An Sp1 Binding Site Polymorphism in the COLIA1 Gene Predicts Osteoporotic Fractures in Both Men and Women
Genetic factors play an important role in the pathogenesis of osteoporosis, and recent studies have shown that a polymorphic Sp1 binding site in collagen type I α1 (COLIA1) gene is associated with bone mass and vertebral fractures in women from the U.K. Information on the predictive value of the COLIA1 Sp1 polymorphism in other populations is limited, however, and no studies have yet been performed in osteoporotic males. In view of this, we analyzed COLIA1 genotypes in relation to bone density and biochemical markers of bone turnover and the presence of osteoporotic fractures in a case-control study of Danish men and women. COLIA1 genotype was determined by polymerase chain reaction analysis of genomic DNA extracted from peripheral blood samples and related to bone mass, biochemical markers of bone turnover, and the presence of fracture in a study of 375 osteoporotic vertebral fracture patients and normal controls. There was no significant effect of COLIA1 genotype on bone mass or biochemical markers when data from the control group (n = 195) and fracture group (n = 180) were analyzed separately. However, the genotype distribution was significantly different in the fracture cases compared with age-matched controls (χ2 = 16.48, n = 249, p = 0.0003) due mainly to over-representation of the ss genotype in the fracture patients (14.3% vs. 1.4%), equivalent to an odds ratio for vertebral fracture of 11.83 (95% confidence interval 2.64–52.97) in those with the ss genotype. Similar differences in genotype distribution between osteoporotic patients and controls were observed in both men (χ2 = 11.52, n = 95, p = 0.0032, OR = 2.04) and women (χ2 = 6.90, n = 154, p = 0.032, OR = 1.37). In keeping with the above, logistic regression analysis showed that the ss genotype was an independent predictor of osteoporotic fracture (p = 0.028). This study confirms that the COLIA1 Sp1 polymorphism is significantly associated with osteoporotic vertebral fractures. The association is seen in both men and women, and the effect on fracture risk appears to be partly independent of bone mineral density. Our results raise the possibility that genotyping at the Sp1 site could be of clinical value in identifying individuals at risk of osteoporotic fractures in both genders.
OSTEOPOROSIS IS A COMMON DISEASE with an increasing rate of occurrence, characterized by reduced bone mass and increased fracture risk,1–4 that affects up to 30% of women and 12% of men at some point during life.5 It would therefore be of importance to improve our ability to predict the risk of osteoporosis earlier in life and thereby increase the possibility of prevention. Although osteoporosis is a multifactorial disease, genetic factors are important in pathogenesis, since they account for 75–85% of the variance in peak bone mass between individuals6–8 and play a role in regulating bone turnover and the rate of bone loss.9,10 The genes responsible for these effects are incompletely defined, but recent work from our laboratories and that of other groups has identified polymorphisms of several candidate genes which are associated with bone mass in clinical studies.11–13 One of the most interesting of these is a novel G-T polymorphism affecting a binding site for the transcription factor Sp1 in the collagen type I α1 gene (COLIA1) which is associated not only with low bone mass but with the clinically important condition of osteoporotic vertebral fracture in U.K. women.14 Information on the predictive value of the COLIA1 Sp1 polymorphism in other populations is limited, however, and studies have not been performed on males.15 In view of this, we decided to investigate the relationship between the COLIA1 Sp1 polymorphism and osteoporosis in a case-control study of Danish men and women with osteoporotic vertebral fractures.
MATERIALS AND METHODS
The study was a case control study. The osteoporotic group consisted of 154 women (mean age 64.9 ± 8.6 years; range 33–80 years) and 26 men (mean age 55.4 ± 11.1 years; range 28–78 years) with severe osteoporosis5 as defined by the presence of at least one nontraumatic fracture of the spine, referred to the Endocrinology Department of Aarhus University Hospital. The normal control group was comprised of 128 normal women (mean age 46.9 ± 13.4 years; range 22–79 years) and 67 normal men (mean age 52.2 ± 15.8 years; range 22–78 years) without diseases and not taking medications that could influence bone mass and turnover and who had been recruited from the local community by invitations posted at working places and at general practitioners. From this initial population of 375 individuals, a subgroup of 77 osteoporotic women and 28 osteoporotic men were identified who could be matched for age with one or more healthy individuals, giving an age-matched case-control study of 105 osteoporotic patients and 144 controls.
Bone mass measurements
All but three study subjects had determined bone mineral density (BMD) at the lumbar spine and the hip. BMD values were assessed using dual-energy X-ray absorptiometry on a Hologic 1000 (Hologic Inc., Waltham, MA, U.S.A.) or a Norland bone densitometer (Norland Corp., Fort Atkinson, WI, U.S.A.). On both machines, the Z score was calculated on the basis of BMD measurements performed in a large local normal population.16
Biochemical markers of bone turnover
Serum samples were collected in the morning after an 8-h fasting period. All samples from osteoporotic women were collected before institution of any antiosteoporotic treatment. Urine samples were 24-h samples.
S cross-linked carboxy-terminal telopeptide of type I collagen (S-ICTP) was measured by an equilibrium radioimmunoassay, the intra-assay coefficient of variation (CV) was 5% and the interassay CV was 6%.17 Hydroxyproline (U-OHP) was measured spectrophotometrically with p-dimethylaminobenz-aldehyde substrate according to the manufacturer's directions (Organon Tecnica, B.V., Boxtel, The Netherlands). To avoid contribution of dietary collagen, participants had to follow a diet free of gelatine for 12 h before and during the 24 h of collection. To compensate for sampling errors, OHP excretion was expressed as a ratio relative to creatinine. S carboxy-terminal propeptide of human type I procollagen (S-PICP) was measured by a radioimmunoassay from Farmos Diagnostica (Oulunsalo, Finland). The intra-assay CV was 3%, and the interassay CV 5%. The detection limit was 1.2 μg/l.18
Collagen type I polymorphism
DNA was isolated from whole blood leukocytes as described by Kunkel et al.19 The G→T polymorphism in the Sp1 binding site in the COLIA1 gene was examined by a polymerase chain reaction (PCR)-based method as previously described.14 In short, we constructed a mismatched primer which introduces a restriction site for the enzyme BalI in the polymorphic alleles with the T substitution. The upstream primer was 5′-GTCCAGCCCTCATCCTGGCC-3′ and the downstream primer was 5′-TAACTTCTGGACTATTTGCGGACTTTTTGG-3′. PCR were performed in a final volume of 50 μl containing 0.2–0.4 μg of genomic DNA and Taq DNA polymerase (Boehringer Mannheim, Mannheim, Germany) using standard conditions on an Omnigene Termal cycler (Hybaid Ltd., Ashford, Middlesex, U.K.): 40 cycles of 94°C for 50 s, 62°C for 10 s, ramping a 1°C for 10 s to 72°C, 72°C for 15 s. Prior to the first cycle, initial denaturation was performed at 94°C for 3 minutes, and the last cycle was followed by an extension step of 5 minutes at 72°C. The PCR products were digested overnight with BalI (Promega Biotech Co., Madison, WI, U.S.A.) according to the manufacturer's instructions and analyzed on a 3% agarose gel. The PCR product containing s alleles are cleaved by BalI, resulting in a band of 246 bp compared with the uncleaved S alleles band of 264 bp.
Differences in prevalence of the genotypes between osteoporotic patients and normal controls were tested using the χ2-test. Differences in BMD Z scores and levels of biochemical markers between groups were tested using Student's t-test for unpaired data. Evaluation of the influence of independent variables on fracture risk was tested by logistic regression. The level of significance was set at p = 0.05.
The genotype distribution in both the osteoporotic patients and the normal controls were in Hardy-Weinberg equilibrium (χ2 = 0.11, p = 0.945 [all], χ2 = 2.63, p = 0.268 [osteoporotic patients], and χ2 = 2.94, p = 0.229 [normal controls]).
Table 1 shows the relationship between COLIA1 genotype and bone mass in the original population of 372 individuals (180 osteoporotic patients and 192 controls), who had BMD measurements performed. Individuals with the ss genotype had significantly lower BMD values at the spine and the hip than in individuals who were heterozygous or homozygous for the S allele with clear evidence of a gene dose effect on bone mass at both sites. Tables 2 and 3 show the results in the controls and in osteoporotic patients separately. There were no significant differences in bone mass or density between individuals with different genotypes when the two groups were analyzed separately. There were no differences in age between individuals with different genotypes when the normal controls and the osteoporotic patients were examined separately. However, in the combined group, individuals with the ss genotype were older: 62.5 ± 10.7 years compared with 53.7 ± 15.5 years for individuals with the Ss genotype and 56.5 ± 13.8 years for those with the SS genotype (p = 0.014 by analysis of variance [ANOVA]).
Table TABLE 1. BONE DENSITY AND BIOCHEMICAL MARKERS OF BONE TURNOVER IN RELATION TO COLIA1 GENOTYPE
Table TABLE 2. BONE DENSITY AND BIOCHEMICAL MARKERS OF BONE TURNOVER IN RELATION TO COLIA1 GENOTYPE IN NORMAL CONTROLS
Table TABLE 3. BONE DENSITY AND BIOCHEMICAL MARKERS OF BONE TURNOVER IN RELATION TO COLIA1 GENOTYPE IN OSTEOPOROTIC FRACTURE PATIENTS
Biochemical markers of bone turnover in the three genotypes are presented in Table 1. No differences in either the marker of collagen production (PICP) or in the two markers of collagen degradation (ICTP and OHP/Cr) were demonstrable. Tables 2 and 3 show the results in the normal controls and the osteoporotic patients, when examined separately.
The distibution of COLIA1 genotypes in relation to the presence of osteoporotic fractures in men and women is shown in Table 4. The genotype distribution differed significantly in osteoporotic fracture cases of both genders when compared with controls, mainly due to over-representation of the ss genotype in fracture patients (χ2 = 16.48, n = 249, d.f. = 2, p = 0.00026), equivalent to an odds ratio (OR) for osteoporotic fracture in individuals with the ss genotype of 11.83 (2.64–52.97). Subgroup analysis showed that the same differences in genotype distribution were observed in both men (χ2 = 11.52, n = 95, d.f. = 2, p = 0.0032, OR = 2.04) and women (χ2 = 6.90, n = 154, d.f. = 2, p = 0.032, OR = 1.37) with osteoporotic fractures. Heterozygotes did not show increased risk of osteoporotic fractures (OR = 1.00, NS). The distribution of alleles were also significantly different between osteoporotic patients and controls (χ2 = 11.10, n = 498, d.f. = 1, p = 0.00086, OR = 2.03). This difference could also be demonstrated in the men (χ2 = 8.13, n = 190, d.f. = 1, p = 0.0044, OR = 2.87), but was not significant in the women (χ2 = 3.31, n = 308, d.f. = 1, p = 0.069, OR = 1.61).
Table TABLE 4. DISTRIBUTION OF COLIA1 GENOTYPES IN OSTEOPOROTIC PATIENTS AND NORMAL CONTROLS
To evaluate if the association between the COLIA1 polymorphism and vertebral fractures is independent of BMD, we performed logistic regression. This demonstrated that the presence of the ss genotype was an independent predictor of fractures as was BMD (Table 5). We also performed logistic regression using lumbar spine BMD and age as variables instead of the lumbar spine BMD Z score. ss genotype, BMD, and age were all independent predictors of fractures (data not shown).
Table TABLE 5. LOGISTIC REGRESSION OF COLIA1 GENOTYPES AND LUMBAR SPINE BMD (Z SCORE) AGAINST THE PRESENCE OF FRACTURE
This study has confirmed that the polymorphic Sp1 site in the COLIA1 gene is an important genetic marker for osteoporosis. Individuals with the low bone density-associated ss genotype were 10 times more frequent in the osteoporotic group (14.3%) when compared with the control group (1.4%), reflecting a relative risk of 11.83 for the presence of osteoporotic vertebral fracture in those with the rare ss genotype. The over-representation of s alleles in osteoporotic patients from Denmark are consistent with those found in osteoporotic U.K. women14,20 and in osteoporotic Dutch women,21 suggesting that the COLIA1 polymorphism has potential clinical value as a predictor of osteoporotic vertebral fracture in diverse populations. A preliminary study by Hustmeyer et al. published in abstract form22 did not find any over-representation of ss in the osteoporotic group. However, the normal controls were premenopausal twins with an average age of 35 years, whereas the osteoporotic patients were postmenopausal women with an average age of 67 years.
The over-representation of s alleles in males with osteoporosis is an important finding in light of recent work, which has shown that a paternal history of osteoporosis is a greater predictor of bone mass in families than a maternal history of osteoporosis.15 While this suggests that genetic factors play an even greater role in the pathogenesis of osteoporosis in males and than in females, we are aware of only one previous molecular-genetic study in male osteoporosis, where vitamin D receptor (VDR) gene polymorphisms were analyzed. In this study, however, no relation between VDR polymorphisms and bone mass or calcium absorption was found in 20 osteoporotic men and 28 controls, and there was no over-representation of specific VDR genotypes in osteoporotic patients.23 These observations contrast markedly with those made here, where we found a highly significant over-representation of COLIA1 s alleles in osteoporotic men when compared with controls and a significant reduction in bone mass at the spine and hip in those with the unfavorable Ss and ss genotypes.
Although our data indicate that the COLIA1 polymorphism is significantly associated with osteoporotic fracture, it is important also to point out that COLIA1 is only one of many genes which predispose to osteoporosis. Thus, while the rare ss genotype conferred a high relative risk of osteoporotic fracture in both genders (relative risk = 10–18), about 52% of the osteoporotic fracture patients were homozygous for the favorable S allele, suggesting the involvement of other genes. These observations are consistent with the results of segregation analysis6 and studies in families which have shown that bone mass is likely to be under the control of several genes with modest effects, rather than a small number of genes with large effects.15,24
We found no effect on bone mass or density in relation to genotype when the controls and osteoporotic patients were considered seperately, but there was an association with spine BMD when data were pooled for both groups. This finding is consistent with the hypothesis that COLIA1 genotype predisposes to fracture at least in part by an effect on bone mass. However, the magnitude of the difference in BMD between genotypes must be treated with caution in view of the over-representation of the ss genotype in the fracture cases. Since the fracture patients had lower BMD values than the controls, combining data from both groups would tend to amplify any genotype-related effects on bone mass which existed. Indeed, the lack of an association between COLIA1 genotype and BMD in the controls, which we found in this study, is in accordance with recent data, published in abstract form, which have shown no association between COLIA1 genotype and bone mass in certain populations.25–27 Nonetheless, data from other large population-based cohorts in The Netherlands21 and France26 has shown an association between COLIA1 genotype and BMD, while also showing that the effect is relatively small. Since the number of individuals studied here in the control and fracture groups was relatively small and comprised of both males and females, we may have had limited power to detect genotype differences in BMD.
The cellular and molecular events that underlie the association between COLIA1 alleles and bone mass remain unclear. Although preliminary studies have shown a 3-fold increase in the affinity of Sp1 binding to the polymorphic site in the COLIA1 and evidence of differences in allele-specific transcription in Ss heterozygotes,14,28 the consequences of these observations for collagen synthesis in vivo remain to be determined. We did not, in this study, find any significant differences in collagen turnover judged by biochemical markers. This does not exclude that individuals with the ss genotype have an altered ratio between the produced amount of COLIA1 and COLIA2. Alterations in this ratio could lead to reduced peak bone mass or altered bone structure, as seen in osteogenesis imperfecta. If individuals with the ss genotype, due to their altered collagen production, have reduced trabecular thickness, they would have higher risk of trabecular perforations during age- or menopause-related high turnover periods. This would reduce bone strength proportionally more than the accompanying loss of bone density. This could be an explanation for the finding that ss genotype predicts fractures independently of bone density. The hypothesis, that the ss genotype is associated with altered collagen production would also fit with our previous findings, that women with spinal fractures have reduced capacity of bone formation evaluated by histomorphometry29 and with a study by Uitterlinden et al.,30 demonstrating increased rate of bone loss in individuals with the ss genotype.
Whatever the underlying mechanisms, our study has demonstrated a highly significant over-representation of the s alleles in severe osteoporosis and for the first time has identified a genetic marker of bone mass in males, raising the possibility that genotyping at the COLIA1 Sp1 site may be of clinical value in identifying subgroups of individuals in both genders, who are at increased risk of osteoporosis.
The authors thank Elsebet Løkke and Grace Taylor for technical assistance.