On exploring cross-sectional stability and persistence of microbiome in a multiple body site colorectal cancer dataset

There are several ways to recover signature microbiome of a disease pathology. One way is to look at the core microbiome, which comprises microbial species prevalent across majority of the samples. At a finer level, certain subcommunities may exhibit stable signature across the sampling space. There can also be similarity of differential patterns across different body sites. In view of above, and leveraging recent advancements in analytical strategies, we revisit a multi-factorial Iranian ColoRectal Cancer (CRC) dataset, and explore stable and persistent patterns in the microbiome. For this purpose, 16S rRNA gene is amplified from saliva and stool samples of CRC patients using healthy controls as a baseline (<i>n</i> = 80). The dataset is supplemented with demographical and nutritional data of the study participants that were collected through filled questionnaire. Our results indicate that certain microbial species i.e., <i>Actinobacteriota</i>, <i>Bifidobacterium</i>, <i>Prevotella</i> and <i>Fusobacterium</i> are consistently present in the CRC patients suggesting their potential as diagnostic biomarkers of disease. Additionally, we identified a group of microbes such as <i>Akkermansia</i>, <i>Selenomonas</i>, <i>Clostridia_UCG-014</i>, <i>Lautropia</i>, <i>Granulicatella</i>, <i>Bifidobacterium</i>, and <i>Gemella</i> that exhibit similar differential response across body sites irrespective of where they are found, whether in saliva or stool samples. This suggest that a part of saliva microbiome can act as a proxy for stool microbiome giving further credence to oral-gut axis. Overall, our findings underscore the importance of exploring stable microbial biomarkers in multifactorial CRC dataset by marginalizing out variabilities, with the potential for improved diagnosis and treatment strategies.