The authors declare that they have no competing financial interests.
The effect of fetal androgen metabolism-related gene variants on external genitalia virilization in congenital adrenal hyperplasia
Article first published online: 8 OCT 2012
© 2012 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Special Issue: Variant
Volume 84, Issue 5, pages 482–488, November 2013
How to Cite
The effect of fetal androgen metabolism-related gene variants on external genitalia virilization in congenital adrenal hyperplasia., , , , , , , , , .
- Issue published online: 17 OCT 2013
- Article first published online: 8 OCT 2012
- Accepted manuscript online: 14 SEP 2012 12:20PM EST
- Manuscript Revised: 10 SEP 2012
- Manuscript Accepted: 10 SEP 2012
- Manuscript Received: 2 JUL 2012
- FAPESP. Grant Number: 2008/57616-5
- FAPESP. Grant Number: 2008/55546-0
- CNPq. Grant Numbers: 305117/2009-2, # 305743/2011-2
- 21-hydroxylase deficiency;
- CYP3A7, PXR and CAR allelic variants;
- external genitalia virilization;
- fetal androgen metabolism;
- genotype/phenotype correlation
The 21-hydroxylase deficiency (21OHD) is caused by CYP21A2 mutations resulting in severe or moderate enzymatic impairments. 21OHD females carrying similar genotypes present different degrees of external genitalia virilization, suggesting the influence of other genetic factors. Single nucleotide variants (SNVs) in the CYP3A7 gene and in its transcription factors, related to fetal 19-carbon steroid metabolism, could modulate the genital phenotype. To evaluate the influence of the 21OHD genotypes and the CYP3A7, PXR and CAR SNVs on the genital phenotype in 21OHD females. Prader scores were evaluated in 183 patients. The CYP3A7, PXR and CAR SNVs were screened and the 21OHD genotypes were classified according to their severity: severe and moderate groups. Patients with severe genotype showed higher degree of genital virilization (Prader median III, IQR III-IV) than those with moderate genotype (III, IQR II-III) (p < 0.001). However, a great overlap was observed between genotype groups. Among all the SNVs tested, only the CAR rs2307424 variant correlated with Prader scores (r2 = 0.253; p = 0.023). The CYP21A2 genotypes influence the severity of genital virilization in 21OHD females. We also suggest that the CAR variant, which results in a poor metabolizer phenotype, could account for a higher degree of external genitalia virilization.
Congenital adrenal hyperplasia (CAH) is an autosomal recessive disorder most commonly caused by 21-hydroxylase deficiency (21OHD). The disease is characterized by the impairment of cortisol synthesis, with or without additional impairment of aldosterone synthesis, leading to excessive ACTH stimulation of the adrenal glands, which ultimately results in adrenal hyperplasia and increased androgen synthesis .
CAH presents a spectrum of clinical manifestations as a consequence of different residual 21-hydroxylase activity. These manifestations range from prenatal external genitalia virilization in females, with or without neonatal salt-wasting (SW) crisis (classical forms), to late-onset hyperandrogenic symptoms. The 21OHD is the most common cause of ambiguous genitalia in patients with 46,XX disorder of sex development and the virilization of external genitalia results from increased 19-carbon steroid production by adrenal glands that, peripherally, is converted to dihydrotestosterone [1, 2].
21OHD is caused by mutations of the CYP21A2 gene, and a strong correlation between genotypes and clinical forms is well established. Mutations that cause the total and the near total inhibition of enzymatic activity are associated with the SW and simple virilizing (SV) forms of the disease, respectively. Such genotype/phenotype correlations can be extended to the hormone level; patients carrying mutations that abolish the 21OH's enzymatic activity present higher serum 17OH-progesterone (17OHP) levels than those carrying milder mutations [3-5].
However, in our cohort and in other studies, genotypes do not correlate with the severity of external genitalia virilization among CAH females. In general, quite variable degrees of external genitalia virilization are observed among females carrying similar genotypes. In these patients, Prader scores, which reflect the severity of virilization of the external genitalia, vary ranging from mild clitoral enlargement (score I) to complete fusion of labioscrotal folds with a phallic urethra (score V) [6-9]. These findings suggest an influence of other genetic factors modulating the genital phenotype. Therefore, we hypothesized that the genetic variability related to fetal androgen metabolism could account for these discrepancies.
CYP3A7, a hepatic cytochrome, is the main fetal metabolizer of 19-carbon steroids. Its catalytic activity is detected as early as 7–9 weeks of gestation, which is within the period of external genitalia differentiation [10, 11]. The enzymatic activity of CYP3A7 is highly variable among individuals, and part of this variability is attributed to genetic factors [12-14]. It has been shown that CYP3A7 gene polymorphisms are associated with changes in serum 19-carbon steroids levels during postnatal life . Variable CYP3A7 activity can also be regulated by transcription factors, such as pregnane X receptor (PXR) and the constitutive androstane receptor (CAR), both of which are members of the most important family of CYP3A transcription factors. Furthermore, both transcription factors are expressed in highly variable levels early during fetal life, which correlate with CYP3A7 expression . For these two genes, PXR and CAR, some functional polymorphisms have also been described [17-19].
This study is the first to investigate whether single-nucleotide variants (SNVs) related to fetal androgen metabolism could have an influence on the degree of genital virilization in a large cohort of females with 21OHD.
Materials and methods
This is a retrospective study conducted using a protocol approved by the Ethical Committee of the Hospital das Clínicas, Universidade de São Paulo. Written informed consent was obtained from all patients and/or caretakers.
A total of 183 females with the classical form of 21OHD were selected from the state of São Paulo: 134 patients were followed at the Hospital das Clínicas, Faculdade de Medicina, Universidade de São Paulo; 45 patients at the Faculdade de Ciências Médicas, Universidade Estadual de Campinas; and 4 at the Irmandade da Santa Casa de Misericórdia, São Paulo.
The SV form was characterized by the presence of prenatal external genitalia virilization. The SW form was characterized by external genital virilization as well as weight loss, vomiting and volume depletion starting from the neonatal period. Basal serum hormone levels are presented as the median value (IQR = interquartile range) as follows: 17OHP, 151 (89–251) ng/ml; testosterone and androstenedione levels in prepubertal patients, 133 (95–297) ng/dl and 8.5 (5–20) ng/ml, respectively. The mean (±SD) age at diagnosis was 24 (±30) days after birth for SW patients and 4.4 (±4.4) years for SV patients.
Degree of external genitalia virilization
The degree of external genitalia virilization before genitoplasty was obtained retrospectively from the medical records by the same physician from each center and were classified according to Prader scores . Prader scores were well documented in 154 medical records and data from the remaining 27 patients were excluded to minimize bias in the classification of Prader III and IV before genitoplasty. Additionally, two other patients, who had been given prenatal dexamethasone treatment, were excluded as well. One patient presented Prader I, 19 with Prader II, 78 with Prader III, 46 with Prader IV and 10 patients with Prader V. A total of 154 cases were considered, 67 of which were the SV form and 87 of which were the SW form.
Basal serum 17OHP levels were measured by radioimmunoassay, testosterone levels were measured by immunofluorometric assay, and androstenedione levels were measured by chemiluminescence assay. Intra-assay and inter-assay coefficients of variation ranged from 5% to 10%.
Genomic DNA was extracted from peripheral blood samples using the salting-out or phenol extraction technique.
The point mutations of CYP21A2 were previously screened by direct sequencing of the entire gene, and large genomic rearrangements were analyzed by either multiplex ligand-dependent probe amplification (MRC Holland, Amsterdam, the Netherlands) or Southern blotting analysis [21-23]. Mutation segregation was determined by genotype analysis of the patients' parents. On the basis of the mutation analysis results, CYP21A2 genotypes were classified into groups A and B [24, 25]. In group A, both alleles carried mutations that cause total or near-total inhibition of 21-hydroxylase enzyme (≤2%), such as: CYP21A2 deletions and large gene conversions, p.W19X, IVS2-13A/C>G (I2 splice), IVS2-2A>G, 8 bp deletion (p.G110Efs), p.Ser170fs (c.511_512insA), exon 6 cluster (p.I236N, p.V237E and p.M239K), p.Leu307fs (c.923_924insT), p.G291S, p.Q318X, p.R356W, p.R408C, p.G424S and p.R444X. Group B was characterized by the p.I172N mutation (3–7% of residual enzymatic activity) with either homozygosity or compound heterozygosity including other severe mutations from group A. The CYP21A2 genotypes were determined in all 154 females with 21OHD.
CYP3A7, PXR and CAR genes
DNA samples were available for CYP3A7 sequencing in 136 patients and for PXR and CAR gene sequencing in 94 patients. Proximal promoter and coding regions of the CYP3A7 gene, intron 1 of the CYP3AP1 pseudogene and coding regions of the PXR gene were polymerase chain reaction (PCR) amplified as previously described [12, 13, 17, 26, 27]. Exons 1–9 of the CAR gene were PCR amplified with primers designed using Primer 3 program (University of Massachusetts Medical School, USA).
PCR products were purified using the ExoSAP-IT purification Kit (USB Corporation, Cleveland, OH), directly sequenced using the ABI BigDye terminator cycle sequencing kit and subjected to capillary electrophoresis on the ABI PRISM 3100 sequencer (Applied Biosystem, Inc., Foster City, CA). Sequences were analyzed for the presence of variants using the Sequencher TM 4.9 software (Genes Codes Corporation, Ann Arbor, MI). All genotypes resulting in SNVs were confirmed by two independent PCRs followed by sequencing. The access numbers were as follows: CYP3A7-ENSG00000160870, PXR-ENSG00000144852, CAR-ENSG00000143257.
SNVs in Hardy–Weinberg equilibrium and with allelic frequency ≥3% were selected for the correlation analysis along with Prader scores. Six SNVs of the CYP3A7 gene, one SNV of the PXR gene and three SNVs of the CAR gene were classified based on the Dominant Genetic Model. To test the role of the 21OHD genotypes and each of the SNVs (as the independent variables), multiple linear regression analyses were performed, using the Prader score as the dependent variable.
Comparisons of numerical variables among groups were performed using Mann–Whitney or Kruskal–Wallis test when appropriate.
Statistical analysis was performed using the SigmaPlot 10.0 software (Sigmaplot Users' Manual, 1993, Jandel Scientific, San Rafael, CA). Statistical significance was established at the level of p < 0.05.
Frequency of CYP21A2 mutations
The two most frequent CYP21A2 point mutations were IVS2–13A/C>G (I2 splice) and p.I172N in 37.6% and 17.8% of all the alleles screened, respectively. Moreover, other severe mutations that lead to the total impairment of enzymatic activity were identified in 28.9% of the alleles screened. Large gene rearrangements, such as CYP21A2 deletions and large conversions, were identified in 11% of all the alleles screened, whereas novel mutations were identified in 2.4% [28-31].
Correlation between 21OHD genotypes, clinical forms, basal hormone levels and Prader scores
Of the 21OHD females who carried a CYP21A2 genotype, 66% were included in group A, 33% corresponded to group B genotype, and 1% carried new mutations, for which the effect on the 21OH's enzymatic activity is unknown. Within group A, 83% of the patients presented the SW form and 17% presented the SV form. Within group B, the percentages of SW and SV patients were 6.5% and 93.5, respectively. CYP21A2 genotypes from groups A and B showed significant association with both the SW and SV forms (p < 0.001).
The medians (IQR) of the basal 17OHP levels of patients in groups A and B were 160 (89–264) ng/ml and 125 (85–208) ng/ml, respectively (p = 0.174). The medians (IQR) of basal androstenedione levels for prepubertal patients in groups A and B were 13.5 (6–30) ng/ml and 8.0 (4–10) ng/ml, respectively (p = 0.003), and basal testosterone levels were 135 (109–345) ng/dl and 92.4 (63–170) ng/dl, respectively (p = 0.013).
The most frequent CYP21A2 mutations and the corresponding Prader scores are shown in Table 1.
|CYP21A2 mutation||Number of patients (%)||Prader I + II (n)||Prader III (n)||Prader IV (n)||Prader V (n)|
|I2 splice/I2 splice||33 (18)||1||19||11||2|
|I2 splice/null||47 (25)||4||19||20||4|
|I2 splice/p.I172N||15 (8)||2||10||2||1|
The CAH patients belonging to group A (severe 21OHD genotype) had a significantly higher degree of external genitalia virilization than those belonging to group B (moderate 21OHD genotype) (p < 0.001). However, despite this correlation, the severity of external genitalia virilization had a significant overlap between both groups (Fig. 1), the median Prader score was III for group A (III–IV) and also III (II–III) for group B. For group A patients, there was no difference in the median Prader score between those who were carrying null mutations in both alleles and those who had compound heterozygosis with the I2 splice mutation.
Interestingly, there were no differences in median basal testosterone levels among the Prader score (P) groups, P I and P II, 138 (97–254) ng/dl; P III, 142 (88–351) ng/dl; P IV, 125 (95–229) ng/dl; and P V 129 (108–135) ng/dl (p = 0.921).
Distribution of CYP3A7, PXR and CAR allelic variants
All SNVs identified in the screening of the CYP3A7, PXR and CAR genes were previously described, and their frequencies are depicted in Table 2.
|SNV||Gene||NCBI||Change of protein or nucleotide||CAH||Caucasiana||Effect on CYP3A activity|
CYP3A7 gene: in the promoter region, four different alleles were identified: SNV 1, CYP3A7*1B (n = 1 allele); SNV 2, CYP3A7*1C (n = 9 alleles); SNV 3, CYP3A7*1D (n = 1 allele); and SNV 4, CYP3A7*1E (n = 6 alleles), all of which were heterozygous.
In the coding regions, the SNV 5, CYP3A7*2 allele (n = 67 alleles) was identified in linkage disequilibrium with the CYP3A7P1 (CYP3A7_39256-->A) pseudogene allele. On the other hand, SNV 6 (n = 8 alleles), SNV 7 (n = 9 alleles), SNV 8 (n = 18 alleles) and SNV 9 (n = 18 alleles) were identified in the intronic flanking regions (Table 2).
In PXR gene, three missense substitutions were identified: SNV 10 (n = 2 alleles), SNV 11 (n = 4 alleles) and SNV 12 (n = 1 allele). In addition, three other mutations were detected in the intronic regions and the 3′UTR: SNV 13 (n = 4 alleles), SNV 14 (n = 36 alleles) and SNV 15 (n = 4 alleles) (Table 2).
In CAR gene, within the coding regions, the silent SNV 16 was observed in 57 of the alleles, whereas in the intronic regions and the 3′UTR, the following mutations were observed: SNV 17 (n = 57 alleles), SNV 18 (n = 3 alleles), SNV 19 (n = 23 alleles) and SNV 20 (n = 2 alleles) (Table 2).
Correlation between 21OHD genotype, CYP3A7, PXR and CAR allelic variants and Prader score
SNVs 2, 5, 6, 7, 8 and 9 (of CYP3A7 gene), SNV 14 (of PXR gene) and SNVs 17 and 19 (of CAR gene) did not correlate with Prader scores (I–V). Only SNV 16 (rs2307424) and the 21OHD genotype were significantly associated with the degree of external genitalia virilization in a multivariate analysis (r2 = 0.253; p = 0.023), explaining 25% phenotypic genital variability. None of the other SNVs analyzed presented a statistical significance result for correlation (p > 0.005).
It is well-known that there is a strong correlation between the impairment of enzymatic activity resulting from CYP21A2 mutations and the clinical forms of CAH as well as the basal levels of 17OHP [4, 5, 32]. In the Brazilian patients included in this study, except for the gene founder effect of some mutations [28, 29], the mutation distribution was similar to that of other ethnic groups and we also observed a strong correlation between genotype and phenotype. However, three studies [6, 8, 9] as well as our previous report comprising 85 CAH females from a single center  have indicated some controversies in the correlation between genotype and degree of external genitalia virilization.
In the analysis of a recent multicentric study comprising 158 CAH females, a correlation was reported between the severity of 21OHD genotype and the degree of external genitalia virilization . In order to verify these abovementioned discordant results, we increased the size of our series with the completion of a multicentric study involving 154 CAH females. In agreement with the latter European study , we also identified a correlation between 21OHD genotype and degree of external genital virilization. Although, our first paper presented a considerable number of patients (n = 85), we cannot rule out the sample size effect . Despite the correlation between 21OHD genotype and genital phenotype in this study, there was great overlap between the degree of external genitalia virilization and 21OHD genotypes, suggesting the involvement of other factors in modulating the genital phenotype.
In the literature, the association of serum androgen levels has also been evaluated with the genital phenotype. In an Indonesian study, that included CAH females with exceptionally late diagnosis, a positive correlation between androgen levels and enlargement of the clitoris during postnatal life was showed . Moreover, Prader scores also varied substantially among different genotypes in that study. The analysis of androgen levels in our cohort indicated a positive correlation between the severity of 21OHD genotypes and either postnatal testosterone or androstenedione levels, but these latter two were not correlated with Prader scores. We suppose that this lack of correlation could be explained by an alternative fetal pathway in the androgen synthesis, which in conditions where 17OHP accumulates, the DHT synthesis bypasses the conventional androstenedione and testosterone precursors and, instead, makes use of androstenediol as a substrate [2, 34].
Although CAH is a monogenic disorder, the interindividual variability of hyperandrogenic manifestations among patients carrying similar 21OHD genotypes could be modulated by a polygenic tract. Therefore, allelic variants in genes related to fetal androgen metabolism could play an important role in the genital phenotype.
During fetal development, 19-carbon steroids are mainly metabolized by CYP3A7, and 90% of the interindividual variability in the enzymatic activity is genetically determined. Several allelic variants have been described for the CYP3A7 gene. For example, the CYP3A7*1C allele has a stretch of 60 bp substituted by the CYP3A4 promoter. This gene variant leads to its delayed postnatal expression, and it correlates with lower serum DHEAS levels in both healthy female carriers as well as polycystic ovary syndrome patients as compared with wild-type carriers [14, 15, 26]. Another frequent variant, the CYP3A7*2 allele, is associated with higher DHEA hydroxylation . Consequently, we hypothesized that these substitutions could correlate with a less severe degree of genital virilization in CAH. The CYP3A7*1C variant was found in only nine CAH patients, whose Prader scores ranged from II to V. Similarly, the CYP3A7*2 allele, although presenting a higher frequency in our study, was also identified in patients with a wide range of Prader scores. These data suggest that both these CYP3A7 alleles, at least in the heterozygous state, did not influence the CAH genital phenotype.
Recently, it was shown that the expression of CYP3A7 is mainly regulated by the nuclear receptors, PXR and CAR, and that variants of these receptors dramatically impact CYP3A family expression [17, 18, 35]. Considering the functional variants previously described, only SNV 14 was identified in a considerable number of CAH patients without any significant influence on the external genitalia virilization.
The literature has been giving emphasis to SNVs that do not change protein sequence, because they also can modulate an individual's phenotype . We analyzed in this study the influence of 10 such SNVs within the CYP3A7, PXR and CAR genes on Prader scores, and we found that only SNV 16 presented a significant effect on phenotype. Thus, SNV 16 and the 21OH genotype, taken together, explained 25% of all observed phenotypic variability. We speculate that SNV 16 contributes to a poor metabolizer phenotype of 19-carbon steroids, as this variant was found more frequently among highly virilized CAH females. In the literature, there are no functional studies evaluating the effect of this SNV on CAR or CYP3A transcriptional activity. In agreement with our hypothesis that SNV 16 could confer a poor liver metabolizer phenotype, a recent study observed an association of this SNV with early treatment discontinuation in patients presenting with HIV/AIDS due to a higher frequency of side effects .
In conclusion, we observed in this multicentric study that the impairment of enzymatic activity predicted by 21OHD genotypes influences the degree of external genitalia virilization in CAH females. In addition, we observed that the CAR variant SNV 16, which is related to fetal 19-carbon steroids metabolism, had a modulatory effect on the genital phenotype. We speculate that, although CAH is a monogenic disorder, the hyperandrogenic phenotype could be modulated by a polygenic pathway, involving either genes related to the fetal androgen synthesis and/or metabolism.
This study was supported by FAPESP # 2008/57616-5, L. C. K. by FAPESP # 2008/55546-0, T. A. B and B. B. M by CNPq # 305117/2009-2 and # 305743/2011-2, respectively.
- 37German Competence Network for HIV/AIDS Coordinators. Cytochrome P450 2B6 (CYP2B6) and constitutive androstane receptor (CAR) polymorphisms are associated with early discontinuation of efavirenz-containing regimens. J Antimicrob Chemother 2011: 66: 2092–2098., , et al.