Transfer of Insulin Lispro Across the Human Placenta: In vitro perfusion studies

The eight human placentae were obtained after vaginal or cesarean section delivery from uncomplicated term pregnancies and transported to the laboratory in heparinized ice-cold PBS. Independent maternal and fetal circulations were established to a peripheral placental lobule within 30 min of the delivery of the infant, as previously described by our laboratory (15).

After residual blood was cleared out of the vessels and intervillous space, a 1-h “closed” (recirculated) circuit control period preceded the experimental period. During the control period, glucose and oxygen consumption and lactate and human chorionic gonadotrophin (hCG) production were measured to determine the baseline of these values for each experiment. All values measured were compared with those of the initial control period to ensure the maintained integrity of the placental preparation throughout the course of the perfusion experiment.

The physical integrity of the placental preparation was assessed by monitoring the stability of the fetal perfusion pressure and volume loss from the fetal reservoir; increases or decreases in pressure (>10 mmHg) and volume loss (>3 ml/h) were criteria for rejection of the preparation. After the 1-h control period, the perfusates in both the maternal and fetal circulations were replaced with fresh media, and insulin lispro at a concentration of 100 μU/ml was introduced into the maternal reservoir in four experiments. In four additional perfusion experiments, the concentrations of insulin lispro in the maternal circulation were 200, 580, and 1,000 μU/ml. The maternal circuit was “open” (nonrecirculated) to deliver a constant concentration of drug to the lobule, while the fetal circuit was “closed” (recirculated).

Samples were obtained from the fetal reservoir and the maternal artery and vein every 20 min for the first 2 h and subsequently every 30 min for the last 2 h to measure insulin lispro, antipyrine, lactate, glucose, and hCG concentrations. Antipyrine is a freely diffusible flow-dependent transfer marker to which drug transfer is commonly compared. Oxygen, glucose consumption, and lactate production were measured during the perfusion to confirm that the tissue maintained its ability for energy metabolism. After 4 h of perfusion with insulin lispo, the perfusates in both maternal and fetal reservoirs were replaced again with fresh media and circulations “closed” for a final 1-h control period. All values measured were compared with those of the initial control period to ensure that the integrity of the preparation was maintained during perfusion.

Perfusate samples were kept at −20°C until analysis. The concentrations of insulin lispro were measured using the insulin lispro competitive-binding radioimmunoassay with overnight equilibrium incubation at room temperature (Linco Research, St. Charles, MO). This assay is highly specific for insulin lispro (100%) and has negligible cross-reactivity with native human insulin or proinsulin (<0.5%). Whereas the limit of quantitation is quoted at 2.5 μU/ml for a 100-μl sample using the manufacturers’ kit, in blank samples of fetal buffer we measured levels of 5.89 μU/ml. Hence, we used this concentration to subtract from all apparent levels measured during the experiment in the fetal circulation. pH and pO₂ of the perfusate samples were measured by using a blood gas analyzer (Radiometer ABL 330; Radiometer, Copenhagen). The perfusate concentrations of lactate, glucose, and hCG, as well as tissue hCG content, were measured as described previously in our laboratory (15). Antipyrine concentrations were determined spectrophotometrically.

The mean fetal concentration-time relation was plotted graphically. Least square regression analysis was used to study the relation between measured fetal drug concentration and time. The rate of appearance on the fetal circulation was calculated by the mean of the slope of insulin lispro appearance (in microunits) versus time (in minutes) and expressed as microunits per minute per grams of tissue. Comparisons among values were made using the Student’s t test.

In 11 women 31–40 years of age, we measured serum insulin lispro levels 60 min after subcutaneous injection of a single lispro dose of 3–52 units. Of 11 women, 9 were pregnant, 2 had gestational diabetes, 5 had type 1 diabetes, and 3 had type 2 diabetes. One of nine women was in the first trimester, and the others were in the second or third trimester of pregnancy.

The mean mass of the perfused cotyledons was 11.8 ± 3.4 g. Placental glucose and oxygen consumption and oxygen delivery and transfer as indicators of metabolic viability of the placental tissue did not change significantly throughout the perfusion experiment (Table 1). The fetal inflow pressures were not different between the control and experimental periods. The overall rate of lactate production during the control period was 0.39 ± 0.21 μmol/ml, significantly higher than the rate of lactate production during the experimental period (0.18 ± 0.04 μmol/ml, P = 0.015). This is most likely due to the washout of lactate that accumulated in the tissue during the period of hypoxia after delivery.

During the 4-h experimental period with a constant concentration of 100 and 200 μU/ml of insulin lispro on the maternal side of the placenta, insulin lispro was undetectable in the fetal side of the placenta. At higher concentrations of insulin lispro in the maternal circulation (580 and 1,000 μU/ml), insulin lispro accumulated linearly in the closed fetal circulation (Fig. 1). The accumulation of insulin lispro on the fetal side started after 80 min for the 580-μU/ml concentration and 60 min for the 1,000-μU/ml concentration. The mean calculated rate of insulin lispro transfer was 0.045 μU · min⁻¹ · g tissue⁻¹ at circulating maternal level of 1,000 μU/ml and 0.019 μU · min⁻¹ · g tissue⁻¹ at 580 μU/ml (Table 2).

In 11 women treated with insulin lispro, serum levels of insulin lispro ranged from 2 to 576 μU/ml with an apparent linear correlation (Fig. 2). One woman receiving 52 units achieved a level of 576 μU/ml.

The placental perfusion model enables the noninvasive collection of information regarding human placental transfer of molecules (15). Because transfer is measured in intact tissue, the model approximates the in vivo situation more closely than subcellular or cell culture systems. Importantly, it obviates the enormous ethical hurdles of studying human placental transfer in vivo. At the end of normal pregnancy, postprandial insulin levels are elevated and plasma insulin concentration may reach 100–200 μU/ml (16).

In the current study, we chose an insulin lispro concentration of 100 μU/ml as our baseline level to mimic the peak levels typically measured after administration of ∼13 units of insulin lispro to healthy volunteers (2). Assuming linear pharmacokinetics, the higher concentrations used in our experiment (200, 580, and 1,000 μU/ml) would correspond to peak serum insulin levels achieved after subcutaneously administered doses of ∼26, 75, and 130 units of insulin lispro, respectively (1). There is an increased need for insulin throughout pregnancy. Women with type 1 diabetes require on average 1.2 units/kg insulin per day; however, the range is broad (16). Women with type 2 diabetes often require larger doses. In one study, women with type 2 diabetes required on average 1.6 units/kg a day or ∼150 units of insulin per day. Women who are very obese may even start with 2 units/kg a day, requiring larger doses as the pregnancy progresses (17). Because a typical multiple daily insulin regimen may include ∼40–60% of the total insulin dose used as a short- or rapid-acting insulin analog, a patient with type 1 diabetes in pregnancy may require up to 10–18 units of insulin lispro per meal and a patient with type 2 diabetes in pregnancy potentially requiring up to 20–30 units of insulin lispro per meal. For those women who are very obese and/or very insulin resistant, requirements can go up to 40–60 units of insulin lispro per meal. Among 11 women monitored by us, serum lispro levels were <200 μU/ml in all those receiving <50 units, whereas higher doses had concentrations associated with placental transfer of the hormone.

The two lower concentrations of insulin lispro used in our experiments (100 and 200 μU/ml) correspond to peak plasma insulin levels achieved after subcutaneous administration of ∼13 and 26 units of insulin lispro and suggest that these results are clinically relevant (17). The higher two concentrations (580 and 1,000 μU/ml) would correspond to subcutaneous doses of insulin lispro of ∼75 and 130 units, respectively, which are less likely to be utilized clinically.

Our study demonstrated a small placental transfer of insulin lispro at higher concentrations, after at least 60 min of constant concentration in the maternal reservoir. This experimental condition of maintaining high levels in the maternal circulation for several hours does not occur in the clinical setting due to the short elimination half-life of insulin lispro. Hence, even the transfer of insulin lispro shown by us in excessive maternal levels with constant concentration on maternal side is not likely to occur in vivo.

Lack of transfer of insulin lispro across the placenta was suggested in a study of insulin lispro in women with gestational diabetes (18). In that study, four women received an intravenous infusion of insulin lispro at a dose of 0.2 units · kg⁻¹ · h⁻¹ during labor, and no insulin lispro was detected in the umbilical cord blood of their infants. In our study, maternal insulin lispro concentrations of ≥580 μU/ml led to a small transfer of insulin lispro across the placenta. To put the results of our study in context, one should compare the placental transfer of insulin lispro to that of human insulin. In a single study by Challier et al. (10), human insulin was shown to appear in the fetal circulation after its introduction in the maternal reservoir. The measured levels by Challier et al., using the same perfusion model that we used, appear to be higher than those measured by us for insulin lispro. In their study, Challier et al. did not address the possibility that the low levels of fetal insulin were due to contamination by the residual blood in the perfused cotyledon. To address this question, we have now measured human insulin levels at the beginning of our perfusion experiment (before insulin lispro was introduced) and thereafter. Insulin levels, measured by a sensitive radioimmunoassay, were below the detection limits. This lends further credibility to the results by Challier et al. showing that human insulin does cross the placenta. Hence, it appears that small amounts of insulin cross the placenta when both human insulin and insulin lispro are present at clinically high concentrations.

Although the use of third-trimester placentae might be considered a potential limitation to our study, there is evidence to suggest that the qualitative aspects of transfer do not differ between placentae taken at earlier versus later phases of gestation (19).

The placentae of diabetic mothers may exhibit chronic disturbance in intervillous circulation, dilatation of capillaries, and relatively immature villous structure. The impact of these changes on the placental transfer of regular insulin or insulin lispro is unknown.

In summary, when used clinically, insulin lispro is not likely to cross the human placenta. There is a small dose-dependent transfer of insulin lispro across the term human placenta at concentrations ≥580 μU/ml. These concentrations likely correspond to peak insulin levels achieved with high doses of insulin lispro at ∼75 units or higher. Only in very rare situations would diabetic women receive doses that would lead to sustained plasma levels sufficiently high (580 μU/ml) for fetal exposure.

The marginal magnitude of the transfer of insulin lispro in terms of amounts, appears to be less than that shown by others to occur with similar concentrations of human insulin, and hence suggests that insulin lispro does not pose a risk for the fetus.

This study was supported by grants from the Canadian Institute of Health Research (CIHR), by Eli Lilly Canada, and by the Research Leadership for Pharmacotherapy During Pregnancy and Lactation. G.K. is a senior scientist of CIHR. R.B. is supported by a Fellowship from the Canadian Society for Clinical Pharmacology.

We thank Loren Grossman, MD, for his valuable comments.