Scott Marshall Ph.D. Managing Director Precision for Medicine
Jared Kohler Ph.D. Senior Vice President Precision for Medicine
Why Trial Success Depends on It
Precision medicine-guided drug development, along with the growing reliance on specialty lab data to address key clinical trial objectives, is fast becoming the rule rather than the exception. Biomarkers are used to understand mechanisms of action (MOA), to evaluate pharmacological effects (pharmacokinetics/pharmacodynamics), to explain differences in treatment response, and in many cases to select or stratify patients. Indeed, a recent study of clinical development success rates from 2006 to 2015 showed the benefit of using biomarkers for patient selection alone: a threefold increase in the likelihood of approval from Phase I and a 20% increase in the transition to approval from Phase III.1
It’s no wonder that pharmaceutical companies are turning to biomarkers in increasing numbers. The use of biomarker data to inform dose selection, stratify patients, and even to adapt clinical trials is an industry-wide goal. Yet, unless that data can be quickly and accurately harmonized, it does nothing to advance pharmaceutical development or approval.
Clinical Trials Have Evolved—Especially in Oncology
Such harmonization can be a challenge—due in part to the sheer volume of data and, substantially, to the diversity of data that are being collected. Historically, hematology, clinical chemistry, and pharmacokinetics have been core to clinical trials. Then, with the advent of antibody-based therapies and immunotherapies, the role of antidrug antibodies, immune monitoring, and other cell-based assays accelerated.
Modern medicine has further incorporated biomarker assessments into disease diagnosis and treatment decisions. While this wealth of information is unquestionably leading to significant healthcare advancements, it also poses challenges—particularly in oncology research. This is because of four factors:
The Assays Themselves Are Diverse
Technological advancements in the areas of next-generation sequencing (NGS), liquid biopsies, and cell-based assays—such as multicolor flow cytometry—have resulted in a boom of commercial assays, ranging from NGS and microarrays to immunoassays and cytokines. Additionally, specialty assays are often used to assess tumor heterogeneity, circulating tumor cells, immune repertoire, and targeted immune cell populations.
The Assay Workflows Are Diverse
To ensure that the wet lab results are useable for downstream analysis, each assay type has unique workflow and quality control parameters.
A Range of Lab Types Are Involved
Traditional assays, such as hematology and clinical chemistry, can be run at central labs. However, the diversity in biomarker assays requires specialized methods, specialized personnel, and specialized expertise. Consequently, specialized labs have emerged to manage these assays; often developing proprietary technology, methods, or panels. Managing the data flow from each of a host of specialty vendors can be trying.
Cancer Biology Is Especially Complex
Disease pathogenesis and heterogeneity, drug response and resistance, tumor genetics, and the impact of the immune system each has specific biological aspects that must be explored and understood. Therefore, cancer researchers execute multiple experiments to effectively address the complexity—creating a different type of complexity in the process.
While successfully harmonizing this disparate array of data can pose a data management challenge, failure carries great risk.
Data Harmonization Is Critical
For two reasons: speeding decisions and meeting regulatory requirements. In clinical drug development for oncology, biomarkers are regularly used to support go/no-go decisions at each stage of a trial. Without tight integration between clinical operations and biomarker data, that decision-making can be postponed, negatively impacting trial timelines, driving up costs, and delaying approvals—to the detriment of both the financial health of the sponsor and the physical health of the target patient population, which may well benefit from a new therapy.
New FDA regulations further amplify the danger. FDA submissions now require datasets to be submitted in formats supported by the FDA and listed in the FDA Data Standards Catalog (e.g., CDISC SDTM, ADaM). In some cases, biomarker data fit into standard CDISC domains, but they often necessitate the creation of custom CDISC domains—or the ability to understand how nuances in a biomarker dataset impact the ability of these data to fit into existing CDISC biomarker domains. Again, disaggregated data may lead to untenable delays in submission and approval.
Traditional Harmonization Methods Are Inadequate for Today’s Data Demands.
For clinical trials without biomarker data, there are traditional workflows through which a sponsor or contract research organization handles the trial’s data management and biostatistics. However, the sponsor may not be equipped to accommodate the special needs of biomarker data, instead funneling it through a separate workflow in which responsibility falls to translational research or translational medicine teams working in conjunction with the clinical trial teams.
Further, data that are generated remotely are typically managed outside the clinical trial database. Each specialty lab vendor transfers its biomarker data to the sponsor. Each individual lab often dictates its own format—or else requires special arrangements and added fees to comply with a sponsor-prescribed format. Thus, each biomarker data type requires a separate workflow. These workflows may take weeks—or even months—to integrate. Worse, they may never be integrated. To put it mildly, this situation is suboptimal for analysis and reporting, ultimately limiting insights that could be gained through harmonization.
Clearly, there is a pressing need for swift, accurate harmonization of disparate datasets.
The Key Is Deep, Diverse Expertise and Specialized Technology
Some might feel that supercomputers or technologies alone offer an obvious solution to the harmonization challenge. They don’t; simply because the core issue is not computing power—it is expertise. Multiple, diverse biological assays—from a host of specialty labs in a range of formats—must be coordinated, harmonized, and reported on. All this must happen in the context of a clinical trial with all its moving parts and rigorous timelines—and must result in an output that enables time-sensitive review, analysis, and decision-making.
Certainly, technology is useful. But it needs to be bolstered by a team of specialists with direct experience around the breadth and depth of modern biomarker assays; by data management experts who can adapt traditional standards and processes for use on specialty lab data; and by innovative data scientists who can design, validate, and operate technologies specifically engineered to address the challenges of biomarker data management.
Building on this foundation of specialists, those standards, agile processes, and secure IT infrastructures must then be developed. Critically, there must be an effective technology platform that provides centralized access to biomarker data, secure data transfer, data integration and collaboration, and interactive reporting. This ability to integrate and visualize biomarker data in real time through user-friendly web-based tools will allow sponsors to harvest insights from all data sources and better inform clinical trial evaluations.
Effective Data Management Is Critical to Meeting Key Trial Objectives
Ineffective management can lead to failure. Generating, tracking, and harmonizing biomarker data and clinical data in real time to inform timely decision-making is critical in modern clinical trials. Paramount to addressing key objectives such as dose finding, patient stratification, and biomarker-guided adaptive trials, this solution enables more targeted trial results, shorter trial duration, and ultimately, increased efficiency and decreased costs in drug development.
These outcomes are too important to put at risk.
Scott Marshall, Ph.D. ([email protected]), is managing director, biomarker and IVD analytics; and Jared Kohler, Ph.D., is senior vice president, translational informatics & biometrics, at Precision for Medicine.
1. David W Thomas, Justin Burns, John Audette, Adam Carroll, Corey Dow-Hygelund, Michael Hay. Clinical Development Success Rates 2006-2015. (2016) A BIO Industry Analysis white paper.