Enhancing QSP Model Creation and Validation with CYTOCON DB: A Standardized Repository for In Vivo Cell and Cytokine Baseline Concentrations
Vladislav Leonov1, Ekaterina Mogilevskaya2, Elita Gerasimuk2, Nikolay Pervushin2, Nail Gizzatkulov2, Oleg Demin2
1InSysBio UK Ltd, 2InSysBio CY Ltd
Objectives: Calibrating Quantitative Systems Pharmacology (QSP) models requires reliable comparisons between simulations and in vivo baseline data, such as cytokine and cell concentrations from patient tissues. However, researchers face significant challenges when integrating such data from scientific literature. These include variability in measurement units (e.g., mg/mL vs. cells/mm²), inconsistent patient demographics (e.g., age, disease stage), and diverse statistical reporting (e.g., medians, ranges). To overcome these barriers, we developed CYTOCON DB. The primary objectives are to (1) unify fragmented literature data into a structured, queryable format, (2) enable direct comparison of model outputs with standardized biological measurements, and (3) reduce manual effort in data extraction and standardization. Methods: CYTOCON DB (Cell and YTOkine CONcentrations Database) is an interactive web-based platform built on the ASP.NET MVC framework, utilizing Microsoft IIS, Microsoft SQL Server, Telerik Kendo UI, and Bootstrap for a robust and user-friendly experience. CYTOCON is a web application built on the ASP.NET MVC framework, supported by Microsoft SQL Server for data storage, and enhanced with Telerik Kendo UI and Bootstrap for user interface design. Key methodological innovations include: 1.Unit Standardization: Cytokine concentrations (e.g., mg/mL) are converted to pM, and cell counts (e.g., cells/mm²) to kcell/L, using predefined formulas embedded in the database. 2.Data Consistency & Transformation: All types of reported averages and dispersion (e.g., median, interquartile range) are normalized to mean and standard deviation (SD). Additionally, weighted mean and SD calculations enable more accurate comparisons. 3.Quality Control: A two-step workflow involves annotators extracting data from literature and reviewers validating entries. Automated outlier detection algorithms help identify potential misprints or inconsistencies in source data and reduce human error. 4.Dynamic Updates: The database evolves continuously, with more than 1,000 new concentration values added monthly from peer-reviewed studies. 5.API Integration: Direct access via R and Python scripts for querying data, generating plots, and integrating results into modeling workflows. Results: Database now aggregate 98,755 concentration values for cytokines and cells, curated from 2,637 scientific papers and public sources. To illustrate the capabilities of CYTOCON DB, we present two use cases: 1.IL-1a, IL-1ß, IL-6, and TNF-a concentrations in serum samples from healthy controls (HC) and SLE patients were retrieved, visualized, and statistically compared. Although mean concentrations of IL-1a, IL-1ß, and IL-6 were not significantly different between groups, TNF-a levels were notably higher in SLE (1.972±4.418 pM) than in HC (0.928±3.305 pM). 2.?ell abundance data—often reported as percentages of a parental subset (e.g., PD1+ CD8+ CD3+ cells relative to CD8+ CD3+ cells)—were transformed into absolute counts by leveraging averaged values from similar entries in the database. When applied to melanoma and NSCLC data, recalculated absolute counts (kcells/L) revealed fewer myeloid dendritic cells and natural killer cells in melanoma tumors compared to NSCLC (e.g., mDC: 1.5×10^6±2.9×10^6 vs. 1.1×10^7±4.6×10^6). Conclusion: CYTOCON DB is a standardized, user-friendly framework for integrating, transforming, and comparing baseline concentrations across diverse data sources, including both healthy and diseased states. By automating unit conversions, normalizing varied statistical measures, and providing rigorous quality control with dynamic updates, the platform significantly streamlines QSP model creation, calibration, and validation. Its robust API capabilities further promote seamless integration into computational workflows, fostering reproducibility and deeper insights into a wide range of disease contexts.
1. Leonov, V., Mogilevskaya, E., Gerasimuk, E., Gizzatkulov, N. & Demin, O. CYTOCON: The manually curated database of human in vivo cell and molecule concentrations. CPT Pharmacometrics Syst Pharmacol (2022) doi:10.1002/psp4.12867 2. “CYTOCON DB Open.” Accessed April 21, 2021. http://www.cytocon-open.insysbio.com/. 3. “CYTOCON DB.” Accessed May 10, 2025. http://www.cytocon.insysbio.com/.