Interrelationships Within the Bacterial Flora of the Female Genital Tract

Analysis of 240 consecutive vaginal swabs using the compatibility profile technique revealed that only 2 bacteria have the ability to be a sole isolate and as such a candidate to be a major aerobic regulator of the bacterial flora of the female genital tract (BFFGT). Compatibility profiles of Lactobacillus and Gardnerella vaginalis have shown that these organisms shared compatibility profiling for the majority of the normal bacterial constituents of the female genital tract. Dominance disruption appears to come from the addition of compatible co-isolates and presumed loss of numerical superiority. These phenomena appear to be the keys to reregulation of BFFGT. Lactobacillus appears to be the major regulator of both G. vaginalis and anaerobic bacteria. When additional organisms are added to the bacterial flora, they may add to or partially negate the inhibitory influence of Lactobacillus on the BFFGT. Inhibitor interrelationships appear to exist between coagulase-negative staphylococci and Staphylococcus aureus and the group B streptococci (GBS) and other beta hemolytic streptococci. Facilitating interrelationships appear to exist between S. aureus and the GBS and selected Enterobacteriaceae.

he female genital tract microbiology is postulated to be a tightly orchestrated, dynamic system which follows defined patterns of regulation. 1,z Most vaginal cultures have 3-6 bacteria. 3,4 Only a minority of cultures have a single bacterium. When group B Streptococcus (GBS) is present in high numbers (>107/cfu/g of vaginal fluid), the concomitant bacterial flora tends to be simplified (M0nif, unpublished data). When present in the vaginal flora at <106/cfu/g of vaginal fluid, co-isolation of multiple bacteria with numerical superiority by one or more of these co-isolates can be demonstrated. Perception of a high degree of governance has been obscured by the various combinations of bacterial combinations which can flnction within the limits imposed by a given microbiological environment.
Our perception of bacterial dominance emanates from studies with the GBS which implied that some form of bacterial interference selectively functions governance of that particular organism, s The composition of each bacterial vaginal flora appears to be tightly regulated until there is an alteration of the microbiological environment or one or more of the governing members of the microbiological flora is removed. 1,7 When antibiotics are administered, a regulatory interrelationship is disrupted which creates a void into which existing microbial organisms can expand to governance or new microbial organisms can move into the void and potentially alter the quantitative interrelationship of those bacteria present.
Bacterial studies of the bacterial flora of the fe- other. Most of the information concerning governance is based on in vitro work which analyzed the presence or absence of bacterial interference. For in vitro data to have in vivo relevance, the quantitative representation of the target bacteria must be assured. The ability of one bacterium to adversely affect the replication of other bacteria has been well documented, and is presumed to be a principal mechanism accounting for the constituency of the BFFGT.

Specimen Handling
All specimens received over a 6 month period were cultured for aerobic and anaerobic bacteria and Candida species. These specimens were received primarily from ambulatory care clinics ( Table 1). Additional workup of anaerobic cultures required confirmation of anaerobic bacteria by inoculating the suspected anaerobic organism onto blood agar and anaerobic CDC media. After overnight incubation in aerobic and anaerobic conditions, respectively, organisms which grew on the latter medium but not the former were worked up as anaer- obes. Gram stains, colony morphology, and the Rapid ANA II system (Innovative Diagnostic Systems, Inc., Norcross, GA) contributed to the identification of these organisms.

Comparative Profiling Technique
The microbiological data were analyzed by the comparative profiling technique developed by G.R.G.M. The technique starts by identifying those bacteria which occurred as a single isolate and then looks at the co-isolates present when only 2 bacteria are recovered. The most prevalent of these bacteria is then added to the initial target bacteria and co-isolates are identified when only 3 bacteria are recovered. This process is again repeated using cultures when 4 bacteria are present, etc. This process of additive bacterial grouping lends to the establishment of a compatibility profile. By inference those bacteria which are not present may be susceptible to bacterial interference by one of the target organisms or its subsequent additive isolates.

Statistical Analysis
The microbiological data were analyzed statistically using the standard or enhanced association Student's t-test for evidence of bacterial interference. Analysis of bacterial interrelationships was restricted to those bacteria isolated in sufficient prevalence so that statistical validity could be achieved. Where appropriate (sample size <5) Fisher's exact test was used. P < 0.05 was considered to be significant.  (8) Gram-negative anaerobic bacilli 7

RESULTS
Seven hundred eighty of the isolates were achieved: 783 aerobic isolates and 104 anaerobic isolates. The overall tabulation of bacterial isolates is listed in Table 2. From the microbiological data, the cultures and bacterial specimen handling techniques were adequate for aerobic bacteria; how-ever, the low prevalence of gram-positive anaerobic rod (anaerobic lactobacilli) Eubacterium, Propionbacterium, Bifidobacterium, and Clostridium was consistent with the concept that the anaerobic data were probably more reflective of high multiplicity replication rather than identification of true prevalence.

Comparative Profiles
When comparative profiles were done using the 781 isolates achieved from 239 vaginal cultures, the only bacteria which achieved a single isolate status were Lactobacillus and Gardnerella vaginalis (see Tables 3, 5). The compatibility profiles were constructed based on these two bacteria. As would be anticipated, the recovery of Lactobacillus was at the lower end of recorded prevalence and that for G.
vaginalis was at the upper end of their recorded prevalence in normal women without overt diseases.
Lactobacillus Lactobaci//us was identified in 131 vaginal cultures (Table 3). In 7 cultures, Lactobacillus was the sole bacteria isolate. On the first level with 2 isolates per culture, compatibility profiling co-isolates were coagulase-negative Staphylococcus (9), group D Enterococcus (8), aerobic gram-positive bacilli (3), Enterobacteriaceae (2), and alpha hemolytic Streptococcus (1). On level two with the combination of Lactobacillus and coagulase-negative Staphylococcus used to identify the third isolate within the cultures with just 3 isolates, the results were as follows: group D Enterococcus (8), Enterobacteriaceae (3), alpha or gamma Streptococcus (3), aerobic grampositive bacilli (1), and group D Streptococcus (1). When the combination of Lactobadllus and group D Enterococcus was used to identify the third isolate within cultures with only 3 isolates, the results were as follows: coagulase-negative Staphylococcus (8), GBS (2), Enterobacteriaceae (1), alpha Streptococcus (1), and anaerobic bacteria (1). On level three when the combination of Lactobadllus, coagulasenegative Staphylococcus, and group D Enterococcus was used to identify bacteria present, the following co-isolates were identified: Enterobacteriaceae (4), GBS (2), anaerobes (3), diphtheroids (1), aerobic gram-positive bacilli (1), and alpha or gamma Streptococcus (1). When Lactobadllus was present, G. vaginalis was a concomitant isolate in 7 of 108 cultures. When no anaerobes were present, only one isolate       (2)]. Of the 10 Klebsiella isolates, 8 were K. pneumoniae and 2 were K. oxytoca. K. pneumoniae and E. coli were co-isolates in 2 cases. Of the 5 P. mirabilis isolates, 3 had co-isolates: 2 E. coli and Acinetobacter. Multiple Enterobacteriaceae isolates occurred in the context of multibacterial (>5) cultures in 3 of the 5 cases.

Coagulase-negative Staphylococcus
Of the 132 patients from'whom a Staphylococcus was isolated, there were only 2 instances of co-isolation of a coagulase-negative Staphylococcus and another Staphylococcus species. The presence of a coagulasenegative Staphylococcus did not appear to inhibit anaerobic staphylococci (Peptococcus magnus and P. asaccharolyticus; currently classified with the Peptostreptococcus).

DISCUSSION
The importance of quantitative microbiological data could not be addressed in this study owing to cost considerations. Nevertheless, some insight can be inferred as to the role of the magnitude of cfu/g of vaginal fluid in bacterial dominance. In vitro studies of bacterial interference have shown that inhibition is related to numerical dominance. When multiple anaerobes were isolated and/or the total number of isolates from a given vaginal specimen exceeded 5, this combination was presumed to indicate the presence of the anaerobic .progression. When the anaerobic progression is functioning, the regulatory influence of traditional dominant aerobes tends to be regulated to a minor role (Monif, unpublished data). When quantitative analysis is done in the anaerobic progression, the number of anaerobes per gram of vaginal fluid exceeded that of aerobic bacteria. 4 Data supporting this thesis are inferred by the in vitro inhibition studies done with the Lactobacillus/G. vaginalis and coagulase-negative Staphylococcus/S. aureus couplings. When numerical disruption is inferred by the presence of anaerobic progression, the prior in vitro demonstrated ability of given aerobic bacteria to inhibit an otherwise susceptible bacteria in the coupling is significantly diminished.
Lactobacilli produce lactic acid, HzOz, and other antimicrobial metabolics which in vitro inhibit other bacteria. 6,7,9 The role of hydrogen peroxide production by the lactobacilli isolates was not addressed in this study. Hydrogen peroxide producing lactobacilli were identified as being present in 90% of women with no vaginal symptomatology. 9 Four percent of cultures contained Lactobacillus species that did not produce hydrogen peroxide. 9 Since only lactobacillis which grow on both aerobic and anaerobic media were identified as lactobacilli, it can be presumed that these represent primarily hydrogen peroxide producing isolates. In the experience of Eschenbach et al., 9 the majority of nonhydrogen peroxide producing lactobacilli grew anaerobically. Klebanoff et al. 7 demonstrated in vitro the ability of hydrogen peroxide generating L. acidophilus to inhibit G. vaginalis and Bacteroides bivius. Other investigations have attributed in vitro growth inhibition of G. vaginalis and other anaerboic bacteria to acid production by lactobacilli. 8 Selected Enterobacteriaceae, diphtheroids, and S. aureus are catalase positive and theoretically could destroy hydrogen peroxide. Only the presence of diphtheroids appeared to alter the probability of isolation of aerobic lactobacilli.
The in vivo data in this study confirm prior in vitro data which documented the ability of Lactobacillus to inhibit G. 2)glgila]is. 1'7'8 When Lactobacillus was present, G. vaginalis was in concomitant isolates in 7 of 113 cultures. When no anaerobes were present, only one concomitant isolate of G. vaginalis was recorded. The other 6 cases occurred in the presence of isolation patterns consistent with the anaerobic progression. The alternate thesis is that these isolates are non-hydrogen producing isolates.
Chaisilwattana and Monif published the only extensive in vitro study on the ability of the GBS to inhibit gram-positive and gram-variable aerobic bacterial constituents of the female genital tract. Given numerical dominance created by the in vitro techniques used, the GBS inhibit group A, B, C, and G streptococci, lactobacilli, G. vaginalis, and most diphtheroid strains. While GBS did not inhibit coagulase-negative staphylococci, S. aureus, and most enterococci, GBS isolates were uniformly inhibited by coagulase-negative staphylococci, but were not inhibited by S. aureus.
The failure of in vitro data concerning GBS/ lactobacilli and G. vaginalis to be predictive of in vivo observations is presumed to be a function of the test technique. The in vitro condition necessarily demonstrates that inhibition cannot be duplicated in vivo. The demonstrated ability of Lactobacillus and G. vaginalis for governance makes it unlikely that the numerical discrepancy with GBS required for inhibition could ever be achieved. The in vivo absence of other beta hemolytic streptococci or diphtheroids except when evidence of the anaerobic progression was present, is consistent with the prior in vitro observations. GBS was recovered in conjunction with both coagulase-negative staphylococci and S. aureus. These rates ofrecovery were statistically different (P < 0.05). In the absence of quantitative data, failure of projected inhibition of GBS by coagulase-negative staphylococci is difficult to interpret, whereas the positive correlation of GBS with S. aureus is totally consistent with the in vitro data. In vitro, the coagulase-negative Staphylococcus suppressed S. aureus. Co-isolation of a coagulase-negative Staphylococcus and S. aureus did not occur unless polymicrobial bacterial isolates were concomitantly present. Until quantitative and qualitative studies are concomitantly done within qualitative studies, con-jecture-based understanding of the interrelationships within the bacterial flora of the female genital tract will be a combination of in vitro and in vivo studies, fragmentary quantitative or qualitative data. The number of observations required and the cost of doing such a study properly make it unlikely such data will be forthcoming in the foreseeable future. Further utilization of bacterial compatibility profiling and selective use of quantitative bacteriology will either sustain or refute the concepts advanced in this paper.