Introduction
Rapid demographic changes caused by a declining birth rate and aging population have become major social concerns in South Korea [1,2]. Assisted reproductive technology (ART) has emerged as an essential treatment modality for women with infertility [3]. The use of ART has continued to increase in South Korea and ART-related clinical outcomes remain important indicators of reproductive healthcare [3]. The percentage of births resulting from infertility treatment has steadily increased and currently accounts for approximately 10% of all live births nationwide [2]. ART, including in vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI), has increasingly improved pregnancy outcomes for couples with infertility [1,3]. Despite advances in laboratory techniques and controlled ovarian stimulation protocols, substantial variability in clinical outcomes persists after ART, highlighting the need to elucidate the key clinical determinants of success [1,3].
Maternal age [4], anti-Müllerian hormone (AMH) level [5-7], and body mass index (BMI) have been widely used as clinical indicators of reproductive potential [8,9]. Among these, the AMH level is considered a reliable marker of ovarian reserve and is strongly associated with ovarian response during controlled ovarian stimulation [5-7]. A higher AMH level has consistently been associated with an increased number of retrieved oocytes [7,10]. However, whether this quantitative advantage translates into improved fertilization rates or clinical pregnancy outcomes remains controversial [11,12]. An elevated AMH level is frequently detected in women with polycystic ovarian morphology or polycystic ovary syndrome (PCOS) [13,14], which is characterized by an increased number of follicles but potentially impaired oocyte quality. In such cases, a higher oocyte yield does not necessarily result in improved fertilization rate or optimal embryo development.
Maternal age is another well-established determinant of ART outcome. Increasing age is associated with diminished oocyte quality, chromosomal instability, and reduced implantation potential. Female fertility declines gradually beginning in the early 30s and sharply after 35 years of age, a threshold widely recognized in clinical practice [4]. BMI has also been implicated in reproductive physiology by influencing ovarian steroidogenesis, endometrial receptivity, and systemic metabolic function [8]. Nevertheless, the individual and combined effects of AMH level, age, and BMI on ovarian response and fertilization outcomes remain unclear despite previous studies [5-9].
Overall, AMH may have dual effects on fertility outcomes: a beneficial quantitative effect on oocyte yield and a potential qualitative limitation on fertilization efficiency [7,10-12]. Therefore, interpreting ovarian reserve solely as a numerical indicator may be clinically insufficient [5,6]. A multivariable approach that simultaneously considers oocyte quantity and fertilization competence is warranted [11,12]. This study aimed to comprehensively evaluate the association of AMH levels, age, and BMI with ovarian response, normal fertilized embryo count, and clinical pregnancy outcomes in women undergoing ART. The study focused on clarifying the dual effect of AMH levels on oocyte yield and fertilization potential using multivariable regression analysis.
Materials and methods
1. Study cohort
This retrospective cohort study included patients undergoing ART cycles performed between 2018 and 2023 at a tertiary fertility center in the Republic of Korea. Initially, data from 6,312 IVF or ICSI cycles were analyzed. Cycles with missing key clinical variables (AMH level or BMI), non-ART cycles, and biologically implausible BMI values (<18 kg/m2 or ≥45 kg/m2) were excluded. Additionally, data from cycles with zero retrieved oocytes were excluded from the fertilization outcome analyses. The dataset used in this study consisted of fully deidentified and preprocessed clinical records. Owing to the anonymization process required for data privacy and research ethics, unique patient identifiers were not available. Therefore, each cycle was considered an independent clinical event. After applying these rigorous criteria, 4,399 cycles were included in the final analysis (Fig. 1).
2. Clinical variables and definitions
Clinical and laboratory data were extracted from electronic medical records and institutional ART databases. The following variables were selected a priori based on their clinical relevance: maternal age, BMI, serum AMH level, fertilization method (IVF or ICSI), number of retrieved oocytes, number of ormally fertilized embryos (two pronuclei [2PN]), and clinical pregnancy outcomes. Maternal age was defined as the age at the initiation of ovarian stimulation. BMI was calculated as weight in kilograms divided by height in meters squared (kg/m2). Serum AMH levels measured during the treatment cycles were analyzed as continuous variables. Serum AMH levels were measured using a Roche cobas pure automated analyzer (Roche Diagnostics, Mannheim, Germany) based on an electrochemiluminescence immunoassay method. For subgroup analyses, the AMH level was categorized as <10 ng/mL and ≥10 ng/mL, to evaluate potential threshold effects. The ovarian response was defined as the number of retrieved oocytes per cycle. The fertilization outcome was defined as the number of normally fertilized embryos (2PN). The fertilization rate was calculated as the number of 2PN embryos divided by the number of retrieved oocytes. Clinical pregnancy was defined as the presence of a gestational sac on ultrasonography. The clinical pregnancy analysis was restricted to cycles in which fresh embryo transfer (ET) was performed. Cycles managed using a freeze-all strategy, including those in which all embryos were cryopreserved without fresh transfer because of ovarian hyperstimulation syndrome risk or other clinical indications, were excluded from the clinical pregnancy outcome analysis.
3. ART procedures
Controlled ovarian stimulation was performed according to established clinical protocols using gonadotropin-releasing hormone agonist or antagonist regimens, selected based on individual patient characteristics and ovarian reserve markers [15,16]. Recombinant or urinary gonadotropins were administered at the discretion of the attending physician, and final oocyte maturation was triggered according to established ART protocols, followed by oocyte retrieval 34-36 hours later [17]. Fertilization was achieved using either conventional IVF or ICSI, depending on semen parameters, previous fertilization history, and clinical judgment [1,3]. For selected cycles requiring micromanipulation, advanced ICSI techniques such as spindle-view ICSI were utilized as appropriate [18]. Embryo culture and assessments were performed according to standard embryology laboratory procedures.
4. Statistical analysis
Continuous variables are presented as means±standard deviations and categorical variables are presented as counts and percentages. Pearson’s correlation analysis and simple linear regression were used to assess the univariate associations among AMH levels, age, BMI, and ovarian response. Multivariable linear regression models were constructed to evaluate independent predictors of ovarian response (number of retrieved oocytes) and fertilized embryo count (2PN). The multivariable model for the fertilized embryo count included the following covariates: AMH level, age, BMI, fertilization method, and number of retrieved oocytes. Logistic regression analysis was performed to evaluate the independent predictors of clinical pregnancy. Odds ratios (ORs) and 95% confidence intervals (CIs) were also determined. For subgroup comparisons (AMH <10 ng/mL vs. ≥10 ng/mL), continuous variables were compared using Welch’s t-test, and categorical variables were compared using the chi-square test. Statistical significance was defined as a two-tailed P<0.05. All statistical analyses were performed using the JASP Software Version 0.19.3 (University of Amsterdam, Amsterdam, Netherlands). Graphical visualizations were generated using GraphPad Prism Version 5 (GraphPad Software, San Diego, CA, USA).
Results
1. Study population and baseline characteristics
Among the 6,312 screened IVF/ICSI cycles, 4,399 cycles met the inclusion criteria and were included in the final analysis (Fig. 1). The baseline characteristics are summarized in Table 1. The mean maternal age was 37.91±5.24 years and the mean AMH level was 2.09±2.47 ng/mL. The mean BMI was 22.70±3.55 kg/m2. The average number of retrieved oocytes per cycle was 10.67±10.30 and the mean number of normally fertilized embryos (2PN) was 6.80±6.43. The overall clinical pregnancy rate was 20.8%.
2. Association of AMH with ovarian response and fertilization
In the multivariable model for the number of normally fertilized embryos (2PN), AMH demonstrated a small but statistically significant negative independent association (β= −0.044; standard error=0.020; P=0.027) (Table 2). Maternal age was also negatively associated (β= −0.019; P=0.019), whereas BMI was not significant. By contrast, the number of retrieved oocytes was the strongest positive predictor of fertilized embryos (Fig. 2A) (β=0.552; P<0.001).
As shown in Fig. 2B, the crude association between AMH level and fertilized embryo count demonstrated an overall positive trend. However, after adjusting for maternal age, BMI, fertilization method, and retrieved oocyte number, the partial regression analysis presented in Fig. 2C demonstrated a small but statistically significant negative independent association between AMH level and fertilized embryo count.
3. Predictors of clinical pregnancy
According to the multivariable logistic regression analysis, increasing maternal age was significantly associated with lower odds of clinical pregnancy (OR, 0.920; 95% CI, 0.906-0.934; P<0.001) (Table 3). Higher AMH levels were independently associated with reduced odds of clinical pregnancy (OR, 0.955; 95% CI, 0.923-0.989; P=0.010). By contrast, BMI (OR, 1.009; 95% CI, 0.989-1.030; P=0.366) and number of fertilized embryos (OR, 1.010; 95% CI, 0.997-1.023; P=0.143) were not significantly associated with clinical pregnancy.
4. Subgroup analysis according to AMH level and maternal age
In the subgroup analysis stratified by AMH level (<10 vs. ≥10 ng/mL), women with AMH ≥10 ng/mL demonstrated significantly higher oocyte yield (34.41±19.57 vs. 11.85±9.5; P<0.001). However, the fertilization rate was significantly lower in the high-AMH group (0.561±0.214 vs. 0.694±0.531; P<0.001). The proportion of IVF and ICSI cycles also differed between the AMH subgroups, with a relatively higher proportion of ICSI cycles observed in the AMH ≥10 ng/mL group (Table 4). The clinical pregnancy rate was also markedly lower in women with AMH ≥10 ng/mL than in those with AMH <10 ng/mL (5.7% vs. 23.10%; P<0.001) (Table 4). Clinical pregnancy outcomes were analyzed only in cycles involving fresh ET. All freeze cycles without fresh ET were excluded from the analysis. Despite restriction to fresh ET cycles, the AMH ≥10 ng/mL group demonstrated a significantly lower clinical pregnancy rate than the AMH <10 ng/mL group. The threshold-dependent differences are shown in Fig. 3. When stratified by maternal age (<35 vs. ≥35 years), younger women showed significantly higher oocyte yield (16.02±11.83 vs. 10.49±8.81; P<0.001) and more normally fertilized embryos (9.31±7.29 vs. 6.23±5.65; P<0.001). The clinical pregnancy rate was significantly higher in women aged <35 years (29.10% vs. 19.9%; P<0.001). By contrast, fertilization rate was higher in women aged ≥35 years (0.709±0.568 vs. 0.643±0.419; P<0.001) (Table 5).
Discussion
AMH demonstrated a strong and independent positive association with ovarian response; however, after multivariable adjustment, the independent association between fertilization rate and clinical pregnancy was limited or negative. Furthermore, higher AMH levels were independently associated with a reduced likelihood of clinical pregnancy. In the subgroup analysis, women with an AMH level ≥10 ng/mL had a significantly higher oocyte yield than those with an AMH level <10 ng/mL; however, they had significantly lower fertilization and clinical pregnancy rates. These findings suggest a quantitative-qualitative dissociation of AMH levels during ART cycles.
Previous studies have demonstrated that the AMH level is a reliable predictor of ovarian response and oocyte yield during controlled ovarian stimulation [5,7,10]. However, Tal and Seifer [9] and Iliodromiti et al. [12] have reported that although AMH levels strongly predict the number of retrieved oocytes, their association with implantation and live birth outcomes is limited. Moreover, studies in women with polycystic ovarian morphology have suggested that elevated AMH level may reflect an increased follicular quantity without a proportional improvement in reproductive efficiency [13,14,19]. Overall, AMH level may primarily reflect ovarian reserve quantity rather than reproductive competence and support the concept that excessive ovarian response does not necessarily translate into improved fertilization efficiency or clinical pregnancy.
A biological explanation for this paradox may be the heterogeneity of the follicular dynamics in women with elevated AMH levels. High AMH levels are frequently observed in women with PCOS and polycystic ovarian morphology [13,14,19]. AMH inhibits primordial follicle recruitment, thereby regulating folliculogenesis [5,6]. Although these patients typically have an increased antral follicle count and enhanced oocyte retrieval, several studies have indicated compromised oocyte developmental competence in this cohort [11,12,20]. Elevated AMH levels have been associated with altered folliculogenesis, impaired granulosa cell differentiation, and disrupted intraovarian signaling pathways that may adversely affect cytoplasmic maturation and steroidogenesis [5,7,20]. Extremely elevated AMH levels, particularly >10 ng/mL, have also been associated with polycystic ovarian morphology and an increased risk of ovarian hyperstimulation [13,21]. Retrospective analyses have suggested that such high AMH levels may reflect excessive follicular recruitment, without proportional improvement in oocyte maturation or embryo developmental competence [11,12,20,21].
In the present study, the subgroup with AMH ≥10 ng/mL represented only a small proportion of the total cohort. However, extremely elevated AMH levels are generally considered to reflect distinct ovarian characteristics, including PCOS morphology and high-responder features [13,14,19]. A previous study has suggested that markedly elevated AMH levels may represent a distinct ovarian response pattern rather than a general infertility [19]. Therefore, we additionally analyzed the AMH ≥10 ng/mL subgroup because excessively elevated AMH levels could potentially influence or distort the overall association between AMH and ART outcomes.
Consistent with these observations, our subgroup analysis demonstrated that women with an AMH level ≥10 ng/mL had an almost threefold higher oocyte yield, but significantly lower fertilization and clinical pregnancy rates than those with an AMH level <10 ng/mL. Together, these findings reinforce the concept that an excessive ovarian response does not necessarily translate into improved reproductive efficiency. Nevertheless, because the AMH ≥10 ng/mL group comprised only 70 cycles (1.6% of the total cohort), these findings should be interpreted cautiously, and larger prospective studies are required to validate the clinical significance of this threshold effect.
Advanced maternal age remains the most established determinant of reproductive outcomes in women undergoing ART. Numerous studies have demonstrated a decline in pregnancy and live birth rates beginning in the early 30s with a more pronounced reduction after the age of 35 years [4,9,22]. In the present study, using 35 years as the clinical threshold, women aged ≥35 years had a significantly lower clinical pregnancy rate than those aged <35 years (19.9% vs. 29.10%), corresponding to a 31% relative reduction or 0.68-fold lower pregnancy rate in the older group. These findings are consistent with the well-recognized impact of age on oocyte quality and chromosomal integrity [4].
Interestingly, although AMH level was strongly associated with ovarian response, its association with clinical pregnancy in the multivariable analysis was similar to that of maternal age. Despite reflecting the quantity of ovarian reserve, a higher AMH level was independently associated with a lower likelihood of clinical pregnancy after adjustment. Thus, AMH level, similar to maternal age, should not be considered a simple prognostic marker. Instead, both variables should be evaluated using multivariable models that incorporate embryo-related parameters.
Large cohort studies have demonstrated that embryo competence and chromosomal integrity are predominantly influenced by maternal age, rather than by ovarian reserve markers alone [22-25]. In addition, an excessive ovarian response does not proportionally increase implantation or live birth rates [11,12,26]. These observations further support the concept that oocyte quantity does not necessarily reflect optimal reproductive potential.
From a clinical perspective, a high AMH level is often considered reassuring because it predicts a robust ovarian response. However, extremely elevated AMH levels, particularly in women with polycystic ovarian morphology, may reflect increased follicular recruitment without a proportional improvement in oocyte developmental potential [11,12]. Furthermore, ovarian reserve markers alone have a limited predictive value for ongoing pregnancy when embryo-related factors are considered [8,9]. Therefore, AMH levels should be regarded primarily as quantitative indicators of ovarian reserve rather than qualitative predictors of reproductive success.
Our study has some limitations. First, its retrospective design limited causal inferences. Second, although a multivariable adjustment was performed, residual confounding factors could not be excluded. Lifestyle factors, including smoking and metabolic status, have not been fully evaluated and may influence ART outcomes [27]. Third, fertilization quality was assessed using 2PN formation rather than oocyte quality, blastocyst development, euploidy rate, or live births, which are direct embryo-related or clinical outcome measures [28,29]. Finally, the subgroup with AMH level ≥10 ng/mL was relatively small; therefore, these findings should be interpreted with caution. Because the clinical records were de-identified for research ethics purposes, we could not account for potential clustering effects in patients who may have undergone multiple cycles. However, considering that each stimulation represents a distinct biological response in a fresh IVF cycle, its effect on the overall trend was expected to be minimal.
In conclusion, although AMH level strongly predicted ovarian response, an extremely elevated AMH level (≥10 ng/mL) was not associated with improved fertilization or clinical pregnancy rates in ART cycles (Fig. 4). Prospective studies with larger cohorts are required to confirm the clinical implications of this quantitative-qualitative dissociation.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
