Missing the forest for the trees: Genome structure vs SNPs

In the July 24 issue of Nature Biotechnology, Jun Wang from the Beijing Genomics Institute reports that large scale variations in the structure (e.g. deletion, duplication) of the human genome may contain more information about an individual than the collection of Single Nucleotide Polymorphisms (SNPs) usually reported in genomic biomarker studies such as Genome Wide Association Studies (GWAS). If confirmed, this report would force the field of genomics to rethink the way genomic data is assembled and reported. Indeed, even the most advanced DNA sequencing methods available today rely on chopping the DNA into small fragments that are easy to read but extremely difficult to reassemble into a full, structurally correct full genome.

Read also the piece in Wired magazine about this article: Your Genome Structure, Not Genetic Mutations, Makes You Different | Wired Science | Wired.com

Thierry Sornasse for Integrated Biomarker Strategy

New approaches to improve biomarker discovery

The American Chemical Society released in its Chemical & Engineering News magazine an overview of the successes and failures of biomarker discovery. The conclusions of this article emphasize the urgent need to pay more attention to the critical step of novel biomarker qualification (i.e. the process of confirming the predictive value of a biomarker candidate).

New approaches to improve biomarker discovery

Thierry Sornasse for Integrated Biomarker Strategy

Medical Devices and the Public’s Health: The FDA 510(k) Clearance Process at 35 Years - Institute of Medicine

Today, the FDA CDRH released a report produced by the independent Institute of Medicine about the FDA 510(k) clearance process for devices. Although this document is not specifically about diagnostic assay devices (it covers all devices from MRI systems to pregnancy test), its main conclusion that the 510(k) clearance process is flawed will most likely have an impact on the approval of future diagnostic method derived from biomarker research.

Medical Devices and the Public’s Health: The FDA 510(k) Clearance Process at 35 Years - Institute of Medicine

Thierry Sornasse for Integrated Biomarker Strategy

Translational Biomarker Development: mind the gap

Of all the disciplines and specialties required to develop new drugs, biomarker development is, in my mind, the specialty that has the most to gain from a modern Translational Medicine organization structure (i.e. a seamless integration of all drug development stakeholders from a project inception to its conclusion).  However, throughout my career, I have experienced situations telling me that biomarker development still appears to be fragmented in many bio / pharma companies. Unlike other specialties, such as pharmacokinetics and toxicology which tend to be formally organized to facilitate connection between Research and Clinical, biomarker development still too often tends to exist in discrete, partially isolated functional groups.


If you have worked in the biomarker field, you might be familiar with the following examples of gaps in connection between biomarker stakeholders:

• A clinical team leader asks for a disease modification biomarker less than six months prior to the start of a clinical study 
• A biology research scientist scrambles to develop and validate a biomarker assay method after his/her favorite biomarker is included in a clinical study protocol 
• Regulatory and Legal produce a Patient Informed Consent form that precludes any post-hoc analysis of clinical samples

Let me suggest a two-prong approach to solve this type of issue:

• Organization structure: many companies would gain in setting up a small translational biomarker group serving as the main interface between all biomarker stakeholders. This "hinge" function should be responsible for identifying, communicating, and resolving requirements and issues associated with biomarkers throughout a program's life cycle. 
• Planning process: based on the FDA guidance "Target Product Profile — A Strategic Development Process Tool", I would argue that biomarker development should follow the same path as a drug program. Thus, I would recommend the creation of a "Target Biomarker Profile" for all drug development programs even if biomarker needs appear to be minimal (determining that nothing needs to be done is not the same as ignoring the issue). This TBP would articulate the intended overall goals for biomarkers associated with a given program, defining the intended decision to be driven by the biomarker, the expected impact level of the biomarker on the program, the timing of deliverable, and last but not least, the intended audience of biomarker data. Similarly to the TPP, the TBP should be an evolving document in which initial assumptions should be revisited and new priorities taken into account. 

Thierry Sornasse for Integrated Biomarker Strategy

Biomarker Development: validate and qualify (or is it the other way around)

Biomarker development is fundamentally the parallel processes of developing a technically reliable biomarker measurement method and establishing the significance of a biomarker relationship with a specific biological process. The former process is generally referred to as "validation" or more precisely "method validation" while the FDA suggested that the latter be referred to as "qualification".

The problem is that the term "qualification" means something quite different to my PK bioanalytical colleagues. Indeed, it refers to the process of testing the performance of an assay under stringent conditions to record the its variability without setting "pass/fail" criteria as it would be done for a validation process. 


To make matter worse, the FDA itself seems to struggle with the term"qualification" in its proposed three stages of biomarker qualification:

• Exploratory: fit for internal decision only
• Assay development à assay qualification
• Limited or no clinical data
• Probable valid: fit for regulatory decision applied to early clinical development
• Well established assay performance
• Initial significance of clinical test result
• Known valid: fit for regulatory decision applied to late clinical development
• Measurement independently reproduced
• Agreement by scientific community.

In other words, when a biomarker is qualified it becomes valid. Confusing isn't it?

In my mind, the only practical way to resolve this semantic issue is to simply be more precise by specifying whether we are referring to a process related to a method:

• Method qualification
• Method validation

Or, referring to a process related to the scientific relevance of a biomarker:

• Scientific qualification

Thierry Sornasse for Integrated Biomarker Strategy

Cancer biomarkers: not such a great model to follow

The development and use of biomarkers in the field of cancer is well ahead of any other indications. In particular, several cancer biomarkers have been co-developed as companion diagnostics associated with specific drugs:

• Her2 expression test (e.g. HercepTest) supporting Herceptin (trastuzumab) prescription for breast cancer
• BCR-Abl / Philedelphia chromosome supporting Gleevec (imatinib) prescription for CML
• BRAF V600E expression supporting PLX4032 (vemurafenib) administration for melanoma (not approved yet)

While these companion diagnostics represent a major progress in the area of personalized medicine, all three cases provide a relatively poor model for the development of advanced biomarkers / companion diagnostics in other diseases. Indeed, for these three biomarkers, there is a direct connection between the biomarker: presence of target and the biological process of interest: sensitivity to the drug. Unfortunately, in many other diseases, this direct relationship does not exist or at least has remained elusive until now. Thus, the experience from the cancer biomarker field teach us very little about the complex process of biomarker biological qualification (i.e. the process of establishing the significance of a biomarker relationship to the biological process of interest). In my mind, the process of biomarker qualification is, and will be for the foreseeable future, the most  important aspect of biomarker development for which the lessons learned from the cancer biomarker field are not a great model to follow.


Thierry Sornasse for Integrated Biomarker Strategy

Beyond the FDA / C-Path Biomarker definition

If you have read any review about biomarkers, you have most likely encountered the following definition put forth by the FDA via the Critical Path Institute:
 "A biomarker is a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention."


While this definition is scientifically accurate, it lacks a key element: an intent. In my mind, this definition should be revised to include the following phrase: "...designed to drive decisions in drug development and/or patient treatment"
Thus, biomarkers are first and foremost decision tools that need to produce reliable predictions about the the biological process of interest in order to make informed decision about a drug or a patient treatment.


Thierry Sornasse for Integrated Biomarker Strategy