Although recurrent group B streptococcal infection has been reported to occur in both infants [3-15] and adults [16-21], no population-based data are available to quantify the risk for recurrent infection. In addition, it is unknown whether recurrent disease is caused by infection with the same strain or reinfection with another strain. After we began an active surveillance project of invasive group B streptococcal infection in Maryland, we noted that some adult patients had more than one episode of group B streptococcal infection. We therefore prospectively studied the problem of recurrent infection and supplemented our study with molecular epidemiologic methods.

Methods

Top Methods Results Discussion Author & Article Info References

Surveillance

Active surveillance for invasive Haemophilus influenzae, meningococcal, Listeria monocytogenes, and group B streptococcal infection was initiated in November 1991 as part of the Maryland Bacterial Invasive Disease Surveillance (BIDS) project. This project is a component of the multistate National Bacterial Invasive Disease Surveillance Group, which is coordinated by the Centers for Disease Control and Prevention (CDC). The surveillance case definition is the isolation of one of the above organisms from a normally sterile body fluid, such as blood or cerebrospinal fluid, from a Maryland resident of any age with clinical signs of infection. All acute-care hospitals in Maryland participate in this project, as do hospitals in Washington, D.C., at which residents of southern Maryland frequently seek medical care. Nonhospital microbiology laboratories that process blood cultures also participate. A brief case report form is completed for each eligible case, and the bacterial isolates are submitted for species confirmation and further testing. Biweekly telephone calls are made to hospital infection control practitioners to ascertain cases not reported spontaneously. To identify unreported cases, annual on-site audits of microbiology laboratories are done by reviewing the laboratory records.

Identification of Patients with Recurrent Group B Streptococcal Infection

At the initiation of the surveillance project in November 1991, a computer program was developed to identify all patients with group B streptococcal infection who, on the basis of their last name and date of birth, had duplicate case report forms in the surveillance database. After noting that six adult patients had more than one admission for invasive group B streptococcal infection, we decided to prospectively study recurrent disease in all age groups, including infants. Recurrent infection was defined as two separate hospital admissions during which group B streptococcus was isolated from a normally sterile body fluid. We also included nonhospitalized patients who had an acute illness during which group B streptococcus was isolated from a normally sterile body fluid and who had a second episode.

Children and adults of any age with a first episode of group B streptococcal infection after 1 November 1991 were included. To calculate the proportion of all patients with group B streptococcal infection who had more than one episode, we included only patients whose first episodes occurred by 30 September 1993; this criterion allowed a minimum follow-up of 1 year. To identify patients not detected by last name and date of birth, we also searched for duplicate street addresses or medical record numbers and matches based on the first and last name or date of birth and ZIP code. Charts were reviewed to verify that each case report form represented a distinct episode of group B streptococcal infection and to obtain clinical information.

Group B streptococcal isolates received from the microbiology laboratories were confirmed for species and were serotyped. Because clinical group B streptococcal isolates are restricted to a few serotypes [22], additional subtyping methods were needed. We chose restriction endonuclease analysis of chromosomal DNA (REAC) because it has been shown to provide discriminatory power among group B streptococcal isolates of the same serotype [23]. Although multilocus enzyme electrophoresis has been shown to be of limited utility in subtyping group B streptococcal isolates obtained from infants [24, 25], we wished to evaluate the method among isolates obtained from adults.

Selection of Control Isolates

Fifty-three group B streptococcal BIDS isolates were selected as controls for REAC analysis to provide information on the genetic heterogeneity of group B streptococci in Maryland [26]. These isolates were obtained from sterile sites in nonpregnant adult residents of Maryland who had invasive group B streptococcal infection during the same time as the patients with recurrent infection.

Laboratory Assays

Serotyping was done by the Lancefield capillary precipitin method [27]; we used antisera to polysaccharide antigens Ia, Ib, II, III, and V and protein antigen c prepared at the CDC.

Group B streptococcal isolates were further characterized by REAC using the restriction enzyme HhAI (Gibco-BRL, Gaithersburg, Maryland). This technique was done using previously published methods [23], although 0.7% rather than 1% gels were used to improve the resolution of the larger bands. Although REAC results are somewhat cumbersome to read because of the many bands, we chose REAC as the main subtyping method because it is relatively easy to do and because it provides greater discrimination among group B streptococcal isolates than does ribotyping [23]. Because of the difficulties in interpreting subtle differences in REAC patterns, strains with similar REAC subtypes were rerun side by side on the same gel.

Group B streptococcal strain extracts were prepared, and multilocus enzyme electrophoresis analysis was done in 11.5% starch at a pH of 8.0 as previously described [28]. Gel slices were stained for the following enzymes: alcohol dehydrogenase, lactate dehydrogenase, nicotinamide-adenine dinucleotide (NAD)-dependent glyceraldehyde 3-phosphate dehydrogenase. NAD phosphate (NADP)-dependent glyceraldehyde 3-phosphate dehydrogenase, NADP-dependent glutamate dehydrogenase, reduced form of NADP diaphorase, indophenol oxidase, nucleoside phosphorylase, aspartate aminotransferase, hexokinase, carbamylate kinase, phosphoglucomutase, esterase, leucine aminopeptidase, arginine aminopeptidase, leu-gly-gly peptidase, phe-leu peptidase, aldolase, and phosphoglucose isomerase [29]. Electrophoretic variants of each enzyme activity were considered to be different alleles of that enzyme and were assigned separate allele numbers. Each strain was characterized by a list of allele numbers for the different enzymes, and each unique list of alleles was designated as an electrophoretic type and assigned a separate electrophoretic type number.

Statistical Analysis

The number of nonpregnant persons 18 years of age or older was estimated using 1990 census-based estimates of the 1 January 1993 population minus the total number of live births among Maryland residents in 1991, the last year for which complete birth data were available. We used the Kruskal-Wallis test to analyze continuous variables and used standard methods to calculate 95% CIs [30]. The probability of at least the number of observed intrapatient REAC subtyping matches was calculated using an independence assumption model in which each permutation of the second episode results was equally probable. We did the calculations by simulation, conditioning on the results of the first episode, and randomly permuting the results of the second episode 700 000 times. With no matches, this yields P values of less than 0.00001 with a probability of 0.999.

Results

Top Methods Results Discussion Author & Article Info References

Seven hundred fifty-one patients with invasive group B streptococcal infection were reported between 1 November 1991 and 30 September 1993; 449 (60%) of these were nonpregnant adults 18 years of age or older. The annual incidence of invasive group B streptococcal infection among nonpregnant Maryland residents during the study period was 6.7 per 100 000 persons, 13 times the incidence of adult invasive meningococcal disease during the same time period. Of the 449 adult patients, 54 (12%) were known to have died during the first identified episode by the time the case report forms were completed. The median duration of follow-up among the 395 survivors was 23 months (range, 12 to 35 months). Among the survivors, 17 (4.3% [95% CI, 2.6% to 6.9%]) had a second episode of invasive group B streptococcal infection. Among the 17 second episodes, 1 occurred during the last 2 months of 1991, 5 occurred during 1992, 6 occurred during 1993, and 5 occurred during the first 9 months of 1994. The patients resided in 17 ZIP code regions and in 8 of the 24 Maryland administrative jurisdictions.

An additional 5 patients with recurrent infection were identified with first-episode onset dates after 30 September 1993; thus, a total of 22 adult patients with recurrent infection were identified (Table 1). For the first episodes, group B streptococcus was isolated from blood in 20 patients and from the synovial fluid of the knee in 2 patients (patients 4 and 11). Group B streptococcus was isolated from blood for 21 of the second episodes and from the synovial fluid for 1 (patient 11). Patients received at least one antibiotic agent to which their group B streptococcal isolates were susceptible. The second episode occurred an average of 24 weeks after the first episode (median, 10 weeks; range, 2 to 95 weeks).

The mean age of patients with recurrent infection was 60 years (range, 27 to 89 years). Eleven patients (50%) were women, and 14 (64%) were white. The patients with a single episode of infection had similar age, sex, and race characteristics (data not shown). All patients had at least one serious underlying medical condition (Table 1), including cancer, diabetes, cirrhosis, and renal transplantation. Patients presented with various clinical syndromes for their first episodes, including bacteremia with no identifiable source, urinary tract infection, cellulitis, septic arthritis, and pneumonia. All patients responded to antimicrobial therapy for the first episode and improved clinically. Second episodes were often severe, including hypotension in 2 patients. Twelve patients were evaluated for endocarditis or other inapparent deep-site infection during the second episode; valvular vegetations were seen on echocardiography in 4 patients, including 1 patient who also had septic arthritis (patient 4). One patient had osteomyelitis of the L3 vertebral body, and another had osteomyelitis of the sternum. Patients 4 and 5 died during the second episodes of infection. Group B streptococcus appeared to be a major contributor to the illnesses of all patients but patient 5, whose illness and subsequent death appeared to be caused by recalcitrant Enterobacter cloacae bacteremia.

On the basis of the chart reviews, several of the patients had other documented episodes of invasive group B streptococcal infection that occurred before the BIDS project began; no isolates were available for these episodes. Patients 3 and 6 had positive blood culture for group B streptococcus 3 and 13 months, respectively, before the first episodes identified by surveillance. In addition, Staphylococcus haemolyticus and an unspeciated gram-positive coccus in chains were isolated from patient 1 between the first and second episodes. Patient 11 had an episode of septic arthritis caused by a non-group A, non-group D, ß-hemolytic streptococcus 20 months after the second episode. Several other patients had episodes of fever and chills during which blood cultures were not obtained.

In 18 of the 22 episodes (82%), group B streptococcus was the only bacterium isolated from a normally sterile body site. However, group B streptococcus and Streptococcus pneumoniae were coisolated from blood during patient 4's second episode. During the first episode, a viridans streptococcus was coisolated from the blood of a patient with a Hickman catheter (patient 5); E. cloacae and a viridans streptococcus were isolated during the second episode. A coagulase-negative staphylococcus grew in one of two blood cultures from patient 8; both blood cultures were positive for group B streptococcus. The group B streptococcal isolates from the first and second episodes were reportedly sensitive to all of the penicillins and cephalosporins studied and to vancomycin.

Of the 18 isolate pairs, 15 (83%) matched by serotype and 13 (72% [CI, 46% to 90%]) had identical REAC patterns (P < 0.001) Table 1. Among the 53 control isolates, 0% to 9.4% had the same REAC pattern as the 13 matching isolate pairs (Table 1). Of the 7 patients with deep-site infection for whom both isolates were available (4 with endocarditis by echocardiogram, 2 with osteomyelitis, and 1 with septic arthritis), all but the 1 patient with sternal osteomyelitis had isolates with matching REAC types. The isoenzyme typing provided little discriminatory power among isolates of the same serotype. For example, all but one of the serotype V isolates were electrophoretic type 4 (data not shown).

Patients with recurrent infection caused by the same strain had a shorter interval between episodes (mean, 14 weeks; range, 2 to 57 weeks) than patients with recurrent infection caused by another strain (mean, 43 weeks; range, 9 to 95 weeks; P = 0.05) Of the 40 group B streptococcal isolates associated with recurrent infection in adults, 16 (40%) were serotype V.

Among 241 infants surviving the first episode of group B streptococcal infection, only 1 case of recurrent infection (0.4% [CI, 0.0% to 2.3%]) was reported, despite similar efforts to detect such cases. The child was admitted at age 27 days with group B streptococcal bacteremia and subsequently had another episode at age 3 months. Isolates from both episodes were serotype III and REAC type 13. Adults were 10.4 times more likely to have a recurrent episode than infants (CI, 1.4 to 77.5; P = 0.003).

Discussion

Top Methods Results Discussion Author & Article Info References

The frequency of recurrent infection among patients with invasive group B streptococcal infection has not been previously quantified. The unique nature of the population-based active surveillance in all acute-care hospitals in Maryland allowed us to quantify the risk for recurrent infection and to obtain the group B streptococcal isolates for molecular epidemiologic analysis. Recurrent group B streptococcal infection occurred in a substantial proportion of Maryland adults with invasive group B streptococcal infection, but this is probably an underestimate of the problem. Patients with recurrent infection with longer intervals between episodes would not have been identified because of the brief follow-up period for many patients. In addition, patients with recurrent infection who were admitted to hospitals not participating in the BIDS project for one of their episodes would not have been detected by this system. Some patients could have had episodes during the study period that were unrecognized because blood cultures were not done or because bacterial growth was suppressed by the administration of anti-biotic agents. In fact, several of the patients had febrile episodes for which they were admitted to the hospital; blood cultures were not done during some of these hospitalizations. In Maryland, recurrent group B streptococcal infection appears to be much more common among adults than among infants.

Cases of recurrent group B streptococcal infection among adults have previously been reported [16-21]. Most reports of recurrent group B streptococcal infection involved cellulitis, often as a chronic complication developing after surgery. To our knowledge, molecular subtyping has not been done on group B streptococcal isolates from adults with recurrent infection. However, an analysis of isolates from an infant with three episodes of serotype III group B streptococcal infection showed that all three were identical by REAC [5].

Adults with invasive group B streptococcal infection clearly have an underlying immune defect because almost all patients have at least one chronic medical condition. Among infants, the level of maternal type-specific antibody correlates with the risk for invasive disease [31], although other host immunologic factors also play a role [32]. However, little is known about the host factors that predispose adults with chronic medical conditions to developing group B streptococcal infection. Studies in this area are warranted to determine whether the vaccines being developed for the prevention of neonatal group B streptococcal disease might be appropriate for use in adults [33].

Several factors may explain why our patients had recurrent infection. Among patients whose isolates had matching REAC types, some had deep-site infections, such as osteomyelitis and endocarditis, that were unrecognized during the first episode. Relapsing bacteremia probably resulted because these patients did not receive the prolonged course of antibiotic therapy required for these types of infection. The fact that all but 1 of the 7 isolate pairs from patients with documented deep-site infection (3 with endocarditis, 1 with endocarditis and septic arthritis, 2 with osteomyelitis, and 1 with septic arthritis) had matching REAC types supports this notion. Others with matching strains may have had persistent gastrointestinal or genital carriage of the invasive strain with subsequent relapse after successful treatment of the infection. Penicillin has been shown to be ineffective in eradicating group B streptococcal carriage in infants [34] and adults [35]. An alternative hypothesis for relapsing infection is the reacquisition of an invasive group B streptococcal strain from, for example, a household contact carrier.

Recurrent infection caused by different group B streptococcal strains, which tended to develop among patients who had a longer time interval between episodes, probably indicates that these patients are highly susceptible to repeated invasive group B streptococcal infection. This finding also suggests that defective host immunologic factors play a role in this setting.

Why were a substantial proportion of adult recurrent disease caused by serotype V, which was previously believed to be a relatively rare human serotype? Evidence suggests that serotype V has recently emerged as a common cause of invasive group B streptococcal infection among adults and infants [36, 37]. Of the 141 isolates from nonpregnant adults that we recently serotyped, 44 (31% [CI, 24% to 40%]) were serotype V [37]. In comparison, 40% of our isolates from adults who had recurrent infection were serotype V, suggesting that infection with this serotype is not strongly associated with recurrent infection.

Several methodologic issues on the molecular techniques we used warrant discussion. First, interpreting the REAC gels was difficult because of the many bands present in all isolates. This required that all apparently related isolates be rerun side by side on the same gel; this process was time consuming. However, the reading of the gels was facilitated by the fact that much of the discriminatory power of REAC was provided by the high-molecular-weight (> 5 kb) bands, which tended to be relatively easy to read. One approach for analyzing many isolates is to use ribotyping as a first step in identifying potentially identical isolates, followed by REAC to further discriminate among identical ribotypes [23]. We used a similar approach by identifying possible matches using the high-molecular-weight bands and subsequently running a second gel to identify identical REAC types.

Although some bacterial species show substantial heterogeneity in electrophoretic enzyme types [26, 38], the utility of this method for differentiating group B streptococcal isolates of the same serotype appears to be limited [24, 25]. Our study in adults confirms previous reports suggesting that multilocus enzyme electrophoresis does not adequately discriminate among strains of the same serotype in infants.

Our study has several other potential limitations. We may have noticed an unusual cluster of recurrent group B streptococcal cases that led us to overestimate the frequency of recurrent infection. However, the absence of geographic and temporal clustering and the continued occurrence of second episodes following the original observation essentially exclude this possibility. We do not know how commonly recurrent group B streptococcal infection occurs in areas other than Maryland. However, a cursory survey of other areas that participate in the National Bacterial Invasive Disease Surveillance Group, including northern California, Missouri, Oklahoma, four counties in Tennessee, and the Atlanta metropolitan area, suggests that this phenomenon is not unique to Maryland. Sixteen adults and eight children with recurrent infection were identified, including an adult in whom four episodes of invasive group B streptococcal infection were documented over a 4-year period. Some of the intrapatient group B streptococcal matches could have been due to chance because the frequency of matching REAC types among the control isolates was as high as 9.4%. However, the probability of 13 or more matches occurring by chance alone was low (P < 0.00001).

All adults with a first group B streptococcal bacteremic episode should have a careful physical examination to identify deep-site infection, such as endocarditis or osteomyelitis, and further evaluation as indicated. Patients with recurrent group B streptococcal bacteremia should be routinely evaluated more thoroughly for deep-site infection; echocardiography should be done during these routine evaluations. Although the precise duration of therapy in patients with these syndromes is unknown, a prolonged course of parenteral therapy with antibiotic agents (for example, 4 to 6 weeks) is warranted. Patients with recurrent disease and no identifiable source of infection should be considered for eradication of group B streptococcal carriage. However, the most appropriate antibiotic regimen [5, 10] and the efficacy of such an approach are unknown.

The frequency of recurrent group B streptococcal infection should be prospectively determined in other populations. We are studying host, treatment, and bacterial factors that may distinguish patients who develop recurrent infection from those with single episodes. These studies are needed to determine the optimal therapy in adults with invasive group B streptococcal infection to minimize the likelihood of recurrent episodes.

Drs. Ali and Johnson and Mr. Lashkerwala: Veterans Affairs Medical Center, Room 4D-148, 10 North Greene Street, Baltimore, MD 21201.

Drs. Dwyer and Libonati: Maryland Department of Health and Mental Hygiene, 201 West Preston Street, Baltimore, MD 21210.

Drs. Reeves and Elliott: Centers for Disease Control and Prevention, 1600 Clifton Road NE, Atlanta, GA 30333.