Failure of gastrointestinal acute graft-versus-host disease (GI-aGvHD) to respond to steroid therapy is associated with limited further therapeutic options. We aimed to assess the safety and efficacy of the first-in-human use of the pooled allogeneic faecal microbiota, MaaT013, for the treatment of steroid-refractory GI-aGvHD. This prospective, international, single-arm, phase 2a study reports clinical outcomes from a 24-patient cohort with grade III-IV, steroid refractory GI-aGvHD treated with the pooled allogeneic faecal microbiota MaaT013. MaaT013 involved pooling faecal matter from 3 to 8 screened donors then transplanting the pooled batches into patients to treat GI-aGVHD. The 24 patients were treated in the HERACLES study (Aug 2018 to Nov 2020) at 26 sites in Europe and an additional 52 patients were treated in a compassionate use/expanded access program (EAP) in France (July 2018 to April 2021). The primary endpoint was GI response at day 28, defined as the proportion of patients with GI-aGvHD who had a complete response (CR) or very good partial response (VGPR). GvHD grading and staging were assessed according to the revised Glucksberg criteria. Adverse events and severe adverse events were monitored for 6 months and 12 months, respectively. The HERACLES study was registered with ClinicalTrials.gov (NCT03359980). Compared with single donors, MaaT013 is characterised by higher microbial richness and reduced variability across batches. At day 28 (D28), the GI-overall response rate (ORR) was 38% in the prospective population, including 5 complete responses (CR), 2 very good partial responses (VGPR) and 2 partial responses (PR). In the EAP, the GI-ORR was 58% (17 CR, 9 VGPR and 4PR). The 12-month overall survival (OS) was 25% in the prospective study and 38% in the EAP. Regarding safety, five infectious complications, including 3 sepsis, could not be excluded from being related to the study procedure in HERACLES. Shotgun sequencing analyses of the identified strains suggest that none were found in MaaT013. In the EAP, 18 pharmacovigilance cases were reported among 52 treated patients, including 11 bacteraemia/sepsis. In HERACLES, we observed in stools from responding patients at D28 a higher microbiota richness and increased levels of beneficial bacteria, in particular butyrate producers, along with increased levels of short-chain fatty acid and bile acids. In contrast, stools from non-responding (NR) patients displayed increased levels of pathogenic pro-inflammatory bacteria along with increased systemic inflammatory parameters. Overall, MaaT013 was safe in this population of highly immunocompromised patients and was associated with responses in some patients with GI-aGvHD and deserves further investigation. MaaT Pharma.