Consuming personal genomics

See allHide authors and affiliations

Science  19 Apr 2019:
Vol. 364, Issue 6437, pp. 213
DOI: 10.1126/science.aax6789

Personal genomics is enjoying increased visibility, yet much remains cloudy about its true potential. We see advertisements from direct-to-consumer companies and media coverage of scientific studies reporting new genomic associations with a range of traits. With this deluge of information, individuals need to become educated consumers to judge how valid these associations are, particularly if individuals are inclined to change their actions based on the results.


“Genomic information has great potential to enrich our lives…”

Some traits, such as adult height, are readily measured. The heritability of this trait is ∼60 to 80%. Attempts to characterize “height genes” have resulted in the identification of tens of thousands of genes, each of which contributes a small amount to this heritability. The plethora of factors is almost inevitable, given the vast number of cellular and physiological steps involved in the development of an adult human being. A model that accounts for ∼40% of height variability predicts individual heights to within 4 cm for 50% of people, but with errors of more than 10 cm for 5%. Thus, a sophisticated genomic analysis can predict height to some extent, but not well enough for use in ordering tailored clothing. Most direct-to-consumer genomic results are based on much less detailed analyses and many involve complex traits, so considerable skepticism is appropriate.

Caution, however, should not dampen optimism for all aspects of this field. An area with exciting recent progress involves genomics as it relates to differences between people in the efficacy or dosing of particular drugs, a field known as pharmacogenomics. The U.S. Food and Drug Administration now includes pharmacogenomic information in the product labeling for more than 200 drugs, but use of this information in clinical practice remains limited. One example is clopidogrel (Plavix), a drug that prevents blood clotting to decrease the risk for heart attack and stroke. The compound present in clopidogrel pills does not, itself, have activity but is converted to an active form by an enzyme called cytochrome P450 2C19 (CYP2C19). Approximately 30% of individuals have one or two copies of a version of the gene encoding CYP2C19 that produces little to no active enzyme, meaning that clopidogrel may not be effectively activated. A pragmatic, multicenter clinical trial was conducted involving patients receiving cardiac stents who were offered a genotype-based prescription of either clopidogrel or related drugs that do not require CYP2C19 for activation. Those with a genotype suggesting inefficient clopidogrel activation who were treated with alternative drugs were significantly less likely to suffer major adverse cardiovascular events than those treated with clopidogrel. With alternative therapy in these patients, outcomes were similar to those of patients without the at-risk genotype who were treated with clopidogrel. These results illustrate the importance of thoughtful studies that help move compelling visions for pharmacogenomics into reality. A major challenge at present is that genotyping is required as an extra step, introducing costs and logistical issues. However, it is easy to envision a future where one-time genomic tests covering the few hundred genes that are associated with commonly used drugs are available to physicians, pharmacists, or consumers, to inform drug-use decisions in real time.

A consumer of genomic information should question the data supporting the interpretation of genome–trait association tests. Some traits given in direct-to-consumer reports such as “lactose intolerance” involve genes with known biological mechanisms and well-characterized variants that largely determine outcomes. Other traits such as “wake-up time” involve models with hundreds of genes with no known mechanism, clear importance of nongenomic factors, and poor predictive value. In addition, many direct-to-consumer tests report only common variants of genes such as BRCA1, potentially giving individuals a false sense of security. Ancestry results can be relatively robust because they involve the segregation of many genetic markers and do not require biological mechanisms.

Genomic information has great potential to enrich our lives, but a critical eye is required to avoid overinterpretation and, potentially, inappropriate actions.

View Abstract


Navigate This Article