|Home | About | Journals | Submit | Contact Us | Français|
HAB conceived the study, performed medical charts abstraction, data analysis and manuscript write-up. LS and SO provided HAB data access and revised the paper for intellectual content. MAB assisted in statistical analyses. HR and YZ were involved in the initial conception and approval of the study.
The purpose of the study is to examine the effects of polycystic ovary syndrome (PCOS) and body mass index (BMI) on selected indicators of in-vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI) treatment success. A retrospective cohort study was conducted using existing data on 69 IVF/ICSI treatment cycles undergone by PCOS women and an individually matched sample of 69 IVF/ICSI treatment cycles undergone by non-PCOS women at a major fertility treatment center. BMI (kg/m2) was analyzed as a continuous and categorical (< 25, 25−29.9 ≥ 30) variable. Results indicated that PCOS was directly associated with the number of oocytes retrieved. Irrespective of PCOS status, continuous BMI was inversely associated with total and mature oocytes retrieved. Multiple linear regression analyses indicated no significant effects of PCOS or continuous BMI on the number of mature oocytes fertilized per mature oocyte retrieved or inseminated. Similarly, multiple logistic regression analyses suggested no significant effect of PCOS and continuous BMI on the odds of pregnancy, miscarriage or live birth. Furthermore, categorical BMI did not influence process and outcome measures of IVF/ICSI treatment success. PCOS and continuous BMI appear to have significant and distinct effects on early stages but not on later stages of IVF/ICSI treatment.
Polycystic ovary syndrome (PCOS) is an endocrine disorder affecting 5−10% of women in their childbearing age (Dunaif 1997, Lord et al 2003, Sheehan 2004). A marker of the “metabolic syndrome”, PCOS has been linked to obesity, type 2 diabetes, dyslipidemia, hypertension and cardiovascular disease (Azziz 2006, Barber et al 2007, Park et al 2007, Soares et al 2007, Weerakiet et al 2007).
PCOS cases can be easily identified among women seeking fertility treatment (Al-Azemi et al 2004). Based on the Centers for Disease Control and Prevention “2001 Assisted Reproduction Technology Success Rates”, nearly 6% of women receiving various fertility treatments are diagnosed with ovulatory dysfunction (6%) (Lebovic et al 2005). Furthermore, in women attending infertility clinics, PCOS accounts for nearly 73% of patients suffering from anovulatory infertility (Al-Azemi et al 2004).
To address ovulatory dysfunction in the context of PCOS, several treatment modalities have been developed over the years. These include lifestyle modification and weight loss, clomiphene citrate, metformin, thiazolidinediones, aromatase inhibitors, gonadotrophin therapy, GnRH agonists, GnRH antagonists, laparoscopic ovarian drilling and in-vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI) (ESHRE/ASRM 2008, Homburg 2003, Sartor and Dickey 2005, Sheehan 2004).
PCOS women undergoing fertility treatment are at increased risk of multiple pregnancies and other adverse outcomes of pregnancy due to ovarian hyperstimulation (Artini et al 2006, Griesinger et al 2006, Zhu et al 2005). Accordingly, after less invasive methods have failed, IVF and ICSI are considered as ideal treatment options aimed at restoring fertility while reducing the chance of twins, triplets and higher-order pregnancies.
Since 1978, over one million infants have been born to infertile couples through assisted reproductive technologies (ART) (Klonoff-Cohen 2005). IVF/ICSI is a special type of ART whereby both oocytes and sperm are handled and fertilization occurs outside of the reproductive system. Several steps are taken in an IVF/ICSI treatment cycle. These include ovarian stimuation, oocyte retrieval, fertilization in a liquid medium, embryo selection and embryo transfer into a uterine environment (Knochenhauer et al 1998). Therefore, indicators of IVF/ICSI treatment success include process (number and quality of oocytes retrieved, inseminated and fertilized) as well as outcome (pregnancy, miscarriage and live birth rates) measures (Abdalla and Thum 2006).
It has been established that 35%−65% of PCOS patients are obese (Al-Azemi et al 2004). Current evidence suggests that, regardless of PCOS, pregnancy achievement and maintenance is adversely affected by obesity, overweight or elevated body mass index (BMI). In particular, obese women may have a lower chance of becoming pregnant and a higher chance of miscarriage after fertility treatment (Azziz 2006, Ehrmann et al 2006, Fedorcsak et al 2001, Park et al 2007, Soares et al 2007, Vignesh and Mohan 2007, Wang et al 2001, Weerakiet et al 2007).
In summary, obesity, overweight and elevated BMI are clinical features of PCOS that may constitute a major impediment for successful fertility treatment in women diagnosed with PCOS. However, few studies have examined the relationship between PCOS, BMI and IVF/ICSI treatment success (Al-Azemi et al 2004, Feorcsak et al 2001, Wang et al 2001). The few studies that have been published so far have yielded conflicting results. Many of these studies were focused on outcome rather than process measures and none had evaluated the separate and joint effects of PCOS and BMI on IVF/ICSI treatment success. The purpose of this study is to examine the net effects of PCOS and BMI on IVF/ICSI treatment success, before and after adjusting for confounders. We hypothesize that, regardless of PCOS status, BMI will be inversely associated with various process and outcome measures of IVF/ICSI treatment success.
A retrospective cohort study was conducted using existing records on a consecutive sample of PCOS women (20−49 years) who received IVF/ICSI treatment at a major fertility treatment center, between January 1st, 2000 and December 31st, 2006 and an age- and period-matched sample of non-PCOS women. All study procedures were approved by an Institutional Review Board.
PCOS women, 18 to 49 years of age, were considered as eligible for entry into the study if they had visited the JI at least once, were clinically diagnosed by physicians as having PCOS, and had undergone at least one IVF/ICSI cycle between January 1st, 2000 and December 31st, 2006. A non-random 1:1 caliper-matched sample was selected among non-PCOS women who had undergone an IVF/ICSI cycle within the same time period at the fertility treatment center. Matching criteria were age (± 1 year) and date at initiation of the IVF/ICSI cycle. Matched pairs were excluded from the analysis if: either of the two patients was below 18 years of age or above 49 years of age; either of the two patients had undergone an experimental type of IVF/ICSI treatment, including PGD and testicular sperm aspiration (TESA); or the specific type of IVF/ICSI treatment was unknown for either of the two patients. Out of 92 PCOS women, 1 had undergone PGD, 1 had undergone TESA and 6 had unknown type of IVF/ICSI treatment. Out of 92 non-PCOS women, 1 was above 49 years of age, 1 had unknown type of IVF/ICSI treatment, 4 had undergone PGD and 11 had undergone TESA. Therefore, the current analyses comprised 69 PCOS and 69 non-PCOS patients.
In the current study, PCOS was defined as a dichotomous (‘yes’ or ‘no’) variable, using data from a computerized database. Physicians at the infertility center diagnosed PCOS women based on the 1990 NIH criteria (Sartor and Dickey 2005). Differential diagnoses included congenital adrenal hyperplasia, androgen-secreting tumors and Cushing syndrome. For matched non-PCOS patients, primary diagnoses were distributed as follows: ‘Male factor’ (52.2%), ‘Endometriosis’ (18.5%), ‘Cervical’ (13.0%), ‘Idiopathic’ (8.7%), ‘Genetic’ (3.3%), ‘Tubal factor’ (1.1%) and ‘Unknown’ (1.1%).
Patient weight and height data at IVF/ICSI initiation were available in non-computerized medical records. Therefore, one of the investigators abstracted weight and height data on the selected patients from those records. Both measures (weight and height) were taken at the time of oocyte retrieval. BMI (continuous) was calculated as the ratio of weight (kg) to height squared (m2). BMI (categorical) was defined as less than 25 (normal weight), 25−29.9 (overweight) and 30+ (obese) kg/m2.
Process and outcome measures of IVF/ICSI treatment success were obtained from a computerized database. Process measures that reflect obstacles to successful IVF/ICSI treatment were defined using the following indicators: Number of oocytes retrieved; Number of mature oocytes retrieved; Number of immature oocytes retrieved; Ratio of mature oocytes inseminated to mature oocytes retrieved; Ratio of mature oocytes fertilized to mature oocytes inseminated; Ratio of mature oocytes fertilized to mature oocytes retrieved. Outcome measures that reflect IVF/ICSI treatment effectiveness were defined as pregnancy, miscarriage and live birth status.
Although approximately 33% of women undergoing IVF/ICSI achieve a pregnancy, IVF/ICSI treatment success is often multifactorial. In addition to medical expertise, pregnancy achievement depends on “ovarian” age and other age-related health characteristics . In the current study, patients were matched on “chronological” age as a proxy measure for “ovarian” age. In the context of IVF/ICSI treatment, a widely recognized indicator of poor prognosis is having an elevated basal FSH level (≥ 10 IU/l) before ovulation induction. A persistently high FSH level in pre-menopausal women is suggestive of a high “ovarian” age, poor ovarian function and potential ovarian failure leading to early onset of menopause (Abdalla and Thum 2006, Beck-Peccoz and Persani 2006, Buffet and Bouchard 2001, Kinney et al 2007). Accordingly, age at initiation of IVF/ICSI treatment and basal (day 3) FSH level were considered as a priori potential confounders for the hypothesized relationships.
Descriptive, bivariate and multivariate analyses were performed using SAS version 9.2. Bivariate associations were tested using correlation coefficients, Student t-test, Chi-square and Fisher's exact tests, where appropriate. For continuous dependent variables, simple and multiple linear regression analyses were performed. For binary dependent variables, crude and adjusted odds ratios (OR) (and 95% confidence intervals (CI)) were estimated using logistic regression. As part of regression modeling, an interaction term combining the effects of PCOS and BMI was evaluated. BMI was included as either a continuous or a categorical variable. Adjusted models included age and day 3 FSH level, as a priori confounders. With the assumption of missing-at-random (MAR) being met, the SAS MI procedure was used to impute missing data. Locally weighted regression (LOWESS) was used as a non-parametric smoothing technique to display relationships between continuous variables with a bandwidth of 0.5 (Schimek and Schimek, 2000).
A total of 69 PCOS and 69 non-PCOS patients were included in the study sample. As a result of matching, PCOS and non-PCOS patients did not differ significantly by age (32.3 ± 4.1 versus 32.5 ± 4.1, P=0.78). Nearly 26% of PCOS women and 22% of non-PCOS women were 35 years of age or more. PCOS patients had a significantly lower basal FSH level (5.4 ± 1.9 IU/l versus 6.3 ± 2.2 IU/l, P=0.0001) and a significantly higher BMI (30.7 ± 8.9 kg/m2 versus 24.7 ± 5.4 kg/m2, P < 0.0001) compared to non-PCOS patients. On the other hand, continuous BMI was positively correlated with age in PCOS patients (r=0.26, P=0.03), but not in non-PCOS patients (r=−0.05, P=0.71). Similarly, continuous BMI was significantly higher in the older age group (age ≥ 35 years: 35.9 ± 9.8 versus age < 35 years: 28.9 ± 7.8, P=0.003) of PCOS women, but not of non-PCOS women (age ≥ 35 years: 24.8 ± 4.3 versus age < 35 years: 24.6 ± 5.4, P=0.92). Finally, continuous BMI was negatively correlated with basal FSH level in PCOS (r=−0.39, P=0.0008) patients, but not in non-PCOS patients (r=−0.6, P=0.63).
Table 1 presents the associations of categorical BMI with age and basal FSH level among PCOS and non-PCOS patients. For PCOS patients, a significantly higher proportion of women aged 35 years or older was found among those with a BMI ≥ 30 kg/m2 versus those with a BMI < 25 kg/m2 (38.2% versus 13.6%, P=0.05). Furthermore, PCOS women with a BMI ≥ 30 kg/m2 had a significantly (P = 0.006) lower FSH level (4.7 ± 1.8 IU/l) compared to PCOS women with a BMI < 25 kg/m2 (6.0 ± 2.2 IU/l). These relationships were not statistically significant among non-PCOS women.
Table 2 evaluates the separate and joint effects of PCOS and BMI on process measures of IVF/ICSI treatment success. As shown in Model I, PCOS was directly related to total number of oocytes retrieved. In Model II, the positive relationship between PCOS and total number of oocytes retrieved remained significant and that of PCOS with number of mature oocytes became significant, after adjusting for age and basal FSH level. While BMI was included as a continuous variable in Models III and IV, dummy variables for overweight (BMI: 25−29.9 kg/m2) and obesity (BMI: 30+ kg/m2) were included in Models V and VI. After controlling for PCOS, continuous BMI was inversely associated with total number of oocytes and number of mature oocytes retrieved (Model III). With few exceptions, adjustment for age and basal FSH level did not affect the relationships of PCOS and continuous BMI with various IVF/ICSI process indicators (Model IV). A statistically significant interaction term (P < 0.05) was found between PCOS and continuous BMI when ratio of mature oocytes inseminated to mature oocytes retrieved was considered (Models III and IV). Findings from Model III were corroborated by LOWESS curves. In particular, the negative relationship between continuous BMI and number of oocytes retrieved was not influenced by PCOS. Also, the effect of continuous BMI on ratios involving mature oocytes retrieved, mature oocytes inseminated and mature oocytes fertilized appear to be non-linear and highly dependent on PCOS (Figures 1 and and2).2). As shown in Models V and VI, categorical BMI had no significant effect on process measures of IVF/ICSI success.
Out of 69 PCOS women undergoing IVF/ICSI treatment, 28 (40.6%) achieved a pregnancy, 7 (10.1%) had a miscarriage and 15 (21.7%) delivered a live born infant. Furthermore, out of 69 non-PCOS women undergoing IVF/ICSI treatment, 23 (33.3%) achieved a pregnancy, 6 (8.7%) had a miscarriage and 13 (18.8%) delivered a live born infant. Table 3 evaluates the separate and joint effects of PCOS and BMI on various outcome measures of IVF/ICSI treatment success. PCOS did not influence the odds of pregnancy, miscarriage or live birth (Models I and II). While BMI was included as a continuous variable in Models III and IV, dummy variables for overweight (BMI: 25−29.9 kg/m2) and obesity (BMI: 30+ kg/m2) were included in Models V and VI. Regardless of PCOS status, continuous BMI, overweight and obesity did not have a significant effect on outcome measures of IVF/ICSI treatment success, either before or after adjustment for a priori confounders (Models III-VI).
The current study assessed the separate and joint effects of PCOS and BMI on selected process and outcome measures of IVF/ICSI treatment success. As expected, PCOS was positively associated with total oocytes retrieved. Irrespective of PCOS status, continuous BMI was inversely associated with total and mature oocytes retrieved. The finding that BMI is associated with a reduced number, while PCOS is associated with an increased number of oocytes retrieved after an IVF/ICSI cycle is biologically plausible. In fact, PCOS and BMI have distinct effects on ovarian function. In recent years, PCOS women were found to have an increased level of the Anti-Müllerian Hormone leading to a greater number of oocytes retrieved upon ovarian stimulation (Fanchin et al 2007, Nardo et al 2007, Wunder et al 2008). Whereas ovarian hyperstimulation (Artini et al 2006, Griesinger et al 2006, Owj et al 2005) is characteristic of PCOS patients, obesity and overweight are associated with multiple metabolic changes that can lead to anovulation. These changes include peripheral aromatization of estrogens, decreased levels of sex hormone-binding globulin resulting in high levels of free estradiol and testosterone, and increased insulin levels that can enhance ovarian production of androgens (Al-Azemi et al 2004, Speroff et al 1999). Alternatively, the reduced number of oocytes retrieved with increasing BMI may be a technical issue, whereby retrieval of oocytes becomes more difficult among overweight and obese patients undergoing IVF/ICSI.
When controlling for PCOS and a priori confounders, overweight and obesity did not influence process or outcome measures of IVF/ICSI success. The latter finding highlights the importance of assessing the full BMI range as a predictor of fertility treatment success. Furthermore, the effect of continuous BMI on the number of mature oocytes inseminated per mature oocyte retrieved was modified by PCOS. Whereas continuous BMI appears to be positively associated with the ratio of mature oocytes inseminated to mature oocytes retrieved among non-PCOS women, this relationship was less clear-cut in the context of PCOS. In clinical practice, the number of mature follicles often does not correspond with the number of truly mature oocytes to be inseminated. Our results suggest that, in the absence of PCOS, the higher the BMI the greater the chance that a mature follicle will yield oocytes that are mature enough to be inseminated. Although PCOS and continuous BMI appear to influence some of the process measures of IVF/ICSI success, they did not have a significant effect on the odds of pregnancy, miscarriage or live birth.
Due to its inverse relationship with the total number of oocytes retrieved, BMI may play a key role in determining ART success among PCOS and non-PCOS women. According to Pasquali et al., the role of PCOS in ovulation and fertility is largely dependent on obesity, insulin-resistance and the “metabolic syndrome”. In particular, obesity can influence the pathophysiology and clinical manifestation of PCOS leading to hyperandrogenism as well as changes in granulosa cell function and follicle development (Pasquali et al, 2006).
Empirical evidence linking PCOS and BMI to process and outcome measures of fertility treatment success is limited. In a recent retrospective study by Matalliotakis et al., 140 women with a BMI ≤ 24 kg/m2 undergoing 291 cycles were compared on various IVF/ICSI outcomes with 138 women with a BMI >24 kg/m2 undergoing 291 cycles Patients with a BMI > 24 kg/m2 demonstrated a decrease in the number of follicles after stimulation, an increase in total dose of gonadotrophin used and a lower number of eggs collected. However, BMI did not affect clinical pregnancy, miscarriage or delivery rates (Matalliotakis et al, 2008). Al-Azemi et al. investigated the effect of obesity on treatment outcome among 270 PCOS patients attending a fertility clinic. Patients were stratified according to BMI (normal-weight: 18−24; overweight: 25−29, obese: 30−34; grossly obese: ≥ 35). Obesity adversely affected the outcome of ovulation induction with clomiphene citrate and gonadotrophins. Pregnancy rate and outcome were also adversely affected by obesity (Al-Azemi et al 2004). In a large cohort study of 1018 infertile women undergoing ART, the effect of PCOS on the risk of miscarriage was examined, after adjusting for BMI and other potential confounders. Univariate analysis showed that women with PCOS had a significantly greater risk of miscarriage as compared to non-PCOS women. However, multivariate analyses suggested that these differences were mainly explained by BMI and ART type received (Wang et al 2001). Fedorcsak et al. examined the impact of obesity and insulin resistance on the outcome of IVF/ICSI in women with PCOS. Insulin-resistant (n=26) and non-insulin-resistant (n=30) patients had similar number of oocytes collected and pregnancy rates. Obesity, independent of hyperinsulinaemia, was found to be related to a lower oocyte count and an increased gonadotropin requirement (Fedorcsak et al 2001).
Existing studies points to the importance of BMI as a prognostic factor for various fertility treatments. The current study is among the first to simultaneously explore the effect of PCOS and BMI on IVF/ICSI treatment success. Our findings highlight the importance of evaluating the full BMI range in relation to IVF/ICSI success. We found opposite effects of PCOS and BMI on indicators of ovulatory function, which is in line with current knowledge. Unlike previously conducted studies, we did not observe a significant effect of PCOS or BMI on final ART treatment outcomes. Nevertheless, our results should be interpreted in light of several limitations. First, the retrospective study design may have limited our ability to obtain accurate data on other potentially confounding factors such as pre-existing health conditions, perimenopausal status, total dose of gonadotropin used, smoking, alcohol and coffee consumption. Second, the small sample size may have precluded our ability to detect an effect of obesity on IVF/ICSI treatment success or an effect of BMI on pregnancy, miscarriage and live birth status. PCOS and BMI appear to have significant and distinct effects on early stages but not on later stages of IVF/ICSI treatment. Large prospective cohort studies are needed to confirm our study findings.
We would like to thank the clinical data manager (Ms. Debi Jones) and the graduate research assistant (Dr. Bethrand Ugwu) at Eastern Virginia Medical School for their technical assistance.
The authors have no affiliation with any organization with a direct or indirect financial or commercial interest in the subject matter discussed in the manuscript that may affect reporting of the work submitted.