Intraovarian injection of autologous platelet-rich plasma combined with peripheral blood mononuclear cells improves ovarian function in advanced-age women with diminished ovarian reserve: a clinical retrospective cohort and single-cell transcriptomic analysis

Abstract

Objective

To perform a preliminary, hypothesis-generating evaluation of the association between intraovarian injection of platelet-rich plasma (PRP) combined with peripheral blood mononuclear cells (PBMCs) and ovarian response parameters in advanced-age women with diminished ovarian reserve (DOR), and to explore potential underlying mechanisms using a complementary animal model.

Methods

A retrospective, non-randomized cohort analysis was conducted on 122 advanced-age DOR patients undergoing IVF-ET. Group allocation was based on patient preference and financial considerations following comprehensive counseling. Patients were accordingly categorized into a blank control group (n = 45), a PRP group (n = 47), or a PRP + PBMCs group (n = 30). Key confounding factors were adjusted for using Inverse Probability of Treatment Weighting (IPTW). In parallel, a small-scale (n = 5 per group) animal study using a 13-month-old DOR rat model was performed to generate mechanistic hypotheses. Ovaries were harvested 10 days post-intraovarian injection for single-cell RNA sequencing (scRNA-seq). Primary hypothesis-generating outcomes were changes in AMH and AFC; oocyte retrieval outcomes were pre-specified secondary endpoints.

Results

In this observational analysis, after statistical adjustment for key confounders using IPTW, the PRP + PBMCs group showed a nominally higher mean number of oocytes retrieved (adjusted mean difference: 0.90, 95% CI: 0.25 to 1.55, p = 0.007) and metaphase II (MII) oocytes (adjusted mean difference: 0.88, 95% CI: 0.20 to 1.56, p = 0.014) compared to the PRP alone group. These statistically significant findings in adjusted analyses should be interpreted as exploratory within the context of this non-randomized study. For the pre-specified primary hypothesis-generating outcomes, changes in serum AMH levels and antral follicle count (AFC) from baseline showed positive trends but did not reach statistical significance in the IPTW-adjusted comparison. No serious adverse events were reported. In the complementary exploratory animal study, rats receiving PRP + PBMCs showed improved follicle counts and hormone levels compared to controls. Single-cell transcriptomic analysis suggested potential alterations in intercellular communication among granulosa cells and activation of pathways related to protein synthesis and energy metabolism.

Conclusion

In this retrospective, hypothesis-generating study, an association was observed between intraovarian PRP + PBMCs injection and increased oocyte yield in adjusted analyses. The exploratory mechanistic data suggest that this potential effect may involve metabolic and immunomodulatory reprogramming of the ovarian microenvironment, a hypothesis that requires direct validation. Importantly, this study was not powered to assess pregnancy or live birth outcomes. Collectively, these preliminary findings do not establish clinical efficacy but serve to highlight the need for and inform a prospective, randomized controlled trial to definitively evaluate this combined approach.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13048-026-02006-z.

Introduction

With the trend of delayed marriage and childbearing, fertility issues in advanced-age women have become increasingly prominent. The proportion of women over 35 years old among those receiving assisted reproductive technology (ART) has risen significantly. Declining oocyte quality and reduced fertility potential due to natural ovarian aging (NOA) are key factors affecting their pregnancy success rates [1]. NOA is a complex, multifactorial process involving genetic regulation, oxidative damage, metabolic dysregulation, and remodeling of the ovarian microenvironment. Current clinical strategies for improving NOA primarily include hormone replacement therapy and antioxidant treatment, but their overall efficacy is limited. While hormone replacement can regulate endocrine levels to some extent, long-term use may increase the risk of thrombosis and endometrial pathologies. Antioxidant therapy can ameliorate ovarian oxidative stress in some patients but shows suboptimal effects on restoring oocyte quality and ovarian reserve function [2]. These conventional treatments fail to effectively address issues such as reduced ovarian vascularization, functional decline of granulosa cells, and weakened interaction between oocytes and granulosa cells in advanced-age women, resulting in limited improvement in fertility.

In recent years, platelet-rich plasma (PRP), as an emerging regenerative medicine treatment, has garnered attention due to its content of various growth factors. These growth factors can promote angiogenesis and tissue repair, thereby improving the ovarian microenvironment. Clinical studies have shown that intraovarian PRP injection can significantly increase patients' anti-Müllerian hormone (AMH) levels and improve levels of basal follicle-stimulating hormone (FSH), basal luteinizing hormone (LH), and basal estradiol (E2) [3]. Some patients showed a significant increase in the number of high-quality embryos after ART cycles following treatment. However, the efficacy of PRP monotherapy remains limited. On one hand, PRP primarily acts by promoting angiogenesis and improving blood flow, but its ability to intervene in the complex cellular interactions and metabolic regulatory mechanisms within the ovary is constrained [4]. On the other hand, individual responses to PRP therapy vary considerably, and some patients show minimal improvement [5].

To overcome the limitations of PRP monotherapy, research has begun exploring the combination of PRP with other cell-based therapies. Peripheral blood mononuclear cells (PBMCs), rich in hematopoietic stem cells and mesenchymal stem cells, possess potent immunomodulatory and tissue repair capabilities and are thus gradually being applied in reproductive medicine [6].

Our team's prior animal experiments have confirmed that intraovarian injection of PRP combined with PBMCs significantly improved ovarian function in an NOA rat model, manifested by restored estrous cycles and hormone levels, increased follicle numbers, and improved embryo yield and litter size. Compared to PRP alone, the synergistic effect of PRP combined with PBMCs more comprehensively improved the ovarian microenvironment and enhanced vascularization [7]. However, despite the significant efficacy observed in animal studies, the application of PRP combined with PBMCs in humans remains experimental, lacking support from large-scale clinical trial data regarding its safety and efficacy. Furthermore, mechanistic research has remained at the tissue level and has not yet delved into the molecular and cellular dimensions.

Therefore, this study combines clinical research with animal experiments. The first part retrospectively collected and analyzed clinical data from DOR patients undergoing IVF at the Reproductive Medicine Center of Qinzhou Maternal and Child Health Hospital between January 2023 and January 2025. Statistical analysis was performed on ovarian reserve function data and laboratory outcomes from the first controlled ovarian stimulation (COS) cycle after treatment among the blank control, intraovarian PRP + PBMCs injection, and intraovarian PRP injection groups, to explore the potential clinical efficacy of PRP + PBMCs. The second part involved intraovarian injections of normal saline, PRP, PBMCs, or PRP + PBMCs into naturally aged (NOA) rats and young control rats, followed by scRNA-seq of ovarian tissues to explore the potential molecular and cellular mechanisms by which PRP + PBMCs improve ovarian function in NOA rats, providing preliminary experimental evidence for its clinical application.

Materials and methods

Clinical trial design and participants

Study participants

This was a retrospective, non-randomized cohort study. We reviewed the records of 122 consecutive female DOR patients aged ≥ 40 years who underwent IVF-ET treatment at the Reproductive Medicine Center of Qinzhou Maternal and Child Health Hospital between January 2023 and January 2025. Patients were not prospectively assigned to groups. Group B (PRP + PBMCs, n = 30) and Group C (PRP alone, n = 47) included all patients who self-selected and underwent these respective experimental interventions during the study period, after comprehensive counseling regarding their experimental nature, potential benefits, and costs. Group A (blank control, n = 45) consisted of eligible patients from the same timeframe who, after counseling, opted for conventional IVF without any ovarian intervention or could not afford the experimental therapies. This patient-driven allocation introduces significant risks of selection bias and confounding by indication (e.g., baseline prognosis, financial capacity, physician/patient preference). To address this critical limitation, we performed an Inverse Probability of Treatment Weighting (IPTW) analysis, adjusting for the following pre-specified potential confounders: age, BMI, baseline AMH, baseline AFC, baseline FSH, and previous IVF attempts. The effectiveness of the balancing was assessed using standardized mean differences. All clinical inferences are presented with this methodological constraint in mind.

Inclusion criteria

Subjects meeting the following criteria were included: 1. AMH level < 1.2 ng/ml, basal FSH level ≥ 10 IU/L, or a total bilateral antral follicle count (AFC) ≤ 5 as assessed by transvaginal ultrasound. 2. Patients agreed to enter the IVF-ET cycle within 3 months after PRP combined with PBMCs ovarian injection treatment.

Exclusion criteria

Individuals meeting any of the following criteria were excluded: 1. Ovarian failure due to non-natural causes such as chromosomal karyotype abnormalities, history of oophorectomy/partial oophorectomy, history of pelvic radiotherapy/chemotherapy, autoimmune oophoritis, etc. 2. Comorbidities significantly affecting IVF-ET success rates, such as moderate to severe endometriosis, significant hydrosalpinx, severe intrauterine adhesions, submucous fibroids, active pelvic inflammatory disease, etc. 3. Poorly controlled endocrine or metabolic diseases, such as uncontrolled thyroid disease, diabetes with HbA1c > 7.0%, severe obesity or emaciation, etc. 4. Factors affecting the safety of PRP/PBMCs treatment, such as platelet dysfunction, coagulation abnormalities, active infection, history of malignancy, severe active autoimmune disease, allergy to related agents, etc. 5. Received treatments potentially interfering with efficacy evaluation within 3–6 months prior to enrollment, such as hormone replacement cycles, antioxidant therapy, IVF cycles, or other ovarian regeneration therapies. 6. Presence of severe male factor infertility, not planning to use ICSI or donor sperm. 7. Other conditions deemed by the investigator as unsuitable for participation, such as poor compliance, mental illness, alcohol or drug abuse.

Ovarian stimulation

Patients underwent transvaginal ultrasound on days 1–3 of the menstrual cycle to assess baseline ovarian status, followed by a mild stimulation protocol for controlled ovarian stimulation (COS), consistent with our center's approach to minimize physiological stress and cost for advanced-age DOR patients. From cycle day 2 or 3, letrozole (LE) 5 mg was administered daily, concurrently with intramuscular injection of human menopausal gonadotropin (hMG) 150 IU. To ensure the quality of ovarian stimulation and cycle outcomes, serum LH and progesterone levels were monitored at regular intervals starting from the first day of gonadotropin administration. This continuous monitoring aimed to promptly detect and manage any premature LH surge or ovulation. The combined administration of LE and hMG continued until the day before the trigger. Ovulation was triggered with 4000 IU of human chorionic gonadotropin (hCG) when at least one follicle reached 18 mm in diameter or three follicles reached 16 mm in diameter. Oocyte retrieval was performed via transvaginal ultrasound-guided puncture 36 h after HCG injection. Gonadotropin-releasing hormone (GnRH) antagonists were not administered at any point during the treatment cycle. LE was administered throughout the cycle, as documented in the literature, to prevent premature LH surges in our mild stimulation protocol. Ibuprofen was also administered prior to ovulation based on limited literature suggesting a potential reduction in premature follicle rupture and was applied uniformly across all treatment groups in this study [8, 9].

Preparation of Platelet-Rich Plasma (PRP)

Thirty milliliters of venous blood were drawn from the patient’s elbow. It was centrifuged at 300 g for 12 min at 4 °C. The supernatant and the interface layer (approximately 1–2 mm) were transferred to a 15 ml conical tube. A second centrifugation was performed at 1200 g for 5 min. The supernatant was discarded, leaving approximately 0.5 mL of supernatant and the pellet, which were mixed to obtain the PRP. The final platelet concentration in the PRP was quantified using an automated hematology analyzer. The median final platelet count was 1.3 × 10^12/L (IQR: 1.2–1.4 × 10^12/L), representing a 3.5-fold median increase over baseline peripheral blood (range: 3.0–5.1 fold). The preparation was leukocyte-poor (L-PRP). It was activated with 10% calcium chloride (CaCl2) immediately prior to injection. The time from blood draw to intraovarian injection was under 2 h. Sterility was maintained throughout the process in a class II biosafety cabinet.

Preparation of Peripheral Blood Mononuclear Cells (PBMCs)

Thirty milliliters of venous blood were drawn from the patient’s elbow. 4 mL of lymphocyte separation medium was added to a sterile conical centrifuge tube, and the fresh whole blood was slowly layered over the separation medium. After centrifugation at 2200 rpm for 16 min, distinct layers formed. The top layer was light red transparent plasma, followed by a thin white ring layer (containing PBMCs), separation medium, and red blood cell pellet. The upper two-thirds of the plasma above the white ring layer were carefully removed without disturbing the white ring layer. Approximately 3 mL of the white ring layer fluid was aspirated into a new centrifuge tube. The collected fluid was diluted to 8 mL with injectable normal saline and mixed thoroughly. It was then centrifuged at 3000 rpm for 5 min. The supernatant was discarded, and the pellet was resuspended in normal saline to a final volume of 2 mL. The final PBMC suspension typically yielded a concentration of 3.5 × 10^6 cells/mL (range: 3.0–4.0 × 10^6), resulting in a total dose of 6–8 × 10^6 cells per ovary. Cell viability, assessed by trypan blue exclusion, was consistently > 95% (range: 95–98%). Immunophenotyping by flow cytometry on a subset of samples (n = 10) confirmed the expected composition: CD45 + leukocytes > 99%, with a predominance of CD3 + T cells (median 65%), CD19 + B cells (median 15%), and CD14 + monocytes (median 18%). Processing time from blood draw to injection was < 4 h.

Intraovarian injection of PRP and PBMCs + PRP

The intraovarian injections of PRP or PRP + PBMCs were performed on an outpatient basis, 3–7 days after the end of menstruation (patients were instructed to abstain from intercourse during this period). Under sedation/anesthesia, using transvaginal ultrasound guidance, a 35 cm 19G single-lumen oocyte retrieval needle (Cook, USA) was used to inject 2 mL of either PRP or the PRP + PBMCs mixture into each ovary, distributed across three points in the cortical and stromal regions. After the procedure, patients rested and were observed for 30–40 min. Upon confirmation of safety, they were discharged and prescribed a 3-day course of antibiotics for infection prophylaxis.

Embryo transfer

All viable embryos were cryopreserved on day 3 (D3) post-fertilization. Approximately 6 weeks later, during the second menstrual cycle following oocyte retrieval, frozen-thawed embryos were transferred in a hormonally prepared cycle.

Outcome variables

The primary outcome measures for hypothesis generation were the changes in serum AMH and AFC levels from before to 6 weeks after treatment. Assessment time points were set as pre-treatment and 6 weeks post-treatment, focusing on parameter changes in the second menstrual cycle after injection. Secondary outcome measures included the number of oocytes retrieved post-treatment, MII oocytes, 2PN fertilized oocytes, high-quality cleavage embryo rate, and adverse events. Given the retrospective design and significant attrition at various stages (see Fig. 1), clinical pregnancy rate is reported descriptively and interpreted with caution. We report pregnancy outcomes per initiated cycle, per oocyte retrieval, and per embryo transfer, with clear denominators, recognizing that live birth data were not available due to limited follow-up period.

Fig. 1.

Clinical trial flowchart

The flowchart of the clinical trial is presented in Fig. 1 and Figure S1.

Animal experiment

Study subjects

Thirteen-month-old naturally aged female Sprague–Dawley (SD) rats were used as a model of natural ovarian aging (NOA). All experimental animals were purchased from the Experimental Animal Center of Guangxi Medical University. The ambient temperature was controlled at 28 ± 2 °C with a 12-h light/dark cycle. Vaginal smears were observed for 5 consecutive days to screen for individuals with disrupted estrous cycles (defined as ≥ 5 days without a complete cycle), as per the NOA model. Twelve-week-old SD female rats with regular estrous cycles served as the young control group. The animal experiment component of this study was approved by the Animal Ethics Committee of Guangxi Medical University (Approval No.: 202501004).

Experimental groups

Twenty-five validated NOA rats were randomly divided into the following 5 groups (n = 5 per group):

• Group A (NOA Control Group): Received intraovarian injection of normal saline only, no cell transplantation.

• Group B (PRP Group): Received intraovarian injection of PRP (4 × 10⁷ platelets/ovary).

• Group C (PBMCs Group): Received intraovarian injection of G-CSF-mobilized PBMCs (4 × 10⁶ cells/ovary).

• Group D (PRP + PBMCs Group): Received intraovarian injection of PBMCs (4 × 10⁶ cells/ovary) combined with PRP (4 × 10⁷ platelets/ovary).

• Group E (Young Rat Group): Received intraovarian injection of normal saline only, no cell transplantation.

PBMCs and PRP preparation methods

PBMCs isolation

Five-week-old male SD rats were subcutaneously injected with G-CSF (100 μg/kg) daily for 5 consecutive days. Twelve hours after the last injection, blood was collected via the abdominal aorta. PBMCs were isolated using Ficoll density gradient centrifugation. Cells were washed twice with PBS, resuspended, and stored at 4 °C. Transplantation was completed within 4 h.

PRP preparation

Blood was collected from the abdominal aorta of male rats from the same batch. PRP was prepared via a two-step centrifugation process (300 g × 10 min • 500 g × 10 min). The platelet concentration was adjusted to 4 × 10⁷/ovary for use.

The preparation protocols for PRP and PBMCs were conceptually aligned with those used in our prior rat model studies, albeit adapted for clinical application.

Intraovarian injection procedure

Rats were anesthetized using inhalational isoflurane. Anesthesia was induced by placing the rats in a chamber with 5% isoflurane delivered in oxygen (1.5 L/min). After loss of consciousness, animals were transferred to a surgical platform, and anesthesia was maintained via a nose cone with 2–2.5% isoflurane. The depth of anesthesia was confirmed by the absence of a pedal withdrawal reflex before and during the surgical procedure. Under aseptic conditions, the bilateral ovaries were exposed. A microsyringe was used to slowly inject the cell/platelet suspension into the ovarian parenchyma, with a total volume of not more than 20 μL per ovary. The needle was left in place for 30 s after injection to prevent backflow. The ovary was then repositioned, and the incision was closed with sutures. Post-operative observations were conducted routinely, with no infections or bleeding complications noted. Rats were sacrificed on day 10 post-transplantation. Serum levels of AMH, E2, and FSH were measured, follicles at various stages were counted, and ovarian tissues were collected for scRNA-seq.

Euthanasia and tissue collection

On day 10 post-transplantation, all rats were euthanised for tissue collection. To ensure a humane endpoint, rats were first deeply anesthetized using the same isoflurane protocol as for surgery (5% for induction, 2–2.5% for maintenance). Once a deep plane of anesthesia was confirmed by the absence of response to a painful toe pinch, euthanasia was performed by cervical dislocation. This method is consistent with the AVMA Guidelines for the Euthanasia of Animals. Blood was collected immediately via cardiac puncture for measurement of serum hormone levels (AMH, E2, and FSH). Ovaries were then harvested promptly; one ovary from each rat was fixed for histological analysis and follicle counting, while the contralateral ovary was processed for single-cell RNA sequencing.

scRNA-seq analysis of rat ovarian cells

This study utilized the MGI DNBSeq platform to conduct an in-depth investigation into the mechanisms by which PRP + PBMCs improve ovarian function in rats at the single-cell level. Ovarian tissues from 25 rats (5 per group) were enzymatically dissociated, ultimately yielding a total of 271,913 ovarian cells (Group A: 44,563; Group B: 54,267; Group C: 48,279; Group D: 56,478; Group E: 68,326). During data preprocessing, the following filtering thresholds were set: the number of genes detected per cell was between 200 and 7500, the number of UMIs per cell was not to exceed 50,000, the proportion of mitochondrial genes per cell was below 20%, and the proportion of erythrocyte genes was below 8%. Cells expressing high levels of mitochondrial genes (> 10% of total UMIs) were excluded, and the scDblFinder package was used to remove doublets, resulting in high-quality single-cell data. To mitigate batch effects, the Harmony algorithm was applied for batch correction during integrated analysis, effectively eliminating technical variations between different samples. For differential expression analysis, the Wilcoxon rank-sum test was used, combined with the Benjamini-Hochberg (BH) method for multiple testing correction to ensure the reliability of the results. To ensure that statistical inference was based on biological replicates, pseudobulk expression profiles were first created by aggregating raw counts across all cells from each individual animal sample (n = 5 per group). Differential expression analysis between experimental groups was then performed at the sample level using these pseudobulk counts. This approach aligns with established best practices to prevent pseudo-replication and inflation of statistical significance.Furthermore, the CellChat package was employed for in-depth analysis of cell–cell communication relationships, aiming to reveal the collaborative mechanisms among different cell types in the process of ovarian function improvement. The single-cell RNA sequencing data generated in this study will be deposited in the Gene Expression Omnibus (GEO) repository upon manuscript acceptance and are available from the corresponding author on reasonable request prior to that.

The overall design of the animal experiment is shown in Fig. 2.

Fig. 2.

Animal experiment flowchart

Statistical analysis

All data in this study were statistically analyzed using SPSS 26.0 and R software. Given the observational, non-randomized nature of the clinical study, our primary analysis employed Inverse Probability of Treatment Weighting (IPTW) to adjust for pre-specified confounders (age, BMI, baseline AMH, AFC, FSH, previous IVF cycles). Balance diagnostics were assessed using standardized mean differences. IPTW weights were estimated using multinomial logistic regression, and weighted linear regression models were subsequently used to estimate adjusted mean differences and 95% CIs.Treatment effects are reported as weighted mean differences with 95% confidence intervals (95% CI). For conventional comparisons, normally distributed data are expressed as mean ± standard deviation and compared using one-way ANOVA, with post-hoc Tukey’s HSD test for pairwise comparisons to control family-wise error rate. Non-normally distributed data are expressed as median (P25, P75) and compared using the Kruskal–Wallis H test, with Dunn’s post-hoc test and Benjamini–Hochberg false discovery rate (FDR) correction. Count data are expressed as frequency (percentage) and compared using the chi-square test or Fisher’s exact test. All tests were two-sided, and a P < 0.05 was considered statistically significant, with acknowledgment that the study is exploratory and findings require validation. We explicitly acknowledge that unmeasured confounding may persist, and causal inferences cannot be drawn from this retrospective design.

Ethical approval

This study was approved by the Ethics Committee of the Reproductive Medicine Center, Qinzhou Maternal and Child Health Hospital (Approval No.: QZSFYSL〔2025〕03). All patients provided written informed consent, agreeing to the use of their clinical data for scientific research and the intraovarian injection procedure. The animal experiment component of this study was approved by the Animal Ethics Committee of Guangxi Medical University (Approval No.: 202501004).

Results

Comparison of baseline characteristics

No significant differences were observed among the three patient groups in terms of age, body mass index (BMI), basal FSH (bFSH), basal LH (bLH), basal E2 (bE₂), pre-treatment AMH, and AFC (all P > 0.05), as shown in Table 1.

Table 1.

Comparison of baseline characteristics among the three groups of patients. (n = 122)

Group	A (n = 45)	B (n = 30)	C (n = 47)	F	P
Indicator
Age (years)	42.42 ± 1.76	41.83 ± 1.23	42.47 ± 1.86	1.489	0.23
BMI (Kg/m2)	23.48 ± 3.05	22.71 ± 1.97	24.08 ± 2.46	2.58	0.08
bFSH (mIU/ml)	13.67 ± 7.29	13.99 ± 4.97	12.47 ± 4.54	0.686	0.506
bLH (mIU/ml)	4.87 (2.99,6.93)	5.52 (3.69,7.99)	4.11 (3.28,6.79)	2.777	0.249
bE2 (pg/ml)	33.44 (19.75,62.75)	27 (22.5,47.0)	40.50 (23.25,68.25)	1.687	0.43
Before AMH (ng/ml)	0.39 (0.15,0.53)	0.29 (0.12,0.63)	0.23 (0.03,0.48)	3.555	0.169
Before AFC (units)	2 (2,3.5)	2 (1,4)	2 (1,3)	4.283	0.117

Intra-group comparison of ovarian reserve parameters before and after treatment in Groups B and C

As the primary hypothesis-generating outcomes, changes in serum anti-Müllerian hormone (AMH) levels and antral follicle count (AFC) were assessed. Intra-group comparative analysis before and after treatment in Group B (PRP + PBMCs) and Group C (PRP) showed that both AFC and AMH levels increased significantly after treatment in both groups (P < 0.05). Post-treatment AFC and AMH levels in Group B were significantly higher than those in Group C (P < 0.05), as shown in Table 2. The median increase in AMH was greater in Group B (0.33 ng/ml) than in Group C (0.12 ng/ml).

Table 2.

Intra-group comparison of ovarian reserve parameters before and after treatment in Groups B and C. (n = 77)

Group	Before AMH (ng/ml)	After AMH (ng/ml)	Median Difference (95% CI)	z	P	Before AFC(units)	After AFC(units)	Median Difference (95% CI)	z	P
Indicator
B (n = 30)	0.29 (0.12,0.63)	0.62 (0.27,0.82)	0.33 (0.01–0.65)	−1.997	0.046	2 (1,4)	3.5 (2,4.25)	1.5 (0.0 −3.0)	−2.187	0.029
C (n = 47)	0.23 (0.03,0.48)	0.35 (0.18,0.62)	0.12 (0.01–0.23)	−2.194	0.032	2 (1,3)	3 (1,4)	1.0 (−0.5 −2.5)	−1.592	0.111
z		−1.928					−1.497
P		0.054					0.134
Median Difference (95% CI)		0.15 (−0.01 −0.31)					0.132–0.146

COS characteristics and oocyte retrieval outcomes

As key secondary outcomes, ovarian stimulation response and oocyte retrieval outcomes were compared. After treatment, no significant differences were observed among the three groups in total gonadotropin dose, duration of stimulation, or hormone levels on hCG day (all P > 0.05), suggesting comparable stimulation management across groups. Given the non-randomized design, our primary analysis employed Inverse Probability of Treatment Weighting (IPTW) to adjust for pre-specified confounders (age, BMI, baseline AMH, AFC, FSH, and previous IVF cycles). The IPTW-adjusted analysis indicated that the PRP + PBMCs combination was associated with a higher mean number of oocytes retrieved (adjusted mean difference: 0.90, 95% CI: 0.25 to 1.55, *p* = 0.007) and metaphase II (MII) oocytes (adjusted mean difference: 0.88, 95% CI: 0.20 to 1.56, *p* = 0.014) compared to PRP monotherapy (Supplementary Table S1). The consistency of this finding is supported by a forest plot of IPTW-adjusted effects (Supplementary Figure S2).

For context, the unadjusted median number of oocytes retrieved was 2.0 (IQR: 1.0–3.5) in Group B (PRP + PBMCs), compared to 1.0 (0.0–2.0) in Group C (PRP alone) and 1.0 (0.5–3.0) in Group A (Control) (P = 0.036, Kruskal–Wallis test) (Table 3).

Table 3.

COS characteristics and oocyte retrieval outcomes. Note: 'a' indicates P < 0.05 for comparison between Group B and Group C. (n = 122)

Group	A (n = 45)	B (n = 30)	C (n = 47)	P value
Indicator
Total Gn dose(IU)	1155 ± 373.6	1101 ± 273.6	1295 ± 473.9	0.0828
Stimulation days(d)	7.553 ± 2.254	7.367 ± 1.79	8.409 ± 2.546	0.0937
HCG Day FSH(mIU/ml)	13.82 ± 8.118	12.97 ± 5.219	12.95 ± 6.259	0.7965
HCG Day E2(pg/ml)	360.4 ± 187.9	449.6 ± 170.9	352.8 ± 185.3	0.0557
HCG Day LH(mIU/ml)	7.06 (5.45, 9.12)	6.36 (4.83, 8.67)	6.2 (3.69, 8.33)	0.1417
HCG Day P(ng/ml)	0.31 (0.23, 0.4)	0.39 (0.34, 0.54)	0.36 (0.26, 0.47)	0.0826
Oocytes retrieved(units)	1.00 (0.50,3.00)	2.00 (1.00, 3.50)	1 (0.00,2.00)a	0.036
MII oocytes(units)	1.00 (1.00,3.00)	2.00 (1.00,4.00)	1.00 (1.00,2.00)a	0.008

^aindicates P < 0.05 for comparison between Group B and Group C

Comparison of laboratory outcomes after treatment among the three groups

After treatment, the normal fertilization rate and high-quality cleavage embryo rate in Group B (PRP + PBMCs) were numerically higher than those in Groups A (Control) and C (PRP alone), but the differences were not statistically significant (P > 0.05) (Fig. 3).

Fig. 3.

Comparison of laboratory outcomes after treatment among the three groups

Regarding clinical pregnancy, the rate per embryo transfer cycle was 26.9% (7/26) in Group A, 40.0% (8/20) in Group B, and 34.4% (11/32) in Group C, again with no statistically significant difference (P > 0.05) (Fig. 4). However, a critical methodological limitation must be emphasized when interpreting these pregnancy outcomes. As detailed in the study flowchart (Figure S1), substantial attrition occurred prior to the embryo transfer stage across all groups. Specifically, 19 patients in Group A, 10 in Group B, and 15 in Group C did not reach embryo transfer due to failure to retrieve oocytes, fertilization failure, or absence of transferable embryos.

Fig. 4.

Comparison of clinical pregnancy rates after treatment among the three groups

Therefore, the denominator for the pregnancy rate (embryo transfer cycles) represents a selectively retained subset of patients who successfully navigated all prior stages of the IVF process. Comparisons based on this selected population are inherently biased and cannot reliably reflect the treatment effect on the entire cohort from the initiation of the cycle. Consequently, these pregnancy outcomes should not be interpreted as evidence of efficacy but rather as preliminary, hypothesis-generating observations. They are reported here for completeness and to highlight the need for future prospective studies with complete follow-up and intention-to-treat analysis.Therefore, pregnancy outcomes are presented for descriptive completeness only and should not be interpreted as evidence of treatment efficacy in this study.

Safety outcomes

No serious adverse events occurred during this study. Only one patient in Group B experienced mild lower abdominal discomfort, which resolved spontaneously within 24 h without medication. No infections, bleeding, or ovarian cyst formation were observed.

Ovarian function recovery

Intraovarian co-injection of PRP and PBMCs resulted in a better restoration of ovarian function in NOA rats compared to PRP injection alone, as evidenced by a more pronounced improvement in hormone levels and a greater increase in the number of follicles at various stages, as detailed in Figs. 5 and 6.

Fig. 5.

Hormone levels (AMH, FSH, E2) were measured after intraovarian injection of normal saline, PRP, PBMCs, or PRP + PBMCs in NOA rats. Note: * indicates P < 0.05, ** indicates P < 0.05

Fig. 6.

Follicle counts at various stages in rat ovaries after intraovarian injection of normal saline, PRP, PBMCs, or PRP + PBMCs. Note: * indicates P < 0.05, ** indicates P < 0.05

scRNA-seq analysis

Cell clustering and annotation

t-SNE and UMAP algorithms were used for non-linear dimensionality reduction analysis of all cells, resulting in 11 distinct cell clusters. Based on the CellMarker database and literature-reported cell marker genes, these 11 cell types were annotated as major ovarian cell types (Fig. 7a, b): Oocytes (Ddx4, Dazl, Figla, Zp3, Zp2), Granulosa Cells (Amh, Fst), Stromal Cells (Col1a1, Mfap4), Theca Cells (Acta2, Srd5a1, Cyp17a1), Endothelial Cells (Flt1), Epithelial Cells (Krt18, Upk3b), Smooth Muscle Cells (Upk3b, Dcn, Pdgfrb, Lum, Col1a1, Cald1, Nr2f2, Fshr), Luteal Cells (Star, Fst), Myeloid Cells (Cd14, Cd68, Csf1r), B Cells (Blnk, Cd79a), T/NK Cells (Cd3d, Cd3e, Nkg7, Gzmk, Gzma, Gzmm, Klrd). Granulosa cells were the most abundant, followed by stromal cells.

Fig. 7.

Cell subpopulation classification analysis, cell annotation, and cell proportions per group. a UMAP plot of cell clusters from scRNA-seq of ovaries from the 5 rat groups. b Marker genes for cell clusters. c Cell type proportions in ovarian tissues from each rat group

To investigate the dynamic changes in the proportions of different cell types in the ovaries of each group, the relative proportion of each cell type within each group (i.e., the number of each cell type divided by the total number of cells in the group) was calculated, and differences between groups were compared. The results are shown in Fig. 7c. In NOA rats, the proportions of three immune cell types – myeloid cells, B cells, and T/NK cells – were significantly increased. In young rats, the proportions of granulosa cells and stromal cells were higher than those in NOA rats and NOA rats treated with various interventions. The overall proportions of granulosa cells and stromal cells in Groups B, C, and D were higher than in Group A.

Granulosa cell subpopulation analysis

Granulosa Cells (GCs) play a crucial role in maintaining ovarian reserve and the growth and development of follicles. Abnormal GCs can directly lead to decreased oocyte quality, thereby affecting reproductive capacity. High levels of glycolysis in GCs are fundamental for active follicular growth [10]. Various products of glycolysis (such as pyruvate and ATP) are transported to the oocyte, supporting its growth and inhibiting the aging process [11]. Oocytes can promote the expression of glycolysis-related genes in GCs via paracrine factors (TGF-β superfamily members GDF9, BMP15, etc.), further enhancing energy supply [12]. Given the important influence of GCs on oocyte quality and follicular development, and their significant heterogeneity during folliculogenesis, we performed subpopulation analysis on GCs. Follicular development can be divided into three main stages: Preantral Follicle (PAF), Small Antral Follicle (SAF), and Large Antral Follicle (LAF). The transcriptional profiles of corresponding GCs change with follicular maturation. This study identified four GC subpopulations. Figure 8A, B show the GC subcluster results, including PAF GC (marked by Wt1 expression), SAF mural GC (mGC) (marked by Cald1 expression), LAF mGC (marked by Inhba expression), and Cumulus Cells (marked by Top2a expression). The expression of marker genes for each subpopulation is shown in Figs. 8C-F.

Fig. 8.

Granulosa cell subpopulation analysis. A Granulosa cell clustering. B Granulosa cell cluster annotation. C-F UMAP plots of granulosa cell markers

Furthermore, we analyzed the proportions of granulosa cell subpopulations in each group. The results showed that the proportion of SAF stage granulosa cells in Group D was higher than in Groups A, C, and B, and was similar to that in Group E (details in Fig. 9). SAF mGCs are the key effector cells of FSH in the follicle. FSH stimulates the transcription and translation of Cnot6 and Cnot6l genes in granulosa cells, mediating the clearance of specific mRNAs, thereby regulating the transition from the preantral to the antral stage [13]. Additionally, AMH is secreted by granulosa cells of preantral and small antral follicles and regulates early follicular growth by inhibiting the activation of primordial follicles [14]. Combining data analysis, the number of SAF mGCs in Group D was higher than in Groups A, C, and B. Correlating with the hormone levels and follicle counts in Figs. 5 and 6, AMH levels in Group D were higher than in Group A, and the number of follicles at various stages was higher in Group D. This indicates that intraovarian injection of PRP + PBMCs can increase the number of SAF mGCs, thereby enhancing AMH levels and the reserve of follicles at various stages in NOA rats.

Fig. 9.

Granulosa cell proportions in each group

Pseudotime analysis of granulosa cells

This study used the Monocle3 algorithm for pseudotime trajectory analysis of granulosa cells. As shown in Fig. 10B, Monocle3 analysis divided granulosa cells into 4 different developmental stages. Mapping the cell clusters from the granulosa cell subpopulation analysis onto the differentiation trajectory revealed that PAF GCs constituted the cluster with the highest differentiation potential, forming the starting point of the granulosa cell differentiation trajectory; SAF mGC and LAF mGC constituted two main developmental stages of the trajectory, forming a temporal differentiation sequence from PAF GC to SAF mGC and LAF mGC (Fig. 10A).

Fig. 10.

Pseudotime analysis of granulosa cell clusters. A Cell pseudotime trajectory. B Cell trajectory

It has been reported that PAF GCs, in the early stages of follicular development, primarily respond to FSH stimulation via paracrine signaling involving CNOT6/6L-mediated mRNA degradation, regulating the transition from preantral to antral follicles. They are a key component of the oocyte-granulosa cell regulatory loop, playing a fundamental role in maintaining ovarian function and fertility [15]. SAF mGCs engage in bidirectional communication with the oocyte via gap junctions and extracellular vesicles and begin secreting steroid hormones. Studies have shown that disruption of the normal differentiation process from SAF mGC to LAF mGC adversely affects follicular development [16].

To explore whether the intraovarian injection of PRP + PBMCs affects the differentiation trajectory of granulosa cell subpopulations, we performed developmental trajectory analysis using Monocle3 (Fig. 11). The results indicated that the differentiation from SAF mGC to LAF mGC was inhibited in NOA rats (Fig. 11a, f), suggesting that this differentiation process might be disturbed in NOA rats, thereby affecting oocyte quality. In Group D, the inhibition of SAF mGC to LAF mGC differentiation was alleviated (Fig. 11d, i). This suggests that PRP + PBMCs can modulate the differentiation of SAF mGC towards LAF mGC, thereby influencing oocyte development from the SAF to LAF stage.

Fig. 11.

Pseudotime analysis of granulosa cells from five rat groups. a-d Cell trajectories of groups A-E. f-j Pseudotime analysis of groups A-E

Cell communication analysis

To investigate how PRP + PBMCs influence the differentiation from SAF mGC to LAF GC, we performed cell–cell communication analysis among the granulosa cell subpopulations. Figure 12C shows that communication between granulosa cell subpopulations was significantly enhanced in Group D compared to Group A, particularly the signaling from SAF mGC to LAF GC. This suggests that intraovarian injection of PRP + PBMCs may affect granulosa cell differentiation and their interactions by modulating the communication signals from SAF mGC to LAF GC.

Fig. 12.

Granulosa cell communication analysis. A Statistics of the number and strength of cell communications between granulosa cell subpopulations in each group. B Chord diagrams of communication strength in each group. C Comparison of the number and strength of communications between cell subpopulations in Group D vs. Group A. D Comparison of the number and strength of communications between cell subpopulations in Group E vs. Group A. E Comparison of the number and strength of communications between cell subpopulations in Group E vs. Group D

KEGG and GO analysis

In addition to cell communication, we performed differential gene expression analysis (DEGs) on SAF stage granulosa cells to investigate whether the PRP + PBMCs combination intervention improves ovarian function in Group D NOA rats by regulating the transcriptome of SAF mGCs. In Group A NOA rats, the upregulated DEGs in SAF mGCs were associated with biological processes or pathways related to immune regulation, such as "regulation of innate immune response" and "autoantigen processing and presentation". In Group D, while retaining immune response modules, the DEGs included new pathways such as the "TGF-β signaling pathway" and "BMP signaling pathway", suggesting potential immune-fibrosis cross-regulation and early stromal remodeling status. The DEGs in Group E were enriched in typical functional characteristics of young rat ovaries, such as glucocorticoid response, FSH response, cAMP signaling, thyroid hormone, and small molecule metabolism.

Further comparison of DEGs in SAF mGCs between Group A and Group E revealed that SAF granulosa cells in Group A had reduced responsiveness to hormones, decreased energy metabolism, and the top 10 signaling transduction pathways in Biological Process (BP) were all related to immune response, indicating that ovarian aging in NOA rats is associated with chronic inflammation. In contrast, GO and KEGG analysis of SAF mGCs in Group D compared to Group A showed that, although chronic inflammation in the ovarian microenvironment of NOA rats persisted, signaling transduction pathways such as "TGF-β" and "BMP" appeared in the BP and Molecular Function (MF) categories. These pathways are involved in immune regulation and the inhibition of fibrosis. The TGF-β pathway, in particular, regulates granulosa cell function, immune cell function, and stromal cell function [17]. Relevant analysis results are detailed in Figs. 13 and 14.

Fig. 13.

GO analysis for Groups A, D, and E. a GO analysis of SAF mGCs in Group E. b GO analysis of SAF mGCs in Group D. c GO analysis of SAF mGCs in Group A

Fig. 14.

KEGG analysis for Groups A, D, and E. a KEGG analysis of SAF mGCs in Group E. b KEGG analysis of SAF mGCs in Group A. c KEGG analysis of SAF mGCs in Group D

Discussion

DOR in advanced-age women represents a persistent challenge in assisted reproduction. While traditional interventions offer limited benefit for enhancing ovarian reserve and oocyte quality [18]. regenerative strategies such as intraovarian platelet-rich plasma (PRP) injection have been explored with mixed results [3, 5]. Indeed, several randomized controlled trials have failed to demonstrate a significant improvement in the number of metaphase II (MII) oocytes or clinical pregnancy rates with PRP monotherapy [19]. To explore a potentially more comprehensive strategy beyond single-agent therapy, we conducted a hypothesis-generating investigation into a combined approach. In this non-randomized, retrospective cohort study, we sought to evaluate the association between intraovarian co-injection of autologous PRP and peripheral blood mononuclear cells (PBMCs) and ovarian response parameters in advanced-age women with DOR, using PRP alone as a comparator.Our adjusted clinical analysis observed an association between the PRP + PBMCs combination and a higher oocyte yield compared to PRP monotherapy. It is paramount to state that this finding is observational and does not imply causality. To generate parallel, complementary mechanistic hypotheses, we employed a rat model of ovarian aging. Thus, the clinical and animal components are distinct yet synergistic: the former identifies a potential associative signal in patients, while the latter provides an exploratory platform for mechanistic inquiry.Within this framework, scRNA-seq data from treated rats suggested a modulation of the ovarian microenvironment, including enhanced granulosa cell communication and an upregulation of TGF-β/BMP pathways. We speculate that these molecular changes could represent a biological substrate that, in theory, might support oocyte maturation. This is congruent with, but does not prove, the clinical observation of an increased MII oocyte count. This putative link remains hypothetical and necessitates direct experimental validation.

An exploratory, post-hoc IPTW analysis suggested that the association between PRP + PBMCs treatment and oocyte yield was also present in the subgroup of patients with ‘extremely low reserve’ (baseline AMH < 0.1 ng/mL), with an adjusted mean difference of 1.20 (95% CI: 0.01 to 2.93, P = 0.048). This cohort represents a population of particular clinical challenge. While this finding hypothetically suggests that the addition of PBMCs might provide unique signals in severely compromised ovaries, it must be interpreted with extreme caution. The analysis was not pre-specified, the sample size was very small (n = 7 for PRP + PBMCs), and it was not adjusted for multiple testing. Therefore, it should be viewed solely as a hypothesis-generating observation to inform future research in this challenging subgroup and does not support clinical use outside of a controlled trial setting.

The potential mechanisms underlying ovarian rejuvenation strategies are complex. While PRP is thought to act primarily through promoting proliferation and mitigating oxidative stress, ovarian aging involves a broader spectrum of alterations including cellular senescence, immune dysregulation, and stromal fibrosis [20]. Notably, the TGF-β signaling pathway is implicated as a key regulator in this process, influencing stromal cell activity, immune balance, and granulosa cell function [21]. Age-related stromal fibrosis is a recognized feature of ovarian aging [22]. and stromal cells are known to be crucial for follicular support and development through paracrine interactions [23]. To explore potential mechanistic synergies of our combined therapy, we conducted scRNA-seq on ovarian tissues from a naturally aged rat model. It is important to note that this model utilized PBMCs from young, male, G-CSF-mobilized donors. which differs from the autologous human clinical scenario and limits direct translational inference. Within this exploratory framework, our analysis in treated (Group D) versus untreated aged (Group A) rats indicated a relative increase in small antral follicle (SAF) mural granulosa cells (mGCs) and an upregulation of TGF-β and BMP signaling pathways within this cell population. These exploratory data lead us to hypothesize that the PRP + PBMCs combination might modulate the ovarian microenvironment by concurrently targeting immune-inflammation and stromal remodeling. PBMCs, as a heterogeneous population, could contribute to this process through dual potential mechanisms: by secreting a repertoire of cytokines and growth factors [7]. that may synergize with PRP, and by exerting immunomodulatory effects that could temper the chronic low-grade inflammatory state associated with aging [20, 21]. The latter is consistent with the downregulation of immune-response pathways we observed in SAF mGCs from Group D. Existing literature suggests that modulating the TGF-β/Smad pathway can inhibit fibrosis and support ovarian function [23]. Therefore, we speculate that the observed molecular changes and the alleviated differentiation block from SAF to LAF GCs in our rat model might be related to a combined effect of PRP + PBMCs on immunomodulation and the TGF-β/BMP signaling axis, potentially influencing stromal fibrosis. This proposed mechanism remains a hypothesis derived from our exploratory animal data and requires direct experimental validation.

Chronic low-grade inflammation is increasingly recognized as a pivotal contributor to the decline of ovarian function with age. The accumulation of pro-inflammatory mediators (e.g., IL-6, TNF-α) and dysregulation of immune cells (e.g., macrophages, Tregs) are thought to disrupt the follicular niche, accelerating atresia and depleting the ovarian reserve, thereby creating a self-perpetuating cycle of aging and immune dysfunction [24]. Consistent with this concept, our exploratory scRNA-seq data from the aged rat model showed a pronounced upregulation of immune and antigen-presentation pathways in granulosa cells from untreated controls (Group A). Intriguingly, in rats receiving the PRP + PBMCs combination (Group D), these inflammatory gene signatures were comparatively attenuated. This observation aligns with the known potential of PBMCs to exert immunomodulatory effects. As a heterogeneous population containing lymphocytes and monocytes, PBMCs could, in theory, help recalibrate the local immune milieu by modulating cytokine secretion (e.g., suppressing IL-6, TNF-α) and influencing immune cell dynamics (e.g., promoting Treg activity) [25, 26]. Thus, the immunomodulatory capacity of PBMCs represents a plausible, biologically grounded component of our overarching mechanistic hypothesis. Combined with their potential to secrete pro-angiogenic and reparative factors (e.g., VEGF, TGF-β) that may synergize with PRP-derived components [7]. PBMCs could contribute to creating a more supportive microenvironment for follicular development. Taken together, these exploratory findings support the hypothesis that the PRP + PBMCs combination might act, in part, by ameliorating the age-associated pro-inflammatory state of the ovary, alongside influencing stromal remodeling pathways as previously discussed. The relative contribution of immunomodulation versus other mechanisms to any potential clinical effect remains to be determined.

The normal progression from small antral follicle (SAF) to large antral follicle (LAF) stages is critical for follicular development, and its disruption—through impaired granulosa cell differentiation, hormone synthesis, or cell–cell communication—can lead to developmental arrest or atresiat [14, 27, 28]. In the context of our exploratory rat model, the observed molecular and cellular changes following PRP + PBMCs treatment appear consistent with a potential restoration of this developmental process. Specifically, our data indicated enhanced cellular communication between SAF and LAF granulosa cell clusters and an alleviation of the differentiation block in pseudotime trajectory analysis following the combination intervention. These findings are congruent with the concurrent upregulation of TGF-β/BMP pathways and downregulation of inflammatory signatures discussed earlier, as both improved intercellular crosstalk and a modified extracellular matrix environment could facilitate granulosa cell maturation. Thus, the improved follicular counts and hormone levels observed in Group D rats are associated with these coordinated changes at the single-cell level. Together, these exploratory results support a unifying hypothesis wherein PRP + PBMCs co-injection may promote a more permissive ovarian microenvironment, potentially facilitating key steps in folliculogenesis that are compromised with age.

In summary, by integrating the adjusted clinical observations with exploratory data from our animal model, we propose a working hypothesis. We speculate that intraovarian PRP + PBMCs co-injection might exert its potential effects by modifying the aged ovarian niche—specifically through attenuating immunoinflammation [24–26]. and influencing extracellular matrix remodeling [23]—thereby potentially enhancing intercellular communication [28] and facilitating the differentiation of granulosa cells during follicular development [27] This proposed mechanism provides a plausible biological framework that could explain the association observed in our clinical cohort between the PRP + PBMCs treatment and a higher yield of oocytes and MII oocytes. Any potential downstream benefit on oocyte quality or fertilization rates remains speculative and requires direct confirmation in future studies.In conclusion, this hypothesis-generating study provides preliminary, non-randomized clinical data alongside exploratory mechanistic insights from an animal model, investigating the potential of combined PRP and PBMCs therapy for advanced-age DOR. Our findings suggest that this combined approach may be associated with an increased retrieval of oocytes compared to PRP monotherapy after statistical adjustment [4], and that its mechanism of action might involve coordinated immunomodulatory and stromal-remodeling effects. However, the study’s major limitations—including its retrospective design, risk of selection bias, small sample size, and the exploratory nature of the mechanistic arm—preclude definitive conclusions regarding efficacy or causality. The clinical pregnancy and live birth outcomes remain unclear. These limitations underscore that our results do not establish clinical utility but rather highlight the need for and inform the design of a rigorous, prospective, randomized controlled trial. Future research should prioritize such a trial to definitively assess the safety and efficacy of this combination therapy. Further mechanistic studies in translationally relevant models are also warranted to validate the proposed hypotheses and elucidate the precise cellular targets.

Beyond the biological rationale, the methodological approach investigated here possesses inherent practical attributes that are relevant to translational considerations, should its efficacy be established in future trials. The preparation of both PRP and PBMCs relies on a straightforward, autologous blood draw and centrifugation protocol, bypassing the complexities, high costs, and ethical considerations often associated with allogeneic stem cell therapies. This technical simplicity and self-sourcing nature could potentially enhance its accessibility and reduce procedural barriers. In the broader context of assisted reproduction, which imposes a significant financial and time burden on patients and healthcare systems [29]. any intervention that proves effective in improving ovarian response might contribute to reducing the overall cycles and costs required to achieve pregnancy. However, these potential socioeconomic benefits remain entirely contingent upon the demonstration of robust clinical efficacy and safety in rigorous prospective studies.

This study is subject to several important limitations that must be considered when interpreting the findings. First and foremost, the clinical component is a retrospective, non-randomized cohort analysis with a limited sample size. Despite employing Inverse Probability of Treatment Weighting (IPTW) to adjust for key measured confounders, the risk of residual confounding and selection bias cannot be eliminated. Unobserved variables, such as detailed lifestyle factors, nuanced medication histories, or subtle differences in ovarian stimulation management, may have influenced the outcomes. Furthermore, the sample size constrains statistical power, increasing the risk of both type I and type II errors and rendering the study underpowered to detect differences in critical outcomes such as fertilization rates, embryo quality, and clinical pregnancy. Second, the animal and single-cell transcriptomic components are explicitly exploratory and have significant translational constraints. A major limitation is the use of allogeneic, G-CSF-mobilized PBMCs from young male donors in the aged female rat model. This paradigm differs fundamentally from the autologous clinical scenario in aged women, introducing potential immunologic and sex-hormone related confounds that limit direct extrapolation. Consequently, while the scRNA-seq findings provide valuable hypotheses regarding ovarian microenvironment modulation, they cannot be directly mapped to human pathophysiology and require validation in more translationally relevant models.Collectively, these limitations underscore that our findings are preliminary and hypothesis-generating. They do not establish efficacy but rather highlight the clear need for a well-powered, prospective, randomized controlled trial to definitively assess the clinical utility and safety of the PRP + PBMCs combination.

Conclusion

In summary, this retrospective, hypothesis-generating study provides preliminary, adjusted observational data suggesting an association between intraovarian PRP + PBMCs co-injection and increased oocyte yield in advanced-age women with DOR. Complementary exploratory data from a rat model offer a plausible, but unvalidated, biological hypothesis involving immunomodulation and stromal remodeling. However, the study's design precludes causal inference, and the mechanistic findings have limited direct translatability. Therefore, these results should not be interpreted as evidence of clinical efficacy but rather as a compelling justification for conducting a rigorous, prospective randomized controlled trial to definitively evaluate the therapeutic potential and safety of this combined approach.

Abbreviations

AMH

Anti-Müllerian Hormone

AFC

Antral Follicle Count

ART

Assisted Reproductive Technology

BMP

Bone Morphogenetic Protein

COS

Controlled Ovarian Stimulation

DEGs

Differentially Expressed Genes

FSH

Follicle-Stimulating Hormone

GCs

Granulosa Cells

GO

Gene Ontology

hCG

Human Chorionic Gonadotropin

IVF-ET

In Vitro Fertilization and Embryo Transfer

KEGG

Kyoto Encyclopedia of Genes and Genomes

LAF

Large Antral Follicle

NOA

Natural Ovarian Aging

PAF

Preantral Follicle

PBMCs

Peripheral Blood Mononuclear Cells

PRP

Platelet-Rich Plasma

SAF

Small Antral Follicle

scRNA-seq

Single-Cell RNA Sequencing

TGF-β

Transforming Growth Factor Beta

Authors’ contributions

HB.W., Z.Z., and LH.R. contributed equally as co-first authors. HB.W and Z.Z. participated in clinical trials, animal experiments, manuscript writing, and experimental design. LH.R.and Z.Z. participated in animal experiments, data curation, and manuscript writing. YY.Y.,HB.W.,and LL.L. are co-corresponding authors who supervised experimental design, clinical diagnosis and treatment, and finalized the manuscript. LL.Y. and GP.L. collected clinical data and information and performed data curation. All authors approved the final manuscript.

Funding

The project was supported by the Scientific and Technological Innovation Major Base of Guangxi(No.No.2022–36-Z05/GXSWBX202401); the Qinzhou Municipal Scientific Research and Technology Development Program (Contract No.:20223630);the Qinzhou Municipal Scientific Research and Technology Development Program (Contract No.:20242420);the Self-funded Research Project, Guangxi Health Commission (Contract No. Z-N20241675).

Data availability

The clinical and statistical summary data supporting the findings of this study are available within the article and its supplementary materials. The raw single-cell RNA sequencing data generated during the rat study will be deposited in the Gene Expression Omnibus (GEO) repository upon manuscript acceptance and will be publicly accessible. Prior to that, the data are available from the corresponding author on reasonable request.

Declarations

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.