Longitudinal profiling of putative SCFA-associated gut taxa and sodium butyrate epithelial assays in acute myeloid leukemia

Introduction Intensive therapy for acute myeloid leukemia (AML) disrupts intestinal homeostasis, but the relationship between microbial composition and epithelial injury remains uncertain. Methods We analyzed 508 longitudinal stool samples from 67 patients using patient-clustered models, conducted an exploratory microbiome Mendelian randomization (MR) screen and host-gene annotation, and evaluated sodium butyrate (NaB) in chemotherapy-injured epithelial and inflammatory cell assays. Results None of 1,521 MR tests passed false-discovery-rate correction. Host annotation identified a 17-gene candidate module. Putative short-chain fatty acid (SCFA)-associated taxa had lower relative abundance on days 15-28 than before chemotherapy. Patients with diarrhea had lower full-follow-up minimum abundance of these taxa (Cliff's delta = -0.518; 95% CI, -0.738 to -0.268; P = 1.20 x 10 -4 ; q = 0.00180). C. difficile positivity was associated with greater maximum single-genus pathobiont relative abundance ( q = 0.0261), but this association did not survive a fixed-window median sensitivity analysis. Period estimates were similar after adjustment, while small pre-event matched analyses did not show temporal precedence. NaB partially improved Caco-2 barrier readouts and was associated with lower THP-1 p-SYK/SYK and Caco-2 p-NF-kappaB/NF-kappaB ratios. Discussion Longitudinal taxonomic associations and NaB-responsive cell phenotypes provide complementary evidence. They do not establish butyrate depletion, microbial causality, receptor dependency, a continuous host pathway, or clinical efficacy.