In the human CYP21A2
gene, approximately 103 mutations have been described which cause congenital adrenal hyperplasia [38
]. The pseudogene, CYP21A1P
, duplicates the active gene, CYP21A2
. The mutations in the active gene may result from unequal crossing over, resulting in deletion of the gene. The common point mutations in the active gene are present in the pseudogene. Thus, point mutations in the active gene may result from gene conversion. Steroid 21-hydroxylase deficiency is observed worldwide and the ethnic specificity of the mutations in the CYP21A2
gene is of interest.
In the CYP21A2 gene, the frequent V281L (1685 G to T) mutation observed in Ashkenazi Jews, the IVS2 AS -13 (A/C to G) mutations in Yupik speaking Eskimos of Western Alaska, and the large gene deletion in Native Americans demonstrate that there is ethnic specificity of the mutations in CYP21A2 gene. This may have resulted from an ancient founder effect, a hot spot in the gene, unequal crossing over during meiosis or gene conversion of point mutations in the pseudogene. The mutation in V281L (1685 G to T) in Ashkenazi Jews also occurs in other populations, which makes a founder effect less likely. However, since the Ashkenazi Jews are an endogamous population, a combination of both a hot spot and the founder effect is possible. The periodic pogroms experienced by the Ashkenazi Jewish population in Europe may also have contributed to a genetic bottleneck effect. This would have resulted in a severe reduction in the population, which would have reduced the diversity of the original gene pool. The mutation could then have occurred in this small population; and then through genetic drift, the reduced population would have slowly increased. Thus, the mutation would have become prevalent in the expanded population.
The V281L (1685 G to T) mutation results in a mild deficiency and is usually observed in the mild, nonclassical phenotype. In certain populations (Asian and Native American), this V281L (1685 G to T) mutation is not observed. This may be because NC21OHD does not occur in that population, or because NC21OHD has not been clinically diagnosed in the population. We have demonstrated that the severe mutations of IVS2 AS -13 (A/C to G) and deletion in the CYP21A2 gene cause the severe form of 21OHD.
In general, our data is consistent with previously published reports of 21OHD mutation frequencies in specific ethnic populations. However, we have found some differences. The mutations in patients from the Emilia-Romagna province in northern Italy reported herein had a large deletion on only 3% of their 106 alleles. A previous report from Emilia-Romagna by Balsamo et al found 13% of 114 alleles with the large deletion mutation [39
]. This is surprising as these patients originated from the same region in Italy. In the Italo-American population from a Southern Italian heritage studied in this paper, 15% of 100 alleles carried the large deletion, which resembles Balsamo et al's report from the northern province of Emilia-Romagna. Despite the similarity in mutation frequency, these two populations are very different in origin. In Iranian patients, we report that 33% of 80 alleles carried the IVS2 AS -13 (A/C to G) mutation, while a previous report from Vakili et al found only 15% out of 60 alleles with this mutation [40
]. Vakili et al also found 10% of 60 alleles carrying the 8 bp deletion (Δ707-714) in exon 3, but this mutation was not detected in our patients. This variance in mutation frequency may be due to regional differences within Iran – our patients come from the area near Tehran, while the patients in Vakili et al's report reside in the Khorasan provinces of northeastern Iran – but has not been established. In our East Indian population, we have found a significant percentage of the Q318X (1994 C to T) mutation, which occurred in 6 of 38 (16%) alleles. This percentage is similar to previous data from Mathur et al, who reported 22% of 46 alleles with the Q318X mutation [41
]. However, we found the large deletion in only 5% of 38 alleles in our East Indian patients, while Mathur at el found 16% of 46 alleles [41
]. The mutation frequencies in our French patients match previous reports [42
], with the exception of the IVS2 AS -13 (A/C to G) mutation. We found 41% of 32 alleles with the IVS2 AS -13 (A/C to G) mutation, whereas previous reports showed only 14% to 23%. The high frequency of the IVS2 AS -13 (A/C to G) mutation in our French population may be due to the lower number of alleles in this study; the previous studies included 3 to 8 times more alleles.
In 21OHD, there is largely a good correlation of genotype with phenotype, so that patients with homozygous mild and heterozygous mild/severe mutations express the nonclassical form of the disease. Patients with classical 21OHD have two severe mutations. However, there are rare examples of genotype/phenotype non-correlation, which will require further exploration of factors that modify the expression of the CYP21A2
gene. Genotype/phenotype noncorrelation occurs in patients with the IVS2 AS -13 (A/C to G) mutation [16
], which may result from the leaky expression of this mutation. The leaky expression could be due to variants in RNA splicing factors. Buchner et al. have shown that a variant of the putative RNA splicing factor SCNM1 resulted in a more severe phenotype in mice with a splice donor mutation in the Scn8a
]. This SCNM1 variant resulted in only 5% of the normal spliced transcript compared to 10% in the normal SCNM1 variant.
The ethnic specificity of the CYP21A2 mutations can guide physicians toward the diagnosis of 21OHD. NC21OHD is underdiagnosed by pediatricians, internists, endocrinologists, gynecologists, and reproductive urologists, as the signs of hyperandrogenism are often mild. Further, newborn hormonal screening for CAH does not detect NC21OHD because of modest or borderline elevation of the index hormone. The high frequency of specific mutations within ethnic groups may stimulate diagnosis of 21OHD by clinicians practicing in communities in which there is a predominant ethnic group described herein. Finally, the genotype/phenotype correlation data presented are valuable in prenatal diagnosis and treatment where diagnosis is usually based on genotype.