Risky Business: The Need for Hypothesis-Generating Research

Abstract

Within the field of biomedical research, unbiased or ‘‘hypothesis-free’’ research is generally viewed negatively. Often characterized as a ‘‘fishing expedition’’ in a pool stocked with the wrong or too few fish, hypothesis-generating research is frequently portrayed as an experimental approach lacking rational design and without scientific merit. In the current risk-averse funding climate, the potential benefits of hypothesis-generating research are viewed as too uncertain to warrant increasingly scarce research funding. With all of the negative attention that hypothesis-free research receives, it is easy to lose sight of the benefits this type of research can yield. Taking an unbiased approach to research can produce a number of unique benefits compared with the more common hypothesis-driven research approach. As the name suggests, a hypothesis-generating approach can be used to identify proteins, pathways, and biomarkers and to characterize disease progression. Rather than delving deeper into pathways or mechanisms that are already extensively characterized, hypothesis-generating research can improve our understanding of the complexity of the molecular changes that underlie disease pathophysiology and the characterization of disease phenotype. Despite most diseases being the result of a complex interplay among multiple genes and gene–environment interactions, research on potential biomarkers and therapeutic targets for many complex disorders tends to center mainly on a single target. Take for example, Alzheimer’s disease (AD). For the past 20 years research has primarily been focused on amyloid beta, one of the key proteins contributing to the plaque formations characteristic of AD. Although this protein undoubtedly has an important role in AD, using it as either a therapeutic target or biomarker has proven difficult. Two recent hypothesis-generating studies, however, have increased the number of known susceptibility loci 10-fold and implicated a number of new pathways in this disease, including immune function and lipid metabolism (Hollingworth et al., 2011; Naj et al., 2011). These studies on the genetic factors contributing to AD provide compelling examples of how hypothesis-generating research can exponentially increase the number of potential therapeutic targets to pursue. Hypothesis-generating research can also improve our understanding of the underlying molecular complexity of single gene disorders. As in the AD research community, much of the research effort on spinal muscular atrophy (SMA) has focused on the gene linked to the cause of over 95% of SMA cases, survival motor neuron (SMN) 1 and a second, highly homologous gene, SMN2. Despite the advent of increasingly sensitive techniques for measuring SMN mRNA and protein levels, the use of SMN transcripts and protein as molecular biomarkers has yielded disappointing and frustratingly variable results (Tiziano et al., 2010). However one recently completed study that took an unbiased approach to identified SMA biomarkers in blood and urine, BforSMA, yielded over 400 candidate biomarkers that await validation (Spinal Muscular Atrophy Foundation, 2010). This willingness to look for biomarkers using an unbiased approach has yielded a plethora of new candidates for investigation and will undoubtedly implicate a number of new pathways in the pathology and progression of this disease. In addition to generating new research hypotheses, unbiased research can reveal previously unknown commonalities among seemingly disparate diseases. With the advent of highthroughput technology and public availability of large biomolecular datasets, it is now possible to probe the molecular characteristics of diseases in an unbiased, systematic manner. Comparing the gene expression signatures in different disease states in a hypothesis-free or unbiased manner has shown some surprising similarities among diseases that have traditionally been viewed as unrelated, such as Crohn’s disease and malaria (Suthram et al., 2010). Understanding the molecular commonalities among seemingly unrelated diseases will help identify new therapeutic targets in diseases for which the underlying cause has proven elusive. Elucidating these commonalities will also make it easier to identify existing therapies that can be repurposed to treat conditions that remain intractable (Terry, 2010). The scientific and technological advances made during the past decade have paved the way for unbiased research at unprecedented levels. The Human Genome Project and subsequent work on understanding the variability present within the human genome has paved the way for conducting powerful and unbiased genome-wide association studies. It is time for our community to revisit its bias against unbiased research and ask ourselves whether the risks traditionally associated with hypothesis-generating research still outweigh the potential benefits. It is our assertion that ignoring the promise of hypothesis-generating research may be the greatest risk of all.