From enrollment to rarefaction: cancer microbiome methods reporting deteriorates across the analysis pipeline

Reproducibility in microbiome research depends on transparent reporting of sample sizes and sequence counts both before and after quality control (QC) processes. This is an expectation formalized in the STORMS (Strengthening The Organization and Reporting of Microbiome Studies) checklist published in November 2021. Whether cancer microbiome studies comply with the STORMS items governing sample attrition and quality control, however, remains largely unexamined. Here, we audited 67 clinical cancer microbiome studies against STORMS items spanning sample size and quality-control reporting across the analytic pipeline, from participants through sequencing and bioinformatic processing. While all studies reported an initial participant count, reporting completeness declined at subsequent analytic stages. Participant exclusions were reported for 70% of studies at the participant stage, microbiome samples lost to post-sequencing QC for 25% of studies, post-extraction/PCR loss for 21% of studies, and just 6% for samples lost at rarefaction. Negative controls were reported in 34% of studies and positive controls in only 10%. Because sample loss, quality-control filtering, rarefaction, and contamination handling directly affect the end result of a cancer microbiome study, differences in how these are implemented can have critical implications for the field’s reproducibility as well as rigor at large. Resolving this issue will require buy-in from scientists, funding institutions, and journals to work together to enforce stricter standards for cancer microbiome methodology reporting.