Genomewide scans of complex human diseases: true linkage is hard to find

Abstract

Many "complex" human diseases, which involve multiple genetic and environmental determinants, have increased in incidence during the past 2 decades. During the same time period, considerable effort and expense have been expended in whole-genome screens aimed at detection of genetic loci contributing to the susceptibility to complex human diseases. However, the success of positional cloning attempts based on whole-genome screens has been limited, and many of the fundamental questions relating to the genetic epidemiology of complex human disease remain unanswered. Both to review the success of the positional cloning paradigm as applied to complex human disease and to investigate the characteristics of the whole-genome scans undertaken to date, we created a database of 101 studies of complex human disease, which were found by a systematic Medline search (current as of December 2000). We compared these studies, concerning 31 different human complex diseases, with regard to design, methods, and results. The "significance" categorizations proposed by Lander and Kruglyak were used as criteria for the "success" of a study. Most (66.3% [n=67]) of the studies did not show "significant" linkage when the criteria of Lander and Kruglyak (1995) were used, and the results of studies of the same disease were often inconsistent. Our analyses suggest that no single study design consistently produces more-significant results. Multivariate analysis suggests that the only factors independently associated with increased study success are (a) an increase in the number of individuals studied and (b) study of a sample drawn from only one ethnic group. Positional cloning based on whole-genome screens in complex human disease has proved more difficult than originally had been envisioned; detection of linkage and positional cloning of specific disease-susceptibility loci remains elusive.

Figure 1

Regression analysis of expected and observed hits on chromosomes 1–22 and X. An average of 4.5 “positive” loci showing some evidence of linkage, chosen by different individual thresholds, were reported in every study. The observed hit ratio (%: hits on every chromosome/all 453 hits) is plotted against the expected hit ratio (%: genes on every chromosome/all 37,701 genes of the genome, when chromosome sizes and gene content are rated as suggested by Venter et al. [2001]).

Figure 2

Number of individuals, plotted against minimum attained P value. The correlation of sample size and study “success” in the genome scans reviewed is illustrated. The vertical lines correspond to the significance thresholds suggested by Lander and Kruglyak (1995), and the mean number of individuals studied in each category is shown.

Figure 3

λ_S, Plotted against minimum attained P value. The correlation of λ_S and study “success” in the genome scans reviewed is illustrated. The vertical lines correspond to the significance thresholds suggested by Lander and Kruglyak (1995), and the mean λ_S in each category is shown.