Pregnancy in women with polycystic ovary syndrome: the effect of different phenotypes and features on obstetric and neonatal outcomes

Transcription

1 Pregnancy in women with polycystic ovary syndrome: the effect of different phenotypes and features on obstetric and neonatal outcomes Stefano Palomba, M.D., a Angela Falbo, M.D., a Tiziana Russo, M.D., a Achille Tolino, M.D., b Francesco Orio, M.D., c and Fulvio Zullo, M.D. a a Departments of Obstetrics & Gynecology, University Magna Graecia of Catanzaro, Catanzaro; and b University Federico II of Naples and c Department of Endocrinology, University Parthenope of Naples, Naples, Italy Objective: To test the hypothesis that the risk of adverse obstetric or neonatal outcomes varies according to different phenotypes of polycystic ovary syndrome (PCOS), and to evaluate the clinical impact of the main features of PCOS. Design: Prospective controlled clinical study. Setting: Academic Departments of Obstetrics and Gynecology, and of Endocrinology, Italy. Patient(s): Ninety-seven pregnant women with PCOS and 73 healthy pregnant subjects were recruited as cases and controls, respectively. Intervention(s): Clinical, biochemical, and ultrasonographic evaluations. Main Outcome Measure(s): Obstetric and neonatal outcomes. Result(s): The relative risk (RR) for adverse obstetric or neonatal outcomes was increased (1.7, 95% confidence interval [CI] ) in patients with PCOS and varied according to the PCOS phenotype (1.93, 95% CI ; 2.23, 95% CI ; 0.54, 95% CI , and 0.48, 95% CI for full-blown, nonpolycystic ovaries [PCO], nonhyperandrogenic, and ovulatory phenotypes, respectively). The RRs were 1.57 (95% CI ) and 0.48 (95% CI ) for oligoanovulatory and ovulatory patients with PCOS, respectively. The risk for adverse obstetric or neonatal outcomes was affected significantly by ovarian dysfunction and biochemical hyperandrogenism, whereas no significant effect was detected for clinical hyperandrogenism and PCO. Conclusion(s): The increased risk for adverse obstetric and neonatal outcomes that was observed in patients with PCOS varies widely according to the different phenotypes and features of PCOS. (Fertil Steril Ò 2010;94: Ó2010 by American Society for Reproductive Medicine.) Key Words: Neonatal outcomes, obstetric outcomes, PCOS, phenotypes, predictors, pregnancy The European Society of Human Reproduction and Embryology (ESHRE)/American Society of Reproductive Medicine (ASRM) Consensus workshop in 2003 established new criteria for diagnosing polycystic ovary syndrome (PCOS) to standardize the working definition of the syndrome (1). Specifically, the workshop specified that the presence of at least two of the following three criteria is required for a diagnosis of PCOS: [1] oligoanovulation, [2] clinical or biochemical signs of hyperandrogenism, and [3] polycystic ovaries (PCO) detected by ultrasound, after the exclusion of other pathologies with a similar clinical presentation. Thus, the syndrome includes four broad phenotypes. Recently, a meta-analysis (2) confirmed that the effects on reproductive function, which are due to PCOS, are not limited to infertility (oligoanovulation or anovulatory dysfunction), but also include pregnancy. A higher incidence of obstetric and neonatal complications has been detected in pregnant women with PCOS in comparison with controls (2). Furthermore, this meta-analysis described Received June 21, 2009; revised October 21, 2009; accepted October 26, 2009; published online December 11, S.P. has nothing to disclose. A.F. has nothing to disclose. T.R. has nothing to disclose. A.T. has nothing to disclose. F.O. has nothing to disclose. F.Z. has nothing to disclose. Reprint requests: Stefano Palomba, M.D., Department of Gynecology & Obstetrics, University Magna Graecia of Catanzaro, Via T. Campanella 182/1, Catanzaro, Italy (FAX: ; stefanopalomba@tin.it). several biases due to the heterogeneity in the diagnosis and treatment of the population studied and, at present, the potential impact of the new PCOS phenotypes on this risk is not known. The aim of the current study was to test the hypothesis that the risk for adverse obstetric or neonatal outcomes varies according to the different PCOS phenotypes, and to evaluate the clinical impact of the three main diagnostic conditions for PCOS (i.e., oligoanovulation, hyperandrogenism, and PCO). MATERIALS AND METHODS The procedures used in the present study were carried out in accordance with the Declaration of Helsinki s principles for the conducting of experiments on human subjects, and the study was approved by the Institutional Review Board (IRB) of the Department of Obstetrics and Gynecology, University Magna Graecia of Catanzaro, Italy. Between February 2003 and April 2008, 97 primigravidas who were suffering from PCOS (PCOS group) and 73 healthy primigravidas (control group) were screened and enrolled in the study. All of the women had been referred by their primary care doctors to the walk-in clinics of the Academic Departments of Obstetrics and Gynecology at our institution. All subjects, case and control groups, were studied before their pregnancy in a 5-year prestudy period as outpatients of the Academic Departments of Obstetrics and Gynecology of Catanzaro and Naples, and Department of Endocrinology of Naples (Italy). In both groups, all clinical (using a standardized clinical chart) and biochemical (obtained using different commercial kits) data were investigated and recorded. All of the enrolled patients in /$36.00 Fertility and Sterility â Vol. 94, No. 5, October doi: /j.fertnstert Copyright ª2010 American Society for Reproductive Medicine, Published by Elsevier Inc.

2 TABLE 1 Definitions used in the study for each specific complication. Outcome Definition Miscarriage Spontaneous abortion occurred during the first 12 weeks of gestation PIH SBP R140 mm Hg or DBP R90 mm Hg PE PIH with proteinuria (R300 mg/24-hour urine collection or 30 mg/dl in single urine sample) of new onset after 20 weeks of gestation GDM Recognition of two abnormal values (i.e., plasma glucose >105 mg/dl at fasting, >190 mg/dl at 1 hour, >165 mg/dl at 2 hours, and >145 mg/dl at 3 hours assessments) SGA Fetal indexes below the 10th percentile adjusted for gestational age for white population LGA Fetal indexes above the 95th percentile adjusted for gestational age for white population AGA Fetal indexes included between the 10th and the 95th percentile adjusted for gestational age for white population Operative delivery Use of forcep or vacuum extractor and the cesarean section Preterm delivery Delivery of a fetus with gestational age less than 37 week according to estimated date of delivery, based on midtrimester ultrasound scan FGR Restriction in fetal growth recognized by serial ultrasonographic fetal biometric measurements (at least two assessments) Antepartum hemorragia Vaginal bleeding similar to menstrual loss before delivery without any evidence of fetal compromise Abruptio placentae Vaginal bleeding with evidence of fetal compromise leading to an emergency delivery and evidence of retroplacental clot on postdelivery examination of the placenta Note: AGA ¼ appropriate for gestational age; DBP ¼ diastolic blood pressure; GDM ¼ gestational diabetes mellitus; FGR ¼ fetal growth restriction; LGA ¼ large for gestational age; PE ¼ preeclampsia; PIH ¼ pregnancy-induced hypertension; SBP ¼ systolic blood pressure; SGA ¼ small for gestational age. both groups became pregnant spontaneously and no pregnancy was commenced using assisted reproductive techniques (ART). In all cases, PCOS was diagnosed according to the criteria specified by ESHRE/ASRM (1). The patients with PCOS were separated into ovulatory and oligoanovulatory groups. An additional categorization into the four different PCOS phenotypes was made, distinguishing [1] patients with hyperandrogenism, oligoanovulation, and PCO (full-blown syndrome); [2] patients with hyperandrogenism and oligoanovulation, but without the appearances of PCO on ultrasound (non-pco PCOS); [3] patients with hyperandrogenism and PCO (ovulatory PCOS); and [4] patients with oligoanovulation and PCO (nonhyperandrogenic PCOS). The health of the control group patients was determined by their medical history, a physical and pelvic examination, and a complete blood chemistry panel. All patients in the control group had regular menstrual cycles (26 32 days in length), no signs of clinical hyperandrogenism, normal range of serum androgens levels, no PCO morphology on transvaginal ultrasonography (TVUS), and no known male or tubal infertility factors. For the PCOS and control groups, the exclusion criteria were as follows: age >35 years; obesity (defined as a body mass index [BMI] higher than 30); multiple pregnancies; a gestational age >7 weeks (calculated from the last menstrual period and confirmed by the ultrasonographic crown rump length measurement); premalignancies or malignancies; any major medical conditions or other concurrent medical illnesses affecting the health status; cigarette smoking; drug/alcohol use; organic pelvic disease; previous pelvic surgery; patients who were not compliant with our study protocol; and current or previous (within the past 6 months) use of any hormonal or antidiabetic drugs. Previous treatments with fertility drugs were considered to be a specific exclusion criterion to avoid any bias due to previous exposure to such medications. We also excluded women who intended to start a diet or a specific program of physical activity. At the start of the study, all subjects were being treated with folic acid (0.4 mg/day). Each subject underwent clinical, biochemical, and ultrasonographic assessments. Clinical visits were performed at study entry, and every 2 weeks during the first trimester, and, thereafter, every 4 weeks until delivery. They consisted of obstetric examination, anthropometric measurements (including height, weight, BMI, and waist-to-hip ratio), heart rate, and blood pressure assessment. Biochemical assessments consisted of liver and renal function analysis, a complete blood count, serum glucose levels, and a complete chemical and organic urine assay, as well as a complete hormonal assay (discussed later). The complete hormonal pattern was only assessed at study entry, whereas all other parameters were assessed monthly. Glucose and insulin concentrations were measured at basal levels and also after an oral glucose tolerance test (OGTT) at study entry and at the 26th week of gestation for gestational diabetes mellitus screening (3). Glucose tolerance was assessed according to the criteria of the World Health Organization (WHO) (4). The glucose and insulin response to OGTT was analyzed by calculating the areas under the curve (AUC) for glucose (AUC glucose ) and insulin (AUC insulin ) according to the trapezoidal method. The AUC glucose :AUC insulin ratio was also calculated for each subject. The baseline clinical and biochemical data were obtained before the seventh week of gestation (the entry point to the study), and all the biochemical assays were centrally performed (University of Naples Federico II ) using the same methods and commercial kits. All subjects underwent ultrasonographic examinations by two experienced gynecologist operators (one in Catanzaro and one in Naples). Transvaginal scans were performed until the 12th week of gestation. Thereafter, transabdominal scans were performed. During the first trimester, the embryonic heart beat was recorded and the crown rump length was measured. Thereafter, fetal growth, placental location and degree, amniotic fluid index, and velocimetry of the umbilical vessels (when required for high-risk pregnancy) were monitored. A careful search for fetal malformations was also performed at 20 weeks of gestation. Obstetric and neonatal outcomes were noted. In particular, miscarriage, gestational diabetes mellitus, pregnancy-induced hypertension (PIH), preeclampsia, and antepartum hemorrage/abruptio placentae, gestational age at delivery, type of delivery, fetal growth, birth weight, Apgar scores, fetal malformations, and intrauterine deaths were recorded. Table 1 shows the criteria that were used to define each specific complication. Obstetric and neonatal outcomes were categorized as either normal or pathological. A normal outcome was defined as the delivery at term of a neonate with an appropriate growth for gestational age without complications Palomba et al. Effects of PCOS phenotypes on pregnancy Vol. 94, No. 5, October 2010

3 TABLE 2 Baseline clinical and biochemical data in PCOS and control groups. PCOS (n [ 93) Controls (n [ 73) P value Age (y) 30 (6 IQR; (7 IQR; BMI (kg/m 2 ) 24.2 (2.7 IQR; (3.4 IQR; WHR 0.78 (0.17 IQR; (0.07 IQR; Ferriman-Gallwey score 9 (7 IQR; (2 IQR; HR (beats/min) 84 (16 IQR; (16 IQR; SBP (mm Hg) 122 (20 IQR; (18 IQR; DBP (mm Hg) 72 (8.5 IQR; (9.5 IQR; FSH (miu/ml) 5.2 (1.7 IQR; (2.4 IQR; LH (miu/ml) 10.2 (5.1 IQR; (6.3 IQR; E 2 (pg/ml) (19 IQR; (13.8 IQR; P (ng/ml) 18.1 (6.2 IQR; (8.0 IQR; T (ng/ml) 1.8 (1.9 IQR; (0.3 IQR; A (ng/ml) 3.8 (3.9 IQR; (0.9 IQR; DHEAS (ng/ml) 2,486.2 (1,012.6 IQR; 1, , ,712.4 (164.2 IQR; 1, , SHBG (nmol/l) 21.0 (24.2 IQR; (5.7 IQR; FAI (%) 10.3 (8.3 IQR; (2.3 IQR; Fasting glucose (mg/dl) 81.0 (11 IQR; (18 IQR; Fasting insulin (mu/ml) 11.3 (7.9 IQR; (1.3 IQR; GIR (mg/10 4 U) 4.5 (0.95 IQR; (0.8 IQR; HOMA 16.3 (8.5 IQR; (4.9 IQR; OGTT AUC glucose (mg/dl/120 min) (38 IQR; , ,011.0 (79 IQR; ,199.0 <0.001 AUC insulin (mu/ml/120 min) 6,788.0 (4,911 IQR; 1, , ,592.0 (3744 IQR; 1, , AUC glucose /AUC insulin ratio 0.2 (0.1 IQR; (0.2 IQR; Note: Data expressed as median (interquartile range [IQR];. The biochemical assays are reported in metric units. A ¼ Androstenedione; AUC ¼ area under curve; BMI ¼ body mass index; DBP ¼ diastolic blood pressure; FAI ¼ free androgen index; GIR ¼ glucose-to-insulin ratio; HOMA ¼ homeostasis model assessment; HR ¼ heart rate; OGTT ¼ oral glucose tolerance test; PCOS ¼ polycystic ovary syndrome; SBP ¼ systolic blood pressure; SHBG ¼ sex hormone-binding globulin; WHR ¼ waist-to-hip ratio. Statistical Analysis Because at the time of study design, no data were available regarding the subject of the study, no power calculation has been performed. The primary end point was a composite outcome consisting of the cumulative incidence of adverse obstetric and neonatal outcomes. The categorical variables were compared by using the Pearson c 2 test or the Fisher s exact test, as required. The normal distribution of the data for continuous variables was evaluated with the use of the Kolmogrov-Smirnov test. Thus, continuous data were expressed as median and interquartile ranges with minimum maximum values, and analyzed using the Mann-Whitney U test. To evaluate the predictive value for adverse obstetric or neonatal outcomes of different features of PCOS, data were analyzed using the general linear model univariate procedure, which provides a regression analysis and analysis of variance (ANOVA) for a single dependent variable against one or more factors or variables. The model included the presence or absence of adverse obstetric or neonatal outcomes per patient as dependent variables, considering each patient as a unit coded with 1 or 0 according to whether they had or did not have any event during the study period. Independent variables were considered as the presence or absence (yes/no; coded as 1/0) of each PCOS feature. Thereafter, the PCOS population was categorized into ovulatory and oligoanovulatory PCOS; a further categorization was made according to the phenotypic spectrum. The relative risk (RR) was calculated with a 95% confidence interval (CI) for adverse obstetric or neonatal outcomes in the presence of each PCOS phenotype and feature. Statistical significance was set at P<.05. A statistical trend was established arbitrarily for P values ranging from The Statistics Package for Social Science (SPSS , 18 Nov 2005; SPSS Inc., Chicago, IL) was used for all statistical analyses. The RR was calculated with StatDirect (StatDirect Ltd, release 2.4.3, UK). RESULTS Four women for each group missed follow-up visits and they were excluded from the final analysis. Thus, our data refer to 93 and 69 women for PCOS and control groups, respectively. In particular, patients with full-blown, non-pco, nonhyperandrogenic, and ovulatory phenotypes were, respectively, 14, 7, 5 and 67. General Data The baseline clinical and biochemical data from the PCOS and control groups are shown in Table 2. The cumulative rate of women with adverse obstetric or neonatal outcomes was significantly higher in the patients with PCOS than in the healthy controls (37/93 [39.8%] vs. 8/69 [11.6%], respectively; P¼.0001) with an RR of 1.7 (95% CI ). Data regarding adverse obstetric or neonatal outcomes in PCOS and control groups were detailed in Table 3. Fertility and Sterility â 1807

4 TABLE 3 Adverse obstetric or neonatal outcomes in overall PCOS population, in patients with PCOS categorized according to ovulatory and oligoanovulatory phenotypes, and in healthy controls. PCOS group Control group Weight gain, kg (IQR; Total 13.0 (4; ) a 8.0 (4; ) 12.5 (3.5; ) Oligoanovulatory 13 (4; ) Miscarriage, n (%) Total 23/93 (24.7) a 6/69 (8.7) 7/67 (10.4) b Oligoanovulatory 16/26 (61.5) PIH, n (%) Total 13/93 (14.0) a 3/69 (4.5) 3/67 (4.5) b Oligoanovulatory 10/26 (38.5) PE, n (%) Total 9/93 (9.6) a 1/69 (1.4) 2/67 (3.0) Oligoanovulatory 7/26 (26.9) GDM, n (%) Total 15/93 (16.1) a 4/69 (5.8) 5/67 (7.5) b Oligoanovulatory 10/26 (38.5) Antepartum hemorrhage, n (%) Total 26/93 (28.0) a 10/69 (14.5) 9/67 (13.4) b Oligoanovulatory 17/26 (65.4) SGA, n (%) Total 18/93 (19.4) a 6/69 (8.7) 7/67 (10.4) Oligoanovulatory 11/26 (42.3) LGA, n (%) Total 9/93 (9.7) a 4/69 (5.8) 5/67 (7.5) Oligoanovulatory 4/26 (15.4) AGA, n (%) Total 3/93 (46.2) a 59/69 (85.5) 58/67 (86.6) Oligoanovulatory 12/26 (46.2) Operative delivery, n (%) Total 37/93 (39.8) a 17/69 (24.6) 17/67 (25.4) b Oligoanovulatory 21/26 (80.8) Gestational age at delivery, wk (IQR; Total 39 (2; 32 41) 39 (2; 37 40) 39 (3; 36 40) Oligoanovulatory 39 (2.5; 32 41) Preterm delivery, n (%) Total 6/93 (6.5) 2/69 (2.9) 3/67 (4.5) Oligoanovulatory 3/26 (11.5) FGR, n (%) Total 9/93 (9.7) 4/69 (5.8) 5/67 (7.5) Oligoanovulatory 4/26 (15.4) Apgar score (IQR; Total 10 (1; 4 10) 10 (0.5; 5 10) 10 (0.5; 5 10) Oligoanovulatory 10 (1.5; 4 10) Fetal malformations, n (%) Total 2/93 (2.2) 0/69 (0) 2/67 (3.0) Oligoanovulatory 0/26 (0) 1808 Palomba et al. Effects of PCOS phenotypes on pregnancy Vol. 94, No. 5, October 2010

5 TABLE 3 Continued. PCOS group Control group Abruptio placentae, n (%) Total 0/93 (0) 0/69 (0) 0/67 (0) Oligoanovulatory 0/26 (0) Note: AGA ¼ appropriate for gestational age; GDM ¼ gestational diabetes mellitus; FGR ¼ fetal growth restriction; IQR ¼ interquartile range; LGA ¼ large for gestational age; PCOS ¼ polycystic ovary syndrome; PE ¼ preeclampsia; PIH ¼ pregnancy-induced hypertension; SGA ¼ small for gestational age. a P<.05 vs. controls. b P<.05 vs. oligo-ovulatory women with PCOS. Data Analysis According to PCOS Phenotypes After categorizing patients with PCOS according to their ovulatory and oligoanovulatory phenotypes, a significant difference in the cumulative rate of women with adverse obstetric or neonatal outcomes was detected (15/67 [22.4%] vs. 22/26 [84.6%], respectively; P<.0001) with an RR of 0.48 (95% CI ) and 1.57 (95% CI ), respectively. After categorizing patients with PCOS into the four ESHRE/ ASRM phenotypes, significant differences between phenotypes were also found in the distribution of women with adverse obstetric or neonatal outcomes (13/14 [92.9%] vs. 6/7 [85.7%] vs. 3/5 [60.0%] vs. 15/67 [22.4%] for the full-blown, non-pco, nonhyperandrogenic, and ovulatory phenotypes, respectively; P¼.003). The RRs for adverse obstetric or neonatal outcomes were 1.93 (95% CI ), 2.23 (95% CI ), 0.54 (95% CI ), and 0.48 (95% CI ) for the full-blown, non-pco, nonhyperandrogenic, and ovulatory phenotypes, respectively. Tables 3 and 4 summarize the specific complications noted in the PCOS population subgroups according to the different phenotypes. Data Analysis According to PCOS Features The distribution of PCOS features in our total population was as follows: 27 of 93 (29.0%) patients had oligoamenorrhea; 65 of 93 (69.9%) and 72 of 93 (77.4%) patients had clinical and biochemical hyperandrogenism, respectively; and 90 of 93 (96.8%) had PCO. TABLE 4 Adverse obstetric or neonatal outcomes in patients with PCOS categorized according to different ESHRE/ASRM phenotypes. Full-blown (n [ 14) Non-PCO (n [ 7) Nonhyperandrogenic (n [ 5) (n [ 67) Weight gain, kg (IQR; 13 (4; ) 13 (3; ) 12.5 (3; ) 12.5 (3.5; ) Miscarriage, n (%) 9/14 (64.3) a 4/7 (57.1) a 3/5 (60.0) a 7/67 (10.4) PIH, n (%) 5/14 (35.7) a 3/7 (42.9) a 2/5 (40.0) a 3/67 (4.5) PE, n (%) 4/14 (28.6) a 2/7 (28.6) a 1/5 (20.0) 2/67 (3.0) GDM, n (%) 5/14 (35.7) a 3/7 (42.9) a 2/5 (40.0) a 5/67 (7.5) Antepartum hemorrhage, n (%) 9/14 (64.3) a 5/7 (71.4) a 3/5 (60.0) a 9/67 (13.4) SGA, n (%) 6/14 (42.9) a 4/7 (57.1) a 1/5 (20.0) 7/67 (10.4) LGA, n (%) 3/14 (21.4) 1/7 (14.3) 0/5 (0) 5/67 (7.5) AGA, n (%) 6/14 (42.9) a,b 2/7 (28.6) a,b 4/5 (80.0) 58/67 (86.6) Operative delivery, n (%) 12/14 (85.7) a 5/7 (71.4) a 4/5 (60.0) a 17/67 (25.4) Gestational age at delivery, 38 (4; 32 41) 39 (2; 33 40) 39 (2; 37 39) 39 (3; 36 40) wk (IQR; Preterm delivery, n (%) 2/14 (14.3) 1/7 (14.3) 0/5 (0) 3/67 (4.5) FGR, n (%) 2/14 (14.3) 1/7 (14.3) 1/5 (20.0) 5/67 (7.5) Apgar score (IQR; 10 (1; 4 10) 10 (0.5; 5 10) 10 (1; 6 10) 10 (0.5; 5 10) Fetal malformations, n (%) 0/14 (0) 0/7 (0) 0/5 (0) 2/67 (3.0) Abruptio placentae, n (%) 0/14 (0) 0/7 (0) 0/5 (0) 0/67 (0) Note: AGA ¼ appropriate for gestational age; ASRM ¼ American Society of Reproductive Medicine; ESHRE ¼ European Society of Human Reproduction and Embryology; GDM ¼ gestational diabetes mellitus; FGR ¼ fetal growth restriction; LGA ¼ large for gestational age; PCO ¼ polycystic ovaries; PCOS ¼ polycystic ovary syndrome; PE ¼ preeclampsia; PIH ¼ pregnancy-induced hypertension; SGA ¼ small for gestational age. a P<.05 vs. ovulatory phenotype. b P<.05 vs. nonhyperandrogenic phenotype. Fertility and Sterility â 1809

6 The general linear model univariate procedure revealed a significant (P<.05) effect on the obstetric or neonatal outcomes if the patient had oligoamenorrhea and biochemical hyperandrogenism, whereas no significant effect was detected if clinical hyperandrogenism and PCO were present. The RRs for adverse obstetric or neonatal outcomes for each PCOS feature was 5.73 (95% CI ) for oligoamenorrhea, 1.07 (95% CI ) for clinical hyperandrogenism, 4.04 (95% CI ) for biochemical hyperandrogenism, and 1.72 (95% CI ) for PCO. DISCUSSION Type I evidence (2, 5) has reported an increased incidence of obstetric and neonatal complications in pregnant patients with PCOS, but no data have been published that evaluate the significance of each PCOS condition on pregnancy and neonatal outcomes, either alone or in combination with other phenotypes. This is therefore the first study that has this aim. Ninety-seven pregnant women affected by PCOS and 73 healthy control subjects were enrolled and followed throughout pregnancy until the delivery. Restrictive criteria were used to exclude the possibility of selection bias and consequent confounding factors; it limited patient recruitment and prolonged the period of enrolment to 5 years. First, we confirmed that pregnant women with PCOS had a higher incidence of obstetric or neonatal complications than healthy subjects (2). A higher prevalence of gestational diabetes mellitus was also observed, even if this risk was recently questioned in a systematic review of 16 studies, which included 2,263 women with PCOS and 92,933 controls (5). Second, by categorizing our population according to the four PCOS phenotypes that could be identified, using the criteria for PCOS specified by ESHRE/ASRM, we found a higher RR for adverse outcomes in patients with the full-blown and non-pco phenotypes than in those with the nonhyperandrogenic and ovulatory phenotypes. Analysis of previous data (6, 7) yielded the results that PCOS phenotypes had different hormonal and metabolic patterns. In particular, with regard to hyperandrogenic patients with PCOS, a fullblown phenotype was associated with impaired insulin sensitivity in contrast with the metabolic findings in the ovulatory phenotype, even if a selection bias was present as a result of a significant difference in BMI between phenotypes (8). Conversely, a lower rate of metabolic disruption has been reported in the nonhyperandrogenic PCOS phenotype (6, 7). Although these results were obtained from studies that were not powered to demonstrate a difference among phenotypes, the opinion of the investigators was that the nonhyperandrogenic PCOS phenotype has an intermediate or milder metabolic risk profile compared with other PCOS phenotypes (6, 7). These last findings suggest that hyperandrogenism plays a key role in metabolic changes, and these results could be explained by well-recognized pathophysiological mechanisms (9 11). In fact, it is well-known that insulin stimulates ovarian androgen production directly in vivo (10) and in vitro (11). In addition, the study on familial clustering of women with PCOS yields results that seem to indicate that hyperandrogenism is a genetic trait that underlies insulin resistance and its long-term risks (9). Furthermore, contrary to the current results, previous data indicated that ovulatory PCOS has metabolic risks that are similar to those of a full-blown phenotype or a non-pco phenotype (6). We carried out additional analyses with the general linear model univariate procedure to define the role of each PCOS feature in the development of pregnancy complications. Our findings seem to demonstrate that ovarian dysfunction and biochemical hyperandrogenism affected obstetric or neonatal outcomes, increasing significantly (more than fourfold) the risk for obstetric or neonatal complication. Conversely, no significant effect was detected for clinical hyperandrogenism and PCO, which increased only slightly the risk for adverse outcomes. Ovarian dysfunction and biochemical hyperandrogenism seem to be related to hormonal and metabolic disruptions (9 11), which are probably linked not only to the long-term cardiovascular risks, as has already been demonstrated, but also to obstetric complications. On the other hand, there is some debate about the metabolic significance of PCO (12 14), whereas clinical hyperandrogenism seems to be related to the receptivity of terminal hair in sexual areas of the skin more than serum levels of androgens. A weakness of the current study was the small sample size for some of the phenotypes or features, which probably biased the observed RR for the nonhyperandrogenic phenotype. On the other hand, our results were based on a carefully selected population of patients with PCOS who had no known risk factors for a complicated pregnancy, and, in terms of reality-based medicine, it is possible to suspect that an increased risk for adverse obstetric or neonatal outcomes would be even higher in patients with PCOS who had other identified risk factors. In conclusion, the current study confirms that PCOS is a composite risk factor for a complicated pregnancy. The increased risk for adverse obstetric or neonatal outcomes, which are observed in women with PCOS, varies widely according to their different PCOS phenotypes and features. Further studies on a bigger sample population are needed to confirm the current findings, particularly with respect to the rarer phenotypes. REFERENCES 1. Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome. Fertil Steril 2004;81: Boomsma CM, Eijkemans MJ, Hughes EG, Visser GH, Fauser BC, Macklon NS. A meta-analysis of pregnancy outcomes in women with polycystic ovary syndrome. Hum Reprod Update 2006;12: American Diabetes Association. Gestational diabetes mellitus. Diabetes Care 2000;23:S Modan M. Evaluation of WHO and NDDG criteria for impaired glucose tolerance. Results from two national samples. Diabetes 1989;38: Toulis KA, Goulis DG, Kolibianakis EM, Venetis CA, Tarlatzis BC, Papadimas I. Risk of gestational diabetes mellitus in women with polycystic ovary syndrome: a systematic review and a metaanalysis. Fertil Steril 2009;92: Shroff R, Syrop CH, Davis W, Van Voorhis BJ, Dokras A. Risk of metabolic complications in the new PCOS phenotypes based on the Rotterdam criteria. Fertil Steril 2007;88: Chang WY, Knochenhauer ES, Bartolucci AA, Azziz R. Phenotypic spectrum of polycystic ovary syndrome: clinical and biochemical characterization of the three major clinical subgroups. Fertil Steril 2005;83: Carmina E, Chu MC, Longo RA, Rini GB, Lobo RA. Phenotypic variation in hyperandrogenic women influences the findings of abnormal metabolic and cardiovascular risk parameters. J Clin Endocrinol Metab 2005;90: Legro RS, Bentley-Lewis R, Driscoll D, Wang SC, Dunaif A. Insulin resistance in the sisters of women 1810 Palomba et al. Effects of PCOS phenotypes on pregnancy Vol. 94, No. 5, October 2010

7 with polycystic ovary syndrome: association with hyperandrogenemia rather than menstrual irregularity. J Clin Endocrinol Metab 2002;87: Nestler JE, Powers LP, Matt DW, Steingold KA, Plymate SR, Rittmaster RS, et al. A direct effect of hyperinsulinemia on serum sex hormone-binding globulin levels in obese women with the polycystic ovary syndrome. J Clin Endocrinol Metab 1991;72: Barbieri RL, Makris A, Ryan KJ. Insulin stimulates androgen accumulation in incubations of human ovarian stroma and theca. Obstet Gynecol 1984;64: 73S 80S. 12. Murphy MK, Hall JE, Adams JM, Lee H, Welt CK. Polycystic ovarian morphology in normal women does not predict the development of polycystic ovary syndrome. J Clin Endocrinol Metab 2006;91: Palomba S, Falbo A, Russo T, Manguso F, Tolino A, Zullo F, et al. Insulinsensitivityafter metforminsuspension in normal-weight women with polycystic ovary syndrome. J Clin Endocrinol Metab 2007;92: Palomba S, Falbo A, Orio F, Tolino A, Zullo F. Efficacy predictors for metformin and clomiphene citrate treatment in anovulatory infertile patients with polycystic ovary syndrome. Fertil Steril 2009;91: Fertility and Sterility â 1811