doi: 10.1161/01.ATV.0000222010.98018.6b
© 2006 American Heart Association, Inc.
Letters to the Editor |
Longitudinal Differences in Familial Combined Hyperlipidemia Quantitative Trait Loci
Robarts Research Institute, London, Ontario, Canada
Robarts Research Institute and Department of Medicine and Biochemistry, Schulich School of Medicine and Dentistry, University of Western Ontario, London, Ontario, Canada
In response:
Previously in Arteriosclerosis, Thrombosis, and Vascular Biology we commented on genome-wide linkage analysis of quantitative trait loci (QTL) as a strategy to solve the genetics of complex traits, such as familial combined hyperlipidemia (FCHL).1 We noted that although this approach has yielded numerous linkage peaks, only rarely have these peaks led to defined molecular genetic determinants.1 We suggested that replication of QTL analysis is an important filtering process to increase confidence in preliminary observations by reducing false-positive leads.1 The usual replication strategy involves repeating the analysis in independent samples and identifying shared peaks. In their letter, Brouwers et al report the use of an alternate replication strategy, namely repeating QTL analysis for FCHL in the same cohort after an elapsed time period of 5 years.2 On first glance, this strategy would appear quite likely to yield replicable results, considering that the same subjects should be stable for both their genotype (definitely) and for their quantitative phenotypes (essentially) over time.
In the baseline study performed using phenotypes collected in 1999,3 Cantor et al identified 4 peaks, using the threshold of P<0.001 to report significant linkage. These peaks were located at chromosome 1p21-31 and 17p11-q23 for apoB, 12p13 for cholesterol, and 4p15-16 for triglycerides. Each QTL was re-examined in their follow-up study of the same subjects using phenotypes collected in 2004.2 First, the unambiguous good news: the peak at 1p21-31 for apoB was replicated, with virtually the same contour and significance observed for the follow-up analysis compared with baseline. This is an ideal example of successful replication. Although it remains unclear whether this QTL will actually lead to the identification of a genomic variant underlying variation in apoB and FCHL, it seems worth pursuing, especially in light of such recent reported associations with FCHL as the Q223R polymorphism in the leptin receptor (LEPR) gene at 1p31.4
For the remaining 3 QTLs, however, replication was modest at best, with both shifting of the peak locations and eroded peak height. For instance, the peak for cholesterol at 12p13 flattened from a Z-score of 3.7 to <1, and the peak for apoB dropped from a Z-score of 4.3 to 1.7, both of which now fall below the threshold for significance. The peak for triglycerides at 4p15-16 had shifted position, and although it was somewhat less eroded, the observed probability value of 0.003 would not have been accepted as a significant positive signal given the prespecified criteria in the original report.
Thus, repeating the analysis produced 1 replicable QTL of 4 that were initially identified as being significant. The glass is thus either one-quarter full or three-quarters empty, depending on one’s perspective and temperament. Although the peak at 1p21-31 for apoB is worth pursuing, it is also probably appropriate that resources are not expended to pursue the other 3 peaks. Such disparities within the same study sample followed longitudinally show that quantitative traits are a moving target.
Typically, replication implies the study of independently ascertained samples; it has somehow been assumed that replication of significant results internally within the same sample would be a foregone conclusion. This longitudinal study provides a glimpse of the apparently tenuous nature of replication in QTL analysis even within the same sample. The need for replication cannot be stressed enough; it must be considered as an essential element of these investigations, perhaps beginning with replication in the same sample studied at different points in time.
Acknowledgments
This work was supported by operating grants from the Canadian Institutes of Health Research (MT14030), the Heart and Stroke Foundation of Ontario, Genome Canada, and the Canadian Diabetes Association. Dr Hegele is a Career Investigator of the Heart and Stroke Foundation of Ontario and holds the Edith Schulich Vinet Canada Research Chair (Tier I) in Human Genetics and the Jacob J. Wolfe Distinguished Medical Research Chair.
References
1. Pollex RL, Hegele RA. Complex trait locus linkage mapping in atherosclerosis: time to take a step back before moving forward? Arterioscler Thromb Vasc Biol. 2005; 25: 1541–1544.
2. Brouwers M, Kono N, van Greevenbroek M, van der Kallen C, Lusis AJ, Cantor RM, de Bruin T. Longitudinal differences in familial combined hyperlipidemia quantitative trait loci. Arterioscler Thromb Vasc Biol. 2006; 26: e108–e109.
3. Cantor RM, de Bruin T, Kono N, Napier S, van Nas A, Allayee H, Lusis AJ. Quantitative trait loci for apolipoprotein B, cholesterol, and triglycerides in familial combined hyperlipidemia pedigrees. Arterioscler Thromb Vasc Biol. 2004; 24: 1935–1941.
4. van der Vleuten GM, Kluijtmans LA, Hijmans A, Blom HJ, Stalenhoef AF, de Graaf J. The Gln223Arg polymorphism in the leptin receptor is associated with familial combined hyperlipidemia. Int J Obes (Lond). In press.
This article has been cited by other articles:
 |
|
 |
|
  M. C. G. J. Brouwers, R. M. Cantor, N. Kono, J. l. Yoon, C. J. H. van der Kallen, M. A. L. Bilderbeek-Beckers, M. M. J. van Greevenbroek, A. J. Lusis, and T. W. A. de Bruin Heritability and genetic loci of fatty liver in familial combined hyperlipidemia J. Lipid Res., December 1, 2006; 47(12): 2799 - 2807. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||