Canine Tooth Microbiome Gingival Index: a new microbiome-derived measure of gingival health validated by nutritional intervention.

INTRODUCTION: Periodontitis affects over 80% of dogs over 3 years of age, progressing irreversibly from gingivitis due to an imbalance in the subgingival microbial community that triggers an immune response. Early diagnosis of gingivitis is challenging, often relying on visible redness or bleeding noticed by pet owners or professionals. Therefore, an easy-to-interpret, clinically relevant, and responsive measurement tool based on the subgingival microbiome is needed to facilitate early diagnosis of oral health issues. We developed the Canine Tooth Microbiome Gingival Index (CTMGI), a single-score metric derived from subgingival plaque microbiome data and machine learning models, and validated its responsiveness via nutritional intervention.

METHODS: We collected subgingival plaque microbiome profiles from 692 tooth samples of 347 dogs, generated through 16S amplicon sequencing. For the machine learning models, the tooth gingivitis scores were dichotomized into healthy (gingivitis score <3) and unhealthy (gingivitis score ≥3), along with other clinical scores such as tooth recession, pocket depth, and attachment loss. The raw data were split into training and test sets, and five distinct machine learning models were employed to identify features that distinguish healthy from gingivitis sites.

RESULTS: The two top-performing models-random forest and logistic regression-yielded 22 unique features. These 22 features included the sum of early and late colonizers, the phyla actinobacteria and proteobacteria, and other bacterial species. The CTMGI was derived from the 22 features, categorized as "positive" or "negative" based on their influence on gingivitis. The CTMGI classification cutoff score was set at -0.12 with a Receiver Operating Characteristics-Area Under the Curve (ROC-AUC) of 0.761, a sensitivity of 0.701, and a specificity of 0.752. A score greater than -0.12 was found to indicate a "Работает" gingival condition; otherwise, it indicated "unhealthy." Furthermore, we conducted a nutritional intervention study to validate the responsiveness of the CTMGI, in which the test food, which has documented oral health benefits, resulted in a significantly higher CTMGI score (1.32) compared to the control food, which offers no oral health benefits (0.66).

DISCUSSION: Overall, this study developed and validated a quantitative, single-score, microbiome-based metric that is clinically translatable for the assessment of early-stage canine gingival health. Furthermore, its demonstrated responsiveness to nutritional intervention suggests that this index can serve as a prognostic measure.