Outpatient-to-Inpatient Transition Causes Marked Dysbiosis in Allogeneic Hematopoietic Cell Transplantation Recipients

Features of gut dysbiosis such as low diversity and intestinal domination have been associated with worse outcomes after allogeneic hematopoietic cell transplantation (allo-HCT). Potential sources of insult to the microbiome vary with time. We hypothesized that outpatient-to-inpatient transition has a different impact on the microbiome than later events such as antibiotic pressure and cumulative conditioning toxicity.We performed 16S rRNA gene profiling of 167 stool samples collected thrice weekly from 20 allograft recipients (12 men, 8 women; median [range] age: 52 [22-68] years) between admission (“day 1”) and 14 days post-HCT. No exclusion criteria were used. Patients were admitted 7-8 days before HCT. None were receiving antibacterial antibiotics at admission. Patients received levofloxacin starting 1 day pre-HCT and cefepime upon neutropenic fever. Conditioning included TBI in all patients (50% myeloablative). The most common underlying diagnosis was acute leukemia (75%). Donors were: HLA-haploidentical relative (40%), HLA-matched sibling (35%), cord blood (20%), and matched unrelated donor (5%). Non-cord graft sources were bone marrow (50%) and peripheral blood (30%).With an estimated Good's coverage of 98.4±0.2%, the 3 most abundant genera (aggregate over time) were Bacteroides (mean relative abundance 27%), Clostridium XIVa (10%), and Parabacteroides (5%). Consistent with previous studies, the remaining 48% of taxa were rare operational taxonomic units (OTUs; each <5%) (Fig. 1A). Although the number of OTUs, or community richness alone, did not change with time (Fig. 1B), the Shannon's index, which also incorporates community evenness, declined markedly with time (Fig. 1C). Bray-Curtis dissimilarity (a measure of non-overlapping species) from baseline increased rapidly initially for ∼7 days, and then decelerated (Fig. 1D).A parsimonious explanation of these findings from a dense sampling protocol and using 3 community-level indices suggests two phases of microbiota injury during the first 3 weeks of hospitalization. In phase 1 (admission to HCT), characterized by outpatient-to-inpatient transition, a large fraction of the microbiome is replaced with new communities. In phase 2 (HCT to engraftment), characterized by antibiotic pressure and conditioning toxicity, microbiome turnover decelerates and the more successful taxa expand. These results identify week 1 of hospitalization as a potential window for microbiota therapeutics to prevent dysbiosis-related adverse outcomes. Features of gut dysbiosis such as low diversity and intestinal domination have been associated with worse outcomes after allogeneic hematopoietic cell transplantation (allo-HCT). Potential sources of insult to the microbiome vary with time. We hypothesized that outpatient-to-inpatient transition has a different impact on the microbiome than later events such as antibiotic pressure and cumulative conditioning toxicity. We performed 16S rRNA gene profiling of 167 stool samples collected thrice weekly from 20 allograft recipients (12 men, 8 women; median [range] age: 52 [22-68] years) between admission (“day 1”) and 14 days post-HCT. No exclusion criteria were used. Patients were admitted 7-8 days before HCT. None were receiving antibacterial antibiotics at admission. Patients received levofloxacin starting 1 day pre-HCT and cefepime upon neutropenic fever. Conditioning included TBI in all patients (50% myeloablative). The most common underlying diagnosis was acute leukemia (75%). Donors were: HLA-haploidentical relative (40%), HLA-matched sibling (35%), cord blood (20%), and matched unrelated donor (5%). Non-cord graft sources were bone marrow (50%) and peripheral blood (30%). With an estimated Good's coverage of 98.4±0.2%, the 3 most abundant genera (aggregate over time) were Bacteroides (mean relative abundance 27%), Clostridium XIVa (10%), and Parabacteroides (5%). Consistent with previous studies, the remaining 48% of taxa were rare operational taxonomic units (OTUs; each <5%) (Fig. 1A). Although the number of OTUs, or community richness alone, did not change with time (Fig. 1B), the Shannon's index, which also incorporates community evenness, declined markedly with time (Fig. 1C). Bray-Curtis dissimilarity (a measure of non-overlapping species) from baseline increased rapidly initially for ∼7 days, and then decelerated (Fig. 1D). A parsimonious explanation of these findings from a dense sampling protocol and using 3 community-level indices suggests two phases of microbiota injury during the first 3 weeks of hospitalization. In phase 1 (admission to HCT), characterized by outpatient-to-inpatient transition, a large fraction of the microbiome is replaced with new communities. In phase 2 (HCT to engraftment), characterized by antibiotic pressure and conditioning toxicity, microbiome turnover decelerates and the more successful taxa expand. These results identify week 1 of hospitalization as a potential window for microbiota therapeutics to prevent dysbiosis-related adverse outcomes. Figure 1.