Valentin Legras & Justin Wilkins
F. Hoffmann-La Roche
Objectives: Cytokine release syndrome (CRS) is caused by substantial, rapid release of cytokines into the circulation from immune cells, often triggered by various kinds of immunotherapy, such as monoclonal antibodies (mAbs). The primary aim of this work was to apply sunburst plots to visualize the effect of Tocilizumab, on CRS induced by Glofitamab (a mAb) over time. We sought to validate the FDA’s requirement for this specific type of plot to provide a clear and comprehensible display of Tocilizumab’s impact on CRS. An additional objective was to prepare an appropriate input dataset, which entailed substantial data processing.
Methods: We employed sunburst plots to represent and analyze the clinical data collected. The hierarchical structure of the sunburst plot was used to represent various aspects of the study data that underwent change within individuals, such as time. A significant portion of our work was dedicated to data preparation, resulting internally in the creation of an Analysis Data Model (ADaM) dataset (ADCRS) that, though not a standard yet, could potentially become an internal standard in the future. We also explored the utility of similar plots such as sankey and alluvial plots, noting their differences and applications.
Results: The sunburst plot revealed that the application of Tocilizumab resulted in the reduction of CRS events in comparison to the absence of any drug administration, which demonstrates the drug’s superior safety profile. To illustrate, we focus on a population who had a CRS Grade 1 on day 8. Out of these patients, 11 patients were given Tocilizumab post their CRS diagnosis, while 80 patients weren’t. By day 15, all the patients who were treated with Tocilizumab no longer exhibited symptoms of CRS whereas only 44% of the non-treated patients showed cessation of CRS symptoms. We also demonstrate that sunburst plots can be successfully applied in pharmacometrics, for instance, to represent patient responses to different drug dosages or exposures over time. Compared to sankey and alluvial plots, the sunburst plot enables tracking of how CRS events evolve over time within groups of patients. However, sunburst plots can become more difficult to read as the number of levels in the hierarchy increases. The advantage of sankey and alluvial plots is that they provide a clearer overall view of the evolution of CRS grades over time.
Conclusions: This work demonstrates that sunburst, sankey and alluvial plots are effective tools for visually presenting complex, dynamic longitudinal data in an understandable manner. The clear display of the reduction in CRS adverse events with Tocilizumab in this example validates its potential use in supporting dosing strategies. The labor-intensive process of creating an effective ADaM dataset was a significant part of our methodology, indicating the potential for this dataset to become a standard. As experience with those data visualization tools grows, we anticipate that they will find additional applications in other contexts within pharmacometrics.
J. Stasko and E. Zhang. Focus+ context display and navigation techniques
for enhancing radial, space-filling hierarchy visualizations. In Information Visualization, 2000. InfoVis 2000. IEEE Symposium on, pages 57–65.
IEEE, 2000.
https://people.cs.vt.edu/~north/infoviz/sunburst.pdf
Reference: PAGE 32 (2024) Abstr 11030 [www.page-meeting.org/?abstract=11030]
Poster: Methodology - Other topics