The practice of genomic medicine: A delineation of the process and its governing principles

Abstract

Genomic medicine, an emerging medical discipline, applies the principles of evolution, developmental biology, functional genomics, and structural genomics within clinical care. Enabling widespread adoption and integration of genomic medicine into clinical practice is key to achieving precision medicine. We delineate a biological framework defining diagnostic utility of genomic testing and map the process of genomic medicine to inform integration into clinical practice. This process leverages collaboration and collective cognition of patients, principal care providers, clinical genomic specialists, laboratory geneticists, and payers. We detail considerations for referral, triage, patient intake, phenotyping, testing eligibility, variant analysis and interpretation, counseling, and management within the utilitarian limitations of health care systems. To reduce barriers for clinician engagement in genomic medicine, we provide several decision-making frameworks and tools and describe the implementation of the proposed workflow in a prototyped electronic platform that facilitates genomic care. Finally, we discuss a vision for the future of genomic medicine and comment on areas for continued efforts.

Keywords: distributed cognition; genomic medicine; integrated care; precision medicine; workflow optimization.

FIGURE 1

Applying biological principles to assess the likelihood of genetic disease. Within the biologic principles of disease (left column), certain clinical attributes (center column) increase the likelihood of a genetic disease. The (right column) lists examples of clinical representations of such biological principles for individuals with intellectual disability. Each example may be representative of more than one biological principal. Drawn from current understanding of evolutionary biology, these principles were synthesized and defined by the Provincial Medical Genetics Program (PMGP) clinical genomic specialists (CGSs), and the examples were drawn from their clinical practices.

FIGURE 2

The genomic medicine process map. This workflow illustrates genomic medicine practice, which differs from medical genetics in that molecular profiling is required for diagnosis. The process begins when a patient or the PCP identifies a health concern. Rectangles denote tasks and orange diamonds denote decision points. Swim lanes correspond to the player responsible for completion of the tasks. References to the frameworks and guidelines presented throughout the manuscript are in purple and are connect to the corresponding decision point with a dotted line. The tasks and decision points are numbered to reference text descriptions (Supplementary material section 1); this numbering does not always reflect a linear sequence of steps due to decisional loops. Endpoints are denoted by pink circles.

FIGURE 3

Principles and frameworks governing genomic medicine. (A) Bayesian framework: A thought process for variant interpretation and diagnosis. Bayes theorem quantifies the probability of “A” being true given some evidence “B.” Applying this framework to genomic diagnosis, we can assess the posterior probability of a patient’s disease (Disease A) being caused by a genetic variant (Variant A) by dividing the product of the prior probability of Disease A and the likelihood of the variant in the context of the disease. As this formula serves to guide a thought process, numerical input and quantification of the probabilities are not required. (B) Assessing eligibility for genomic testing using biologic and utilitarian frameworks. Given that assessment of biological principles suggests that a patient has an increased likelihood of genetic disease (Figure 1), application of a utilitarian framework allocates resources to enable testing that accomplishes societal expectations of benefit, impact, and utility for the patient, the family members, and the payer.

FIGURE 4

Schematic representation of variant analysis and resolution. (A) A hypothetical patient phenotype (gray circle) is plotted on an imaginary grid in relation to disease phenotypes associated with genomic loci within which the patient carries variants (ochre, purple, green dots). (B) Given that the patient has a rare disorder, variants with high unaffected population frequencies (green dots) are eliminated from consideration. (C) The remaining variants are filtered considering precedent (prior association with phenotype), variant type (e.g., non-synonymous, frameshift, premature stop), and in silico prediction of deleteriousness. The remaining two variants, which meet a threshold of potential deleteriousness and occur in loci associated with diseases phenotypically overlapping that of the patient, are prioritized and reported by the laboratory. (D) The CGS constructs a posterior probability for a diagnosis through review of the reported variants for evidence that variation of the genomic loci cause disease, for concordance of the patient phenotype with that of the diseases, and for genetic, molecular, and biochemical congruence of the variant with the disease mechanism. For this hypothetical patient, the clinical genomic specialists (CGS) judges one variant (ochre dot) a sufficient sole molecular diagnosis.

FIGURE 5

Clinical variant interpretation logic. The left column includes the key lines of evidence to consider in variant interpretation; it starts with assessment of gene-disease causality and progresses to the integration of prior risk. A detailed description of the principles establishing gene-disease association, phenotype-disease concordance, disease-mechanism concordance, and variant deleteriousness, and a Clinical Variant Analysis Tool were reported previously to facilitate the capture and synthesis of these multiple lines of logic in assessing causality of variants (27). A depiction of how phenotype specificity informs the posterior probability that Variant A contributes to the patient presentation is presented in Supplementary material section 6. The prior risk refers to the likelihood of the Disease A in the patient, as determined before the initiation of genomic testing (Figure 3A). The middle column lists possible assignments for each row; the left-most option is most supportive of variant contribution to disease, whereas the right-most option is least supportive of variant contribution to disease. The third column highlights methods that can be pursued to collect further evidence to inform the corresponding assessment. The last row denotes the clinical conclusion based on the integration of the other rows (56).