Fast and reliable detection of patients with severe and heterogeneous illnesses is a major goal of precision medicine^1,2. Patients with leukaemia can be identified using machine learning on the basis of their blood transcriptomes³. However, there is an increasing divide between what is technically possible and what is allowed, because of privacy legislation^4,5. Here, to facilitate the integration of any medical data from any data owner worldwide without violating privacy laws, we introduce Swarm Learning—a decentralized machine-learning approach that unites edge computing, blockchain-based peer-to-peer networking and coordination while maintaining confidentiality without the need for a central coordinator, thereby going beyond federated learning. To illustrate the feasibility of using Swarm Learning to develop disease classifiers using distributed data, we chose four use cases of heterogeneous diseases (COVID-19, tuberculosis, leukaemia and lung pathologies). With more than 16,400 blood transcriptomes derived from 127 clinical studies with non-uniform distributions of cases and controls and substantial study biases, as well as more than 95,000 chest X-ray images, we show that Swarm Learning classifiers outperform those developed at individual sites. In addition, Swarm Learning completely fulfils local confidentiality regulations by design. We believe that this approach will notably accelerate the introduction of precision medicine.

Administration of tryptophan and some of its metabolites reduces the severity of colitis in mice, whereas removing tryptophan from the diet increases susceptibility to colitis. Transfer of the intestinal microbiome transfers the colitogenic phenotype from tryptophan starved animals to normally nourished mice. We aimed to systematically evaluate serum levels of tryptophan and its metabolites in patients with inflammatory bowel diseases (IBD), and study their association with clinical and serologic features. We studied 535 consecutive patients with IBD (211 with ulcerative colitis [UC], 234 with Crohn’s disease [CD]; 236 male), enrolled in Germany from August 2013 through April 2014 and followed until July 2016. Serum samples were collected from patients and 291 matched individuals without IBD (controls); levels of tryptophan were measured using high-performance liquid chromatography. Metabolites of tryptophan were measured in serum from 148 patients and 100 controls by mass spectrometry. We measured levels of interleukin 22 in serum from 28 patients by enzyme-linked immunosorbent assay. Paired stool and serum samples were collected from a subset of patients with active UC (n = 10) or CD (n = 8) to investigate associations between serum levels of tryptophan and composition of the fecal microbiota, analyzed by 16S ribosomal DNA amplicon sequencing. We used real-time polymerase chain reaction to measure levels of messenger RNAs in colonic biopsies from 60 patients with UC, 50 with CD, and 30 controls. We collected information on patients’ disease activity scores, medications, laboratory assessments, and clinical examinations during recruitment and follow-up visits. Serum levels of tryptophan were significantly lower in patients with IBD than in controls (P = 5.3 × 10⁻⁶) with a stronger reduction in patients with CD (vs control; P = 1.1 × 10⁻¹⁰) than UC (vs control; P = 2.8 × 10⁻³). We found a negative correlation between serum levels of tryptophan and disease activity or levels of C-reactive protein. Levels of messenger RNAs encoding tryptophan 2,3-dioxygenase-2 and solute carrier family 6 member 19 (also called B0AT1) were significantly decreased in colonic biopsies from patients with IBD compared with controls, whereas level of messenger RNA encoding indoleamine 2,3-dioxygenase-1 was significantly increased. The composition of the fecal microbiota associated with serum levels of tryptophan. Analysis of tryptophan metabolites revealed activation of the kynurenine pathway, based on high levels of quinolinic acid, in patients with IBD compared with controls. Serum concentration of interleukin 22 associated with disease activity in patients with IBD; there was an inverse association between levels of interleukin 22 and serum levels of tryptophan. In an analysis of serum samples from more than 500 patients with IBD, we observed a negative correlation between serum levels of tryptophan and disease activity. Increased levels of tryptophan metabolites—especially of quinolinic acid—indicated a high activity of tryptophan degradation in patients with active IBD. Tryptophan deficiency could contribute to development of IBD or aggravate disease activity. Interventional clinical studies are needed to determine whether modification of intestinal tryptophan pathways affects the severity of IBD.

A coding variant of the inflammatory bowel disease (IBD) risk gene ATG16L1 has been associated with defective autophagy and deregulation of endoplasmic reticulum (ER) function. IL-22 is a barrier protective cytokine by inducing regeneration and antimicrobial responses in the intestinal mucosa. We show that ATG16L1 critically orchestrates IL-22 signaling in the intestinal epithelium. IL-22 stimulation physiologically leads to transient ER stress and subsequent activation of STING-dependent type I interferon (IFN-I) signaling, which is augmented in Atg16l1ΔIEC intestinal organoids. IFN-I signals amplify epithelial TNF production downstream of IL-22 and contribute to necroptotic cell death. In vivo, IL-22 treatment in Atg16l1ΔIEC and Atg16l1ΔIEC/Xbp1ΔIEC mice potentiates endogenous ileal inflammation and causes widespread necroptotic epithelial cell death. Therapeutic blockade of IFN-I signaling ameliorates IL-22–induced ileal inflammation in Atg16l1ΔIEC mice. Our data demonstrate an unexpected role of ATG16L1 in coordinating the outcome of IL-22 signaling in the intestinal epithelium.

Radiotherapy is a mainstay cancer therapy whose antitumor effects partially depend on T cell responses. However, the role of Natural Killer (NK) cells in radiotherapy remains unclear. Here, using a reverse translational approach, we show a central role of NK cells in the radiation-induced immune response involving a CXCL8/IL-8–dependent mechanism. In a randomized controlled pancreatic cancer trial, CXCL8 increased under radiotherapy, and NK cell positively correlated with prolonged overall survival. Accordingly, NK cells preferentially infiltrated irradiated pancreatic tumors and exhibited CD56 ^dim -like cytotoxic transcriptomic states. In experimental models, NF-κB and mTOR orchestrated radiation-induced CXCL8 secretion from tumor cells with senescence features causing directional migration of CD56 ^dim NK cells, thus linking senescence-associated CXCL8 release to innate immune surveillance of human tumors. Moreover, combined high-dose radiotherapy and adoptive NK cell transfer improved tumor control over monotherapies in xenografted mice, suggesting NK cells combined with radiotherapy as a rational cancer treatment strategy.

COVID-19 shares the feature of autoantibody production with systemic autoimmune diseases. In order to understand the role of these immune globulins in the pathogenesis of the disease, it is important to explore the autoantibody spectra. Here we show, by a cross-sectional study of 246 individuals, that autoantibodies targeting G protein-coupled receptors (GPCR) and RAS-related molecules associate with the clinical severity of COVID-19. Patients with moderate and severe disease are characterized by higher autoantibody levels than healthy controls and those with mild COVID-19 disease. Among the anti-GPCR autoantibodies, machine learning classification identifies the chemokine receptor CXCR3 and the RAS-related molecule AGTR1 as targets for antibodies with the strongest association to disease severity. Besides antibody levels, autoantibody network signatures are also changing in patients with intermediate or high disease severity. Although our current and previous studies identify anti-GPCR antibodies as natural components of human biology, their production is deregulated in COVID-19 and their level and pattern alterations might predict COVID-19 disease severity.