Multicenter Microbiota Analysis Indicates That Pre-HCT Microbiota Injury Is Prevalent across Geography and Predicts Poor Overall Survival

Intestinal microbiota composition is associated with important outcomes after allo-HCT including survival, relapse, GVHD, and infections. These observations have been made almost exclusively by characterizing the microbiota in the first weeks after transplantation, and in single-center studies. We previously reported that intestinal diversity measured peri-neutrophil engraftment is predictive of overall survival in a multicenter cohort. Here, we hypothesized that pre-HCT microbiota configuration may also be an important determinant of post-transplantation outcomes. We report a multicenter analysis conducted at 4 independent international institutions to test this hypothesis.

We collected 1922 stool samples ~weekly from 991 unique allo-HCT patients at four international transplant centers: Cohorts 1 and 2 in the US, cohort 3 in Europe, and cohort 4 in Japan. The patients-all adult recipients of allo-HCT-varied in underlying diagnosis, donor-graft sources, conditioning intensity, and GVHD prophylaxis. Samples from all 4 centers were sequenced and analyzed at a central laboratory using the V4-V5 region of 16S rRNA. For patients with multiple samples within the pre-HCT sampling period, which we defined as day -20 to 0, median values were used.

On average, patients from all 4 transplantation centers had reduced microbiota diversity pre-HCT, as measured by median α-diversity (inverse Simpson) values that were 1.7-to-2.5-fold lower than those of healthy volunteers. This comparison was made both in volunteers whose samples we sequenced ourselves and in a publicly available dataset (Fig A, p<0.005). Since α-diversity measurements do not consider which taxa are present in a community, we asked whether the composition of pre-HCT microbiotas are similar to healthy microbial communities. While the intestinal communities of most healthy volunteers could be matched to the Enterotypes classifier, pre-HCT samples from all four centers had configurations that were poorly characterized by this independent classification scheme (Fig B). Thus, the post-HCT microbiota injuries that we previously observed comparably across geography are preceded by community structures that are already abnormal pre-HCT, consistent with our prior observation that pre-HCT antibiotic exposure is a risk factor for poor outcomes.

We next asked how similar these pre-HCT communities are across geography. We found that Bray-Curtis distances between institutions were reproducibly much smaller in magnitude than the changes observed over time during transplantation (Fig C, p<0.005). We also observed in the largest cohort (#1) that pre-HCT diversity is associated with patient survival. Among 753 patients, those in the lowest quartile of pre-HCT α-diversity had a lower overall survival than those in the highest quartile (Fig D, p<0.009).

In order to characterize these clinically relevant low-diversity phenotypes, we defined domination as a microbiota injury in which any operational taxonomic unit comprises >30% of bacterial abundance. The dominating taxa belonged to multiple genera, the most common being Enterococcus, Streptococcus, Lactobacillus, Escherichia, and Klebsiella (Fig E) as annotated by Greengenes and NCBI databases. At all four institutions, the cumulative incidence of intestinal domination by any organism was >50% by day 0 and was >87% by day +28 (Fig F). We performed additional analyses in the largest cohort and found that the low-diversity state is associated with exposure to broad-spectrum antibiotics, conditioning intensity, and low calorie intake.

We demonstrate that HCT patients at 4 institutions on 3 continents presented with microbiota configurations that were similar to one another and distinct from those of healthy individuals. Severe microbiota injury as revealed by domination is a common event whose development begins before allograft infusion, and pre-HCT microbiota injury predicts poor overall survival. These observations suggest the pre-HCT period as a window of opportunity to (a) assess microbiota injury as part of comorbidity evaluation, (b) inform selection of antibiotic prophylaxis, gut-decontamination, GVHD-prophylaxis, or conditioning regimens, and (c) intervene with microbiota injury-remediation or prevention strategies.

Figure.

View largeDownload slide

Figure.

View largeDownload slide

Close modal