Prior Intervention Criteria in Clinical Trial Matching
Clinical trials are designed to test the efficacy of investigational or approved drugs in new settings for eligible patients. The eligibility of patients for these trials are governed by protocol documents that list out all inclusion and exclusion criteria for each treatment arm.
Inclusion criteria are conditions potential cancer patients must meet to be eligible to enroll in the clinical trial. These are often specific disease, biomarker combinations, or additional clinical parameters.
Exclusion criteria are conditions potential cancer patients must not have in order to enroll in the clinical trial.
Examples of common exclusion criteria include laboratory results within a certain range, brain metastasis status, or pregnancy status. Cancer clinical trials additionally require that patients must have received a certain number of prior cancer treatments, or that they have exhausted all standard of care treatments for their particular condition. Additionally, the trials also contain requirements for the current disease setting of the patient or their overall number of treatment lines received. Generally, there is a lack of standard vocabulary or criteria used to define these eligibility requirements, along with differences in descriptions of drugs, drug categories, or constitution of standard of care.
All clinical trials containing biomarker-driven or relevant eligibility criteria are curated by members of GenomOncology’s Content Team on a nightly basis, resulting in a database of approximately 10,000 candidate trials for matching. This process ensures that all patients whose molecular results are analyzed using one or more of GenomOncology's software offerings will be able to be matched against a fully curated, up-to-date, and accurate set of clinical trial eligibility criteria. The results of this matching process outputs relevant trials these patients would be eligible for, either for inclusion on pathology reports or for use in clinical decision support in oncology settings.
While this process ensures that accurate demographic, biomarker, and disease-driven matching takes place, the prior treatment criteria requirements have not been part of the overall curation and matching process. In order to ensure that GenomOncology software can produce the most comprehensive results for patient matching, these eligibility criteria will now be curated in a structured and matchable manner on clinical trials within our software.
To tackle the problem of capturing the wide variety of prior treatment eligibility requirements possible, the team first analyzed the landscape of these criteria to determine common characteristics for shaping the overall data model. It was determined that each trial needed the following elements to represent the full prior treatment eligibility:
Each curated treatment arm is intended for a specific disease state (i.e. Metastatic disease)
Each curated treatment arm is intended for patients in a particular treatment setting (i.e. Subsequent line)
Each curated treatment arm has specific prior treatment requirements in specific disease state or treatment settings (i.e "patient must have received 1 line of endocrine therapy (including CDK4/6 inhibitor, but excluding fulvestrant or mTOR inhibitor) and/or 1 line of chemotherapy in the metastatic setting")
GenomOncology's existing ontologies have made the addition of these elements to clinical trials a seamless process. We already have an existing database of disease states and treatment settings, along with the ability to relate individual values to each other across the ontology. Our drug database contains ~ 6,000 entries from NCIthesaurus, and it is continuously updated and maintained through our regular trial curation efforts. In addition, logic is now in place to allow for the "bridging" of patient input drugs to all of the relevant ontology concepts in the database. For example, if a patient has received treatment with erlotinib, it can be exploded into higher-level concepts (i.e. tyrosine kinase inhibitor, antineoplastic agents, EGFR TKI etc.) and its drug type or category (targeted therapy).
The second aspect required to make this process effective is the ability to successfully interpret a patient's clinical data to enable the proper inputs for the matching process. Cancer patients whose results are processed in one or more GenomOncology solutions will often have a later stage of disease along with a detailed history of treatments received. These clinical histories, either from data warehouses or data lakes, can be inputted into GenomOncology software through a variety of mechanisms and combined with existing patients' genomic and demographic data. Specific logic can be applied for each client's unique clinical data extract, or generic broadly applicable data can be associated. The intention, regardless of the complexity, is to ensure that each patient will have all of the candidate data elements for matching to trials available as inputs into the matching algorithm. This process also ensures that patient records stored within the GenomOncology database accurately represent the full data available for patients within the system.
By combining sophisticated data ingestion mechanisms with the new prior treatment curation in the clinical trial database, the most accurate and clinically relevant treatment options can be surfaced using GenomOncology software. Learn how prior medical interventions impacted treatment matching in a recent patient use case.