Dear Editor,
It was with great interest that I read the article by Yaylacıoğlu Tuncay et al.1 entitled “The role of FOXP3 polymorphisms in Graves’ disease with or without ophthalmopathy in a Turkish population” recently published in the Turkish Journal of Ophthalmology. The authors compared three FOXP3 polymorphisms (rs3761549, rs3761548 and rs3761547) between Graves’ disease patients with or without ophthalmopathy and between Graves’ disease patients and controls.
As mentioned by the authors, the forkhead box P3 (FOXP3, MIM: 300292) gene has been mapped to human chromosome Xp11.23. Considering that males and females have one and two X chromosomes, respectively, the genotypic patterns are quite different between the two sexes. Females have three genotypes (including two homozygotes and one heterozygote) and males have only two hemizygous genotypes. Therefore, it is impossible to pool the genotypes of the sex groups. Unfortunately, the authors made a fatal mistake by pooling the genotypes of both sexes. I have already discussed this point in a debate.2 However, I think it is necessary to explain the following points: 1) how we should estimate allele frequencies, and 2) how we should make statistical comparisons in case-control studies for such loci.
According to the STrengthening the REporting of Genetic Association studies statement, in genetic association studies, researchers should compare the observed and expected genotypic values based on Hardy-Weinberg equilibrium (HWE).3
Suppose we have two alleles A1 and A2, at an X-linked polymorphic locus. If the frequencies of A1A1, A1A2, and A2A2 genotypes in females are “a”, “b” and “c”, respectively, and the frequencies of A1 and A2 hemizygous genotypes in males are “d” and “e”, then the allele frequency can be calculated by the counting method. The number of females and males is equal to n1 (=a+b+c) and n2 (=d+e), respectively.
The allele frequencies of A1 and A2 are denoted by p and q, respectively, and are estimated using the following formulas:
p=(2a+b+d)/(2n1+n2)
q=(b+2c+e)/(2n1+n2)
Then, the expected values for the A1A1, A1A2, and A2A2 genotypes in the female samples, calculated using the estimated p and q, are equal to p2, 2pq, and q2, respectively. Finally, the observed and expected values for the genotypes should be compared using the chi-squared test. The degree of freedom is 1.
Because the number of X chromosomes differs between the sexes of the participants, to compare cases with controls (in case-control studies), participants should be stratified by sex. However, when allelic frequency is estimated in the manner described above, it is possible to compare allelic frequencies (and not genotypic frequencies) between all cases and controls, regardless of their sex.
Considering that Yaylacıoğlu Tuncay et al.1 included both sexes among their participants, the reported results should be interpreted with caution. It is recommended that the authors present their data on the genotypes of each polymorphism according to the sex of the participants and reanalyze their data to address the major issues mentioned above. As emphasized elsewhere,3, 4, 5 the researchers should also show that the frequencies of the observed genotypes are not significantly different from their expected frequencies based on HWE. I wish the esteemed authors all the best in their research.