SS P08 18March2014 Standard Scripts Project 08 Meeting
When: 17-18 March 2014
Place: Silver Spring, MD (PhUSE CSS Break-out Session)
Facilitator: Mary Nilsson
Scribe: Mary Nilsson
Attendance: Wei Wang, Simin Baygani, Sheryl McCoy, Mercidita Navarro, Mary Doi, Jane Diefenbach, Terry Walsh, Kim Musgrave, Francois Vandenhende, Frank Senk, Sasha Ahrweiler (part), Steve Wilson (part)
- Face-to-face discussion about white paper topics
White Paper 1 (Central Tendency for Vitals, ECGs, labs):
- Talk about timing for version 2
- How to get people more comfortable with box plots. (Add box plot explanations in CSRs?)
White Paper 2 (Outliers/Shifts for Vitals, ECGs, labs):
- Defining reference limit - performing lab vs large clinical trial based limit
- Ordering of labs
- Baseline definition
- If time, start discussion around integrated display (e.g., create forest plot for all measurements vs special topics only?)
White Paper 3 (Adverse Events):
- Definition of treatment emergence (Probably have to save for after the conference, loop in ADaM? ADaM does have some example derivation)
- Should standard tables include MedDRA HLT?
- CTCAE grade - use of it outside oncology?
- What to do (if anything) with relatedness information.
- Anything with number of events? (Many times difficult given collection methods)
White Paper 4 (Demographics/Disposition/Medications):
- Demographics - what should always be included?
- Combine demographics with other baseline (disease characteristics), or separate tables, or doesn't matter? [How to deal with this for Script-athon]
- ATC codes for medications - needed? If desired, how to implement mapping (all versus one)?
- Note: It appears the FDA used to require ATC on submission dataset but dropped it in the last guidance (Maybe Steve can help find someone to give us clarity on this)
- Need any medication summary for all medications, or just meds of interest?
- Disposition - How to handle studies that allow subjects to stay in study (following normal Schedule of Events) after stopping treatment. (Not sure we'll be ready for this discussion)
White Paper 5 (Hepatotoxicty):
- Everyone OK with the eDish plot?
- Do we need an edish plot with AST?
- Individual patient display - review a couple options.
- Discuss some options for the 3x/5x/10x tables.
- How to incorporate baseline?
- Standard SMQ layout (for this topic and applies to other topics)
- Discuss ways to address AE data and lab data alignment and/or interpretation.
White Paper 6 (PK):
- Gap analysis related to process map for SDTM/ADaM PP and PC datasets creation
- Labels and calculation rules for PK parameters; development of standard scripts for NCA?
- PK TFLs: urine samples; graphics for PK parameters; any addition shells
- Should we include statistical inference shells/methods in a subsequent version of WP?
- Publication strategy for recommended guidelines in scientific journal.
Note: The disclaimer is in effect. All of this represents the views of participants and not necessarily the view of the organization in which they work. Also, these should be considered DRAFT until reviewed by attendees.
Mary Doi and Eileen Navarro are FDA Medical Liaisons for our project.
FYI - The FDA does have plans to create a ISS guidance. ISE guidance is in development now.
We had wonderful discussion and a great mix of expertise among the attendees!
Discussed scope of white papers. Essentially, the scope includes information that would normally be included in a Statistical Analysis Plan plus the table, figure, and/or listing shells. "TFL requirements" is currently considered out-of-scope. Examples of things that would be in TFL requirements but not in a SAP - font size, margins, portrait vs landscape, which variables to use from the datasets. Detailed information that might be required for performing the recommended analysis/display will generally be in scope but the details on whether that will be done via an ADaM variable versus TFL programming is considered out-of-scope. We would like to include recommendations for "difficult/controversial" decisions related to details around an analysis/display. In the cases where we really can't come up with a recommendation, a textual paragraph describing some of the issues going into the decision would make sense. Alternatively, some of these "difficult/controversial" decisions can be stated as being out-of-scope for version 1 of the white paper. The idea is that we are the group to start tackling some of these, but we realize we can't tackle all of them in our version 1's. The introductions in the white papers can likely be improved to clarify scope.
Discussed the need for a medical writer. Mary Nilsson will continue to work through the PhUSE Steering Committee. There was a suggestion that perhaps a medical writer can be hired through TransCelerate (David Jones contact). We can all still ask around.
Discussed titles of white papers. We agreed to change "Submission Documents" to "Summary Documents" based on feedback on the Demog/Disp/Med white paper, but applies to all of them. We also discussed a comment to say Tables/Figures/Listings instead of Analyses and Displays. We had TFLs at first and changed it to Analyses and Displays based on early feedback (to make it more obvious that it applies to the analytical methods). We decided to stay with Analyses and Displays.
Discussed Agilewords and where to put comments. This discussion was during the break-out session with the entire working group, but putting notes here for convenience. Agilewords allows to see all comments. Not sure how long the versions are kept though. May need to keep a copy if want to keep comments. However, many of our comments are coming via email. These comments (or a high level summary of the main comments) can be put in the Discussion section of the wiki page. White paper leads can make a judgment call on this. No need to feel obligated.
Discussed the Outliers/Shifts white paper.
Discussed reference limits. Wei and Mary Nilsson outlined the difference between limits for individual patient care purposes (sensitive is good) versus for Drug vs Comparator purposes (sensitive limits may not be ideal) per research done by our company with an internal medical subject matter expert. The concept is also covered in the FDA safety reviewer guidance but covered in general and not in the context of lab limits. It appears many companies use the limits supplied by the lab only. At least one company uses multiple limits - from the lab, marked abnormality, and clinical relevant change. Discussed the possibility that Tox Grades (already supported in SDTM) as a possible method for getting wider limits (more specific). The white paper will likely need to simply discuss some of the options and not make a specific recommendation in version 1. If and when a position can be agreed upon, there was some thought that it would belong in CDISC as limits are usually included in SDTM. Phil Pochon from CDISC can be considered someone to loop in (involved with labs).
Discussed ordering of labs in reporting. There's general agreement that things should at least be separated by high level categories (chemistry, hematology, urine). There was also general agreement that additional subcategories would be useful but perhaps harder to implement. Jane mentioned that SDTM has CAT and SCAT. CAT is defined by SDS team. SCAT (for subcategories) would currently be defined by sponsors. The CDISC lab group does not plan on taking this on. The white paper may be able to come up with a recommendation for SCAT (e.g., as an appendix to the white paper). Terry has a start on ordering and will forward to Wei. Wei also has a bit of a start from a project.
Discussed defining baseline for purposes of the scatterplot/shift/TE display. The white paper currently proposes max baseline when max during treatment is of interest and min baseline when min during treatment is of interest. This would be as opposed to "last" baseline. Nancy created max and min baselines in the example data for the Scriptathon. John Troxell did propose max/min for ADaM (Intrachange conference presentation). It appears that max and min baseline is not a foreign concept for some companies, however, last baseline is still commonly used. It appears that there isn't an objection to keep the white paper as it is on this topic, recognizing that nothing is "required".
Discussed Adverse Events White Paper.
Discussed different ways to summarize AEs. MedDRA PT sorted by decreasing frequency seems common and useful. MedDRA PT sorted within SOC seems common and it is expected per ICH guidance and in the FDA safety reviewer guidance. It seems uncommon to utilize HLT and HLGT across companies (except when it's included as part of an adverse event of special interest, and/or per therapeutic area specific). However, such displays are easy for the FDA via their MedDRA at a glance tool. So, such exploratory looks at the data (all HLTs and HLGTs) might be common. It also appears looking at all SMQs might be common with FDA's tools (not common across industry). They can look at primary and secondary coding. They can easily look at all of these within subgroups as well. However, it's a separate decision on whether to include such looks in their reviewer report. They might do this thorough exploratory work for the pivotal studies, by study (as opposed to in an integrated manner). For the white paper, it is likely that mock shells won't include these. A textual paragraph can be added about the idea that such explortory looks at the data may benefit safety signal detection and that a tool would be useful as opposed to the creation of a whole bunch of tables. We do think a SMQ table (assuming to be used for topics of interest) can still be in scope for the white paper.
Discussed CTCAE grades versus severity (mild/moderate/severe). This would be a table that displays max severity across the treatment period. It's acknowledged that CTCAE grades are widely used in oncology. There are examples where they are used outside of oncology (e.g., Sascha is aware of an RA example). The white paper can mock-up both. The idea is that a particular compound will likely use just one of them - no expectation that both would be needed. The white paper can include a discussion about when one might be used over the other - likely CTCAE is used in populations where you might expect a lot of serious AEs. When CTCAE grade is used, what to do for events that don't have CTCAE grades would need to be specified. We did not get into this in our discussion. Analytical planning should include what to do with missing grades/severities. Discussed that if and when p-values or CIs are included, it would likely just be for comparing percentages of a severe event. Comparing percentages of just mild or just moderate would not be useful. (This led to a discussion about p-values and CIs in general. As a reminder, see the final white paper on central tendency for the p-value/CI paragraph. We are not trying to resolve the debate - We plan to include them in our mocks and for those compound teams that don't find them useful, they can be deleted.)
Discussed AE relationship. Appears it's common to repeat AE table with relatedness. Such a table is generally expected by at least one regulatory agency. The white paper may need to say that analytical planning would need to be specific on when relatedness is missing (if allowed in collection) and when relatedness is unknown (if allowed in collection). Sounds like collection is not very standard in this regard. Generally missing and unkown might be able to be treated as "yes". However, if there are a lot of missings or unknowns, treating them as "yes" may not be very useful. (Relates to the concept that being more conservative from a total count perspective may actually make it more difficult to see treatment vs comparator differences. You can miss signals by adding too much noise.)
Discussed number of events briefly. Acknowledged this is challenging due to collection considerations. (Collection varies - sometimes you get each headache, sometimes you get one "intermittent headaches".) Appears there is variability on if and when this is created. Probably easier for solicited AEs. Some only do this in an exposure-adjusted analysis. We believe such a summary is required at least for CT Registries. We will need to discuss this further.
Discussed Demog/Disp/Med White Paper.
Discussed single table vs separate tables for demographics and baseline characteristics. General agreement that it is likely best to have the demographics remain separate so it can have a similar appearance across studies. May help with standardization. However, we would still prefer that programming is flexible to add more. For the white paper, there can be a sentence or two that it should be assumed that a baseline characteristics table would be created, but is out-of-scope for the white paper.
Discussed what should be in the demographics table. Demographics probably should include weight since it's in the FDA safety reviewer guidance. The text can mention that BMI and BMI category can be considered. Text might be needed to address cases where you might want region instead of country.
Discussed medication classifications. Mary Doi generally likes medication classes - helps with looking at the data in various ways for exploratory purposes. Indication might be the most helpful though. There was discussion around coding indication that might end up giving us more value than anything else. We can discuss further! Acknowledged that getting ATC coding to be done correctly is quite the challenge. Coding is not 1:1 and requires medical/pharmacy skills. Noted that ATC classes and subclasses are not in SDTM. The white paper probably won't incude ATC coding in the shells.
Discussed Hepatotoxicity White Paper.
Discussed baseline categories. These would need to be incorporated in the tables somehow.
Discussed need for AST vs Total Bili (in addition to ALT vs Total Bili). FDA's Jumpstart package includes this. Mary Doi looks at AST in addition to ALT, and they can filter on Alk Phos <=2. In the past, Bob Temple has made the point that these can not be used in isolation (Confirm that my notes are correct on this).
Discussed the individual patient display. We might be able to use the standard patient profile plot. (Wei can forward.) There might be one in CTSPedia. We did not discuss yet which subjects would require an individual patient display. Mary Nilsson can forward some advice from an internal subject matter expert.
Discussed SMQ layout. Mary Doi mentioned that it's usually the sub-SMQs that seem the most useful. Mary Nilsson gave Terry a list from Lilly, but currentlly unsure if that is really directly from the liver guidance. Further discussion is needed.
Discussed situation when you have data from multiple sources (e.g., AEs and labs). Hepatotoxicity is an example, but applies to other topics. There was some discussion on possibly displaying a summary from the multiple sources in a display. Wei has an example she can forward. Terry has an example in the slides he passed out at the conference. There's general agreement that such a display may be very useful! Further discussion is needed. It will likely get addressed in the liver white paper only for now. Version 2's of the other white papers might be able to add something.
Discussed other miscellaneous topics. It was noted that some companies have a follow-up form that is completed for subjects meeting certain criteria. Some discussion about this may be warranted in the white paper. Mary Doi mentioned there's a DILI meeting going on later in the week. We should try to get a summary from someone. Sounds like they might be discussing the idea of 3X baseline. Also, we need a different name for the eDish plot - eDish really refers to a software package - it's not a nickname for the plot (news for some of us). Maybe hepatotoxicity grid? (Post-meeting note from Mary Nilsson: eDISH stands for Evaluation of Drug-Induced Serious Hepatotoxicity?)
Discussed addition of a QT Studies White Paper.
Francois mentioned that he is aware of someone who would be willing to author a QT Studies white paper. Even though our intent is to focus on finishing the white papers we have in progress, we're not going to say "no" to anyone willing to get another one started! Of note, QT Studies is part of the scope for TransCelerate. We'll need to collaborate closely with them.
- Last revision by Jmbodart,10/13/2015