A Statistical Approach to the Ecology of Porphyromonas gingivalis

Abstract

Previous studies have suggested that infections with Porphyromonas gingivalis, associated with periodontal disease, may consist of one clonal type. It has also been shown that each individual patient carries a unique clonal type of P. gingivalis, as assessed by DNA fingerprinting. This issue was further examined by random collection of multiple isolates ofP. gingivalis from multiple sites in several patients, and characterization of these isolates by DNA fingerprinting. Although most patients appeared to be infected exclusively by one clonal type of P. gingivalis, at least one patient was found to harbor two distinct clonal types. This indicates that the simultaneous presence of different clonal types of P. gingivalis can occur. A statistical method was developed for retrospective analysis of these data for estimation of whether the remainder of these patients were actually infected with single or multiple clonal types ofP. gingivalis. With this statistical method, a confidence interval was calculated for estimation of the true proportion of a single observed clonal type in the potential population of P. gingivalis that might be recovered from an infected patient. Statistically, the sampling of small numbers of sites per patient or isolates per site leads to a wide confidence interval for the estimated true proportion of the observed clonal type in the infecting P. gingivalis population. For example, when five sites in an oral cavity yield indistinguishable P. gingivalis isolates, then the true proportion of this clonal type in the total P. gingivalis population in the infected oral cavity is estimated to be in the interval between 55% and 100% (at a 95% confidence level). If the isolates from all sampled sites from a periodontal patient appear to be of the same clonal type, it can be calculated that at least 29 different sites must be sampled for this observed clonal type to represent at least 90% of the P. gingivalis population in this patient (95% confidence level); this was termed the LTP, or lowest true proportion. Such calculations can also be used for estimation of the LTP of a clonal type infecting a single oral site in a periodontal patient. These results confirm the commonly held opinion that studies investigating the ecology of P. gingivalis and other suspected pathogens must be designed on a large scale in order for statistically meaningful results to be obtained.