Group B streptococcus (GBS) remains a significant cause of neonatal sepsis in the United States, even after the initiation of prophylaxis measures in the 1980s.1 The Centers for Disease Control and Prevention (CDC) released revised guidelines for the prevention of perinatal group B streptococcal disease in 2002 with prophylaxis recommendations for penicillin-allergic women.2 For women not at significant risk of anaphylaxis, cefazolin is an acceptable alternate agent. For those at high risk of anaphylaxis with clindamycin-sensitive GBS, clindamycin is acceptable chemoprophylaxis. Women with clindamycin-resistant GBS and a history of anaphylaxis, urticaria, or angioedema associated with administration of penicillins or with a penicillin allergy and a history of asthma currently are recommended to receive 1 g of vancomycin intravenously every 12 hours during active labor. However, there is very little published regarding the pharmacokinetics of vancomycin across the placenta and into the fetal compartment.
In the setting of infection, Bouget et al1 and Reyes et al2 observed transplacental passage of vancomycin in women with amnionitis when administered over several days. More recently, an ex vivo model demonstrated minimal vancomycin transplacental passage, calling into question the efficacy of vancomycin for chemoprophylaxis in the setting of severe penicillin allergy.3 The objectives of the current study were to characterize the transplacental passage of vancomycin into amniotic fluid and cord blood in term, clinically uninfected pregnant women.
MATERIALS AND METHODS
Term, clinically uninfected pregnant women undergoing scheduled cesarean delivery between May 2004 and March 2005 were candidates for the study. Subjects were eligible for inclusion if they were at least 18 years of age, had an uncomplicated term pregnancy (38–42 weeks of gestation), and were undergoing scheduled cesarean delivery. The Institutional Review Board (IRB) of the University of Michigan Hospital and Health Centers approved the study protocol, and informed consent was obtained from all enrolled patients. Specific information on the expected frequency and descriptions of vancomycin reactions, including red man syndrome, were provided to each potential study subject before their inclusion. Exclusion criteria included the need for antimicrobial therapy of any type (including any signs or symptoms of chorioamnionitis), history of allergy or red man syndrome to vancomycin, lack of or inability to give informed consent, or history of impaired hepatic or renal function.
Four women were enrolled in a pilot portion of the study to evaluate amniotic fluid samples without prior administration of vancomycin to determine whether the fluorescence polarization vancomycin assay technique could be used in this fluid. At the time of cesarean delivery, amniotic fluid samples were collected after uterine incision, but before amniotomy, by aspiration with a sterile needle and syringe. Care was taken to avoid gross contamination by maternal blood.
Thirteen women undergoing scheduled cesarean delivery were enrolled into the active study group. These subjects were to be allocated to receive 1 g of vancomycin intravenously, 1, 4, or 6 hours before scheduled delivery. Subjects were assigned to one of these groups in a nonrandomized, nonconsecutive manner. Women were selected and placed into the different time groups based on their availability to participate in the time frames of the study.
Because of the limited sampling technique (only one maternal and cord blood sample per study subject) used in this trial, a rigorous pharmacokinetic modeling analysis that would yield elimination and volume of distribution information could not be performed. Instead, relationships between patient weight, vancomycin administration, and maternal blood, cord blood, and amniotic solution were assessed. Data were analyzed with Fisher exact test or t test. P<.05 was considered statistically significant.
One gram of vancomycin was to be infused over 1 hour for each patient according to her projected time of delivery. Vancomycin infusion rates were regulated by a medication infusion pump. Only 6 of 13 women received the full 1 g of vancomycin because 53.8% (7 of 13) had some manifestation of red man syndrome during the infusion. An IRB-approved protocol for management of occurrences of red man syndrome was placed at the bedside for care providers' immediate access, along with doses of diphenhydramine and acetaminophen. Infusions were to be stopped for any signs or symptoms of red man syndrome or other adverse effects due to vancomycin, including pruritus, flushing, rash, hypotension, oxygen desaturation, or shortness of breath. If an infusion was stopped, the amount of delivered vancomycin was determined. Serial maternal blood pressure assessment, continuous fetal monitoring, intravenous fluid bolus, and a combination of acetaminophen and diphenhydramine were initiated with the cessation of infusion if red man syndrome developed. Additional treatments (eg, oxygen therapy, ephedrine) were available and used if needed. Because three of the first four women enrolled experienced red man syndrome, the IRB-approved a protocol change to extend the vancomycin infusion time to 90 minutes.
Amniotic fluid samples were collected by aspiration with a sterile needle and syringe at the time of cesarean delivery, just after uterine incision but before amniotomy. Care was taken to avoid contamination by maternal blood. Umbilical cord blood was obtained from segments of cord that were collected just after delivery of the neonate. Maternal blood was collected at the time of delivery in all cases. Actual time from completion of infusion to time of delivery and times to sample collection were noted. All blood and amniotic fluid samples were immediately refrigerated. Samples were then centrifuged and supernatants were transferred into plastic centrifuge storage tubes and frozen at –70ºC until analyzed.
Vancomycin assays were performed in duplicate on all serum and amniotic fluid samples. Reported results reflect the mean of the two measured values. Assays were performed on a Cobas Integra 400 Plus (Roche, Indianapolis, IN), which used fluorescence polarization immunoassay technology. The intraday coefficient of variation of this assay was less than 6% for maternal and cord blood.
Samples from the preliminary (no drug administered) portion of the study had serial vancomycin added to known volumes of amniotic fluid to achieve concentrations varying from 0 to 40 mcg/mL and were analyzed for validation of the fluorescence polarization immunoassay technique in amniotic fluid.
Thirteen women were enrolled and given vancomycin in this study. The women ranged in age from 21 to 41 years, with a mean age of 29.9±5.2 years. Ten of the study subjects were white women, and three were African American. The subjects' weights ranged from 65 to 163 kg (mean weight 93±24.2 kg). Four of the 13 cesarean deliveries were performed for breech presentation; the remaining nine deliveries were scheduled repeat cesarean deliveries. All but three of the women were multiparous.
Vancomycin concentrations in maternal serum ranged from 2.6 to 19.8 mcg/mL (Table 1). In cord blood samples, vancomycin concentrations ranged from 2.8 to 9.4 mcg/mL and persisted above the GBS vancomycin breakpoint of 1 mcg/mL.
The actual time from end of vancomycin infusion to maternal vancomycin sampling varied from 26 minutes to 509 minutes (Table 1). Consequently, instead of four discrete sampling groups, there was a range of time represented in this pharmacokinetic sampling. Changes in time intervals from scheduled times were due to circumstances including obstetric emergencies on the unit, early discontinuation of drug infusion, difficulty establishing regional anesthesia, and increased time to delivery in repeat cesarean delivery.
A strong correlation (r 2=0.93, P<.001, unpaired two-tailed t test) was observed between the ratio of cord blood to maternal blood vancomycin concentrations and time from the end of the vancomycin infusion (Fig. 1). Cord blood concentrations approached maternal blood concentrations 4 hours after the vancomycin infusion ended. Normalizing the ratio of cord vancomycin concentration to maternal vancomycin concentration to a maternal dose (milligram per kilogram basis) did not improve this model fit. Duration of vancomycin infusion (60 or 90 minutes) had a negligible effect on model fit.
As stated earlier, three of the first four women receiving vancomycin with a 60 minute infusion experienced red man syndrome, which led the investigators to extend the infusion time to 90 minutes. Although all subsequent subjects were to receive a 90-minute infusion, one subject received her infusion over 60 minutes and, incidentally, experienced red man syndrome. Three of the eight women receiving 90-minute vancomycin infusions also experienced red man syndrome. In total, red man syndrome was experienced in four of the five subjects in the 60-minute infusion group and in three of the eight subjects in the 90-minute infusion group (P=.26, Fisher exact test). The mean infusion rates (milligram per kilogram per minute) for the subjects who developed red man syndrome compared with those who did not were not significantly different (0.18±0.05 versus 0.13±0.06 mg · kg-1 · min-1, respectively; P=.15, unpaired two-tailed t test).
Table 2 outlines the clinical course of red man syndrome in these subjects. One subject developed moderately severe symptoms, which included hypotension, shortness of breath, and oxygen desaturation requiring treatment with ephedrine. No short-term sequelae (ie, at the time of discharge) or increased length of stay were required for any study subjects or their neonates. Because vancomycin infusions were stopped when red man syndrome was observed, the amount of delivered vancomycin is reported in Table 1.
Amniotic fluid was obtained from four women who did not receive vancomycin in the pilot study. Known amounts of vancomycin were added to these amniotic fluid samples, which were then analyzed by florescence polarization immunoassay. This commercially available vancomycin assay was developed for use in human serum, but we attempted to use it on amniotic fluid. The fluorescence polarization immunoassay reported that vancomycin was present in concentrations above the minimum detectable limit of the assay, even in amniotic fluid control samples that were not spiked with vancomycin. Vancomycin-spiked amniotic fluid samples did have concentrations similar to the amount that was spiked, but repeated tests revealed similar results in the vancomycin-free samples. We also attempted to assay vancomycin in amniotic fluid via high-performance liquid chromatography technique but had similarly poor results. Consequently, despite the fact that amniotic fluid samples were collected from the study subjects, our inability to get the assays to perform well in this matrix led us not to assay the amniotic fluid samples.
This study demonstrates that vancomycin readily crosses the noninfected term placenta in vivo. Figure 1 illustrates a significant correlation (r 2=0.93, P<.001) between time and the ratio of cord serum vancomycin concentrations to maternal concentrations. Because of the strong correlation between cord and maternal concentrations, it is possible that cord vancomycin concentrations might be estimated from a more easily obtained maternal vancomycin value. It is important to recognize that this strong correlation between the ratio of cord and maternal blood vancomycin concentrations was noted only after a single maternal vancomycin dose was administered. The impact of repeated maternal vancomycin administration on the proportion of vancomycin that crosses the placenta at any given time cannot be predicted with the data in this study.
The results of our study directly contradict a recent report that was unable to demonstrate significant vancomycin transplacental passage in an ex vivo model.3 One possible explanation for this discrepancy is that the heparin used in the ex vivo model may have formed a heparin-vancomycin complex that could not be transported across the placenta, as suggested by the authors.3
The in vitro minimal inhibitory concentration (MIC) breakpoint for vancomycin against GBS is 1 mcg/mL or less.4 Group B streptococcus strains with an MIC greater than 1 have not been reported. Data from our small study demonstrated that vancomycin concentrations exceeded the breakpoint for GBS within 30 minutes of completion of the vancomycin infusion and remained above the breakpoint through 500 minutes. In addition, vancomycin concentrations above the GBS breakpoint were observed regardless of dose received (range 420–1,000 mg, 5.4–14.7 mg/kg per dose). Although the GBS vancomycin breakpoints are known, the optimal or desired serum concentration for the cord blood to prevent GBS disease in the newborn (eg, sepsis, pneumonia, or meningitis) has never been studied.
Our patient population experienced a 53.8% rate of red man syndrome, a rate that was higher than anticipated but consistent with published reports of the frequency of red man syndrome.5–8 The incidence of red man syndrome in the literature ranges from 1–2% up to 805,6 but has not been studied previously in pregnancy. Red man syndrome is a drug reaction caused by vancomycin-induced degranulation of mast cells and basophils resulting in the release of histamine.9 This occurs independently of immunoglobulin E formation and is, therefore, not an anaphylactic reaction. It is unclear if pregnant women are predisposed to red man syndrome. Because the degree of histamine release from a vancomycin infusion may be impacted by both dose and rate of infusion,6,10 we slowed the vancomycin infusion rate after many of our initial subjects experienced red man syndrome. Although the lengthening of the infusion rate did not change the red man syndrome rate to a statistically significant degree, our study was not adequately powered to address this issue. Others have addressed this issue in nonpregnant, otherwise healthy adults and have found that extending the infusion time of a vancomycin dose can reduce the incidence of red man syndrome.10 The relatively high rate of red man syndrome seen in this study may also be related to the fact that all study subjects were clinically noninfected. In 216 prior studies of healthy (ie, uninfected) volunteers, higher rates of red man syndrome were seen (30–90%). The presence of infection induces histamine release, and these higher extracellular histamine levels are believed to down-regulate the effect of vancomycin on basophils and mast cells.5
Our study subjects exhibited a wide range of weight (65–163 kg), but we did not dose vancomycin on a milligram per kilogram basis. Further, vancomycin infusions were discontinued as soon as red man syndrome was experienced. Consequently, the administered doses range from 420 to 1,000 mg and from 5.4 to 14.7 mg/kg on a weight-normalized basis. Despite this large dose range, the delivered vancomycin dose on a milligram per kilogram basis was not different between the red man syndrome and non–red man syndrome groups (P=.21, unpaired two-tailed t test). Indeed, no demographic or clinical parameter predicted the onset of red man syndrome.
Pharmacokinetic studies of this type are very difficult to perform, and there are limitations to our study. We performed very limited sampling (one maternal plasma vancomycin concentration and one cord concentration per subject). Although it would be preferable to be able to sample more frequently, it is unfeasible. Further, a randomized, double-blinded study with better defined and distributed blood sampling times would be preferable. Again, the uncertainties of the timing of infusions and the actual births made this impossible. Although we were unable to enroll patients in our study in the manner we originally intended (ie, four patients into each of the four groups), we do not feel this introduced bias into our study because patients were not enrolled in any certain group due to any specific demographic or pharmacokinetic factor. Although perhaps not ideal, this manner of enrollment was the best that could be performed in the dynamic setting of a surgical suite.
The results of this study leave a few unanswered questions. The main question is “What is the appropriate dose of vancomycin for a pregnant woman for GBS prophylaxis to maximize therapeutic outcome without causing red man syndrome?” Given the altered pharmacokinetic parameters exhibited by women during pregnancy, dosing vancomycin on a milligram per kilogram basis might be more appropriate than standardized dosing.11 It appears, however, that 1 g of vancomycin may be more than is necessary to achieve reasonable fetal serum levels (ie, more than 1 mcg/mL) and that a lower dose may achieve therapeutic levels and lessen the risk of red man syndrome, although it would require additional study to determine the effective dose (in milligrams per kilogram). Also unanswered by this trial is the appropriate vancomycin redosing interval when used as intrapartum prophylaxis for GBS. We demonstrated cord blood vancomycin concentrations above the breakpoint for GBS out to 8 hours with a single vancomycin infusion. We do not know, however, if this effect would persist out to 12 hours, which is when the CDC recommends redosing vancomycin. Nor can we tell what cord blood concentrations would be attained when vancomycin is administered repeatedly.
Given the relatively common occurrence of adverse reactions to vancomycin in this study population, consideration should be given not only to means of reducing red man syndrome (ie, decreasing overall dose, slower infusions, potentially concomitant antihistamines), but also to the study of alternate agents with activity against group B streptococcus, such as linezolid. Consideration could also be given to providing chemoprophylaxis to neonates born to women with clindamycin-resistant GBS and high risk of penicillin anaphylaxis instead of the pregnant woman herself. In reality, this would affect approximately 3,600 pregnant women annually in the United States (assuming a 15% rate of GBS colonization, 20% rate of clindamycin resistance, 15% rate of penicillin allergy overall with a 33% rate of anaphylaxis, and 4 million births per year).
Although our study was limited by a small sample size and limited pharmacokinetic sampling, we were able to demonstrate transplacental passage of vancomycin in concentrations above the MIC breakpoint for GBS in all of the study participants. In addition, we also demonstrated a highly significant correlation between the ratio of maternal serum concentrations and cord blood concentrations of vancomycin versus time after vancomycin infusion. The high rate of red man syndrome observed in this study suggests that vancomycin should be infused for not less than 90 minutes when used in this clinical setting.