Igor Sauer
Chief consultant and Head of Experimental Surgery
Research Interests
Make your next move.
Prepare your application
Review drafts before you send them.
Explore related searches
Articles (12)
Year:
2026
Year:
2026
Federated learning (FL) has emerged as a promising solution to deal with the privacy concerns which often limit access to data for training machine learning (ML) models. It encompasses the exchange of model parameters between data owners and a central model aggregation instance without the need to share sensitive data directly. Although FL does not require sharing sensitive information, it may not guarantee complete privacy, as model parameters can potentially reveal sensitive details about the underlying training data. To enhance privacy, FL approaches can additionally integrate differential privacy (DP), which entails the careful addition of noise to the model. In this paper, we investigate the relationship between FL, DP and highly non-independent and identically distributed (non-IID) data. We present a specific real-life healthcare application of predicting patient mortality and revision surgery after visceral operations. Our contributions encompass assessing the performance of FL with and without DP in the presence of highly non-IID data, analyzing the fairness for individual training participants and investigating the efficacy of personalization to alleviate the privacy-utility trade-off introduced by DP. Our findings indicate that the addition of DP into FL given non-IID datasets disparately favors participants with larger datasets, while the model has limited utility for those with less data. The addition of a subsequent model fine-tuning step can remedy this disparity and enables all participants to reach a model performance comparable to the centralized training scenario.
Year:
2025
Background Liver transplantation remains the only curative option for end-stage liver disease, yet high waitlist mortality and organ scarcity continue to challenge the field. The randomized multicentric ExTra trial aims to improve donor utilization by evaluating whether normothermic machine perfusion (NMP) allows for safe transplantation of liver grafts that would otherwise be discarded. Methods The ExTra trial is a prospective, randomized, multicentre study conducted across German transplant centres. Participants listed for liver transplantation with a ReMELD-Na-Score ≤ 21 who are not eligible for Standard or Non-Standard Exceptions will be randomized for a one-year intervention period into two groups: the experimental arm, in which they have - in addition to regular organ allocation - the extra option to receive a graft that was initially deemed non-transplantable based on clinical criteria, but meets specified quality criteria after at least four hours of NMP, and the control arm, which consists of conventional allocation procedures alone. The primary endpoint is time-to-transplant, defined as the period between randomization and transplantation. Secondary outcomes include competing events (death, disease progression, or recovery without transplantation), graft and patient survival, quality of life, and cost-effectiveness. Additional analyses will assess early graft function, ICU and hospital stay, biliary complications, and 1-year post-transplant outcomes. A centralized biobank will collect perfusate, tissue, and blood samples for biomarker research. Safety will be monitored through quarterly reviews by the Data and Safety Monitoring Board (DSMB). Discussion The ExTra trial seeks to establish NMP as a tool for improving organ utilization and reducing waitlist mortality by quality assessment of severely marginal liver grafts. Nearly all German transplant centres have expressed their interest in participating, though limited access to NMP devices and lack of reimbursement remain key barriers to broader implementation of this technology in Germany. By generating high-quality clinical and translational data, the trial has the potential to reshape the liver graft allocation process, provide key insights in liver physiology, and contribute to the development of biomarkers for organ assessment by NMP. The findings will be essential for future integration of NMP into clinical routine. Trial registration : ClinicalTrials.gov NCT06874296. Registered March 2025
Year:
2025
BACKGROUND Thyroid surgery entails significant challenges due to the intricate anatomical structure of the region, including the recurrent laryngeal nerve and the parathyroids. OBJECTIVE To support the training of resident surgeons through the development of a virtual reality simulator for thyroid surgery. METHODS We created an anatomical model integrating structures segmented from CT data with manually inserted elements not clearly visible in CT scans (e.g., nerves, parathyroids). The model was integrated into a photorealistic operating room environment created using photogrammetry. The simulation incorporates organ and tissue deformation consistent with clinical observations and surgical exploration. RESULTS The prototype simulator enables users to manipulate key anatomical components and explore the positioning of parathyroids, nerves (e.g., the recurrent laryngeal nerve), and relevant vascular structures, creating an immersive learning experience. CONCLUSIONS We have developed a proof-of-concept virtual reality simulator for thyroid surgery training that has received initial qualitative validation and feedback from experienced thyroid surgeons. CLINICALTRIAL This study primarily involved the development and qualitative assessment of an immersive virtual reality simulator for thyroid surgery training. The anatomical models were created using anonymized publicly available CT scan data, ensuring no identifiable patient information was utilized. For simulator validation, three surgeons, who are also co-authors of this paper, with varying levels of experience participated in the evaluation after receiving a brief explanation of the simulator functionality. Written informed consent was obtained from all participating surgeons prior to their involvement. The photogrammetry process for creating the virtual operating room environment and patient positioning utilized a surgical training mannequin rather than actual patients. No experimental procedures were performed on human subjects or animals, and no sensitive personal data was collected during the research process. This type of technological development and simulator validation research does not typically require formal institutional review board approval according to our institutional guidelines. All research activities were conducted in compliance with the principles outlined in the Declaration of Helsinki and adhered to the institutional regulations for educational technology development and evaluation. The surgeons' feedback was documented and analyzed with their consent, with responses presented in an anonymized format to maintain confidentiality while preserving the educational value of their expert assessment.
Year:
2025
T-cell mediated rejection (TCMR) remains a significant challenge after kidney transplantation and is associated with reduced allograft outcome. Previous research highlighted the critical role of TCMR-induced renal epithelial injury. Yet, the detailed cellular origin of these injury responses and the associated clinical implications remain poorly understood. To induce acute TCMR, we used mouse models of allogeneic (C57BL/6 to BALB/c and BALB/c to C57BL/6) kidney transplantation and syngeneic controls (C57BL/6 to C57BL/6 and BALB/c to BALB/c). Molecular changes were analyzed 7 days post-transplant using single-nucleus RNA sequencing and spatial transcriptomics. Results were compared with snRNA-seq data from three human TCMR biopsies and three stable allografts without rejection. The clinical impact of TCMR-induced epithelial injury was evaluated using marker gene sets on bulk transcriptomic data from 1292 kidney allografts, including 95 TCMR samples, with allograft outcome. Mouse kidneys from allogeneic transplants exhibited all hallmark histological features of TCMR. Single-nucleus RNA sequencing revealed TCMR-induced injured cell states and significant gene expression changes particularly in proximal tubules (PT) and thick ascending limbs (TAL). Spatial transcriptomics showed a heterogeneous spatial distribution of these injured cell states and proximity to leukocytes. Cross-species analysis confirmed similar injured PT and TAL cell states in human TCMR. Kidney allograft outcomes strongly correlated with TCMR-induced injured epithelial cell states. Distinct from other transplant biopsies, severe TAL injury emerged as a key factor for allograft survival after TCMR and was associated with reduced leukocyte proximity, suggesting potential non-immune mechanisms of epithelial damage.
Year:
2025
Risk prediction prior to oncologic esophagectomy is crucial for assisting surgeons and patients in their joint informed decision making. Recently, a new risk prediction model for 90-day mortality after esophagectomy using the International Esodata Study Group (IESG) database was proposed, allowing for the preoperative assignment of patients into different risk categories. However, given the non-linear dependencies between patient- and tumor-related risk factors contributing to cumulative surgical risk, machine learning (ML) may evolve as a novel and more integrated approach for mortality prediction. We evaluated the IESG risk model and compared its performance to ML models. Multiple classifiers were trained and validated on 552 patients from two independent centers undergoing oncologic esophagectomies. The discrimination performance of each model was assessed utilizing the area under the receiver operating characteristics curve (AUROC), the area under the precision–recall curve (AUPRC), and the Matthews correlation coefficient (MCC). The 90-day mortality rate was 5.8%. We found that IESG categorization allowed for adequate group-based risk prediction. However, ML models provided better discrimination performance, reaching superior AUROCs (0.64 [0.63–0.65] vs. 0.44 [0.32–0.56]), AUPRCs (0.25 [0.24–0.27] vs. 0.11 [0.05–0.21]), and MCCs (0.27 ([0.25–0.28] vs. 0.15 [0.03–0.27]). Conclusively, ML shows promising potential to identify patients at risk prior to surgery, surpassing conventional statistics. Still, larger datasets are needed to achieve higher discrimination performances for large-scale clinical implementation in the future.
Year:
2024
Cell-death-inducing DNA fragmentation factor-alpha (DFFA)-like effector b (CIDEB) was first identified as an apoptosis-inducing protein. Further research revealed a pivotal role in lipid metabolism, regulating very-low-density lipoprotein (VLDL), lipid droplets (LD), sterol response element-binding protein (SREBP), and chylomicrons. Recent studies have uncovered that rare germline variants in CIDEB protect against liver diseases, including MAFLD, cirrhosis, and viral hepatitis. Furthermore, CIDEB influences steps of the hepatitis C virus (HCV) replication cycle. This review summarizes the current knowledge about CIDEB’s roles in apoptosis, lipid metabolism, and viral hepatitis, and highlights its critical role in liver diseases.
Year:
2024
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.
Year:
2024
CAR-T therapies have shown remarkable success in treating hematological malignancies. However, effectiveness against solid tumors remains limited due to the immunosuppressive tumor microenvironment (TME), such as TGF-β signaling and upregulated immune checkpoints (ICs). Furthermore, identifying universal, tumor-specific targets for CAR-T cells in solid tumors is challenging, but using reinvigorated, immunosuppressive-resistant tumor-infiltrating lymphocytes (TILs) could be a promising alternative approach. Unlike nucleases, which may induce genotoxic DNA double-strand breaks, multiplexed Zinc Finger Repressors (ZFR) offer a safer alternative for knocking out TME-related immunosuppressive factors. We epigenetically repressed PD-1 expression both in CAR-T cells and TILs from colorectal liver metastases. PD-1 repression did not affect T cell viability, proliferation, or functionality. In a murine B cell lymphoma model, PD-1-repressed CD19-CAR-T cells exhibited enhanced anti-tumor activity and improved survival. Notably, PD-1 repression alone did not increase cytotoxicity against a PD-L1-positive colorectal cell line in vitro. To further increase anti-tumor potency in this context, ZFR-expressing lentiviral vectors targeting PD-1 and other ICs (LAG-3, TIM-3, TIGIT) or TGFBR2 were developed, improving significantly the cytotoxic activity in TILs. This strategy highlights the potential to enhance tumor-reactive T cells and improve anti-cancer immunotherapies by epigenetically repressing immunosuppressive factors in the TME using multiplexed ZFRs.
Year:
2024
Year:
2023
Year:
2023
Collaborators (14)
Simon Moosburner
Charité - Universitätsmedizin Berlin
Peter Tang
Charité - Universitätsmedizin Berlin
Robert Toivo Tapani Ahlborn
Charité - Universitätsmedizin Berlin
Wenzel Schöning
Charité - Universitätsmedizin Berlin
Madhukar S. Patel
University of Texas Southwestern Medical Center
Axel Winter
Charité - Universitätsmedizin Berlin
Dominik P. Modest
-
Robin Philippus van de Water
Hasso Plattner Institute
Kai Schmidt-Ott
Ärzte
Medizinische Hochschule Hannover
Bert Arnrich
Professor
Hasso Plattner Institute
Il‐Kang Na
Charité - Universitätsmedizin Berlin
Moritz Queisner
Professor
Charité - Universitätsmedizin Berlin
Frederike Butz
Charité - Universitätsmedizin Berlin
Michael Fähling
Professor
Charité - Universitätsmedizin Berlin
Make your next move.
Prepare your application
Review drafts before you send them.