Microbiome alteration via fecal microbiota transplantation is effective for refractory immune checkpoint inhibitor-induced colitis

Abstract

Immune checkpoint inhibitors (ICIs) target advanced malignancies with high efficacy but also predispose patients to immune-related adverse events like immune-mediated colitis (IMC). Given the association between gut bacteria with response to ICI therapy and subsequent IMC, fecal microbiota transplantation (FMT) represents a feasible way to manipulate microbial composition in patients, with a potential benefit for IMC. Here, we present a large case series of 12 patients with refractory IMC who underwent FMT from healthy donors as salvage therapy. All 12 patients had grade 3 or 4 ICI-related diarrhea or colitis that failed to respond to standard first-line (corticosteroids) and second-line immunosuppression (infliximab or vedolizumab). Ten patients (83%) achieved symptom improvement after FMT, and three patients (25%) required repeat FMT, two of whom had no subsequent response. At the end of the study, 92% achieved IMC clinical remission. 16S rRNA sequencing of patient stool samples revealed that compositional differences between FMT donors and patients with IMC before FMT were associated with a complete response after FMT. Comparison of pre- and post-FMT stool samples in patients with complete responses showed significant increases in alpha diversity and increases in the abundances of Collinsella and Bifidobacterium, which were depleted in FMT responders before FMT. Histologically evaluable complete response patients also had decreases in select immune cells , including CD8⁺ T cells, in the colon after FMT when compared with non-complete response patients (n = 4). This study validates FMT as an effective treatment strategy for IMC and gives insights into the microbial signatures that may play a critical role in FMT response.

Fig. 1:. Comparisons of alpha and beta diversity of baseline pre-FMT samples from IMC patients treated with FMT and healthy individuals.

(A) Alpha diversity by inverse Simpson index on samples from healthy individuals (FMT donors), pre-FMT samples from Complete Response (CR), and pre-FMT samples from non-Complete Response (non-CR) patients, compared using Mann-Whitney test. (B) Principle coordinates analysis (PCoA) of FMT donors (n=11), CR (n=6), and non-CR (n=3) patient samples; significance was assessed with PERMANOVA statistical testing. Axis percentage labels represent variance explained. (C) Differentially abundant bacterial genera in samples collected pre-FMT from CR compared with non-CR using DESeq2 with adjustment for multiple comparisons using the Benjamini-Hochberg method. (D) Differentially abundant bacterial genera in samples collected pre-FMT CR compared with samples from healthy FMT donors, analyzed as in (C).

Fig. 2:. Changes in beta diversity predict IMC response to FMT.

(A) Principle coordinates analysis (PCoA) of samples from Complete Response (CR; blue), and non-Complete Response (non-CR; orange) patients are shown. Pre-FMT (square), post-FMT (Day 30 +/− 20; circle), and corresponding FMT donor sample (star) for each patient are indicated in the figure. Solid lines indicate linked pre- and post-FMT samples, while dashed lines indicate linked post-FMT and FMT donor samples. Axis percentage labels represent variance explained. (B) Beta diversity distances were quantified using weighted UniFrac values comparing pre- and post-FMT samples against FMT donor samples in CR and non-CR IMC patients. Data are shown using both bar and paired dot plots. Comparisons were analyzed using Mann-Whitney statistical test.

Fig. 3:. Increases in alpha diversity and specific bacterial taxa are associated with FMT response.

(A) Comparison of alpha diversity by inverse Simpson index pre-FMT (baseline) and post-FMT (Day 30 +/− 20) in Complete Response (CR) and non-Complete Response (non-CR) IMC patients using paired Wilcoxon statistical test. (B) Volcano plot of differentially abundant bacterial genera comparing pre-FMT and post-FMT samples from CR patients using paired Wilcoxon statistical test with adjustments for multiple comparisons (Benjamini-Hochberg method). (C) Relative abundance of significantly changed bacterial genera identified in (B) quantified in paired samples for CR and non-CR patients.

Fig. 4:. Associations of select taxa with pre-FMT clinical features of IMC patients.

(A) Spearman’s rank correlation of clinical variables with bacterial taxa pre-FMT. (B) Binary clinical associations using DESeq2. Log2fold change represents effect size and directionality is indicated in parentheses next to each clinical feature (red/blue). Previously identified taxa of interest (Fig.3) are present at the bottom of each heatmap based on response to FMT therapy (CR, green; non-CR, orange). All available clinical features were considered (see Tables 1 and 2). Significance was assessed using calculated values normalized to each sample mean and scaled for visualization using heatmaps generated in R.

Fig. 5:. Immune infiltrate changes in FMT treated IMC patients.

Multiplex immunofluorescence quantification of (A) total CD3+ and (B) CD3+ CD8+ T cells. Total area of colonic epithelium and stroma of IMC patients were used to quantify changes in immune cells. Individual patients are depicted using shapes. (C) Representative immunofluorescent images depicting CD3+ & DAPI, CD8+ & DAPI, and merged for IMC patients treated with FMT. Scale bars represent 50 um. P- values not reported due to small sample size.