Cost-Effectiveness of Clomiphene Citrate-Based Minimal Stimulation Compared with Controlled Ovarian Stimulation Protocol in Women with Diminished Ovarian Reserve

INTRODUCTION

Assisted reproductive technology (ART) is an effective treatment for unexplained infertility [Pandian et al., 2015]. Optimal ovarian stimulation is important in ART. Controlled ovarian stimulation (COS) protocols use various gonadotropin (Gn) preparations and adjust the dosage and duration of administration to produce multiple oocytes. A positive correlation has been observed between the live birth rate and the number of oocytes acquired per cycle, with a higher number of eggs retrieved increasing the cumulative live birth rate without compromising the first embryo transfer outcome [Fanton et al., 2023; Polyzos et al., 2018]. In contrast, the benefit of oocyte retrieval on cumulative live births with high doses of Gn in patients with diminished ovarian reserve (DOR) is small [Datta et al., 2020; Lensen et al., 2018; Montoya-Botero et al., 2021]. Ovarian stimulation protocols for patients with DOR have not been established [Baldini et al., 2023; Xia et al., 2021; Zhao et al., 2023].

Depending on the patient’s background, selecting an appropriate protocol is crucial, as several methods exist [Esteves et al., 2018; Ferraretti et al., 2011]. Among these, the POSEIDON criteria are characterized by the ability to group patients before the first oocyte retrieval using age, antral follicle count (AFC), and anti-Mullerian hormone (AMH), with a strong correlation with clinical results [Esteves et al., 2021].

In Japan, insurance coverage for ART commenced in April 2022, increasing societal interest in fertility treatments. According to the Japan Society of Obstetrics and Gynecology, Japan is characterized by an aging population of patients undergoing in vitro fertilization (IVF), with 39.9% of treated patients over 40 years of age [Katagiri et al., 2024]. The number of treatment cycles for patients with DOR often increases as the population ages. The International Committee for Monitoring Assisted Reproductive Technology reports that Japan has the second highest number of ART treatment cycles globally, at 400,000 cycles per year [Chambers et al., 2021]. Increased treatment cycles lead to an increased economic burden on patients and society, especially as Japan has a national health insurance system. In Japan, where infertility treatment is now covered by insurance, selecting appropriate protocols for patients with DOR based on treatment outcomes and cost-effectiveness is necessary.

This study aims to provide information on the outcomes and cost-effectiveness of COS and mild stimulation for patients with DOR in Japan.

METHODS

Ethics approval

The institutional review board of Showa University approved this retrospective cohort study (number: 2023-015-A). Written informed consent for the retrospective analysis of de-identified data was obtained from all patients undergoing IVF treatment at our hospital.

Study design

Electronic records of women who had undergone oocyte retrieval at the Showa University Hospital Assisted Reproduction Centre in Japan between May 2019 and November 2023 were screened. We included patients with poor prognoses based on the POSEIDON criteria. Only the POSEIDON 3 or 4 groups were included in the analysis. The POSEIDON 3 group comprised women younger than 35 with an $\mathrm{\text{AFC}}<5$ or $\mathrm{\text{AMH}}<1.2$ng/mL based on ovarian biomarker analysis. The POSEIDON 4 group comprised women older than 35 with an $\mathrm{\text{AFC}}<5$ or $\mathrm{\text{AMH}}<1.2$ng/mL [Esteves et al., 2018]. Each patient was included only once in the analysis (Fig. 1).

One treatment group received clomiphene citrate (CC) with exogenous Gn (EGn), whereas the other group underwent standard ovarian stimulation with ovulation suppression using a Gn-releasing hormone (GnRH) antagonist or progestin. All embryos from both groups were cultured to the blastocyst stage and frozen using the vitrification method. All the patients were treated independently by physicians. Supernumerary blastocysts were cryopreserved and transferred before a new IVF treatment cycle. To compare the cumulative live birth rate per cycle, we excluded patients with surplus embryos for whom transfers were not completed. In both groups, cryopreserved embryos underwent single embryo transfer (SET). For each IVF protocol, the number of stimulation days, medical examinations, the total dose of follicle-stimulating hormone (IU), the number of eggs retrieved, the number of fertilized embryos, the total number of frozen embryos, high-quality blastocysts, medical examinations before oocyte retrieval, and blood hormone levels were measured. In addition, pregnancy outcomes, including ongoing pregnancy and live birth rates after ovarian stimulation between the two groups, were retrospectively analyzed.

Stimulation protocols in IVF

A detailed protocol for CC-based minimal stimulation using CC alone exists [Kato et al., 2018]. Ovarian stimulation initiated after an ultrasound examination during menstruation confirmed that luteinized, unruptured follicles had not developed beyond 10mm. CC (50mg/d; Fuji Pharma Co., Ltd.) was orally administered with an extended regimen from Day 3 of the retrieval cycle to the day before induction of ovulation (trigger). The date of the first visit after the start of the stimulation cycle was designated as Days 8–11, based on the menstrual cycle and basic hormone levels during menstruation. Monitoring, including ultrasonography and hormonal profiling, was performed when the primary follicle was less than 15mm, with at least three follicles larger than 10mm present. A minimum dose of follicle-stimulating hormone (FSH) (FOLYRMON-P; Fuji Pharma Co., Ltd.) was continuously administered. When the leading follicle exceeded 17mm in diameter, ovulation triggering was performed using a nasal spray containing the GnRH agonist buserelin (Suprecur; Mochida Pharmaceutical Co., Ltd.). Human chorionic Gn was administered when the serum luteinizing hormone level on the trigger day was less than 1.0. Oocyte retrieval was performed 34–36h after induction using a 21G needle (Kitazato Corporation). No follicle washing was performed.

In contrast, the COS group included patients in whom GnRH antagonist or progestin was used to suppress ovulation. In both groups, the baseline scan for COS was the same as described earlier. The starting date of EGn administration was adjusted between menstrual days 2–4, with doses of 150 or 225 IU/d initiated daily or at the physician’s discretion. When the leading follicle exceeded 14mm, a combined injection of FSH and 0.25mg GnRH antagonist (Cetrotide; Merck & Co.) was initiated. When progestin was used for ovulation suppression, dydrogesterone (Duphaston; 30mg/d; Viatris) was administered on the day of EGn administration. When at least two follicles exceeded 17mm, ovulation was triggered using a nasal spray containing the GnRH agonist buserelin and recombinant human chorionic Gn (Ovidrel; Merck & Co.). Oocyte retrieval was performed as previously described. Fertilization was performed via IVF or intracytoplasmic sperm injection (ICSI), depending on the semen parameters. IVF or ICSI was performed approximately 3 and 5h after oocyte retrieval, respectively. Fertilization assessment was performed 16–20h after insemination. All fertilized eggs were cultured for 5–6 days and cryopreserved as blastocysts. An experienced embryologist evaluated embryo quality for all available embryos on Day 5 or 6 of culture. On the basis of Gardner’s classification, 4 BB or more were considered high-quality blastocysts. All the embryos were cryopreserved.

Embryo transfer

Only SETs were performed during the study period. Endometrial preparation for frozen-thawed embryo transfer was performed as natural or hormone replacement cycles, depending on patient availability. The natural cycle began on Day 3 of menstruation with letrozole (Letrozole, 2.5mg/d; Nippon Kayaku Co., Ltd.) administered for 5 days. A visit to the hospital was indicated on Days 12–14 of menstruation to confirm that the follicle diameter was at least 18mm and then triggered with GnRHa. The SET was performed 5 days after ovulation. In contrast, hormone replacement cycle implantation was initiated on Day 2 of the menstrual cycle using estradiol (ESTRANA Tapes; HISAMITSU) and continued until the endometrium reached a thickness of 8mm on ultrasound monitoring. Oral and vaginal progesterone tablets were administered three times daily, and SET was performed 5 days after the completion of five full days of progesterone. Progesterone replacement was administered in all patients with oral dydrogesterone (Duphaston, 30mg/d; Viatris) and intravaginal progesterone (Lutinus, 300mg/d; Ferring Pharma Co., Ltd.). Clinical and ongoing pregnancy rates were defined by ultrasonographic observation of the gestational sac 5–6 weeks after embryo transfer and the fetal heartbeat at 7 weeks after embryo transfer, respectively. The clinical and ongoing pregnancy rates were analyzed.

Cost calculations

The cost from the start of stimulation to embryo freezing was calculated and averaged for each stimulation protocol. The cost of drugs used for ovarian stimulation was calculated by multiplying the health insurance drug prices. The total cost of consultations, ultrasounds, blood tests, and other necessary procedures was calculated by multiplying the current insurance points (medical fee points) by 10 yen per point. The cost from oocyte retrieval to embryo freezing was calculated for each stimulation protocol, and the live birth rate was evaluated as the primary effect measure. The difference in costs divided by the difference in live birth rate was used as the incremental cost-effectiveness ratio (ICER) and cost per additional birth. Costs related to embryo transfer, pregnancy, and neonatal were excluded.

Statistical analyses

Statistical analyses were performed using JMP software (SAS Inc., Cary, NC, USA). Continuous parameters were compared using Student’s t-test or one-way analysis of variance, and statistical significance was determined using Tukey’s test for post-hoc analysis. Proportional data were analyzed using Pearson’s Chi-square test. Statistical significance was set at $p<0.05$.

A propensity score-matching model was applied to balance baseline characteristics, including body mass index, AMH, and AFC, between the CC and COS groups. The propensity scores were calculated using a logistic regression model. Patients in the CC group were matched with those in the COS group in a 1:1 ratio based on the propensity score with a standard caliper width of 0.2.

RESULTS

Patient characteristics

Table 1 shows the cohort characteristics of the CC and COS groups. In total, 24 and 42 patients in the CC and COS groups, respectively, underwent oocyte retrieval. Baseline characteristics, including age (41 vs. 42 years, $p=0.68$) and serum AMH level (0.68 vs. 0.79, $p=0.60$), were similar between the CC and COS groups. However, the AFC (2 vs. 3, $p<0.05$) was higher in the COS group than in the CC group (Table 1).

Stimulation cycle characteristics and outcomes

Compared with the COS group, the CC group had significantly fewer medical examination days ($p<0.0001$), a shorter stimulation duration ($p<0.0001$), and a lower total EGn dose ($p<0.0001$). The median numbers of cumulus–oocyte complexes ($p=0.1686$), metaphase-two oocytes ($p=0.5175$), total number of blastocysts ($p=0.5694$), and high-quality blastocysts ($p=0.9169$) were similar between the CC and COS groups. The rates of clinical pregnancy, ongoing pregnancy, and live births were comparable between the two reference groups. Table 2 shows the characteristics and outcomes of the stimulation cycles.

Comparison of costs required using stimulation methods

Table 3 shows the average costs per treatment cycle. The cost per cycle for each stimulation method was calculated by multiplying the amount of drug used by the health insurance drug price on average. The costs for oocyte retrieval and embryo culture were calculated similarly by multiplying the medical fee points by 10 yen per point. The CC group had significantly lower drug cost requirements because they used fewer EGn and no GnRH antagonists ($p<0.0001$). When applied to the current insurance scores, the CC group had significantly lower costs during ovarian stimulation for the cycles in which oocytes were retrieved ($p<0.0001$), frozen embryos ($p<0.05$), and live birth rate than the COS group ($p<0.05$). The relationship between live birth rate and the cost of ovarian stimulation was more favorable in the CC group than in the COS group (ICER: 46,024/8.9%) (Fig. 2A). Similarly, the comparison with high-quality blastocysts was favorable (ICER: 46,024/0.6%) (Fig. 2B).

Clinical outcomes of both groups after propensity score matching

After post-matching analysis, the baseline characteristics and outcomes, including body mass index, numbers of cumulus–oocyte complexes, total blastocysts, high-quality blastocysts, and cumulative live birth rate, were comparable between the CC and COS groups. The total amount of EGn and the total cost of ovarian stimulation were significantly higher in the COS group than in the CC group. Table 4 shows a detailed comparison of the groups.

DISCUSSION

Several reports on clinical outcomes have compared mild stimulation and COS in patients with DOR. Furthermore, mild stimulation costs less than COS [Polinder et al., 2008]. However, a comparison of medical costs between CC-based minimal stimulation and COS based on Japanese insurance scores has not yet been reported. Our study demonstrates that CC-based minimal stimulation is less expensive for patients in POSEIDON groups 3 and 4 than COS for ovarian stimulation without reducing the cumulative live birth rate. GnRH antagonists and progestins are effective ovulation suppressors; however, EGn should be simultaneously administered to avoid interference with follicular development [Chen et al., 2022]. This inference inevitably leads to an increase in the total amount of EGn used. Conversely, continuous dosing of mild with CC was possible with minimal doses of EGn owing to its effect on ovulation suppression. In the CC group, the dose of EGn was reduced, eliminating the need for a new ovulation suppressant, suggesting that costs could be reduced. In this study, no significant differences were observed in the number of oocytes or blastocysts frozen between the two groups, suggesting that the administration of EGn for COS in patients with DOR may only suppress follicle development by suppressing ovulation. In contrast, some reports suggested that EGn in patients with inadequate endogenous Gn secretion in the late follicular phase can restore clinical outcomes despite inadequate follicle growth when CC is initiated alone [Karakida et al., 2020]. The fact that only a small amount of EGn was administered to the CC group in this study does overrule EGn use, which can be considered if multiple developments are observed in the late follicular phase. The significantly higher number of consultations in the COS group may be attributed to the need for more frequent follicle size and blood tests to determine when to start ovulation suppression. In contrast, the CC group required less frequent monitoring because continuous administration of CC suppressed ovulation [Kato et al., 2012; Teramoto et al., 2007]. This finding may have led to fewer visits in the CC group than in the COS group. When multiple AFCs were present, significantly more patients in the COS group had a higher AFC, as COS was often performed even in patients with low AMH levels. Therefore, the amount of EGn used increased. This study analyzed data matched by propensity scores to balance baseline characteristics such as age, AMH, and AFC. Post-match data analysis showed that the amount of EGn, costs associated with ovarian stimulation, and costs of blastocyst freezing were significantly lower in the CC group without reducing the live birth rate. This finding suggests that the COS group may receive EGn, which is unnecessary for follicle development. These findings are consistent with those of previous studies reporting that ovarian stimulation with EGn adversely affects oocyte and embryo development [Van der Auwera et al., 2001; Lee et al., 2017; Uysal et al., 2018; Wang et al., 2015; Yu et al., 2019].

However, this study has some limitations, including a small number of cases and a retrospective study design. In addition, despite propensity-score matching, some differences in patient backgrounds, such as ovarian responsiveness, may exist. Therefore, further randomized controlled trials are required to determine the appropriate ovarian stimulation protocol for patients with DOR. The social insurance situation should be considered, and cost-effectiveness should be evaluated. In addition, this study aims to assess the cost differences between ovarian stimulation methods and does not consider embryo transfer costs. As the proportion of embryo transfer methods differed between the two groups, if the transfer methods are standardized, the results of the statistical analysis of cost-effectiveness may change. The optimal ovarian stimulation protocol for patients with DOR must be further investigated.

In conclusion, in Japan, where infertility treatment is covered by insurance, the initial treatment of patients with DOR should be initiated with mild stimulation, which is cost-effective.

ACKNOWLEDGMENTS

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CONFLICT OF INTEREST

The authors declare no competing interests.

ORCID

Shogo Nishii  https://orcid.org/0000-0003-0787-3552

Ryosuke Akino  https://orcid.org/0000-0002-9137-2146

Chen Linji  https://orcid.org/0009-0007-6042-463X

Miwa Sakamoto  https://orcid.org/0009-0000-1952-9295

Akihiko Sekizawa  https://orcid.org/0000-0003-2249-1333