P921 FMT in UC is associated with a decrease in Bacteroides-2 enterotype and response with baseline RNA and microbiota signatures

Abstract

Abstract Background The efficacy of faecal microbiota transplantation (FMT) in UC has been reported to be donor-, patient- and procedure-dependent (Rees et al., 2022). The RESTORE-UC trial [NCT03110289] aimed to improve the outcome of FMT in patients with active UC by donor preselection on microbiota level, a strict anaerobic preparation and repeated FMT administration. The trial was prematurely stopped for futility (Caenepeel et al., 2022). We investigated changes in resp. biopsy and faecal samples obtained host transcriptomics and microbiota profiles from baseline to primary endpoint (PE) at week 8 to understand reasons for the observed lack of efficacy. Methods Active UC patients (total Mayo score 4-10 with endoscopic sub-score ³2, n=72) were randomly allocated to receive 4 anaerobic-prepared superdonor (S) FMT or autologous (A) FMT. Primary endpoint was defined as steroid-free clinical remission (Total Mayo ≤ 2, with no sub-score >1). Host RNA extractions were performed from mucosal biopsies collected at week 0 (n=63) and 8 (n=54) using the Qiagen AllPrep DNA/RNA Mini kit, and sequenced using Illumina HiSeq4000. Sequencing data was further processed (read-count filtering, normalization) and further analyzed using DESeq2 package. Corresponding faecal samples (resp. n=62 and n=50) were submitted to DNA extractions using MagAttract PowerMicrobiome DNA/RNA kit on an automated extraction platform, followed by library prep and 16S rDNA-sequencing using Illumina MiSeq. The obtained sequences were subjected to the DADA2 pipeline in R and the Quantitative Microbial Composition (QMP) was quantified by flow cytometry. Results Eight responders were observed: 3 after S-FMT and 5 after A-FMT, as well as 58 non-responders considering the intention-to-treat population. Responders to FMT tended to cluster at baseline (adonis p=0.34) and PCA analysis on the top 500 mucosal genes with the highest variance (Fig.1), showed a significant host effect at week 8 between responders and non-responders (adonis p<0.05). Likewise, PCA analysis of the currently available QMP (Fig.2) showed a trend towards clustering by response at week 0 and 8 resp. adonis p=0.18 and p=0.11). An overall decrease in Bacteroides-2 enterotype prevalence was observed (Fig.3) over both treatment groups and independently from reaching the PE (All McNemar's p=0.077). Conclusion A trend towards clustering of responders on host mRNA level and QMP was observed at baseline as well as at the primary endpoint, showing a potential role for pre-FMT patient selection by phenotype. Moreover, the dysbiotic enterotype Bacteroides-2 seems to be decreased after FMT. However, further analyses are mandatory to identify specific predictors of response and the origin of the microbiota shift.