Interestingly, the modeling and simulation technology to achieve these goals has been available for decades (12) and originated from within multiple disciplines almost simultaneously (13); its practitioners used to be few and far between, but this corpus of techniques is now achieving something of a renaissance (14). Some companies have made this an integral part of their approach to drug development and candidate selection (15). These seemingly disparate technologies are necessary if we are to solve the problems that are listed here related to disease diagnosis and therapy, as applied to drug development.
• Most diseases are multifactorial, caused by several simultaneous mutations that are not amenable to direct quantification or prediction: one example for all is the prevalence of diabetes type 2 and its links to obesity and the metabolic syndrome. However, it may be difficult to translate even "simpler" causative links to promising therapeutic strategies. Moreover, most diseases may be caused by a combination of genetic and environmental factors, with the former providing predisposition and the latter providing the triggering of the adverse condition. It may not be appropriate to rely on either one or the other for predictive purposes, and especially for individualized medicine; rather, both need to be taken into account.
• There is a perceived excessive reliance on animal models and on otherwise simplified disease components, as opposed to an increased understanding of the disease mechanisms in humans. While cancer research has been recently singled out (16), the case could be made for other diseases as well. Preclinical models are often oversimplified, difficult to scale to humans (17), ultimately irrelevant, or all of these. This issue deeply affects both academic and governmental research and private R&D labs.
• A related issue with modern biomedical research is attention to the functional aspects of collected biological information. There appears to be a disconnection between the databasing and cataloging effort ongoing in bioinformatics and related fields and the translation of these findings into action items relevant for human health. While it is true that the knowledge base about human pathophysiology has increased severalfold, it is less clear how much of this knowledge is actually relevant and can be turned into lifesaving treatments or blockbuster drugs or both. The accumulation of knowledge may not be regarded as sufficient for much longer, as both the public and investors clamor to see tangible results of time spent or venture capital supplied.
• The lack of a common language between biologists and quantitative scientists, such as engineers, mathematicians, and statisticians, may be the single most important obstacle to the progress of biomedical research. If available, this language could be used for predictive statements and integration of data and experiment, but its absence is a major flaw. It has been said that, in the biological sciences, "you're not licensed to theorize unless you put the time in and get the data" (18), which seems a hardly efficient approach to science, since individual researchers' strengths and weaknesses may prevent such approaches. In a recent commentary, the author argues this very point, and goes on to state that it is "common experience that once the number of components in a system reaches a certain threshold, understanding the system without formal analytical tools requires geniuses, who are so rare even outside biology. In engineering, the scarcity of geniuses is compensated, at least in part, by a formal language that successfully unites the efforts of many individuals, thus achieving a desired effect, be that design of a new aircraft or of a computer program. In biology, we use several arguments to convince ourselves that problems that require calculus can be solved with arithmetic if one tries hard enough and does another series of experiments" (19).
With regard to the last statement, a possibility is that the most powerful and important role for bioinformatics and computational biology is not at all where everyone thinks it is. While databasing and the collection of genetic information and sequence results may be important per se in the end, the real promise for drug development will lie in their integration with predictive, quantitative models of drug-disease systems. Such models have concentration and effect measurements as inputs, often linked with biomarkers, demographic variables, genetic information and the like, and as outputs they return dosage information and probabilistic statements of outcomes. These models are now arising in a variety of areas, and their results in the determination of what we have called model-based biomarkers earlier are worth sharing here. Instead of focusing on drug development per se, we review recent findings in a few areas, discussing also validation aspects, or tests that can be conducted on a biomarker to evaluate whether it is measuring something of value. The commonality here is the great potential for translational research, that is, the generation of testable hypotheses or therapies from basic science research.
Was this article helpful?