Page History

...

Here is a mock example modeling his data:

Background Information:

1. This was sent to the lab team for discussion and we decided that the INF-y response tests should be modeled in IS (as agreed to with the same TB TAUG INF-y response tests here: https://wiki.cdisc.org/x/azXFBg).
2. Craig said mentioned that there are NO other variables to capture the units and dosing concentrations for the test agents that are added to the samples. In fact, this is a gap in the CDISC model that we lack variables (or a domain) to help mapping in vitro drug/non-drug exposure data. The EX/EC and AG domains are used for mapping in vivo drug and non-drug therapeutic and other types of agents exposure data (respectively). These domains have variables for mapping dosing concentrations and units, but we can't use them for in vitro exposure data. Craig said he would .  One option is to pre-coordinate the concentrations and dosing units of the added agents added into the ISCNAGT (test condition agent) variable, see the texts in red in the above example.
   • this This is a problem because a single CDISC domain variable should be designed for a single meaning and a single purpose. The Test Condition Agent variable is used for the name/identity of the agentadded study/control drugs or other materials. I think we would be breaking rules to also include the concentration and unit of the agent in this variable as well.
3. While brain storming Stefan's use-case, I remembered that in the MS domain, there are MSAGENT (Agent Name), MSCONC (Agent Concentration), MSCONCU (Agent Concentration Units) variables that were developed added back in SDTMIG v3.2 to represent the study and control drugs used for microbial resistance testing. The 3 variables, MSAGENT, MSCONC, and MSCONCU are used to map the name of the study/control drugs, their concentrations and units respectively. Basically, for these MS tests, you dump antibiotics study drugs, likely an antibiotic (at varying concentrations) on a plate to see if the bacteria would be resistant or susceptible for the drug by measuring a few specific outputs, at differing drug concentrations. Here Below is a MS example from IG3.4 :

4. Here is an MS example from IG3.4:

After E. faecalis was identified in the subject sample, drug susceptibility testing was performed at each of the labs using both the sponsor's investigational drug and amoxicillin. Because an identified organism is the subject of the test, the NHOID variable is populated with "ENTEROCOCCUS FAECALIS". Between the 2 labs (MSNAM), a total of 3 susceptibility testing methods were used: epsilometer, disk diffusion, and macro broth dilution (MSMETHOD). Epsilometer and disk diffusion both use agar diffusion methods, in which an agar plate is inoculated with the microorganism of interest and either a strip (epsilometer) or discs (disk diffusion) containing various concentrations of the drug are placed on the agar plate. The epsilometer test method provides both a  minimum inhibitory concentration (MSTESTCD = "MIC"), the  lowest concentration of a drug  that inhibits the growth of a microorganism, and a qualitative interpretation (MSTESTCD = "MICROSUS") such as susceptible, intermediate, or resistant. The disk diffusion test method provides the diameter of the zone of inhibition (MSTESTCD = "DIAZOINH") and a qualitative interpretation such as susceptible, intermediate, or resistant (MSTESTCD = " MICROSUS" ). The quantitative and qualitative results are grouped together using MSGRPID.

...

1. to show how the 3 variables are used together with the MSTESTs:

PowerPoint Illustration for the in vitro testing exposure data problem:

Questions/Discussion Needed:

I want to run this by the team to see if we actually have a "duplicated purpose variable" issue here. we can take advantage of the existing standard variables? Do we also have "duplicate variables" in different domains that serve the same purpose?

1. Are we using Is the Test Condition Agent (ISCNDAGT) variable the same as the Agent Name (MSAGENT)? Do we use them for the same purpose? - they
   • They both are used to map the name of the thing/entity (e.g. study/control drug, stimulating test materials like TB antigens
   ) you add
   • , etc) that are added to a sample to "induce" a response.
   • The "induced" response can be stimulatory (in immune response) or inhibitory (in bacterial resistance response).
2. If yes, do we need to merge these two variables? Seeing as we are making changes for IG4.0, this might be worth considering, and will need multiple team's review and approval.
   • This also means there would be one less standard variable out there (as people complain that there are too many new variables in IG3.4).
   •  
3. As a separate conversation, can we Can we also then extend the use of the standard variables --CONC (Agent Concentration), --CONCU (Agent Concentration Units) to the other specimen-based domains for ? For mapping concentration and unit of the in vitro test agent exposure agents? There are now use-cases needing these variables in IS, MS and CP. Right now --CONC and --CONCU are limited to MS only. They also only qualify --AGENT.
   • This way we can take advantage of existing standard variables, instead of creating new ones.
   Other concerns....
   • This also means we don't need to tell users to pre-coordinate unit and dosing concentration of the exposure agent into the same variable - which would be breaking rules.

MB-IS Team/ISMT Final Recommendations:

1.We need a variable to map the added agent: drug, diagnostic agents, other substances (used to induce a in vitro response). Sometimes this is all we need, and the dataset stops here.

• • There are two variables now serving the same function: CNDAGT and AGENT. – Yes true, but that ship had sailed, we can't do much at this point without causing significant changes.
  • Can we merge them? One less variable people have to worry about. – No to this one because the kind of changes required to replace AGENT (MS) is too large and not worth people going back to fix their data. The –TSTCND (test condition) and –CNDAGT (test condition agent) variables can be available for all the other specimen-based domains when needed, EXCEPT, the --CNDAGT (test condition agent) variable should NOT be used in MS (until use-cases arise that require both AGENT and CNDAGT to be needed for MS, we haven't seen any right now). We need to update their scopes in the newest SDTM Model.

2.We need a variable for the agent concentration.

• • MSCONC works but this is limited to MS.
  • Expand it to other specimen-domains, no need to create new variable. – Yes, we agree to extend --CONC to other specimen-based domains. Meanwhile people can use it as a NSV for their respective domains if they are using IG3.2-3.4. We may aim for IG4.0 for this change. If ISMT/GGG agrees, we need to update the variable Role, Notes and Definition so that it qualifies --CNDAGT as well. Also CONC is already a standard variable, we can just take advantage of it.

3.We need a variable for the agent dosing unit.

• • MSCONCU works but this is limited to MS.
  • Expand it to other specimen-domains, no need to create new variable. – Yes, we agree to extend --CONCU to other specimen-based domains. Meanwhile people can use it as a NSV for their respective domains if they are using IG3.2-3.4. We may aim for IG4.0 for this change. If ISMT/GGG agrees, we need to update the variable Role, Notes and Definition so that it qualifies --CNDAGT as well. Also CONCU is already a standard variable, we can just take advantage of it.

If we extend CONC and CONCU to other specimen domains...

Here is a mock example of Stefan's data using --CONC and --CONCU: