Salivary microbiome profiles for different clinical phenotypes of pituitary adenomas by single-molecular long-read sequencingOriginal paper
What was studied?
This study profiled the salivary microbiome of patients with pituitary adenoma and healthy people, and across four clinical tumor phenotypes. Saliva is stable and easy to sample. It used single-molecule long-read, full-length 16S rRNA sequencing on the PacBio Sequel platform. This third-generation method resolves taxa to genus and species. Alpha and beta diversity, differential genera, random forest and ROC analyses, plus BugBase and PICRUSt2 functional predictions were performed.
Who was studied?
The cohort was 42 pituitary adenoma patients and 20 healthy individuals in Hefei, China, recruited between 2019 and 2021. This was a human study. The four phenotypes were ACTH-secreting (6), growth-hormone-secreting (9), prolactin-secreting (18), and nonfunctioning (9). Patients on recent antibiotics or with oral disease were excluded. Saliva was collected before surgery. Diagnoses followed the 2017 WHO pituitary adenoma classification.
What were the most important findings?
Salivary microbial diversity was higher in pituitary adenoma patients than in healthy individuals. Microbes from 82 genera were detected overall, with distinct genera differing between the two groups. Among the four tumor phenotypes, 55 genera were shared. Genera uniquely enriched numbered one in ACTH, five in GH, three in NF, and three in PRL groups. Single genera distinguished the ACTH, GH, and PRL subtypes from nonfunctioning tumors with ROC AUC values of 0.91, 0.93, and 0.92.
What are the greatest implications of this study?
Salivary microbiome patterns may offer noninvasive markers to help distinguish pituitary adenoma phenotypes. Anaerobic and Gram-positive microbes were more common in patients. The work supports an oral-gut-brain axis link to a neuroendocrine tumor. Functional predictions pointed to heightened microbial metabolism in patients. The sample was small due to strict inclusion criteria, and the findings are associations from predictive bioinformatics. Larger studies are needed, and causation is not established.