Camille Vong (1), Alison Margolskee (2), Fariba Khanshan (4), Andrew Stein (2), Theodoros Papathanasiou (1), Yu-Yun Ho (3) and Michael Looby (1)
(1) Analytics, Novartis Pharma AG, Basel, Switzerland (2) Analytics, Novartis Pharma AG, Cambridge, MA, USA (3) Analytics, Novartis Pharma AG, Cambridge, East Hanover, NJ, USA
Objectives: Checklists are useful tools not only for beginners, but also for experts in any field. As John D. Cook stated: “We resist checklists because they insult our intelligence, and yet they greatly reduce errors. Experienced people in every field can skip a step, most likely a simple step, without some structure to help them keep track.”[1]
As pharmacometricians, we sometimes jump into complex modeling before thoroughly exploring our data. Exploratory plots help uncover issues with data, useful insights, or aspects to explore further, and can sometimes answer questions without the need for complex models.
To help pharmacometricians remember the key steps in exploring PKPD data, we created the PKPD Exploratory Graphics (xGx) Cheat Sheet [2], a 1-page (front and back) question-based reference sheet, designed around the principles from the Exploratory Graphics (xGx) tool, an open-source R-based tool available on GitHub [3].
Methods: The key steps in dose-exposure-response exploration were organized by structured PK, PD, and PKPD exploration. Key questions and example graphs were created to accompany each step. Guiding Principles from xGx were simplified and organized into Technical Considerations. Commonly used functions from the xgxr package[4] were compiled along with brief descriptions of their usage. Finally, a high-level checklist was written to summarize the key steps in PKPD data exploration. All of these components were streamlined and optimized to fit into a 1-page cheat sheet.
Results: Key steps in PKPD exploration include:
- Identify data type and choose appropriate graph types (PD)
- Identify axis scale that reflects distribution of data (PK, PD)
- Provide an overview of the data (PK, PD)
- Determine whether data corrections are needed (PD)
- Assess trends over time (PK, PD)
- Assess trends by dose (PD)
- Assess PK linearity (PK)
- Assess extent and sources of variability (PK, PD)
- Get an overview of the relationship between exposure and response (PKPD)
- Explore delays between exposure and response (PKPD)
Technical considerations include: extent and sources of variability (between vs. within subject, explained vs. unexplained), PD data corrections (e.g. baseline correction), and effective use of scales.
Useful functions include: xgx_theme(), xgx_geom_ci(), xgx_geom_pi(), xgx_scale_y_log10(), xgx_scale_y_reverselog10(), xgx_scale_y_percentchangelog10(), xgx_annotate_status(), xgx_annotate_filenames(), and xgx_save_table().
Conclusions: The PKPD Exploratory Graphics (xGx) Cheat Sheet was created to help pharmacometricians follow a structured approach to PKPD data exploration. The key steps to PKPD exploration along with questions and example graphs were compiled into a 1-page reference sheet. The cheat sheet also includes technical considerations, useful functions from the xgxr package, and a high-level checklist. Cheat sheets, similar to checklists, can be useful for people of all levels of experience, from beginners learning PKPD, to experts who can sometimes skip simple but important steps.
References:
[1] https://www.johndcook.com/blog/2019/12/03/distracted/
[2] https://opensource.nibr.com/xgx/Resources/PKPD_Exploratory_Graphics_(xGx)_Cheat_Sheet.pdf
[3] https://opensource.nibr.com/xgx/
[4] https://cran.r-project.org/web/packages/xgxr/index.html
Reference: PAGE 30 (2022) Abstr 10144 [www.page-meeting.org/?abstract=10144]
Poster: Methodology - Other topics