We here a great deal these days about biomarkers and their importance in drug discovery, clinical trials, and even regulatory paths to approval. I would like to highlight a few articles on the topic that were recently published in the journals Cancer Research and Clinical Cancer Research (continue reading below the jump)
*Inspiration and many more examples courtesy of the Biotech Strategy Blog (written by Pieter Droppert) and the Pharma Strategy Blog (authored by Dr. Sally Church).
Dr. Kern sums up the list of pitfalls as follows:
- the biomarker is indeed a valid classifier of disease/risk, but has no clinical utility
- the biomarker appears promising but data used to generate this result were biased in some way or the study population was deficient
- biomarker is implausible or of limited utility due to normal variation
- biomarker assay deficiencies (technical or statistical)
- biomarker has already been discredited by someone else ("quietly" as he puts it)
- non-biomarker deficiencies (investigators, institutions, grantors, publication process)
Beyond the fact biomarker studies are highly unlikely to influence disease treatment in the future, they also will rarely deliver benefit to the patients involved. Therefore, it is important to carefully scrutinize the ethics of these studies, and this was the subject of a recent article and commentary in Clinical Cancer Research.
Georgina Freeman and Jonathan Kimmelman write on the subject of this overarching theme: "In principle, nondiagnostic biopsies for pharmacodynamic (PD) studies are carried out to inform decision-making in drug development. Because such procedures have no therapeutic value, their ethical justification requires that results be published."
As you may have guessed, these publications simply do not occur often enough. Somewhere around 41-53% of studies involving non-diagnostic tissue samples were either not published at all or reported incompletely. The authors surveyed authors and clinical trial investigators to figure out what caused the low publication rate:
- Strategic consideration in publication / unclear interpretation of results (59% of survey respondents)
- Poor patient accrual (44%)
- Poor quality or incomplete biomarker assay (41%)
- Inadequate tissue (35%)
- Publication pressure (29%)
- Funding constraints (9%)
- Industry pressure (6%)
The authors suggest a number of process changes to improve these results including better plans (and IRB oversight of) for recruitment, assay validation, and sample handling as well as higher publication standards/guidelines along the lines of CONSORT for clinical trials and REMARK for tumor marker prognostic studies.
Dr. Robert Iannone's commentary offers further perspective on the topic. He first notes that all the of trial studied in the above paper were completed before 2005 (in order to assess publication outcomes), and thus the authors may not have captured some improvements since in the ensuing years especially as drug development increasing relies on our molecular understanding of cancer biology. Going forward, Dr. Iannone recommended the following guidelines for all biomarker studies, which overlap with those proposed by Drs. Freeman and Kimmelman above:
- Clear objectives and hypotheses
- Adequate statistical power (accounting for assay variability, effect size, sample size)
- Adequate tissue acquisition, handling, storage
- Validated assay in place
- Information to be gained justifies patient risk
- Funding is assured