Detection of group B streptococcal bacteremia in simulated intrapartum antimicrobial prophylaxis

1. Introduction 

1996 CDC guidelines for prevention of perinatal group B streptococcal (GBS) disease recommend either a screening-based or risk factor-based approach to identify pregnant women in labor who should receive intrapartum antimicrobial prophylaxis (IAP) with penicillin, ampicillin, clindamycin or erythromycin ACOG Committee on Obstetric Practice 1996, AAP Committee on Infectious Diseases and Committee on Fetus and Newborn 1997, Centers for Disease Control and Prevention 1996a, Centers for Disease Control and Prevention 1996b. After institution of these guidelines, an estimated minimum of 18% of deliveries were administered IAP (Centers for Disease Control and Prevention, 1996a). The effect of IAP on subsequent management of the newborn has remained largely unassessed. Despite data indicating a decreased incidence of early onset GBS disease in geographic areas where more hospitals have established GBS prevention policies (Centers for Disease Control and Prevention, 1998), and a published algorithm for management of neonates exposed to IAP (Centers for Disease Control and Prevention, 1996a), surveyed pediatricians have indicated they would paradoxically hospitalize at risk newborns longer for observation post-IAP, in part because the sensitivity of neonatal blood culture is questionable in the context of IAP Kenagy and Baker 1990, Peralta-Carcelen et al 1996.

Placental transfer of antibiotics varies and is dependent on factors such as stage of pregnancy, drug molecular weight, and transplacental mechanism of transport (Pacifici et al., 1995). Cord to maternal ratios for ampicillin serum concentrations approach unity within 2 h of IV administration. Fetal serum concentrations following administration of the currently recommended 2 g IV ampicillin at least 4 h prior to delivery presumably exceed the known minimum of 5 mcg/mL that occurs following IV administration of 500 mg ampicillin to women in labor Bray et al 1966, MacAulay et al 1966, Pacifici and Nottoli 1995, Yow et al 1979. Gentamicin placental transfer occurs incompletely. Cord to maternal serum ratio is approximately 0.15, with a peak of 0.8 mcg/mL within 2 h after administration of 80 mg IM to the mother Pacifici and Nottoli 1995, Weinstein et al 1976, Yoshioka et al 1972.

The BACTEC Peds Plus media was designed to overcome factors such as prior antibiotic exposure that might inhibit isolation of pediatric pathogens. Modifications included decreased broth volume (to optimize blood to broth ratio for smaller volume draws), increased CO2 content in the headspace of the vial (to improve recovery of capnophilic organisms like Streptococcus pneumoniae), and the addition of non-ionic adsorbing resins and cationic exchange resins (to adsorb antibiotic when present) (Morello et al., 1991). In previous clinical studies of Peds Plus media, recovery of staphylococci and yeasts was demonstrably better with Peds Plus than with Du Pont Isolator 1.5, bioMerieux VITAL, BACTEC NR-6A, and approximately equivalent to Roche Septi-Chek Eisenach et al 1992, Morello et al 1991, Welby et al 1992, Zaidi et al 1997. This difference was even more pronounced when antibiotic was present. The sensitivity of resin-containing media has not specifically been evaluated in the setting of neonates born to mothers given IAP. We hypothesized that resin-containing media would improve detection of neonatal GBS bacteremia post-IAP.

We designed an in vitro model to simulate conditions of IAP. Our in vitro model simulated IAP by preincubating low and high inocula of S. agalactiae (to imitate low and high inoculum states of sepsis) with ampicillin and gentamicin at concentrations consistent with known pharmacokinetic data regarding their transplacental transfer. Commercially available, standardized defibrinated horse blood was used instead of human blood. The small volume of blood (0.5 mL) inoculated into each culture bottle was based upon the likely sample size that would be obtained from a neonate.

2. Materials and methods 

<500 or >500 CFU/mL of Streptococcus agalactiae (American Type Culture Collection, Manassas, VA, strain 13813; or a clinical strain from an infant with GBS meningitis) were grown for 18 h in supplemented brain heart infusion (500 mL brain heart infusion supplemented with 5 mg NADH and 1.5 mL filtered defibrinated horse blood). Final concentration of bacteria was determined in duplicate by counting and averaging the number of viable organisms on subculture to blood agar plates. Bacteria were then mixed with defibrinated horse blood (Remel, Lenexa, KS), 0, 1, 5, 10 or 50 mcg/mL ampicillin (IM/IV preparation, Apothecon, Princeton, NJ), and 0 or 2 mcg/mL gentamicin sulfate (IM/IV preparation, Elkins-Sinn, Cherry Hill, NJ) in a volume ratio of 23:1:1 blood:S. agalactiae:antibiotic. Specific concentrations of antibiotic tested against both low and high inocula S. agalactiae included all concentrations of ampicillin alone, and the 10 mcg/mL ampicillin together with 2 mcg/mL gentamicin. Blood:S. agalactiae:antibiotic mixture was agitated for 4 h at 37°C. One-half mL of each blood:S. agalactiae:antibiotic mixture was inoculated into twelve bottles each of BACTEC Peds Plus/F (Becton-Dickinson, Sparks, MD), BACTEC Standard 10 Aerobic/F (Becton-Dickinson, Sparks, MD) or Trek ESP 80A (Trek Diagnostics, Westlake, OH) aerobic media. Bottles were incubated at 35°C in their respective continuous detection systems (fluorescence-sensitive BACTEC 9050 or pressure-sensitive ESP blood culture system), and subcultured either when flagged positive or after 5 days as a terminal subculture. S. agalactiae was identified by gram stain, blood agar plate colonial morphology, and CAMP testing. Detection of S. agalactiae and time to that detection were recorded for 12 bottles of each medium per blood:S. agalactiae:antibiotic mixture. Statistical significance was evaluated using Fisher’s exact test or Wilcoxon rank-sum test where appropriate.

The minimum inhibitory concentration (MIC) of ampicillin for a fixed concentration of S. agalactiae ATCC strain 13813 and for the clinical strain were measured using serial tube dilutions of antibiotic in media used in the experiment. 5 × 105 CFUs of S. agalactiae were suspended in 1 mL of supplemented BHI with ampicillin from 8 mcg/mL serially diluted down to approximately 0.02 mcg/mL. The MIC was the lowest antibiotic dilution for which no growth of S. agalactiae was visualized after an overnight incubation at 37°C.

3. Results 

The ampicillin MIC for S. agalactiae ATCC strain 13813 was 0.03 mcg/mL. The ampicillin MIC for S. agalactiae clinical strain was 0.06 mcg/mL.

Colony counts demonstrated that final concentrations of bacteria ranged between 76-350 CFU/mL for the low inocula group, and between 870-3500 CFU/mL for the high inocula group. With both high and low inocula, the Peds Plus media detected S. agalactiae more consistently than the other two aerobic media (Fig. 1, Fig. 2). This difference was more pronounced in the presence of ampicillin that was >40× the MIC. However, the sensitivity of Peds Plus for detection of S. agalactiae decreased as higher ampicillin concentrations were used. When S. agalactiae was detected in the presence of 10 mcg/mL ampicillin, detection was more rapid by Peds Plus than by ESP 80A (Table 1). In another series of experiments, when S. agalactiae detection in the presence of 10 mcg/mL ampicillin and 2 mcg/mL gentamicin was compared with 10 mcg/mL ampicillin alone, no gentamicin effect was observed. However, the difference in detection between Peds Plus, ESP 80A and Standard media persisted (Table 2).

View Large Image Download to PowerPoint

Fig. 1. Sensitivity of blood culture media for detection of low inoculum (<500 CFU/mL) GBS bacteremia.

View Large Image Download to PowerPoint

Fig. 2. Sensitivity of blood culture media for detection of high inoculum (>500 CFU/mL) GBS bacteremia.

Table 1. Median time to GBS detectiona in 10 mcg/mL ampicillin—comparison of Peds Plus and ESP 80Ab

	GBS inoculum (CFU/ml)	Peds Plus	ESP 80A	P valuec
	<500	19.0 hrs	43.0 hrs	0.0032
	>500	16.0 hrs	42.5 hrs	0.0007

a Median time to detection calculated based upon 10 bottles of Peds Plus and 5 bottles of ESP 80A which detected low inoculum GBS; and 12 bottles of Peds Plus and 6 bottles of ESP 80A which detected high inoculum GBS. b BACTEC Standard medium did not enable detection of GBS growth in 10 mcg/ml ampicillin and is therefore not included in this table. c Wilcoxon rank-sum test.

Table 2. Lack of gentamicin effect on GBS detection at 10 mcg/ml ampicillin

	GBS inoculum (CFU/ml)	Gentamicin (mcg/ml)	Peds Plus GBS Recovery (%)a	Standard GBS Recovery (%)a	ESP 80A GBS Recovery (%)a
	<500	0	75	0	17
		2	75	0	0
	>500	0	100	0	83
		2	100	0	75

a Percent GBS recovered (n = 12 bottles).

All percentages of S. agalactiae detection were calculated based upon twelve bottles of each media type tested per blood:S. agalactiae ATCC strain 13813:antibiotic mixture. Similar results were obtained in a smaller series of experiments using the S. agalactiae clinical strain to compare Peds Plus versus Standard aerobic media (results not shown).

4. Discussion 

GBS recovery rates and time to detection were compared among three commercially available aerobic blood culture media. Peds Plus was significantly more sensitive to the presence of S. agalactiae following preincubation with concentrations of ampicillin likely to be achieved in serum of neonates whose mothers received IAP with standard doses of ampicillin. When 0.5 mL of blood containing known concentrations of S. agalactiae and antibiotic was inoculated into culture media, antibiotic was further diluted by broth volume (80-fold by either Peds Plus or Standard, 160-fold by ESP 80A), perhaps explaining why 10 mcg/mL ampicillin was insufficient to completely inhibit growth in ESP 80A which had the largest broth volume of 80 mL (the 160-fold dilution would have reduced the antibiotic to the MIC of 0.06 mcg/mL). Four hours of incubation with higher concentrations of ampicillin was able to sufficiently inhibit the growth of S. agalactiae so that the resin-containing medium inconsistently detected the organism’s presence. However, the sensitivity of Peds Plus continued to exceed that of either non-resin-containing medium at higher ampicillin concentrations.

Previous clinical studies on the Peds Plus resin-containing medium have shown significantly earlier and increased recovery of staphylococcal species and yeast Morello et al 1991, Zaidi et al 1997. Significantly increased detection of other pathogens has not been shown (although Morello et al. showed a trend toward increased recovery of several bacteria including S. agalactiae), in part because of relatively small numbers of other pathogens found. Our in vitro study supports this idea that resin-containing devices may both shorten time to detection and increase detection overall for certain pathogens compared to standard media, and extends the idea to S. agalactiae specifically. An explanation for the poor performance of BACTEC Standard medium might lie in the small amount of blood (0.5 mL) used in the experiment, since the optimal inoculation blood volume as recommended by BACTEC is 8–10 mL with a 3 mL minimum. The observation that S. agalactiae in the presence of antibiotic took longer to be detected in ESP 80A compared to Peds Plus might relate to the dilutional issue (it might take longer for sufficient quantities of S. agalactiae to overcome inhibition of growth by still-present ampicillin) or to the issue that gas pressure change might take longer to manifest than fluorescent detection of CO2. However, the idea that pressure detection methodology is inferior to fluorometric detection of CO2 is not supported by the superior performance of ESP 80A compared to BACTEC Standard medium in the same series of experiments.

An explanation for the failure of gentamicin to alter the results compared to ampicillin alone lies in S. agalactiae-killing kinetics. Although addition of gentamicin improves the rate of killing, the minimum time period necessary for bactericidal activity is 8 h Cooper et al 1979, Schauf et al 1976, Swingle et al 1985. Therefore the short incubation (4 h) fails to permit sufficient gentamicin exposure to kill S. agalactiae, as demonstrated by detection of GBS growth once the mixture is added to resin-containing media (Peds Plus) or further diluted (ESP 80A).

The idea that continuously monitoring blood culture systems are able to support the growth of and detect a single viable bacterium was not supported by these data. Although only ampicillin and gentamicin were tested directly in this model, presumably these data are generalizable to IV penicillin, which has similar pharmacokinetics to ampicillin. Whether they are generalizable to clindamycin or erythromycin (the recommended second-line agents for IAP in penicillin-allergic mothers) is speculative. Although transplacental transfer of clindamycin does occur, peak umbilical cord concentrations in term infants have only been about half of maternal serum concentrations Philipson et al 1973, Weinstein et al 1976. Erythromycin gives variable levels with transplacental transfer, perhaps because most studies have examined fetal levels following oral administration of erythromycin, which is known to have unreliable oral absorption Kiefer et al 1955, Philipson et al 1973.

Limitations to this study include the fact that the model, although designed to imitate IAP given to a mother, did not permit simulation of fetal elimination of drug after single dose IV administration to the mother. The defibrinated horse blood used in our experiments might also have influenced detection rates of S. agalactiae. A note of caution should also be sounded in regards to using a single aerobic blood culture medium for detection of neonatal bacteremia. Although obligate anaerobic organisms seldom play a role in neonatal or pediatric bacteremia Dunne et al 1994, Zaidi et al 1995, the detection of even such a rare event would not be possible if only the Peds Plus bottle were used for blood culture. No comparison of anaerobic blood culture media was made in this study.

5. Conclusions 

In this in vitro simulation of GBS bacteremia in the neonate whose mother receives IAP, four-hour exposure to ampicillin or ampicillin and gentamicin fails to sterilize blood. Our model suggests that pediatric resin-containing media shortens time to detection of S. agalactiae, is consistently more sensitive than non-resin-containing media, and remains able to detect S. agalactiae despite addition of gentamicin to the IAP regimen. Limitations to our model include the use of defibrinated horse blood instead of human peripheral blood, and model design that did not permit simulation of fetal elimination of antibiotic during incubation. Resin-containing media may prove to be superior to non-resin-containing media in the detection of true S. agalactiae bacteremia in the clinical setting of neonates born to mothers who have received IAP for prevention of early onset GBS disease. However, additional clinical studies on neonates exposed to IAP are needed to determine whether this increased detection of GBS bacteremia is clinically relevant. (Spaargaren et al 1998)