| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Surgical and Perioperative Sciences (M.J., P.S.), Urology and Andrology, and Public Health and Clinical Medicine (G.H.), Umeå University Hospital, 901 85 Umeå, Sweden; International Agency for Research on Cancer (J.D.M., S.R.), 69372 Lyon, France; Department of Medical Epidemiology and Biostatistics (F.W., K.B., H.-O.A., H.G.), Karolinska Institutet, 17177 Stockholm, Sweden; Julius Center for Health Sciences and Primary Care (M.V., C.H.v.G.), University Medical Center Utrecht, 85500 Utrecht, The Netherlands; Department of Epidemiology (H.-O.A.), Harvard University, Boston, Massachusetts 02215; and Division of Cancer Epidemiology (R.K.), German Cancer Research Center (DKFZ), 69120 Heidelberg, Germany
Address all correspondence and requests for reprints to: Mattias Johansson, Department of Surgical and Perioperative Sciences, Urology and Andrology, Umeå University, 901 85 Umeå, Sweden. E-mail: Mattias.Johansson{at}oc.umu.se.
| Abstract |
|---|
|
|
|---|
Materials and Methods: We analyzed haplotypes and three haplotype-tagging single nucleotide polymorphisms (htSNPs) in the 3' region of the IGF-I gene in relation to circulating levels IGF-I in 698 control subjects from the CAncer Prostate in Sweden (CAPS) study and 575 cases and controls from the prospective Northern Sweden Health and Disease Cohort (NSHDC) study. We also performed a meta-analysis of these two and four other association studies on genetic variation in the 3' region of the IGF-I gene in relation to circulating IGF-I levels.
Results: The IGF-I haplotype previously associated with prostate cancer risk, labeled "TCC," was associated with elevated levels of IGF-I in the CAPS study (P = 0.02), but not in the NSHDC study. In contrast, two of the three IGF-I htSNPs tagging this haplotype, rs6220 and rs7136446, were associated with elevated levels of IGF-I in the NSHDC (P = 0.03 and P = 0.04, respectively), but not in the CAPS study. In the meta-analysis, the TCC haplotype and the rs6220 SNP were associated with elevated levels of circulating IGF-I (P = 0.001 and P < 0.0001, respectively).
Conclusions: Genetic variation in the 3' region of the IGF-I gene seems to influence circulating levels of IGF-I. This observation is consistent with the hypothesis that variation in the IGF-I gene plays a role in prostate cancer susceptibility by influencing circulating levels of IGF-I.
| Introduction |
|---|
|
|
|---|
| Subjects and Methods |
|---|
|
|
|---|
The CAncer Prostate in Sweden (CAPS) study is a population-based case-control study extensively described previously (9). In brief, cases of prostate cancer were recruited in two rounds (CAPS1 and CAPS2) from four of six regions in Sweden through a rapid ascertainment scheme at each regional oncological center between March 2001 and October 2003. In total, 2975 cases donated a blood sample and filled out a questionnaire concerning demographic, medical, and lifestyle data. Control subjects were randomly selected from the Swedish population register within groups of men matching the case distribution for age (groups of 5-yr interval) and residency. A total of 3153 control subjects were invited to the study. Of these men 1896 (60%) agreed to participate and answered the same questionnaire as the cases. All study participants were asked to donate blood at the nearest health clinic or hospital. The samples were mailed overnight to the Medical Biobank at Umeå University. Leukocytes, erythrocytes, plasma, and serum were separated and stored at –70 C. At the time of this study, plasma was available for the first part of the CAPS study, CAPS1, including 698 control subjects (mean age at blood draw 69.7 yr) available for plasma analysis of IGF-I.
Written informed consent was obtained from all participants, and the research ethical committee at the Karolinska Institutet and Umeå University Hospital approved the study.
The Northern Sweden Health and Disease Cohort (NSHDC)
The NSHDC is a long-term population-based study earlier described in detail (6). In short, since 1985, all residents of the Västerbotten County are invited to a health survey at the age of 40, 50, and 60 yr. The health examination includes measurement of weight, height, and blood pressure, followed by a blood draw. The blood sample is fractioned into plasma, buffy coat, and erythrocyte aliquots, and cryopreserved at –80 C. Subjects included in the present study were originally included in a case-control study of prostate cancer nested within this cohort (6). The original study included measurements of plasma IGF-I in 281 prospective prostate cancer cases and 560 controls matched on age (6 months) and date of blood draw (2 months). In total, 575 subjects were available for genotyping in the present study (mean age at blood draw: 57.8 yr for cases and 58.6 yr for controls), of which 239 had been included as cases in the original study.
All participants signed an informed consent form, and the study was approved by the ethical committee of Umeå University Hospital.
Single nucleotide polymorphism (SNP) selection, genotyping, and hormone measurements
SNPs were selected using a haplotype-tagging approach as previously described (8).
Genotyping was performed by the 5' nuclease assay (TaqMan) (10). Primer and probe sequences are available on request. Genotyping call rates ranged between 95 and 99%, and duplicate concordance rates were higher than 99.7%. All SNPs conformed to Hardy-Weinberg equilibrium.
Measurements of plasma levels of IGF-I in subjects from CAPS were performed by an ELISA by Diagnostic Systems Laboratories (Webster, TX) as described previously (10). IGF-I measurements from prevalent cases were not included in the present study. The mean intrabatch coefficient of variation was 4.1%. IGF-I measurements in the NSHDC were performed using double-antibody, immunoradiometric assays from Immunotech (Marseille, France), as described by Stattin et al. (6). The intraassay coefficient of variation was 13.5%. All hormone analyses were performed at the International Agency for Research on Cancer (IARC, Nutrition and Hormones Group, Lyon, France).
Statistical analysis of CAPS and NSHDC data
We analyzed the relationship between IGF-I levels and polymorphic variants using linear regression models. For each SNP a variable indicating the number of rare alleles carried by an individual was included as a covariate in the regression model. For each haplotype two variables ("dosage variables"), ranging from 0–1.0, indicating the probability for carrying one or two copies of the haplotype (heterozygosity or homozygosity), were calculated using the "tagSNPs" software (11). The dosage variables were then included as covariates in the linear regression model, testing all haplotypes simultaneously, using the homozygote carriers of the most common haplotype as reference category. These statistical analyses were performed in Statistical Analysis System software (SAS Institute Inc., Cary, NC) (12).
Meta-analysis of results from CAPS, NSHDC, including other studies
We conducted a meta-analysis to estimate the combined effect in this and previous studies on the relationship between genetic variation and IGF-I levels. Epidemiological studies published before February 2007 assessing the relationship between SNPs in the 3' region of the IGF-I gene and circulating levels of IGF-I were included in the analysis. We searched the PubMed database and identified two prospective breast cancer studies in which the relevant SNPs had been investigated in relation to IGF-I levels (10, 13). In the first study, Al-Zahrani et al. (13) analyzed nine tagging SNPs within the IGF-I gene in relation to breast cancer risk and circulating levels of IGF-I in 600 men and women within the Medical Research Council Ely study. In the second study, Canzian et al. (10) analyzed five SNPs within the IGF-I gene and circulating levels of IGF-I in 807 cases and 1588 controls participating in the European Prospective Investigation into Cancer and Nutrition (EPIC). Canzian et al. (10) selected SNPs because of potentially functional roles, i.e. SNPs in exons, exon-intron junctions, etc. Through personal communication we also included a third study, analyzing htSNPs in relation to mammographic breast density as well as to circulating levels of IGF-I among 656 women participating in Prospect-EPIC, a Dutch breast cancer screening cohort, which is part of the EPIC study (Verheus, M., J. D. McKay, R. Kaaks, F. Canzian, C. Biessy, M. Johansson, D. E. Grobbee, P. H. M. Peeters, and C. H. v. Gils, unpublished data). Verheus et al. (unpublished data) selected htSNPs with the criteria R2h
0.95, and also included three additional SNPs. This resulted in 18 SNPs, of which seven were located in the region of the 3' block.
Because absolute levels of IGF-I differed substantially between studies, we calculated the within study mean difference in IGF-I levels between wild-type homozygotes and heterozygotes, and between wild-type homozygotes and rare type homozygotes, respectively. In the meta-analysis the estimated differences were then used to calculate the combined genotype-specific effect on IGF-I levels. To investigate heterogeneity between studies, Cochrans Q tests were performed. We used the random effects model when heterogeneity was significant, otherwise the fixed effects model. To assess global significance, we estimated study-specific β-coefficients with corresponding confidence intervals (CIs) based on the genotype-specific level differences. The β-coefficients were then included in meta-analysis as described previously, and the resulting P values are hereafter referred to as P trend.
All reported P values are two sided. All meta-analyses were performed using the "StatsDirect" software (Cheshire, UK) (http://www.statsdirect.com) (14).
| Results |
|---|
|
|
|---|
In total, nine SNPs from three linkage disequilibrium (LD) blocks were selected as htSNPs. In the present study, only the three SNPs tagging block 3 were used because variation in this particular block was previously associated with prostate cancer (8). Details concerning IGF-I gene structure, LD pattern, haplotype-tagging SNPs (htSNPs), and haplotypes in block 3 can be seen in Fig. 1
. We also included one additional SNP (rs2946834), located downstream of block 3, because this SNP has been previously related to elevated levels of IGF-I (13).
|
The "TCC" haplotype, previously shown by us to be associated with an increased risk of prostate cancer (8), was associated with elevated levels of IGF-I for heterozygote carriers in the CAPS study (P = 0.02) but not in the NSHDC study (P = 0.12; Table 1
). The mean increase in IGF-I levels for heterozygote TCC carriers in the CAPS study was 25.6 (95% CI 4.5–46.8) and 18.8 ng/ml (95% CI –4.7 to 42.2) in the NSHDC study.
|
|
In the Dutch Prospect-EPIC study, Verheus et al. (unpublished data) selected 18 SNPs of which seven were located in the region of block 3, although they extended the block to include the SNP rs2946834 as well. The haplotype corresponding to the TCC haplotype, previously demonstrated by us to be associated with prostate cancer risk (8), was identified as TTCAGCC (underlined alleles identify the TCC haplotype) with a frequency of 6% in the Dutch population, compared with a frequency of 5% in the CAPS study (Table 3
). Verheus et al. (unpublished data) also identified an additional haplotype that was not tagged by our three htSNPs, defined by the SNPs rs1520220 and rs5742714.
|
|
|
Canzian et al. (10) analyzed only one SNP (rs6220) within the 3' block and found it to be associated with elevated levels of IGF-I under a dominant model (P = 0.03), but not under a trend model (Ptrend = 0.17).
Only one SNP (rs6220) was analyzed in all studies, and a meta-analysis of the genotype specific IGF-I mean differences compared with the major homozygotes gave a highly significant combined result of 5.6 ng/ml increase for the heterozygotes (95% CI 1.4–9.9) and 11.1 ng/ml for the minor homozygotes (95% CI 4.3–17.9; Ptrend, < 0.0001; Fig. 3
). This result was similar using both random and fixed effects models. All other SNPs analyzed in more than two studies are presented in Fig. 3
. Combining the other SNPs that were analyzed in more than two studies yielded modest but significant associations for all SNPs in the meta-analysis. There was no evidence of heterogeneity between studies apart from the SNP rs2946834 (phet = 0.05).
Overall, we found significant associations with elevated IGF-I levels in the meta-analysis for the SNPs rs6220 (Ptrend < 0.0001), rs1520220 (Ptrend = 0.04), rs2033178 (Ptrend = 0.02), rs7136446 (Ptrend = 0.01), and rs2946834 (Ptrend = 0.005), as well as for the TCC haplotype (P = 0.001).
| Discussion |
|---|
|
|
|---|
Elevated levels of circulating IGF-I have consistently been associated with increased risk of prostate cancer in prospective and case-control studies (4, 5, 6, 7). Although nutrition strongly influences levels of IGF-I in the circulation (15, 16), twin studies have suggested that a large part of the variation is due to hereditary factors (17, 18, 19). Recently we reported a 46% increase in prostate cancer risk associated with a haplotype in a LD block spanning the 3' region of the IGF-I gene (8). We hypothesized that the risk increase may be due to an increase in IGF-I levels caused by rare genetic variants in that region of the gene.
When we analyzed the TCC haplotype in relation to levels of IGF-I, the association was significant for heterozygote carriers in both the CAPS and Verheus et al. (unpublished data) studies, but not in the NSHDC study. Combining all three studies in meta-analysis gave a significant result overall. However, homozygote carriers did not show any association with increased levels of IGF-I in this study or with prostate cancer risk, as reported in our previous study. Because a dose response effect on levels usually would be anticipated for an increasing number of rare alleles, the lack of association among the homozygotes may indicate that the association among the heterozygotes is due to chance or that the TCC haplotype has to be inherited with another genetic variant to produce elevated IGF-I levels. Still, no firm conclusion can be drawn regarding the homozygote TCC carriers because they were rare (0.5%), and estimates of the association displayed a wide confidence interval.
In our previous study, all three SNPs in block 3 showed borderline association with increased prostate cancer risk depending on the analyzed subgroup (8). Odds ratios for individual htSNPs, previously not reported, can be seen in Table 4
. Both rs2033178 and rs7136446 were significantly associated with overall risk, whereas rs6220 was only significantly associated with risk in advanced cases. In analysis of the SNPs in relation to IGF-I levels, none of the block 3 htSNPs showed significant association in the CAPS study. In contrast, both rs6220 and rs7136446 showed significant association with elevated IGF-I levels in the NSHDC study. In the Verheus et al. (unpublished data) study, both rs6220 and rs2033178 were significantly associated with elevated IGF-I levels. After combining the three studies in our meta-analysis, all SNPs showed a modest but statistically significant association with elevated levels of IGF-I. The only SNP analyzed in all five studies, rs6220, gave a highly significant result in meta-analysis and displayed a dose-response trend in IGF-I levels for an increasing number of rare alleles.
|
One potential limitation in the present study is that the study groups have been selected. As noted by Reilly et al. (20), this selection may introduce bias when estimating the genotype effect on intermediate phenotypes, such as circulating levels of IGF-I, that the study was not originally designed to investigate. Another important limitation is the possible existence of additional unpublished studies that we were unaware of. In addition, the inclusion of both sexes in the meta-analysis may also cause some concern because there are systematic differences between men and women in the regulation of circulating IGF-I. To a large part, this can be attributed to differences in GH secretion patterns, i.e. due to sexual dimorphism (21). The question is whether these differences modify the effect that a germline genetic polymorphism may have on circulating levels of IGF-I. The association noted in our study could be the result of increased expression/transcription of the gene or to a coding variant that leads to a protein with longer half-life in the circulation, for example. It is reasonable to assume that such an effect is present in both genders. The association between IGF-I levels and polymorphisms in the 3' region of IGF-I gene noted in both men and women supports this assumption.
In conclusion, our study supports the hypothesis that genetic variation in the 3' region of the IGF-I gene influences levels of circulating IGF-I and, therefore, prostate cancer risk. Because none of the genetic variants that we investigated outshines the others in strength of association, we are unlikely to have tested the causative polymorphism. Functional studies are required to identify the causal genetic variant. This study also gives further support for the causal link between high levels of IGF-I in circulation and increased prostate cancer risk along the lines of Mendelian randomization (22).
| Acknowledgments |
|---|
| Footnotes |
|---|
Disclosure Statement: The authors have nothing to declare.
First Published Online October 2, 2007
Abbreviations: CAPS, CAncer Prostate in Sweden; CI, confidence interval; EPIC, European Prospective Investigation into Cancer and Nutrition; htSNP, haplotype-tagging SNP; LD, linkage disequilibrium; NSHDC, Northern Sweden Health and Disease Cohort; SNP, single nucleotide polymorphism.
Received April 18, 2007.
Accepted September 26, 2007.
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
F. Porcellati, P. Rossetti, P. Candeloro, P. Lucidi, P. Cioli, A. M. Andreoli, E. Ghigo, G. B. Bolli, and C. G. Fanelli Short-Term Effects of the Long-Acting Insulin Analog Detemir and Human Insulin on Plasma Levels of Insulin-Like Growth Factor-I and Its Binding Proteins in Humans J. Clin. Endocrinol. Metab., August 1, 2009; 94(8): 3017 - 3024. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. Palles, N. Johnson, B. Coupland, C. Taylor, J. Carvajal, J. Holly, I. S. Fentiman, I. dos Santos Silva, A. Ashworth, J. Peto, et al. Identification of genetic variants that influence circulating IGF1 levels: a targeted search strategy Hum. Mol. Genet., May 15, 2008; 17(10): 1457 - 1464. [Abstract] [Full Text] [PDF] |
||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |