The centralized microbiota profiling involved extracting bacterial DNA, and then using polymerase chain reaction to amplify 16sRNA for sequencing and subsequent taxonomic identification.
“Samples can be segregated into clusters according to microbiota composition,” said Dr. Peled, a medical oncologist at MSKCC. The investigators used an algorithm called t-distributed stochastic neighbor embedding, or tSNE, to help detect patterns in microbiota composition and diversity before and throughout the HCT process. Visualizations using tSNE allow for two-dimensional representations of complicated associations and interrelatedness in data.
“Color-coded by diversity and time, we see that these early samples tend to be more diverse,” in the tSNE analyses, Dr. Peled said. The later clusters, he said, show evidence of lower diversity and injury.
Individual samples can also be coded in a way that shows clusters by abundance of various bacterial taxa, Dr. Peled said. “The early, diverse cluster tends to be dominated, or filled, by anaerobic commensals such as Firmicutes and Clostridia, which we and others have found are associated with good outcomes after transplant.”