A Road Map for Evaluation and Appropriate Implementation of Genome Sequencing to Improve Population Health
Posted on by
This blog is a summary of our recently published paper in PLOS Medicine, and is an update of my 2011 blog on “binning” the human genome.
A common vision for genomic medicine is that genome sequencing will be routinely used in health systems to provide health care and preventive services tailored to each individual. For the most part, sequencing is not yet routinely used in general practice, but only among people with certain diseases (e.g. ill newborns, cancer, rare diseases), or genetic predisposition to certain diseases (e.g., BRCA1 and BRCA2 testing for hereditary breast and ovarian cancer susceptibility). Currently there is limited direct evidence of clinical utility of sequencing to guide clinical care and disease prevention at the population level. Yet, several health systems in the United States and other countries are beginning to integrate sequencing into patient care and disease prevention independent of disease, leading to unclear benefits, harms and healthcare costs. In addition, direct-to-consumer genetic testing has been on the rise even with no or limited evidence of clinical validity and utility.
Normally, evidentiary frameworks for genetic testing require establishing the clinical validity and utility of testing for a specific intended use. This approach may present an insurmountable challenge in evaluating genome sequencing, as the human genome sequence can be used to answer numerous questions relevant to health care over time. Multiple “tests” can be used to direct health care related activities (e.g., diagnosis, risk assessment, treatment, prevention), for multiple diseases (e.g. heart disease, cancer) and deployed throughout life.
The early adoption of genome sequencing by some health systems provides a unique opportunity to develop an evidence based process to accelerate evaluation and appropriate implementation. A fundamental question is whether or not we can use a genome sequencing platform embedded into learning health systems that can accelerate simultaneous evaluation of multiple testing scenarios with differing levels of evidence.
With this background, a group of authors from CDC, NIH institutes, health systems and academia just published in PLOS Medicine, a proposal for an evidence-based road map to accelerate the evaluation of clinical utility of genome sequencing and its appropriate implementation in health systems. The road map has 5 components with a robust translational research agenda (see Table 1 for definitions of tiers 1-3) and is described more fully in the paper.
- Ongoing knowledge integration of the human genome
- Accelerated Implementation science agenda for tier 1 applications
- Accelerated clinical utility research for tier 2 genes and variants
- Accelerated evaluation of validity and utility of tier 3 applications
- Enhanced development of workforce, tools and resources
The authors suggested that a translational research collaboration can be built onto well characterized populations with already available sequence data (in a biobank/research environment), risk factor information, intervention information, and clinical outcomes. The framework calls for collaboration among several organizations, to recruit adequate numbers of individuals. This is crucial in order to understand the consequences of knowledge of multiple risk variants for multiple clinical scenarios. In addition to observational studies, randomized clinical trials can be designed to assess individual, family, system and population outcomes based on returning versus not returning the results of selected genes/variants for specific clinical scenarios, depending on existing level of evidence. This implies that the return of the results of genome sequencing to patients and providers will occur in a research controlled fashion specified based on pre-agreed study protocols. The collaborative agenda will allow learning health systems to evolve the necessary capacity for appropriate integration of genomics alongside other medical services such as screening and treatment, as well as to prepare the healthcare workforce and the public.
A major challenge in implementing this research collaboration is to have clear lines of demarcation will have to be drawn between the “research” and the “clinical practice” arms. For example, there has to be agreement on what genes/variants to return to participants and their providers and medical records, in the course of routine clinical practice. As applications from the human genome sequence move to tier 1, they could be integrated into a practice environment where evidence generation continues to occur. For tier 2 and 3 applications, the return of results will have to be done in the appropriate research environment (e.g. randomized clinical trial of the use of pharmacogenomics variant for certain medications). As evidence accumulates over time, an important question is how do health systems handle changing information and continuously re-evaluate new information? Once integrated into care, it is notoriously difficult to de-implement practices that are subsequently found to lack effectiveness. In addition, ethical, legal, and social implications and other challenges lie ahead, necessitating community engagement to ensure that underrepresented groups are not left behind, and the collection of data to inform the development of appropriate regulations and standards for test utilization, patient privacy and data security.
In conclusion, a translational collaborative effort is urgently needed to understand the health benefits and potential harms and costs of genome sequencing for all, by studying the implementation of what we know can work (tier 1), evaluating the utility of promising applications (tier 2), and critically assessing the validity of emerging genomic information for improving health and preventing disease (tier 3).
We welcome our readers’ comments and suggestions on how to how to accelerate the evaluation and appropriate implementation of human genome sequencing to improve population health.
Table 1 *
| >Level of Evidence (Genes/Variants) | Examples | Research Framework | Research Questions |
|---|---|---|---|
| Tier 1 * | HBOC**, Lynch Syndrome, FH** | Accelerated implementation science | Assessing patient, provider, health systems success factors of optimal implementation and outcomes of existing recommendations, and reducing health disparities |
| Tier 2* | Selected pharmacogenetic traits, monogenic risk variants | Accelerated collaborative evaluation of clinical utility (RCTs) ** | Assessing benefits, harms and costs from return of genomic information compared to standard of care; selected hybrid effectiveness/implementation studies to assess how genes and variants may be integrated into practice. |
| Tier 3* | Genetic risk scores, gene-environment interaction | Accelerated collaborative evaluation of clinical validity and utility | Assessing added value of using genomic information compared to existing approaches (prediction, discrimination, interventions, outcomes) *** |
Tier 1: Applications with synthesized evidence supporting use, Tier 2: Applications with insufficient evidence to support routine use, Tier 3: Applications with either no synthesized evidence available, or evidence supporting recommendations against use. Based on the paper by Dotson et al.
** Abbreviations: HBOC: Hereditary breast and ovarian cancer, FH: Familial hypercholesterolemia, RCTs: Randomized clinical trials
*** Tier 2 and 3 genes/variants also provide an opportunity to evaluate how to prepare a health system to anticipate new findings, both positive and negative related to genome sequencing—what ancillary studies can focus on current use of genome sequencing, factors affecting uptake/de-implementation as warranted, so that future discoveries can be integrated into the learning health system