Clinical Outcomes of Postponing Oocyte Retrieval Beyond 40 h After Trigger With Diclofenac Pretreatment Compared With the Routine Paradigm: A Retrospective Propensity Score‐Matched Study

ABSTRACT

Purpose

This study aimed to compare outcomes of delaying oocyte retrieval beyond 40 h post‐trigger with diclofenac versus the conventional workflow.

Methods

A total of 4912 cycles were retrospectively screened. Eighty‐four delayed cycles and 168 routine cycles from patients who received a single trigger in their first cycle were matched at a 1:2 ratio for age, AMH, COS protocol, and ICSI use and compared.

Results

The ratio of the number of oocytes retrieved to the number of follicles ≥ 14 mm on trigger day was significantly higher in the delayed group (1.27 ± 0.42 vs. 1.14 ± 0.47, p = 0.031). The rates of maturation (78.9% vs. 78.8%), fertilization (64.0% vs. 64.6%), and blastulation (63.8% vs. 68.5%) were comparable (all p > 0.05). Biochemical pregnancy rates in fresh cycles were numerically lower in the delayed group (38.4% vs. 47.0%), while clinical pregnancy rates were similar between groups in both fresh (38.4% vs. 41.1%) and frozen–thawed cycles (51.2% vs. 52.8%) (all p > 0.05).

Conclusions

Delaying oocyte retrieval beyond 40 h post‐trigger with diclofenac could be feasible to enhance clinical operational flexibility. Further research is warranted to identify subgroups that may benefit most and to optimize strategies adapted to this extension.

Abbreviations

2PN

Two Pronuclei

ADAMTS1

A Disintegrin And Metalloproteinase with Thrombospondin Motifs 1

AMH

Anti‐Müllerian Hormone

AREG

amphiregulin

ART

Assisted Reproductive Technology

BMI

body mass index

COC

cumulus–oocyte complex

COS

controlled ovarian stimulation

COX‐2

cyclooxygenase‐2

EGF

epidermal growth factor

EREG

epiregulin

FSH

follicle‐stimulating hormone

GLM

Generalized Linear Model

GnRH

gonadotropin‐releasing hormone

hCG

human chorionic gonadotropin

ICSI

intra‐cytoplasmic sperm injection

IVM

in vitro maturation

LH

luteinizing hormone

MII

metaphase II

MMPs

matrix metalloproteinases

NSAIDs

nonsteroidal anti‐inflammatory drugs

OPU

oocyte pick‐up

PGE2

prostaglandin E2

PPOS

progestin‐primed ovarian stimulation

PSM

propensity score matching

SMD

standardized mean difference

TORI

trigger‐to‐oocyte retrieval interval

1. Introduction

In physiological conditions, ovulation is triggered by the luteinizing hormone (LH) surge, which initiates a cascade of morphological and functional alterations in the dominant follicle. During this process, the oocyte resumes and completes the first meiotic division, arresting at metaphase II (MII) phase [1]. The essential role of prostaglandins in ovulation mechanism has long been established [2]. Under LH stimulation, prostaglandin synthesis is driven by epidermal growth factor (EGF)‐like mediators such as amphiregulin (AREG) and epiregulin (EREG), which upregulate cyclooxygenase‐2 (COX‐2) and increase prostaglandin E2 (PGE2) production. PGE2, acting through EP2/EP4 signaling, orchestrates gene expression patterns underlying oocyte maturation, cumulus‐oocyte complex (COC) expansion, follicular rupture, and luteinization‐related processes [3]. In routine assisted reproductive technology (ART) practice, exogenous human chorionic gonadotropin (hCG) is administered to mimic the endogenous LH surge once leading follicles reach 18–20 mm in diameter. Building on this established physiological sequence, the clinical paradigm typically schedules oocyte pick‐up (OPU) at 35–38 h after the hCG trigger, depending on the ovarian stimulation regimen [4, 5]. This standardized timeframe aims to maximize the retrieval of mature oocytes while minimizing the risks of spontaneous ovulation and oocyte postmaturity (i.e., aging).

In controlled ovarian stimulation (COS), an antecedent LH surge may precipitate spontaneous ovulation, typically occurring when LH levels rise by more than 20 mIU/mL within 24 h, which is associated with increased cycle cancelation rates [6]. Nonsteroidal anti‐inflammatory drugs (NSAIDs), which inhibit cyclooxygenase (COX) activity and downstream prostaglandin synthesis, have been investigated for their potential to postpone ovulation. As early as 1996, Nargund et al. reported that continuous administration of indomethacin from cycle Day 10 of a natural cycle effectively delayed ovulation by approximately one week, ultimately enabling planned conception and a live birth [7]. Subsequently, researchers explored the application of NSAIDS as preventive or rescue strategies to suppress premature LH surges and premature ovulation in natural cycles [8, 9]. The effectiveness of NSAIDs as strategies to secure a controlled retrieval window has been supported by these investigations.

Recently, a prospective controlled trial demonstrated that ibuprofen could safely delay ovulation in natural cycles, permitting OPU at 42 h post‐hCG without compromising oocyte maturity or fertilization competence [10], thereby suggesting the feasibility of a more flexible retrieval schedule beyond the conventional time frame under certain pharmacological intervention. Further intriguing insights emerged from an anecdotal report described by Ben‐Shlomo et al., in which OPU was accidentally performed 47 h after hCG administration due to a patient misunderstanding. Notably, most follicles remained intact, yielding a high proportion of MII oocytes and resulting in an ongoing pregnancy. This observation challenged the prevailing paradigm of oocyte retrieval timing in multi‐follicular environments [11]. The authors further posited that prolonging the in vivo incubation may be potentially superior to extended in vitro culture for cytoplasmic maturation, given the sustained exposure to intrafollicular signaling [11, 12]. However, these findings remain limited by the paucity of systematic and reproducible studies.

Despite these advances, current research on the optimal trigger‐to‐oocyte‐retrieval‐interval (TORI) has largely examined the modest deviation within the routine 36–38‐h window [13]. A significant evidence gap exists regarding the feasibility and safety of a planned delay of OPU beyond 40 h, achieved through pharmacological intervention to avoid spontaneous ovulation. To address this gap, this study sought to evaluate the clinical feasibility of delaying OPU to 40 h following diclofenac pretreatment, by comparing laboratory and clinical outcomes with those obtained from OPU within the conventional 36–38 h window. The findings are expected to provide preliminary evidence supporting strategies that enhance operational flexibility in ART practice.

2. Materials and Methods

2.1. Study Design and Subjects

This retrospective study included ovarian stimulation cycles conducted at the Reproductive Medicine Center of the Sixth Affiliated Hospital of Sun Yat‐sen University between January 2025 and October 2025. Cycles were excluded if they involved: (1) fertility preservation; (2) in vitro maturation; (3) oocyte donation; (4) lack of trigger injection or with trigger time undefined; (5) canceled OPU due to premature ovulation. Patients aged 20–49 years who underwent trigger injection and oocyte retrieval were classified based on TORI and whether diclofenac was taken: those with TORI within 36–38 h after trigger were assigned to the routine group, and those with TORI within 40–45 h after diclofenac pretreatment (planned postponement) were assigned to the delayed group. Cycles with a TORI outside both the 36–38 h and 40–45 h ranges, as well as those with a TORI within 36–38 h but receiving diclofenac, were not targeted in this study. Cycles that did not undergo insemination after oocyte retrieval and thus adopted oocyte cryopreservation were further excluded. Only each patient's first cycle with a single trigger was included in the propensity score‐matched (PSM) cohort analysis to avoid the potential influence of decision making based on anticipated stimulation performance and treatment outcomes (Figure 1). In the PSM cohort, planned postponement of the OPU procedure in the delayed group was mainly determined by clinical resource allocation and patients' scheduling considerations. Patients' electronic medical records, surgical records and laboratory records were reviewed and analyzed including demographic characteristics, ovarian reserve parameters, ovarian stimulation variables, hormone level and the outcomes of oocyte retrieval, insemination, embryo development, and pregnancy outcomes. The study was approved by the Ethics Committee of the Sixth Affiliated Hospital of Sun Yat‐sen University (approval number: 2026ZSLYEC‐009).

FIGURE 1.

The flowchart of the study selection process.

2.2. Ovarian Stimulation, Insemination and Embryo Culture

Ovarian stimulation, insemination and embryo culture followed routine clinical protocols implemented in practice and were executed in accordance with the described methodologies in our prior studies [14, 15], unless otherwise specified. Embryos were morphologically evaluated on Day 3 and graded based on the cell number, uniformity and fragmentation according to the grading criteria modified from Racowsky et al. [16]. Four‐cell embryos exhibiting uniform blastomeres and no fragmentation (Grade 1) and embryos with more than four cells showing uniform or mildly uneven blastomere size with fragmentation below 20% (Grades 1–2) were defined as usable embryos. Good‐quality embryos were defined as usable embryos with 6–10 cells. The decision to extend culture for the Day 3 embryos that were neither transferred nor cryopreserved until the fully expanded blastocyst formed on Day 5 or Day 6 was made based on clinical indication and contingent upon the patient's informed consent. Blastocysts were morphologically evaluated and scored according to the Gardner criteria [17]. The stage of expansion was graded from 1 (early blastocyst, blastocoel less than half of total embryo volume) to 6 (hatched blastocyst, fully escaped from the zona pellucida). The inner cell mass (ICM) was graded from C (very few cells) to A (many cells, compacted and tight). The trophectoderm (TE) was graded from C (very few cells with poor epithelium formation) to A (Many cells with a cohesive epithelium). Blastocysts reaching grade 3 and higher, with either ICM or TE reaching grade B and higher (i.e., ≥ 3bc or 3CB) were considered usable for transfer or cryopreservation. Blastocysts with an expansion grade of 3 or higher and both ICM and TE grades of at least B (i.e., ≥ 3BB) were defined as good‐quality.

2.3. Trigger and Oocyte Retrieval

Typically, when two or more follicles reached ≥ 18 mm, final oocyte maturation was triggered with 250 μg of recombinant human chorionic gonadotropin (r‐hCG) (OVIDREL, Merck Serono, Germany) or 0.1–0.2 mg triptorelin acetate (Decapeptyl, Ferring Pharmaceuticals, Germany) or both when a dual trigger was adopted. In the routine group, retrieval surgery was performed 36–38 h after trigger under transvaginal ultrasound guidance. For the planned delayed group, 75 mg of oral diclofenac was prescribed to be administered at approximately 3:00 PM on the day following the trigger, after screening for contraindications including NSAID allergy and gastrointestinal history.

2.4. Outcomes

Laboratory outcome variables for observation included the proportion of cycle without retrieved oocytes, the number of oocytes retrieved, the ratio of the number of oocyte retrieved to the number of follicle ≥ 14 mm on trigger day (values exceeding 1.0 indicated that additional oocytes were recovered from follicles below 14 mm), the number and rate of MII oocytes, the number and rate of two pronuclei (2PN) zygotes, the number and rate of cleaved 2PN zygotes, the number and rate of Day 3 usable embryo, the number and rate of Day 3 good‐quality embryo, the number of blastocysts cultured and formed as well as blastulation rate. MII rate was calculated as the number of MII oocytes identified at ICSI divided by the number of oocytes retrieved in ICSI cycles. The 2PN fertilization rate was calculated as the number of 2PN zygotes divided by the number of oocytes retrieved. The 2PN cleavage rate was calculated as the number of cleaved embryos on Day 2 derived from 2PN zygotes divided by the number of 2PN. The Day 3 usable embryo rate or Day 3 good‐quality embryo rate was calculated as the number of Day 3 usable or good‐quality embryos divided by the number of cleaved 2PN zygotes. The blastulation rate was calculated as the number of blastocysts formed (blastocysts of Grade 3 or higher on Day 5 or Day 6) divided by the number of embryos for extended culture (i.e., embryos not transferred or cryopreserved on Day 3 and thus subjected to continuous culture). The usable blastocyst rate was defined as the number of usable blastocysts divided by the number of blastocysts formed. Clinical outcome variables included biochemical and clinical pregnancy rates, defined as the cases of pregnancy diagnosed by the detection of a positive beta hCG in serum and by ultrasonographic visualization of one or more gestational sacs per embryo transfer cycles, respectively [18, 19].

2.5. Statistics

The Kolmogorov–Smirnov test was conducted to assess whether continuous data followed a normal distribution. Continuous variables were presented as mean (±standard deviation, SD), and comparisons between groups were performed using t‐test for normally distributed data or the Mann–Whitney U test for nonnormally distributed data. Categorical variables were presented as frequencies and percentages, and the chi‐square test or Fisher's exact test was used for difference comparisons. A two‐tailed p‐value < 0.05 was considered statistically significant. All statistical analyses were performed using R software (version 4.4.3, http://www.r‐project.org). To reduce confounding and enhance comparability between groups, PSM was employed using the Matchit package in R. The analytic sub‐population for outcome comparison was restricted to patients undergoing their first ovarian stimulation cycle who received a single trigger for final oocyte maturation. Propensity scores were estimated using a generalized linear model (GLM) incorporating age and anti‐Müllerian hormone (AMH) as continuous covariates. A 1:2 nearest‐neighbor matching algorithm without replacement was applied. Additionally, exact matching was enforced on the following categorical variables to ensure perfect balance within matched sets: COS protocols (PPOS, GnRH‐antagonist, or others), ICSI use (yes/no), age category (< 35 or ≥ 35 years), and ovarian reserve category (AMH < 1.2, 1.2–3.5, or > 3.5 ng/mL). Covariate balance before and after matching was assessed using standardized mean differences (SMDs), with SMDs < 0.1 considered indicative of adequate balance. All subsequent comparative analyses were performed on the matched cohort.

3. Results

Baseline characteristics for the total cohort (n = 4, 912 cycles) prior to matching are summarized in Table 1. The delayed OPU group (n = 424) differed significantly from the routine OPU group (n = 4488) in the TORI, with mean ± SD of 40.35 ± 0.61 h versus 37.34 ± 0.52 h, respectively (p < 0.001). In the routine group, the majority of cycles (2766/4488, 61.62%) had a TORI of approximately 37 h, whereas in the delayed group, the majority (318/424, 75.00%) had a TORI of around 40 h (Figure S2A). Notable disparities were also observed in several prognostic factors: patients in the delayed OPU group were significantly older (37.51 ± 4.93 vs. 36.03 ± 5.35 years, p < 0.001), and a higher proportion were aged ≥ 35 years (70.0% vs. 59.9%, p < 0.001). Ovarian reserve was poorer in the delayed OPU group, as reflected by lower AMH levels (1.57 ± 1.79 vs. 2.20 ± 2.65 ng/mL, p < 0.001) and a higher prevalence of a diminished ovarian reserve (AMH < 1.2 ng/mL) (59.4% vs. 46.9%, p < 0.001). The distributions of infertility types did not differ significantly between groups. The delayed OPU group had a markedly higher proportion of nonfirst stimulation cycles (75.0% vs. 44.5%, p < 0.001). A single trigger was more commonly employed than dual triggers in both groups (74.1% vs. 83.2%, p < 0.001). Significant differences were also observed in the distributions of COS protocols and ICSI utilization (all p < 0.001). These between‐group imbalances in the unmatched cohort underscored the necessity for PSM to ensure valid comparisons.

TABLE 1.

Baseline characteristics of cycles with delayed or routine OPU before PSM.

Variables	Delayed OPU	Routine OPU	p
Cycles (n)	424	4488	/
TORI (h)	40.35 ± 0.61	37.34 ± 0.52	< 0.001
Age (years)	37.51 ± 4.93	36.03 ± 5.35	< 0.001
Aged < 35 years (n/%)	127 (30.0%)	1799 (40.1%)	< 0.001
Aged ≥ 35 years (n/%)	297 (70.0%)	2689 (59.9%)
AMH (ng/mL)	1.57 ± 1.79	2.20 ± 2.65	< 0.001
AMH < 1.2 ng/mL (n/%)	252 (59.4%)	2103 (46.9%)	< 0.001
1.2 ≤ AMH ≤ 3.5 ng/mL (n/%)	119 (28.1%)	1499 (33.4%)
AMH > 3.5 ng/mL (n/%)	53 (12.5%)	886 (19.7%)
AFC	7.66 ± 6.27	9.52 ± 7.50	< 0.001
Type of infertility (n/%)
Primary infertility	169 (39.9%)	1969 (43.9%)	0.123
Secondary infertility	255 (60.1%)	2519 (56.1%)
Etiologies (n/%)
Female factors	223 (52.6%)	2324 (51.8%)	0.017
Male factors	12 (2.8%)	286 (6.4%)
Combined factors	168 (39.6%)	1716 (38.2%)
Unknown	21 (5.0%)	162 (3.6%)
First cycle (n/%)
Yes	106 (25.0%)	2489 (55.5%)	< 0.001
No	318 (75.0%)	1999 (44.5%)
Trigger strategies (n/%)
Single trigger	314 (74.1%)	3733 (83.2%)	< 0.001
Dual trigger	110 (25.9%)	755 (16.8%)
COS protocols (n/%)
PPOS	211 (49.8%)	1554 (34.6%)	< 0.001
GnRH‐Ant	63 (14.9%)	1458 (32.5%)
Others	150 (35.4%)	1476 (32.9%)
ICSI (n/%)
Yes	228 (53.8%)	1885 (42.0%)	< 0.001
No	196 (46.2%)	2603 (58.0%)

The 1:2 matching procedure yielded a well‐balanced comparative cohort comprising 84 delayed OPU cycles and 168 routine OPU cycles, all from patients undergoing their first ovarian stimulation cycle with a single trigger (Table 2). As intended by design, the TORI remained significantly longer in the Delayed OPU group (40.23 ± 0.46 vs. 37.44 ± 0.53 h, p < 0.001). In the routine group, the majority of cycles (90/168, 53.57%) had a TORI of approximately 37 h, whereas in the delayed group, the majority (60/84, 71.43%) had a TORI of around 40 h (Figure S2B). The matched cohort achieved adequate balance across key baseline variables. Specifically, age (35.55 ± 4.77 vs. 35.40 ± 4.89 years, p = 0.770) and AMH levels (1.72 ± 1.58 vs. 1.71 ± 1.54 ng/mL, p = 0.935) were highly similar between groups. No significant differences were observed in body mass index (BMI), baseline hormone profiles, total gonadotropin dose, stimulation duration, or hormone levels on the trigger day (all p > 0.05). The type of infertility and etiologies did not differ significantly between groups (p = 0.684 and p = 0.251, respectively). The distributions of age categories (< 35 or ≥ 35 years) and ovarian reserve categories (AMH < 1.2, 1.2–3.5, > 3.5 ng/mL), COS protocols (PPOS, GnRH‐antagonist, or others) and ICSI utilization were exactly identical between the matched cohorts, as enforced by the exact matching criteria. All SMDs were < 0.1 (Figure S1).

TABLE 2.

Baseline characteristics of the matched cohort (first cycle with single trigger).

Variables	Delayed OPU	Routine OPU	p
Cycles/patients (n)	84	168	/
TORI (h)	40.23 ± 0.46	37.44 ± 0.53	< 0.001
Age (years)	35.55 ± 4.77	35.40 ± 4.89	0.770
Aged < 35 years (n/%)	36 (42.9%)	72 (42.9%)	NS
Aged ≥ 35 years (n/%)	48 (57.1%)	96 (57.1%)
AMH (ng/mL)	1.72 ± 1.58	1.71 ± 1.54	0.935
AMH < 1.2 ng/mL (n/%)	43 (51.2%)	86 (51.2%)	NS
1.2 ≤ AMH ≤ 3.5 ng/mL (n/%)	26 (31.0%)	52 (31.0%)
AMH > 3.5 ng/mL (n/%)	15 (17.9%)	30 (17.9%)
AFC	7.65 ± 4.98	8.04 ± 5.90	0.966
BMI (kg/m2)	23.02 ± 3.56	23.44 ± 4.55	0.675
FSH (IU/L)	8.93 ± 4.41	8.51 ± 3.54	0.314
E2 (pg/mL)	54.44 ± 57.60	55.81 ± 80.61	0.707
LH (IU/L)	5.68 ± 2.81	5.46 ± 2.87	0.232
Gn dose (IU)	2528.19 ± 1094.62	2651.82 ± 1043.91	0.644
Gn days	10.08 ± 3.34	10.49 ± 3.16	0.679
E2 on trigger day (pg/mL)	2184.24 ± 1869.33	2345.64 ± 2088.52	0.621
LH on trigger day (IU/L)	3.65 ± 4.55	3.16 ± 3.66	0.485
P on trigger day (ng/mL)	4.36 ± 6.44	3.45 ± 6.42	0.396
Type of infertility (n/%)
Primary infertility	33 (39.3%)	72 (42.9%)	0.684
Secondary infertility	51 (60.7%)	96 (57.1%)
Etiologies (n/%)
Female factors	44 (52.4%)	105 (62.5%)	0.251
Male factors	4 (4.8%)	9 (5.4%)
Combined factors	29 (34.5%)	48 (28.6%)
Unknown	7 (8.3%)	6 (3.6%)
COS protocols (n/%)
PPOS	34 (40.5%)	68 (40.5%)	NS
GnRH‐Ant	7 (8.3%)	14 (8.3%)
Others	43 (51.2%)	86 (51.2%)
ICSI (n/%)
Yes	30 (35.7%)	60 (35.7%)	NS
No	54 (64.3%)	108 (64.3%)

Embryological and laboratory outcomes of the matched cohort were summarized in Table 3. The number of cycles without retrieved oocytes was 1 (1.19%) and 3 (1.79%) in the delayed and routine groups, respectively. Compared to the routine group, the delayed OPU group had a significantly higher ratio of the number of oocytes retrieved to the number of follicles ≥ 14 mm on trigger day (1.27 ± 0.42 vs. 1.14 ± 0.47, p = 0.031). The number of oocytes retrieved (9.36 ± 7.62 vs. 8.38 ± 6.27, p = 0.588) and the MII oocyte yield (7.47 ± 5.95 vs. 6.68 ± 5.01, p = 0.539) were numerically higher in the Delayed OPU group, though these differences did not reach statistical significance. A similar MII rate was observed (78.9% vs. 78.8%, p > 0.999). Fertilization outcomes were highly comparable, with a mean of 5.99 ± 5.45 2PN zygotes in the delayed OPU group versus 5.41 ± 4.60 in the routine OPU group (p = 0.631), and 2PN fertilization rates of 64.0% and 64.6%, respectively (p = 0.827). The number of cleaved 2PN embryos (5.89 ± 5.35 vs. 5.31 ± 4.55, p = 0.590) and the 2PN cleavage rates were similar between groups (98.4% vs. 98.1%, p = 0.864). Regarding subsequent embryo development indicators, absolute values tended to be numerically higher in the delayed OPU group, including the number of Day 3 usable embryos (4.75 ± 4.63 vs. 4.36 ± 4.01, p = 0.747), good‐quality embryos (3.71 ± 3.69 vs. 3.53 ± 3.58, p = 0.647), embryo for extended culture (4.01 ± 4.61 vs. 3.48 ± 3.81, p = 0.744), blastocysts formed (2.56 ± 3.41 vs. 2.39 ± 2.84, p = 0.794), usable blastocyst (2.38 ± 3.29 vs. 2.13 ± 2.52, p = 0.504). Distribution of blastocysts formed on Day 5 (36.3% vs. 36.9%) and Day 6 (63.7% vs. 63.1%) were similar in the two groups. The corresponding rates were numerically lower in the delayed OPU group: Day 3 usable embryo rate (80.6% vs. 82.2%, p = 0.516), Day 3 good‐quality embryo rate (63.0% vs. 66.5%, p = 0.217), and blastulation rate (63.8% vs. 68.5%, p = 0.161). The usable blastocyst rate was slightly higher in the delayed group (93.0% vs. 89.3%, p = 0.170). Subgroup analysis stratified by hourly intervals for the number of oocytes retrieved, MII rate (i.e., maturation rate), fertilization rate and blastulation rate was shown in Figure S2C–F presented as median and interquartile range (IQR).

TABLE 3.

Laboratory outcomes of the matched cohort (first cycle with single trigger).

Variables	Delayed OPU	Routine OPU	p
Cycles without retrieved oocytes (n/%)	1/84 (1.19%)	3/168 (1.79%)	0.673
No. of oocytes retrieved	9.36 ± 7.62	8.38 ± 6.27	0.588
No. of oocytes/no. of follicles ≥ 14 mm	1.27 ± 0.42	1.14 ± 0.47	0.031
No. of MII oocytes	7.47 ± 5.95	6.68 ± 5.01	0.539
MII rate (n/%)	224/284 (78.9%)	401/509 (78.8%)	> 0.999
No. of 2PN	5.99 ± 5.45	5.41 ± 4.60	0.631
2PN rate (n/%)	503/786 (64.0%)	909/1408 (64.6%)	0.827
No. of cleavaged 2PN	5.89 ± 5.35	5.31 ± 4.55	0.590
2PN cleavage rate (n/%)	495/503 (98.4%)	892/909 (98.1%)	0.864
No. of usable embryos	4.75 ± 4.63	4.36 ± 4.01	0.747
Day 3 usable embryo rate (n/%)	399/495 (80.6%)	733/892 (82.2%)	0.516
No. of good‐quality embryos	3.71 ± 3.69	3.53 ± 3.58	0.647
Day 3 good‐quality embryo rate (n/%)	312/495 (63.0%)	593/892 (66.5%)	0.217
No. of embryos for extended culture	4.01 ± 4.61	3.48 ± 3.81	0.744
No. of blastocysts formed	2.56 ± 3.41	2.39 ± 2.84	0.794
No of blastocysts formed on Day 5	78/215 (36.3%)	148/401 (36.9%)	0.947
No of blastocysts formed on Day 6	137/215 (63.7%)	253/401 (63.1%)
Blastulation rate (n/%)	215/337 (63.8%)	401/585 (68.5%)	0.161
No of usable blastocysts	2.38 ± 3.29	2.13 ± 2.52	0.504
Usable blastocyst rate (n/%)	200/215 (93.0%)	358/401 (89.3%)	0.170

Table 4 summarizes the clinical outcomes of embryo transfer cycles for embryos derived from the delayed and routine OPU procedures. In fresh embryo transfer cycles, there was no case of blastocyst transfer in the delayed group, whereas two blastocyst transfers were conducted in the routine group. The delayed group showed lower rates of biochemical pregnancy (38.4% vs. 47.0%, p = 0.472) and clinical pregnancy (38.4% vs. 41.1%, p = 0.818), although these differences were not statistically significant. For frozen–thawed embryo transfer cycles, 69.8% of transfers were with blastocysts in the delayed group, compared with 82.0% in the routine group. Pregnancy outcomes were highly comparable between the two groups, with no significant differences observed in biochemical pregnancy (60.4% vs. 60.6%, p = 0.982) or clinical pregnancy rates (51.2% vs. 52.8%, p = 0.859). Notably, no ectopic pregnancies were recorded in the delayed OPU group in either fresh or thawed transfer cycles.

TABLE 4.

Clinical outcomes of delayed or routine OPU‐derived embryo transfer cycles.

Variables	Delayed OPU	Routine OPU	p
Fresh embryo transfer
Cycles (n)	26	51	/
SET (n/%)	20 (76.9%)	32 (62.7%)	/
DET (n/%)	6 (23.1%)	19 (37.2%)
Transfer with cleavage‐stage embryo (n/%)	26 (100%)	49 (96.1%)	/
Transfer with blastocyst (n/%)	0	2 (3.9%)
Biochemical pregnancy rate (n/%)	10/26 (38.4%)	24/51 (47.0%)	0.472
Clinical pregnancy rate (n/%)	10/26 (38.4%)	21/51 (41.1%)	0.818
Ectopic pregnancy cases (n)	0	1	/
Frozen–thawed embryo transfer
Cycles (n)	43	89	/
SET (n/%)	29 (67.4%)	64 (71.9%)	/
DET (n/%)	14 (32.5%)	25 (28.1%)
Transfer with cleavage‐stage embryo (n/%)	13 (30.2%)	16 (18.0%)	/
Transfer with blastocyst (n/%)	30 (69.8%)	73 (82.0%)
Biochemical pregnancy rate (n/%)	26/43 (60.4%)	54/89 (60.6%)	0.982
Clinical pregnancy rate (n/%)	22/43 (51.2%)	47/89 (52.8%)	0.859
Ectopic pregnancy cases (n)	0	0	/

4. Discussion

This study employed a propensity score matching approach to construct a highly homogeneous cohort, aiming to evaluate the clinical feasibility and potential benefits of extending the TORI to 40 h under NSAID pretreatment. Our findings demonstrated that, under diclofenac prophylaxis, delayed oocyte retrieval did not result in evident deterioration of oocyte or embryonic developmental potential, with a statistically significant improvement in retrieval from follicles < 14 mm. These results indicated a plausible decoupling between follicular rupture and intrafollicular oocyte maturation, thereby suggesting that the established timing paradigm for oocyte retrieval in ART may be amenable to extension.

In conventional IVF, an approximate 36‐h hCG‐to‐OPU interval is typically scheduled to balance the harvest of oocytes following meiotic completion against the risk of spontaneous ovulation [20]. This risk may increase when the interval is prolonged beyond 38 h, potentially resulting in cycle cancelation [21]. NSAIDs were initially introduced to manage the unpredictable LH surge and the high risk of premature ovulation in the natural cycle, where they significantly reduced the premature ovulation rate from 6.8%–20% to 3.6% [9]. In our study, among cycles including both natural and controlled stimulation cycles, the cancelation rate due to premature ovulation in cycles with diclofenac treatment was 0.53% (4/741), compared to an overall rate of 1.3% (82/6504). By achieving a functional steady state of follicle in vivo, NSAID intervention challenges the temporal constraint on oocyte retrieval and widens the time window beyond the conventional range [10]. This phenomenon supports the hypothesis that oocyte nuclear maturation and follicular rupture, though initiated by the same LH/hCG surge, are mediated by distinct downstream effectors [22]. As a potent COX inhibitor, diclofenac effectively arrests follicular wall degradation by suppressing PGE2‐mediated activation of matrix metalloproteinases (MMPs) and ADAMTS1, without apparently compromising maturation pathways [23].

Our study observed a numerically higher number of oocytes retrieved (9.36 ± 7.62 vs. 8.38 ± 6.27) along with a significantly higher ratio of oocytes retrieved to follicles ≥ 14 mm on trigger day in the delayed group (1.27 ± 0.42 vs. 1.14 ± 0.47, p = 0.031). This suggested that the benefit of a modest TORI extension lies possibly in improving the retrievability of oocytes from smaller or borderline follicles. The successful retrieval of an oocyte depends not only on nuclear maturation but also on the complete detachment of the COC from the follicular wall. Kawashima et al. demonstrated that after the LH/hCG surge, cumulus cells activate calpain 2 via EGFR‐ERK1/2 and Ca²⁺ signaling. Activated calpain degrades paxillin and talin, disrupts focal adhesion complexes, and allows cumulus cells to detach and form membrane protrusions. Calpain activity increases within 2–5 h after hCG and reaches its maximum at 8 h, and membrane protrusions are observed 2–3 h before rupture of the follicle wall [24]. In COS cycles, follicular asynchrony often results in a cohort of 12–14 mm follicles when lead follicles reach maturity [25]. Bomsel‐Helmreich et al. observed that follicles ≤ 16 mm at 26 h post‐hCG were still in prophase, lagging approximately 12 h behind larger follicles [26]. Conventional 36‐h intervals may provide insufficient time for all follicles to achieve adequate cumulus expansion, possibly leading to oocyte retention during aspiration [27]. Thus, an extension of TORI can possibly boost the time‐dependent detachment, which may be particularly critical in follicles with slower signal transduction or delayed development. While a macaque study showed COX inhibitors may partially attenuate PGE2‐mediated cumulus expansion [28], our findings suggest that the additional incubation time may compensate for this inhibitory effect induced by drugs, thereby ensuring successful aspiration of marginal follicles and ultimately optimizing the total oocyte yield. The clinical relevance of a 2–3 h TORI extension was supported by Deng et al., who compared patients with TORI of 36.00–36.99 h versus 38.00–39.32 h. They found that the extended interval was associated with a significantly higher number of oocytes retrieved (14.63 ± 6.55 vs. 12.53 ± 7.21, p < 0.001). Multivariable regression confirmed a positive correlation between TORI and oocyte yield [29]. These findings are by and large consistent with our own results. Furthermore, extending the TORI from 36 h to 40–42 h was reported to successfully retrieve mature oocytes in cases of genuine empty follicle syndrome and lead to live births, supporting the requirement of longer interval for oocyte detachment and cumulus expansion for some patients [30]. In our study, the incidence of cycles with no retrieved oocytes was numerically lower in the delayed group (1.19% vs. 1.79%). The lack of significance in the difference was most likely attributable to the low baseline incidence in this low‐risk cohort and the limited statistical power due to sample size.

Another potential benefit of delayed oocyte retrieval lies primarily in harnessing the in vivo environment to facilitate oocyte maturation and subsequent embryonic developmental potential [1, 31]. Vandenberghe et al. systematically compared the laboratory outcomes across different intervals between trigger to oocyte injection from < 36 h to ≥ 41 h and found that significant improvements in maturation rate, fertilization rate, and utilization rate were observed as the interval increased [32]. However, that extension was primarily attributable to prolonged in vitro culture rather than extended intrafollicular residence. By comparison, prolonged intra‐follicular residence more likely preserves metabolic homeostasis through sustained paracrine signaling and antioxidant exposure [33]. Relative to the oxidative stress encountered in conventional in vitro culture, the physiological buffering afforded by an extended in vivo residence appears safer for human oocytes [12, 34]. Nonetheless, conclusions regarding the advantages of a longer interval (> 36 h) from two published meta‐analyses remain controversial, especially for the maturation rate [13, 31]. Our findings showed that a prolonged interval over 40 h yielded a comparable maturation rate with the conventional group (78.9% vs. 78.8%), which was concluded similarly with Deng et al. (0.88 ± 0.13 vs. 0.86 ± 0.15, p = 0.3563) [29]. This may possibly be explained by the fact that the proportion of MII oocytes from mature follicles has already plateaued by the conventional retrieval window of 36–38 h for that human oocyte nuclear maturation follows a defined timeline, with MII largely achieved by 35 h post‐hCG [26]. Furthermore, it was observed that the delayed group retrieved more oocytes from smaller (< 14 mm) follicles. Although most of these additional oocytes were mature, their maturation rate could be slightly lower than that of oocytes from larger follicles. This potential dilution effect might offset any modest improvement in MII rate among dominant follicles, resulting in an unchanged overall percentage. However, due to the lack of follicle‐specific maturity data, this hypothesis remains speculative and requires future validation. A comprehensive panel of laboratory outcomes including fertilization, cleavage, usable embryo, and high‐quality embryo rates remained comparable between the two groups. At a minimum, the absence of significant differences in oocyte and embryo competence supports the noninferiority and clinical feasibility of the delayed OPU approach.

Our study observed differential clinical outcomes between fresh and frozen–thawed embryo transfers in the delayed OPU group. In fresh transfer cycles, the biochemical pregnancy rate in the delayed OPU group was 8.6% lower than that in the conventional OPU group, whereas in thawed transfer cycles, the outcomes were highly comparable between the two groups. Although this difference did not reach statistical significance, it prompted us to consider whether adjustments to the transfer strategy are warranted when delayed OPU is performed, and whether endometrial‐embryo synchrony may be altered by the prolonged interval from trigger to oocyte retrieval, which is derived from the speculation that the extended duration of in vivo LH exposure experienced by granulosa cells retained within the follicles during delayed OPU may possibly accelerate luteinization, potentially altering progesterone synthesis and disrupting the endometrial implantation window [35, 36]. However, this hypothesis requires validation through more prospective clinical studies investigating progesterone trajectories and endometrial receptivity assessment.

Several limitations of this study merit consideration. First of all, the retrospective nature of this study remains an inherent limitation. While PSM effectively balances observable covariates such as age and AMH, it may not fully account for unmeasured confounders, including latent clinical justifications in real‐world decision‐making. However, selection bias was mitigated by restricting the analysis to first‐cycle patients and by the fact that a proportion of delays were operationally planned (e.g., for resource allocation or patient scheduling) rather than pathologically driven. By employing exact matching for stimulation protocols and fertilization methods, we further ensured high comparability in clinical trajectories between cohorts. Furthermore, as the delayed and routine analytic groups in this study were defined according to the actual recorded TORI, premature ovulation rates could not be accurately determined in either group because canceled cycles did not undergo OPU and therefore lacked an actual TORI. This retrospective limitation should be addressed in future prospective studies with predefined intended retrieval schedules and systematic monitoring of follicular rupture. The relatively small sample size may have resulted in insufficient statistical power to detect a minor but potentially clinically meaningful decline in safety parameters such as blastulation rate. Nevertheless, the comparable fertilization and maturation rates combined with the higher absolute number of blastocysts (2.56 vs. 2.39) suggested no substantial detrimental impact on early embryonic development. The long‐term safety of this strategy warrants further validation through large‐scale prospective trials. In addition, most delayed cycles in this study had a TORI of only slightly over 40 h. The feasibility of more extended delays requires investigation in future studies.

If the feasibility and safety of delayed oocyte retrieval are further supported by future prospective evidence, the broadened timelines could offer substantial clinical value by transforming rigid IVF workflows into more flexible and resource‐efficient schedules, especially in high‐volume centers [29]. Traditionally, IVF centers are constrained by the requirement for an approximate 36‐h interval, which sometimes necessitates late‐night trigger injections (e.g., 9–10 pm) or early‐morning surgical procedures. A 3‐h extension allows the trigger to be given in the early evening (e.g., 6–7 pm) while still performing retrieval during regular daytime hours, or keeps the usual trigger time but shifts retrieval to late morning. This avoids late‐night or early‐morning medical visit and thereby alleviates fatigue, sleep disruption, traffic burden, and the need for overnight accommodation, particularly beneficial for those who live far from the clinic or have limited access to transportation. Moreover, the modest extension provides a safety buffer against unexpected clinical delays (e.g., such as equipment issues, operating room availability, or scheduling conflicts), reassuring patients that even a slightly later retrieval does not compromise laboratory or clinical outcomes. As our data suggested and if it was further validated, the extension may improve oocyte yield from smaller follicles, thereby potentially reducing the number of required stimulation cycles for patients with recurrent empty follicle syndrome or severely diminished ovarian reserve. As research on delayed OPU progresses, several strategic avenues warrant further investigation, including: determining whether laboratory protocols and embryo transfer strategies require adaptation for oocytes retrieved via delayed OPU; identifying the clinical subgroups that derive the maximal benefit from this protocol; establishing the minimum effective dose and optimal administration timing for NSAIDs to balance efficacy and safety; characterizing the molecular and transcriptomic profiles of oocytes and granulosa cells obtained following delayed OPU; and establishing offspring follow‐up cohorts to evaluate whether pharmacological exposure during delayed retrieval has any long‐term epigenetic consequences.

In conclusion, a planned OPU delay beyond 40 h post‐trigger with diclofenac pretreatment demonstrated no discernible impact on laboratory and clinical outcomes, with potentially improved oocyte yielding from sub‐optimal follicles. These findings provide preliminary evidence for the feasibility of a more flexible OPU scheduling framework that may improve operational efficiency in ART practice. Large‐scale prospective studies are warranted to confirm these findings and to establish optimized protocols for this approach.

Funding

This study was supported by the National Natural Science Foundation of China (grant no. 82571904).

Ethics Statement

The study was approved by the Ethics Committee of the Sixth Affiliated Hospital of Sun Yat‐sen University (approval number: 2026ZSLYEC‐009).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Love plot demonstrating standardized mean differences of covariates before and after propensity score matching.

Figure S2: Distribution of TORI and hourly subgroup analysis of core laboratory outcomes between the delayed and routine groups. (A, B) Histograms showing the distribution of TORI before (A) and after (B) PSM. (C–F) Boxplots with jittered individual data points displaying the median and IQR for each laboratory outcome, stratified by hourly TORI subgroups. (C) Number of oocytes retrieved. (D) Maturation rate. (E) Fertilization rate. (F) Blastulation rate. Color in blue represents the routine group, and red represents the delayed group.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.