Many monogenic disorders cause a characteristic facial morphology. Artificial intelligence can support physicians in recognizing these patterns by associating facial phenotypes with the underlying syndrome through training on thousands of patient photographs. However, this ‘supervised’ approach means that diagnoses are only possible if the disorder was part of the training set. To improve recognition of ultra-rare disorders, we developed GestaltMatcher, an encoder for portraits that is based on a deep convolutional neural network. Photographs of 17,560 patients with 1,115 rare disorders were used to define a Clinical Face Phenotype Space, in which distances between cases define syndromic similarity. Here we show that patients can be matched to others with the same molecular diagnosis even when the disorder was not included in the training set. Together with mutation data, GestaltMatcher could not only accelerate the clinical diagnosis of patients with ultra-rare disorders and facial dysmorphism but also enable the delineation of new phenotypes.

Purpose Phenotype information is crucial for the interpretation of genomic variants. So far it has only been accessible for bioinformatics workflows after encoding into clinical terms by expert dysmorphologists. Methods Here, we introduce an approach driven by artificial intelligence that uses portrait photographs for the interpretation of clinical exome data. We measured the value added by computer-assisted image analysis to the diagnostic yield on a cohort consisting of 679 individuals with 105 different monogenic disorders. For each case in the cohort we compiled frontal photos, clinical features, and the disease-causing variants, and simulated multiple exomes of different ethnic backgrounds. Results The additional use of similarity scores from computer-assisted analysis of frontal photos improved the top 1 accuracy rate by more than 20–89% and the top 10 accuracy rate by more than 5–99% for the disease-causing gene. Conclusion Image analysis by deep-learning algorithms can be used to quantify the phenotypic similarity (PP4 criterion of the American College of Medical Genetics and Genomics guidelines) and to advance the performance of bioinformatics pipelines for exome analysis.

Immune checkpoint inhibitors (ICI) represented a step forward in improving the outcome of patients with various refractory solid tumors and several therapeutic regimens incorporating ICI have already been approved for a variety of tumor entities. However, besides remarkable long-term responses, checkpoint inhibition can trigger severe immune-related adverse events in some patients. In order to improve safety of ICI as well as T cell therapy, we tested the feasibility of combining T cell-based immunotherapy with genetic disruption of checkpoint molecule expression. Therefore, we generated H-Y and ovalbumin antigen-specific CD8⁺ T cells with abolished PD-1, LAG-3, and TIM-3 expression through CRISPR/Cas9 technology. CD8⁺ T cells, subjected to PD-1, LAG-3, and TIM-3 genetic editing, showed a strong reduction in immune checkpoint molecule expression after in vitro activation, while no relevant reduction in responsiveness to in vitro stimulation was observed. At the same time, in B16-OVA tumor model, transferred genetically edited OT-1 CD8⁺ T cells promoted longer survival compared to control T cells and showed enhanced expansion without associated toxicity. Our study supports the notion that antigen-specific adoptive T cell therapy with concomitant genetic disruption of multiple checkpoint inhibitory receptors could represent an effective antitumor immunotherapy approach with improved tolerability profile.

Despite remarkable clinical efficacy, little is known about the system-wide immunological alterations provoked by PD1 blockade. Dynamics of quantitative immune composition and functional repertoire during PD1 blockade could delineate cohort-specific patterns of treatment response and therapy-induced toxicity. We longitudinally assessed therapy-induced effects on the immune system in fresh whole blood using flow cytometry-based cell quantifications, accompanied by analyses of effector properties of all major immune populations upon cell-type specific stimulations. 43 cancer patients undergoing PD1 blockade were recruited with assessments performed pre-treatment and before cycles 2/4/6, which resulted in the collection of more than 30,000 cytometric data values. We observed no intrinsic immune pattern correlating with clinical outcome before PD1 blockade initiation, but cohort-specific immune alterations emerged during therapy. The most striking evolving changes in therapy responders were an increase in activated T and NK cell subsets, which showed high IFNγ and TNFα expression upon ex vivo stimulation. Patients affected by severe immune-related adverse events (s-irAE) presented with an analogously increased number of activated CD4 ⁺ and CD8 ⁺ T cells compared to patients with no/mild irAE, but lacked the functional divergences observed between responders versus non-responders. Instead, their monocytes showed discriminatory functional deficits with less IL10 production upon stimulation, which led to an abrogated inhibition of T cell proliferation in vitro and thus may account for the observed T cell expansion in patients with s-irAE. Our holistic explorative approach allowed the delineation of clinically relevant cohorts by treatment-triggered immune changes, potentially enabling better patient stratification and further revealed new mechanistic insights into the pathogenesis of s-irAE.

T cells and their effector functions, in particular the canonical cytotoxicity of CD8 ⁺ T cells involving perforin, granzymes, Fas ligand (FasL), and tumor necrosis factor related apoptosis inducing ligand (TRAIL), are crucial for tumor immunity. Here, we reveal a previously unidentified mechanism by which CD40L-expressing CD8 ⁺ T cells induce cytotoxicity in cancer cells. In murine models, up to 50% of tumor-specific CD8 ⁺ T cells expressed CD40L, and conditional CD40L ablation in CD8 ⁺ T cells alone led to tumor formation. Mechanistically, CD40L ⁺ CD8 ⁺ T cells can induce cell death in CD40-expressing cancer cells by triggering caspase-8 activation. We demonstrate that a gene signature for resistance to CD40 signaling–induced cell death strongly correlates with worse survival in different human cancer cohorts. Our results introduce CD40L as a rather counterintuitive, noncanonical cytotoxic factor that complements the capabilities of CD8 ⁺ T cells to combat cancers and has the potential to enhance the efficacy of immunotherapies.

Antibody-dependent cell-mediated cytotoxicity (ADCC) by NK cells is a key mechanism in anti-cancer therapies with monoclonal antibodies, including cetuximab (EGFR-targeting) and avelumab (PDL1-targeting). Fc gamma receptor IIIa (FcγRIIIa) polymorphisms impact ADCC, yet their clinical relevance in NK cell functionality remains debated. We developed two complementary flow cytometry assays: one to predict the FcγRIIIa-V158F polymorphism using a machine learning model, and a 15-color flow cytometry panel to assess antibody-induced NK cell functionality and cancer-immune cell interactions. Samples were collected from healthy donors and metastatic colorectal cancer (mCRC) patients from the FIRE-6-Avelumab phase II study. The machine learning model accurately predicted the FcγRIIIa-V158F polymorphism in 94% of samples. FF homozygous patients showed diminished cetuximab-mediated ADCC compared to VF or VV carriers. In mCRC patients, NK cell dysfunctions were evident as impaired ADCC, decreased CD16 downregulation, and reduced CD137/CD107a induction. Elevated PD1+ NK cell levels, reduced lysis of PDL1-expressing CRC cells and improved NK cell activation in combination with the PDL1-targeting avelumab indicate that the PD1-PDL1 axis contributes to impaired cetuximab-induced NK cell function. Together, these optimized assays effectively identify NK cell dysfunctions in mCRC patients and offer potential for broader application in evaluating NK cell functionality across cancers and therapeutic settings.