Initially, Tannerella forsythensis was thought to be a relatively uncommon subgingival species. The studies of Gmür et al. using monoclonal antibodies to enumerate the species directly in plaque samples suggested that the species was more common than previously found in culture-based studies (182). According to the extensive literature bacterial culture is considered inadequate for the detection of T. forsythensis (161, 192). Therefore, studies using this method will not be discussed here.

Recently, 16S rDNA sequence analysis confirmed that T. forsythensis (formerly Bacteroides forsythus) was not a species within the genus Bacteroides sensu stricto. A new genus, Tannerella, was proposed for B. forsythus, with one species, Tannerella forsythensis (220). Interestingly, the nearest genetic neighbour, oral clone BU063, has been associated with oral health (221). The oligonucleotide probe used in the present study is not complementary with that clone. It is conceivable that earlier studies mistakenly pooled several strains of T. forsythensis with different pathogenic potentials.

In the current study T. forsythensis was one of the species with a high prevalence (96%) in patients with generalized aggressive periodontitis (Fig.1). Additionally, the patients showed high load of T. forsythensis, i.e. 73.3% of patients had all pockets sampled positive for this species (Fig.6). T. forsythensis could be detected significantly more often in periodontal pockets than in healthy control sites (88.6% and 34.1% respectively) (Fig.2). However, no significant differences in detection frequency between different pocket depths could be seen (Fig. 5). Gmür et al. using a quantitative approach could demonstrate that the levels of T. forsythensis were strongly related to increasing pocket depth (182). Young patients with aggressive periodontitis studied by Kamma et al. (170) revealed a similarly high prevalence of T. forsythensis (98.5%). In that investigation indirect immunofluorescence assay was used. The site prevalence (83% positive pockets) was in good agreement with our study. However, no control sites were sampled.
In young adults with advanced periodontitis a high prevalence of T. forsythensis (60-70%) has been reported by using DNA probe approach (191, 206, 212). Similar results were obtained with a PCR-assay (11, 14).
Studies with the aim of correlating clinical parameters and the presence of certain bacterial species concluded that T. forsythensis was positively correlated with the clinical signs of a disease and can be regarded as a risk indicator for attachment and bone loss (145, 146, 222). Persistence of T. forsythensis post-therapy and the subsequent deterioration of periodontal conditions suggests evidence for the species association with recurrent periodontitis (156). It has been shown that with the reduction of T. forsythensis below the detectable level clinical improvement was significant (151). However, clinical improvement is never a consequence of a reduction / elimination of single species.

Choi et al. used a detection method similar to our study for the evaluation of the prevalence of T. forsythensis in patients with chronic periodontitis and in healthy persons (176). High prevalence (96.6%) was reported in chronic periodontitis patients. The prevalence was significantly lower in healthy persons. However, the comparison between the groups is unreliable because of incorrect statistics.

More sensitive microbiological tests have been of advantage for detecting T. forsythensis in healthy subjects. The current study particularly identifies T. forsythensis as a much more frequent member of the microflora of healthy sites than previously suspected on the basis of culture investigations. This contradicts the hypothesis that the mere presence of T. forsythensis may be taken as an indicator of active periodontal breakdown. T. forsythensis could be detected in 85.7% of the elderly (Fig.1). Almost half of the sampled sites in this patient group were positive (Fig.3). It has been reported that the colonization of supragingival plaque of healthy subjects, even in early childhood, is frequent (34, 35, 223-226). These data suggest the indigenous nature of T. forsythensis. However, in special periodontal / host conditions or specific proportional bacterial constellations T. forsythensis can induce periodontal breakdown. The presence of putative pathogens increases the risk of disease development. The presence of T. forsythensis has been especially correlated with attachment loss, increasing the risk by a factor of 2.45 - 5.3 (36, 145, 227). Even 8.16 times greater odds of attachment loss by presence of T. forsythensis has been reported in a longitudinal study (222). A significant increase in the proportions of T. forsythensis has been observed in sites with periodontal breakdown (182).
Little information is available on the virulence of T. forsythensis. Beside the LPS and the production of a trypsine-like protease (228) its ability to penetrate host cells and to induce apoptosis has received attention (84, 229). Recently, a cysteine protease gene (prtH) has been identified (226). It was shown that 85% of T. forsythensis isolates from diseased sites revealed a prtH genotype, however only 10% of such strains could be detected in healthy sites. There is a need for more comprehensive research of the virulence and taxonomic variability of T. forsythensis.

The evidence for associating T. forsythensis with severe periodontal disease has mainly been based on epidemiological data and evaluation of the correlation between the presence of the species and clinical conditions.
The data of the present study support the evidence of T. forsythensis being a periodontopathogen. The species was significantly more frequently detected in periodontal pockets than in control sites of GAP patients (Fig. 2). Also, the significantly higher prevalence and load of T. forsythensis in GAP patients than in healthy elders (Fig. 3, 6) indicates its strong association with aggressive periodontitis in young adults.

Porphyromonas gingivalis is the most intensively investigated periodontal species. It has frequently been associated with severe periodontitis (10, 15, 31, 167, 230). The culture-based approach can be considered adequate for the detection of P. gingivalis, as was shown by Conrads through comparative analysis using DNA hybridization (193).
The prevalence of P. gingivalis in the generalized aggressive periodontitis group was high, 63.6% in the present study (Fig.1). Interestingly, most of these positive subjects showed colonization with P. gingivalis in all sampled pockets (Fig.6). The species was significantly more frequently detected in periodontal pockets than in control sites (59.1% and 22.7%, respectively) (Fig. 2). The site-prevalence was significantly lower in the elderly group than in the GAP population (Fig.3). Porphyromonas gingivalis could hence be associated with the development of aggressive periodontitis.
Loesche et al. recovered high proportions of P. gingivalis along with high proportions of spirochetes from one EOP patient group, so-called "type B" patients (13). This group resembles the GAP entity as derived from clinical descriptions. These patients had significantly higher proportions of P. gingivalis than in four other patient categories. Also Alabandar et al. associated P. gingivalis and Treponema denticola with more severe and progressive forms of EOP (230). Kamma et al. recovered more frequent and significantly higher proportions of P. gingivalis from deeper periodontal pockets as compared to shallow sites in RPP patients (15). Her recently published data obtained by using indirect immunofluorescence technique revealed higher detection frequency of P. gingivalis in aggressive periodontitis group than the results of our study (170). The subject prevalence was as high as 89.4%, and 80% of the sampled pockets were positive for P. gingivalis.
P. gingivalis has also been strongly associated with chronic periodontitis (176, 231, 232) and recurrent periodontitis (142). Several investigators have found a significant correlation between proportions of P. gingivalis and attachment loss (39, 137, 182, 233).

Contradictory data exists on the presence of P. gingivalis in healthy periodontal conditions. In the current study, a relatively high prevalence (62%) was observed in the elderly with no significant difference to the periodontitis group (Fig.1). However, only 32% of the sampled sites of elderly were positive (Fig. 3), showing infrequent colonization of P. gingivalis in the control group. The difference to the periodontitis group was highly significant (p<0.0001). The species was rarely found in shallow sites (Fig.4).
Healthy persons investigated by Tanner et al. were not colonized by P. gingivalis (36). This was proven by bacterial culture and checkerboard hybridization methods. Healthy subjects and the elderly investigated by Haffajee et al. showed very low site-prevalence of P. gingivalis (4% and 5% respectively) (34). Absence of P. gingivalis in pre-school children and students confirmed by PCR assay demonstrates that this bacterium is usually not part of the resident oral flora in young healthy people (201, 224). Other authors have found a prevalence of 10-30% in older healthy subjects (232, 234). It raises the question whether these individuals may carry different, possibly less-virulent strains of P. gingivalis, or does the host response determine the outcome?
There is extensive evidence of the pathogenic nature of P. gingivalis (s. 1.2.6.). However, it is complicated to prove the relevance of the reported virulence factors in vivo. Additionally, it has been proven that virulence varies among strains. P. gingivalis is able to invade human gingival epithelial cells in vitro (76). In a mouse model different strains of P. gingivalis exhibited varying levels of invasiveness (235). Proteolytic enzymes have been suggested as a possible virulence factor. However, no differences in proteolytic enzyme production between invasive and non-invasive strains could be demonstrated (236).
As yet it is unknown which virulence factor correlates with more virulent P. gingivalis strain. In animal studies, strains W 50 or W 83 were highly virulent. Various strain-typing approaches (RFLP, ribotyping, serotyping, multilocus enzyme electrophoresis) have been used to identify highly virulent genotypes and to correlate them with disease. As yet, there is still no convincing evidence to associate specific genetic clone clusters with periodontal disease and hence numerous P. gingivalis genotypes were associated with disease (234, 237). Several studies confirm that patients are usually colonized by a single, unique genotype, present in sites with different clinical status. (237, 238). Healthy subjects were likely to harbour several strains (234).
Obviously, there are various virulence factors of P. gingivalis playing a role in the etiopathogenesis of periodontitis. Therefore, an attempt of epidemiological studies to detect virulent P. gingivalis strains by targeting only one suspected virulence gene is questionable.
Currently it is not possible to associate any genotype with aggressive periodontal disease. P. gingivalis can hence be regarded as an opportunistic pathogen. This evidence is supported by the low prevalence of P. gingivalis in healthy adults. Under suitable yet unknown environmental / host conditions certain strains of P. gingivalis can multiply and express virulence factors inducing disease development. Frequent detection of P. gingivalis together with T. forsythensis in active/deep periodontal pockets supports the evidence that certain consortia are of particular importance in progressive disease. The present epidemiological data have confirmed a strong association of T. forsythensis and P. gingivalis with generalized aggressive periodontitis.

Actinobacillus actinomycetemcomitans is considered to be of major etiologic relevance in the localized form of aggressive periodontitis (LAP) (14, 26, 49, 50, 239). The role of A. actinomycetemcomitans in generalized aggressive periodontitis is still unclear. The study presented here showed a low prevalence of this species (36.4%) in GAP patients (Fig.1). However, in 11% of the patients A. actinomycetemcomitans was present in all sampled pockets (data not shown). Generally, the species was evenly distributed in periodontal pockets and shallow sites (Fig.2). There is no preferential niche in diseased sites (Fig.5). In patients with a high load of A. actinomycetemcomitans the species can play an important role (240). It cannot be ruled out that some of the GAP patients previously had the localized form of aggressive periodontitis.
Kamma et al. (170) found a low prevalence (25%) of A. actinomycetemcomitans in patients with aggressive periodontitis, similarly to our results. Tanner et al. detected high proportions of A. actinomycetemcomitans in young adults with advanced disease (241). Older subjects with advanced disease did not harbour this organism. Similarly, older GAP patients investigated by Loesche et al. were negative for A. actinomycetemcomitans (13). Slots et al. observed that the prevalence of A. actinomycetemcomitans decreased significantly with age. The authors reported that 74.4% of the 15-24 year-age group were colonized by the microorganism compared to 38.7% of those 25-34 years of age (242).
Van Winkelhoff et al. observed a negative association between A. actinomycetemcomitans and Fusobacterium spp. in subgingival plaque (243). The authors followed LJP patients and found that A. actinomycetemcomitans was not recovered after these patients reached middle-age and developed more widespread periodontal breakdown. It may be that fusobacteria possess the capability to inhibit the colonization of A. actinomycetemcomitans as the genus Fusobacterium emerges as a prominent taxon of generalized EOP (243). Similarly, in the study presented here Fusobacterium spp. showed a high site-prevalence (96.7%) and negative association with A. actinomycetemcomitans (Table 2).
There are, however, studies demonstrating a low prevalence of A. actinomycetemcomitans even in LJP patients (13, 244). Obviously, A. actinomycetemcomitans plays an important role in some cases of localized and generalized forms of aggressive periodontitis. The evidence of this association can be traced back to the virulence of the species, epidemiological data and correlation with improvement after treatment.
Single clones of exceedingly high virulence may be implicated in etiopathogenesis of periodontitis. A. actinomycetemcomitans strains differ in their ability to produce leukotoxin. This exotoxin lyses polymorphonuclear leukocytes and macrophages (106). Highly leukotoxic strains produce 10-20 times more leukotoxin than other strains (245). Bueno et al. found that subjects harbouring A. actinomycetemcomitans with a 530-bp deletion in the leukotoxin promoter region were more likely to convert to LJP than subjects who had A.a.-variants containing the full-length leukotoxin promoter region (246). In addition to the predominant colonization of A. actinomycetemcomitans in younger patients, it has been observed that very young patients (mean age 12.7 y.) harboured highly toxic strains (245). Young adults (mean age 25.5 y.) were colonized by minimal toxic strains. However, since the strains without the 530-bp deletion have been recovered from LJP lesions as well (245), high leukotoxin production may not be a prerequisite for pathogenicity. No epidemiological study looking of the distribution of highly virulent strains has been performed with GAP patients.
A. actinomycetemcomitans strains also vary in their capability to invade epithelial cells (247). The failure of non-surgical therapy to effectively control A. actinomycetemcomitans from subgingival sites may stem from the ability of the organism to invade gingival tissue and thereby evade the effect of mechanical debridement and periodontal healing (248). Investigation of biofilm formation in a plaque-free-zone of the bottom of a pocket demonstrated no participation of A. actinomycetemcomitans (46).
Studies evaluating the correlation between treatment outcome and the presence of bacteria in LJP and "severe periodontitis" found a clear association between the improvement of periodontal conditions and the elimination of A. actinomycetemcomitans (249, 250). The failure to eliminate the combination of A. actinomycetemcomitans and T. forsythensis resulted in attachment loss (151). These microorganisms have been detected frequently in refractory periodontal lesions (231). An exceedingly high proportion of subgingival A. actinomycetemcomitans in periodontal sites undergoing active breakdown gives substantial credence to the notion of it being a key bacterial pathogen in certain cases of LJP (231). In chronic periodontitis patients the microorganism was not found to be related to an increased risk of disease recurrence (137).

Healthy elderly individuals in our study showed a very low prevalence (9.5%) of A. actinomycetemcomitans (Fig.1, 3), demonstrating that the species is not a common colonizer in healthy periodontal conditions. This is in agreement with other reports about healthy individuals (34, 176, 231). Healthy children were shown to be negative for A. actinomycetemcomitans by PCR (201, 251). Even patients with chronic periodontitis are seldom colonized with this species (34, 192). Gmür et al. studied dental hygienists and observed a prevalence of 33% in supragingival plaque in spite of their better than average personal plaque control. However, the detected cell numbers were <1% of the sampled microbiota (35). A. actinomycetemcomitans has also been found in tongue and saliva samples, even without subgingival colonization (252).
The strains identified in healthy subjects or patients with chronic periodontitis were shown to be minimal leukotoxic strains(245).
Chronic periodontitis patients investigated within the comprehensive epidemiological study with the identical methods described here showed a site-prevalence of only 7% (Dr. Moter, personal communication). Interestingly, Choi et al. could detect this organism in almost 90% of chronic periodontitis patients in a Korean population using a similar detection method (176). The high prevalence can be explained by ethnic, immunological, nutritional factors.
Apparently, A. actinomycetemcomitans (probably special strains) can be regarded as etiologically relevant in aggressive periodontal disease in young adults, however, the evidence is not as strong as for the localized form (14, 230, 239, 240, 253).

Campylobacter rectus could be detected in more than half of the GAP patients; however, only 31.8% of the pockets were positive (Fig.1,2). Still a significant difference was observed in the colonization pattern of periodontal pockets and control sites in the study population (Fig. 2). When compared to the occurrence in the healthy elderly group a highly significant difference could be seen in site-prevalence (Fig. 3). The species could be rarely detected in healthy population (prevalence 23.8%) (Fig.1).
High levels of C. rectus, especially when found together with T. forsythensis have been related to an increased risk of disease progression (31, 33, 36, 39, 137, 138, 254). The highest association among all organisms in the current study was observed for these two species (Table 2).
Kamma et al. recovered C. rectus exclusively in the deep pockets of GAP patients (15). The microorganism has been seen predominately in the middle and deep pocket zones forming large clumps when examined by scanning immunoelectron microscopy (46). C. rectus may advance to the most apical border of plaque area by the use of its motility and associate with biofilm formation in the plaque-free-zone and with disease progression.
The high prevalence of C. rectus in aggressive periodontitis patients has been reported in the literature in contrast to low detection in healthy populations (14, 15, 30, 170, 255). The data of Gmür et al. confirm that supragingival plaque is a natural habitat for C. rectus in periodontally healthy persons with good oral hygiene (35, 225). The species could be identified in 48% of the investigated subjects using an immunoassay. Between the elderly and the healthy control group no difference was seen (34). C. rectus was identified in less than 20% of the subgingival samples. The site-prevalence was not higher in patients with chronic periodontitis (19.3%) (Dr. Moter, personal communication).
Patients with aggressive periodontitis showed significantly higher and more frequent elevation of serum IgG antibody to this organism as compared to chronic periodontitis patients or healthy controls (256).
C. rectus may be considered as a putative pathogen that occasionally, probably in constellation with T. forsythensis, contributes to the development of aggressive periodontitis. It may need to exceed relatively high critical threshold values in the subgingival flora to lead to a progression of the disease (257).

According to the literature Fusobacterium nucleatum is the most commonly isolated organism in subgingival samples, especially in deep periodontal pockets. It has often been recovered in high proportions in different patient groups (10, 31, 33, 241).
The data of the present study suggest a strong association of F. nucleatum with GAP. The species was present in 91% of the GAP patients (Fig.1) and significantly more frequently in periodontal pockets than in shallow sites (Fig. 2). All 4 sampled pockets were positive for this organism in half of the patients (Fig.6). The load in the elderly subjects was significantly lower. The site-prevalence in the elderly was only 25% (Fig. 3). Haffajee et al. in contrast, identified F. nucleatum ss. polymorphum in 58% of the sites of the elderly subjects using checkerboard hybridization method (34).
It is however, difficult to interpret the role of F. nucleatum from our data, as the probe simultaneously detected F. periodonticum. There are reports of F. periodonticum being a frequent colonizer of the supra- and subgingival plaque of healthy individuals and periodontitis patients (30).
Furthermore, the phenotypic and genetic heterogeneity of F. nucleatum has led to an attempt to classify the strains into taxonomically relevant groups, such as subspecies, but these efforts have not resulted in a widely accepted taxonomy (258). Within the species F. nucleatum, a number of investigators have identified a range of distinct genetic clusters which have subsequently been designated as subspecies. Unfortunately the results often do not correlate with each other. Morris et al. could demonstrate that the species F. nucleatum consists of at least three distinct species (186). There is a need for revision of the previously designated genetic divisions as well as phenotypic characterization of F. nucleatum and genus Fusobacterium as a whole.
Therefore, no differentiation between the subspecies was undertaken in the present study.
Loesche et al. did not associate F. nucleatum with GEOP, because the species formed only ca. 3% of the total bacterial count (13). In the study of Kamma et al. F. nucleatum was detected most often in the medium and deep lesions of GAP patients (15). Chronic periodontitis patients exhibited high prevalence of F. nucleatum (204). A prevalence of 81.3% was observed in the reported epidemiological study (Dr. Moter, personal communication).
The adherence factors of F. nucleatum have been attributed to the potential pathogenicity of certain strains, however, no correlation could be established between any particular subspecies (259). F. nucleatum could not be detected in the plaque-free zone at the bottom of periodontal pockets suggesting that this microorganism does not primarily participate in the apical progression of plaque (46). The species was usually located in the middle and deep pocket zones in an unattached plaque area.
F. nucleatum has been implicated in disease progression, as significantly higher proportions were detected in active lesions than in inactive sites (39, 140).
Our data provide evidence that F. nucleatum might be associated with aggressive periodontitis. The species is only infrequently detected in healthy subjects (Fig.1, 3).

Contradictory reports exist about Prevotella intermedia. This microorganism has been found in high proportions in adults with moderate to severe periodontal breakdown and in EOP patients (13, 183). At the same time, a high prevalence has been reported in treated and maintained patients, in some of them in high proportions (13). Colonization of P. intermedia in children has also been observed (201, 224).
Recently, strains of Bacteroides intermedius with similar phenotypic traits have been classified into two species, Prevotella intermedia and Prevotella nigrescens. This distinction makes earlier studies on these organisms difficult to interpret, since data from two different species may have been inadvertently pooled (260). It has been reported that P. intermedia is associated with periodontitis, whereas P. nigrescens is a natural inhabitant of the gingival sulcus and the supragingival plaque (35, 40, 201, 261).
In the present study no clear association for P. intermedia with GAP or periodontal health was found. There was no significant difference in subject-based prevalence and load of P. intermedia between the groups (Fig.1, 6). The species was more often detected in the pockets of GAP patients than in the control sites (Fig. 2), and when compared to the positive sites of elderly (Fig. 3). The difference however, was always only moderately significant (p<0.05).
Mullally et al. (14) associated P. intermedia with GEOP because of its high prevalence (58.8%), which is lower than in our study. However, no control group was included in the study, reducing the relevance of the results.
Kamma et al. (15) found no difference in the detection frequency of P. intermedia between shallow sites and deep pockets, however, the proportions were significantly higher in deep pockets.
Earlier studies using unreliable "predominant cultivable microbiota" method recovered higher proportions of P. intermedia in active sites when compared to inactive sites, however, without significant difference (39).
Concerning the probable differences in virulence of P. intermedia clones no correlation could be found since the same genotypes were found at both diseased and non-diseased sites (237).
Choi et al. reported a high prevalence of P. intermedia (90%) in patients with chronic periodontitis and significantly lower in healthy subjects (only 5%) in a Korean population (176).
The data of the elderly group in our study revealed frequent detection of the species in well-maintained subjects (prevalence 66.7%) (Fig.1) incriminating P. intermedia as a common colonizer of the oral cavity in healthy oral conditions. Conrads et al. (201) found P. intermedia very frequently in plaque samples from children with PCR detection assay.

However, contradictory results have been gained. Molecular genetic analysis revealed no significant difference in site-prevalence between healthy, elderly and chronic periodontitis groups (34). Interestingly, despite more frequent supra- and subgingival colonization of P. intermedia in periodontitis patients than in healthy persons, the quantitative assessment showed no significant difference in proportions (30).
Although in some studies the etiologic role of P. intermedia in periodontitis was considered to be high (11, 13, 30, 230), a relatively low risk (1.6) for periodontal breakdown in the presence of P. intermedia has been reported (48). It seems that higher mean counts of P. intermedia than for P. gingivalis are needed for the progression of the disease. Rams et al. assessed the relative risk for periodontal breakdown with respect to the proportions of certain species and found that much higher proportions of P. intermedia are needed to reach a 2.5 relative risk for periodontitis recurrence when compared to other putative pathogens (233). Analysis of the humoral host response against P. intermedia has shown similar antibody levels in individuals with and without periodontal disease (262). This lack of association with periodontal disease could be explained in part by the frequent colonization of P. intermedia in locations other than periodontal pockets (183).
It is difficult to evaluate the role of P. intermedia given the variable and contradictory literature. Our data incriminates this organism as one of moderate importance in GAP patients, but it is obviously not a key putative pathogen. The majority of data implicates P. intermedia as a common part of oral microflora in healthy persons. According to the literature an increase of its proportions is considered critical in etiopathogenesis of periodontal disease.

Since Eikenella corrodens has been suggested in earlier research as a pathogen because of its assotiation with initiation and progression of the disease in juvenile and refractory cases (49, 241, 263, 264), it was of interest to compare its distribution in GAP patients and the elderly. E. corrodens was more frequently identified in GAP patients than in elderly subjects (Fig.1, 3, 6). In the majority of the elderly only one sampled site was positive for E. corrodens (Fig.6). Interestingly, no significant difference was found between the number of colonized pockets and control sites in the diseased group (Fig.2). E. corrodens tends to colonize GAP patients, probably when the environmental conditions become favorable. It is questionable whether it plays a role in disease progression.
The results are consistent with the data of Kamma et al. who could not find significant qualitative and quantitative difference between colonization of deep pockets and shallow sites by E. corrodens in RPP patients either (15). Albandar et al. detected this species in 91% of EOP patients and in 89% of subjects without disease (230). Similar relations between chronic periodontitis and health has been observed frequently (11, 34, 36). The level of this bacterium seems to be independent of disease classification or the rate of progression.
In conclusion, E. corrodens is a commensal species that does not play a significant role as a primary opportunistic periodontal pathogen.

Veillonella parvula was an infrequently detected species in the GAP and elderly groups, with a prevalence of 25% and 43%, respectively (Fig. 1). No significant difference in detection frequency between the two groups (Fig. 1, 3) or between diseased and healthy sites in periodontitis patients could be demonstrated (Fig. 2, 5). This is in accord with the results of Kamma et al. that revealed a similar presence of this bacterium in shallow and deep sites (53% and 33%, respectively) of RPP patients, however, as comparison to our study with a much higher site-prevalence (15).
Studies of experimental gingivitis included V. parvula in the group of species which increased in proportion as inflammation developed (23). However, a majority of publications show only a weak association between periodontitis and this microorganism (39, 140). V. parvula together with E. corrodens and Capnocytophaga spp. was shown to be a common member of microbiota in subgingival sites of diseased and healthy subjects and was detected in similar proportions in both groups (36, 217). V. parvula has been detected more frequently in inactive sites when compared to active sites (39, 137, 140). Haffajee et al. showed the site-prevalence of V. parvula in the elderly (60%) being higher than in the younger healthy group (48%), however without significant difference (34). Low detection frequency of the species in our study raises the question of the accuracy of the performed identification. Low sensitivity of the oligonucleotide probe cannot be ruled out.
Results of our study suggest that V. parvula should not be regarded as an adequate marker of healthy flora.

The prevalence of Capnocytophaga ochracea in the GAP patients of the current study was distinctly lower (16%) than of the elderly (95%) (Fig. 1). As few as 3% of the pockets of the GAP subjects were positive, in contrast 55% of the sites of the elderly demonstrated colonization by C. ochracea (Fig. 3). The differences were highly significant. Even the shallow sites of GAP patients were rarely colonized, while 70% of the 1-3 mm sites of the elderly were positive (Fig. 4). Interestingly, patients with aggressive periodontitis exhibited seldom C. ochracea in subgingival plaque.
Frequent detection of C. ochracea in gingival sulci of children by Conrads et al. was interpreted as a physiological condition (201).
Increased levels of the species appeared to be consistent with a decreasing risk of new attachment loss. Similarly lower levels were found prior to breakdown (36, 137, 140, 265).
In several studies, contradictory to our results, C. ochracea tends to be a common colonizer in bacterial plaque regardless of whether the samples were obtained from supra-or subgingival plaque of diseased or healthy subjects, from active or inactive sites (15, 30, 34, 39). In some medical case reports C. ochracea has been made responsible for endocarditis and cervical abscess (266). Tanner et al. in earlier research has associated C. ochracea with juvenile periodontitis as high proportions of the species were recovered from diseased sites in young adults (241). Our data clearly confirm the association of C. ochracea with periodontal health. The use of this species may be a good indicator for periodontal health.

The comparison of the microflora of shallow sites from GAP patients with the flora from subjects with healthy periodontium can give valuable information as to whether microorganisms originate from the adjacent pockets as a consequence of a "spill over", or belong to the resident microflora. Both hypotheses are presented in literature (267). Riviere et al. tested the hypothesis that certain bacteria at healthy sites would be detected more frequently in subjects with periodontitis than in subjects without periodontitis (255). Using an immunological assay he could show statistically significant differences only for P. gingivalis and Treponema spp. The data reported by Ximenez-Fyvie et al. support Riviere's hypothesis (30). The authors obtained quantitative data using checkerboard hybridization with whole-genomic probes. Higher mean counts of periodontitis-associated bacteria were observed in shallow sites of periodontitis patients when compared to healthy subjects. Similarly Haffajee et al. using the same detection method showed that T. forsythensis, P. gingivalis, T. denticola and Selenomonas noxia were found more frequently, and at higher levels, in shallow pockets of periodontitis subjects than at similar sites in the healthy group (34). The data of the present study contradicts this hypothesis. We found no significant difference between the groups for any species, except for C. ochracea, in colonization of shallow sites (Fig. 4). The results suggest rather that the putative periodontopathogens belong to a resident flora. Thus, the mere presence of a putative pathogen has limited value as an adjunct to clinical diagnosis and treatment planning. However, the risk that the disease may occur at these sites is highly dependent on the host, as well as the variation in bacterial virulence.

"Profiles" of microbial complexes have been recognized upon clustering of the detected species from distinct clinical conditions (140, 177). A high degree of association between organisms may indicate a symbiotic relationship in periodontal pockets. A pathogen may more readily colonize subgingival sites already occupied by other organisms, due to gingival inflammation or growth factors produced by other organisms. However, some organisms may occur together in periodontitis lesions merely because they both induce destructive disease without interacting with each other. Putative pathogens acting together may produce additive or even synergistic damage to the periodontal tissues. Therefore a therapeutic regimen leading to concomitant suppression or elimination of symbiotic microorganims may achieve particularly great clinical benefits.
The definition of five subgingival plaque bacterial complexes by Socransky et al. (215) was based on the analysis of the microbial community of over 13,000 plaque samples from 185 subjects (EOP patients excluded) by using whole-genomic DNA probes in checkerboard hybridization assays. 5 clusters were formed:

1. Red cluster - P. gingivalis, T. forsythensis, T. denticola

2. Orange cluster - F. nucleatum, P. intermedia, P. nigrescens, Peptostreptococcus micros, Campylobacter spp., E. nodatum, S. constellatus

3. Green cluster - Capnocytophaga spp., Campylobacter concisus, Eikenella corrodens, A. actinomycetemcomitans serotype a

4. Yellow cluster - Streptococcus spp.

5. Purple cluster - Actinomyces odontolyticus, Veillonella parvula

The members of the red complex have frequently been detected together and exhibit a very strong correlation with pocket depth (215). The biological basis of the association among these species is not known. However, strong interspecies adherence has been demonstrated among these putative pathogens (268).
The co-existence of T. forsythensis and P. gingivalis has been frequently reported (11, 151, 177, 182, 192, 215, 223). Slots et al. (269) showed an odds ratio of 18.6. It has been speculated that T. forsythensis precedes the colonization by P. gingivalis, since T. forsythensis alone is detected more frequently (182, 191). The data of the present study are consistent with this report. P. gingivalis was detected alone in only 4 samples. The odds ratio of detecting these species together was 23.5. Since T. forsythensis and P. gingivalis were strongly associated with GAP, a co-infection with both microorganisms may lead to a particularly aggressive form of periodontal disease.

The red complex was shown to be closely associated with the orange cluster (215). The odds ratios of the present study show mutual relationships among the species belonging to these 2 complexes. T. forsythensis and C. rectus were most frequently detected together in periodontal pockets (OR 35.6). Colonization by these two bacterial species has often been associated with the induction of a shift from periodontal health to disease (17, 31, 33, 36, 39, 48, 144, 254). Similarly to our results Kamma et al. (170) observed the strongest positive association between T. forsythensis and C. rectus (OR 109.5).
The statistical analysis of Haffajee et al. revealed that combinations of species are better predictors of new attachment loss. Significantly higher levels of T. forsythensis and C. rectus, and significantly lower levels of C. ochracea were found in active subjects prior to breakdown (265).

T. forsythensis is often detected together with F. nucleatum (36, 223). In culture F. nucleatum enhances the growth of T. forsythensis (228). These species were frequently identified in the same pockets (OR 9.1) (Table 2). The species mentioned, together with C. rectus, have often been found in sites which exhibited active disease and those which responded poorly to therapy (140).
Mullally et al. demonstrated a strong association between P. intermedia and E. corrodens in EOP patients (14). This constellation however could not be confirmed in our study. Rather E. corrodens was associated with A. actinomycetemcomitans (OR 5.8) (Table 2). According to Socransky et al. they both belong to a green cluster (215). This pair has frequently been identified in lesions of LJP (33, 49).
Ashimoto et al. found positive associations between C. rectus and E. corrodens, as well between P. gingivalis and E. corrodens when investigating heterogeneous patient groups (11). Our results and those of Socransky et al. (177) indicated a negative association between these species. The evaluation of heterogeneous population groups and the use of different detection methods may explain these discrepancies to some extent.