Evaluation of Methods To Increase the Sensitivity and Timeliness of Detection of Streptococcus agalactiae in Pregnant Women

Abstract

Direct culture of rectovaginal specimens on Granada agar was compared to culture on sheep blood agar plate (SBAP) and AccuProbe detection of group B streptococcus from overnight LIM broth enhancement cultures (LIM-SBAP). Both broth-enhanced methods demonstrated excellent sensitivity (97.5% for LIM-SBAP and 93.5% for AccuProbe), while Granada agar demonstrated a sensitivity of only 40.3%.

Streptococcus agalactiae (group B streptococcus) is a significant cause of perinatal and neonatal infections in the United States. Colonization occurs in 10 to 30% of pregnant women (1, 7, 10) and is responsible for 1.8 neonatal infections per 1,000 live births per year (8). Neonatal colonization with group B streptococcus occurs in about 11% of babies (2). Screening of pregnant women for group B streptococcus is aimed at identifying those pregnancies at highest risk for developing maternal and neonatal diseases. The Centers for Disease Control and Prevention presently recommends culturing samples from pregnant women at 35 to 37 weeks gestation for S. agalactiae from vaginal and anorectal swabs after 18 to 24 h of growth in a selective enrichment medium, such as LIM or SBM (8). This process requires a minimum of 48 h to complete. A more rapid and sensitive method would be beneficial, especially in dealing with patients who present for delivery without any prenatal care. Although rapid detection techniques are available, none of these methods have the sensitivity needed to identify colonized women (5). Similarly, direct culture onto solid medium, such as sheep blood agar plates (SBAP), underestimates the actual incidence of rectovaginal colonization (3, 4, 15). Recently, new techniques have become available that may increase the sensitivity of group B streptococcus detection or decrease the turnaround time until detection. Granada agar is a differential medium that was developed to identify colonies of group B streptococci by their orange pigment production (9) while the AccuProbe S. agalactiae rRNA probe assay (Gen-Probe, San Diego, Calif.) was developed to detect bacterial rRNA. This study investigated the use of both Granada agar and the AccuProbe assay as potential methods for increasing the detection of S. agalactiae in pregnant women and in decreasing the turnaround time for reporting the results.

Four hundred sixty-three serially collected anovaginal swab specimens were tested in parallel by three different group B streptococcus detection methods. Swabs were inoculated directly onto Granada agar (Hardy Diagnostics, Santa Maria, Calif.) and were then placed into LIM broth (Remel, Lenexa, Kans.). Granada agar was incubated anaerobically at 35°C and was examined after 24 and 48 h of incubation. S. agalactiae colonies were detected by their production of an orange pigment. Such colonies were confirmed as S. agalactiae by using the Streptex latex agglutination reagent (Murex Biotech Ltd., Dartford, England). LIM broth was incubated aerobically at 35°C for 18 to 24 h. The broth was then subcultured onto an SBAP (Remel) for detection of beta-hemolytic colonies after 24 h of additional growth on the solid medium. Beta-hemolytic colonies were again confirmed as S. agalactiae by using the Streptex latex agglutination reagent. Immediately after inoculating the SBAP, the LIM broth was thoroughly mixed by using a vortex-type mixer and a 50-μl aliquot that was tested by using the AccuProbe S. agalactiae rRNA probe assay (Gen-Probe). The diagnostic sensitivity, specificity, and negative predictive values were calculated by applying Bayes' theorem (19). Test efficiency was calculated as the total number of positive and negative specimens correctly identified by each method divided by the total number of specimens tested.

Of the 463 specimens, 124 (27%) were found to be positive for S. agalactiae by one or more detection methods, while 339 (73%) were determined to be negative for group B streptococcus. Table 1 summarizes the total number of true- and false-positive specimens, the number of unique positive results, and the number of true- and false-negative results detected by each assay. The sensitivity, specificity, negative predictive value, positive predictive values, and overall testing efficiency of each test are summarized in Table 2. Test specificity was 100% for the LIM-SBAP culture method because all positive results were confirmed to be group B streptococci serologically. Granada agar was 99.7% specific, with a single specimen that produced pigmented colonies that were later identified as Enterococcus sp. For the probe, it was expected that the probe should detect more positives than LIM-SBAP. Our previous experience with group B streptococcus probe detection demonstrated unique positives beyond those detected by culture alone. On reculture of the LIM broth at 48 h, group B streptococcus was detected in the majority of cases (unpublished observations). Therefore, the one unique probe-positive result was considered a true positive, although this was not confirmed by reculture in this study. Both the 48-h LIM-SBAP culture and AccuProbe assay from the overnight LIM broth growth demonstrated excellent sensitivity and overall testing efficiency (97.3 and 99.4% for the LIM-SBAP and 93.5 and 98.2% for the AccuProbe assay, respectively). These findings confirm those of Kircher et al. (12), Bourbeau et al. (6), and Williams-Bouyer et al. (18), who also found AccuProbe to be about 95% sensitive compared to the standard LIM-SBAP culture detection for group B streptococcus.

TABLE 1.

Characterization of testing results by diagnostic procedures

S. agalactiae detection method	True positive	Unique positive	False positive	True negative	False negative
LIM-SBAP	121	2	0	339	3
AccuProbe	116	1	0	339	8
Granada agar	50	2	1	338	74
Total	124	5	NAa	339	NA

^aNA, not applicable.

TABLE 2.

Performance characteristics for LIM-SBAP broth-enhanced culture, AccuProbe, and Granada agar direct culture in detecting S. agalactiae^a

Method	% Sensitivity	% Specificity	% Negative predictive value	% Test efficiency
LIM-SBAP	97.5	100	99.1	99.4
AccuProbe	93.5	100b	97.7	98.2
Granada agar	40.3	99.7	82.0	83.8

^aTP, true positive; FN, false negative; TN, true negative; FP, false positive. Values were calculated as follows: sensitivity, TP/(TP + FN); specificity, TN/(TN + FP); negative predictive value, TN(TN + FN); test efficiency, (TP + TN)/total no. tested.

^bThe one unique positive detected by probe was assumed to be a true positive.

It is noteworthy that no assay detected all positive specimens. Even the Granada agar detected two specimens containing group B streptococcus not discovered by either of the broth-enhanced techniques. There are several explanations for the discrepancies among the various group B streptococcal detection methods. Low starting colony counts in the direct specimens probably accounts for some of the discrepancies leading to false-negative results, particularly when the direct detection method (Granada agar) and the LIM-enhanced methods (LIM-SBAP and AccuProbe) are compared. Possibly also underlying the high false-negative rate for the Granada agar is the issue of the medium's stability. This medium has a short shelf life and is extremely sensitive to changes in its storage conditions (11). During this study, two lots of medium were received from the manufacturer. The first lot of medium demonstrated detection of 68.8% of the positive specimens compared to that of the other detection methods. Despite adequate performance of the medium with control bacterial strains, the second lot of medium detected only 18.9% of the group B streptococcus-positive specimens detected by the other methods. Both lots of plates were stored at 4°C and were evaluated under identical conditions per the manufacturer's recommendations. The medium is also cumbersome to use from the standpoint of requiring anaerobic culture conditions. Approximately 3 to 4% of group B streptococcus isolates fail to produce the caratinoid pigment even under optimal growth conditions (9, 13, 14). The occasional failures seen in the LIM broth enhanced detection methods may be due to the inhibition of the group B streptococcal growth in the LIM broth when Enterococcus sp. have overgrown the medium (17), leading to false negatives with both LIM broth-dependent protocols. Additionally, about 2% of all group B streptococci fail to produce hemolysis on the SBAP used for the screening protocols (1). Although it is possible that the one probe-positive LIM-SBAP-negative specimen was a nonhemolytic strain of group B streptococci, we could not rule out low colony counts in the overnight culture as the reason for the LIM-SBAP failure. The specimen or isolate was no longer available for analysis to resolve this discrepancy. The two unique positive specimens found on the Granada agar were both found to be beta-hemolytic on SBAP on subsequent subculture. In comparison, Granada agar only detected 40.2% of all positive specimens. At least three specimens produced nonpigmented colonies on the Granada agar that were later confirmed to be group B streptococcus by latex agglutination during the analysis of discordant testing results. All of these isolates were hemolytic on SBAP. This is consistent with the observation that about 3% of group B streptococcal isolates lack pigment production (13). In addition to these nonpigmented isolates, the Granada agar failed to detect group B streptococcus in many other specimens. In the hands of other researchers, Granada agar was as sensitive (17) or was more sensitive (11) than selective broth SBAP culture. In marked contrast to these findings, the present study demonstrated a sensitivity of detection of only 40.2%, which is the rate of detection seen for direct culture on SBAP (5, 16). This places the Granada agar into the same low-sensitivity category seen with the rapid immunoassays and direct culture without broth enhancement, which are generally considered inadequate for predicting group B streptococcus colonization (5). The finding that Enterococcus sp. may produce pigmented colonies on this agar further reduces its reliability.

The cost for performing testing varied according to the method. The standard LIM broth culture with SBAP and serologic confirmation costs approximately $1.88 per positive and $0.99 per negative specimen in reagents alone. The Granada agar method cost $1.38 per plate, with additional costs for anaerobic storage and serologic confirmation for a total of $3.38 for each positive and $1.50 for each negative specimen. The highest cost was seen for the AccuProbe methodology, with the cost of the LIM broth and probe reagents coming to $19.71 per positive or negative specimen. These costs reflect only reagents and expendables and do not include the costs of required quality control or technologist time. Both of the culture-based assays require setup and 3 days of manipulation for the individual specimens. The probe assay requires only 2 days of specimen manipulation, with the testing occurring in batches and not with individual specimens.

In conclusion, this study demonstrated that the standard LIM-SBAP culture method is highly sensitive, as is the use of a bacterial rRNA probe on the overnight LIM broth growth from anovaginal samples. The probe offers the advantage of saving 24 h in turnaround time in detecting S. agalactiae. Granada agar was insensitive and cumbersome to use. Some positive cultures failed to produce the orange-pigmented colonies that are regarded as characteristic of S. agalactiae on this selective medium. Overall, direct detection of S. agalactiae on Granada agar missed 60% of true-positive cultures and, therefore, is not an adequate detection system for this important pathogen.