DOI: 10.1128/msystems.00049-23 ISSN:

A new hypothesis on BV etiology: dichotomous and crisscrossing categorization of complex versus simple on healthy versus BV vaginal microbiomes

Zhanshan (Sam) Ma
  • Computer Science Applications
  • Genetics
  • Molecular Biology
  • Modeling and Simulation
  • Ecology, Evolution, Behavior and Systematics
  • Biochemistry
  • Physiology
  • Microbiology

ABSTRACT

It has been estimated that bacterial vaginosis (BV) influences as many as one-third of women, but its etiology remains unclear. A traditionally held view is that dominance by Lactobacillus is the hallmark of a healthy vaginal microbiome (VM) and lack of dominance may make women BV-prone. A more recent characterization is that the human VMs can be classified into five major types, four of which possess type-specific dominant species of Lactobacillus . The remaining one (type IV) is not dominated by Lactobacillus and contains a handful of strictly anaerobic bacteria. Nevertheless, exceptions to the first hypothesis have been noticed from the very beginning, and there is not a definite relationship, suggested yet, between the five VM types and BV status. Here, we propose a novel hypothesis that assumes the existence of four VM types from dichotomous crisscrossing of “complex versus simple (high-diversity or low-dominance versus low-diversity or high-dominance)” on “healthy versus BV” (the four essential dimensions of VMs). We comprehensively test the hypothesis with 7,958 VM samples by demonstrating the statistically significant differences between the four VM types in community diversity, composition, dominance, heterogeneity, and stochasticity, and by identifying unique/enriched species in each VM type. We further verified the categorization (hypothesis) by using six machine learning (ML) algorithms, including deep learning artificial intelligence (AI) technology, to reclassify the randomly mixed VM samples into four respective types and achieved 85%–100% classification precisions. Our hypothesis provides a foundation for further investigating the etiology and automatic diagnosis of BV based on inexpensive amplicon sequencing and AI technology.

IMPORTANCE

BV may influence as many as one-third of women, but its etiology remains unclear. A traditional view is that dominance by Lactobacillus is the hallmark of a healthy vaginal microbiome and lack of dominance may make women BV-prone. Recent studies show that the human VMs can be classified into five major types, four of which possess type-specific dominant species of Lactobacillus . The remaining one (type IV) is not dominated by Lactobacillus and contains a handful of strictly anaerobic bacteria. Nevertheless, exceptions to the first hypothesis have been noticed from the very beginning, and there is not a definite relationship, suggested yet, between the five VM types and BV status. Here, we propose and test a novel hypothesis that assumes the existence of four VM types from dichotomous crisscrossing of “complex versus simple (high diversity or low dominance versus low diversity or high dominance)” on “healthy versus BV.” Consequently, there are simple BV versus complex BV.

More from our Archive