IS Domain Scope Update for the SDTMIG v3.4: A Development History and the Difficulties of Standardizing Complicated Biological Processes

Section 1: An expanded scope and new definition for the IS domain in the SDTMIG v3.4

The current Immunogenicity Specimen Assessments (IS) domain in the SDTMIG v3.4 is designed to represent data pertaining to specimen-based assessments that measure the “presence, magnitude and scale of the immune response upon an antigen stimulation or encounter.” Not only does the new domain definition better align with the scientific definition of “immunogenicity”, but it also expands the scope of the IS domain from the previous versions of SDTMIG (i.e., v3.2 and v3.3), where the IS domain was defined to represent data pertaining to “assessments that describe whether a therapy provoked/caused/induced an immune response”.

The “antigen” (of interest) in the updated IS domain definition is any entity capable of inducing a host immune response, and may be (but is not limited to), drug/test article (i.e., study/non-study therapy), allergen, microorganism (e.g., bacteria, virus, fungi, parasite, etc.), self-antigen (i.e., autoantigen), and others.

The assessments that “detect the presence of, and quantify the magnitude and scale of the immune response upon an antigen stimulation or encounter”, and are in scope for inclusion in the IS domain per the SDTMIG v3.4, are the following:

1. Activated immune cells, specifically and mainly (but are not limited to) those measured via ELISPOT:

a. Cytokine-secreting cells

b. Antibody-secreting cells

2. Activated immune cell products:

a. Antibody (i.e., humoral immune response) -

i. Antidrug antibody.

ii. Anti-microbial antibody for both microbiology and vaccine studies.

iii. Anti-allergen antibody, for allergy studies.

iv. Anti-autoantigen antibody (also called autoantibody), for autoimmune disease studies.

v. Others, such as alloantibody for transplantation studies.

3. Other currently unclassified immune response tests.

More details and description of the changes and updates made to the IS domain in the SDTMIG v3.4 can be found in section 6.3.5.5 Immunogenicity Specimen Assessments (IS), specifically, assumptions 2 and 3.
 

Section 2: Problems that led to, and rationales behind, the IS domain scope update for the SDTMIG v3.4

Domain and variable level limitations

Prior to the release of the SDTMIG v3.4, most specimen-based immunogenicity assessment data had been mapped to the Microbiology Specimen (MB) and Laboratory Test Results (LB) domains, which do not have the sufficient structure and standard variables to support the meaningful representation of such data. As a result, various supplemental qualifiers had to be created to map key immunogenicity testing information in both LB and MB.

• Note: The IS domain in the SDTMIG v3.4 has new standard variables for the stable and consistent representation of the key information.

The MB and LB domains also employ a pre-coordinated TEST/TESTCD values modeling approach, and there is a lack of clear definition and boundaries on what is considered as the “analyte” under assessment, which is typically mapped to the --TEST/--TESTCD variables. Consequently, too much information is pre-coordinated into the TEST/TESTCD values, overloading the --TEST/--TESTCD variables. In addition, CDISC Conformance Rules dictate that <=8-character TESTCD/PARMCD and <=40-character TEST/PARM values be available for data submission to regulators. Because of the 8- and 40-character limit, LB and MB TEST/TESTCD values are heavily abbreviated, and in some instances, TESTCD values are replaced with the NCI C-codes to ensure uniqueness, which significantly undermine their meaningfulness and clarity, and further make data/dictionary management of the TEST/TESTCD values difficult.

• Note: the updated IS domain in the SDTMIG v3.4 takes a post-coordination modeling approach, with the introduction of several new variable qualifier standard variables to help to resolve the overloading of the --TEST/--TESTCD variables issue.

Multiple SDTM Findings domains were utilized to represent data of a common topic, due to changes in SDTMIG versioning and the limited IS domain scope

The specimen-based assessments that measured the presence of and/or quantify the magnitude and scale of the immune responses, and are in scope for the IS domain in the SDTMIG v3.4, had historically been mapped to various specimen-based domains, such as LB, MB and/or IS.

More specifically:

1. Section 1, bullet 1 (activated immune cells) has been modeled in IS.
2. Section 1, bullet 2 (activated immune cell products), If submitting under the SDTMIG v3.2/v3.3, and based on the previous IS domain scope:

a. 2a-i (antidrug antibody) should be modeled in IS. Previously there had been discussions on representing this data in LB.

b. 2a-ii (anti-microbial antibody) could be mapped to multiple domain

• based on the version of the SDTMIG implemented:
  • Use LB or IS, if submitting data under the SDTMIG v3.2 or SENDIG v3.0 and v3.1.
  • Use MB or IS, if submitting data under the SDTMIG v3.3.
• based on the IS domain scope per the SDTMIG v3.2/3.3 (where the IS domain was only defined for “study therapy-induced” immune response data):
  • Use MB or LB, if antibody presence and/or quantity were collected prior to study treatment.
  • Use IS, if antibody presence and/or quantity were collected after study treatment.

              c. 2a-iii (anti-allergen antibody) in specimen-based assessments should be mapped to multiple domains,

• Use LB, if it is due to natural/previous exposure to allergens.
• Use IS, if it is measured after administration of a study immunotherapy (e.g., increased/decreased anti-allergen IgE antibody titers in response to, for example, allergy shots).

              d. 2a-iv (autoantibody) and 2a-v (alloantibody) are still mapped to LB per the SDTMIG v3.2/3.3.

• There is a similar confusion/lack of clarity with alloantibodies in organ transplantation studies. If the organ recipient has certain anti-doner alloantibodies due to previous exposure (e.g., from blood transfusion, pregnancy, or prior transplant), this data belongs in LB. If the alloantibody levels change after receiving organ transplantation, then they belong in IS.

In summary, the “baseline” immune response testing data prior to study treatment exposure would most likely end up in a different domain. The section below describes and discusses the SDTM modeling of a specific use-case, which models Respiratory Syncytial Virus (RSV) vaccine-induced humoral protective immune response via the quantification of RSV antibody production. In this example, when one measures “TEST = RSV IgG Antibody”, this piece of data can mean two things:

• It is an anti-microbial antibody, produced in response to the RSV natural infection, and can be used to detect prior or current RSV infection, therefore it should be mapped to the MB domain (note the MB domain is used to store data pertaining to detection, identification, and characterization of non-host microorganisms), as per the SDTMIG v3.2/v3.3.
• It can also be a vaccine-induced antibody. If it is produced in response to the RSV study vaccine, it should be mapped to IS, per the SDTMIG v3.2/v3.3.

The data collected in the form of controlled terminology is the same for the above two scenarios (e.g., MB/ISTEST = RSV IgG Antibody, per the SDTMIG v3.2/v3.3).

This then raises the question: for a RSV vaccine trial, when baseline measurement is performed, and there is RSV IgG antibody presence (and a titer is also frequently collected at this time) - this is prior to RSV study vaccine administration. Should users map the baseline data to MB or IS or both?

Users are required to perform the following mental process:

1. There is a presence/quantity of RSV IgG antibody at baseline, this could be due to a prior or ongoing natural infection, therefore could be mapped to MB.

• Note: MB is used to model non-host microorganism detection, identification and quantification type of data. In this specific scenario, the RSV antibody levels at baseline are collected to compare with the RSV vaccine-induced protective antibody levels after study vaccine administration – the baseline measurement is not meant for microbial identification/detection purpose, consequently, using MB is a stretch on its domain scope.
• Note: A separate, but closely related argument has also been made about whether anti-microbial antibody tests should be modeled in MB in the first place. The MB domain is used for the direct assessments of the non-host microorganism of interest itself that indicates a presence (or absence) of the microorganism at the time of specimen testing. The presence of the anti-microbial antibody is the host’s immune response toward a pathogen, and is a surrogate measure for prior or current infection. If a particular anti-microbial antibody is present, it doesn’t necessarily signal a current and ongoing infection, nor the presence of the microorganism at the time of specimen testing.

2. The presence of the RSV antibody at baseline could be the result of a prior vaccine exposure (or from both prior infection and vaccination). Because it is not induced by the “study” vaccine, it does not belong in IS. Where would one map it? Using MB would be technically incorrect.

3. Lastly, the version of the SDTMIG also affects the domain used for mapping this baseline record – it should go into LB if submitted under the SDTMIG v3.2, and MB if submitted under the SDTMIG v3.3 .

Following the above process, the SDTM modeling for “RSV IgG antibody quantification” test, based on the SDTMIG v3.2/v3.3, would look like Mock 1 or 2 examples (note variables have been omitted from the tables for simplicity):

Mock 1 and 2 approaches would lead to using separate domains to store baseline and post-exposure data. To keep pre- and post-vaccination records in the same domain, the scope of the IS domain would require expansion. See Mock 3 example:

Mock 3 is the approach that was ultimately adopted for the SDTMIG v3.4, with some modifications, by adding several variable qualifiers to relieve the --TEST/--TESTCD variables from being overloaded. See Example 1 in the Appendix section.

Finally, CDISC principles (per https://www.cdisc.org/kb/articles/domains-are-topic-based-except-when-theyre-based-structure and https://www.cdisc.org/guiding-principles ) dictate that domains are topic-based, and records with a common "topic" should be mapped into a single domain. The main argument and drive behind the IS domain scope update in the SDTMIG v3.4 is that the measurement of the magnitude and scale of the immune response upon stimulation, whether in the form of antibody or immune cell activation, is an assessment of immunogenicity, and therefore should be mapped into a single domain, regardless of the source of that stimulation (i.e., study therapy vs. non-study therapy vs. antigen exposure). This argument aligns with CDISC’s topic-based domain principle and absolves users from having to triage the same kind of data into different domains based on the source of the stimulation and the version of the SDTMIG used for data submission.

A general disagreement and confusion over the prior narrow definition of the IS domain – what’s considered as therapy?

Users have inquired whether certain specimen-based immunogenicity assessment data, which had been previously modeled in LB, such as the interferon gamma release assay (IGRA) that measures the host’s immune reactivity toward a pathogen/antigen, should be reconsidered for modeling in the IS domain. IGRA is typically performed for (in vitro blood) diagnostic purposes as it is not strictly therapeutic in the traditional sense of the word (i.e., as an intervention to treat), which may have been the reason why this type of testing had been modeled in LB and precluded from the prior versions and scope the IS domain. The SDTMIG v3.4 IS domain expanded scope allows the transition of such data from LB to IS with the support of the IS domain standard variables.

• Note: Based on user feedback, most users have already mapped this type of data to the IS domain, despite the guidance provided by the SDTMIGs and TAUGs.

Multiple SDTM Classes are utilized to model systemic vs. localized immune responses, in part to meet regulatory requirements

When modeling vaccine and allergy testing immune response data:

• Specimen-based, (typically) systemic, immune response data should now be modeled in the IS domain instead of the LB domain. Examples include, but are not limited to, assessments of anti-allergen antibodies and vaccine-induced antibodies.
• Surface, (typically) localized, immune response data are mapped to the Events and Findings About domains:
  • If mapping a “wanted, expected” dermal immune response to substances that are intended to provoke such a response (e.g., allergens/antigens used in dermal allergy testing), use the Skin Response (SR) domain.
  • If mapping an “unwanted, unexpected” or symptomatic allergic, reactogenicity event (e.g., vaccine-related reactogenicity events, such as an injection-site allergic AE), use the Adverse Events (AE) or Clinical Events (CE) domains, and the FACE, FAAE structures for findings about the reactogenicity event.
     

Section 3: A Summary on the outcomes

An effort was made to update and expand the scope of the IS domain for the SDTMIG v3.4 to resolve issues outlined in Section 2 of this discussion. This effort included the migration, harmonization, consolidation, and re-structuring of the specimen-based, immunogenicity assessment data from three domains (LB, MB and IS) into a single IS domain.

Immunogenicity studies have complicated experimental designs and biological processes, yielding complicated data. New standard variables were developed and introduced for the IS domain in the SDTMIG v3.4 to a) meet the urgent need for clear and consistent representation of specimen-based, immune response testing data, and b) resolve the long-standing issue of overloading the --TEST/--TESTCD variables. The following list outlines the new standard variables published for the IS domain in the SDMTIG v3.4 and SDTM v2.0 :

• MSCBCE (Molecule Secreted by Cells), this variable is restricted to the IS domain.
• BDAGNT (Binding Agent), TSTCND (Test Condition), and CNDAGT (Test Condition Agent) are used in IS, LB and Cell Phenotyping (CP) domains.
• TSTOPO (Test Operational Objective) is used in both the LB and IS domains.
   

Appendix

Example 1 : ISTEST, ISBDAGNT and ISTSTDTL post-coordination approach to resolve --TEST/--TESTCD variables overloading issue in the SDTMIG v3.4.

(Note: Variables are omitted from the tables for simplicity)

Example 2 : LB Pre-coordination (SDTMIG v3.2/v3.3) vs. IS post-coordination (SDTMIG v3.4) approaches on modeling allergy antibody tests.

(Note: Variables are omitted from the tables for simplicity)