Could polymorphisms of some hormonal receptor genes, involved in folliculogenesis help in predicting patient response to controlled ovarian stimulation?

Association of FSHR gene polymorphisms with poor ovarian response in patients undergoing IVF: A systematic review and meta-analysis

BACKGROUND

The results of studies on the association between polymorphisms in the FSHR gene and the risk of POR undergoing IVF have been inconsistent with each other, so we conducted a meta-analysis of all the available studies to explore the association between polymorphisms in the FSHR gene and the risk of POR.

METHODS

Literature that met the inclusion criteria was collected by searching six electronic databases and basic data from included studies were extracted. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated to assess the strength of association between follicle-stimulating hormone receptor (FSHR) gene polymorphism and poor ovarian response (POR) risk. Begg's and Egger's tests were used to determine whether there was publication bias, and sensitivity analysis and TSA analysis were used to verify the stability and reliability of the results.

RESULTS

We included 24 articles, 22 of which explored rs6166, including 2,206 cases and 3,897 controls. 6 articles explored rs6165, including 444 cases and 875 controls. Under additive, heterozygote, and dominant models, rs6166 was significantly associated with POR (S vs. N: OR = 1.29, 95 % CI = 1.05-1.59, P = 0.017; NS vs. NN: OR = 1.33, 95 % CI = 1.02-1.74, P = 0.038; NS + SS vs. NN: OR = 1.38, 95 % CI = 1.04-1.84, P = 0.025). In ethnicity-based subgroup analyses, the additive, homozygote, heterozygote, and dominant models increased Asian POR risk. Among the five genetic models, rs6165 was significantly associated with POR (T vs. C: OR = 1.64, 95 % CI = 1.25-2.16, P = 0.000; TT vs. CC: OR = 2.76, 95 % CI = 1.43-5.32, P = 0.003; CT vs. CC: OR = 1.58, 95 % CI = 1.19-2.10, P = 0.001; TT vs. CC + CT: OR = 2.32, 95 % CI = 1.67-3.23, P = 0.000; CT + TT vs. CC: OR = 1.80, 95 % CI = 1.22-2.65, P = 0.003). In ethnicity-based subgroup analyses, all five genetic models increased the risk of POR in Caucasians.

CONCLUSION

According to the current meta-analysis, the rs6166 S allele was significantly associated with an increased risk of POR, especially in Asian populations. The rs6165 T allele was significantly associated with an increased risk of POR, especially in Caucasian populations.

m.4216 T > C polymorphism in JT cluster determines a lower pregnancy rate in response to controlled ovarian stimulation treatment

PURPOSE

To analyze the influence of Caucasian mitochondrial haplogroups on controlled ovarian stimulation outcome (COS), embryo (E), and pregnancy success.

METHODS

In a Caucasian population (n = 517) undergoing COS, mitochondrial haplogroups and physiological parameters were determined. Patients were classified, according to Bologna criteria, as good (>3)/poor ≤3) responder, on dependence of recruited oocytes (RO), and in pregnancy/non-pregnancy groups. Haplogroups were determined by sequencing mitochondrial hypervariable sequence I and confirmed by polymerase chain reaction (PCR), followed by restriction fragment length polymorphisms (RFLP).

RESULTS

The rank of total dose of FSH (TD FSH) was similar in all clusters/haplogroups, except in JT, which is narrower (950-3,650 IU), particularly in T (1,350-3,650 IU). The statistical analysis showed higher RO and E in JT when compared to U, although it was only Uk which accumulated significantly in pregnancy respect to JT. Pearson's correlations between TD FSH and RO showed negative statistical significance in all population (P = 0.001), H (P = 0.03), JT (P = 0.01), and T (P = 0.03). The percentage of contribution of TD FSH on RO was almost nine times in the JT cluster as compared to all population one.

CONCLUSIONS

JT cluster shows a different influence of TD FSH on RO. JT cluster shows higher RO and E than U, but it is Uk which exhibits a significant higher pregnancy rate than JT. The negative influence of the JT cluster on pregnancy success strongly suggests that the m.4216 T > C polymorphism could be responsible.

Updates in diagnosing polycystic ovary syndrome-related infertility

INTRODUCTION

Polycystic ovary syndrome (PCOS) is a condition that affects approximately 13% of reproductive age women and is characterized by androgen excess, menstrual irregularity and altered ovarian morphology. PCOS presents a complex etiology and pathophysiology, which still requires a detailed investigation of biochemical signatures to identify the molecules and mechanisms that govern it.

AREAS COVERED

This narrative review summarizes the main molecular alterations found in the ovarian follicular fluid, endometrium and placenta of women with PCOS, and the genotypes potentially associated with the outcome of infertility treatments in PCOS.

EXPERT OPINION

PCOS is associated with multiple alterations in growth factors, sex steroid hormones, reactive oxygen species, proinflammatory cytokines and adipokines, which contribute to follicle arrest/ anovulation or suboptimal corpus luteum function, and ultimately to menstrual irregularity and hyperandrogenic symptoms. A panel of PCOS biomarkers should include, besides ovarian products, markers of adipose tissue function, insulin resistance, vascular health, and low-grade chronic inflammation. The effects of ovarian stimulation drugs on infertile women with PCOS are likely to be modified by genetic factors, but the available evidence is heterogeneous; therefore, future studies should evaluate standard treatments and pre-specified outcomes of interest to provide more conclusive answers.

The estrogen receptor polymorphisms and controlled ovulation hyperstimulation outcomes: a meta-analysis

Objective: Several studies evaluated the relationship between estrogen receptor (ER) polymorphisms and the outcomes of controlled ovulation hyperstimulation (COH). However, the results remained obscure. The objective of this study was to perform a meta-analysis to investigate the links between ER polymorphisms and COH outcomes.Methods: Eligible studies were identified from the following electronic databases: PubMed, Embase and the Chinese CNKI till Nov. 2021. The odds ratio (OR) and 95% confidence intervals (CIs) for dichotomous data and the weighted mean difference (WMD) with 95% confidence intervals (CIs) for continuous variables were used to calculate correlations between ER polymorphisms and COH outcomes. Based on the inclusion and exclusion criteria, a total of thirteen papers were eventually enrolled in the current meta-analysis.Results: The following were the key findings of this meta-analysis: (1) PvuII polymorphism had a significant relationship with IVF pregnancy outcome in all models, except the recessive model(CC VS TT: OR, 5.51, 95% CI, 1.13,26.84; CC + CT VS TT: OR, 3.73, 95% CI, 1.21,11.57; CT VS TT: OR, 3.19, 95% CI, 1.11,9.16;C VS T: OR, 2.19, 95% CI, 1.15, 4.19), with large or extreme heterogeneity; (2) XbaI polymorphism had a significant association with IVF risk in heterozygous and dominant models(AG VS GG: OR, 0.27, 95% CI, 0.12,0.61; AAAG VS GG: OR, 0.27, 95% CI, 0.12,0.59), with no heterogeneity; (3) (TA)n polymorphism was linked with IVF risk in the homozygous and recessive models(LL VS SS:OR = 3.74, 95%CI = 1.53,9.12;LL VS LS + SS:OR = 2.75, 95%CI = 1.18, 6.38), with no heterogeneity; (4) for Alul polymorphism and its association with POR risk, significant relationship was observed in the recessive model(AA VS AG + GG:OR = 2.27, 95%CI = 1.46, 3.54), with no heterogeneity;(5) PvuII T/C mutation did not predict the follicle number, oocyte number, ratio of follicles to oocytes and the risk of IVF;(6) for XbaI polymorphism and pregnancy outcome of IVF, no significant association was observed under all models; and (7) RsaI polymorphism does not increase the risk of POR under all models.Conclusion: In summary, our meta-analysis found evidence supporting that PvuII polymorphism may serve as a marker in predicting pregnancy rate in IVF-ET, XbaI and (TA)n polymorphisms may be related with infertility, and Alul polymorphism may predict the poor ovarian response to COH. More well-designed investigations are warranted to corroborate these findings.

Correlation of TP53 (rs1625895), TP73 (rs3765730), MMP9 (rs17576), and MTHFR (rs868014) polymorphisms with low ovarian reserve

OBJECTIVE

To investigate the influence of the Single Nucleotide Polymorphisms (SNPs) TP53 rs1625895, TP73 rs3765730, MMP9 rs17576, and MTHFR rs868014 on ovarian reserve (OR) in infertile patients.

STUDY DESIGN

A prospective cross-sectional study was carried out in 145 infertile women. The patients were divided into two groups according to ovarian reserve, characterized by association between AMH levels and AFC:After patient distribution, both groups were compared (LOR X NOR) regarding the genotypes of the SNPs TP53 T/C rs1625895, TP73 G/A rs3765730, MMP9 Gln/Arg rs17576, and MTHFR A/G rs868014.

RESULT(S)

The frequency of the TP53-T/T genotype was greater in the LOR and the TP53-C/C genotype was more frequent in patients with NOR. This association was confirmed by the frequency of alleles, where the presence of the T allele was significantly higher in patients who exhibited LOR (P = 0.0003). The frequency of the TP73-G/G genotype and of the G allele was higher in the LOR group (P = 0.01). Considering the MMP9 gene, the frequency of the Gln/Gln genotype was higher in the LOR group. However, the Gln/Arg genotype and the Arg allele prevailed in the NOR group (P = 0.006). The frequency of the MTHFR-A/A genotype was higher in the LOR group, whereas that of the MTHFR-GG genotype was higher in the NOR group. The presence of allele A was significantly higher in the LOR group (P = 0.002). The regression analysis shows that patients who present the TP53-T/T, TP73-G/G, MMP9-Gln/Gln, and MTHFR-A/A genotypes are 3.6X, 3.1X, 3.2X, and 3.7X more likely of having LOR, respectively. In addition, the association of the TP53/TT + TP73/GG genotypes increased the chance of women being included in the LOR group in 5.7-fold.

CONCLUSION(S)

The genotypes TP53-T/T, TP73-G/G, MMP9-Gln/Gln, and MTHFR-A/A increase the chance of women to exhibit LOR. These polymorphisms could be useful as genetic markers of low ovarian reserve in infertile patients.

The FSHR G-29A variant is not associated with the ovarian response to exogenous FSH stimulation

A common genetic variant in the follicle stimulating hormone receptor gene (FSHR) 5'-untranslated region has been previously reported to influence FSHR gene expression. However, studies on the ovarian response to exogenous gonadotropin stimulation are limited. The aim of this study was to evaluate the association of variants at positions -29 of the FSHR gene with the ovarian response to exogenous FSH stimulation in Chinese women. The genotypes of the FSHR gene were assayed using the Sequenom MassARRAY system. Total RNA and protein was extracted from granulosa cells, and FSHR expression at the mRNA and protein levels was assessed using quantitative PCR and western blotting. Our data revealed that there was no association between the FSHR genotype at the -29 position and the outcome of controlled ovarian stimulation. The expression of FSHR, at both the mRNA and protein levels, was similar amongst the different FSHR genotypes assessed, but was significantly reduced in the low responders. These results indicate that the variants caused by mutations at position -29 are not associated with ovarian response, and the low ovarian response to gonadotropin stimulation may be caused by decreased FSHR expression.

Genetic determination of the ovarian reserve: a literature review

The ovarian reserve is one of the most important indicators of female fertility. It allows for the evaluation of the number of viable oocytes. This parameter is actively used in pregnancy planning and in assisted reproductive technology application, as it determines chances of successful fertilization and healthy pregnancy. Due to increased attention towards diagnostic tests evaluating the ovarian reserve, there has been a growing interest in factors that influence the state of the ovarian reserve. True reasons for pathological changes in the ovarian reserve and volume have not yet been explored in depth, and current diagnostic screening methods often fall short in efficacy. In the following review we analyze existing data relating to the study of the ovarian reserve through genetic testing, determining specific characteristics of the ovarian reserve through genetic profiling. We explore existing studies dedicated to finding specific genetic targets influencing the state of the ovarian reserve.

Variants Ala307Ala and Ser680Ser of 307 and 680 FSHr polymorphisms negatively influence on assisted reproductive techniques outcome and determine high probability of non-pregnancy in Caucasian patients

PURPOSE

To determine the influence of different genotypes of Ala307Thr and Asn680Ser FSHr polymorphisms on controlled ovarian stimulation (COS) outcome and pregnancy.

METHODS

This study collected blood and physiological and clinical parameters of 517 Caucasian patients (Statistical power ≥ 80%) that underwent COS treatment. Genotypes of Ala307Thr and Asn680Ser polymorphisms were determined using PCR amplification followed by Bsu36I and BsrI digestion, respectively.

RESULTS

Ala307Ala and Ser680Ser genotypes associated to worse parameters of COS outcome (preovulatory follicles P = 0.05, in both), justifying their lower pregnancy rate than Non-Ala307Ala, P = 0.01 and Non-Ser680Ser, P = 0.004, respectively or together, (P = 0.003). Within the Non-Ala307Ala group, Thr307Thr genotype showed higher number of fertilized oocytes (P = 0.04) and embryos (P = 0.01) than Non-Thr307Thr, but no influence on pregnancy rate. Ala307Ala and Ser680Ser patients doubled probability of non-pregnancy than Non-Ala307Ala (odds ratio = 2.0) and Non-Ser680Ser (odds ratio = 2.11), respectively. Ala307Ala and Ser680Ser genotypes tend to appear together (P < 0.0001), which increases the probability of non-pregnancy.

CONCLUSIONS

Ala307Ala and Ser680Ser genotypes of 307 and 680 FSHr polymorphisms associate to worse COS outcome than its respective Non-Ala307Ala and Non-Ser680Ser. Within the Non-Ala307Ala genotypes, Thr307Thr, although shows higher Fertilized Oocytes and Embryos, do not influence on pregnancy rate. Ala307Ala and Ser680Ser genotypes double the probability of Non-Pregnancy than their respective Non-Ala307Ala and Non-Ser680Ser genotypes. Furthermore, the strong tendency of these genotypes to appear together worsens the probability of pregnancy in these patients.

Oocyte Morphology and Reproductive Outcomes - Case Report and Literature Review

Oocyte quality could be negatively affected by many factors including smoking, alcohol consumption, obesity, woman's age, endometriosis and controlled ovarian stimulation (COS), during assisted reproductive technology (ART), in addition to genetic factors, such as hormone receptor polymorphisms, for example. We know that the increase in the reactive oxygen species (ROS) due to systemic disorders causes biochemical and morphological changes to the oocytes, interfering with their quality. The oocyte dysmorphism can be expressed through intra and/or extra cytoplasmic changes. In general, the size and number of oocytes' morphological abnormalities are directly related to preimplantation development failure. This case report is based on four in vitro fertilization (IVF) cycles performed by a patient with oocyte dysmorphism in all oocytes captured. The literature review on this topic aims to relate the characteristics of the oocytes, presented in the case report, with research results about the quality and morphology of the oocytes.

Association between GnRH Receptor Polymorphisms and Luteinizing Hormone Levels for Low Ovarian Reserve Infertile Women

The choice of ovarian stimulation protocols in assisted reproduction technology (ART) cycles for low ovarian reserve patients is challenging. Our previous report indicated that the gonadotrophin-releasing (GnRH) agonist (GnRHa) protocol is better than the GnRH antagonist (GnRHant) protocol for young age poor responders. Here, we recruited 269 patients with anti-Müllerian hormone (AMH) < 1.2 ng/mL undergoing their first ART cycles for this nested case-control study. We investigated the genetic variants of the relevant genes, including follicular stimulating hormone receptor (FSHR; rs6166), AMH (rs10407022), GnRH (rs6185), and GnRH receptor (GnRHR; rs3756159) in patients <35 years (n = 86) and patients ≥35 years of age (n = 183). Only the genotype of GnRHR (rs3756159) is distributed differently in young (CC 39.5%, CT/TT 60.5%) versus advanced (CC 24.0%, CT/TT 76.0%) age groups (recessive model, p = 0.0091). Furthermore, the baseline luteinizing hormone (LH) levels (3.60 (2.45 to 5.40) vs. 4.40 (2.91 to 6.48)) are different between CC and CT/TT genotype of GnRHR (rs3756159). In conclusion, the genetic variants of GnRHR (rs3756159) could modulate the release of LH in the pituitary gland and might then affect the outcome of ovarian stimulation by GnRHant or GnRHa protocols for patients with low AMH levels.

An in silico model using prognostic genetic factors for ovarian response in controlled ovarian stimulation: A systematic review

PURPOSE

To study the use of in silica model to better understand and propose new markers of ovarian response to controlled ovarian stimulation before IVF.

METHODS

A systematic review and in silica model using bioinformatics. After the selection of 103 papers from a systematic review process, we performed a GRADE qualification of all included papers for evidence-based quality evaluation. We included 57 genes in the silica model using a functional protein network interaction. Moreover, the construction of protein-protein interaction network was done importing these results to Cytoscape. Therefore, a cluster analysis using MCODE was done, which was exported to a plugin BINGO to determine Gene Ontology. A p value of < 0.05 was considered significant, using a Bonferroni correction test.

RESULTS

In silica model was robust, presenting an ovulation-related gene network with 87 nodes (genes) and 348 edges (interactions between the genes). Related to the network centralities, the network has a betweenness mean value = 102.54; closeness mean = 0.007; and degree mean = 8.0. Moreover, the gene with a higher betweenness was PTPN1. Genes with the higher closeness were SRD5A1 and HSD17B3, and the gene with the lowest closeness was GDF9. Finally, the gene with a higher degree value was UBB; this gene participates in the regulation of TP53 activity pathway.

CONCLUSIONS

This systematic review demonstrated that we cannot use any genetic marker before controlled ovarian stimulation for IVF. Moreover, in silica model is a useful tool for understanding and finding new markers for an IVF individualization.

PROSPERO

CRD42020197185.

Association between FSHR and ESR1 gene variants and ovarian response to gonadotropin in Egyptian women undergoing ICSI treatment

The unpredictable variability in patients' responses to gonadotropins represents one of the most intractable IVF treatment problems. Identifying the genetic variants associated with ovarian responses to gonadotropins is an important step towards developing individualised pharmacogenetics protocols for ovarian stimulation. The purpose of the current study was to evaluate correlations between FSHR rs6165, FSHR rs616, and ESR1 rs2234693 gene variants and the degree of ovarian response to gonadotropin in Egyptian women undergoing ICSI treatment. Two hundred and eighty Egyptian women (mean age of 20-35) undergoing ICSI treatment were enrolled in a cross-sectional study conducted between January 2017 and May 2019. The women were classified into three groups based on ovarian response: normal responders (retrieved oocytes = 4-15) (n = 80), poor responders (retrieved oocytes < 4) (n = 92), and high responders (retrieved oocytes> 15) (n = 108). Genomic DNA was extracted from blood samples, and PCR and DNA sequencing were performed to identify genetic variations in the different study groups. FSHR and ESR1 genetic variants were then compared in normal, poor, and high responders. DNA sequencing results showed significant differences in the frequencies of FSHR rs6166 and ESR1 rs2234693 genotypes in poor responders compared with normal responders (P ≤ 0.001 and P ≤ 0.001, respectively). In contrast, no significant differences in the frequencies of FSHR rs6166, FSHR rs6165, or ESR1 rs2234693 genotypes were observed in high responders compared with normal responders (P ≤ 0.074, P ≤ 0.353, and P ≤ 0.060, respectively). These results suggest that FSHR and ESR1 gene variants could predict the degree of ovarian response to Controlled ovarian hyperstimulation in Egyptian women.

Pharmacogenomic Biomarkers of Follicle-Stimulating Hormone Receptor Malfunction in Females with Impaired Ovarian Response-A Genetic Survey

Follicle-stimulating hormone receptor (FSHR) plays an essential role as one of the most important molecules in response to some of infertility related medications. Impaired ovarian reserve and poor response to such treatments are partially dependent on the FSHR molecule itself. However, the function and drug sensitivity for this receptor may change due to various allele and polymorphisms in the FSHR gene. Studies indicated some of the FSHR-mediated treatments utilized in clinical centers display different outcomes in specific populations, which may arise from FSHR altered genotypes in certain patients. To support the increased demands for reaching the personalized drug and hormone therapy in clinics, focusing on actionable variants through Pharmacogenomic analysis of this receptor may be necessary. The current study tries to display a perspective view on genetic assessments for Pharmacogenomic profiling of the FSHR gene via providing a systematic and critical overview on the genetics of FSHR and its diverse responses to ligands for infertility treatment in females with impaired ovarian responses and show the potential effects of the patient genetic make-up on related binding substances efficacy. All identified functional drug-related alleles were selected through a comprehensive literature search and analyzed. Advanced technologies for the genetic evaluation of them are also discussed properly.

Association of two single nucleotide polymorphisms rs10407022 and rs3741664 with the risk of primary ovarian insufficiency in a sample of Iraqi women

Primary ovarian insufficiency (POI) can be a devastating disease impacting women below the age of forty. This involves a major decrease in the amount and quality of oocytes, or ovarian reserve in a woman. The distribution of single-nucleotide polymorphisms, rs10407022 and rs3741664, in Iraqi people and its association with primary ovarian insufficiency is the main objective of this study. The mean of FSH and LH levels of patients with POI was higher than control, while the mean of AMH levels of patients was lower compared to control. For rs10407022, the GT and TT genotypes were positively associated with the risk of POI. For the rs3741664, the AG genotype was negatively associated with the risk of POI. The results lead to the main conclusion that rs10407022 and rs3741664 polymorphisms may significantly affected the serum levels of AMH and FSH and thus affect POI etiology.

Investigating apoptotic, inflammatory, and growth markers in poor responders undergoing natural in vitro fertilization cycles: a pilot study

This study investigates follicular fluid (FF) from patients with poor and normal ovarian response undergoing natural assisted reproductive technology cycles. We report about (1) cell-free DNA (cfDNA), which reflects apoptosis; (2) corticotropin-releasing hormone (CRH); (3) interleukin (IL)-15, which reflects inflammation; (4) granulocyte colony-stimulating factor (G-CSF); (5) vascular endothelial growth factor (VEGF); and (6) insulin-like growth factor I (IGF-I), which reflects follicular growth. Forty-four poor responders and 44 normal responders-according to the Bologna criteria-were recruited. FF samples were prepared for cfDNA quantification employing Q-PCR and for CRH, IL-15, G-CSF, VEGF, and IGF-I quantification employing ELISA. Statistically nonsignificant different levels of FF cfDNA, CRH, IL-15, VEGF, and IGF-I were observed. Interestingly, statistically significant higher G-CSF levels were observed in normal responders (302.48 ± 474.36 versus 200.10 ± 426.79 pg/mL, P = 0.003). Lower cfDNA integrity was observed in cycles resulting in clinical pregnancy for both groups (normal: 0.07 ± 0.04 versus 0.25 ± 0.17 ng/μL, P < 0.001; poor: 0.10 ± 0.06 versus 0.26 ± 0.12 ng/μL, P < 0.001). The results predominantly showcase similarities between normal and poor responders pertaining to inflammatory, apoptotic, and growth factors. This may be attributed to the employment of natural cycles in order to exclude controlled ovarian stimulation as a factor-indicating its detrimental effect. As G-CSF levels presented significantly higher in normal responders, its vital role in understanding a compromised ovarian response is highlighted.

Expression of genes that regulate follicle development and maturation during ovarian stimulation in poor responders

RESEARCH QUESTION

Sex hormone-binding globulin (SHBG), androgen receptor (AR), LH beta polypeptide (LHB), progesterone receptor membrane component 1 (PGRMC1) and progesterone receptor membrane component 2 (PGRMC2) regulate follicle development and maturation. Their mRNA expression was assessed in peripheral blood mononuclear cells (PBMC) of normal and poor responders, during ovarian stimulation.

DESIGN

Fifty-two normal responders and 15 poor responders according to the Bologna criteria were enrolled for IVF and intracytoplasmic sperm injection and stimulated with 200 IU of follitrophin alpha and gonadotrophin-releasing hormone antagonist. HCG was administered for final oocyte maturation. On days 1, 6 and 10 of stimulation, blood samples were obtained, serum hormone levels were measured, RNA was extracted from PBMC and real-time polymerase chain reaction was carried out to identify the mRNA levels. Relative mRNA expression of each gene was calculated by the comparative 2^-DDCt method.

RESULTS

Differences between mRNA levels of each gene on the same time point between the two groups were not significant. PGRMC1 and PGRMC2 mRNA levels were downregulated, adjusted for ovarian response and age. Positive correlations between PGRMC1 and AR (standardized beta = 0.890, P < 0.001) from day 1 to 6 and PGRMC1 and LHB (standardized beta = 0.806, P < 0.001) from day 1 to 10 were found in poor responders. PGRMC1 and PGRMC2 were positively correlated on days 6 and 10 in normal responders.

CONCLUSIONS

PGRMC1 and PGRMC2 mRNA are significantly decreased during ovarian stimulation, with some potential differences between normal and poor responders.