Advance Precision Medicine Through Integrated Multi-Omics Patient Data

Evotec’s Molecular Patient Database E.MPD Patient Cohorts in Detail GCKD (Roman Martin, guest interview with Prof. Eckardt), UMG (Kay Schreiter, guest interview with Priv.-Doz. Dr. med. Björn Tampe) Drug Discovery Target ID & Pathway Analysis Bed-to-Bench-to-Bed concept Patient Stratification ML and feature reduction concept Application in autoimmune diseases A Note From Our CSO Welcome to the first edition of our new e-book – a refreshed format created to share forward-thinking insights into Shared R&D, innovative technologies, and the science that powers Evotec. Building on the legacy of Drug Discovery Update (DDup), this publication now offers a broader view of our capabilities and collaborative approach across the R&D continuum. In this first issue, we are excited to highlight E.MPD, Evotec’s Molecular Patient Database – a groundbreaking platform at the heart of our precision medicine approach. E.MPD integrates high-quality molecular data with clinical insights, enabling a molecular understanding of disease biology and accelerating the development of personalized therapies. This platform exemplifies our commitment to driving data-driven innovation, fostering collaboration, and developing innovative treatment options that make a real difference for patients in the future. Aligned with our purpose to develop life-changing medicines and our vision to accelerate the journey from concept to cure, this format offers a closer look at how we work, what drives us, and the opportunities our platforms create. Whether you’re familiar with Evotec or just getting to know us, we hope this publication provides valuable insights and inspiration. Thank you for joining us on this journey. Warm regards, Cord Dohrmann The Authors Questions? Contact us! Cord Dohrmann Chief Scientific Officer info@evotec.com Roman Martin Senior Research Scientist Hanna Schebet Senior Research Scientist Ricardo Castro Research Associate Lavanya Muthukrishnan Research Scientist Tobias Bohnenpoll Senior Research Scientist Olivier Radresa SVP Head of Nephrology Kay Schreiter Principal Scientist Uwe Andag EVP Head of Therapeutic Areas Evotec’s Molecular Patient Database E.MPD 3 Contents 1 Page 6 Evotec’s Molecular Patient Database (E.MPD) Evotec’s Molecular Patient Database E.MPD 4 2Patient Cohorts in Detail 2.1 Insights from Prof. Kai-Uwe Eckardt: CKD Research Page 8 2.2 Advances with Priv.-Doz. Dr. med. Björn Tampe: Vasculitis Treatments Page 12 3.1 Target Identification & Pathway Analysis Chronic Kidney Disease (CKD) Page 20 Cardiovascular Disease (CVD) Page 22 3.2 Bed-to-Bench-to-Bed Concept: A Case Study Page 24 3Drug Discovery 6.1 Empowering Multi-Omics Analysis Page 26 6Evotec’s Vision 6.2 Future Directions in Precision Medicine Page 27 4.2 Application in Autoimmune Diseases Chinook Collaboration Page 15 Systemic Lupus Erythematosus (SLE) Case Study Page 18 4.1 Machine Learning in Precision Medicine Page 14 4 Patient Stratification and AI 5Molecular Innovations 5.1 Unbiased Molecular Classification of CKD Page 10 5.2 ADPKD Case Study: PanOmics Research for Biomarker Identification Page 21 Evotec’s Molecular Patient Database E.MPD 5 E.MPD: Evotec’s Molecular Patient Database by Uwe Andag and Philipp Skroblin Using human data as a starting point for drug discovery programs should greatly improve the probability of success during development of novel innovative treatment options for patients. With this in mind, Evotec began a few years ago to build an in-house molecular patient database in close collaboration with academic partners and major hospitals. Since then, Evotec’s Molecular Patient Database (E.MPD) has become a major component in daily research and development processes at Evotec, underpinning our long-term vision that starts with the patient and ends with the patient’s benefit. The E.MPD basically started with acquisition of data relating to chronic kidney disease (read more in DDup #7), and has successfully been expanded into other important therapeutic areas such as metabolic and immune-mediated diseases. The database is a result of a multi-omics (“PanOmics”) analysis of patient samples combined with key information from patient’s clinical records. The PanOmics analysis includes classical genomics and transcriptomics, high-resolution and spatial transcriptomics as well as high-throughput, state-of-the-art proteomics to validate our findings at a protein level. Evotec’s Molecular Patient Database E.MPD 6 By applying a holistic PanOmics analysis approach to patient samples that involves PanHunter, Evotec’s proprietary software for handling large data sets, we were able to generate more than 200 billion data points from over 25,000 patients and a significant number of healthy controls, so far. Why have we made this effort to develop our own database when there are already databases that are readily available in the public domain? Compared to many public domain data sets, the following reasons highlight why a fully QCed internal database is superior for subsequent data analysis: – Evotec controls the acquisition of high-quality PanOmics data leveraging inhouse state-of-the-art high-throughput and high-resolution multi-omics technologies, which allows for data comparability of different sample & analysis batches, within diseases and between disease states. – Evotec engages closely with clinicians during the acquisition of high-quality clinical patient samples and data prior to and during clinical trials. This is important for study planning and a clear understanding of patient-centric data. The relationship, between the clinic and our research teams, is also critical in moving from bedside to bench and back to patients again during the development of the successful therapeutic candidate and the discovery of disease population specific biomarkers. – A comprehensive understanding of the progression from health to disease enables the development of first-in-class therapies, facilitates innovative biomarker discovery and clinical trial support, and can lead to the development of next generation diagnostics. The E.MPD approach enables and drives precision medicine, offering innovative opportunities for a new paradigm of patient diagnosis, innovative therapy development, optimizing clinical trial outcomes and minimizing adverse effects by improving all stages of drug discovery and development. Welcome to Evotec’s world of PanOmics in the E.MPD, Evotec’s Molecular Patient Database! E.MPD data are the result of a multiomics analysis of patient samples combined with key information from patients clinical records Evotec’s Molecular Patient Database E.MPD 7 Interview with Prof. Dr. Kai-Uwe Eckardt by Roman Martin Prof. Dr. Eckhardt, thank you for joining us today. Could you start by providing an overview of the GCKD cohort and its primary objectives? Thank you for having me. The German Chronic Kidney Disease (GCKD) cohort study is one of the world’s largest studies focused on chronic kidney disease (CKD). It was established to investigate the various factors contributing to the progression and complications of CKD. Our primary objectives include identifying genetic, environmental, and lifestyle factors that influence CKD. The comprehensive collection of bio-samples and clinical data will enable us to discover biomarkers for rapid progression and hopefully also targets for novel therapies. Could you explain the design and structure of the GCKD cohort study and why it is special compared to other large kidney cohort studies? The GCKD study was designed as a prospective observational study. We recruited over 5,000 participants with moderately advanced CKD from nephrology outpatient clinics across Germany. With that multicentric approach we were able to compile one of the largest CKD cohorts worldwide. The participants underwent comprehensive baseline assessments, including de- Prof. Dr. Kai-Uwe Eckardt is a distinguished nephrologist and researcher with extensive experience in the field of chronic kidney disease (CKD). He holds a degree in human medicine from the University of Münster and completed his clinical training in internal medicine and nephrology at several prestigious institutions, including Hannover Medical School, University of Zürich, University of Oxford, University of Regensburg, Charité - Berlin and University Medical Centre Erlangen. Between 2004 and 2017, he was Chair of the Department of Nephrology and Hypertension at the Friedrich-Alexander-Universität of Erlangen-Nürnberg. Since 2017 he has held the position of clinic director for internal medicine with focus on nephrology and intensive care at the Charité. Prof. Dr. Kai- Uwe Eckardt is the funding principal investigator of the German Chronic Kidney Disease (GCKD) study, one of the largest and most comprehensive studies of its kind worldwide. His work has been published in numerous high-impact scientific journals, and he has received multiple awards for his contributions to nephrology and translational medicine. About tailed questionnaires, physical examinations, and laboratory tests. Standardization as well as manual curation ensure high data quality. We collected a number of blood and urine samples at baseline and during follow-up visits every two to three years, to establish a central biobank. The longitudinal design allows us to monitor disease progression and the emergence of complications over time. With the first patients recruited in 2009, GCKD is also among the longest running studies in the kidney disease field. The high level of retention of our patients was only possible through the great dedication of all GCKD members. In 2021 you entered a collaboration with Evotec – what are the key benefits of this partnership for GCKD? We were very happy to collaborate with Evotec, one of the leading companies in pharmaceutical renal research. Funding provided by Evotec and the retention of our highly motivated participants was key to extend the follow up. This allowed continuation of the GCKD study for an additional 4 years, including one additional visit for collection of follow-up samples from nearly 1,500 participants. Evotec’s Molecular Patient Database E.MPD 8 Besides this, we are intrigued by the capacity of Evotec for performing and analyzing large scale omics data. Here we rely on the high-quality sample processing pipelines at Evotec as well as the advanced multi-omics data analysis platform, PanHunter, which will enable direct access to the omics data for GCKD scientists. The fact that the agreement with Evotec was to share all data generated from the study samples was crucial for GCKD. I am confident that the joint analysis will also in future inspire multiple sub-projects of Evotec and GCKD affiliates. What are the key benefits that you expect from the mentioned joint sample analysis, from your point of view? How will this impact medical care in future? We are especially excited about the transcriptomics results from approx. 4,500 blood samples, including two time points, that we provided to Evotec. This data will allow us to examine gene expression patterns. Use of Evotecs multi-omics data analysis platform will provide fast access to the data and state-of-the-art analysis tools for GCKD scientists. We anticipate that this data will inspire close scientific collaborations. Together, we aim to gain a deeper understanding of the molecular mechanisms driving CKD progression. The longitudinal design of the study will also provide highly valuable data showing how gene expression patterns change in response to various clinical and live-style factors. Application of machine learning algorithms developed together with Evotec will hopefully lead to the identification of new molecular biomarkers for early detection of high risk disease progression, as well as potential therapeutic targets. Together these insights will pave the way for personalized medicine approaches tailored to individual patient profiles. Are there further omics analysis planned apart from RNA sequencing? Indeed, we believe that the full benefit of omics technologies requires analysis at multiple levels. We are therefore also generating data using other omics technologies. This is possible because of the rich biosample collection that GCKD established. We also provided plasma and urine samples to Evotec for additional analyses in the framework of the collaboration. Importantly, the results of such analyses will also be shared with GCKD investigators. Apart from that, we are now collecting biopsy samples in a retrospective sub-project together with Evotec. GCKDs rich biobank even enabled the support a recent subproject with Evotec in which we used snap-frozen blood for single cell analysis. What challenges do you foresee in this collaboration, and how do you plan to address them? One of the main challenges is the integration and analysis of large and complex datasets generated from multi-omics approaches. Ensuring data quality, reducing batch effects and grant consistency across different omics platforms is crucial. To address this, experts from GCKD are collaborating closely with experts from Evotec. We will employ Evotec‘s production-ready sample processing pipelines as well as the robust bioinformatics toolkits provided by PanHunter. Within GCKD we are collaborating across different centers to incorporate diverse scientific expertise, in particular, we are able to rely on the great expertise in data science and systems biology from Prof. Dr. Köttgen‘s group at the University of Freiburg. Another challenge is translating molecular findings into clinical practice. This requires validation studies and close collaboration with clinicians to ensure that our findings are applicable and beneficial in a real-world setting. GCKD provides a unique network including the leading experts in kidney disease and also collaborate with various international partners, e.g. within the CKD Prognosis Consortium. A continual dialogue between academics, clinics and industry partners in a multidisciplinary approach will be key to overcoming these challenges associated with CKD. Finally, how do you envision the long-term impact of this collaboration on CKD research and patient care? In the long term, this collaboration has the potential to significantly advance our understanding of CKD. By identifying new biomarkers and therapeutic targets, we can develop more effective and personalized treatments. This could lead to earlier diagnosis, improved patient stratification and better disease management. Ultimately, our goal is to reduce the burden of CKD on patients and healthcare systems, and collaborations like this are essential to achieving that vision. Evotec’s Molecular Patient Database E.MPD 9 by Kay Schreiter Interview with Priv.-Doz. Dr. med. Björn Tampe on ANCAVasculitis Evotec’s Molecular Patient Database E.MPD 10 ANCA (Antineutrophil Cytoplasmic Antibody)- associated vasculitis is a heterogeneous autoimmune disorder characterized by the production of autoantibodies against molecules in the cytoplasm of neutrophils, white blood cells that play an important role in the immune system fighting infection and inflammation. The production of autoantibodies leads to inflammation of small blood vessels, which involves multiple organs, especially the kidney. Current therapies are restricted to the depletion of B-cells, a type of white blood cell that neutrophils interact with (Rituximab) and the inhibition of the part of the innate immune system known as complement (Avacopan). Early diagnosis and current immunosuppressive therapies have shifted the focus to managing remission, addressing complications such as renal involvement, infections, and cardiovascular risk, as well as predicting relapses, which remain significant challenges. A comprehensive understanding of autoimmune diseases like ANCA-associated vasculitis enables proper patient stratification and innovative biomarker discovery for treatment selection, treatment monitoring and, relapse prediction. A better understanding of the system biology of these diseases can lead to the improved development of next generation diagnostics to enhance clinical trial support and to even better therapies. Evotec’s Molecular Patient Database E.MPD 11 5 minutes with Björn Tampe interviewed by Kay Schreiter Why is it important to study autoimmune diseases, especially ANCA-associated vasculitis? Autoimmune diseases are very common. Worldwide, 5 to 10 % of the population is affected by at least one of 80 to 100 different autoimmune diseases. One of these autoimmune diseases is ANCA-associated vasculitis, which is a prototypical disease induced by auto- antibodies. For ANCA-associated vasculitis, a very good treatment protocol exists and the progression of disease is, for the most part, well understood. This enables efficient patient monitoring and supports the standardized collection of patient samples, which is the fundamental basis of the non-interventional patient cohort study with Evotec. The course of the disease, which is not a heterogeneous one, can be described very well and can be compared between different patients. This offers the opportunity to gain deep insights into this disease, which is translatable to the understanding of other autoimmune diseases. About The treatment of ANCA-associated vasculitis encompasses an induction and a maintenance therapy. The challenge in this context is a immunosuppression balanced between disease control but also risk of infectious complications. Therefore, non-invasive biomonitoring is required enabling a tailored use of immunosuppressive regimens. This becomes essential when disease remission is achieved and the prediction of potential relapse of the disease is needed. Commonly used treatment regimens with immunosuppressants include the approved treatment with Rituximab, a B-cell depleting therapy, that is given for long-term therapy and is also used in the treatment of other autoimmune diseases. Avacopan, which has recently been approved for ANCA-associated vasculitis, provides a treatment alternative. It is an oral complement-system inhibitor. What is unique about this ANCA-vasculitis cohort or different from other initiatives? In addition to the non-interventional study with Evotec, University Medical Center Göttingen (UMG) is also involved in clinical trial studies, where the UMG is among the top hospitals and specialist clinics for the recruitment of ANCA-associated vasculitis patients. Nevertheless, the cohort with Evotec is quite unique as we will achieve one of the largest number of patients enrolled worldwide for ANCA-associated vasculitis outside of clinical trials, despite the rarity of the disease. It is also noteworthy that the diagnostic kidney biopsy, which is carried out for each patient in the study, is accompanied by a simultaneous blood sampling and single cell data, which are generated directly from a fresh blood sample at the Evotec research laboratory. This is essential Priv.-Doz. Dr. med Björn Tampe is an established specialist in nephrology, immunology and critical care medicine. After his degree in human medicine from the University of Göttingen, he spent his postdoc at the Harvard Medical School in Boston. Thereafter, he joined the Department of Nephrology and Rheumatology at the University Medical Center Göttingen where he is executive assistant medical director and head of intensive care medicine. He focuses on translational research in kidney disorders, connecting experimental findings to clinical implications. This includes autoimmunity in vasculitis, immune checkpoint inhibitor-associated complications, inflammation and organ fibrosis. He has more than 100 peer reviewed publications and received several awards for his contributions to nephrology and translational medicine. Evotec’s Molecular Patient Database E.MPD 12 for our collaborative research project where correlation of the biopsy directly with the blood can lead to new biomarkers for diagnostics. The proximity of UMG to Evotec and the timely analysis of the unique patient samples using the latest omics technologies available at Evotec in Göttingen is absolutely crucial here. The fastest possible analysis makes it possible to understand cellular relationships before the cells and systemic processes in the blood can change due to the transport or storage of the samples. This gives us a direct image of the systemic changes in the patient’s samples. What novel insights do you expect from the ANCA-associated vasculitis cohort study? A comprehensive mechanistic understanding of patient stratification. A new classification of patients based on multi-omics data with regard to pathogenesis, treatment response and relapse. To date, there are no predictions about a possible relapse for these patients and no good definition of a partial or complete relapse. So far, no procedures have prevailed in everyday clinical practice. As part of today’s treatment regimen patients receive a fixed dose of Rituximab regardless of the B-cell serotype and the course of the disease. So we will also gain molecular and cellular insights into the current treatment regimen, which will enhance the understanding of the off-label use of these drugs in other autoimmune diseases. When do you think clinical trials and ultimately the patients will benefit from the ANCA-associated vasculitis cohort? Left untreated, the disease is fatal. Today there are good treatment methods, but they greatly influence the immune system. Patients no longer die from the disease, but often succumb to the consequential risks of treatment such as infections. It is therefore very important to taper the immunosuppression during treatment. Additionally, treatment methods are still based on the physicians’ individual decisions. Ideally, better non-invasive biomarkers are needed to replace the invasive kidney biopsy and the hope is that Evotec can develop molecular approaches that can meet this expectation. It is still not known in detail how much the activity of the complement system, an integral part of the human immune system, plays a role in treatment response. Active and better biomarkers are needed that lead to better treatment decisions especially for lowering corticosteroid dosing during treatment. Apart from funding, where do you see Evotec’s key contribution? Together with Evotec, we have developed a standardized operation procedure that brings the clinical samples directly from the patient to the research laboratory at Evotec in order to generate comprehensive panomics data. We are now also pursuing this standardized concept of consistent sample collection for other diseases and are steadily building up biobanks at the UMG. In addition, these studies contribute to improving the care structure for patients and bundling the care of a rare disease at one site – better care through centralization and networking with research for better disease insights with Evotec. In recent years, we have seen an increase in the number of patients from 20 to about 70–80 ANCA-associated vasculitis patients per year who visit our clinic. Are there plans to expand the program? The collaboration between our University Hospital and Evotec creates an infrastructure that enables patient follow-up over a longer period of time. The UMG creates sustainable patient loyalty in this study and offers patients follow-up treatment and comprehensive monitoring for years. There is strong patient interest and patients sometimes accept long journeys to become part of the study with Evotec. “The treatment of ANCA-associated vasculitis encompasses an induction and a maintenance therapy. The challenge in this context is an immunosuppression balanced between disease control but also risk of infectious complications. Therefore, non-invasive biomonitoring is required enabling a tailored use of immunosuppressive regimens. This becomes essential when disease remission is achieved and the prediction of potential relapse of the disease is needed.” Evotec’s Molecular Patient Database E.MPD 13 Chronic kidney disease (CKD) is a highly prevalent and heterogeneous group of disorders characterized by progressive loss of kidney function resulting from a variety of etiologies involving different cellular and molecular disease mechanisms. Despite this heterogeneity, conventional classification of CKD follows a reductionist approach based on indirect systemic measures such as estimated glomerular filtration rate (eGFR) and albuminuria and is insufficient to predict patient outcomes and treatment responses. We argue that the reductionist classification system reflects a lack of mechanistic disease understanding, which impedes the discovery and development of new therapies and contributes to the high unmet medical need in kidney disease. Recent advances in kidney transcriptomic profiling of large patient cohorts provide a mechanistic approach to CKD classification and offer a promising path to a new era of kidney precision medicine.1,2 Kidney-centric, unbiased clustering of patients by their biopsy transcriptomes revealed four novel molecular groups with different biological pathway enrichment, that were associated with disparate progression rates, histopathology and biomarker signatures.2 Importantly, these molecular groups were consistently identified across multiple independent patient cohorts, suggesting biological relevance and generalizability. Despite these significant advances, the underlying cellular and molecular features that define these novel CKD endotypes remained poorly understood. Evotec’s Molecular Patient Database (E.MPD) combines rich clinical data and patient samples from more than 12,000 CKD patients into a unique resource for human data-driven research. A representative selection of ~ 300 kidney biopsy transcriptomes from the National Unified Renal Translational Research Enter- Unbiased Molecular Classification of Chronic Kidney Disease by Tobias Bohnenpoll prise (NURTuRE) cohort were clustered using self-organizing maps (SOM), an unsupervised machine learning algorithm that groups transcriptomes by molecular similarity (Figure 1).3,4 We identified 4 clusters that were consistent with the previously described molecular groups.2 PHATE dimension reduction revealed a molecular gradient partitioned by SOM clusters that was correlated with eGFR decline, interstitial fibrosis and tubular atrophy (IFTA), suggesting a previously undescribed pseudotemporal relationship along the disease continuum.4,5 Embedding of cell type specific signatures revealed continuous tissue remodeling along this common progression axis, including tubular atrophy, fibrosis and immune infiltration as well as podocyte loss and parietal epithelial cell expansion (Figure 2).6 We derived a quantitative model of pseudotemporal gene expression dynamics that allows differentiation of early disease initiating and driving events from late consequences, thus enabling a mechanistic interpretation of CKD progression. Our work demonstrates the power of E.MPD and highlights the importance of curated, high-quality cell type and mechanistic signatures and quantitative gene expression models to elucidate tissue remodeling dynamics and cell state transitions in CKD progression. The validated, unbiased molecular groups provide a complementary classification system that has the potential to improve disease staging and patient stratification by incorporating a kidney-centric and mechanistic view. Importantly, the pseudotemporal interpretation of gene expression dynamics supports target and biomarker discovery by enabling prioritization of early disease initiating and driving mechanisms. Together, these results contribute to the ambitious goal of early and precise diagnosis and intervention in a disease area of high unmet need. Evotec’s Molecular Patient Database (E.MPD) combines rich clinical data and biosamples from more than 12,000 CKD patients into a unique resource for human data-driven research. Evotec’s Molecular Patient Database E.MPD 14 Fig 1. Unbiased molecular classification of chronic kidney disease A data-driven selection of kidney biopsy transcriptomes (n = 303) from the NURTuRE patient cohort was clustered using self-organizing maps3, resulting in 4 molecular groups (EB, E, B and C) that aligned with clinical (eGFR) and histopathological (IFTA) parameters of disease progression. PHATE5 dimension reduction and trajectory analysis allowed the integration of molecular groups into a model of pseudotime disease progression, providing a novel approach to molecular disease stratification and staging. N=303 NURTuRE CKD patients Kidney biopsy transcriptomics Self-organizing maps AI algorithm MCD 22% FSGS 14% Vasculitis 15% IgAN 11% MN 9% CKDu 7% GN 5% TIN 5% DKD 4% Other 8% Broad range of INS/CKD etiologies Phate 2 Phate 1 SOM cluster EB E B C pseudotime Clinical stratification Phate 2 Phate 1 Kidney function (eGFR) 50 100 Histological stratification Phate 1 Histology (IFTA) 0 1 2 3 Phate 2 Molecular stratification Phate 2 Phate 1 Mol. disease stage 1 2 3 4 Unbiased classification (SOM) Evotec’s Molecular Patient Database E.MPD 15 Fig 2. Tissue remodeling dynamics along a common pseudotime disease progression axis. 1 Lake, B. et al. An atlas of healthy and injured cell states and niches in the human kidney. Nature 619, 585–594 (2023). 2 Reznichenko, A. et al. Unbiased kidney- centric molecular categorization of chronic kidney disease as a step towards precision medicine. Kidney International 105, 1263–1278 (2024). 3 Löffler-Wirth, H. et al. oposSOM: R-package for high-dimensional portraying of genome-wide expression landscapes on bioconductor. Bioinformatics 31, 3225–3227 (2015). 4 Bohnenpoll, T. et al. FC080: A Systems Nephrology Framework for the Molecular Classification of Chronic Kidney Disease. Nephrology Dialysis Transplantation 37, gfac114.004 (2022). 5 Moon, K. R. et al. Visualizing structure and transitions in high-dimensional biological data. Nat Biotechnol 37, 1482–1492 (2019). 6 Bohnenpoll, T. et al. Unsupervised Characterization of the NURTuRE Cohort Reveals Gene Expression and Tissue Remodeling Dynamics Along a Synthetic CKD Progression Axis [Abstract]. J Am Soc Nephrol 33, 2022: 883 (2022). Phate 2 Phate 1 Proximal tubule Phate 2 Phate 1 Podocytes Phate 2 Phate 1 Immune cells Phate 2 Phate 1 Parietal epithelium Embedding of cell typespecific gene expression signatures, representing proximal tubules, immune cells, podocytes or parietal epithelial cells revealed CKD tissue remodeling dynamics. The pseudotime disease progression model captured changes in tissue composition and cell state transitions for common and rare cell populations allowing for a mechanistic interpretation of disease progression. Signature expression 0 0,5 1 Evotec’s Molecular Patient Database E.MPD 16 Decoding SLE Heterogeneity: A PanOmics Approach to Patient Stratification and Target Identification Hanna Schebet, Omar Elakad, Ilayda Beyreli Kokundu, Julian Reinhard, René Rex, Kay Schreiter, Helge Stark, Kulwadee Thanamit, Deniz Yuezak, Christiane Honisch, Uwe Andag, Philipp Skroblin Systemic Lupus Erythematosus (SLE) is a highly debilitating chronic disease that is difficult to diagnose and treat. Evotec’s Molecular Patient Database E.MPD 17 I t is an autoimmune condition where the immune system attacks healthy tissue, resulting in widespread inflammation and organ damage. SLE is inherently heterogeneous, presenting a broad range of symptoms such as skin rashes, fatigue, joint pain, swelling, and fever that can differ significantly between individuals. It has highly variable disease courses and progression patterns with potential involvement of nearly any organ system, often leading to severe manifestations in the kidneys, heart, lungs, and brain. Clinical variability of SLE is mirrored by underlying molecular heterogeneity, particularly evident at the blood transcriptome level, involving multiple immune cells and signaling pathways disrupted in different combinations across patients. This complexity presents major obstacles to accurate diagnosis, disease management, therapeutic target identification, and successful drug development. Diagnosis of SLE is frequently delayed by several years after disease onset, and many patients experience incorrect or missed diagnoses, risking organ damage due to unmanaged disease. Even post-diagnosis, patients often face challenges due to the lack of personalized treatment options. Medicines beneficial for one individual may be ineffective or harmful for another. Indeed, the failure of many late-stage clinical trials is largely attributed to unaddressed patient heterogeneity. To address these challenges, omics-driven approaches leveraging big data analytics are increasingly essential to support the development of efficient personalized therapies with companion diagnostics. In collaboration with leading medical experts in autoimmune disorders, we have expanded Evotec’s Molecular Patient Database (E.MPD) with multiomics profiles from over 10,000 donors with SLE, other related autoimmune diseases as well as carefully selected healthy controls. The multi-omicsdata, which include whole blood and single cell RNA sequencing, genomic and highly sensitive proteomic profiling, are complemented by the detailed clinical data including disease activity scores, treatments, organ involvement to enable comprehensive patient assessment. Applying state-of-the-art machine learning (ML) and bioinformatics tools to the E.MPD data, we stratified SLE patients based on underlying molecular and cellular signatures central to disease pathology, allowing us to truly understand disease complexity (Figure 1). The molecular subtypes of SLE patients are more homogeneous and vary in disease mechanisms and clinical characteristics. To identify novel biomarkers and drug targets specific to each molecular subtype, we have developed a comprehensive ML-driven pipeline that allows us to accurately classify SLE samples and distinguish them from related autoimmune disorders and healthy controls with high accuracy as well as to identify key predictive features important for the classification (Figure 2). These features serve as the basis for systematic target and biomarker prioritization, strengthening the foundation for precision medicine in complex autoimmune conditions like SLE. Our precision medicine approaches hold great promise for transforming management of heterogeneous diseases. By advancing our understanding of disease heterogeneity, we are paving the way for more effective, personalized treatments and ultimately, better patient outcomes. The diagnosis of SLE is often delayed by many years after the disease onset and many patients receive incorrect diagnosis or no diagnosis at all with the risk of extensive organ damage due to unmanaged disease. Evotec’s Molecular Patient Database E.MPD 18 Figure 1: Stratifying SLE Patients Using Transcriptomic Signatures Each molecular endotype reveals a distinct pattern of immune dysregulation in SLE Figure 2: Using ML to Identify Biomarkers & Targets from Omics Data Machine Learning reveals key features for accurate disease classification and target discovery Whole Blood Transcriptomics SLE patients from E.MPD 1. Accurate ML classification Separating Target Disease vs Control 2. Feature Reduction Key for Classification Omics features 3. Target/Biomarker Candidates Evidence-based Scoring for Prioritization Unsupervised Stratification and Endotyping clustering based on signature enrichment Target 1 ↑ Closest to Target 2 ↑ Healthy This pipeline enables biomarker discovery, patient stratification, and tailored therapeutic strategies in heterogeneous autoimmune diseases like SLE. Our robust ML pipeline accurately distinguishes disease from control and identifies key predictive features, providing a powerful foundation for precision medicine efforts. 1 2 3 4 5 6 7 Metabolic SjS PsA UC Ps RA CD Controls ~27 000 Patients TB NS Healthy SLE IMIDs Chronic Kidney Disease CKD AKI SLE Decision Axis Diagnosis not SLE Predicted as SLE Sensitivity = 0.998, Specificity = 0.896 Number of Omics Features 0 20 40 60 80 100 -1 0 1 2 3 0.950 0.925 0.900 0.875 0.850 0.825 0.800 80 candidates for downstream analysis Integrated Evidence Sources ML score (f1-score) Optimal ML performance Genetic evidence Expression Data Literature Knowledge AI-powered review Evotec’s Molecular Patient Database E.MPD 19 Human Target ID & Drug Discovery (ASN23, Evotec TH-PO423) Ricardo Castro, Hendrik Urbanke, Philipp Skroblin, Jürgen Stumm, Maximilian Naujock, Mitja Mitrovic, I-Ju Lo, Michaela Bayerlova, Manuel Landesfeind, Sven Sauer, Antje Schmidt, Olivier Radresa, Uwe Andag Integration of Multi-Omics networks for Biomarkers Identification and Precision Medicine: an ADPKD case-study Epidemiologically, autosomal dominant polycystic kidney disease (ADPKD) is one of the most common monogenic human diseases. Mutations in PKD1 gene (TRPP1) account for 85% of ADPKD diagnoses while mutations in PKD2 (TRPP2) account for about 15% of all cases. ADPKD affects 1:500 to 1000 people, being 10 times more common than sickle cell disease, 15 times more common than cystic fibrosis, and 20 times more common than Huntington’s disease. Diagnosis is readily achieved through imaging and genetic testing. Tolvaptan, the only FDA approved treatment, contributes to clinically measurable improvements in kidney volume and kidney filtration rate (eGFR). Yet, adverse events linked to its primary mode-of-action, including high aquaresis and diuresis, together with adverse liver events have triggered a high 23% discontinuation rate in clinical trials. On average, PKD1 patients show renal failure at the age of 54.3 yr, while PKD2 patients show an onset of failure at the age of 74 yr. 3;4 By leveraging the E.MPD genomic data (whole exome sequencing), we confirmed that our PKD cohorts accurately recapitulate known predictors of diseases progression. Namely, it showcases both the well-documented differential effects of target gene on the disease onsets (PKD1 vs PKD2 gene mutations), as well as the increased severity of splicing mutations (putatively non-coding) compared to less-severe truncating mutations (Fig 1). 1 Gabow P (1993) Autosomal Dominant Polycystic Disease. N Engl J Med (329): 332-342 2 Torres VE, et al (2012) Tolvaptan in patients with ADPKD. N Engl J Med (367): 2407-2418 3 Harris PC & Torres VE (2009) Polycystic kidney disease. Ann Rev Med (60): 321–337. 4 Hateboer N, et al (1999) Comparison of phenotypes of polycystic kidney disease types 1 and 2. Lancet (353): 103 ADPKD epidemiology and disease progression Evotec’s Molecular Patient Database E.MPD 20 Enabling ADPKD Panomic research through multilayered networks Physiologically, ADPKD is characterized by the development of fluid-filled cysts on the kidneys (and liver/pancreas). While PKD1 and PKD2 are the main known risk factors, the precise determinants for the differential progression seen across patients are still largely unknown. Critically, the field is still missing a precise biological description of disease trajectories. The missing piece is a direct consequence of a lack of patient biopsies and associated molecular data, ideally followed-up over the time course of disease. With a specific objective to address this shortcoming, our Renal and Computational Biology Teams have established a bioinformatic workflow for the integration of multiple omics data from diverse sources into one integrated multi-omic network as part of our PanOmics offerings. The Team validated their findings using both real-life patient data and human iPSCderived in vitro models. In a resounding achievement, the integrated strategy, anchored in omics data and AI-driven computational analyses, offers a unique opportunity to conduct various types of analysis interrogating complex molecular mechanisms at play in disease progression. Such multi-omics network builds synergies between the input data layers, allowing to identify interactions previously inaccessible when studying each data set individually. With this new innovative tool at hand, our future partners will be well positioned to interrogate and elucidate key biological mechanisms underpinning ADPKD progression in patients (Fig 2). Kaplan-Meier plots show the renal event-free survival in patients carrying PKD1 and PKD2 mutations (50% survival at 55 years and 75 years, respectively). Fig 1. Calculation of renal event-free survival using genomic data from Polycystic patients and their characteristic mutations. Kaplan-Meier plots show the renal event-free survival in patients carrying PKD1 and PKD2 mutations (50% survival at 55 years and 75 years, respectively). 20 1.00 0.75 0.50 0.25 0.00 40 p = 0.013 60 80 Age (years) Strata PKD1 PKD2 Renal event-free survival 1.00 0.75 0.50 0.25 0.00 p = 0.038 Strata non-coding truncating Renal event-free survival Age ( years) Evotec’s Molecular Patient Database E.MPD 21 The multi-omics network is largely constituted by proprietary and exclusive data from ADPKD patients (E.MPD). Namely, blood RNAseq data and serum proteomics. It also incorporates the latest publicly available datasets on ADPKD patients at single cell resolution. 5;6 At a molecular level, we matched ADPKD patients with data from blood transcriptomics and serum proteomics by age, sex, BMI, eGFR and blood urea. We also included all currently available public datasets to integrate additional information on cell-specific expression of genes (snRNAseq) and proteins (LC-Ms/ Ms) associated with the cystogenesis. We generated similarity networks from the comparison between these “disease groups” and the “matched healthy controls” for each omic layer, while cyst proteomics data were annotated through high-confidence protein- protein interactions (PPI). Once all data layers were integrated into a network construct, we performed graph embedding using a random walk algorithm to reduce the dimensionality of the data and uncover biologically relevant associations. To identify functionally relevant multi-omic features, the inferred low dimensional network was then clustered and biological functions were annotated using over-representation analysis using Reactome7 as a reference database. As expected, the multi-omic network is enriched for genes with a known ADPKD association. It also contains communities associated with distinct disease-relevant biological pathways (Fig 3). 5 Muto Y, et al (2022) Defining cellular complexity in human autosomal dominant polycystic kidney disease by multimodal single cell analysis. Nat Commun 13:6497 6 Lai X, et al (2008) Characterization of the renal cyst fluid proteome in autosomal dominant polycystic kidney disease (ADPKD) patients. Proteomics Clin Appl 2:1140- 115 7 Milacic M, et al (2024) The Reactome Pathway Knowledgebase. Nucleic Acids Research (52): D672–78 Fig 2. Workflow for the integration of an ADPKD multi-omics network. Community detection & biological association Multi-omic network inference Graph embeddings for 2D Projection Single Omic Bulk transcriptomics, Proteomics, scRNA-Seq, etc. Single-Omic similarity network Intra-Omic networks + Inter-Omic networks (PPI. common features, …) Multi-layer network Evotec’s Molecular Patient Database E.MPD 22 Fig 3. Multi-omic integration workflow. No significant overlap Multi-layer network Multi-layered network generated from proprietary and public omic datasets. Multi-omic network inference multi-omic network inferred from the graph embeddings inferred from (A) snRNA-Seq (POD) snRNA-Seq (PT) Cyst proteomics NURTuRE Blood RNA-Seq NURTuRE Serum proteomic Community detection of the nodes contained in (B) each node represents a group of features. Groups with a significant enrichment for biological pathways are colored in shades of red. Two such groups are highlighted in boxes, in red “specific granule lumen” and in blue “integrin cell surface interactions”. POD = podocytes, PT = proximal tubules. Team overlap 0 0,25 0,5 Community detection and biological association snRNA-Seq (POD) snRNA-Seq (PT) Cyst proteomics NURTuRE Blood RNA-Seq NURTuRE Serum proteomic Evotec’s Molecular Patient Database E.MPD 23 Applied outcome: Biomarker identification and experimental validation Amongst several readouts of clinical relevance, the workflow enables the identification of novel biomarkers for early diagnosis of ADPKD. It can also be used to identify or prioritize candidate targets and biological pathways along a precision medicine discovery cascade (Fig 4). In a completely unbiased way, FN1 and LRP1 biomarkers were both identified from the inferred PKD network, showing strong B. High-resolution imaging of cystic human kidney organoids. hiPSC 2D specification 3D aggregate Organoid maturation Cyst formation Day 0 1 11 13 27 37+ A. Time course of human iPSC derived PKD1 KO organoids cystogenesis. Fig 4. Validation of FN1 and LRP1 biomarkers in human iPSC derived v PKD Kidney organoid and in the urine of ADPKD patients. association with the disease state (i.e. cystogenic multi-omic networks). These two candidate biomarkers have been subsequently validated using real-life evidence from a separate dataset of PKD patient as well as in human iPSC-derived in vitro models. Namely, both LRP1 and FN1 showed increased protein levels directly seen in the urine of ADPKD patients, but that stayed comparatively low in the matched controls (Fig 4 E). LRP1 and FN1 also showed a large and significant upregulation of expression in our in vitro human iPSC derived cyst-forming ADPKD Kidney organoids (Fig 4 B-D). Evotec’s Molecular Patient Database E.MPD 24 C & D. FN1 and LRP1 up-regulation in cystic-forming kidney organoids versus controls. E & F. Independent confirmation of increased FN1 and LRP1 levels in the urine of PKD patients compared to matched controls (age, sex and eGFR). LRP1 Scaled Average Expression 0 1 PTEC Endothelium Tubular progenitors Mesenchymal Mesangium Kidney progenitors LoH Podocytes Proliferating Control PKD1 KO no cysts PKD1 KO cystforming Percent Expression 5 15 25 FN1 Scaled Average Expression 0 1 PTEC Endothelium Tubular progenitors Mesenchymal Mesangium Kidney progenitors LoH Podocytes Proliferating Control PKD1 KO no cysts PKD1 KO cystforming 10 15 20 Percent Expression T-test, p = 0.0021 n = 6 n = 6 250 500 750 value Condition matched control cystic LRP1 T-test, p = 0.036 n = 6 n = 6 0 10000 20000 30000 value Condition matched control cystic FN1 In conclusion We have shown the integration of multiple layers of omics data and peer-reviewed information into a multi-omics network that recapitulates all available information on the molecular interactions associated with ADPKD progression. In the absence of complementary sources of molecular data from patients biopsies, such network fills a critical gap and can directly support patient stratification, prognosis and clinical trial design. It can also be interrogated to identify biological mechanisms linked to candidate targets or biomarkers of clinical relevance. Newly generated ideas should then be further validated using real-life patient data and the vast panel of translational platforms and innovative bioassays available at Evotec. Our joint teams offer competitive solutions in data collection, analysis and wet lab validation that are directly applicable to all discovery stages and the successful progression of clinical trials. We’ve already announced a series of innovative partnerships with pharma and biotech strategic collaborators that directly use these resources to open new avenues in precision medicine. Not the least, by means of its feature-centric nature, our workflow allows for the continuous extension and flexible integration of new datasets as they become available. With the everincreasing amount of patient data, such a holistic profiling tool offers a unique opportunity to unravel the full potential of molecular disease profiling. Evotec’s Molecular Patient Database E.MPD 25 PanOmics-Driven Drug Discovery in Cardiovascular Diseases Lavanya Muthukrishnan For decades, cardiovascular diseases (CVD) have been the leading cause of death globally. Despite recent advances with Semaglutide1 and Empagliflozin2-based anti-diabetic drugs, there is still a lack of effective treatments to fully reverse heart failure (HF). An improved patient stratification strategy is required to significantly augment the probability of success of future clinical trials. Molecular understanding of the health to disease continuum of human diseases will be essential to achieve this goal. Through our successful collaboration with Quality in Organ Donation biobank3, we obtained access to biopsies from heart, kidney and liver, along with clinical variables from thousands of donors. Our PanOmics platforms are generating high-quality cutting- edge bulk, single cell/nuclei and spatial omics data from such patient samples, building up Evotec’s Heart Atlas, using our innovative software solution, PanHunter4. An in-depth analysis of these patient data guided by strong disease expertise is leading to proper understanding of key disease mechanisms and the identification of cell type-specific novel target candidates. These target candidates are then confirmed and validated on Evotec’s state-of-the-art technology platforms, including iPSC/3D models. Combining Evotec’s Heart Atlas with data from other biobanks with CVD outcomes such as UK Biobank5 and NURTuRE6, leads to understanding of the interplay of the cardio-renal-metabolic axis at molecular, cellular and histopathology levels. Evotec’s Molecular Patient Database E.MPD 26 Fig 1. Evotec’s HCA Evotec’s HCA comprises 400 samples from 200 donors; 8 heart regions; 9 CVD subtypes including cardiomyopathies, myocardial infarction and aortic stenosis; 8 comorbidities including hypertension, obesity, diabetes, kidney disease; 2.5 million cells, 15 major cell types Heart Cell Atlas: an integrative single cell resource for human heart diseases Evotec’s Heart Atlas is a high-resolution, multi-omics CVD patient data source leveraging the power of both public domain and proprietary datasets. Our Heart Atlas has two major components: Heart Cell Atlas (HCA) is a single cell heart disease atlas, carefully curated by integrating multiple public human heart sc/snRNAseq datasets. The defining feature of the HCA is the integration of multiple CVD sub-types, which enables studying both common as well as unique disease patho-mechanisms at a cellular level (Fig 1). Another important and unique aspect of the HCA is the ability to study early disease insults like diabetes, obesity and hypertension in heart cell types e.g metabolic reprogramming and re-activation of fetal pathways. 7,8,9 We have observed a significant correlation between the molecular disease progression signatures of cardiomyocytes with clinical cardiac echo measurement, which demonstrates the clinical value of our atlas. We leverage the HCA in many ways, ranging from proper disease understanding and identification of key drivers of a disease, generating disease and cell typespecific signatures, to identification of innovative target candidates on cellular level. -15 10 0 -10 UMAP_1 UMAP_2 -10 -5 0 5 10 15 Adipocytes B_cells Cardiomyocytes Endocardium Endothelial_cells Epicardium Fibroblasts Lymphatic Macrophages Mast_Cells Neuronal NK_cells Pericytes T_cells VSMC Abbreviations: NK: Natural Killer VSMC: Vascular Smooth Muscle Cells Evotec’s Molecular Patient Database E.MPD 27 Heart Spatial Atlas (HSA) is a unique, proprietary spatial transcriptomics dataset generated using left ventricular heart tissues from healthy, diabetic, obese and CVD donors (cardiomyopathy, myocardial infarction, atrial fibrillation). The ability to study disease mechanisms at a high-resolution within the native tissue context makes spatial omics analysis attractive and valuable for prioritizing target candidates, since diseases manifest via interactions between multiple cell types in defined injury niches. 10,11 Fig 2. Spatial transcriptomics data on a left ventricular heart tissue from a dilated cardiomyopathy donor Vascular-endothelial Fibroblasts Cardiomyocytes injured Cardiomyocytes health) Macrophages Colors represent different tissue domains, white shapes highlight vasculo-fibrotic niches validated by Sirius Red staining. The molecular fibrosis signature is enriched exactly in the fibrotic tissue areas detected via histochemical Sirius Red staining. Furthermore, this damaged tissue area contains multiple injured cell states (e.g “cardiomyocytes injured”) as shown by spatial transcriptomics analysis. Heart Spatial Atlas: Linking histopathology with molecular disease signatures in cell types within tissues 1 https://www.nejm.org/doi/full/10.1056/NEJMoa2306963 2 https://www.nejm.org/doi/full/10.1056/NEJMoa2107038 3 https://quod.org.uk/ 4 https://www.evotec.com/en/panomics/panhunter 5 https://www.ukbiobank.ac.uk/ 6 https://nurturebiobank.org/ 7 https://academic.oup.com/nar/article/46/6/2850/4831081 8 https://www.jacc.org/doi/abs/10.1016/j.jacc.2019.07.076 9 https://www.nature.com/articles/s41392-020-00413-2 10 https://www.nature.com/articles/s41592-020-01033-y 11 https://www.nature.com/articles/s41392-022-00960-w Using Evotec’s sophisticated analysis workflows, we have identified specific tissue niches linked to disease severity that delivered novel target candidates for the initiation of drug discovery projects. To our knowledge, Evotec’s HSA is the only human spatial heart atlas capturing multiple CVDs and early heart disease insults like diabetes and obesity. We are excited to be able to provide an end-to-end solution for target discovery in CVD and are constantly striving towards a unique database by continuously expanding our heart atlas with both additional samples and omics modalities, e.g. proteomics. We are looking forward to implementing new technologies that have the potential to further improve data-driven target & biomarker identification in CVD. Summary Evotec’s Molecular Patient Database E.MPD 28 Uwe Andag EVP Head of Therapeutic Areas info@evotec.com Contact us Evotec’s Molecular Patient Database E.MPD 29 Editor Evotec SE Chief Editor Susanne Kreuter/Björn Steinhoff Design Alessandri, Design & Brand Manufactory Evotec SE (Headquarters) Essener Bogen 7 22419 Hamburg, Germany www.evotec.com