Investigation of the role of X chromosome inactivation and androgen receptor CAG repeat polymorphisms in patients with recurrent pregnancy loss: a prospective case-control study

Abstract

Background: Previous research has revealed that skewed X chromosome inactivation (SXCI) and androgen receptor (AR) CAG polymorphisms are associated with increased risk of recurrent pregnancy loss (RPL); however, the results are conflicting, and the underlying mechanisms remain unclear. This study investigated the role of SXCI and AR CAG polymorphisms in patients with RPL and explored whether the underlying mechanisms were related to the ovarian reserve and preimplantation embryo aneuploidy.

Methods: This was a prospective case-control study carried out in a tertiary hospital-based reproductive medicine center. An external validation RPL cohort was recruited during the study period. Data on baseline and cycle characteristics were collected. X-chromosome inactivation (XCI) was measured using a human AR assay. AR polymorphisms were assessed using quantitative fluorescent polymerase chain reactions and direct sequencing. Blastocysts of the patients with RPL were tested by single nucleotide polymorphism microarray based preimplantation genetic testing for aneuploidy.

Results: In total, 131 patients with idiopathic RPL and 126 controls were included for the case-control study. Patients with RPL exhibited a significantly more skewed XCI distribution pattern (67.71 ± 10.50 vs. 64.22 ± 10.62, p = 0.011), as well as significantly shorter bi-allelic mean (18.56 ± 1.97 vs. 19.34 ± 2.38, p = 0.005) and X-weighted bi-allelic mean (18.46 ± 2.02 vs. 19.38 ± 2.53, p = 0.001) of AR CAG repeats. Multivariate logistic regression models indicated that CAG repeat < 20, SXCI, and duration of stimulation were independently associated with the risk of RPL. However, SXCI and AR CAG polymorphisms were not associated with ovarian reserve or preimplantation embryo aneuploidy in the RPL group, and the same results were attained in a separate validation cohort of 363 patients with RPL.

Conclusion: SXCI and AR CAG polymorphisms are related to RPL; however, these two factors do not lead to RPL by affecting the ovarian reserve or increasing embryo aneuploidy. The roles of SXCI and AR CAG in RPL may involve other mechanisms that require further investigation.

Trial registration: NCT02504281, https://www.

Clinicaltrials: gov (Date of registration, 21/07/2015; date of enrolment of the first subject, 30/07/2015).

Keywords: Androgen receptor CAG repeat polymorphisms; Ovarian reserve; Preimplantation embryo aneuploidy; Recurrent pregnancy loss; X-chromosome inactivation.

Fig. 1

Flowchart of the study. PL = pregnancy loss; PGT-A = preimplantation genetic testing for aneuploidy; ICSI = intracytoplasmic sperm injection; AR = androgen receptor; DOR = diminished ovarian reserve

Fig. 2

 A) Comparison of the XCI skew between the RPL and control groups. The XCI skew is significantly higher in patients with RPL than in the controls (mean ± standard deviation, 67.71 ± 10.50 vs. 64.22 ± 10.62, p = 0.011). XCI = X-chromosome inactivation; RPL = recurrent pregnancy loss B) Comparison of the AR CAG repeat polymorphism between the RPL and control groups. The short allele (16.87 ± 2.39, vs. 17.77 ± 2.37, p = 0.003) and BAM (18.56 ± 1.97 vs. 19.34 ± 2.38, p = 0.005) were significantly shorter in patients with RPL than in controls. The difference in bioactive CAG repeats (XWBM) was more pronounced (18.46 ± 2.02 vs. 19.38 ± 2.53, p = 0.001). AR CAG = androgen receptor CAG; RPL = recurrent pregnancy loss; XWBM = X-weighted-biallelic-mean. C) Percentage of women with XCI skew > 90, 85–90, and < 85 in the RPL and control groups. The prevalence of highly skewed XCI (≥ 85%) and extremely skewed XCI (≥ 90%) were not significantly different between the two groups. XCI = X-chromosome inactivation; RPL = recurrent pregnancy loss D) Comparison of patients with RPL and controls according to CAG repeats. The median CAG repeat length of the control group was 19.50 or 19.72 calculated as BAM or XWBM, such that 20 was used as the cut-off value. The distribution was significantly different between the two groups (chi-square test, p-value = 0.000 for BAM, p-value = 0.000 for XWBM). RPL = recurrent pregnancy loss; XWBM = X-weighted biallelic mean; BAM = biallelic mean.