Persistent microbiome dysbiosis following hematopoietic cell transplantation: Gut GvHD as the most severe manifestation of microbial disruption

Abstract

Abstract Background Hematopoietic cell transplantation conditioning regimens severely disrupt the gut microbiome, potentially contributing to transplant-related complications including graft-versus-host disease (GvHD). Gut GvHD, targeting the primary site of microbial residence, may be particularly linked to microbiome disruption. While previous studies suggested gradual microbiome recovery, comprehensive longitudinal analyses with clinical correlations, remain limited. Methods We analyzed 385 fecal samples from 83 transplant patients and 74 healthy controls using 16S rRNA gene sequencing. Samples were collected at pre-transplant, day 0, 15, 30, and 100 post-transplant time points. We calculated inverse Simpson diversity (primary endpoint), performed beta diversity analysis using Bray-Curtis distances, identified differentially abundant taxa, and used mixed-effects models to analyze individual recovery trajectories. Clinical outcomes including overall GvHD and gut-specific GvHD were correlated with microbiome metrics using three-way comparisons (gut GvHD vs other GvHD vs no GvHD). Results Transplant patients showed profound gut microbiome disruption with a 33.6-fold reduction in diversity compared to controls (median inverse Simpson: 1.91×10⁻⁵ vs 6.41×10⁻⁴, p<2.22×10⁻¹⁶). We observed no significant recovery over 100 days, with diversity remaining at only 3% of control levels across all time points (Spearman rho=0.016, p=0.77). Gut GvHD emerged as the most severe form of microbiome disruption. Among 42 total GvHD patients, 14 developed gut GvHD and demonstrated distinctly worse microbiome profiles than both other GvHD patients and non-GvHD patients (three-way Kruskal-Wallis p=5.62×10⁻⁶). Gut GvHD patients had the lowest median diversity (9.21×10⁻⁶), representing 2.2-fold lower diversity than non-GvHD patients (p=6.18×10⁻⁶) and significantly worse than other GvHD types (p=6.07×10⁻⁴). Overall GvHD severity correlated with progressive diversity loss (rho=-0.175, p=0.002). Beta diversity analysis revealed complete separation between controls and transplant samples (PERMANOVA p=0.001) with no community recovery over time. Mixed-effects modelling demonstrated significant individual variation in recovery trajectories (p<2.2×10⁻¹⁶), identifying 77 patients with heterogeneous patterns. Taxonomic analysis revealed loss of 59 beneficial taxa (primarily Clostridiales, Ruminococcaceae, and Lachnospiraceae families) with only 2 taxa increased in transplant patients. Risk stratification identified early predictors of poor recovery, with gut GvHD patients consistently in the highest risk categories. ConclusionsTransplant conditioning causes profound microbiome dysbiosis without spontaneous recovery over 100 days, contradicting previous findings of gradual restoration. Gut GvHD represents the most severe manifestation of microbiome disruption, suggesting an intimate pathophysiologocal connection between gut microbial communities and localized GvHD development. These findings challenge current supportive care approaches and highlight the urgent need for gut GvHD-specific microbiome interventions, potentially including fecal microbiota transplantation.