Drivers and determinants of strain dynamics following fecal microbiota transplantation

Abstract

Fecal microbiota transplantation (FMT) is a therapeutic intervention for inflammatory diseases of the gastrointestinal tract, but its clinical mode of action and subsequent microbiome dynamics remain poorly understood. Here we analyzed metagenomes from 316 FMTs, sampled pre and post intervention, for the treatment of ten different disease indications. We quantified strain-level dynamics of 1,089 microbial species, complemented by 47,548 newly constructed metagenome-assembled genomes. Donor strain colonization and recipient strain resilience were mostly independent of clinical outcomes, but accurately predictable using LASSO-regularized regression models that accounted for host, microbiome and procedural variables. Recipient factors and donor-recipient complementarity, encompassing entire microbial communities to individual strains, were the main determinants of strain population dynamics, providing insights into the underlying processes that shape the post-FMT gut microbiome. Applying an ecology-based framework to our findings indicated parameters that may inform the development of more effective, targeted microbiome therapies in the future, and suggested how patient stratification can be used to enhance donor microbiota colonization or the displacement of recipient microbes in clinical practice.

Fig. 1. Study design and workflow overview.

a, We analyzed a dataset of 316 FMT time series across ten disease indications and 22 cohorts, totaling 1,492 fecal metagenomes. Species pangenomes were built from reference genomes and newly generated MAGs and profiled across samples for taxonomic, functional and strain population composition, based on microbial SNVs and differential gene content. b, Each allogenic FMT was represented as a triad of donor pre-FMT (blue hues), recipient pre-FMT (yellow) and post-FMT (purple) samples; each sample’s strain population is indicated as an overlapping circle. c, FMT strain-level outcomes for each species were scored using patterns of determinant SNVs and gene content (Supplementary Table 5). d, Ternary diagram of the strain population space for conspecific recipient strain persistence, donor strain colonization, donor–recipient coexistence and influx of novel strains.

Fig. 2. Community-wide FMT outcomes vary across patients and indications.

a, Microbiome-level outcomes of 228 scorable allogenic FMT time series, summarized across all strain populations observed in donor and recipient (rec.). Fractions are normalized to the number of species observed in the recipient post FMT. b, Contextual data on indication, procedure and clinical outcome for each FMT time series in a.

Fig. 3. Drivers and determinants of FMT community-level outcomes.

a, Ex ante predictability of microbial community-wide outcomes for individual FMTs (summarized across all trackable strain populations in a triad of donor, recipient pre-FMT and recipient post-FMT samples; Fig. 2) using cross-validated LASSO linear models with regularized subsets of different variable categories or a combination of all variables (‘full’ model) knowable before the intervention (Methods and Supplementary Table 6). Within each category, only the most relevant predictors are included. Predictive performance for each outcome index is shown as R² on the left, and variable importance and directionality for the most predictive factors as cross-validated LASSO coefficients on the right. b, Association of FMT outcomes with LASSO-regularized sets of post hoc variables (measured after the intervention).

Fig. 4. Strain-level FMT outcomes vary between species but are predictable ex ante.

a, Strain-level outcomes for selected species are shown for conspecific FMT triads—that is, time series where the focal species was present in both donor and recipient pre FMT. Outcomes are scored as recipient strain persistence (dominance by recipient strains, yellow), donor takeover (blue), donor–recipient coexistence (orange) or influx of novel or previously undetected strains (purple), as indicated in the schematic on the left. Each dot corresponds to one scored FMT. b, Stacked bars representing outcomes for each species across scorable FMTs, scaled to the number of FMTs where the species was observed in the recipient following the intervention. Dashed lines indicate averages for recipient strain persistence within taxonomic groups (x axis). Outcome frequencies across all species are summarized on the left. c, Frequency of colonization by donor or novel (previously undetected) strains per species, as subsets of the data in b. Averages per taxonomic group are represented by dotted lines. d, Prediction accuracies of LASSO models for different binarized FMT outcomes (indicated on the left; Methods) as AUROC, averaged across cross-validation folds per species.

Fig. 5. Drivers and determinants of FMT strain-level outcomes for individual species.

a, Logistic LASSO models were trained to predict FMT binarized outcomes (recipient resilience, yellow; recipient turnover, purple; donor takeover, blue) for n = 307 species across FMT time series, using different subsets of ex ante variables (knowable before the intervention). Each dot represents data for one species. Data are shown for full models (choosing from all available variables) and models trained on variable subsets categorized by type (procedural, community-level diversity and so on). Predictive performance of species models is shown as average AUROC across LASSO cross-validation folds in marginal box plots, ranging from 0.5 to 1.0; center line, median; box limits, upper and lower quartiles; whiskers, maxima/minima within 1.5× interquartile range from upper/lower quartiles. b, Variable importance across full models to predict takeover by donor strains. Each edge indicates the importance of a predictor variable (top row) when predicting donor takeover for a given species (bottom row). Dot size for predictors indicates summed variable importance across all species; dot size for species (bottom) indicates total number of relevant predictors. Edge color and width indicate direction and strength of the association, respectively. c, Variable importance for individual predictor categories, as subsets of the data in b.

Fig. 6. FMT strain-level outcomes are shaped by both neutral and adaptive processes.

Each of the tested variables used to predict FMT outcome can be linked to putative underlying ecological processes, as suggested previously. Factors are organized by scope (pertaining to the donor, recipient or donor–recipient complementarity, top) and resolution (host, community, species and strain level; left to right). Underlying ecological processes can be roughly ranked along the gradient, from neutral/stochastic to adaptive/selective; each process is illustrated with a toy example on the right. Circle size corresponds to average variable importance, calculated across all tested species from LASSO coefficients and overall model performance (less predictive models penalize variable importance). Recipient factors and, in particular, donor–recipient complementarity measures across all resolutions, were generally far more relevant to species-level outcome than donor factors. neg, negative; pos., positive; abd, abundance.