Rev Osteoporos Metab Miner. 2017; 9 (2): 72-81
1 IMIM (Instituto Hospital del Mar de Investigaciones Médicas) – Red Temática de Investigación Cooperativa en Envejecimiento y Fragilidad (RETICEF) – Instituto de Salud Carlos III FEDER – Barcelona (España)
2 Departamento de Medicina Interna – Parque de Salud Mar – Universidad Autónoma de Barcelona – Barcelona (España)
3 IMIM (Instituto Hospital del Mar de Investigaciones Médicas) – Departamento de Oncología Médica – Medical Oncology Department – Parque de Salud Mar – Barcelona (España)
Objetives: Identify putative functional variants in the CYP11A1 and CYP17A1 genes associated with musculoskeletal effects (accelerated bone mass loss and arthralgia) derived from treatment with aromatase inhibitors (AI).
Material and methods: The B-ABLE cohort is a prospective study of postmenopausal women with breast cancer undergoing AI treatment. Bone mineral density in the lumbar spine and femoral neck was measured by densitometry and joint pain using visual analogue scale. From single-nucleotide polymorphisms (SNPs) in genes CYP11A1 (rs4077581, rs11632698 and rs900798) and CYP17A1 (rs4919686, rs4919683, rs4919687, rs3781287, rs10786712, rs6163, rs743572), previously associated with musculoskeletal events, haplotypes were constructed for each pacient from the cohort, and those haplotypes that showed greatest phenotypic differences were chosen (p<0.05). Within each haplotype, patients with extreme phenotypes were chosen for the sequencing of respective genes and identifying functional genetic variants. Finally, a multiple linear regression analysis was carried out considering the models of dominant, recessive and additive genetic inheritance.
Results: No mutation was found in coding regions. A genetic variant (D15S520), in the basal promoter region of gene CYP11A1, was found associated with femoral neck bone loss at 24 month of AI treatment.
Conclusions: Variants in regulatory regions of the CYP11A1 gene could modulate the expression of this gene, thus explaining part of the phenotypic variability found in bone loss of patients undergoing AI treatment.
The use of aromatase inhibitors (AI) as adjuvant therapy after surgery, and/or radiotherapy, and/or chemotherapy, has achieved a significant increase in survival in postmenopausal women diagnosed with breast cancer with hormone receptors (estrogen and/or progesterone) positive (HR), in the initial stages1,2.
The action of aromatase on testosterone and androstenedione produces estradiol and estrone3. These two components constitute the main source of estrogen in postmenopausal women. This aromatization process is performed in peripheral tissues, such as adipose tissue and muscle. Approximately two-thirds of breast tumors have been shown to have aromatase activity, locally producing estrogens in the tumor itself that stimulate the growth of breast tumor cells4. AI directly blocks estrogen production in the tumor and also causes a drastic reduction in circulating estrogen levels5.
Sustained estrogen deprivation due to AI therapy causes an accelerated loss of bone mass, increasing the risk of osteoporotic fracture6. AIs may also produce other adverse musculoskeletal effects, such as arthralgia and muscle pain, which may hinder adherence to therapy during the years of prescribed treatment7,8.
Furthermore, patients treated with AI reportedly present a large inter-individual variability in the appearance and intensity of musculoskeletal symptoms, suggesting that there are factors that may increase their appearance. In this sense, vitamin D levels (Vit D) have been linked to the appearance of arthralgias9. Likewise, there is probably also a genetic basis that modifies, in part, the effect of AI. Several studies have linked genetic variants associated with increased pain and loss of bone mass in women treated with AI of the B-ABLE cohort10,11.
Specifically, single nucleotide polymorphisms (SNPs) in the CYP11A1 gene: rs4077581, rs11632698 and rs900798 were associated with loss of bone mineral density (BMD) at the femoral neck (FN) at 2 years of treatment with IA11. The CYP11A1 gene encodes the cholesterol side chain cleavage enzyme (alternative name: P450scc) that catalyzes the first and limiting step of steroidogenesis, converting cholesterol to pregnenolone. In addition, P450scc can also hydroxylate vitamin D2, D3 and its precursors12,13, suggesting a broad spectrum of functions in cellular metabolism.
On the other hand, seven SNPs of the CYP17A1 gene (rs4919686, rs4919683, rs4919687, rs3781287, rs10786712, rs6163, rs743572) were associated with increased pain at 1 year of treatment with IA10. CYP17A1 (17α-hydroxylase/17,20 lyase) is a key enzyme in the steroidogenic pathway that produces progestins, mineralocorticoids, glucocorticoids, androgens, and estrogens.
None of the SNPs of the CYP11A1 and CYP17A1 genes, previously genotyped, cause non-synonymous changes in protein, nor are they known to have any regulatory function of gene expression.
It is possible that functional variants of the genes involved in both the coding region that would modify enzyme activity and in regulatory regions that would regulate gene expression levels could be implicated in AI side effects. Therefore, the aim of this study is to identify putatively functional variants in the CYP11A1 and CYP17A1 genes.
Material and methods
The B-ABLE cohort (Barcelona-Aromatase induced Bone Loss in Early breast cancer) is the population of a prospective study that includes postmenopausal patients with RH positive breast cancer and treated at the Hospital del Mar de Barcelona. Participants receive AI (letrozole, exemestane or anastrozole) over 5 years, or alternatively after 2 or 3 years of treatment with tamoxifen (3 and 2 years of AI, respectively), according to the American Society of Clinical Oncology’s recommendations, starting within 6 weeks post op or 1 month after the last cycle of chemotherapy14.
Exclusion criteria were: alcoholism, grade 3b renal insufficiency, rheumatoid arthritis, bone metabolic diseases other than osteoporosis, Paget’s disease, osteomalacia, primary hyperparathyroidism, hyperthyroidism, insulin-dependent diabetes mellitus, previous or ongoing treatment with antiresorptive agents, oral corticosteroids or any other drug that could affect bone metabolism except tamoxifen.
Bone mineral density
At the outset and every 12 months until the end of treatment, levels of BMD at the lumbar (LS L1-L4), femoral neck (FN) and total hip (TH) were measured using the dual X-ray energy densitometer (DXA) QDR 4500 SL® (Hologic, Waltham, Massachusetts, USA). The variation coefficient for this technique in our center is 1% in LS and 1.65% in FN. Densitometries with artifacts, degenerative disc disease with osteophytes, osteoarthritis with hyperostosis of the facet joints, vertebral fractures and/or aortic calcifications, and all those that could cause a false increase in BMD, were excluded as in the description of Blake et al.15 It was then analyzed by the relative loss of bone mass.
Visual Analogue Scale
Joint pain was measured using the visual analogue scale (VAS), at baseline, at 3 months and then every 12 months until the end of the study. Joint pain was assessed: hands, shoulders, knees, hips, ankles and feet, on a scale of 1 to 10 with decimals. Subsequently it was analyzed by means of the VAS absolute change.
Data were collected from a large number of clinical variables at the time of recruitment, including age, menarche and menopause ages, lactation time, number of deliveries, previous chemotherapy and radiotherapy, adjuvant treatments, weight, smoking habits and calcium intake through the INDI-CAD survey16.
Construction of haplotypes
Previous studies in the B-ABLE cohort genotyped SNPs located in the CYP11A1 and CYP17A110,11 genes. SNPs that showed a statistically significant association with the evaluated phenotypes were chosen for the construction of haplotypes (Figure 1).
To establish the relationship of the haplotypes of the CYP11A1 gene to the SNPs rs4077581, rs11632698 and rs900798 in the B-ABLE cohort, the haplotype frequencies were calculated with the haplo.em analysis and the most common haplotypes (frequency >0.01).
The CYP17A1 gene haplotypes were constructed in the same manner with the SNPs rs743572, rs6163, rs10786712, rs3781287, rs4919687, rs4919686 and rs4919683. Each haplotype was assigned a code to facilitate its nomenclature during the study.
DNA Extraction and Sanger Sequencing
DNA extraction was performed from peripheral blood using the Wizard® Genomic DNA Purification Kit (PROMEGA). The coding regions, 5’UTR, 3’UTR and proximal promoter (up to -601 bp for CYP11A1 and -589 bp for CYP17A1) were amplified with the primers described in Table 1.
Sequencing was performed using the Sanger method. The sequences were analyzed with the Sequence Scanner program (v1.0) and alignment with the reference sequence (NCBI Reference Sequence: CYP11A1 NG_007973.1 and CYP17A1 NG_007955.1) was carried out through the Multiple Sequence Alignment (EMBL-EBI).
The frequency of the CYP11A1 and CYP17A1 SNPs was estimated using the expectation-maximization algorithm. The association between haplotypes and phenotypes (change in BMD in CF and increased pain) was analyzed using the haplo.glm, based on glm regression analysis, adjusting for age, body mass index (BMI), previous tamoxifen therapy And chemotherapy. The most common haplotype was used as the reference haplotype and the additive model was assumed to obtain a p-value and the β-coefficient relative to the reference haplotype.
The potential differences between the characteristics of the patients selected according to their haplotype and with extreme phenotypes were evaluated with Student’s t-test for independent samples.
The association between the genetic variants found in the sequencing and the extreme phenotypes were analyzed by multiple linear regression, contemplating dominant, recessive and additive genetic inheritance models.
All statistical analyzes were defined as significant with P<0.05. These were performed using the SPSS (version 22) and R for Windows (version 2.15.2) statistical programs using packages, foreign, rms, multtest, plyr, boot, haplo.stats and SNPassoc.
The study protocols have been approved by the Ethical Committee for Clinical Research of the Marine Health Park (2013/5283/I). Approved protocols for obtaining DNA from blood samples were explained to potential participants, who signed an informed consent before being included in the study.
Baseline characteristics of patients in the B-ABLE cohort
Table 2 shows the demographic characteristics, BMD values and the evolution of the musculoskeletal symptomatology by VAS, for the CYP11A1 and CYP17A1 genes, in which the haplotypes were constructed.
The scheme of the procedure to reach the final analysis of genetic association with extreme phenotypes of BMD and musculoskeletal symptomatology by VAS is shown in figure 2.
Construction of the haplotypes of the CYP11A1 gene and the CYP17A1 gene
Table 3 shows the constructed haplotypes and the association analysis of the CYP11A1 and CYP17A1 genes with the BM change in CF at 2 years and increased pain at 12 months of AI treatment, respectively.
In the CYP11A1 gene, the haplotype that showed a major phenotypic difference with respect to the reference haplotype (11.1) was 11.2, where patients carrying haplotype 11.1 in homozygosity had a loss of BMD 4.41 times greater than haplotype carriers 11.2 in homozygosis (Table 4).
In the case of the CYP17A1 gene, haplotypes 17.3 and 17.4 showed statistically significant differences with respect to the reference haplotype (17.1). Patients homozygous for haplotype 17.1 showed an increase in pain 3.26 times more than patients homozygous for haplotype 17.4 (Table 4).
Selection of patients for the genetic study by Sanger sequencing
Based on the results of the haplotype-association analysis, we selected patients from the B-ABLE cohort who had haplotypes (with a 99% probability) showing greater phenotypic differences: for the CYP11A1 gene, The haplotypes 11.1 and 11.2 in homozygosis. For the CYP17A1 gene, we selected patients with haplotypes 17.3 and 17.4, both in homozygosis and in heterozygosity. In addition, patients with haplotype 17.1 and any other haplotype (with the exception of 17.3 and 17.4) were selected (Figure 2 and Table 3).
Later, within each CYP11A1 gene haplotype group, patients who showed an extreme phenotype in CF BMD (greater or less loss of BMD at 24 months of treatment) (n=40) were selected. The same procedure was performed for the haplotype groups of the CYP17A1 gene in which patients with the extreme phenotype for arthralgia (greater or lesser pain increase at 12 months of treatment) (n=39) were selected (Table 5).
Identification of genetic variants and analysis of association with extreme phenotypes
Following sequencing of the CYP11A1 and CYP17A1 genes, several SNPs were found in both genes. None of them corresponded to a non-synonymous change, or in splicing sites and, therefore, a change in the protein sequence was ruled out.
However, in the basal promoter region of the CYP11A1 gene, a genetic variant (D15S520) associated with the BMD variation in CF at 24 months was found (Coefficient β=-6.32; 95% confidence interval (CI): [-8.55, -4.09], p=3.71e-06).
The D15S520 polymorphism is a microsatellite in the -373 bp position that is used as a genetic marker (Sequence Tagged Sites, STS) and consists of the tandem repeat of pentanucleotide (TAAAA) n. In our patients, the number of repetitions observed was 4, 6, 8 and 9.
The haplotype 11.1 was found to correlate with the allele of 4 replicates of the pentanucleotide. In contrast, patients carrying haplotype 11.2 had different alleles of the microsatellite that could be homozygosis or heterozygous, but never the allele with 4 replicates.
AIs have a number of side effects, including the onset or increase of arthralgias and loss of bone mass, thus increasing the risk of fractures. All this can affect compliance with therapy, decrease the quality of life of patients and increase the risk of breast tumor recurrence.
In previous studies, genetic variants of the CYP11A1 and CYP17A1 genes were associated with loss of BMD in FN11 and increased joint pain10, respectively. None of the SNPs associated with these events produced a change in the protein structure and, therefore, a possible functionality of these SNPs in the determination of the event was discarded.
In order to identify putative functional genetic variants that explain the association of these genes with musculoskeletal effects, the coding and regulatory regions of the CYP11A1 and CYP17A1 genes were sequenced.
No variant was found in the coding region that would cause a change in the amino acid sequence of the protein and, therefore, could involve a structural change of the enzyme. However, a genetic variant, D15S520, located in the regulatory region of CYP11A1, was found to be associated with loss of bone mass.
The D15S520 is a microsatellite based repeating pentanucleotide (TAAAA) n located in the CYP11A1 promoter, at 528 bp upstream from the start of gene translation. In our study, this polymorphism was found to be significantly associated with loss of bone mass at 24 months of AI treatment. It has been observed that all patients carrying the 11.1/11.1 haplotype were also carriers of the 4/4 genotype. In the B-ABLE cohort, these patients had a greater predisposition to lose bone mass (-3.014%) than those with haplotypes 11.2/11.2 (-0.683%).
This microsatellite was previously associated with the risk of breast cancer17,18, although there is some controversy concerning the results19,20. The study by Sakoda et al.18 suggested that women with 4 repetitions in homozygosis would have a lower risk of breast cancer. One hypothesis would be that the allele of 4 replicates would affect the expression of the CYP11A1 gene by decreasing estrogen production. As a consequence, lower estrogen exposure would reduce the risk of breast cancer21, but during treatment with AI, the remaining estrogen levels may be lower than those of the carriers of the other alleles, thus increasing the loss of bone mass.
The detection of genetic variants that partly explain the action of AIs on the musculoskeletal system would allow for the development of personalized therapies in order to avoid, or at least anticipate, the side effects of AI. This could improve adherence to the treatment of these patients, which currently stands between 75.5-78.5%, thus avoiding relapses and a new contralateral breast cancer22.
The main limitation of this study is that it does not prove that this microsatellite is really a functional variant, since there are no functional studies of the CYP11A1 promoter that validate this hypothesis. However, the fact that no functional variable was found in the coding regions of any of the genes studied seems to indicate that the observed association between these genes and the phenotypes has to be caused by genetic variants located in regulatory regions. Another limitation of the study is the use of the EVA parameter for the evaluation of the musculoskeletal symptomatology. EVA assumes that pain is a one-dimensional experience that can be measured on a single-point intensity scale. However, the toxicity reported by the patient more comprehensively captures the side effects of therapies (ie, pain) in daily experience and is more consistent with the patient’s quality of life than the clinician-verified toxicity, Thus, being appropriate for the investigation of the musculoskeletal symptomatology. Likewise, the VAS scale ratio allows detecting the percentage differences between the VAS measurements obtained at multiple points in time. Other advantages of the VAS are its ease and brevity of punctuation, minimal intrusiveness and conceptual simplicity.
In conclusion, the D15S520 variant of the CYP11A1 gene promoter could modulate the expression of this gene, thus explaining some of the phenotypic variability found in the loss of bone mass of patients under treatment with AI. Furthermore, no variant has been found in CYP17A1 to explain the increase or decrease in joint pain observed in patients receiving AI. The promoter regions of these genes should be further studied to detect possible genetic variants that could be involved in the regulation of their expression.
Conflict of interest: The authors declare that they have no conflicts of interest in relation to this work.
Funding: This work has been funded by the FEIOMM 2010 and 2012 grants, the Thematic Network for Cooperative Research in Aging and Fragility (RETICEF, RD12/0043/0022), and the support of PI13/00444 Of Science and Innovation). The Generalitat of Catalonia (DIUE 2014 SGR 775) and ERDF funds have also contributed to its financing.
1. Nabholtz JM. Long-term safety of aromatase inhibitors in the treatment of breast cancer. Ther Clin Risk Manag. 2008;4(1):189-204.
2. Gonnelli S, Petrioli R. Aromatase inhibitors, efficacy and metabolic risk in the treatment of postmenopausal women with early breast cancer. Clin Interv Aging. 2008;3(4):647-57.
3. Ma CX, Reinert T, Chmielewska I, Ellis MJ. Mechanisms of aromatase inhibitor resistance. Nat Rev Cancer. 2015;15(5):261-75.
4. Bolufer P, Ricart E, Lluch A, Vazquez C, Rodriguez A, Ruiz A, et al. Aromatase activity and estradiol in human breast cancer: its relationship to estradiol and epidermal growth factor receptors and to tumor-node-metastasis staging. J Clin Oncol. 1992;10(3):438-46.
5. Fabian CJ. The what, why and how of aromatase inhibitors: hormonal agents for treatment and prevention of breast cancer. Int J Clin Pract. 2007;61(12):2051-63.
6. Amir E, Seruga B, Niraula S, Carlsson L, Ocana A. Toxicity of adjuvant endocrine therapy in postmenopausal breast cancer patients: a systematic review and meta-analysis. J Natl Cancer Inst. 2011;103(17):1299-309.
7. Dent S, Di Valentin T, Vandermeer L, Spaans J, Verma S. Long term toxicities in women with early stage breast cancer treated with aromatase inhibitors: data from a tertiary care center. Breast Cancer Res Treat. 2006;100S1(4057):S190-1.
8. Henry NL, Giles JT, Ang D, Mohan M, Dadabhoy D, Robarge J, et al. Prospective characterization of musculoskeletal symptoms in early stage breast cancer patients treated with aromatase inhibitors. Breast Cancer Res Treat. 2008;111(2):365-72.
9. Prieto-Alhambra D, Javaid MK, Servitja S, Arden NK, Martinez-Garcia M, Diez-Perez A, et al. Vitamin D threshold to prevent aromatase inhibitor-induced arthralgia: a prospective cohort study. Breast Cancer Res Treat. 2011;125(3):869-78.
10. Garcia-Giralt N, Rodriguez-Sanz M, Prieto-Alhambra D, Servitja S, Torres-Del Pliego E, Balcells S, et al. Genetic determinants of aromatase inhibitor-related arthralgia: the B-ABLE cohort study. Breast Cancer Res Treat. 2013;140(2):385-95.
11. Rodriguez-Sanz M, Garcia-Giralt N, Prieto-Alhambra D, Servitja S, Balcells S, Pecorelli R, et al. CYP11A1 expression in bone is associated with aromatase inhibitor-related bone loss. J Mol Endocrinol. 2015;55(1):69-79.
12. Nguyen MN, Slominski A, Li W, Ng YR, Tuckey RC. Metabolism of vitamin d2 to 17,20,24-trihydroxyvitamin d2 by cytochrome p450scc (CYP11A1). Drug Metab Dispos. 2009;37(4):761-7.
13. Tuckey RC, Janjetovic Z, Li W, Nguyen MN, Zmijewski MA, Zjawiony J, et al. Metabolism of 1alpha-hydroxyvitamin D3 by cytochrome P450scc to biologically active 1alpha,20-dihydroxyvitamin D3. J Steroid Biochem Mol Biol. 2008;112(4-5):213-9.
14. Servitja S, Nogues X, Prieto-Alhambra D, Martinez-Garcia M, Garrigos L, Pena MJ, et al. Bone health in a prospective cohort of postmenopausal women receiving aromatase inhibitors for early breast cancer. Breast. 2012;21(1):95-101.
15. Blake G, E. Adams J, Bishop N. DXA in Adults and Children. In: Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism. Eighth Edition ed: John Wiley & Sons, Inc.; 2013. p. 249-63.
16. Orozco López P, Zwart Salmerón M, Vilert Garrofa E, Olmos Domínguez C. Predicción de la ingesta total de calcio a través del consumo de lácteos en la población adulta de España. Estudio INDICAD 2001. Aten Primaria. 2004;33(5):237-43.
17. Zheng W, Gao Y-T, Shu X-O, Wen W, Cai Q, Dai Q, et al. Population-Based Case-Control Study of CYP11A Gene Polymorphism and Breast Cancer Risk. Cancer Epidemiol Biomarkers Prev. 2004;13(5):709-14.
18. Sakoda LC, Blackston C, Doherty JA, Ray RM, Lin MG, Stalsberg H, et al. Polymorphisms in steroid hormone biosynthesis genes and risk of breast cancer and fibrocystic breast conditions in Chinese women. Cancer Epidemiol Biomarkers Prev. 2008;17(5):1066-73.
19. Setiawan VW, Cheng I, Stram DO, Giorgi E, Pike MC, Van Den Berg D, et al. A systematic assessment of common genetic variation in CYP11A and risk of breast cancer. Cancer Res. 2006;66(24):12019-25.
20. Yaspan BL, Breyer JP, Cai Q, Dai Q, Elmore JB, Amundson I, et al. Haplotype analysis of CYP11A1 identifies promoter variants associated with breast cancer risk. Cancer Res. 2007;67(12):5673-82.
21. Dumitrescu RG, Cotarla I. Understanding breast cancer risk — where do we stand in 2005? J Cell Mol Med. 2005;9(1):208-21.
22. Font R, Espinas JA, Gil-Gil M, Barnadas A, Ojeda B, Tusquets I, et al. Prescription refill, patient self-report and physician report in assessing adherence to oral endocrine therapy in early breast cancer patients: a retrospective cohort study in Catalonia, Spain. Br J Cancer. 2012;107(8):1249-56.