacental clearance is estimated at 5 mL/min or ∼15% of placental clearance and thus in the order of 1% of maternal clearance.

The important question is how much of this fetal nonplacental clearance can be attributed to fetal metabolic clearance. The approximation of fetal clearance calculated is of the same order of magnitude as that approximated from in vitro data of fetal hepatic enzyme capacity assuming a 1:30 size ratio and fetal enzyme activity as 20% of maternal activity(9,10,24). Although fetal metabolism may not account for all fetal nonplacental clearance, it certainly appears likely to be a major contributor.

In general, because of the limited glucuronyl-transferase activity exhibited in fetal tissue, it has been assumed that fetal metabolism would have an insignificant effect of fetal drug and metabolite concentrations. Although fetal nonplacental clearance calculated above is small in comparison with maternal clearance, it is not so small in relation to placental clearance and, it is this relationship that determines the fetal to maternal concentration ratio of AZT during maternal infusion (Eq. 7). Thus, even small amounts of fetal nonplacental clearance, some of which is fetal metabolism, would lead to substantial concentrations of metabolite in the fetus, and account for decreased concentrations of drug in the fetus.

To understand the implications of this research it must be considered in the context of the entire pregnancy. Low levels of fetal drug metabolism, similar to those seen in this study, likely exist from an earlier point in gestation and only gradually increases as the fetus grows and matures; however, some time near parturition, a developmentally regulated surge in metabolic capacity is initiated(24). The present study was not designed to evaluate temporal changes in fetal metabolism, and no data currently exist in primates that delineate the specific mechanisms or the timing of this rapid induction of metabolism(24). Changes in placental clearance will influence the effect that changes in fetal metabolism will have on the fetal to maternal ratio (Eq. 7). Placental clearance is expected to show a gradual increase as the placental surface area enlarges and the trophoblast interface between the maternal and fetal circulations becomes thinner. As term approaches, placental clearance may reach a limit and in some cases may become restricted. So, although increases in placental clearance may counteract small increases in fetal metabolism before term, fetal enzyme induction shortly before delivery and/or a fall off in placental clearance at this time could result in lower drug concentrations in the fetus. Thus, the most likely time a clinically meaningful fall in fetal plasma concentrations will occur is at term or during parturition. To address these issues of developmental regulation of fetal metabolism, longitudinal studies have been designed that evaluate the changes in fetal metabolism and placental clearance over the late gestational period. In addition, metabolic inhibitors and inducers will be used to determine the influence of changes in fetal metabolism on fetal drug concentrations.

The effects of fetal drug metabolism on fetal drug concentration takes on clinical relevance when therapy is being directed at the fetus as is the case with AZT. This is of particular concern in the prevention of mother to infant transmission of HIV infection, as labor and delivery is the very time that risk of viral transmission is thought to be maximal(2,3,5). In addition, drug action in the fetus may be an important contributor to efficacy(5). Glucuronyl-transferase enzymes are known to be pharmacologically induced with phenobarbital shortly before delivery; thus, there is a very real possibility that changes in fetal metabolism occur during this crucial time period(25). In addition, glucocorticoids, used therapeutically in preterm deliveries to enhance lung maturity, and chronic opioid abuse, a risk factor for HIV infection, are both drugs that might be expected to enhance or induce fetal metabolism(24,26). It is difficult to determine whether these concerns are real, as all but a single case report of AZT levels during labor and delivery have used oral administration, where it is difficult to evaluate the fetal-maternal paired sample(27–30). However, continuous infusion of AZT to HIV-infected pregnant women is now the standard of care during labor and delivery. Thus, there is a sizeable population available where steady-state maternal and fetal paired samples could be obtained, allowing comparison of the fetal to maternal ratio under varying conditions, including prematurity with and without prenatal steroids, i.v. drug abuse, and elective (not in labor) versus elective (in labor) cesarian section. Understanding the extent to which fetal drug metabolism can affect fetal drug levels will enhance our ability to effectively treat the fetus.

In a previous study where the duration of AZT infusion was only 24 h, the fetal to maternal AZT-glu ratio was found to be significantly less than 1(our unpublished observation). In the present study, after 48 h of infusion, this AZT-glu ratio could not be distinguished from 1. It is more likely the 48-h infusion gives a better representation of steady state for AZT-glu in the fetus as 4-5 half-lives of AZT-glu in the amniotic fluid have elapsed. Thus, the proportion of glucuronide excreted from the fetus into the amniotic fluid and that subsequently ingested and absorbed back into the fetus would be constant. The marked similarity between fetal and maternal levels of the glucuronide after 48 h of infusion strongly suggests transfer of the metabolite across the placenta. This supports a previous finding of a strong correlation between maternal and fetal AZT-glu concentrations (our unpublished observation). Thus, although during fetal infusion, the glucuronide present in the fetus is mostly of fetal origin, it is likely that during maternal infusion a significant amount of the glucuronide is of maternal origin.

Fetal glucuronyl-transferases are generally considered the most down-regulated of fetal biotransformation enzymes. In evolutionary terms, down regulation seemed a reasonable adaptation to the fetal environment because of the presumption of very limited permeability of the glucuronide conjugate across the placenta. The present study suggests that fetal glucuronidation occurs to a greater extent than previously appreciated and that at least some of the glucuronide conjugates can cross the placental barrier. Fetal metabolism cannot simply be passed over as insignificant and may, in fact, have significant implications for fetal drug therapy.