Peer review, the bastion of quality scientific research, seems to be impeding progress in the biomedical sciences. I recently attended the 2013 Molecular Medicine Tri-Con which focuses on the how recent developments in genetic analysis are applied to advance healthcare diagnosis and treatment. There was so much going on it was hard to keep track of all the ways the explosion of genetic data is changing healthcare. In fact, though, that is precisely the problem.
The Problem with Peer-Reviewed Publications
There is a real bottleneck translating the expanding amount of genomic-related discoveries into useable clinical information. As new research and computational tools provide more precise and complex measurements about biology that enable broader system-wide studies, the data sets have become massive and analysis of them involves consortiums of labs around the world looking at bits and pieces. Biotechnology has moved into the realm of big science.
While these technical advances have greatly contributed to our ability to understand how life works, a laudable scientific goal, biomedical research also has the very a practical goal of improving healthcare. Here is where the bottleneck occurs. It has become too difficult for clinical practitioners to access and make practical use of this avalanche of complex results to improve patient care. The peer-reviewed publication process, the traditional route to evaluate and communicate scientific results is just not adequate as a vehicle for evaluating and integrating new results into disease models and treatment scenarios. In the era of computer modeling, social networking, and database screening, published scientific literature is simply too “old school” to keep up, even though almost all publications are easily searched and accessible on-line through PubMed.
How Does Peer review Work?
In reporting results, typically a group of scientists will put together a written article describing their studies with detailed experimental details and figures showing key results. These are submitted for publication to relevant scientific journals where the editors then coordinate a review of the manuscript by other researchers, typically 3 individuals that work in the same general area but are not associated with the study. These scientific peers provide comments, feedback on additional work that might be needed to support the conclusions, and recommendations whether to publish the article or not. If the reviewers and editor believe the paper has potential, the author is asked to make revisions per the reviewers’ comments, and the journal publishes it. Not surprisingly, it is typically a lengthy process that takes several months to a year, not including the work put into drafting the initial manuscript before submission. Most published work is at least few years old.
Life’s Dynamics Cannot Be Described in a Few Pages
While the time lag is a significant problem in a world where information is moving faster every day, it is not really the core issue. The real problem is the amount and complexity of the data. As mentioned above, technology developments in the biomedical research have enabled researchers to collect massive amounts of data on the characteristics and changes in thousands of bits of discrete information of living systems such as DNA variations between individuals, changes in gene expression, and the concentration or activity of different cellular proteins. This information cannot be effectively communicated in standard scientific articles and, in fact, it cannot even be effectively analyzed by a single laboratory. Simply sorting out the noise from the signal in these types of studies can be a challenge.
The Volume of Data Continues to Increase
Meanwhile, the data continues to grow. In 2012, the number of new MedLine citations, which archive most life science-related publications worldwide, surpassed 1 million. Ten years ago it was about half that. Genomics-related research is only a portion of these, but it is a significant portion and growing. For example, publications that mention SNP, an abbreviation for a single nucleotide polymorphism which describes one type of gene mutation, grew from 745 in 2002 to over 4,500 last year.
Is There a Better Option to Scientific Publication?
Unfortunately, there is no viable alternative to peer reviewed publications. The peer review process, which has become the standard to evaluate and ensure the integrity of scientific research, became a standard over the past couple hundred years. It is not something that can be easily or lightly replaced.
Also, the point is not just to replace traditional scientific publishing with something that is simply faster or more modern. The issue is to provide a mechanism that not only documents quality science, but helps integrate the growing volume of new data and results with related findings from other groups and current scientific models. In other words, in place of simple peer-reviewed publications, there needs to be some process for labs to provide novel results, get appropriate credit for the work, have it subjected to review to ensure quality, and integrate it with existing data from other researchers to facilitate the development of comprehensive predictive scientific models. This is an ambitious set of requirements.
Software to Mine and Organize Published Data
There are commercial software products available to help manage the massive amount of knowledge required to analyze new findings data from companies such as, Thompson Reuters, Ingenuity Systems, and Elsevier. However, these sorts of applications are actually a symptom of the core problem as much as a partial solution. Essentially, these packages provide tools to help investigators analyze their own data in light of previously reported results because published bioscience has become too vast and complex to analyze withou advanced computational tools.
Crowdsourcing and Social Media Solutions
A more effective alternative would seem to involve some sort of open-source crowdsourcing and, in fact, attempts are being made in this vein. For example, the Biotechnology and Biological Sciences Research Council (BBSRC) in the UK used this approach to work out the genetics of a deadly new strain of bacteria in 2011. The success of this project encouraged them to use a similar crowdsourcing approach to tackle a new plant pathogen.
There is also sbv IMPROVER, a collaboration between Philip Morris R&D and IBM which has initiated two focused crowdsourcing projects in the scientific community to expedite the verification and interpretation of complex data sets. They completed one project last year using research data to identify profiles of activated genes that are diagnostic for specific diseases, such as psoriasis and lung cancer. Their current project evaluates how results from studies of animal disease models can be more accurately extrapolated to humans.
Still No Solution that Bridges Basic Research with the Clinic
However, these newer approaches to disseminate and distribute data and analysis are really just pilot studies and, so far, focus primarily on basic research. Where does this leave clinical healthcare professionals trying to understand the disease risk of particular genetic mutations their patients might have or the significance of certain protein markers in their blood? It is clear, the days where physicians could rely on reading a few recent publications to keep up to date on the latest research have ended. How can the increasingly complex and rapidly moving research in bioscience be captured and presented so that it provides useful and practical information to healthcare providers and patients so that they can make informed decisions about treatment options? This is still very much an open question.
There is at least one non-profit organization, Cancer Commons, that is trying to tackle exactly this problem of making sure doctors and cancer patients have access to the latest genetic research relevant to their diagnosis and treatment. I will write a bit more about their innovative approach in the near future.
(Published Feb, 28, 2013)