Infertility clinics around the globe are encountering an increasing number of female patients with advanced age and/or diminished ovarian reserve, the majority of whom require assisted reproductive technology (ART) interventions, such as intrauterine insemination (IUI) and in-vitro fertilization (IVF) [1, 2]. Assessing ovarian reserve before proceeding to controlled ovarian stimulation (COS) is critical since it aids in identifying women at risk of a poor or excessive ovarian response [3, 4, 5, 6]. The success of patient counseling and treatment planning heavily relies on this initial yet essential step, leading to the development of numerous markers and methods, along with the refinement of the entire ART process [7, 8].

Numerous factors, including age, basal follicle stimulating hormone (FSH), estradiol (E2), anti-müllerian hormone (AMH), ovarian volume, and antral follicle count (AFC), have been used to predict ovarian reserve [9]. Among these factors, both AMH and AFC exhibit superior predictive value compared to other measures [7, 10, 11, 12, 13]. They are closely correlated and are frequently used interchangeably [14, 15]. Indeed, there is mounting evidence that these indicators are associated to the initial and final ovarian response, number of retrieved oocytes, and ultimately even the live birth rate [13, 16, 17].

However, within this interrelated status of AMH and AFC, discordance is commonly encountered in clinical practice. IVF specialists often confront situations in which the serum AMH level is either higher or lower than expectations based on the AFC, even when the AMH measurement and AFC are performed at the same clinical center during the early follicular phase of the same menstrual cycle [18, 19]. The observed discordance cannot be solely attributed to technical issues in antral follicle counting or the analytical variability of the AMH assays [3]. This incongruity poses challenges and uncertainties in the pre-treatment patient counseling and COS planning, which is particularly critical in older patients where even a single oocyte retrieval result can lead to significant difference in cascade to IVF pregnancy outcome.

Undoubtedly, while a single IVF procedure does not guarantee a successful pregnancy, when the discordance between AMH and AFC is present, improper management can undermine the clinician’s confidence in the patient’s long-term ART journey. This can have enduring negative implications for subsequent procedures. Although there have been limited studies on this discordance, even these studies have not provided a consistent answer regarding which ovarian reserve marker is reliable in such cases [18, 19, 20, 21]. Therefore, the objective of this study is to determine which marker is more reliable in infertility patients with the AMH and AFC discordance by comparing their ART outcomes between different patient groups with congruent or discordant values of AMH and AFC, as well as evaluating possible patient-related factors that could contribute to such discordance in ovarian reserve markers.

The clinical and biochemical characteristics of all patients were presented in Table 1. The mean age of the entire patient was 35.20 $\pm$ 4.42 years, ranging from 23 to 47 years. The mean AMH level was 3.64 $\pm$ 3.98 ng/mL, and AFC was 9.24 $\pm$ 5.52, indicating the inclusion of patients with a diverse range of ovarian reserve markers across various age groups in the study. The HTP group (n = 46) was younger than the LTP group (n = 46), while there were no statistically significant differences between the two groups in terms of gravida, BMI, infertility duration, infertility type, or FSH. Regarding the history of ovarian pathology, the HTP group exhibited the higher prevalence of polycystic ovary syndrome (PCOS), while the LTP group predominantly had cases of endometriosis.

Table 2 compared COS protocols and ART outcomes among the patient groups. In the HTP group, 61% of patients used the antagonist protocol in COS, while the LTP group predominantly utilized the long protocol (67%). The mean total FSH usage in the HTP group was lower than that in the LTP group, with the LTP group having the highest mean total FSH usage among the three groups. The HTP group displayed significantly higher numbers of pre-ovulatory follicles, total oocytes retrieved, metaphase II oocytes, total embryos, and blastocysts, along with higher FOI and FORT values when compared to the LTP group.

The results of logistic regression analysis for correlation between patient groups and ART outcomes were presented in Table 3. Events that exceeded the median values for ART outcomes in the entire patient were considered as the outcomes. When comparing the HTP group to the LTP group, the OR [95% confidence interval (95% CI)] were found to be 3.248 [1.375–7.673] for pre-ovulatory follicles, 8.028 [3.165–20.361] for FOI, 13.081 [4.837–35.373] for FORT, 3.899 [1.637–9.285] for total oocytes retrieved, 4.760 [1.968–11.512] for metaphase II oocytes, 4.760 [1.968–11.512] for total embryos, and 3.949 [1.649–9.456] for blastocysts. These ORs remained statistically significant even after adjustments for AFC alone and when adjusting age, BMI, AFC and total FSH usage together. Table 4 described the results of logistic regression analysis aimed at identifying factors influencing discordance. While age did not exhibit statistical significance, presence of PCOS was associated with OR of 1.24 with 95% CI of 1.043–1.475.

Infertility specialists frequently encounter clinical situations where a patient’s AMH and AFC do not align, and most of them tend to consider such discordance as a technical limitation [18, 19]. However, in 2018, a study was published indicating that when such discordance occurs, pregnancy rates were lower in concordant patients [20]. Furthermore, in real clinical settings, healthcare providers face challenges when they encounter discordance, particularly in the pre-treatment patient counseling and COS planning processes. Alebic et al. [3] suggested that this discordance goes beyond being a mere technical constraint and instead reflects patient-specific differences in the production of AMH within the follicles. When the actual measurement of AMH was higher than what could have been expected from the AFC, it might indicate an up-regulation of AMH secretion, which was a typical clinical feature of PCOS. Conversely, when AMH was lower than what could have been expected from the AFC, it might suggest down-regulated AMH secretion, which could be an early indication of diminished ovarian reserve and premature ovarian insufficiency. In other terms, when challenged against AFC, the serum AMH level could not only a quantitative but also a qualitative follicle marker, in relation with clinical and endocrine parameters [3].

The aim of the current study did not focus on exploring any underlying mechanisms of the discordance between AMH and AFC. Nevertheless, several patient-related clinical factors influencing the discordance were confirmed in the study; it was noted that patients with higher AMH in comparison to their AFC were more inclined to have PCOS, and the presence of PCOS increased the discordance rate among the infertility patients. This implies that PCOS could be regarded as one of the factors contributing to the observed discordance. Moreover, the group with higher AMH levels compared to the expected values from AFC had the lowest mean age and the lowest total FSH dosage among the groups, and their ART outcomes were more favorable than those of the group with lower AMH compared to AFC. In contrast, the group with lower AMH compared to AFC turned out to have the highest mean age and highest total FSH usage, also having less favorable ART outcomes than the other groups. Even after adjusting for factors known to affect ovarian response, such as age, BMI, AFC, and total FSH usage, the group with higher AMH compared to AFC demonstrated superior results in terms of oocyte yield and good-quality embryo rate compared to the group with lower AMH compared to AFC. This suggests that AMH might be a more robust predictor of ovarian response than AFC when the discordance is present.

In the study conducted by Li et al. [21] which involved 1046 patients of various age groups, in the presence of discordance between AMH and AFC, the group with higher AMH compared to AFC exhibited a higher number of retrieved oocytes and a greater cumulative live-birth rate. Similarly, Guo et al. [24] confirmed that in the group with higher AMH compared to AFC, there was a greater yield of oocytes and favorable embryos. Furthermore, in a recent investigation by Aslan et al. [25] involving 662 Turkish diminished ovarian reserve patients, when discordance between AFC and AMH was present, those with normal AMH and low AFC showed a better ovarian response than the patients with low AMH and normal AFC. This led to the conclusion that serum AMH possess higher predictive value for stimulation success in cases of discordance with AFC [25].

Various patient factors have been examined to understand their possible influence on such discordance between AMH and AFC. In the study of Zhang et al. [26], patients were classified according to Bologna criteria, and the length of the menstrual cycle was observed as a possible discordant factor. When the discordance arose, the group with higher AFC to detected AMH had significantly higher oocyte yield, embryo quality, and clinical pregnancy rates than the group with lower AFC to AMH. They concluded that the AFC should be the preferable marker for predicting ovarian response in order to design the most effective personalized COS regimen in this group of patients [26]. Their conclusions are contrasting to ours, probably due to the difference in the adopted study design; Zhang’s study [26] excluded women who had undergone ovarian surgery and/or been diagnosed with PCOS, whereas in the current study, more than one-third of the patients were diagnosed with endometriosis or PCOS. Regarding endometriosis and PCOS, as they have been widely known as the leading causes of infertility, the current study tends to more practically reflect the real-world clinical data of poor ovarian response (POR) patients, thus being more clinically appropriate in deciding ART management of POR patients. Yet, despite the varying results, previous literature has commonly considered the age and PCOS history of a patient as possible discordant components. These studies do agree that discordance is not merely a technical issue; when discordance exists, the COS outcome and clinical pregnancy rate can be reduced. When possible discordant factors are observed, careful analysis of AMH and AFC might enhance the success of the ART treatment [21, 24, 25, 26].

To our knowledge, the current study is the first analysis of Korean infertility patients investigating the discordance between AMH and AFC among ethnically and racially homogenous study groups. Also, while most of the previous studies related to discordance primarily focused on ART outcomes, specifically the number of oocytes, FOI, and FORT, our study employed a comprehensive approach by not only evaluating the quantity but also the quality of oocytes and embryos, as well as FOI, and FORT. Moreover, it is noteworthy that previous studies addressing discordance typically categorized patients using criteria such as the Bologna or Poseidon classification, which might have not effectively accounted for the diverse range of ages and values for AMH and AFC observed in clinical practice. On the other hand, we employed an AMH prediction model based on actual AFC from the study groups when classifying the patients, thereby overcoming such limitation.

Several limitations still exist and should be thoroughly considered in the current study. Firstly, the study’s retrospective nature imposed constraints on addressing potential heterogeneity in COS protocols; the adjustment of these protocols was not feasible within the current framework. It is important to emphasize that the primary focus of the study did not involve evaluating the specific protocols themselves; nevertheless, the selection of the most appropriate stimulation protocols was consistently made by the same experienced reproductive endocrinologist and embryologists within a tertiary hospital setting throughout the study. Secondly, the sample size was relatively small due to the study being conducted at a tertiary infertility clinic in a single center, and it did not include the assessment of live birth rates as one of the outcome measures. Thus, there is a compelling need for a multi-centered study design with a more diverse population that incorporates live birth rate data. Despite these limitations, the findings of this study can serve as a foundational step towards improving effective treatment plans for infertility patients and providing valuable guidance to specialized clinicians faced with the challenge of discordance between AMH and AFC.

Serum AMH provided a more accurate prediction of the patient’s ovarian reserve, especially in the presence of discordance between AMH and AFC, resulting in more favorable ART outcomes in the patients with higher AMH compared to their AFC. Regarding the discordance between AMH and AFC, the presence of PCOS could be considered as one of the significant factors contributing to this observed discordance; special clinical attention should be given to this specific patient group when tailoring ART management plans in order to magnify their chances of a successful pregnancy.

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

EY and HL designed the research study. EY performed the research. JJ and YN provided help and advice on data gathering and analysis. HL analyzed the data. EY, HL, JJ, and YN wrote the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

This study was submitted to the Institute Review Board (IRB) with the assigned number 2209-006-118, granted ethical approval by the IRB of Pusan National University Hospital. Given its retrospective observational nature, informed consent was not required.

Every statistical analysis performed in this study was professionally discussed and consulted with the Department of Biostatistics, Biomedical Research Institute, Pusan National University Hospital.

This research received no external funding.

The authors declare no conflict of interest.