Aromatase gene (CYP19A1) variants, female infertility and ovarian stimulation outcome: a preliminary report

18 Progress has been made towards ascertaining the genetic predictors of ovarian stimulation in 19 IVF. Aromatase cytochrome P450, encoded by the CYP19A1 gene, catalyses a key step in 20 ovarian oestrogen biosynthesis. Hence, the aromatase gene is an attractive candidate for 21 genetic studies. This study aimed to examine the genetic influences of CYP19A1 TCT 22 trinucleotide insertion/deletion (Ins/Del) and (TTTA)n microsatellite intronic polymorphisms 23 on ovarian stimulation outcome and aetiology of female infertility. IVF patients (n = 152) 24 underwent ovarian stimulation according to recombinant FSH and gonadotrophin- 25 releasing hormone antagonist protocol. Del/Del homozygous patients with shorter TTTA 26 repeats exhibited decreased ovarian FSH sensitivity in ovarian stimulation, which may reflect 27 variations in aromatase gene expression during early antral follicle development. 28 Accordingly, this study demonstrates correlations between Del allele and shorter (TTTA)n 29 repeat sizes with smaller ovaries (r = −0.70, P = 0.047) and fewer antral follicles (r = 0.21, P 30 = 0.018) on days 3–5 of spontaneous menstrual cycle, respectively. Furthermore, Del 31 variation linked with low-repeat-number (TTTA)n alleles are involved in enhanced genetic 32 susceptibility to unexplained infertility (adjusted OR = 4.33, P = 0.039) and endometriosis (r 33 = −0.88, P = 0.026), which corroborates evidence on the overlapping patient profiles of 34 ovarian dysfunction in both types of female infertility. 35 36 Keywords: aromatase, female infertility, IVF, ovarian stimulation 37 38 39

40 Oestrogens and FSH act synergistically to induce follicular growth and maturation. Oestrogen 41 biosynthesis depends on the collaboration between follicular theca and granulosa cells. 42 Androgens produced in steroidogenic theca cells diffuse into the granulosa layer and are then 43 aromatized into oestrogens (Ryan and Petro, 1966). The conversion of androgens to 44 oestrogens is catalysed by FSH-inducible aromatase cytochrome P450 (Whitlock, 1986). 45 Ovarian aromatase activity is continuously required for follicular cycle progression from 46 growth and maturation to ovulation, and even for luteal function (Ryan, 1982). 47 IVF includes administration of FSH to stimulate multiple follicle development by 48 suppressing the dominant follicle selection and the atresia of subordinate antral follicles. Age 49 and reduced ovarian reserve negatively impact the ovarian response to FSH stimulation 50 during ovarian stimulation in IVF (Kligman and Rosenwaks, 2001). Previous work 51 demonstrated a causative association between anti-FSH autoantibodies and poor ovarian 52 stimulation outcome (Haller et al., 2008). Additionally, variations in FSH receptor (FSHR) 53 and oestrogen receptor (ESR1) genes influence FSH activity during ovarian stimulation 54 (Georgiou et al., 1997; Perez Mayorga et al., 2000; Altmäe et al., 2007). Another focus of 55 interest is the aromatase enzyme, because it catalyses the key step in ovarian oestrogen 56 biosynthesis. Thus, aromatase is an attractive candidate for genetic studies. 57 Aromatase is encoded by the CYP19A1 gene (15q21.1), spanning over 123 kb and comprised 58 of nine (II-X) coding exons. Aromatase is expressed in ovarian, placental, testicular, adipose, 59 bone and brain tissues (Sebastian and Bulun, 2001). Tissue specificity is regulated by the use 60 of nine alternate untranslated first exons located in the large 93 kb gene regulatory unit. 61 Ovarian aromatase expression is controlled by promoter PII within 1 kb upstream of exon II 62 (Sebastian and Bulun, 2001). 63 Inappropriate activation of promoters may underlie the aetiology of oestrogen-driven 64 diseases, such as endometriosis and breast cancer. Although eutopic endometrial tissue lacks 65 aromatase expression, elevated CYP19A1 transcription via the recruitment of ovarian-specific 66 promoter PII is characteristic of pelvic endometriotic lesions (Noble et al., 1996; Zeitoun et 67 al., 1999). In addition to dysregulated promoter activation, several CYP19A1 gene variants 68 increase susceptibility to certain diseases. Common (TTTA)n polymorphism comprised of 7–69 13 repeats in intron 4 has attracted the most attention. Polycystic ovarian syndrome (PCOS) 70 is described by an accumulation of incompletely developed follicles due to low 71 concentrations of local oestrogens and aromatase enzymatic activity. Women with PCOS 72 possess, at greater frequency, shorter CYP19A1 alleles with ≤9 TTTA repeats. Importantly, 73 these PCOS patients show the highest serum testosterone and testosterone/oestradiol ratio 74 during the early follicular phase of the menstrual cycle (Xita et al., 2008). In contrast, longer 75 alleles of 10 or 12 repeats have been suggested as breast cancer risk alleles with excessive 76 aromatase activity (Kristensen et al., 1998; Haiman et al., 2000). 77 TCT trinucleotide insertion (Ins) or deletion (Del) variation occurs upstream of (TTTA)n 78 microsatellite. The three base pair deletion segregates exclusively with (TTTA)7 variant, and 79 generates two alleles: Del-(TTTA)7 and Ins-(TTTA)n (Probst- Hensch et al., 1999). Del-80 (TTTA)7 associates with increased follicular phase serum testosterone and 81 testosterone/oestradiol ratio in premenopausal women, suggesting lower ovarian aromatase 82 activity (Baghaei et al., 2003). 83 While the prevalent interest in CYP19A1 gene variants has emphasized associations with 84 cancers of female reproductive organs, other possible outcomes also seem obvious targets to 85 study. The described genetic variations in CYP19A1 may affect gene expression or aromatase 86 enzymatic activity, and thus result in alterations in regulation of folliculogenesis that impact 87 ovarian stimulation outcome during infertility treatment. Identification of the genetic 88 predictors of ovarian response in IVF would enable clinicians to individualize ovarian 89 stimulation regimen, minimize the risks of cycle cancellation and ovarian hyperstimulation, 90 and maximize the chance of pregnancy. The present study examines the associations between 91 CYP19A1 (TTTA)n repeat and Ins/Del polymorphisms, and ovarian stimulation outcome 92 among Estonian IVF patients. 93 94

95 Patients 96 The study was approved by the Ethics Committee of the University of Tartu, and informed 97 consent was obtained from all 152 participating normally ovulating women undergoing IVF 98 treatment. The patients were 34.0 ± 4.9 (mean ± SD) years old, and had been infertile for at 99 least a year prior to entering the study. Their indications for IVF were as follows: tubal factor 100 infertility (44.1%, n = 67), male factor infertility (31.6%, n = 48), endometriosis (9.2%, n = 101 14), unexplained infertility (9.2%, n = 14), and infertility due to other reasons such as uterine 102 myomas (5.9%, n = 9). The endometriosis stages according to the American Society for 103 Reproductive Medicine revised classification system (ASRM, 1997) were as follows: 104 minimal to mild (III) stages in 10 patients and moderate to severe (IIIIV) stages in four 105 patients. CYP19A1 (TTTA)n allelic variants are known to interfere with follicular 106 steroidogenic properties in the genesis of polycystic ovarian phenotype (Xita et al., 2008), 107 and thus PCOS patients were excluded from the study. 108 Mean ultrasound parameters for right and left ovaries (volume and early antral follicle count) 109 and serum FSH concentration (9.3 ± 5.3 IU/l) were determined between days 3–5 of 110 spontaneous menstrual cycle, which allowed indirect ovarian follicular reserve assessment. 111 Ovarian volume (4.9 ± 2.1 cm3) was calculated using the formula: 0.5(A B C), where A is 112 the longitudinal, B the anteroposterior and C the transverse diameter of the ovary (Sample et 113 al., 1977). Early antral follicles (4.5 ± 1.4 follicles) were counted in longitudinal cross-114 section. All hormonal analyses were conducted using chemiluminescence immunoassay 115 (Immulite 2000; Siemens Healthcare Diagnostics, Deerfield, IL, USA). The within-run (intra-116 assay) precision coefficients of variation (CV) ranged from 2.3 to 3.7 and 6.3 to 15.0% and 117 the total (inter-assay) precision CV ranged from 5.4 to 6.7 and 6.4 to 16.0% for FSH and 118 oestradiol respectively. Associations between ovarian reserve parameters and ovarian 119 stimulation variables have also been described in previous studies (Altmäe et al., 2007; 120 Haller et al., 2008). 121 122 Ovarian stimulation regimen and IVF 123 Ovarian stimulation was conducted according to the gonadotrophin-releasing hormone 124 (GnRH) antagonist regimen. All patients commenced ovarian stimulation with the 125 recombinant FSH (Gonal-F; Serono, Rome, Italy) mean starting dose of 178.3 ± 40.6 IU on 126 day 1–3 of menses, continuing 9.6 ± 0.7 days until 1 day before human chorionic 127 gonadotrophin (HCG) (Ovitrelle; Serono) administration. Daily GnRH antagonist 128 administration (0.25 mg, Cetrotide; Serono or Orgalutran; N.V. Organon, Oss, The 129 Netherlands) was initiated when at least one follicle reached the size of 14 mm. The GnRH 130 antagonists were given for up to 4–5 days, including the day of HCG administration. Final 131 follicular maturation was achieved using 250 g of HCG, followed by ovarian puncture 36 h 132 later. 133 The number of follicles punctured at oocyte retrieval (14.5 ± 6.6) and the number of 134 cumulusoocyte complexes obtained (12.4 ± 6.5) were counted for all participants. Serum 135 oestradiol concentrations on the day of oocyte retrieval (4159.8 ± 4620.3 pmol/l), serum 136 oestradiol concentration per punctured follicle at oocyte retrieval (295.1 ± 264.6 pmol/l) and 137 serum oestradiol concentration per oocyte retrieved (388.9 ± 538.4 pmol/l) were also 138 determined. In addition, follicular fluid oestradiol concentration (2375.4 ± 6924.1 nmol/l) 139 was determined for all women. 140 Both IVF (45.4%, n = 69) and intracytoplasmic sperm injection (ICSI, 54.6%, n = 83) 141 patients participated. The number of mature oocytes (10.1 ± 5.6) was calculated for both IVF 142 and ICSI patients. The maturity of IVF oocytes was assessed 1 day after insemination by 143 counting the fertilized and unfertilized metaphase II (M II) oocytes. ICSI oocytes were 144 considered mature if they had reached M II stage by 4–6 h after oocyte retrieval. The total 145 number of embryos with two pronuclei (embryos = 7.1 ± 4.1) was calculated 16–18 h after 146 microinjection or insemination. The patients had, on average, 3.0 ± 2.8 (42.3 ± 29.6%) good-147 quality day 2 embryos, characterized by having at least four blastomeres and <20% cellular 148 fragments. 149 The following parameters were calculated from the total amount of FSH used for ovarian 150 stimulation (1893.5 ± 482.5 IU) to determine the amount of FSH (IU) administered: (i) per 151 day (196.0 ± 40.8 IU); (ii) to mature one ovarian puncture follicle (184.6 ± 158.1 IU); (iii) to 152 obtain one oocyte (239.2 ± 228.4 IU); (iv) per mature oocyte (303.3 ± 308.8 IU), (v) per 153 embryo (393.3 ± 355.3 IU); and (vi) per good-quality embryo (835.7 ± 693.8 IU). 154 Two day-2 embryos were transferred into the uterus in the majority (87.6%) of IVF and ICSI 155 cycles (2.1 ± 0.3 embryos per transfer). Vaginal progesterone (Lugesteron; Leiras, Turku, 156 Finland) was used for luteal support. Single (n = 31) and twin (n = 16) clinical pregnancies 157 were recognized by the presence of gestational sac(s) with fetal heartbeat on transvaginal 158 sonography at 6–7 weeks of gestation. Implantation (19.6%) and clinical pregnancy (30.9%) 159 rates were calculated per embryo transfer. 160 161 CYP19A1 Ins/Del and (TTTA)n genotyping 162 Genomic DNA was extracted from peripheral EDTA blood using the salting-out method 163 (Aljanabi and Martinez,1997). Polymerase chain reaction (PCR) of CYP19A1 region 164 encompassing both TCT Ins/Del and (TTTA)n polymorphisms was accomplished with 165 fluorescently labelled forward (5-JOE-GGTAAGCAGGTACTTAGTTAG-3) and reverse (5-166 CAAGGTCGTGAGCCAAGGTC-3) primers. Amplification of 50 ng DNA was performed 167 in a total volume of 15 l containing 0.25 mol/l dNTPs (MBI Fermentas, Vilnius, Lithuania), 168 2.5 mmol/l MgCl , 1 PCR buffer (Solis BioDyne, Tartu, Estonia), 10 pmol of primers 169 (Metabion, Martinsried, 0 Germany) and 1U HotStart thermostable DNA polymerase 170 HotFirePol (Solis BioDyne), in an Eppendorf thermal cycler (Eppendorf, Hamburg, 171 Germany). The reactions were initiated with DNA denaturation and enzyme activation at 96C 172 (10 min), followed by 35 cycles of denaturation at 96C (30 s), annealing at 57C (30 s), 173 elongation at 72C (30 s), and final extension at 72C (5 min). The sizes of fluorescently 174 labelled PCR products were estimated using ABI Prism 377 automated DNA sequencer and 175 Genescan 2.1 software (PE Applied Biosystems, Forster City, CA, USA). Rox 500 (PE 176 Applied Biosystems) was used as an internal size standard. DNA sequencing was used to 177 verify the results of fragment size analysis in 8.0% of patients, using forward (5-178 TCATTACAGCTCTCGATTCG-3) and reverse (5-CAAGGTCGTGAGCCAAGGTC-3) 179 primers. 180 Statistical analysis 181 Linear parameters are reported as mean ± SD. R2.3.1A Language and Environment software 182 (Free Software Foundation, Boston, MA, USA) was used for linear and logistic regression 183 analyses. Regression coefficients derived from linear regression analyses are reported in the 184 table and text as r-values. Biallelic mean of (TTTA)n repeat was used in statistical analyses, 185 representing the arithmetic mean of two parental CYP19A1 variants. Women with tubal factor 186 infertility were used as the control group. PS program (PS Power and Sample Size 187 Calculations, Free- Software, Version 2.1.30, 2003, Nashville, TN, USA) was used for 188 statistical power calculations. Statistical significance was set at P < 0.05 for all tests. 189 190

191 CYP19A1 allelic variants and aetiology of female infertility 192 The distribution of CYP19A1 (TTTA)n and Ins/Del allele frequencies is shown in Figure 1. 193 (TTTA)n microsatellites ranged from 7 to 13 repeats and Del segregated only with (TTTA)7. 194 The two most prevalent allelic variants were (TTTA)11 (34.9%) and Del-(TTTA)7 (33.2%). 195 Del and Ins alleles occurred with incidences of 33.2 and 66.8% respectively, while genotypes 196 were distributed as follows: Ins/Ins (43.4%, n = 66), Ins/Del (46.7%, n = 71) and Del/Del 197 (9.9%, n = 15). The most common combined (TTTA)n and Ins/Del genotypes were Del-198 (TTTA)7/ (TTTA)11 (23.7%) and (TTTA)11/(TTTA)11 (11.8%). 199 The average biallelic mean of (TTTA)n variation was 8.9 ± 1.3 repeats. Linear regression 200 models were used to examine the associations between (TTTA)n biallelic means and the 201 causes of female infertility. Patients with endometriosis showed markedly shorter biallelic 202 means of (TTTA)n repeats (8.3 1.1 repeats, r = 0.88, P = 0.026) when compared with the 203 control group of women with tubal factor infertility (9.1 1.4 repeats). Female patients with 204 unexplained and male factor infertility showed (TTTA)n length means similar to the control 205 group. 206 The possible role of CYP19A1 Ins/Del variation in the aetiology of female infertility was 207 studied by applying logistic regression models. The presence of Del allele appeared as a 208 genetic risk factor for unexplained infertility (Del/Del and Ins/Del frequency of 78.6%, odds 209 ratio (OR) = 3.78, P = 0.056) when compared with the tubal factor infertility group (Del/Del 210 and Ins/Del frequency of 49.3%). Del allele showed a significant relationship (OR = 4.33, P 211 = 0.039) with unexplained infertility when the model was further corrected for early-212 follicular-phase serum FSH concentrations. Other causes of infertility were, however, 213 unrelated to Ins/Del variant. 214 215 CYP19A1 variants and ovarian stimulationIVF outcome 216 CYP19A1 (TTTA)n biallelic means showed a positive correlation (r = 0.21, P = 0.018) with 217 follicular count at days 3–5 of spontaneous menstrual cycle irrespective of patient’s age. The 218 ovarian volume and serum FSH of early follicular phase, also recommended as ovarian 219 reserve markers, were unrelated to CYP19A1 (TTTA)n polymorphic locus. During ovarian 220 stimulation, age-adjusted linear regression model revealed a negative correlation between 221 CYP19A1 (TTTA)n biallelic means and the amounts of FSH used to mature one ovarian 222 puncture follicle (r = 18.38, P = 0.039). The (TTTA)n biallelic means were 9.0 ± 1.3 and 8.8 223 ± 1.3 in the groups of women with and without clinical pregnancy respectively. Age adjusted 224 logistic regression model was not powerful enough (<80.0%) to suggest a significant 225 association between 0.2 repeats of difference in (TTTA)n biallelic mean and increased 226 chance of pregnancy (OR = 1.07, not significant). 227 Correlations between Ins/Del variation and clinical parameters influencing the outcome of 228 ovarian stimulation were assessed with linear regression models that accounted for patient’s 229 age. Women with at least one Del allele (Del/Del and Ins/ Del genotypes) possessed 230 markedly smaller ovaries (4.6 1.8 cm3, r = 0.70, P = 0.047) compared with women with 231 Ins/Ins genotype (5.3 2.4 cm3). On the contrary, Ins/Del variation did neither predict serum 232 FSH concentration nor follicle count at days 3–5 of the spontaneous menstrual cycle. 233 Associations between Ins/Del genotypes and ovarian stimulation variables are presented in 234 Table 1. Women with Ins/Del and Ins/ Ins genotypes needed lower FSH doses to mature one 235 ovarian puncture follicle (Ins/Del, r = 89.67, P = 0.024 and Ins/Ins, r = 101.27, P = 0.011) 236 and to obtain one mature oocyte (Ins/Del, r = 240.84, P = 0.004 and Ins/Ins, r = 211.61, P = 237 0.012) when compared with patients with the reference Del/Del genotype according to age-238 adjusted linear regression models. In addition, Ins/Del heterozygotes tended to yield more 239 oocytes (r = 3.04, P = 090) and mature oocytes (r = 2.58, P = 0.094), while carriers of both 240 Ins/Del and Ins/Ins genotypes required, albeit not significantly, lower doses of FSH to obtain 241 one oocyte (Ins/Del, r = 104.38, P = 0.084, and Ins/Ins, r = 105.03, P = 0.083) if contrasted 242 with the reference Del/Del patients. Contrary to the expectation, IVF patients with Ins/Ins 243 genotype showed marginally lower serum oestradiol (r = 2580.60, P = 0.062) and 244 substantially reduced serum oestradiol per follicle punctured (r = 173.90, P = 0.032) as 245 shown by linear regression models adjusted by ovarian volume. Age adjusted logistic 246 regression models did not demonstrate any relation between Ins/Del gene variants and IVF 247 pregnancy outcome: clinical pregnancy rates for patients with Ins/Ins 27.0%, Ins/Del 33.8% 248 and Del/Del 40.0% genotypes were observed. However, these statistical models were 249 insufficiently powered (<80.0%) to conclusively rule out the lack of association between 250 CYP19A1 Ins/Del and IVF pregnancy success. 251 252

253 Progress has been made towards ascertaining the genetic predictors of ovarian stimulation 254 success. The genes involved in steroid biosynthesis and hypothalamicpituitaryovarian axis 255 that possess numerous functional polymorphisms are promising targets for genetic studies. 256 Along with substantial impacts on ovarian stimulation variables, these variations also play 257 crucial roles in the pathogenesis of certain forms of female infertility. The present study is the 258 first one to show associations between CYP19A1 gene variants and the outcome of ovarian 259 stimulation- IVF in normally ovulating infertile women. Patients with shorter CYP19A1 260 (TTTA)n repeats and Del/Del homozygosity exhibit decreased ovarian FSH sensitivity during 261 ovarian stimulation, along with the greater risk for endometriosis and unexplained infertility 262 respectively. 263 The functional importance of linked intronic (TTTA)n and Ins/Del genetic markers is far 264 from completely understood. Gene introns are known to contain sequences for transcription 265 and splicing regulation, which may lead to different mRNA levels and isoforms, and result in 266 modified protein activity (Gasch et al., 1989; Carstens et al., 1998). Alternatively, 267 microsatellite variations could be in linkage disequilibrium with other functional gene 268 variants, and thus indirectly modify gene expression and protein function. TTTA 269 microsatellite in intron 4 has been reported to be in linkage disequilibrium with CT single 270 nucleotide polymorphism (SNP) rs10046 in 3-untranslated region of exon 10 (Kristensen et 271 al., 2000). Linked T-variant and long (TTTA)12 allele are associated with elevated aromatase 272 transcript levels in breast cancer tissue (Kristensen et al., 2000). 273 Two previous studies have also addressed the effect of CYP19A1 variants on ovarian 274 stimulation outcome. In both of these studies, no genetic interactions were observed between 275 CYP19A1 C/T SNP (rs10046) and FSH hormone response during ovarian stimulation (de 276 Castro et al., 2004) or the aetiology of severe ovarian hyperstimulation syndrome (Binder et 277 al., 2008). Considering the known linkage between (TTTA) n microsatellite and C/T SNP, 278 the published studies seemingly contradict the present results. However, genotyping C/T SNP 279 in exon 10 only partially predicts TTTA-repeat length in intron 4 (Kristensen et al., 2000), 280 which makes direct comparisons of study results impossible. In addition, differences in the 281 study populations and ovarian stimulation regimens may account for the discrepancies noted 282 between the clinical outcomes. Intra-cycle GnRH antagonists were used for the rapid down- 283 regulation of pituitary function in all patients, unlike the GnRH agonist long protocol utilized 284 by de Castro and colleagues (de Castro et al., 2004). 285 Ovarian stimulation outcome was shown to correlate with the follicle count observed in 286 ovaries on ultrasound scan during the preceding early follicular phase of an unstimulated 287 cycle (Gougeon, 1996). Thus, the ovarian follicular response and oocyte maturity in IVF may 288 depend on aromatase gene Ins/ Del and (TTTA)n genotypes through selective CYP19A1 gene 289 expression in small antral follicles. In line with this possibility, correlations have been 290 demonstrated between Del allele and shorter TTTA repeat sizes, with smaller ovaries 291 showing fewer antral follicles on days 3–5 of a spontaneous menstrual cycle. Both ovarian 292 size and follicle count are regarded as ovarian reserve markers. Therefore, it is unexpected 293 that CYP19A1 gene variants would not demonstrate any association with serum FSH 294 concentration, which is probably the most acknowledged marker of ovarian senescence and 295 responsiveness. 296 Multiple factors govern ovarian response, along with the most prominent negative effect of 297 increased patient age (Kligman and Rosenwaks, 2001). A previous study suggested that 298 serum anti-FSH antibodies are associated with poor ovarian response to FSH stimulation in 299 IVF, with a potential local antagonizing effect in maturing follicles (Haller et al., 2008). 300 Diminished response to FSH stimulation is also associated with decreased granulosa cell 301 aromatase activity (Hurst et al., 1992) and lower follicular fluid oestradiol concentration 302 (Bahçeci et al., 2007). However, aromatase mRNA and protein levels in granulosa cells in 303 respect to ovarian FSH-sensitivity are not known. 304 Genetic studies should provide the basis for the pharmacogenetic approach to ovarian 305 stimulation as has recently been demonstrated for patients with unfavourable FSHR genotype 306 using higher initial and total FSH doses to overcome relative ovarian insensitivity (Behre et 307 al., 2005). However, whether or not the aromatase gene variants have enough influence on 308 ovarian stimulation outcome in order to be applicable in determining FSH doses in hormonal 309 stimulation remains a challenge for future studies. 310 Serum oestradiol concentration during ovarian stimulation represents the sum of oestradiol 311 production of all growing follicles. It was found that women with the Del homozygous 312 genotype have higher values of serum oestradiol and oestradiol per follicle punctured. This 313 finding apparently contradicts decreased FSH sensitivity observed in Del/Del patients 314 compared with women carrying at least one Ins allele. However, the Del variation with 315 accompanying shorter TTTA- repeats is associated with smaller ovaries exhibiting fewer 316 antral follicles at the beginning of the natural cycle. The aim of ovarian stimulation is to 317 produce the maximum number of high quality oocytes with the utmost developmental 318 capacity to sustain fertilization, implantation and pregnancy. Ovarian stimulation is closely 319 followed with monitoring of follicular development two or three times during the stimulation. 320 This vigilance leads to the appropriate adjustments of FSH doses guided by the ultrasound 321 images of the ovaries. Considering the increased doses of FSH required per maturing follicle, 322 it is postulated that the higher serum oestradiol concentrations demonstrated in Del/Del 323 patients might be explained with the exaggerated oestradiol production caused by the 324 excessive FSH stimulation of follicles. 325 Although no genetic influences on pregnancy outcome were detected, such predictions may 326 not be meaningful, as aromatase is not expressed in the normal endometrium (Kitawaki et al., 327 1997). Alternatively, the present negative finding can also be a result of the low level of 328 statistical power to reveal minor differences between study groups with insufficient size. 329 Indeed, post-hoc analysis indicated the unsatisfactory power to conclusively prove the 330 absence of a relationship between CYP19A1 variants and IVF pregnancy success. 331 Intriguingly, in this context, aromatase inhibitors and lower concentrations of oestrogens may 332 contribute to better implantation potential by improving endometrial development, without 333 having a negative anti-oestrogenic effect on folliculogenesis (Verpoest et al., 2006). 334 Literature offers lines of evidence on the overlapping patient profiles of folliculogenesis 335 abnormalities in women with endometriosis and unexplained infertility, as extensively 336 reviewed by Cahill and Hull (2000). Furthermore, although diagnostic laparoscopy is 337 included in the routine evaluation of female infertility, endometriosis can be underestimated 338 due to the non-visible precursor stages of endometriotic lesions, ending up with misdiagnosis 339 of unexplained infertility. Although the correlations between CYP19A1 variants and the 340 occurrence of endometriosis and unexplained infertility are presented, the limited size of both 341 patient groups merits consideration of the present findings. Earlier studies have failed to 342 obtain evidence on the significance of shorter CYP19A1 TTTA-repeats in the increased risk 343 of endometriosis (Kado et al., 2002; Hur et al., 2007). However, a study of Japanese 344 endometriosis patients showed a preponderance of Del/Del genotype among these patients 345 compared with controls (Kado et al., 2002). This is relevant because Del variant is known to 346 be in strong linkage disequilibrium with short (TTTA)7 allele (Probst-Hensch et al., 1999), 347 which supports the view of low-repeat-number TTTA-alleles as susceptibility factors for 348 endometriosis. In addition, a study conducted among the Greek population suggests 349 CYP19A1 (TTTA)10 allele is consistent with the endometriosis phenotype (Arvanitis et al., 350 2003). However, this inference has been undermined in a recent meta-analysis that attributes 351 the association to chance (Guo, 2006). 352 In conclusion, future studies are clearly needed to confirm these preliminary results on the 353 importance of aromatase gene variants in the aetiology of female infertility and ovarian 354 stimulationoutcome. 355 356

357 The authors acknowledge all voluntary participants of the study. The study was supported by 358 the Estonian Science Foundation (grants nos 6498 and 6585), Estonian Ministry of Education 359 and Science (core grants nos. 0182641s04, 0180142Cs08, and PBGMR07903), EU FP6 360 Grant LSHB-CT-2004–503243, Kristjan Jaak Stipendiumid, and Swedish Institute. 361 362 363 364

365 Aljanabi SM, Martinez I 1997 Universal and rapid salt-extraction of high quality genomic 366 DNA for PCR-based techniques. Nucleic Acids Research 25, 4692–4693. 367 Altmäe S, Haller K, Peters M et al. 2007 Allelic estrogen receptor 1 (ESR1) gene variants 368 predict the outcome of ovarian stimulation in in vitro fertilization. Molecular Human 369 Reproduction 13, 521–526. 370 Arvanitis DA, Koumantakis GE, Goumenou AG et al. 2003 CYP1A1, CYP19, and GSTM1 371 polymorphisms increase the risk of endometriosis. Fertility and Sterility 79 (Suppl. 1), 702–372 709. 373 ASRM 1997 Revised American Society for Reproductive Medicine classification of 374 endometriosis: 1996. Fertility and Sterility 67, 817–821. 375 Baghaei F, Rosmond R, Westberg L et al. 2003 The CYP19 gene and associations with 376 androgens and abdominal obesity in premenopausal women. Obesity Research 11, 578–585. 377 Bahçeci M, Ulug U, Turan E et al. 2007 Comparisons of follicular levels of sex steroids, 378 gonadotropins and insulin like growth factor-1 (IGF-1) and epidermal growth factor (EGF) in 379 poor responder and normoresponder patients undergoing ovarian stimulation with GnRH 380 antagonist. European Journal of Obstetrics, Gynecology, and Reproductive Biology 130, 93–381 98. 382 Behre HM, Greb RR, Mempel A et al. 2005 Significance of a common single nucleotide 383 polymorphism in exon 10 of the follicle- stimulating hormone (FSH) receptor gene for the 384 ovarian response to FSH: a pharmacogenetic approach to controlled ovarian 385 hyperstimulation. Pharmacogenetics and Genomics 15, 451–456. 386 Binder H, Dittrich R, Hager I et al. 2008 Association of FSH receptor and CYP19A1 gene 387 variations with sterility and ovarian hyperstimulation syndrome. Reproduction 135, 107–116. 388 Cahill DJ, Hull MG 2000 Pituitaryovarian dysfunction and endometriosis. Human 389 Reproduction Update 6, 56–66. 390 Carstens RP, McKeehan WL, Garcia-Blanco MA 1998 An intronic sequence element 391 mediates both activation and repression of rat fibroblast growth factor receptor 2 pre-mRNA 392 splicing. Molecular and Cellular Biology 18, 2205–2217. 393 de Castro F, Moron FJ, Montoro L et al. 2004 Human controlled ovarian hyperstimulation 394 outcome is a polygenic trait. Pharmacogenetics 14, 285–293. 395 Gasch A, Hinz U, Renkawitz-Pohl R 1989 Intron and upstream sequences regulate expression 396 of the Drosophila beta 3-tubulin gene in the visceral and somatic musculature, respectively. 397 Proceedings of the National Academy of Sciences of the USA 86, 3215–3218. 398 Georgiou I, Konstantelli M, Syrrou M et al. 1997 Oestrogen receptor gene polymorphisms 399 and ovarian stimulation for in-vitro 400 fertilization. Human Reproduction 12, 1430–1433. 401 Gougeon A 1996 Regulation of ovarian follicular development in 402 primates: facts and hypotheses. Endocrine Reviews 17, 121–155. Guo SW 2006 Association 403 of endometriosis risk and genetic 404 polymorphisms involving sex steroid biosynthesis and their receptors: a meta-analysis. 405 Gynecologic and Obstetric Investigation 61, 90–105. 406 Haiman CA, Hankinson SE, Spiegelman D et al. 2000 A tetranucleotide repeat 407 polymorphism in CYP19 and breast cancer risk. International Journal of Cancer 87, 204–408 210. 409 Haller K, Salumets A, Uibo R 2008 Anti-FSH antibodies associate with poor outcome of 410 ovarian stimulation in in vitro fertilization. Reproductive BioMedicine Online 16, 350–355. 411 Hur SE, Lee S, Lee JY et al. 2007 Polymorphisms and haplotypes 412 of the gene encoding the estrogen-metabolizing CYP19 gene in Korean women: no 413 association with advanced-stage endometriosis. Journal of Human Genetics 52, 703–711. 414 Hurst BS, Zacur HA, Schlaff WD et al. 1992 Use of granulosaluteal cell culture to evaluate 415 low and high clinical responses to menotropin stimulation. Journal of Endocrinological 416 Investigation 15, 567–572. 417 Kado N, Kitawaki J, Obayashi H et al. 2002 Association of the CYP17 gene and CYP19 gene 418 polymorphisms with risk of endometriosis in Japanese women. Human Reproduction 17, 419 897–902. 420 Kitawaki J, Noguchi T, Amatsu T et al. 1997 Expression of aromatase cytochrome P450 421 protein and messenger ribonucleic acid in human endometriotic and adenomyotic tissues but 422 not in normal endometrium. Biology of Reproduction 57, 514–519. 423 Kligman I, Rosenwaks Z 2001 Differentiating clinical profiles: predicting good responders, 424 poor responders, and hyperresponders. Fertility and Sterility 76, 1185–1190. 425 Kristensen VN, Harada N, Yoshimura N et al. 2000 Genetic variants of CYP19 (aromatase) 426 and breast cancer risk. Oncogene 19, 1329–1333. 427 Kristensen VN, Andersen TI, Lindblom A et al. 1998 A rare CYP19 (aromatase) variant may 428 increase the risk of breast cancer. Pharmacogenetics 8, 43–48. 429 Noble LS, Simpson ER, Johns A et al. 1996 Aromatase expression in endometriosis. Journal 430 of Clinical Endocrinology and Metabolism 81, 174–179. 431 Perez Mayorga M, Gromoll J, Behre HM et al. 2000 Ovarian response to follicle-stimulating 432 hormone (FSH) stimulation depends on the FSH receptor genotype. Journal of Clinical 433 Endocrinology and Metabolism 85, 3365–3369. 434 Probst-Hensch NM, Ingles SA, Diep AT et al. 1999 Aromatase and breast cancer 435 susceptibility. Endocrine-related Cancer 6, 165–173. 436 Ryan KJ 1982 Biochemistry of aromatase: significance to female reproductive physiology. 437 Cancer Research 42, 3342s-3344s. 438 Ryan KJ, Petro Z 1966 Steroid biosynthesis by human ovarian granulosa and thecal cells. 439 Journal of Clinical Endocrinology and Metabolism 26, 46–52. 440 Sample WF, Lippe BM, Gyepes MT 1977 Gray-scale ultrasonography of the normal female 441 pelvis. Radiology 125, 477–483. 442 Sebastian S, Bulun SE 2001 A highly complex organization of the regulatory region of the 443 human CYP19 (aromatase) gene revealed by the Human Genome Project. Journal of Clinical 444 Endocrinology and Metabolism 86, 4600–4602. 445 Verpoest WM, Kolibianakis E, Papanikolaou E et al. 2006 Aromatase inhibitors in ovarian 446 stimulation for IVF/ICSI: a pilot study. Reproductive BioMedicine Online 13, 166–172. 447 Whitlock JPJ 1986 The regulation of cytochrome P-450 gene expression. Annual Review of 448 Pharmacology and Toxicology 26, 333–369. 449 Xita N, Georgiou I, Lazaros L et al. 2008 The synergistic effect of sex hormone-binding 450 globulin and aromatase genes on polycystic ovary syndrome phenotype. European Journal of 451 Endocrinology 158, 861–865. 452 Zeitoun K, Takayama K, Michael MD et al. 1999 Stimulation of aromatase P450 promoter 453 (II) activity in endometriosis and its inhibition in endometrium are regulated by competitive 454 binding of steroidogenic factor-1 and chicken ovalbumin upstream promoter transcription 455 factor to the same cis-acting element. Molecular Endocrinology (Baltimore, Md.) 13, 239–456 253. 457 458 459 460 461 462 463 464 465 466 467 468 469