QT subgroup summary notes (cumulative)

[mli1]

Notes from Mar 2 2021 meeting:
1- There is a Cardiac Safety Research Consortium which is very influential (website is https://cardiac-safety.org/): Members are from the industry, regulatory agencies and academia. Deliverables are standard tables, ECG dataset, White paper, etc. We may benefit from getting to know what they have done and see what else we can do, in terms of better visualization
2- Within FDA, there is a dedicated QT group. Veronica is contacting the project manager to see if anyone might be interested to join our effort developing generic visualization tools that can be used by the entire industry.
3- Team agreed this group to focus on how to develop a tool to visualize selected cardiac safety data at an aggregated level and also at individual level.
4- Types of cardiac safety data includes: ECG (QT data, corrected by HR, with various formula which can be built upfront; Many ECG reading came with QTcF, QTcB, team can also build QTcI into the tool for users to choose); cardiac function data (for example, cardiac echography); other events that are associated with cardiac safety, for example, syncope. Team will focus on ECG data first.
5- Team agreed the tool should be applicable to all the phases and all therapeutic areas, with the potential to use the full set of the tool, or only part of the tool, depending on the availability of the data. For example, QT data with PK data and prebuilt modeling work: if PK data is not available, then this set of plots will not be shown.
6- Some questions were also raised: how to deal with triplicate ECG data (by the mean or what is the best way??); Holter data (extract certain timepoints so they can be fit into the prebuilt tool); food effect on QT (this might need a standalone plot, as it does require certain structure of the data, including the food intake data)
7- Team to get familiar with ICH E14 Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs and FDA Pre-Marketing Risk Assessment Guidance 2005
8- To recruit more experts into the team. Veronica to reach out a true leading expert who has been working on renal and cardiac safety (extensively involved in PhUSE and leading ADBMI??), can also contribute to renal explorer development. Mengchun to reach out to Jonathan Seltzer as he is a working group member and a cardiologist by training, and he was known to have involved in the Cardiac Safety Research Consortium.

[mli1]

Notes from 24 Mar 2021:
1- Jonathan is working within CSRC to assemble an expert team who can join ISG for a 1 hour introductory meeting to learn what we have been doing in general and our thoughts on interactive cardiac safety graphics tool. And we can start from there if a collaborative partnership should be formed, with strong expertise input from CSRC. (Thank you so much Jonathan for your great effort!!!)
2- Here are a few things, as we discussed, would already need expertise inputs:
a. Team agreed to start with Rich’s static QT graphics. When designing the standard interactive QT graphics, one question is how to deal with “multiple values”. Learning from interactive eDISH, the default setting is, each dot on the graphics is the worse value per subject, after click on the dot, the whole trajectory will show how things have evolved with time. If interactive QT graphics is to follow a similar idea, then the team needs to agree on rules dealing with the multiple values per visit (for example, triplicates ECG). Based on team members’ experience, it can be done in different ways, we can look into mean/medium over the time course(big picture), sometimes we do need to take the most extreme values and look into them (worst case scenario). This is something we can ask for CSRC’s input.
b. Exposure Vs QT. With the platform V2.0 coming out soon, it will be very easy to incorporate ECG data with exposure data (if available). Key PK parameters such Cmax/Cmax at steady status should be examined with the worst QT. from aggregated level, we should also build in capacity to check the QT response among the different quartiles. A grouping function might be applied (different quartile with different color, for example; or with different panels representing different quartiles)
c. Tables are nice too. Similar to the table we have for eDISH, we can also build in some default table, for example, how many were above 500ms, how many had delta >60ms etc.
d. Food effect as positive control (if a trial can pick up food effect then a positive control such as moxi might be not needed). See attached literature shared by Barbara. [ Post meeting note based on follow up talk between Mengchun and Jim: with food effect, be aware the heart rate might change, in which case the QT correction formula will be applied. In a paper Jerry and Mengchun co-authored, in certain therapeutic area, heart rate will decreased when the underlying disease is under control, which leads to the observation that even in the SOC arm (where none of the agent was known to cause QT prolongation, QTcF was still presenting an increase over time). This might be a good topic for CSRC advice as well.

Note: all the discussion points above are just team member’s personal opinion, not representing his/her organization’s position on each scientific question.