Embryo Screening: Polygenic Traits and Disease Risk

Several people asked me to comment on this paper, which appeared recently on biorxiv. It seems to be an update of earlier (simulation) analyses by Gwern [16] and Shulman and Bostrom [15] (cited in the paper) on potential gains from embryo selection using quantitative trait predictors (e.g., height, cognitive ability). In the paper the authors analyze real families using actual genetic and phenotype data.

The main limitations given current technology are the number of embryos available from which to select, and the accuracy of the polygenic predictors. The latter will almost certainly improve significantly for some traits in the near future, and for all traits eventually. The number of embryos available for selection may also increase if new methods allow oocytes (eggs) to be produced using stem cell technology (already demonstrated in mice; video).

Screening human embryos for polygenic traits has limited utility
E. Karavani et al.

Genome-wide association studies have led to the development of polygenic score (PS) predictors that explain increasing proportions of the variance in human complex traits. In parallel, progress in preimplantation genetic testing now allows genome-wide genotyping of embryos generated via in vitro fertilization (IVF). Jointly, these developments suggest the possibility of screening embryos for polygenic traits such as height or cognitive function. There are clear ethical, legal, and societal concerns regarding such a procedure, but these cannot be properly discussed in the absence of data on the expected outcomes of screening. Here, we use theory, simulations, and real data to evaluate the potential gain of PS-based embryo selection, defined as the expected difference in trait value between the top-scoring embryo and an average, unselected embryo. We observe that the gain increases very slowly with the number of embryos, but more rapidly with increased variance explained by the PS. Given currently available polygenic predictors and typical IVF yields, the average gain due to selection would be ≈2.5cm if selecting for height, and ≈2.5 IQ (intelligence quotient) points if selecting for cognitive function. These mean values are accompanied by wide confidence intervals; in real data drawn from nuclear families with up to 20 offspring each, we observe that the offspring with the highest PS for height was the tallest only in 25% of the families. We discuss prospects and limitations of PS-based embryo selection for the foreseeable future.

The authors of the paper seem to define "utility" in terms of expected gain in trait value. However, there is also utility in eliminating very negative outcomes, even if they have small probability. This does not shift the average (expected gain) very much but may still be highly desirable. For example, the odds of my house being destroyed by fire or earthquake in the next decade are small, but the outcome is negative enough that I will act to insure against it. If there is a 1% chance of a $100k house being destroyed, the expected loss is only $1k over the period. But without insurance the outcome might be devastating to a family.

One can compare this to screening for Down Syndrome, which has an incidence of roughly 1% (depending on parental age, etc.) but very serious consequences (see podcast discussion below).

At Genomic Prediction we have focused on screening against disease risk rather than on selection for quantitative traits, for both ethical and practical reasons. Even noisy (imperfect) predictors allow the identification of individuals who are high risk outliers -- e.g., are 5x times more likely to get the disease than a typical person.

When considering disease risk the key metric is not the polygenic score itself, because odds ratios are nonlinear functions of the score (or score percentile). For example (note, this is entirely hypothetical), consider 3 embryos with disease risk percentile scores (e.g., Breast Cancer, Type 1 Diabetes, Atrial Fibrillation, Coronary Artery Disease) given by column:

    #1    33   57   64   51

    #2    62   39   36   49

    #3    26   22   52   99.5

Even though the linear averages of the four risk percentiles for all three embryos are similar (contrived to be near 50), embryo #3 has unusually high risk for one condition (e.g., Coronary Artery Disease) and embryos #1 and #2 might be preferred.

Quantifying the utility to the family from this kind of screening is much more complex than for quantitative traits such as height or cognitive ability.

For more on ethical questions related to genetic engineering and embryo selection, see this podcast discussion with Sam Kerstein, chair of the philosophy department at the University of Maryland.