Rearrangements involving 11q23.3/KMT2A in adult AML: mutational landscape and prognostic implications - a HARMONY study.

Balanced rearrangements involving the KMT2A gene (KMT2Ar) are recurrent genetic abnormalities in acute myeloid leukemia (AML), but there is lack of consensus regarding the prognostic impact of different fusion partners. Moreover, prognostic implications of gene mutations co-occurring with KMT2Ar are not established. From the HARMONY AML database 205 KMT2Ar adult patients were selected, 185 of whom had mutational information by a panel-based next-generation sequencing analysis. Overall survival (OS) was similar across the different translocations, including t(9;11)(p21.3;q23.3)/KMT2A::MLLT3 (p = 0.756). However, independent prognostic factors for OS in intensively treated patients were age >60 years (HR 2.1, p = 0.001), secondary AML (HR 2.2, p = 0.043), DNMT3A-mut (HR 2.1, p = 0.047) and KRAS-mut (HR 2.0, p = 0.005). In the subset of patients with de novo AML < 60 years, KRAS and TP53 were the prognostically most relevant mutated genes, as patients with a mutation of any of those two genes had a lower complete remission rate (50% vs 86%, p < 0.001) and inferior OS (median 7 vs 30 months, p < 0.001). Allogeneic hematopoietic stem cell transplantation in first complete remission was able to improve OS (p = 0.003). Our study highlights the importance of the mutational patterns in adult KMT2Ar AML and provides new insights into more accurate prognostic stratification of these patients.

Covering Hierarchical Dirichlet Mixture Models on binary data to enhance genomic stratifications in onco-hematology.

Onco-hematological studies are increasingly adopting statistical mixture models to support the advancement of the genomically-driven classification systems for blood cancer. Targeting enhanced patients stratification based on the sole role of molecular biology attracted much interest and contributes to bring personalized medicine closer to reality. In onco-hematology, Hierarchical Dirichlet Mixture Models (HDMM) have become one of the preferred method to cluster the genomics data, that include the presence or absence of gene mutations and cytogenetics anomalies, into components. This work unfolds the standard workflow used in onco-hematology to improve patient stratification and proposes alternative approaches to characterize the components and to assign patient to them, as they are crucial tasks usually supported by a priori clinical knowledge. We propose (a) to compute the parameters of the multinomial components of the HDMM or (b) to estimate the parameters of the HDMM components as if they were Multivariate Fisher's Non-Central Hypergeometric (MFNCH) distributions. Then, our approach to perform patients assignments to the HDMM components is designed to essentially determine for each patient its most likely component. We show on simulated data that the patients assignment using the MFNCH-based approach can be superior, if not comparable, to using the multinomial-based approach. Lastly, we illustrate on real Acute Myeloid Leukemia data how the utilization of MFNCH-based approach emerges as a good trade-off between the rigorous multinomial-based characterization of the HDMM components and the common refinement of them based on a priori clinical knowledge.

A dynamic time-to-event model for prediction of acute graft-versus-host disease in patients after allogeneic hematopoietic stem cell transplantation.

BACKGROUND: Acute graft-versus-host disease (aGvHD) is a major cause of death for patients following allogeneic hematopoietic stem cell transplantation (HSCT). Effective management of moderate to severe aGvHD remains challenging despite recent advances in HSCT, emphasizing the importance of prophylaxis and risk factor identification. METHODS: In this study, we analyzed data from 1479 adults who underwent HSCT between 2005 and 2017 to investigate the effects of aGvHD prophylaxis and time-dependent risk factors on the development of grades II-IV aGvHD within 100 days post-HSCT. RESULTS: Using a dynamic longitudinal time-to-event model, we observed a non-monotonic baseline hazard overtime with a low hazard during the first few days and a maximum hazard at day 17, described by Bateman function with a mean transit time of approximately 11 days. Multivariable analysis revealed significant time-dependent effects of white blood cell counts and cyclosporine A exposure as well as static effects of female donors for male recipients, patients with matched related donors, conditioning regimen consisting of fludarabine plus total body irradiation, and patient age in recipients of grafts from related donors on the risk to develop grades II-IV aGvHD. Additionally, we found that higher cumulative hazard on day 7 after allo-HSCT are associated with an increased incidence of grades II-IV aGvHD within 100 days indicating that an individual assessment of the cumulative hazard on day 7 could potentially serve as valuable predictor for later grades II-IV aGvHD development. Using the final model, stochastic simulations were performed to explore covariate effects on the cumulative incidence over time and to estimate risk ratios. CONCLUSION: Overall, the presented model showed good descriptive and predictive performance and provides valuable insights into the interplay of multiple static and time-dependent risk factors for the prediction of aGvHD.

Data-driven grading of acute graft-versus-host disease.

Despite advances in allogeneic hematopoietic cell transplantation, acute graft-versus-host disease (aGVHD) remains its leading complication, yet with heterogeneous outcomes. Here, we analyzed aGVHD phenotypes and clinical classifications in depth in large, multicenter cohorts involving 3019 patients and addressed prevailing gaps by developing data-driven models. We compared, tested and verified these along with all conventional classifications in independent cohorts and found that data-driven grading outperformed conventional grading in Akaike information criterion and concordance index metrics. Data-driven classifications refined aGVHD assessment with up to 12 severity grades, which were associated with distinct nonrelapse mortality (NRM) and confirmed the key role of intestinal aGVHD. We developed an online calculator for physicians to implement principal component-derived grading (PC1). These results provide substantial insight into the evaluation of aGVHD phenotypes and multiorgan involvement, which relegates the exclusive reporting of overall aGVHD severity grades in transplant registries and clinical trials. Data-driven aGVHD grading provides an expandable platform to refine classification and transplant risk assessment.

Optimizing platelet transfusion through a personalized deep learning risk assessment system for demand management.

ABSTRACT: Platelet demand management (PDM) is a resource-consuming task for physicians and transfusion managers of large hospitals. Inpatient numbers and institutional standards play significant roles in PDM. However, reliance on these factors alone commonly results in platelet shortages. Using data from multiple sources, we developed, validated, tested, and implemented a patient-specific approach to support PDM that uses a deep learning-based risk score to forecast platelet transfusions for each hospitalized patient in the next 24 hours. The models were developed using retrospective electronic health record data of 34 809 patients treated between 2017 and 2022. Static and time-dependent features included demographics, diagnoses, procedures, blood counts, past transfusions, hematotoxic medications, and hospitalization duration. Using an expanding window approach, we created a training and live-prediction pipeline with a 30-day input and 24-hour forecast. Hyperparameter tuning determined the best validation area under the precision-recall curve (AUC-PR) score for long short-term memory deep learning models, which were then tested on independent data sets from the same hospital. The model tailored for hematology and oncology patients exhibited the best performance (AUC-PR, 0.84; area under the receiver operating characteristic curve [ROC-AUC], 0.98), followed by a multispecialty model covering all other patients (AUC-PR, 0.73). The model specific to cardiothoracic surgery had the lowest performance (AUC-PR, 0.42), likely because of unexpected intrasurgery bleedings. To our knowledge, this is the first deep learning-based platelet transfusion predictor enabling individualized 24-hour risk assessments at high AUC-PR. Implemented as a decision-support system, deep-learning forecasts might improve patient care by detecting platelet demand earlier and preventing critical transfusion shortages.

Time series clustering of T cell subsets dissects heterogeneity in immune reconstitution and clinical outcomes among MUD-HCT patients receiving ATG or PTCy.

Introduction: Anti-T-lymphocyte globulin (ATG) or post-transplant cyclophosphamide (PTCy) prevent graft-versus-host disease (GVHD) after hematopoietic cell transplantation (HCT), yet individual patients benefit differentially. Methods: Given the sparse comparative data on the impact of cellular immune reconstitution in this setting, we studied flow cytometry and clinical outcomes in 339 recipients of 10/10 matched-unrelated donor (MUD) HCT using either ATG (n=304) or PTCy (n=35) for in vivo T cell manipulation along with a haploidentical PTCy control cohort (n=45). Longitudinal cellular immune reconstitution data were analyzed conventionally and with a data science approach using clustering with dynamic time warping to determine the similarity between time-series of T cell subsets. Results: Consistent with published studies, no significant differences in clinical outcomes were observed at the cohort level between MUD-ATG and MUD-PTCy. However, cellular reconstitution revealed preferences for distinct T cell subpopulations associating with GVHD protection in each setting. Starting early after HCT, MUD-PTCy patients had higher regulatory T cell levels after HCT (p <0.0001), while MUD-ATG patients presented with higher levels of γδ T- or NKT cells (both p <0.0001). Time-series clustering further dissected the patient population's heterogeneity revealing distinct immune reconstitution clusters. Importantly, it identified phenotypes that reproducibly associated with impaired clinical outcomes within the same in vivo T cell manipulation platform. Exemplarily, patients with lower activated- and αß T cell counts had significantly higher NRM (p=0.032) and relapse rates (p =0.01). Discussion: The improved understanding of the heterogeneity of cellular reconstitution in MUD patients with T cell manipulation both at the cohort and individual level may support clinicians in managing HCT complications.

Impact of chronic graft-versus-host-disease on intensive care outcome in allogeneic hematopoietic stem cell recipients.

Chronic graft-vs-host-disease (cGvHD) is the most relevant long-term complication after allogeneic stem cell transplantation (HSCT) with major impact on non-relapse mortality, but data on intensive care unit (ICU) outcome are missing. In this retrospective, multicenter study we analyzed 174 adult HSCT recipients with cGvHD requiring intensive care treatment. Skin, pulmonary, liver, and intestinal involvement were present in 76.7%, 47.1%, 38.1% and 24.1%, respectively, and a total of 63.2% had severe cGvHD. Main reasons for ICU admission were respiratory failure (69.7%) and sepsis (34.3%). Hospital- and 3-year OS rates were 51.7% and 28.6%, respectively. Global severity of cGvHD did not impact short- and long-term survival. However, patients with severe liver cGvHD or the overlap subtype had a reduced hospital survival, while severe pulmonary cGvHD was associated with worse long-term survival. In multivariate analysis need for invasive ventilation (HR 1.08 (95% CI 1.02-1.14)) or hemodialysis (HR 1.73 (95% CI 1.14-2.62)) and <1 year since HSCT (HR 1.56 (95% CI 1.03-2.39)) were independently associated with a poorer survival. While the global severity of cGvHD does not per se affect patients' survival after intensive care treatment, pre-existing severe hepatic, intestinal or pulmonary cGvHD is associated with worse outcomes.

Time-dependent prediction of mortality and cytomegalovirus reactivation after allogeneic hematopoietic cell transplantation using machine learning.

Allogeneic hematopoietic cell transplantation (HCT) effectively treats high-risk hematologic diseases but can entail HCT-specific complications, which may be minimized by appropriate patient management, supported by accurate, individual risk estimation. However, almost all HCT risk scores are limited to a single risk assessment before HCT without incorporation of additional data. We developed machine learning models that integrate both baseline patient data and time-dependent laboratory measurements to individually predict mortality and cytomegalovirus (CMV) reactivation after HCT at multiple time points per patient. These gradient boosting machine models provide well-calibrated, time-dependent risk predictions and achieved areas under the receiver-operating characteristic of 0.92 and 0.83 and areas under the precision-recall curve of 0.58 and 0.62 for prediction of mortality and CMV reactivation, respectively, in a 21-day time window. Both models were successfully validated in a prospective, non-interventional study and performed on par with expert hematologists in a pilot comparison.

Early report on the severity of COVID-19 in hematologic patients infected with the SARS-CoV2 omicron variant.

INTRODUCTION: Patients with hematologic disease are at high risk of morbidity and mortality from COVID-19 due to disease-inherent and therapy-related immunodeficiency. Whether infection with the SARS-CoV2 omicron variant leads to attenuated disease severity in these patients is currently unknown. METHODS: We assessed clinical and laboratory parameters in 61 patients with underlying hematologic conditions with a SARS-CoV2 omicron variant infection confirmed by nucleic acid amplification testing. RESULTS: Fifty patients reported symptoms of COVID-19, most commonly fatigue (37 patients, 60.66%) and cough (32 patients, 52.46%). 39.34% of patients reported fever. Dyspnea was reported by 10 patients and 7 patients (11.48%) required oxygen therapy. Anosmia and ageusia were relatively rare, occurring in less than 10% of patients. Severity of SARS-CoV2 infection could be assessed in 60 patients. Five cases of critical illness leading to ICU admission occurred during the observation period. Overall mortality was 9.84% in this patient cohort, with heterogeneous causes of death. The majority of omicron-infected hematologic patients experienced mild symptoms or remained asymptomatic. DISCUSSION: In this study, symptoms of COVID-19 tended to be milder than described for previous SARS-CoV2 variants. However, the extent to which attenuated severity of omicron-variant SARS-CoV2 infection is caused by altered viral pathogenicity or pre-existing host immunity cannot be inferred from our data and should be investigated in larger prospective studies.

Impact of CMV reactivation on relapse of acute myeloid leukemia after HCT is dependent on disease stage and ATG.

Cytomegalovirus (CMV) reactivation is a frequent complication after allogeneic hematopoietic cell transplantation (HCT), whose impact on clinical outcome, in particular on leukemic relapse, is controversial. We retrospectively analyzed 687 HCT recipients with acute myeloid leukemia (AML) and ciclosporin-based immunosuppression to better understand the differential impact of CMV on transplant outcomes depending on AML disease stage and in vivo T cell depletion with antithymocyte globulin (ATG). Without ATG, CMV reactivation associated with significantly reduced relapse, yet its effect was more pronounced for advanced disease AML (P = .0002) than for patients in first complete remission (CR1, P = .0169). Depending on the disease stage, ATG exposure abrogated relapse protection following CMV reactivation in advanced stages (P = .796), while it inverted its effect into increased relapse for CR1 patients (P = .0428). CMV reactivation was associated with significantly increased nonrelapse mortality in CR1 patients without ATG (P = .0187) but not in those with advanced disease and ATG. Following CMV reactivation, only patients with advanced disease had significantly higher event-free survival rates as compared with patients without CMV. Overall, our data suggest that both ATG and disease stage modulate the impact of post-HCT CMV reactivation in opposite directions, revealing a level of complexity that warrants future studies regarding the interplay between antivirus and antitumor immunity.

Combined Analysis of Early CD4+ T Cell Counts and CMV Serostatus May Improve CMV Risk Assessment after Allogeneic Hematopoietic Cell Transplantation.

The incidence and severity of viral complications after cellular therapy are highly variable. Recent publications describe relevant interactions between the human Cytomegalovirus (CMV) and host immunity in recipients of allogeneic hematopoietic cell transplantation (HCT). Although immune monitoring is routinely performed in HCT patients, validated cut-off levels correlating with transplant outcomes such as survival or CMV reactivation are mostly limited to day +100, which is later than the median time for CMV reactivation in the absence of medical prophylaxis. To address this gap in early risk assessment, we applied an unsupervised machine learning technique based on clustering of day +30 CD4+ helper T cell count data, and identified relevant cut-off levels within the diverse spectrum of early CD4+ reconstitution. These clusters were stratified for CMV recipient serostatus to identify early risk groups that predict clinical HCT outcome. Indeed, the new risk groups predicted subsequent clinical events such as NRM, OS, and high CMV peak titers better than the most established predictor, i.e., the positive CMV recipient serostatus (R+). More specifically, patients from the R+/low CD4+ subgroup strongly associated with high CMV peak titers and increased 3-year NRM (subdistribution hazard ratio (SHR) 10.1, 95% CI 1.38-73.8, p = 0.023), while patients from the R-/very high CD4+ subgroup showed comparable NRM risks (SHR 9.57, 95% CI 1.12-81.9, p = 0.039) without such an association. In short, our study established novel cut-off levels for early CD4+ T cells via unsupervised learning and supports the integration of host cellular immunity into clinical risk-assessment after HCT in the context of CMV reactivation.

Real-world efficacy of docetaxel plus nintedanib after chemo-immunotherapy failure in advanced pulmonary adenocarcinoma.

Aim: This real-world analysis evaluated docetaxel plus nintedanib in patients with advanced pulmonary adenocarcinoma after chemotherapy and immune checkpoint inhibitor failure, for whom treatment options are limited. Methods: Data were sourced retrospectively from seven German centers. Results: Of 93 patients, overall response rate was 41.4% (disease control rate: 75.9%). Of 57 patients given third-line docetaxel plus nintedanib, overall response rate was 50.0% (disease control rate: 82.7%). Median overall survival following third-line docetaxel plus nintedanib was 8.4 months. Adverse events were consistent with the known safety profile of docetaxel plus nintedanib. Conclusion: To date, this was the largest retrospective, real-world analysis of docetaxel plus nintedanib after chemotherapy-immunotherapy failure, indicating that docetaxel plus nintedanib offers meaningful clinical benefits in this setting.

Lay abstract The standard of care for patients with lung adenocarcinoma has advanced with the introduction of immunotherapy in the first-line setting. However, limited clinical data are available to help guide treatment decisions after failure of chemotherapy and immunotherapy. Nintedanib is an oral antiangiogenic agent that is approved in the EU and other countries in combination with docetaxel for the treatment of patients with advanced/metastatic lung adenocarcinoma after first-line chemotherapy. This study is a retrospective, real-world analysis of docetaxel plus nintedanib in 93 patients with advanced lung adenocarcinoma who progressed on immunotherapy (either in sequence or in combination with chemotherapy). The results suggest that docetaxel plus nintedanib offers a meaningful clinical benefit in this setting. Safety findings were generally consistent with the known safety profile of docetaxel plus nintedanib.

A bi-centric experience of extracorporeal carbon dioxide removal (ECCO2 R) for acute hypercapnic respiratory failure following allogeneic hematopoietic stem cell transplantation.

Acute respiratory failure (ARF) is the main reason for ICU admission following allogeneic hematopoietic stem cell transplantation (HSCT). Extracorporeal CO2 removal (ECCO2 R) can be used as an adjunct to mechanical ventilation in patients with severe hypercapnia but has not been assessed in HSCT recipients. Retrospective analysis of all allogeneic HSCT recipients ≥18 years treated with ECCO2 R at two HSCT centers. 11 patients (m:f = 4:7, median age: 45 [IQR: 32-58] years) were analyzed. Acute leukemia was the underlying hematologic malignancy in all patients. The time from HSCT to ICU admission was 37 [8-79] months, and 9/11 (82%) suffered from chronic graft-versus-host disease (GVHD) with lung involvement. Pneumonia was the most frequent reason for ventilatory decompensation (n = 9). ECCO2 R was initiated for severe hypercapnia (Pa CO2 : 96 [84-115] mm Hg; pH: 7.13 [7.09-7.27]) despite aggressive mechanical ventilation (invasive, n = 9; non-invasive, n = 2). ECCO2 R effectively resolved blood gas disturbances in all patients, but only 2/11 (18%) could be weaned off ventilatory support, and one (9%) patient survived hospital discharge. Progressive respiratory and multiorgan dysfunction were the main reasons for treatment failure. ECCO2 R was technically feasible but resulted in a low survival rate in our cohort. A better understanding of the prognosis of ARF in patients with chronic GVHD and lung involvement is necessary before its use can be reconsidered in this setting.

Cytomegalovirus kinetics after hematopoietic cell transplantation reveal peak titers with differential impact on mortality, relapse and immune reconstitution.

Even in the era of PCR-based monitoring, prophylaxis, and preemptive therapy, Cytomegalovirus (CMV) viremia remains a relevant cause of non-relapse mortality (NRM) after allogeneic hematopoietic cell transplantation (HCT). However, studies using binary analysis (presence/absence of CMV) reported contradicting data for NRM, overall survival and leukemia relapse. Here, we analyzed CMV replication kinetics in 11 508 whole blood PCR samples of 705 patients with HCT between 2012 and 2017. Using two independent models based on CMV peak titers and on the time point of first CMV reactivation, we stratified patients into risk cohorts. Each cohort had distinct cellular immune reconstitution profiles and differentiated for relevant clinical outcomes. Patients with high CMV peak titers had significantly reduced overall survival (HR 2.13, 95% CI 1.53-2.96; p < .0001), due to high NRM. Early impaired T cell reconstitution was a risk factor for high CMV peak titers, however relevant CMV viremia also related to boosted T cell reconstitution. Importantly, intermediate CMV peak titers associated with a significantly reduced relapse probability (HR 0.53, 95% CI 0.31-0.91; p = .022). In short, CMV kinetics models distinguished relevant clinical outcome cohorts beyond the R+ serostatus with distinct immune reconstitution patterns and resolve in part contradicting results of previous studies exclusively focused on the presence or absence of CMV.

Permissive HLA-DPB1 mismatches in HCT depend on immunopeptidome divergence and editing by HLA-DM.

In hematopoietic cell transplantation (HCT), permissive HLA-DPB1 mismatches between patients and their unrelated donors are associated with improved outcomes compared with nonpermissive mismatches, but the underlying mechanism is incompletely understood. Here, we used mass spectrometry, T-cell receptor-ß (TCRß) deep sequencing, and cellular in vitro models of alloreactivity to interrogate the HLA-DP immunopeptidome and its role in alloreactive T-cell responses. We find that permissive HLA-DPB1 mismatches display significantly higher peptide repertoire overlaps compared with their nonpermissive counterparts, resulting in lower frequency and diversity of alloreactive TCRß clonotypes in healthy individuals and transplanted patients. Permissiveness can be reversed by the absence of the peptide editor HLA-DM or the presence of its antagonist, HLA-DO, through significant broadening of the peptide repertoire. Our data establish the degree of immunopeptidome divergence between donor and recipient as the mechanistic basis for the clinically relevant permissive HLA-DPB1 mismatches in HCT and show that permissiveness is dependent on HLA-DM-mediated peptide editing. Its key role for harnessing T-cell alloreactivity to HLA-DP highlights HLA-DM as a potential novel target for cellular and immunotherapy of leukemia.

Optimizing anti-T-lymphocyte globulin dosing to improve long-term outcome after unrelated hematopoietic cell transplantation for hematologic malignancies.

Prophylaxis of graft-versus-host disease (GVHD) after allogeneic hematopoietic stem cell transplantation (HCT) remains challenging. Because prospective randomized trials of in-vivo T cell depletion using anti-T-lymphocyte globulin (ATLG) in addition to a calcineurin inhibitor and methotrexate (MTX) led to conflicting outcome results, we evaluated the impact of ATLG on clinical outcome, lymphocyte- and immune reconstitution survival models. In total, 1500 consecutive patients with hematologic malignancies received matched unrelated donor (MUD) HCT with cyclosporin and MTX (N = 723, 48%) or with additional ATLG (N = 777, 52%). In the ATLG cohort, grades III-IV acute (12% vs 23%) and extensive chronic GVHD (18% vs 34%) incidences were significantly reduced (P < .0001). Nonrelapse mortality (27% vs 45%) and relapse (30% vs 22%) differed also significantly. Event-free and overall survival estimates at 10 years were 44% and 51% with ATLG and 33% and 35% without ATLG (P < .002 and <.0001). A dose-dependent ATLG effect on lymphocyte- and neutrophil reconstitution was observed. At ATLG exposure, lymphocyte counts and survival associated through a logarithmically increasing function. In this survival model, the lymphocyte count optimum range at exposure was between 0.4 and 1.45/nL (P = .001). This study supports additional ATLG immune prophylaxis and is the first study to associate optimal lymphocyte counts with survival after MUD-HCT.

Serologic response to meningococcal vaccination in patients with cold agglutinin disease (CAD) in the novel era of complement inhibition.

Cold agglutinin disease (CAD) is a rare, potentially life-threatening acquired autoimmune hemolytic anemia characterized by hemagglutination and hemolysis due to immunoglobulin-mediated (usually IgMκ) classic complement pathway activation. Complement inhibition (CI) represents a novel treatment option to control hemolysis. Due to CI patients (pts) are susceptible to encapsulated bacteria e.g. N. meningitidis. Therefore, meningococcal vaccination on CI is mandatory. In this study serologic response to the tetravalent conjugate vaccine Menveo® was analyzed in CAD pts on eculizumab treatment (DECADE trial) using rabbit serum as complement source (rSBA). Protective rSBA titers varied for meningococcal serogroups and over time reflecting an early decline to even non-protective rSBA titers. These data highlight the importance of serologic analyses under chronic CI. Currently, re-vaccination with a tetravalent meningococcal conjugate vaccine every 3 years is recommended on chronic CI. However, re-vaccination on CI might further rely on serologic analyses, implying even early booster vaccinations similar to adults with (functional) asplenia.

Ileostomy for steroid-resistant acute graft-versus-host disease of the gastrointestinal tract.

Steroid-resistant acute graft-versus-host disease (GVHD) of the gastrointestinal tract associates with important morbidity and mortality. While high-dose steroids are the established first-line therapy in GVHD, no second-line therapy is generally accepted. In this analysis of 65 consecutive patients with severe, steroid-resistant, intestinal GVHD (92% stage 4), additional ileostomy surgery significantly reduced overall mortality (hazard ratio 0.54; 95% confidence interval, 0.36-0.81; p = 0.003) compared to conventional GVHD therapy. Median overall survival was 16 months in the ileostomy cohort compared to 4 months in the conventional therapy cohort. In the ileostomy cohort, both infectious- and GVHD-associated mortality were reduced (40% versus 77%). Significantly declined fecal volumes (p = 0.001) after surgery provide evidence of intestinal adaptation following ileostomy. Correlative studies indicated ileostomy-induced immune-modulation with a > 50% decrease of activated T cells (p = 0.04) and an increase in regulatory T cells. The observed alterations of the patients' gut microbiota may also contribute to ileostomy's therapeutic effect. These data show that ileostomy induced significant clinical responses in patients with steroid-resistant GVHD along with a reduction of pro-inflammatory immune cells and changes of the intestinal microbiota. Ileostomy is a treatment option for steroid-resistant acute GVHD of the gastrointestinal tract that needs further validation in a prospective clinical trial.