What is the true place of a double stimulation and double oocyte retrieval in the same cycle for patients diagnosed with poor ovarian reserve? A systematic review including a meta-analytical approach

Konstantinos Sfakianoudis; Konstantinos Pantos; Sokratis Grigoriadis; Anna Rapani; Evangelos Maziotis; Petroula Tsioulou; Polina Giannelou; Adamantia Kontogeorgi; Agni Pantou; Nikolaos Vlahos; Michael Koutsilieris; Mara Simopoulou

doi:10.1007/s10815-019-01638-z

. 2019 Dec 3;37(1):181–204. doi: 10.1007/s10815-019-01638-z

What is the true place of a double stimulation and double oocyte retrieval in the same cycle for patients diagnosed with poor ovarian reserve? A systematic review including a meta-analytical approach

Konstantinos Sfakianoudis ¹, Konstantinos Pantos ¹, Sokratis Grigoriadis ^2,³, Anna Rapani ^2,³, Evangelos Maziotis ^2,³, Petroula Tsioulou ^2,³, Polina Giannelou ^1,², Adamantia Kontogeorgi ², Agni Pantou ¹, Nikolaos Vlahos ³, Michael Koutsilieris ², Mara Simopoulou ^2,^3,^✉

PMCID: PMC7000611 PMID: 31797242

Abstract

Purpose

This systematic review including a meta-analytical approach aims to investigate the safety and efficacy of employing a double ovarian stimulation (DuoStim) and a subsequent double oocyte retrieval in the same menstrual cycle, in poor ovarian reserve (POR) patients.

Methods

A systematic search of literature was performed in the databases of PubMed/MEDLINE, Embase, and Cochrane Central Library up until March 2019. Both prospective and retrospective cohort studies considered suitable for inclusion reported on women with POR undergoing a DuoStim in the follicular (FPS) and luteal phase (LPS) of the same menstrual cycle. Following the systematic review of the literature, a meta-analytical approach was attempted.

Results

This study indicates that DuoStim is correlated with a higher number of retrieved oocytes, mature MII oocytes, and good-quality embryos in comparison to conventional stimulation. Additionally, LPS seems to be correlated with an equal or an even higher overall performance in comparison to FPS.

Conclusion

DuoStim favors an enhanced clinical outcome in regard to the total number of yielded oocytes, mature oocytes, and available embryos, along with the quality of obtained embryos. Sourced data indicate that LPS is not correlated with a higher aneuploidy rate. This option may present as promising for the time-sensitive nature of POR patients’ management, by enabling a higher oocyte yield during a single menstrual cycle.

Keywords: Luteal phase stimulation, Follicular phase stimulation, Double stimulation, DuoStim, Poor ovarian response

Introduction

Despite common belief that the female menstrual cycle presents with a single follicular wave, several reports have revealed that more than one wave of follicular recruitment may be observed during the same menstrual cycle [1]. A research published by Baerwald et al. based on serial transvaginal ultrasonography examinations along with hormonal testing in healthy women established the wave-like follicular development during the menstrual cycle, providing solid evidence on the occurrence of such events [2]. The phenomenon of multiple ovarian follicular waves in numerous animal species seems to be consistent with the events observed during the menstrual cycles in humans. Animal studies have provided valuable insight and contributed towards a better understanding of the physiology entailed in this unorthodox phenomenon. The elevation of follicle-stimulating hormone (FSH) levels described in domestic animals prior to the recruitment of each follicular wave has also been observed and confirmed in healthy human individuals [3]. Furthermore, inhibin B seems to present with a second unusual peak 2 days following the mid cycle luteinizing hormone (LH) surge, serving as an indicator of a luteal phase follicular development [4]. The involvement of FSH and inhibin B may imply mechanisms associated with follicular waves; however, the all-inclusive process of physiological phenomena has not been fully elucidated yet [5].

Understandably, this new model of folliculogenesis, suggesting the concept that women exhibit two or three waves of folliculogenesis during the same menstrual cycle, opens a new line of investigation on optimal treatment in the context of in vitro fertilization (IVF). Its application in the context of IVF treatment has been further investigated, as a promising step towards optimizing and personalizing the IVF procedure for certain groups of patients. As a result, various studies have developed protocols implementing the phenomenon in ART [6]. In the last decade, research has been published evaluating the potential of luteal phase stimulation either separately or in combination with regular stimulation protocols [7–9]. Thus, the approach of double stimulation (DuoStim) has been suggested implementing conventional follicular phase stimulation (FPS) combined with luteal phase stimulation (LPS) [10]. Overall, the approach of collecting oocytes during the luteal phase has been introduced as beneficial for certain groups of infertile patients, including advanced maternal age patients, or patients with inadequate number of oocytes retrieved during conventional FPS [10]. Benefits associated with this practice may be heightened in patients presenting with poor ovarian response (POR) as well or in cases of urgent fertility preservation for women diagnosed with ovarian cancer scheduled for imminent chemotherapy, oophorectomy, or any gonadotoxic therapy [11, 12].

Hitherto, literature is lacking robust data that may assist clinicians in exploring the concept and safely adopt the practice of luteal phase oocyte retrieval in clinical practice. Guidance on comprehending how this phenomenon could be implemented to enable clinicians’ better practice is imperative. Interestingly, various studies emerging lately explore the value of multiple follicular waves. Reports demonstrate the superiority of the LPS in some cases compared to the standard FPS [13]. On the contrary, early publications indicate that the oocytes retrieved during luteal phase stimulation present with certain “weaknesses” regarding characteristics such as attrition or reduced number of granulosa cells compared to the follicular phase oocytes [14]. Nevertheless, given the fact that performing a second oocyte retrieval during a single menstrual cycle undoubtedly increases the number of recruited oocytes in a shorter timeframe [15], this approach may serve as a valuable asset in the hands of clinicians managing the challenging cases of patients for whom time is of the essence.

This systematic review including a meta-analytical approach, set out to compare the efficacy of LPS versus FPS and subsequent double oocyte retrieval practice under the prism of performing the DuoStim approach within the same menstrual cycle. The rationale behind this study is to determine the role and effectiveness of this method by presenting advantages and disadvantages through a systematic analysis and critical review, along with concurring on the management of POR patients employing this approach. Furthermore, reporting on the value of performing a DuoStim during the same menstrual cycle versus the conventional single follicular phase stimulation (ConStim) is attempted.

Materials and methods

Search strategy

A systematic search of literature was conducted in databases of PubMed/MEDLINE, Embase, and Cochrane Central Library, with the aim to elicit full-length articles published in peer-reviewed journals up to the 27th of March 2019. The keywords along with respective combinations that included in the search strategy were “In Vitro Fertilization,” “IVF,” “Intracytoplasmic Sperm Injection,” “ICSI,” “Assisted Reproductive Techniques,” “Assisted Reproductive Technologies,” “ART,” “Medically Assisted Reproduction,” “MAR,” “Luteal Phase,” “Double,” “Oocyte Retrieval,” “Egg Retrieval,” “Oocyte Pickup,” “OPU,” “Ovulation,” “Stimulation,” “Follicular Wave,” “DuoStim,” “Dual Stimulation,” “Second Follicular Wave,” and “Lupor.” The preliminary search yielded 4890 studies from the three databases (PubMed/MEDLINE: n = 1401, Embase: n = 2552, Cochrane Central Library: n = 937). From the total yield, 2037 studies were duplicates and thus removed, while 239 studies from the Cochrane database were directly excluded, as they were categorized as reviews (n = 206), protocols (n = 26), editorial (n = 1), and clinical answers (n = 6). The initial screening of titles and abstracts of all records resulted in the removal of 2533 studies, in an attempt to obtain only the relevant articles. Thenceforth, full texts were screened seeking for ultimately relevant articles, along with citation mining that was performed at the reference lists of all included articles to identify other article of relevance. This thorough screening resulted in five prospective studies [6, 9, 10, 16, 17] and six retrospective cohort studies [7, 18–22] that were eligible to be included. In Fig. 1, the flow chart of Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) is depicted. Screening and selection of literature was performed independently by three authors (AR, PT, and PG). Any disagreements between the authors were resolved by an arbitration mediated by the senior authors. No protocol was submitted to the Prospero International Prospective Register of Systematic Reviews (PROSPERO), providing details on conducting of this study.

Fig. 1 — PRISMA flowchart regarding the search results

Study selection

The authors of the present study opted for including both prospective and retrospective cohort studies that were published in English in peer-reviewed journals. The population comprised of women with POR that underwent ovarian stimulation in the follicular and luteal phase of the same menstrual cycle, during IVF/ICSI cycles. This design served our purpose on concurring on the effectiveness of the DuoStim during the same menstrual cycle, in comparison to the conventional protocol, entailing the employment of a single controlled ovarian stimulation (COS) throughout the follicular phase [23]. Additionally, in this distinct group of patients subjected to DuoStim, we further present associations on embryology and clinical data originating from the comparison of FPS versus LPS.

Excluded studies

Studies employing as a population of interest time-sensitive patients presenting with premature ovarian insufficiency [24] or diagnosed with cancer [25, 26] were excluded from the present study. Other exclusion criteria were on the grounds of performing a double oocyte retrieval in the same menstrual cycle during a natural cycle instead of a stimulation cycle [8]. Moreover, the study of Leahomschi et al. was excluded due to the publication language being Czech [27]. Additionally, following personal communication with the team of Professor Cimadomo, it was elucidated that the two following studies were part of the same project. This led to the decision of including only the data resulting from the multicenter study [10] which was encompassing the data resulting from the smaller intra-patient paired case-control study [28].

Data extraction

Data extraction was performed independently by two authors (EM and SG) based on the selection criteria.

Outcomes

The reported primary outcome measures of interest were number of oocytes retrieved and number of MII oocytes retrieved. The reported secondary outcomes were days of stimulation, cycle cancelation rate, number of cycles with zero MII oocytes retrieved, fertilization rate, blastocyst formation rate, clinical pregnancy rate, miscarriage/early pregnancy loss, and live birth/ongoing pregnancy rate. The authors herein define clinical pregnancy and live birth/ongoing pregnancy rates as the reported respective number per cycle. The miscarriage rate is defined as the number of miscarriages per patients achieving a clinical pregnancy. Comparisons of the above-mentioned outcomes are presented in two arms, the first being FPS versus LPS and the second being the DuoStim protocol versus ConStim only.

Bias assessment

Assessment of bias was performed on the selected studies regarding (i) confounding, (ii) selection of the participants, (iii) classification of intervention, (iv) variations from intended intervention, (v) missing data, (vi) measurement of outcomes, and (vii) reporting selection. The bias assessment was performed independently by three authors (SG, AR, and EM) employing ROBINS-I tool for assessing risk of bias in non-randomized interventional studies [29].

Statistical analysis

Following the descriptive presentation of extracted data, a statistical analysis was performed. Statistical comparisons of the above-mentioned outcomes were performed in two arms, the first being FPS versus LPS and the second being the DuoStim protocol versus ConStim. Furthermore, subgroup analysis was performed in both two arms for all of the examined outcomes in regard to the specific patient population that was studied, namely POR subgroup and mixed population subgroup. POR subgroup includes patients diagnosed with POR according strictly to the Bologna criteria and the mixed population subgroup includes patients categorized as POR failing to be categorized according to specific universally accepted criteria. Odds ratio (OR) with 95% confidence intervals (95%CI) was employed for the analysis of dichotomous data. Odds ratio greater or less than 1 represents a statistically significant difference. Mean difference (MD) with 95%CI was employed regarding the analysis of continuous data. Mean difference greater or less than 0 represents a statistically significant difference. P value < 0.05 was considered statistically significant. Either the fixed effect or the random effects model was employed for combining the results according to heterogeneity. Heterogeneity of the exposure effect was evaluated through I² statistic. An I² value 80% or greater indicated high heterogeneity and thus the statistical analysis should not be considered as trustworthy, while an I² value 60% or greater represented significant heterogeneity, leading to the employment of the random effects model. A chi-squared test for heterogeneity was also conducted and the P values are provided. All statistical analysis was performed using the Review Manager (RevMan) software (built 5.3).

Results

Comparison between FPS and LPS regarding IVF/ICSI outcome for patients presenting with a prognosis of POR

Nine studies investigating the efficacy of LPS in patients with a prognosis of POR met the inclusion criteria for this systematic review [6, 7, 9, 10, 16, 19–22]. Four out of nine studies performed a prospective data analysis [6, 9, 10, 16] and the other five were of retrospective nature [7, 19–22]. Data extraction was performed, as presented in the “Materials and methods” section, in order to provide information regarding the studies’ general characteristics and outcomes obtained following FPS and LPS, respectively.

Studies’ general characteristics

Great variation was observed among the studies included in this systematic review regarding the inclusion criteria employed respectively for the selection of the studies’ participants. Four studies included participants characterized as POR patients according to the Bologna criteria [16, 19, 21, 22]. In the other five, different inclusion criteria were employed. In the prospective study of Kuang et al., patients had to meet at least two out of the following five criteria: advanced maternal age defined as over 40 years of age; a history of ovarian surgery; previous IVF attempts using conventional COS protocols that resulted to less than three oocytes; antral follicle count (AFC) of less than five on menstrual cycle days 2–3; and basal serum FSH levels between 10 and 19 IU/L [6]. In the prospective study of Ubaldi et al., patients included had been submitted to preimplantation genetic diagnosis for aneuploidy testing (PGD-A) and had presented with a medical history including: anti-Müllerian hormone (AMH) levels of less than 1.5 ng/mL, AFC of less than six follicles, and/or less than five oocytes retrieved in previous IVF attempts [9]. In the retrospective study of Liu et al., all enrolled patients were 38 years or older, were presenting with normal menstruation, and had at least one follicle 6–11 mm in diameter observed during oocyte retrieval following FPS [20]. In the retrospective study of Rashtian et al., patients had to meet the following inclusion criteria: basal serum FSH levels > 15 IU/m on menstrual cycle day 3; total AFC of less than eight; and at least one failed IVF attempt following conventional COS [7]. In the prospective study of Vaiarelli et al., patients had to meet at least two out of the following four criteria: advanced maternal age defined as over 35 years of age; AMH levels of less than 1.5 ng/mL; AFC of less than six follicles; previous IVF attempts yielding less than five metaphase II (MII) oocytes [10]. Regarding the sample size of this systematic review, a total of 1026 and 988 patients were enrolled in the FPS group and LPS group, respectively. In the FPS group, the number of patients enrolled ranged from 38 to 353, and in the LPS group, the number of patients enrolled ranged from 30 to 366. Data regarding the studies’ general characteristics is presented in Table 1. Assessment of bias was performed and the results are presented in Figs. 2 and 3.

Table 1.

Studies’ general characteristics as obtained following data extraction regarding FPS and LPS respectively

Studies	Type of study	Infertility etiology	Number of patients
Studies	Type of study	Infertility etiology	FPS	LPS
Kuang et al. 2014 [6]	Prospective	Mixed Population^a	38	30
Ubaldi et al. 2016 [9]	Prospective	Mixed Population^a	43	43
Liu et al. 2017 [20]	Retrospective	Mixed Population^a	116	116
Zhang et al. 2017 [21]	Retrospective	POR^b	153	153
Rashtian et al. 2017 [7]	Retrospective	Mixed Population^a	69	69
Jin et al. 2018 [19]	Retrospective	POR^b	72	76
Vaiarelli et al. 2018 [10]	Prospective	Mixed Population^a	353	336
Zhang et al. 2018 [22]	Retrospective	POR^b	61	61
Madani et al., 2019 [16]	Prospective	POR^b	121	104

Open in a new tab

FPS follicular phase stimulation, LPS luteal phase stimulation

^aMixed population including patients with a prognosis of poor ovarian response

^bPOR population including patients diagnosed with POR according to the Bologna criteria

Fig. 2 — Assessment of risk of bias of studies included in the systematic review regarding FPS vs LPS arm

Fig. 3 — Summary of risk of bias assessment regarding each item for each study included in the systematic review regarding FPS vs LPS arm

Stimulation protocols employed

Heterogeneity was also observed among the studies regarding the stimulation protocols employed for both FPS and LPS, respectively. In the great majority of the studies, FPS was achieved via the employment of mild and minimal stimulation protocols based on citrate clomiphene (CC) and letrozole administration. In some of the studies, exogenous gonadotropin administration was also employed in order to further support follicular development. Human menopausal gonadotropin (HMG) and recombinant follicle-stimulating hormone (rFSH) and luteinizing hormone (rLH) are presented to have been used. In some of the studies, gonadotropin-releasing hormone (GnRH) antagonist was used as a part of the standard protocol employed and in some others was used only if was necessary in order to prevent early LH surge. Regarding LPS, similar protocols to FPS were used. In order to induce ovulation, different approaches among the studies were used including administration of triptorelin, buserelin, recombinant human chorionic gonadotropin (rHCG), GnRH agonist, and GnRH antagonist. Extracted data regarding the stimulation protocols employed are presented in Table 2. Seven studies reported results comparing the total dose of exogenous gonadotropins required for FPS and LPS, respectively [6, 7, 16, 19–22]. In three out of seven studies, a statistically significant higher dose of HMG was required for LPS in comparison to FPS [6, 7, 16]. Similarly, in the study of Zhang et al., a higher dose of gonadotropins was required for LPS in comparison to FPS. However, three studies reported no statistically significant difference between the two groups regarding the dose of exogenous gonadotropins required for stimulation in each phase [7, 19, 20]. Variation was also observed among the studies regarding the initiation time of LPS protocol. Initiation of LPS protocol ranges among the studies between 0 and 7 days following first oocyte retrieval. Moreover, different criteria were used among the studies regarding the number and the size of pre-existing small follicles required for LPS. This size ranges among the studies between 2 and 11 mm. The aforementioned data are presented in Table 2.

Table 2.

Extracted data regarding the stimulation protocols employed for FPS and LPS respectively

Studies	Stimulation protocols employed
Studies	FPS	First trigger	Pre-existing follicles required for LPS	Initiation of LPS (days following 1st OR)	LPS	Second trigger
Kuang et al. 2014 [6]	CC, Let, HMG	GnRH-ag, Ibup	At least 2 antral follicles 2–8 mm	0–1 days	Let, HMG, Medrox	GnRH-ag, Ibup
Ubaldi et al. 2016 [9]	rFSH, rLH, GnRH-antag	Bus	NA	5 days	rFSH, rLH, GnRH-antag	Bus
Liu et al. 2017 [20]	rFSH, rLH	rHCG	At least one antral follicle 6–11 mm	1–3 days	HMG	rHCG
Zhang et al. 2017 [21]	CC, Gon	GnRH-ag	Antral follicles 5–10 mm	1 day	Gon	rHCG
Rashtian et al. 2017 [7]	Let, CC, Gon, GnRH-antag	GnRH-ag	At least 1 antral follicle < 11 mm	1 day	Let, CC, Gon	rHCG
Jin et al. 2018 [19]	Let or CC, HMG, GnRH-antag	GnRH-ag	NA	1–3 days	CC, HMG	rHCG
Vaiarelli et al. 2018 [10]	E2 valerate, rFSH, rLH, GnRH-antag	Bus	NA	5 days	rFSH, rLH, GnRH-antag	Bus
Zhang et al. 2018 [22]	CC, HMG	rHCG	At least 2 antral follicles 2–8 mm	2–7 days	CC, HMG	rHCG
Madani et al. 2019 [16]	CC, Let, HMG	Bus, Ibup	At least 2 antral follicles 2–8 mm	1 day	Let, HMG	Bus

Open in a new tab

FPS follicular phase stimulation, LPS luteal phase stimulation, OR oocyte retrieval, CC clomiphene citrate, HMG human menopausal gonadotropin, rFSH recombinant FSH, rLH recombinant LH, rHCG recombinant human chorionic gonadotropin, Gon exogenous gonadotropins, GnRH-ag GnRH agonist, GnRH-antag GnRH antagonist, Bus buserelin, Ibup ibuprofen, Let letrozole, Medrox medroxyprogesterone acetate, NA not applicable

Stimulation days

Regarding the length of the stimulation period required for FPS and LPS, respectively, eight studies reported results comparing the days of stimulation for each group [6, 7, 9, 10, 16, 19–22]. Stimulation duration ranged from 3.9 to 10.2 days during FPS and from 8.7 to 10.8 days during LPS. Only two studies reported a statistically significant higher number of stimulation days during LPS [21, 22] compared to FPS. The remaining six studies did not report any statistically significant difference. Extracted data regarding days of stimulation are presented in Table 3. Seven studies were included in the statistical analysis comparing days of stimulation between FPS and LPS. In the study of Zhang et al., data provided regarding the stimulation period refer only in the duration of gonadotropin administration and not in the duration of the total stimulation protocol, and thus, this study was excluded from the analysis [21]. Total heterogeneity was statistically significant (I² = 60%, P value < 0.02); thus, the random effects model was employed. Pooled data analysis failed to reveal any statistically significant difference (MD = − 0.06, 95%CI = − 0.54, 0.43) between FPS and LPS regarding days of stimulation. Following subgroup analysis, no statistically significant difference was observed in any of the two subgroups. Statistical analysis is presented in Fig. 4.

Table 3.

Extracted data regarding duration of stimulation and cycle cancelation rate for FPS and LPS respectively

Studies	Stimulation days		Cycle cancelation rate
Studies	FPS	LPS	FPS	LPS
Kuang et al. 2014 [6]	10.2 ± 2.4	10.8 ± 3.1	20/38 (52.6%)	13/30 (43.3%)
Ubaldi et al. 2016 [9]	9.6 ± 2.4	10.3 ± 2.5	4/43 (9.3%)	3/43 (6.9%)
Liu et al. 2017 [20]	8.3 ± 3.5	7.5 ± 3.4	43/116 (37.1%)	33/116 (28.4%)
Zhang et al. 2017 [21]	3.9 ± 3.2	8.7 ± 3.5	19/153 (12.4%)	5/153 (3.3%)
Rashtian et al. 2017 [7]	8.3 ± 0.6	7.9 ± 0.5	ΝΑ	ΝΑ
Jin et al. 2018 [19]	7.5 ± 2.4	8.3 ± 4.0	25/72 (34.7%)	21/76 (27.6%)
Vaiarelli et al. 2018 [10]	NA	NA	17/353 (4.8%)	26/336 (7.7%)
Zhang et al. 2018 [22]	6.9 ± 4.1	7.9 ± 4.0	NA	NA
Madani et al. 2019 [16]	8.0 ± 2.6	7.8 ± 2.8	12/121 (9.9%)	23/104 (22.1%)

Open in a new tab

FPS follicular phase stimulation, LPS luteal phase stimulation, NA not applicable

Fig. 4 — Forest plot of days of stimulation resulting from the comparison of FPS and LPS respectively

Cycle cancelation rate

Seven studies reported results comparing the cycle cancelation rate for each phase [6, 9, 10, 16, 19–21]. Four were of prospective nature and the other three of retrospective. Cycle cancelation rate ranged from 4.8 to 52.6% during FPS and from 3.3 to 43.3% during LPS. No statistically significant difference was observed between the two stimulation phases in any of the studies regarding cycle cancelation rate. Extracted data regarding cycle cancelation rate are presented in Table 3. Total heterogeneity was statistically significant (I² = 69%, P value = 0.002); thus, the random effects model was employed. Pooled data analysis failed to reveal any statistically significant difference (OR = 1.16, 95%CI = 0.67, 2.00) between FPS and LPS regarding the cycle cancelation rate. Following subgroup analysis, no statistically significant difference was observed in any of the two subgroups. Statistical analysis is presented in Fig. 5.

Fig. 5 — Forest plot of cycle cancelation rate resulting from the comparison of FPS and LPS respectively

Number of oocytes retrieved

Eight studies reported results comparing the number of oocytes retrieved in each phase [6, 7, 9, 16, 19–22]. Three were of prospective nature and the other five of retrospective. Mean number of oocytes retrieved ranged from 1.5 to 5.1 in FPS and from 1.3 to 5.7 in LPS. One study reported results as a median and range [19]. Five studies reported a statistically significant higher oocyte yield during LPS [6, 19–22]. The remaining three studies presented with no statistically significant differences regarding the number of oocytes retrieved [7, 9, 16]. Extracted data regarding number of oocytes retrieved are presented in Table 4. Seven studies were included in the statistical analysis comparing the number of oocytes retrieved between FPS and LPS. The study in which results reported as a median and range was excluded from the statistical analysis [19]. Total heterogeneity was statistically significant (I² = 76%, P value = 0.0003); thus, the random effects model was employed. Pooled data analysis revealed a statistically significant higher mean number of retrieved oocytes following LPS in comparison to FPS (MD = − 0.63, 95%CI = − 0.97, − 0.28). Following subgroup analysis, a statistically significant higher mean number of retrieved oocytes was observed following LPS in comparison to FPS (MD = − 0.88, 95%CI = − 1.63, − 0.14) in the mixed population subgroup. In the POR subgroup, a higher number of retrieved oocytes was also observed following LPS, but a statistically significant difference could not be established marginally (MD = − 0.52, 95%CI = − 1.10, 0.05). Statistical analysis is presented in Fig. 6.

Table 4.

Extracted data regarding the number of oocytes retrieved and the number of MII oocytes obtained following FPS and LPS respectively

Studies	Number of oocytes retrieved		Number of MII oocytes		MII oocyte rate^a
Studies	FPS	LPS	FPS	LPS	FPS	LPS
Kuang et al. 2014 [6]	1.7 ± 1.0	3.5 ± 3.2	1.4 ± 1.0	2.7 ± 2.7	53/62 (85.5%)	82/105 (78.1%)
Ubaldi et al. 2016 [9]	5.1 ± 3.4	5.7 ± 3.3	3.4 ± 1.9	4.1 ± 2.5	ΝΑ	ΝΑ
Liu et al. 2017 [20]	2.3 ± 2.0	3.5 ± 3.5	1.9 ± 1.7	2.8 ± 3.1	ΝΑ	ΝΑ
Zhang et al. 2017 [21]	2.2 ± 1.6	3.3 ± 2.6	2.03 ± 1.53	3.16 ± 2.55	311/333 (93.4%)	483/499 (96.8%)
Rashtian et al. 2017 [7]	1.6 ± 0.2	1.9 ± 0.2	1.6 ± 0.2	1.6 ± 0.2	ΝΑ	ΝΑ
Jin et al. 2018 [19]	1 (1–2)	2 (1–4)	ΝΑ	ΝΑ	25/35 (71.4%)	46/58 (79.3%)
Vaiarelli et al. 2018 [10]	NA	NA	4.0 ± 2.5	4.7 ± 3.0	ΝΑ	ΝΑ
Zhang et al. 2018 [22]	1.3 ± 0.9	1.8 ± 1.1	NA	NA	325/416 (78.1%)	374/551 (67.9%)
Madani et al. 2019 [16]	1.5 ± 1.1	1.5 ± 2	1.4 ± 1.0	1.2 ± 1.6	NA	NA

Open in a new tab

FPS follicular phase stimulation, LPS luteal phase stimulation, NA not applicable

^aMII oocytes per number of total oocytes obtained following oocyte retrieval

Fig. 6 — Forest plot of number of oocytes retrieved resulting from the comparison of FPS and LPS respectively

Number of metaphase II oocytes

Nine studies reported results regarding the number of MII oocytes retrieved in each phase. Four out of nine studies included performed a prospective data analysis [6, 9, 10, 16] and the other five were of retrospective nature [7, 19–22]. Five studies presented their results in the format of mean number of MII oocytes retrieved, two studies in the format of MII retrieval rate and two studies in both formats. Mean number of MII oocytes retrieved ranged from 1.4 to 4.0 regarding FPS and from 1.2 to 4.7 regarding LPS. MII oocyte rate per number of total oocytes obtained following oocyte retrieval ranged from 71.4 to 93.4% regarding FPS and from 67.9 to 96.8% regarding LPS. Four out of nine studies reported a statistically significant higher number of MII oocytes retrieved during LPS [6, 10, 20, 21], and one reported a statistically significant higher number of MII oocytes retrieved during FPS [22], whereas the remaining four reported no statistically significant difference [7, 9, 16, 19]. Extracted data regarding number of MII oocytes obtained are presented in Table 4. Seven studies included in the statistical analysis comparing the mean number of MII oocytes retrieved between FPS and LPS. Total heterogeneity was statistically significant (I² = 88%, P value < 0.001); thus, the random effects model was employed. Pooled data analysis revealed a statistically significant higher mean number of MII oocytes retrieved following LPS in comparison to FPS (MD = − 0.58, 95%CI = − 1.02, − 0.14). Following subgroup analysis, a statistically significant higher mean number of retrieved MII oocytes was observed following LPS in comparison to FPS (MD = − 0.63, 95%CI = 1.18, − 0.09) in the mixed population subgroup. In the POR subgroup, no statistically significant difference was established (Fig. 7a). Four studies included in the statistical analysis comparing the MII oocyte rate between FPS and LPS. Total heterogeneity was statistically significant (I² = 79%, P value = 0.002); thus, the random effects model was employed. Pooled data analysis failed to reveal any statistically significant difference (OR = 1.00, 95%CI = 0.49, 2.02) between FPS and LPS regarding the MII oocyte rate. Following subgroup analysis, no statistically significant difference was observed in any of the two subgroups (Fig. 7b). Statistical analysis is presented in Fig. 7.

Fig. 7 — Forest plot regarding the number of MII oocytes obtained resulting from the comparison of FPS and LPS respectively. a Comparison of mean difference of MII oocytes. b Comparison of MII oocyte rate

Embryology data

Great variation was observed among the studies regarding the criteria used to evaluate the embryology data obtained following FPS and LPS respectively. In the study of Kuang et al., no statistically significant difference was established between the two phases regarding the number of top-quality embryos, the number of cryopreserved embryos, and the cleavage rate defined as the number of fertilized zygotes reaching cleavage stage. In the same study, a statistically significant higher number of cleaved embryos was observed in the LPS group probably due to the higher number of fertilized oocytes obtained in this group [6]. In the study of Ubaldi et al., no statistically significant difference was observed between FPS and LPS in regard to the number of fertilized oocytes, the number of biopsied blastocysts, the number of euploid blastocysts, the blastocyst quality, the type of aneuploidy, and the day of blastocyst formation [9]. In the study of Liu et al., a statistically significant increased number of fertilized oocytes, number of cleaved embryos, and number of top-quality embryos were observed in the LPS group. No statistically significant difference could be established in regard to the top-quality embryos rate, the embryo survival rate, and the number of cryopreserved embryos between the two stimulation phases [20]. A statistically significant higher mean number of 2 pro-nuclei (2PN) zygotes and mean number of good-quality embryos was observed in the study of Zhang et al. in the LPS group [21]. In the study of Vaiarelli et al., a higher number of fertilized oocytes and blastocysts were observed in the LPS group as a result of the higher number of MII oocytes collected in this group. No statistically significant difference was observed between the two phases in regard to the mean number of euploid blastocysts obtained. The mean fertilization, blastocyst, and euploid blastocyst rates, calculated per number of MII oocytes collected, did not differ significantly between the two phases. Similarly, the overall fertilization, blastocyst, and euploid blastocyst rates of the MII oocytes obtained following FPS and LPS, respectively, did not differ significantly [10]. No statistically significant difference was observed between FPS and LPS in terms of the cleavage rate and the top-quality embryo rate in the study of Zhang et al. [22].

Fertilization rate

Seven studies reported results regarding the fertilization rate observed in each phase. Fertilization rate is defined as the number of 2PN zygotes obtained per inseminated oocytes. Three out of seven studies included performed a prospective data analysis [6, 9, 10] and the other four were of retrospective nature [19–22]. Fertilization rate ranged from 69.7 to 86.9% regarding FPS and from 73.1 to 86.5% regarding LPS. No statistically significant difference was observed between FPS and LPS in neither of the studies regarding fertilization rate. Extracted data regarding fertilization rate observed among the studies are presented in Table 5. Seven studies included in the statistical analysis comparing fertilization rate between FPS and LPS. Total heterogeneity was not statistically significant (I² = 48%, P value = 0.07); thus, the fixed effect model was employed. Pooled data analysis failed to reveal any statistically significant difference (OR = 1.08, 95%CI = 0.95, 1.24) between FPS and LPS regarding fertilization rate. Following subgroup analysis, no statistically significant difference was observed in any of the two subgroups. Statistical analysis is presented in Fig. 8.

Table 5.

Extracted data regarding fertilization rate and clinical pregnancy rate observed following FPS and LPS respectively

Studies	Fertilization rate^a		Clinical pregnancy rate
Studies	FPS	LPS	FPS	LPS
Kuang et al. 2014 [6]	37/53 (69.8%)	62/82 (75.6%)	8/13 (61.5%)	5/7 (71.4%)
Ubaldi et al. 2016 [9]	99/142 (69.7%)	136/173 (78.6%)	6/7 (85.7%)	6/8 (75%)
Liu et al., 2017 [20]	199/270 (73.7%)	297/406 (73.1%)	4/16 (25%)	7/34 (20.6%)
Zhang et al. 2017 [21]	236/311 (75.8%)	353/483 (73.1%)	3/28 (10.7%)	14/36 (38.9%)
Rashtian et al. 2017 [7]	ΝΑ	ΝΑ	ΝΑ	ΝΑ
Jin et al. 2018 [19]	86/108 (79.6%)	161/186 (86.5%)	7/20 (35.0%)	12/32 (37.5%)
Vaiarelli et al. 2018 [10]	878/1129 (71.4%)	1072/1442 (74.3%)	36/81 (44.4%)	45/83 (54.2%)
Zhang et al. 2018 [22]	86/99 (86.9%)	51/65 (78.5%)	4/29 (13.8%)	3/14 (21.4%)
Madani et al. 2019 [16]	NA	NA	NA	NA

Open in a new tab

FPS follicular phase stimulation, LPS luteal phase stimulation, NA not applicable

^a:Fertilization rate defined as the number of 2PN zygotes obtained per inseminated oocytes

Fig. 8 — Forest plot of fertilization rate resulting from the comparison of FPS and LPS respectively

Clinical pregnancy rate

Seven studies reported results regarding the clinical pregnancy rate (CP) observed in each phase. Three out of the seven studies included performed a prospective data analysis [6, 9, 10] and the other four were of retrospective nature [19–22]. CP rate ranged from 10.7 to 85.7% regarding FPS and from 20.6 to 75% regarding LPS. No statistically significant difference was observed between FPS and LPS in six of the studies regarding CP. In the study of Zhang et al., a statistically higher CP rate observed in the LPS phase [21]. Extracted data regarding CP rate observed among the studies are presented in Table 5. Seven studies were included in the statistical analysis comparing clinical pregnancy rate between FPS and LPS. Total heterogeneity was not statistically significant (I² = 0%, P value = 0.54); thus, the fixed effect model was employed. Pooled data analysis revealed a trend towards a statistically significant higher clinical pregnancy rate following LPS in comparison to FPS (OR = 0.65, 95%CI = 0.42, 1.00). Following subgroup analysis, a statistically significant higher clinical pregnancy rate was observed following LPS in comparison to FPS in the POR subgroup (OR = 0.46, 95%CI = 0.22, 0.99). In the mixed population subgroup, no statistically significant difference could be established between FPS and LPS. Statistical analysis is presented in Fig. 9.

Fig. 9 — Forest plot of clinical pregnancy rate resulting from the comparison of FPS and LPS respectively

Live birth rate/ongoing pregnancy rate

Seven studies reported results regarding the live birth rate/ongoing pregnancy rate (LB/OP) observed in each stimulation phase. Three out of the seven studies included performed a prospective data analysis [6, 9, 10] and the other four were of retrospective nature [19–22]. LB/OP rate ranged from 10.3 to 71.4% regarding FPS and from 14.3 to 62.5% regarding LPS. No statistically significant difference was observed between FPS and LPS in six of the studies regarding LB/OP. In the study of Zhang et al., a statistically higher LB/OP rate was observed in the LPS phase [21]. Extracted data regarding CP rate observed among the studies are presented in Table 6. Seven studies were included in the statistical analysis comparing LB/OP rate between FPS and LPS. Total heterogeneity was not statistically significant (I² = 0%, P value = 0.88); thus, the fixed effect model was employed. Pooled data analysis failed to reveal any statistically significant difference (OR = 1.08, 95%CI = 0.52, 2.25) between FPS and LPS regarding fertilization rate. Following subgroup analysis, no statistically significant difference was observed in any of the two subgroups. Statistical analysis is presented in Fig. 10.

Table 6.

Extracted data regarding live birth/ongoing pregnancy rate and miscarriage/early pregnancy loss rate observed following FPS and LPS respectively

Studies	Live birth/ongoing pregnancy rate		Miscarriage/early pregnancy loss rate
Studies	FPS	LPS	FPS	LPS
Kuang et al. 2014 [6]	7/13 (53.8%)	4/7 (57.1%)	1/8 (12.5%)	1/5 (20%)
Ubaldi et al. 2016 [9]	5/7 (71.4%)	5/8 (62.5%)	1/6 (16.7%)	1/6 (16.7%)
Liu et al. 2017 [20]	4/16 (25%)	6/34 (17.6%)	0/4 (0%)	1/7 (14.3%)
Zhang et al. 2017 [21]	3/28 (10.7%)	10/36 (27.7%)	0/3 (0%)	4/14 (28.6%)
Rashtian et al. 2017 [7]	ΝΑ	ΝΑ	ΝΑ	ΝΑ
Jin et al. 2018 [19]	7/20 (35.0%)	11/32 (34%)	0/20 (0%)	1/12 (8.3%)
Vaiarelli et al. 2018 [10]	32/81 (39.5%)	41/83 (49.4%)	4/36 (11.1%)	4/45 (8.9%)
Zhang et al. 2018 [22]	3/29 (10.3%)	2/14 (14.3%)	1/4 (25%)	0/3 (0%)
Madani et al. 2019 [16]	NA	NA	NA	NA

Open in a new tab

FPS follicular phase stimulation, LPS luteal phase stimulation, NA not applicable

Fig. 10 — Forest plot of live birth/ongoing pregnancy rate resulting from the comparison of FPS and LPS respectively

Miscarriage rate/early pregnancy loss

Seven studies reported results regarding the miscarriage rate/early pregnancy loss rate (M/EPL) observed in each stimulation phase. Three out of the seven studies included, performed a prospective data analysis [6, 9, 10] and the other four were of retrospective nature [19–22]. M/EPL rate ranged from 0 to 25% regarding FPS and from 0 to 20% regarding LPS. No statistically significant difference was observed between FPS and LPS in six of the studies regarding M/EPL. In the study of Zhang et al., a statistically higher M/EPL rate observed in the LPS phase [21]. Extracted data regarding M/EPL rate observed among the studies are presented in Table 6. Seven studies were included in the statistical analysis comparing M/EPL rate between FPS and LPS. Total heterogeneity was not statistically significant (I² = 0%, P value = 0.91); thus, the fixed effect model was employed. Pooled data analysis failed to reveal any statistically significant difference (OR = 0.81, 95%CI = 0.32, 2.05) between FPS and LPS regarding fertilization rate. Following subgroup analysis, no statistically significant difference was observed in any of the two subgroups. Statistical analysis is presented in Fig. 11.

Fig. 11 — Forest plot of miscarriage/early pregnancy loss rate resulting from the comparison of FPS and LPS respectively

Number of cycles leading to no oocytes retrieved

Five studies reported on number of cycles leading to no oocytes retrieved for each stimulation phase. Four out of seven studies included performed a prospective data analysis [6, 9, 10, 16] and the last one was of retrospective nature [16]. In the study of Zhang et al., a statistically significant higher number of cycles leading to no oocytes retrieved was observed in the FPS phase in comparison to LPS phase (12.4% vs 3.3%, respectively) [21]. In the study of Madani et al., a statistically significant higher number of cycles leading to no oocytes retrieved was observed in the LPS phase in comparison to FPS phase (16.3% vs 10.7%, respectively) [16]. No statistically significant difference was reported in neither of the other three studies regarding the number of cycles leading to no oocytes retrieved.

Comparison between DuoStim and ConStim regarding IVF/ICSI outcome for patients presenting with a prognosis of POR

Five studies investigating the efficacy of double ovarian stimulation (DuoStim) in patients with a prognosis of POR, met the inclusion criteria and were included in this systematic review [17–20, 22]. Four out of the five studies included performed a retrospective data analysis [18–20, 22] and the last one was of prospective nature [17]. Data extraction was performed, as presented in the “Materials and methods” section, in order to provide information regarding the studies’ general characteristics and outcomes obtained following DuoStim and ConStim, respectively.