St. Joseph Hospital (C.J.R.), Bangor, Maine 04401; and College of Physicians and Surgeons (J.P.B.), Columbia University, New York, New York 10032
Address correspondence and requests for reprints to: John P. Bilezikian, Department of Medicine, Columbia University College of Physicians and Surgeons, 630 West 168th Street, PH 8 West, New York, New York 10032.
![]() |
Introduction |
---|
![]() ![]() ![]() |
---|
Although the concept of genetic mapping by comparing the inheritance pattern of a trait with the inheritance pattern of a chromosomal region was not new, the advent of recombinant DNA provided the impetus to isolate genes solely on the basis of chromosomal location without respect to function (a.k.a. positional cloning). With the application of PCR technology to rapidly identify various alleles in these loci, the 1990s blossomed into the decade of polymorphisms. Indeed, in a literature search over the past 3 yr, we found more than 7500 peer reviewed manuscripts that include the word "polymorphism" in their title or abstract. And the beat goes on. Many geneticists now believe that the key to defining heritable determinants of complex (i.e., non-Mendelian) traits lies in understanding quantitative differences in single nucleotide polymorphisms. These nucleotide changes are densely spaced throughout the genome, do not change much between generations, and are easily detected with automated systems, thereby opening the door to even faster mapping and screening (2).
Yet, as the 1990s draw to a close, what is the status report of our journey to discovering the genetic underpinnings of complex disorders such as osteoporosis? Do highly polymorphic regions of the human genome offer clues to the pathophysiology of common diseases? The study by Takacs et al. (3) in this issue of The Journal of Clinical Endocrinology & Metabolism provides such insights about those perplexing polymorphisms, by an analysis of the insulin-like growth factor (IGF)-I gene and osteoporosis (3). The IGF-I gene would seem to be an ideal "candidate" to examine for the following reasons. First, it is a protein that has been linked indirectly to acquisition and maintenance of bone mass (4). Second, IGF-I is easily measured in serum and, hence, can be related to bone turnover and bone mass (5). Third, the particular polymorphic microsatellite region of interest is only 1 kilobase upstream of the transcription start site for exon 1 in the IGF-I gene (6). Hence, nucleotide differences located near possible promoter regions of the gene could have an impact on protein expression. Practically, the multiple alleles in this microsatellite, based on variable numbers of CA repeats, are easily identified by PCR (3, 7). Fourth and somewhat ironically, this particular microsatellite was the first to be successfully amplified by PCR in seminal studies by Weber and May (7) in 1989. Their study led to widespread analyses of microsatellite regions by PCR and, in essence, initiated the ongoing search for valid associations between genotypes and phenotypes of complex diseases.
The study by Takacs et al. (3) examined the relationship between bone mineral density (BMD) and alleles in the same microsatellite of the IGF-I gene, as reported previously by Rosen et al. (8). Using both linkage analysis from 542 sister sib pairs and association studies with 363 premenopausal women, these investigators were unable to show any relationship between femoral or spinal BMD and the IGF-I gene locus. Despite the fact that this seems to be a negative study, it is very important for several reasons. In the field of osteoporosis, choosing candidate genetic loci and relating their frequency to skeletal phenotypes has led to numerous, but conflicting, results. Polymorphisms in candidate genes such as the vitamin D receptor, the estrogen receptor, interleukin-1, collagen A1a, and parathyroid hormone genes have shown variable relationships to bone mass, bone loss, and fracture incidence (9, 10). Even under the best of circumstances, most of these candidate genes still account for a relatively small proportion of the variability noted in several different osteoporotic cohorts. This observation does not argue against the importance of genetics in osteoporosis, but rather recognizes the problems inherent in studies of this sort (9). Limitations of association studies include the relatively small numbers of individuals studied, the multigenic components of the bone density phenotype, environmental influences on genetic make-up, gene-gene interaction (epistasis), and generic pleiotropism (i.e., multiple effects of one gene on more than one phenotype; Ref. 10). Thus, the work by Takacs et al. (3) illustrates many of the pitfalls of association studies. Complex phenotypes are not controlled by single genes, even ones such as IGF-I in which circumstantial evidence for a role in bone acquisition is very strong.
But, there are several other lessons to be drawn from their nice study. First, association studies in heterogeneous populations can be problematic. As Takacs et al. (3) point out in their discussion, combining data from African-American and Caucasian subjects revealed a strong effect of the 192-bp allele of the IGF-I polymorphism on femoral BMD, but this effect completely disappeared when race was analyzed separately (3). Clearly, there are differences in allelic frequency for this CA repeat sequence between blacks and whites, as well as recognized differences in BMD between races. Although quite obvious in this setting, population stratification is less distinct in other cohorts of the same race. Therefore, extreme caution should be the rule in performing association studies in heterogeneous populations.
Second, association studies compare unrelated affected and unaffected individuals from a given population. They are case controlled. In contrast, linkage studies involve modeling to explain inheritance patterns of phenotypes and genotypes observed in a pedigree. They are "vertical" in concept (i.e., among generations). In other words, association-type studies focus on population frequencies, whereas linkage studies focus on concordant (or discordant) inheritance patterns. Thus, association and linkage studies are very different approaches, and the terms should not be used interchangeably (11). On the other hand, when association and linkage are used together, such as in this study, to define a relationship between genotype and phenotype, these methods can provide very powerful information about cause and effect (11).
It should be emphasized that for any a priori hypothesis to
be validated when associating polymorphic nucleotide regions to
phenotype, the variance in the gene under study must be shown to cause
a relevant alteration in function or level of the gene product. Yet,
for most of the polymorphisms said to be related to osteoporosis
(excluding collagen type I 1), the true effect of that polymorphism
in functional terms has not been defined (9, 12). So, even if alleles
in noncoding regions of the estrogen or vitamin D receptor can be
related mathematically to bone density, for example, understanding true
cause and effect remains a daunting challenge. Clearly the same lesson
holds for IGF-I, where one would have to show not only that changes in
the IGF-I product result from a given polymorphism, but also that this
change is causally related to alterations in acquisition or maintenance
of bone mass.
We are left with an important but unanswered question from this study and our previous work. Do alterations in the length of a CA repeat region of the IGF-I gene affect IGF-I protein expression, and, if so, is this related in a causal manner to bone size, bone volume, or bone mass? Unfortunately, Takacs et al. (3) were not able to measure serum IGF-I concentrations to prove or disprove the central element of that hypothesis. Our work and one subsequent small preliminary study were able to meet that central tenet, although once again the number of subjects in those studies is relatively small and will require testing in larger cohorts (8, 13). But, other provocative questions are also raised by these studies. For example, is the regulation of IGF-I in the liver and bone similar, and how does this relate to the IGF-I polymorphism and to understanding the relationship between circulating and skeletal IGF-I? The pressing issue of the determinants of IGF-I heritability and their relationship to the IGF-I polymorphism will need additional scrutiny (14). In the meantime, perplexing D(i)PS and single nucleotide polymorphisms will continue to be featured in our ongoing but exciting trip through the human genome.
Received September 8, 1999.
Accepted September 30, 1999.
![]() |
References |
---|
![]() ![]() ![]() |
---|
HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |