The Haystack Is Made of Needles

Abstract

Developing genetic tests that have clinical utility and validated biomarkers presents many challenges. Much has been written about these challenges for the development of genetic test evidence (Khoury et al., 2010; Horn and Terry, 2012) and biomarker validation (Lesko and Atkinson, 2001; Surh, 2009). One consistent thread through these challenges is the lack of well-characterized cohorts. This is already generally true for common conditions (< 5% of patients with cancer participate in clinical trials [Michaels, 2012]), and it seems likely this problem will only worsen as cancers are stratified by genomic tests. Finding a robust cohort has always been difficult for rare diseases, simply by virtue of their rarity (Griggs et al., 2009; Gliklich and Leavy, 2011). Finding individuals to create these cohorts can feel like finding a needle in a haystack. And yet, this way of framing the problem unnecessarily boxes us in. As a solution, we are designing and pursuing research innovations that turn that framing on its head. Notice what happens to the possibilities when we consider that we have before us not a needle in a haystack but a haystack composed of needles! Instead of trying to enroll participants in a clinical trial by seeking only those few rare individuals with a particular variation or individuals affected by a rare disease (the needles in a haystack), we need to engage all individuals (because the haystack is made of needles). This simple concept has been discounted as too expensive, producing excessively noisy and dirty data, and being fundamentally impractical in an age when researchers are burdened by national and institutional regulations that make it difficult to engage anyone but a welldefined cohort formed around a hypothesis. As more people are becoming aware of this hidden potential, we are seeing a shift from hypothesis-constrained cohort identification toward the creation of hypothesis-generating registries and repositories (Ritchie et al., 2010; Manolio et al., 2012). Although this is a welcome shift, it generally originates in institutions seeking robust resources in the form of registries to support their own research, or those that are bravely encouraging open data sharing (Norman et al., 2011). As a result of this limited scope, each has limited value for creating meaningful cohorts for advancing widespread research objectives. We speculate that given the right tools, it is possible to create a novel type of registry that overcomes these limitations and identifies more robust cohorts for both common and rare diseases. This shift in capability requires looking at biomedical research from the perspective of the people who will benefit most: each of us. This work starts from the premise that a carefully maintained haystack made of needles will be of use to researchers and appealing to participants. Genetic Alliance has a long history of engaging the public with comprehensive and accessible health information and in building capacity in disease advocacy organizations to create platforms to better engage their respective populations. Two major results of these endeavors are Disease InfoSearch (www.diseaseinfosearch.org) and the Genetic Alliance Registry and BioBank (GARB; www.biobank.org). Since their creation, Genetic Alliance has experienced some of the challenges described earlier. As increasingly powerful and less costly technologies are developed for genomic testing, data analysis, and exchange, Genetic Alliance has also been mindful of the ethical issues inherent in data sharing; it led the coalition that moved the Genetic Information Nondiscrimination Act of 2008 (GINA) through the US Congress over 12 and a half years. GARB was created in 2003, the only layperson-owned and -managed cross-disease registry and biorepository platform in the world, but unfortunately it has not yet reached critical mass. The many registries created by disease advocacy organizations are highly trusted, but they lack the cross-disease features needed to support a systems approach to diseases; ease of discovering cohorts; and appropriate, individualized access to clinical trials for their participants. We observed a common characteristic in all these databases of personally identified information. In these registries— whether owned by disease advocacy organizations, universities, government agencies, or industry (and indeed, even in the case of GARB)—the individuals who elect to participate hand over control of their data and samples to a proxy. Although that proxy is often considered a trusted entity, the fact is that the individual who contributed data is no longer a participant in the registry. Data context is lost because the experience of the individual and the community in which she or he lives are not available. More often than not, the opportunity for engaging individuals in their own health, and moving translation forward toward effects on actual health outcomes, is neglected in favor of more conventional research goals.