Comparison of family history and SNPs for predicting risk of complex disease - PubMed (original) (raw)

Comparison of family history and SNPs for predicting risk of complex disease

Chuong B Do et al. PLoS Genet. 2012.

Abstract

The clinical utility of family history and genetic tests is generally well understood for simple Mendelian disorders and rare subforms of complex diseases that are directly attributable to highly penetrant genetic variants. However, little is presently known regarding the performance of these methods in situations where disease susceptibility depends on the cumulative contribution of multiple genetic factors of moderate or low penetrance. Using quantitative genetic theory, we develop a model for studying the predictive ability of family history and single nucleotide polymorphism (SNP)-based methods for assessing risk of polygenic disorders. We show that family history is most useful for highly common, heritable conditions (e.g., coronary artery disease), where it explains roughly 20%-30% of disease heritability, on par with the most successful SNP models based on associations discovered to date. In contrast, we find that for diseases of moderate or low frequency (e.g., Crohn disease) family history accounts for less than 4% of disease heritability, substantially lagging behind SNPs in almost all cases. These results indicate that, for a broad range of diseases, already identified SNP associations may be better predictors of risk than their family history-based counterparts, despite the large fraction of missing heritability that remains to be explained. Our model illustrates the difficulty of using either family history or SNPs for standalone disease prediction. On the other hand, we show that, unlike family history, SNP-based tests can reveal extreme likelihood ratios for a relatively large percentage of individuals, thus providing potentially valuable adjunctive evidence in a differential diagnosis.

PubMed Disclaimer

Conflict of interest statement

CBD, DAH, UF, and NE are employed by 23andMe and own either stock or stock options in the company. PLOS co-founder Michael B. Eisen is a member of the 23andMe Scientific Advisory Board and also holds stock options in the company.

Figures

Figure 1

Figure 1. Test pedigree.

The test pedigree used throughout this paper, consisting of an extended family with multiple aunts and uncles. An arrow designates one particular individual as the “index individual” (or consultand) whose disease risk we wish to predict.

Figure 2

Figure 2. Area under the curve (AUC) plots.

Each cell of the formula image grid corresponds to a different combination of disease characteristics: rows correspond to differing heritabilities (formula image) and columns correspond to differing frequencies (formula image). Within each cell, the subplots compare the AUC when using a complete family history model that accounts for the disease status of every individual in the pedigree (solid red line), a restricted family history model that only considers the number (0, 1, or formula image1) of affected first-degree relatives of the index individual (dashed red line), or genetic risk factors for the index individual only (blue line). For each subplot, the horizontal axis indicates the proportion (formula image) of heritability explained by known SNP associations, and the vertical axis indicates the AUC. Arrows indicate points of equivalence—values of formula image and formula image for which family history and SNP-based methods give the same AUC.

Figure 3

Figure 3. Proportion of heritability explained.

Subpanels (A) and (B) contain contour plots showing the proportion of heritability explained (formula image) by a complete family history model and a restricted family history model, respectively. Horizontal and vertical axes correspond to varying disease frequency (formula image) and heritability (formula image). Lines in each subplot depict the level curves of formula image, i.e., the combinations of formula image and formula image for which the proportion of heritability explained by family history is constant. SNP-based risk models for specific diseases are illustrated by circles (when the SNP-based model outperforms family history) and squares (when family history outperforms the SNP-based model). The circle or square for each SNP-based model has been colored to indicate the estimated proportion of heritability explained by SNPs, using the same color scheme as the contour plot (e.g., blue indicates 25–30% of heritability explained whereas red indicates formula image5% of heritability explained). Note that the performance of SNP-based models shown here reflects only currently known genetic factors for European populations and will change as more associations are discovered.

Figure 4

Figure 4. Distribution of likelihood ratios (LRs).

The subplots show density histograms for the distribution of LRs achieved by complete (solid red line) and restricted (dotted red line) family history models, and probability density functions for the distribution of LRs achieved by genetic factors accounting for either 10% (dotted blue line), 30% (dashed blue line), or 100% (solid blue line) of the heritability of the disease. As a technical aside, since all plots are shown on a logarithmic scale, the density histograms and density functions shown here were derived for formula image rather than formula image itself (see Text S1).

Figure 5

Figure 5. Worked example.

(A) In the family structure shown, the shaded box represents the index individual, whose risk of developing the disease we wish to predict. (B) There are formula image possible combinations of disease status for the individuals in the family. Using the liability threshold model, we compute the probability of each combination; in this example, we assume formula image and formula image. (C) From the joint distribution, we can then compute the disease risk of the index individual for any given family history pattern, as well as the likelihood of particular family history patterns among cases and controls. (D) These quantities then allow us to construct the receiver operating characteristic (ROC) curve for a complete family history-based classifier, from which sensitivity, specificity, PPV, NPV, and AUC can be computed.

Similar articles

Cited by

References

    1. Hindorff LA, Sethupathy P, Junkins HA, Ramos EM, Mehta JP, et al. (2009) Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proc Natl Acad Sci USA 106: 9362–9367. - PMC - PubMed
    1. Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, et al. (2009) Finding the missing heritability of complex diseases. Nature 461: 747–753. - PMC - PubMed
    1. Maher B (2008) Personal genomes: The case of the missing heritability. Nature 456: 18–21. - PubMed
    1. Eichler EE, Flint J, Gibson G, Kong A, Leal SM, et al. (2010) Missing heritability and strategies for finding the underlying causes of complex disease. Nat Rev Genet 11: 446–450. - PMC - PubMed
    1. So HC, Gui AH, Cherny SS, Sham PC (2011) Evaluating the heritability explained by known susceptibility variants: a survey of ten complex diseases. Genet Epidemiol 35: 310–317. - PubMed

Publication types

MeSH terms

Grants and funding

This study was funded by 23andMe. Other than the authors of this paper, who are all employed by 23andMe, the funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

LinkOut - more resources