Candidate microRNA networks in cyclophosphamide-associated ovarian injury and premature ovarian insufficiency: a structured narrative review.

Chemotherapy-induced ovarian insufficiency involves a complex interplay of apoptotic, inflammatory, and oxidative pathways. Among these, miRNAs act as post-transcriptional regulators that determine whether a GC will survive or undergo apoptosis after exposure to gonadotoxic agents such as CYC. However, the available evidence is heterogeneous. Only a subset of candidate miRNAs has been evaluated directly in CYC or active-metabolite models, whereas others derive from human POI cohorts, granulosa-cell apoptosis models, ovarian aging, cisplatin injury, high-fat/high-sugar oxidative-stress models, or regenerative/exosome studies [ 11 , 27 , 70 , 71 ]. Therefore, the miRNA pattern discussed below should be interpreted as a candidate injury-response network rather than a validated CYC-specific signature. Direct CYC-related evidence is strongest for miR-15b, let-7a, miR-122-5p, miR-22-3p, whereas miR-10a, miR-21, and miR-144-5p are supported mainly by preclinical therapeutic or exosome-delivery models in chemotherapy-associated ovarian injury, although even among these, the experimental context differs substantially across studies. In contrast, miR-23a/27a, miR-146a/146b-5p, miR-483-5p, and miR-15a provide relevant mechanistic or chemotherapy-related context, but they should not be presented as uniformly CYC-induced without direct validation in CYC-exposed ovarian cells or tissues. Importantly, the biological source of each miRNA signal varies across studies. Some experiments evaluated granulosa-like cell lines or isolated GCs, whereas others used whole ovarian tissue, serum, plasma, or extracellular vesicles. Therefore, miRNA expression changes should be interpreted according to the analyzed matrix and the likely contributing cell type. Whole-ovary and circulating miRNA profiles cannot be assumed to represent granulosa-cell-specific regulation unless supported by cell-specific validation. This is particularly relevant for circulating miRNAs, because a substantial fraction of extracellular miRNAs may circulate in Argonaute2-containing ribonucleoprotein complexes rather than inside vesicles [ 56 ]. In CYC-associated ovarian injury models, several miRNAs have been implicated in granulosa-cell apoptosis, impaired autophagy, oxidative stress, inflammatory signaling, and defective follicular support. The most directly supported pro-injury candidates include miR-15b and miR-122-5p. miR-15b is upregulated after CYC exposure in mouse ovarian GCs and contributes to reduced α-Klotho expression, impaired autophagy, increased oxidative stress, and activation of TGF-β/SMAD signaling [ 20 ]. miR-122-5p is increased in ovarian tissue-derived exosomes from CYC/busulfan (BUS)-induced POI mice and promotes apoptosis of CYC-damaged GCs by targeting BCL9, a regulator of Wnt/β-catenin-related survival signaling [ 19 ]. Protective or restorative candidates include let-7a, miR-10a, miR-21, miR-22-3p, and miR-144-5p. let-7a is downregulated after exposure of neonatal mouse ovaries to 4-hydroperoxycyclophosphamide (4-HC) and its replacement reduces chemotherapy-induced ovarian apoptosis [ 27 ]. miR-10a-enriched amniotic fluid stem cell-derived exosomes attenuate chemotherapy-induced granulosa-cell apoptosis and follicular atresia, with miR-10a showing a dominant protective effect in this system [ 21 ]. miR-21 has mainly been evaluated as a therapeutic cargo in chemotherapy-induced ovarian damage models, where miR-21-overexpressing mesenchymal stem cells or extracellular vesicles reduce granulosa-cell apoptosis through PTEN/PDCD4 or PTEN/caspase-3-related mechanisms [ 22 , 23 ]. miR-22-3p is reduced in CYC-induced POF/POI models and its restoration improves ovarian weight, hormone profiles, follicle counts, and apoptosis-related endpoints through CMKLR1 or the c-fos/MALAT1/miR-22-3p/STAT1 axis [ 24 , 63 ]. Finally, BMSC-derived exosomal miR-144-5p improves ovarian function after CYC-induced ovarian failure by reducing granulosa-cell apoptosis through PTEN targeting [ 64 ]. Other miRNAs should be discussed more cautiously. miR-23a/27a promote apoptosis in human GCs through XIAP and SMAD5-related mechanisms, supporting their relevance to granulosa-cell vulnerability, but the available studies are not CYC-specific [ 66 , 72 ], miR-146a is linked to inflammatory signaling and granulosa-cell apoptosis in POF-related models through IRAK1/TRAF6/NF-κβ-related pathways, whereas miR-146b-5p is protective in a high-fat/high-sugar oxidative-stress POF model; neither should be described as directly induced by CYC in ovarian cells unless supported by additional CYC-specific evidence [ 60 , 67 , 68 ]. miR-483-5p has been implicated in cisplatin-induced POI by targeting FKBP4, supporting chemotherapy-related ovarian vulnerability, but it should not be presented as CYC-specific [ 69 ]. Similarly, miR-15a is involved in Tripterygium glycoside-induced granulosa-cell senescence through Lats1/Hippo-YAP/TAZ signaling, but this is not equivalent to CYC-induced miR-15b regulation [ 65 ]. Together, these miRNAs converge on pathways that regulate granulosa-cell survival, apoptosis, oxidative stress, autophagy, inflammatory responses, and follicular support (Fig.  1 ). Nevertheless, the current evidence supports a candidate network of CYC/chemotherapy-associated ovarian injury rather than a definitive conserved CYC-specific miRNA signature. Fig. 1 Representative overview of ovarian miRNAs and CYC-associated miRNA dysregulation. Under physiological conditions, ovarian miRNAs such as miR-21, miR-145, miR-143, miR-26b, miR-181a, and the let-7 family participate in key processes of follicular growth, oocyte maturation, and female reproductive physiology. Following CYC exposure, miRNA dysregulation may alter pathways involved in follicular development/survival (e.g., miR-21, miR-10a, and miR-22-3p), apoptosis/autophagy (e.g., miR-15b, miR-122-5p, and contextually miR-23a/27a), and oxidative stress or hormonal dysfunction (e.g., contextually miR-146a, miR-483-5p, and therapeutically miR-144-5p). Because the evidence is heterogeneous, this figure distinguishes direct CYC-related evidence from contextual or therapeutic miRNA evidence Representative overview of ovarian miRNAs and CYC-associated miRNA dysregulation. Under physiological conditions, ovarian miRNAs such as miR-21, miR-145, miR-143, miR-26b, miR-181a, and the let-7 family participate in key processes of follicular growth, oocyte maturation, and female reproductive physiology. Following CYC exposure, miRNA dysregulation may alter pathways involved in follicular development/survival (e.g., miR-21, miR-10a, and miR-22-3p), apoptosis/autophagy (e.g., miR-15b, miR-122-5p, and contextually miR-23a/27a), and oxidative stress or hormonal dysfunction (e.g., contextually miR-146a, miR-483-5p, and therapeutically miR-144-5p). Because the evidence is heterogeneous, this figure distinguishes direct CYC-related evidence from contextual or therapeutic miRNA evidence miR-15b is among the best-supported CYC-associated miRNAs. In a CYC-induced POF mouse model and in mouse ovarian GCs, CYC increased miR-15b expression and reduced α-Klotho expression. α-Klotho is involved in antioxidant defense, autophagy, and cellular protection. Mechanistically, miR-15b-mediated α-Klotho suppression was associated with reduced autophagy-related proteins, increased ROS, decreased SOD and ATP, and activation of downstream TGF-β1/SMAD signaling. These findings support a model in which CYC-induced miR-15b contributes to granulosa-cell oxidative injury and impaired autophagy, thereby promoting ovarian dysfunction [ 20 ]. miR-122-5p has direct support from a CYC/BUS-induced POI model. In ovarian tissue-derived exosomes from POI mice, miR-122-5p was significantly increased. Functional experiments showed that miR-122-5p promoted apoptosis of CYC-damaged primary ovarian GCs by targeting BCL9, a component involved in Wnt/β-catenin-related survival signaling. Exosomes from the POI ovary aggravated granulosa-cell apoptosis, whereas exosomes from control ovaries or miR-122-5p inhibitor conditions attenuated apoptosis. Thus, miR-122-5p is best described as an exosomal pro-apoptotic candidate in CYC-associated ovarian injury rather than as a validated human biomarker of CYC-induced POI [ 19 ]. miR-23a and miR-27a are relevant to granulosa-cell apoptosis, but the evidence should be framed as contextual rather than CYC-specific. In human GCs, miR-23a and miR-27a promote apoptosis by targeting SMAD5 and activating FasL/Fas/caspase-8/caspase-3 signaling. Separately, miR-23a has been reported as differentially expressed in plasma from patients with POF and can promote granulosa-cell apoptosis by decreasing XIAP expression and increasing cleaved caspase-3. These findings support their role in granulosa-cell apoptotic regulation, but they do not demonstrate direct induction by CYC exposure [ 66 , 72 ]. miR-146a should be presented as an inflammation-related and POF-associated miRNA rather than a confirmed CYC-induced ovarian miRNA. In POF-related human granulosa-cell studies, miR-146a is upregulated and promotes granulosa-cell apoptosis by targeting IRAK1 and TRAF6, molecules involved in TLR/NF-κβ-related signaling. The classical inflammatory biology of miR-146a/b is supported by evidence that miR-146a is NF-κβ-inducible and regulates IRAK1/TRAF6 in immune cells. However, extrapolation from immune-cell biology to CYC-exposed ovarian tissue should be cautious [ 60 , 68 ]. In contrast, miR-146b-5p has been studied in a high-fat/high-sugar-induced oxidative-stress POF model, where its overexpression attenuated ovarian granulosa-cell aging through inhibition of p38-MAPK14 and reduced γH2A.X phosphorylation. This provides relevant oxidative-stress context but not direct CYC evidence [ 67 ]. miR-483-5p should not be described as elevated in CYC-treated ovaries based on the currently reviewed evidence. The uploaded primary study evaluates cisplatin-induced POI, where miR-483-5p aggravates ovarian injury by targeting FKBP4. FKBP4 overexpression can attenuate apoptosis in cisplatin-treated GCs, supporting a role for miR-483-5p in chemotherapy-related ovarian vulnerability. However, this evidence is cisplatin-specific and should be distinguished from CYC-induced injury [ 69 ]. let-7a has direct support in a CYC-metabolite model. In postnatal day 3 mouse ovaries exposed to 4-HC, let-7a was among the most markedly downregulated miRNAs. Delivery of a let-7a mimic reduced chemotherapy-induced ovarian apoptosis and decreased expression of cell-death-related genes. These data support let-7a as a candidate protective miRNA in early follicular ovarian injury induced by active CYC metabolites. However, further in vivo validation is needed to determine whether let-7a replacement preserves long-term follicular development and fertility [ 27 ]. miR-21 is a well-established anti-apoptotic miRNA in periovulatory GCs, where depletion of miR-21 increases granulosa-cell apoptosis and reduces ovulation [ 34 ]. In chemotherapy-induced ovarian damage models, the strangest evidence is therapeutic rather than endogenous CYC regulation. miR-21-overexpressing mesenchymal stem cells reduced apoptosis in phosphoramide mustard-damaged GCs and improved ovarian structure, follicle counts, estradiol levels, and FSH levels in CYC-induced POF rats through downregulation of PTEN and PDCD4 [ 22 , 74 ]. AFMSC-derived extracellular vesicles enriched in miR-21 also improved follicle counts, AMH levels, estrous cyclicity, and fertility-related outcomes while reducing PTEN and caspase-3; inhibition of miR-21 attenuated these benefits [ 23 ]. Therefore, miR-21 should be described as a therapeutic anti-apoptotic cargo in chemotherapy-induced ovarian injury rather than as definitively downregulated by CYC [ 22 – 34 ]. miR-10a appears to have a protective role in chemotherapy-associated ovarian injury, but its mechanism should be stated cautiously. In 4-HC-exposed neonatal mouse ovaries, miR-10a was reduced together with other miRNAs such as let-7a [ 27 ]. In a separate chemotherapy-induced POF model, amniotic fluid stem cell-derived exosomes protected GCs and ovarian follicles, and miR-10a-enriched exosomes or miR-10a delivery recapitulated much of the protective effect. The proposed targets included apoptosis-related and inflammatory mediators such as Bim, IRAK1, and TRAF6, rather than a uniformly validated CYC-specific DNA-repair mechanism. Thus, miR-10a is best framed as a protective exosomal miRNA that attenuates chemotherapy-induced apoptosis and follicular atresia [ 21 , 27 ]. BMSC-derived exosomal miR-144-5p has direct therapeutic evidence in CYC-induced ovarian failure. In CYC-treated rats and CYC-damaged GCs, exosomal miR-144-5p decreased granulosa-cell apoptosis by targeting PTEN. BMSC or BMSC-derived exosome transplantation improved estrous cyclicity, follicle counts, E2 and AMH levels, and reduced FSH and LH. Inhibition of miR-144-5p attenuated the anti-apoptotic effect, supporting a causal contribution of this miRNA cargo. This evidence supports miR-144-5p as a candidate exosome-mediated therapeutic molecule, but not yet as a clinically validated treatment [ 64 ]. miR-22-3p has both circulating human POI and CYC-associated preclinical support. Plasma miR-22-3p is decreased in Han Chinese patients with POF and shows modest discrimination between POF and controls, with an inverse relationship with FSH [ 73 ]. In CYC-induced POF mice, miR-22-3p is reduced in ovarian tissue, whereas CMKLR1 is increased; miR-22-3p overexpression or CMKLR1 knockdown improves hormone levels, ovarian weight, follicle counts, and apoptosis-related endpoints [ 63 ]. In another POI model, c-fos and MALAT1 reduce miR-22-3p availability, leading to increased STAT1 expression and impaired ovarian function; miR-22-3p suppresses STAT1 and participates in this ceRNA regulatory network [ 24 ]. These findings support miR-22-3p as a candidate protective regulator in POI/CYC-associated ovarian injury. The experimental evidence supporting these candidate miRNAs, including model type, CYC or injury protocol, tissue/cell source, miRNA measurement methods, targets, pathways, functional outcomes, and putative affected cell types, is summarized in Table  1 . Table 1 Experimental evidence supporting candidate miRNAs involved in cyclophosphamide/chemotherapy-associated ovarian injury and related POI models miRNA Direction in source model Evidence context Experimental model CYC or injury protocol Tissue/cell or sample source miRNA measurement / validation methods Validated or proposed target(s) Main pathway affected Functional effect Putative cell type affected Ref. miR-15b Upregulated after CYC Direct CYC evidence CYC-induced POF mouse model; isolated mouse ovarian GCs; α-Klotho-deficient mice used as mechanistic comparator CYC 70 mg/kg i.p., followed by 30 mg/kg every 2 days for 2 weeks Mouse ovarian GCs; whole ovary qPCR for miRNA, qPCR/Western blot for target and pathway proteins; oxidative stress, SOD, ATP, and autophagy assays α-Klotho/Kl α-Klotho-TGF-β/SMAD; autophagy; oxidative stress CYC increased miR-15b and reduced α-Klotho, impairing autophagy and ROS scavenging; miR-15b overexpression reproduced oxidative stress and autophagy dysfunction GCs [ 20 ] let-7a Downregulated after 4-HC exposure Direct CYC-metabolite evidence Postnatal day 3 mouse ovaries cultured ex vivo 4-HC 20 µM for 1 h and/or 24 h Whole neonatal mouse ovary, enriched in primordial follicles; granulosa/stromal/oocyte compartments inferred Rodent TLDA miRNA cards; custom TLDA validation; let-7a mimic delivery; TUNEL; apoptosis-related gene expression Proposed targets include HMGA2 and STAT3; apoptosis-related genes Cell death/apoptosis regulation let-7a mimic reduced chemotherapy-induced ovarian apoptosis; apoptosis mainly observed in stromal and granulosa compartments, with oocytes also affected Whole neonatal ovary; apoptosis observed mainly in stromal and GCs, with oocytes also affected [ 27 ] miR-122-5p Upregulated in POI ovarian-derived exosomes and CYC-damaged ovarian tissue Direct CYC/BUS evidence CYC/BUS-induced POI mouse model; primary ovarian GCs; ovarian tissue-derived exosomes CYC 20 mg/kg/day i.p. plus BUS 6.75 mg/kg/day; in vitro granulosa-cell injury with CYC 2 µM for 24 h Ovarian tissue-derived exosomes; primary ovarian GCs High-throughput miRNA sequencing; qRT-PCR; flow cytometry; luciferase assay; Western blot/IHC for BCL9 BCL9 Wnt/β-catenin-related survival signaling Exosomal miR-122-5p was increased in POI ovarian exosomes; miR-122-5p mimic promoted granulosa-cell apoptosis, whereas miR-122-5p inhibitor attenuated CYC-induced apoptosis GCs; exosome-mediated ovarian microenvironment [ 19 ] miR-22-3p Downregulated in CYC-induced POF ovarian tissue Direct CYC evidence CYC-induced POF mouse model CYC-induced POF model; lentiviral miR-22-3p, sh-CMKLR1 and oe-CMKLR1 injected into bilateral ovaries Whole ovarian tissue RT-qPCR; dual-luciferase reporter assay; Western blot; ELISA; H&E follicle counting; TUNEL; Bax/Bcl-2 assessment CMKLR1 Chemerin/CMKLR1-related inflammatory / apoptotic signaling miR-22-3p overexpression or CMKLR1 silencing improved E2/AMH/FSH profile, ovarian weight and follicle counts, and reduced atretic follicles and ovarian-cell apoptosis Ovarian tissue; granulosa-cell apoptosis assessed [ 63 ] miR-22-3p Downregulated in POI mouse ovarian tissue and CYC-treated KGN cells Direct CYC/BUS plus in vitro CYC evidence CYC/BUS-induced POI mouse model; CYC-induced KGN granulosa-like cell model; human bPOI granulosa-cell microarray dataset CYC 120 mg/kg plus BUS 12 mg/kg i.p.; KGN cells treated with CYC 250µM for 48 h Mouse ovarian tissue; KGN granulosa-like cells; human granulosa-cell dataset GEO microarray analysis; RT-qPCR; Western blot; dual-luciferase assay; RIP; RNA pull-down; ChlP; ELISA; H&E; TUNEL; flow cytometry; EdU STAT1, regulated through c-fos/MALAT1/miR-22-3p/STAT1 ceRNAaxis STAT1-mediated inflammatory/apoptosis signaling c-fos/MALAT1 activation reduced miR-22-3p and aggravated POI phenotype; miR-22-3p suppressed STAT1 and partially rescued KGN proliferation/apoptosis phenotype Granulosa-like KGN cells; ovarian tissue [ 24 ] miR-10a Downregulated after 4-HC in PND3 ovaries; enriched in AFSC-derived exosomes Direct chemotherapy/CYC-associated evidence PND3 mouse ovary exposed to 4-HC; CYC-treated mice; nitrogen mustard-damaged mouse GCs; AFSC-derived exosomes 4-HC 20 µM in PND3 ovary model; CYC-treated mice; nitrogen mustard 3 µM used in vitro as phosphoramide mustard-like injury Whole neonatal ovary; mouse GCs; AFSC-derived exosomes TLDA/qPCR in 4-HC model; exosome RNA analysis; miRNA knockdown/delivery; liposomal miRNA delivery; apoptosis and follicle atresia assays Proposed targets include Bim, IRAK1, and TRAF6 Anti-apoptotic signaling; inflammatory/apoptotic gene regulation AFSC-derived exosomes protected damaged GCs; miR-10a delivery reduced apoptosis and follicular atresia in CYC-treated mice, with miR-10a showing dominant protective effect GCs; ovarian follicles [ 21 , 27 ] miR-21 Therapeutic overexpression; endogenous CYC-specific downregulation not clearly established Direct chemotherapy-induced ovarian damage evidence, mainly therapeutic Rat model of chemotherapy-induced ovarian damage; phosphoramide mustard-damaged GCs; MSCs overexpressing miR-21 Rat POF model induced by CYC; phosphoramide mustard used in vitro to mimic ovarian chemotherapy microenvironment Rat ovarian tissue; GCs; MSCs/miR-21-MSCs Lentiviral miR-21 overexpression; qRT-PCR; Western blot; TUNEL; hormone assays; follicle counts; estrous-cycle assessment PTEN, PDCD4 PI3K/Akt-related survival; apoptosis inhibition miR-21-overexpressing MSCs reduced granulosa-cell apoptosis, increased ovarian weight/follicle counts/E2, decreased FSH, and downregulated PTEN/PDCD4 GCs; ovarian follicles [ 22 ] miR-21 EV-mediated therapeutic enrichment Chemotherapy-induced POD model; therapeutic/exosome evidence Rat POD model; AFMSC-derived extracellular vesicles; EVs modified with miR-21 mimics/inhibitors CYC 40 mg/kg i.p. single dose to induce POD; AFMSC/EV treatment 21 days after CYC injection; intraovarian injection of AFMSCs, EVs, or EVs carrying miR-21 mimic/inhibitor Ovarian tissue; AFMSC-derived EVs EV isolation/characterization; miR-21 mimic/inhibitor transfection; AMH, follicle counts, caspase-3 and PTEN protein assessment; estrous cycle and mating trials PTEN, caspase-3 pathway Anti-apoptotic signaling AFMSC-derived EVs and EVs carrying miR-21 mimics restored AMH, follicle counts, estrous cyclicity and fertility-related outcomes; miR-21 inhibition blunted benefit Granulosa-cell/follicular compartment inferred [ 23 ] miR-144-5p Downregulated in POF-related contexts; delivered by BMSC exosomes Direct CYC evidence, therapeutic/exosome CYC-induced POF rat model; CYC-damaged GCs cocultured with BMSC-derived exosomes CYC 50 mg/kg i.p. on day 1, then 8 mg/kg/day for 14 days Rat ovarian tissue; GCs; BMSC-derived exosomes qRT-PCR; exosome coculture; flow cytometry; Western blot; hormone assays; follicle counts PTEN PTEN/PI3K/Akt pathway BMSC-derived exosomal miR-144-5p reduced CYC-damaged granulosa-cell apoptosis, restored estrous cycle, improved E2/AMH, lowered FSH/LH, and increased follicle counts GCs; ovarian follicles [ 64 ] miR-23a Upregulated in plasma of POF patients; proapoptotic in GCs POI/POF mechanistic evidence; not CYC-specific Plasma from POF patients and controls; cultured human GCs No CYC exposure Plasma; human ovarian GCs miRNA microarray; qRT-PCR; GO/KEGG analysis; Hoechst apoptosis assay; Western blot XIAP Caspase-3-mediated apoptosis miR-23a decreased XIAP mRNA/protein, increased cleaved caspase-3 and promoted granulosa-cell apoptosis GCs [ 72 ] miR-23a/miR-27a Upregulated in POF-related circulating data; experimentally overexpressed in human GCs POI/POF mechanistic evidence; not CYC-specific Human GCs from IVF/ET patients; 293T luciferase system No CYC exposure Human GCs qRT-PCR; Hoechst apoptosis assay; Western blot; dual-luciferase reporter assay SMAD5 SMAD5/FasL-Fas/caspase-8/caspase-3 pathway miR-23a/27a overexpression increased apoptosis; inhibitors reduced apoptosis; SMAD5 knockdown activated Fas/FasL and caspase signaling GCs [ 66 ] miR-146a Upregulated in plasma and GCs from POF patients POI/POF mechanistic evidence; not CYC-specific Human plasma and GCs from POF patients; COV434 granulosa-cell model No CYC exposure Plasma; human GCs; COV434 cells qPCR; gain/loss-of-function; dual-luciferase assay; Western blot; TUNEL/flow cytometry IRAK1, TRAF6 TLR/NF-κβ-related signaling; caspase cascade miR-146a promoted granulosa-cell apoptosis through IRAK1/TRAF6 and caspase activation; inhibition reduced apoptosis GCs; inflammatory signaling context [ 68 ] miR-146a/b NF-κβ-inducible in immune cells Mechanistic inflammatory context; not ovarian and not CYC-specific Human monocytes / immune-cell models No CYC exposure Monocytes miRNA expression profiling; promoter analysis; 3’UTR luciferase reporter assays IRAK1, TRAF6 TLR/NF-κβ negative-feedback loop Supports biological plausibility that miR-146 links NF-κβ activation to IRAK1/TRAF6 regulation; ovarian extrapolation should be cautious Immune/inflammatory cells [ 60 ] miR-146b-5p Downregulated in HFHS-induced oxidative-stress POF model; therapeutic overexpression protective Oxidative-stress POI model; not CYC-specific HFHS-induced POF mouse model; mouse ovarian GCs No CYC; high-fat/high-sugar diet and HFHS-treated OGCs Mouse ovarian GCs; ovarian tissue qRT-PCR; Northern blot; luciferase assay; Western blot; co-IP; oxidative stress assays; hormone assessment p38-MAPK14 Dab2ip/Ask1/p38-MAPK; γH2A. X phosphorylation miR-146b-5p overexpression reduced OGC aging and oxidative stress and attenuated POF-like changes GCs [ 67 ] miR-483-5p Upregulated in cisplatin-induced POI and POI serum; not demonstrated as CYC-specific Chemotherapy-related, cisplatin-specific evidence Cisplatin-induced POI mouse/rat model; primary GCs and KGN cells; POI patient serum Cisplatin, not CYC Ovarian tissue; serum; GCs; KGN cells iTRAQ; TargetScan; qRT-PCR; Western blot; dual-luciferase assay; in situ hybridization; ELISA; TUNEL FKBP4 Steroid/stress-response signaling; apoptosis miR-483-5p increased ovarian sensitivity to cisplatin; FKBP4 overexpression reduced apoptosis induced by cisplatin and miR-483-5p Oocytes and GCs [ 69 ] miR-15a Upregulated in Tripterygium glycoside-induced ovarian toxicity Drug-induced ovarian toxicity context; not CYC-specific Tripterygium glycoside-induced POF model; mouse ovarian GCs No CYC; Tripterygium glycosides Mouse ovarian GCs; ovarian tissue qPCR; Western blot; luciferase reporter assay; MTT; SA-β-gal; flow cytometry Lats1 Hippo-YAP/TAZ signaling miR-15a overexpression inhibited granulosa-cell proliferation and induced senescence; TG-induced toxicity involved miR-15a/Lats1 signaling GCs [ 65 ] Abbreviations AFMSC amniotic fluid mesenchymal stem cell, AFSC amniotic fluid stem cell, AMH anti-Müllerian hormone, BMSC bone marrow mesenchymal stem cell, BUS busulfan, ChlP chromatin immunoprecipitation, CYC cyclophosphamide, EV extracellular vesicle, FC fold-change, FSH follicle-stimulating hormone, GC granulosa cell, H&E hematoxylin and eosin, HFHS high-fat/high-sugar, i.p. intraperitoneal, OGC ovarian granulosa cell, PM phosphoramide mustard, PND3 postnatal day 3, POD premature ovarian dysfunction, POF premature ovarian failure, POI premature ovarian insufficiency, qRT-PCR quantitative reverse transcription polymerase chain reaction, RIP RNA immunoprecipitation, ROS reactive oxygen species, TG Tripterygium glycosides, TLDA TaqMan low-density array, TUNEL terminal deoxynucleotidyl transferase dUTP nick-end labeling Experimental evidence supporting candidate miRNAs involved in cyclophosphamide/chemotherapy-associated ovarian injury and related POI models Abbreviations AFMSC amniotic fluid mesenchymal stem cell, AFSC amniotic fluid stem cell, AMH anti-Müllerian hormone, BMSC bone marrow mesenchymal stem cell, BUS busulfan, ChlP chromatin immunoprecipitation, CYC cyclophosphamide, EV extracellular vesicle, FC fold-change, FSH follicle-stimulating hormone, GC granulosa cell, H&E hematoxylin and eosin, HFHS high-fat/high-sugar, i.p. intraperitoneal, OGC ovarian granulosa cell, PM phosphoramide mustard, PND3 postnatal day 3, POD premature ovarian dysfunction, POF premature ovarian failure, POI premature ovarian insufficiency, qRT-PCR quantitative reverse transcription polymerase chain reaction, RIP RNA immunoprecipitation, ROS reactive oxygen species, TG Tripterygium glycosides, TLDA TaqMan low-density array, TUNEL terminal deoxynucleotidyl transferase dUTP nick-end labeling

The marked inter-individual variability in ovarian susceptibility to CYC suggests that genetic and epigenetic background may influence the ovarian response to alkylating injury. Potential determinants include differences in CYC bioactivation and detoxification, DNA-damage response capacity, antioxidant defenses, baseline follicular reserve, and regulation of apoptosis or follicular activation pathways [ 6 , 61 ]. In this context, miRNA-related variants are biologically plausible modifiers because they may alter miRNA biogenesis, mature miRNA sequences, miRNA expression, or miRNA-mRNA binding [ 43 , 97 ]. To date, however, evidence directly linking miRNA gene polymorphisms to CYC-induced POI is very limited. Some candidate-gene studies have explored variants in specific miRNAs or miRNA-processing pathways in idiopathic POI or ovarian reserve phenotypes, but these data cannot be directly extrapolated to CYC-exposed women. For example, polymorphisms affecting miRNA-mediated regulation of reproductive genes may influence POI susceptibility, but whether such variants modify ovarian sensitivity to CYC metabolites has not been established [ 97 ]. Similarly, broader genetic studies of POI have highlighted pathways related to germ-cell development, meiosis, folliculogenesis, DNA repair, and ovarian reserve, whereas validated SNPs in miRNA genes specifically predicting CYC-induced gonadotoxicity have not yet been incorporated into clinical risk models [ 6 , 61 ]. This gap is clinically relevant. A future prognostic strategy could combine clinical variables, such as age, baseline AMH, antral follicle count, cumulative CYC dose, and treatment indication, with pharmacogenomic markers of CYC metabolism, DNA-repair variants, and longitudinal miRNA profiles [ 1 , 6 , 13 – 15 , 77 ]. Such an integrated model may eventually identify women at increased risk of ovarian reserve decline and support individualized counseling regarding oocyte or embryo cryopreservation, ovarian tissue cryopreservation, or pharmacologic ovarian protection. At present, this approach should be considered exploratory and not ready for clinical decision-making. Future studies should evaluate whether miRNA-related SNPs, expression quantitative trait loci, or variants in miRNA-binding sites modify the relationship between CYC exposure and ovarian outcomes. Ideally, these studies should be prospective, include well-defined CYC dose exposure, use standardized POI definitions, and integrate genomic, transcriptomic, and circulating miRNA data.

This work was designed as a structured narrative review aiming to synthesize and critically appraise current evidence regarding the role of miRNAs in CYC-induced POI. A comprehensive literature search was performed comprising articles between January 2015 and December 2024 using PubMed, Scopus, and Web of Science databases. A targeted update was subsequently performed during the editorial revision process to incorporate recent reviews and highly relevant studies published in 2025–2026. The following combinations of keywords and MeSH terms were used: “microRNA” OR “miRNA” AND “ovary” OR “granulosa cell” OR “folliculogenesis” AND “cyclophosphamide” OR “chemotherapy” OR “alkylating agent” AND “premature ovarian failure” OR “premature ovarian insufficiency” OR “diminished ovarian reserve”. Additional references were identified through manual screening of the bibliographies of relevant papers. We reviewed original articles (in vitro, animal, or human studies) assessing miRNA expression, regulation, or function in ovarian tissue or GCs following CYC exposure. We included mechanistic studies that elucidated signaling pathways (e.g., transforming growth factor beta (TGF-β)/small mothers against decapentaplegic (SMAD), phosphoinositide 3-kinase (PI3K)/protein kinase B (Akt), nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κβ), Wingless-related integration site (Wnt)/β-catenin)) modulated by specific miRNAs during chemotherapy-induced injury. We only included articles in English. Studies focused exclusively on non-CYC chemotherapeutic agents, or non-CYC ovarian dysfunction models were excluded from the core CYC evidence synthesis; however, they were considered as contextual evidence when they provided mechanistic information on granulosa-cell apoptosis oxidative stress, DNA-damage response, extracellular vesicle-mediated miRNA transfer, or POI-related pathways. Studies lacking molecular miRNA data and conference abstracts without full text were excluded. Two reviewers independently screened titles and abstracts, retrieved full texts, and extracted the following variables: experimental model (species, cell type), CYC dose and duration, analytical techniques (quantitative reverse transcription polymerase chain reaction (qRT-PCR), ribonucleic acid (RNA) sequencing (RNA-seq), microarray), identified miRNAs, validated targets, affected signaling pathways, and functional outcomes (apoptosis, oxidative stress, follicle counts, hormonal levels). When studies did not involve direct CYC exposure, their evidence level was classified as contextual rather than direct. Findings were synthesized narratively and organized by biological process (apoptosis, oxidative balance, autophagy, deoxyribonucleic acid (DNA) repair, and steroidogenesis). Recurrent miRNAs across models were highlighted to delineate candidate regulatory nodes rather than definitive clinical signatures. The structure and critical interpretation of this review were guided by the Scale for the Assessment of Narrative Review Articles (SANRA) [ 59 ].

Beyond direct DNA adducts, CYC provokes epigenetic remodeling of ovarian cells. CYC-induced DNA damage and oxidative stress may influence both transcriptional and post-transcriptional regulatory programs in ovarian cells. In GCs, acute CYC exposure has been shown to reduce EZH2 and H3K27me3, resulting in aberrant transcriptional activation of genes involved in transcriptional regulation and apoptosis [ 9 ]. In parallel, inflammatory pathways can regulate miRNA transcription, as illustrated by NF-κβ-dependent induction of miR-146a in immune-cell models; however, this evidence should be considered contextual rather than direct ovarian CYC evidence [ 60 ]. Beyond their post-transcriptional regulatory roles, miRNAs also intersect with broader epigenetic mechanisms. The miR-29 family, for instance, modulates global DNA methylation by targeting DNMT3 family members and TET enzymes in early embryonic development, supporting the concept of miRNA-epigenetic crosstalk rather than a CYC-specific ovarian mechanism [ 42 ]. More broadly, miRNA expression and activity may be influenced by DNA methylation, histone modifications, RNA-binding proteins, competing endogenous RNAs, and RNA editing, adding an additional layer of complexity to miRNA interpretation in ovarian injury models [ 43 ]. Recent findings have demonstrated that CYC exposure alters histone methylation patterns in both GCs and oocytes. In oocytes, CYC exposure has been associated with long-term alterations in epigenetic modifications and maternal-factor transcription [ 44 ]. In GCs, acute CYC exposure reduces EZH2 and H3K27me3 rather than increasing H3K27me3, and this loss of H3K27me3 is associated with transcriptional overactivation of genes involved in apoptosis and transcriptional regulation [ 9 ]. The latter study identified the EZH2-H3K27me3 axis as a key epigenetic regulator of granulosa-cell apoptosis after CYC treatment. Taken together, these epigenetic alterations establish a reinforcing loop in which chemotherapy-induced oxidative stress reshapes chromatin architecture and miRNA expression, ultimately driving ovarian cells toward a persistently pro-apoptotic transcriptional program [ 6 , 9 , 44 , 61 , 62 ]. Therefore, the available evidence supports a candidate injury-response miRNA network rather than a validated CYC-specific signature: ▪ Direct CYC-associated pro-injury candidates include miR-15b and miR-122-5p. miR-15b is linked to α-Klotho suppression, impaired autophagy, ROS accumulation, and TGF-β/SMAD activation, whereas miR-122-5p promotes apoptosis of CYC-damaged GCs through BCL9/Wnt-related signaling [ 19 , 20 ]. ▪ Protective or restorative miRNA-based mechanisms include let-7a replacement after 4-HC exposure, miR-10a delivery through AFSC-derived exosomes, miR-21-overexpressing MSCs or EVs, miR-22-3p restoration, and BMSC-derived exosomal miR-144-5p [ 21 – 24 , 27 , 63 , 64 ]. These should be interpreted as preclinical protective or therapeutic strategies rather than uniformly reduced endogenous miRNAs after CYC. ▪ Direct CYC-associated pro-injury candidates include miR-15b and miR-122-5p. miR-15b is linked to α-Klotho suppression, impaired autophagy, ROS accumulation, and TGF-β/SMAD activation, whereas miR-122-5p promotes apoptosis of CYC-damaged GCs through BCL9/Wnt-related signaling [ 19 , 20 ]. ▪ Protective or restorative miRNA-based mechanisms include let-7a replacement after 4-HC exposure, miR-10a delivery through AFSC-derived exosomes, miR-21-overexpressing MSCs or EVs, miR-22-3p restoration, and BMSC-derived exosomal miR-144-5p [ 21 – 24 , 27 , 63 , 64 ]. These should be interpreted as preclinical protective or therapeutic strategies rather than uniformly reduced endogenous miRNAs after CYC. Contextual miRNAs, including miR-23a/27a, miR-146a/146b-5p, miR-483-5p, and miR-15a, provide relevant mechanistic information on granulosa-cell apoptosis, inflammatory signaling, oxidative stress, or chemotherapy-associated ovarian vulnerability, but they should not be presented as direct CYC-induced miRNAs unless supported by CYC-specific ovarian evidence [ 60 , 65 – 69 ]. This dysregulation disrupts key survival pathways, including α-Klotho/TGF-β-SMAD/autophagy-ROS signaling, BCL9/Wnt-related survival signaling, PTEN/PI3K/Akt signaling, CMKLR1-and STAT1-related inflammatory/apoptotic pathways, leading to a feed-forward loop of oxidative stress and apoptosis. Although delayed ovarian dysfunction after CYC is biologically plausible and may relate to persistent follicular depletion, stromal injury, and epigenetic alterations in oocytes, persistent miRNA reprogramming after cessation of CYC exposure remains insufficiently demonstrated and should be considered a hypothesis requiring longitudinal validation [ 44 , 70 , 71 ].

CYC-induced POI is a complex and multifactorial process involving DNA damage, oxidative stress, apoptosis, impaired repair responses, stromal remodeling, vascular injury, and disruption of the ovarian follicular niche. miRNAs are plausible epigenetic mediators within this process because they regulate several pathways involved in granulosa-cell survival, follicular atresia, inflammation, oxidative balance, autophagy, and tissue repair. The available evidence supports a candidate miRNA network associated with CYC- and chemotherapy-related ovarian injury rather than a validated CYC-specific signature. Direct CYC evidence is strongest for miR-15b and miR-122-5p as pro-injury candidates, and for let-7a and miR-22-3p as protective or restorable miRNAs in specific experimental settings. miR-10a, miR-21, and miR-144-5p are mainly supported as therapeutic or exosome-delivered protective cargos in preclinical chemotherapy-associated ovarian injury models. In contrast, miR-23a/27a, miR-146a/146b-5p, miR-483-5p, and miR-15a should be interpreted as contextual evidence from POI, granulosa-cell apoptosis, oxidative-stress, cisplatin, or non-CYC ovarian injury models. However, this pattern should not yet be interpreted as a validated diagnostic, prognostic, or therapeutic signature. Differences in species, CYC exposure protocols, biological matrices, sample timing, and miRNA detection methods limit direct translation to clinical practice. Circulating miRNAs may eventually complement established ovarian reserve markers, but they should currently be considered exploratory biomarkers. Their ovarian specificity, optimal sampling window, diagnostic thresholds, and incremental value beyond AMH, antral follicle count, gonadotropins, and menstrual pattern remain to be established. Similarly, miRNA mimics, antagomirs, MSC-derived exosomes, and engineered extracellular vesicle platforms represent promising experimental strategies, but current evidence is largely preclinical. No miRNA-targeted therapy has yet been validated for prevention or treatment of CYC-induced POI in humans. Future studies integrating longitudinal clinical cohorts, standardized ovarian reserve assessment, pharmacogenomic data, extracellular vesicle characterization, and human-relevant experimental models may clarify whether miRNAs can be used for risk stratification, mechanistic monitoring, or targeted ovarian protection in women exposed to alkylating agents. Until then, miRNA-based approaches should be framed as hypothesis-generating and translationally promising but not clinically established.

Several limitations should be considered when interpreting the evidence summarized in this review: First, the available literature is heterogeneous in terms of species, experimental model, CYC dose, dosing schedule, timing of sample collection, tissue or biofluid analyzed, and miRNA quantification method. These differences limit direct comparison across studies and may explain inconsistencies in miRNA direction or magnitude [ 9 , 19 – 24 , 27 , 63 , 64 , 73 – 76 ]. Second, many studies use immortalized granulosa-like cells, isolated GCs, or whole-ovary rodent tissue. Although these models are useful for mechanistic exploration, they do not fully reproduce the human ovarian niche. Whole-ovary studies capture multicellular damage but cannot identify the precise cellular source of miRNA changes unless paired with cell-specific, single-cell, or spatial validation. Conversely, cell-line studies offer mechanistic precision but lack stromal, endothelial, immune, and oocyte–granulosa interactions [ 9 , 19 , 20 , 24 ]. Third, human data remain scarce and mostly observational. Existing studies are often limited by small sample size, heterogeneous causes of diminished ovarian reserve or POI, variable prior treatments, absence of pre-CYC baseline samples, and inconsistent adjustment for age, cumulative CYC dose, baseline ovarian reserve, autoimmune disease activity, and systemic inflammation. These limitations prevent causal inference and preclude the definition of clinically actionable miRNA thresholds [ 56 , 73 – 77 ]. Fourth, therapeutic evidence is predominantly preclinical. Although miRNA mimics, antagomirs, MSC-derived exosomes, and engineered vesicle-delivery platforms have shown encouraging biological effects in experimental models, there is insufficient evidence to conclude that they preserve fertility, prevent POI, or restore ovarian function in women exposed to CYC. Reproductive endpoints such as pregnancy and live birth are rarely evaluated, and long-term safety data are lacking [ 21 – 23 , 64 , 82 – 87 , 90 – 93 , 95 , 98 ]. Fifth, human-mouse orthology and miRNA pleiotropy remain major translational barriers. Conservation of mature miRNA sequences does not ensure conservation of target sites or downstream functional networks. Moreover, a single miRNA may regulate multiple genes across different tissues, increasing the risk of off-target effects when systemic delivery is used [ 43 , 79 , 83 , 90 – 93 ]. First, the available literature is heterogeneous in terms of species, experimental model, CYC dose, dosing schedule, timing of sample collection, tissue or biofluid analyzed, and miRNA quantification method. These differences limit direct comparison across studies and may explain inconsistencies in miRNA direction or magnitude [ 9 , 19 – 24 , 27 , 63 , 64 , 73 – 76 ]. Second, many studies use immortalized granulosa-like cells, isolated GCs, or whole-ovary rodent tissue. Although these models are useful for mechanistic exploration, they do not fully reproduce the human ovarian niche. Whole-ovary studies capture multicellular damage but cannot identify the precise cellular source of miRNA changes unless paired with cell-specific, single-cell, or spatial validation. Conversely, cell-line studies offer mechanistic precision but lack stromal, endothelial, immune, and oocyte–granulosa interactions [ 9 , 19 , 20 , 24 ]. Third, human data remain scarce and mostly observational. Existing studies are often limited by small sample size, heterogeneous causes of diminished ovarian reserve or POI, variable prior treatments, absence of pre-CYC baseline samples, and inconsistent adjustment for age, cumulative CYC dose, baseline ovarian reserve, autoimmune disease activity, and systemic inflammation. These limitations prevent causal inference and preclude the definition of clinically actionable miRNA thresholds [ 56 , 73 – 77 ]. Fourth, therapeutic evidence is predominantly preclinical. Although miRNA mimics, antagomirs, MSC-derived exosomes, and engineered vesicle-delivery platforms have shown encouraging biological effects in experimental models, there is insufficient evidence to conclude that they preserve fertility, prevent POI, or restore ovarian function in women exposed to CYC. Reproductive endpoints such as pregnancy and live birth are rarely evaluated, and long-term safety data are lacking [ 21 – 23 , 64 , 82 – 87 , 90 – 93 , 95 , 98 ]. Fifth, human-mouse orthology and miRNA pleiotropy remain major translational barriers. Conservation of mature miRNA sequences does not ensure conservation of target sites or downstream functional networks. Moreover, a single miRNA may regulate multiple genes across different tissues, increasing the risk of off-target effects when systemic delivery is used [ 43 , 79 , 83 , 90 – 93 ]. Future research should prioritize: (i) longitudinal human cohorts with pre-treatment and serial post-CYC samples; (ii) standardized definitions of POI and ovarian reserve decline; (iii) harmonized CYC exposure reporting, including cumulative dose and dosing schedule; (iv) parallel assessment of AMH, antral follicle count, gonadotropins, menstrual pattern, and fertility outcomes; (v) rigorous pre-analytical control of circulating miRNA studies, including hemolysis assessment and normalization strategy; (vi) distinction between total circulating, extracellular vesicle-enriched, and protein-bound miRNA fractions; (vii) validation in human-relevant systems such as primary ovarian cells, organoids, single-cell sequencing, and spatial transcriptomics; and (viii) safety-focused studies of ovarian-targeted delivery systems [ 1 , 13 – 15 , 56 , 74 , 77 , 80 , 81 , 90 – 93 ]. Multi-omics integration, including miRNA, mRNA, methylome, proteome, and extracellular vesicle cargo analysis, may help identify robust regulatory nodes. However, computational predictions and artificial intelligence-assisted network analyses should be experimentally validated before being proposed as clinically useful biomarkers or therapeutic targets [ 43 , 79 – 81 , 90 – 93 ].

The identification of miRNA networks involved in granulosa-cell apoptosis, oxidative stress, autophagy, DNA-damage responses, and stromal–vascular remodeling has generated interest in miRNA-based ovarian protection. However, this field remains predominantly preclinical. Therefore, therapeutic strategies discussed in this section should be interpreted as experimental approaches rather than clinically established interventions for CYC-induced POI. The elucidation of miRNA networks governing GC fate has opened a novel avenue for therapeutic intervention in chemotherapy-associated ovarian injury and CYC-induced POI. Rather than solely preventing DNA damage, the focus now extends toward modulating post-transcriptional regulators to preserve follicular function. Strategies include (i) synthetic miRNA mimics to restore depleted protective miRNAs, (ii) antagomirs to silence overexpressed pro-apoptotic miRNAs, and (iii) biological carriers such as MSC-derived extracellular vesicles/exosomes for miRNA delivery [ 21 – 23 , 82 – 89 ]. These strategies should not be considered substitutes for established fertility preservation methods such as oocyte, embryo, or ovarian tissue cryopreservation. At present, they are best understood as mechanistic tools and candidate ovarian-protective platforms requiring further validation in human-relevant models. Synthetic miRNA mimics and inhibitors are chemically engineered oligonucleotides designed to normalize aberrant miRNA expression. ▪ miR-21-based strategies have demonstrated the ability to reduce granulosa-cell apoptosis in chemotherapy-induced ovarian damage models, mainly when delivered through MSCs or extracellular vesicles, by modulating PTEN, PDCD4 and caspase-3-related apoptotic pathways [ 22 , 23 ]. ▪ miR-10a mimics or miR-10a-enriched exosomes attenuate chemotherapy-induced granulosa-cell apoptosis and follicular atresia, potentially through regulation of pro-apoptotic and inflammatory mediators such as Bim, IRAK1 and TRAF6 [ 21 ]. ▪ Conversely, miR-122-5p inhibition attenuates CYC-induced granulosa-cell apoptosis by preserving BCL9/Wnt-related survival signaling [ 19 ]. ▪ miR-21-based strategies have demonstrated the ability to reduce granulosa-cell apoptosis in chemotherapy-induced ovarian damage models, mainly when delivered through MSCs or extracellular vesicles, by modulating PTEN, PDCD4 and caspase-3-related apoptotic pathways [ 22 , 23 ]. ▪ miR-10a mimics or miR-10a-enriched exosomes attenuate chemotherapy-induced granulosa-cell apoptosis and follicular atresia, potentially through regulation of pro-apoptotic and inflammatory mediators such as Bim, IRAK1 and TRAF6 [ 21 ]. ▪ Conversely, miR-122-5p inhibition attenuates CYC-induced granulosa-cell apoptosis by preserving BCL9/Wnt-related survival signaling [ 19 ]. Although highly specific in vitro, these molecules face challenges related to stability, off-target effects, and targeted delivery. Nanoparticle conjugation or encapsulation within biodegradable polymers has been proposed to improve oligonucleotide stability, biodistribution, and tissue delivery, although ovarian-specific targeting remains unresolved [ 82 – 87 , 90 – 93 ]. These findings support biological plausibility, but the evidence remains largely derived from cell-based systems and rodent models. Several barriers limit clinical translation. First, each miRNA can regulate multiple transcripts, increasing the risk of off-target effects in ovarian and non-ovarian tissues. Second, chemically modified oligonucleotides may differ in stability, immunogenicity, and intracellular uptake. Third, ovarian-specific delivery remains unresolved, particularly because the ovary contains multiple relevant compartments, including follicles, stromal cells, vascular cells, and immune cells. Fourth, the therapeutic window after CYC exposure is uncertain; early intervention may target acute oxidative and DNA-damage responses, whereas delayed intervention may occur after irreversible follicular depletion or stromal fibrosis. Consequently, miRNA mimics and antagomirs should be described as mechanistic tools and candidate therapeutic platforms rather than validated ovarian-protective treatments. Future studies should define dose, timing, biodistribution, ovarian cell-type specificity, durability of effect, and reproductive endpoints, including ovulation, pregnancy, and live birth, rather than relying only on follicle counts or hormone changes. Exosomes (30–150 nm vesicles secreted by virtually all cell types) act as natural miRNA carriers with high biocompatibility and potentially low immunogenicity [ 39 , 94 ]. Stem cell-derived extracellular vesicles/exosomes carrying or enriched in protective miRNAs, notably miR-21, miR-10a, and miR-144-5p, have shown promising preclinical efficacy in chemotherapy-induced ovarian damage models, including CYC-related models [ 21 – 23 , 64 , 87 – 89 , 95 ]. In rodent models, administration of stem cell-derived extracellular vesicles/exosomes restored estradiol production, normalized FSH/luteinizing hormone (LH) ratios, increased follicular density, and reduced granulosa-cell apoptosis through pathways involving PTEN, PDCD4, caspase-3, Bcl-2, or related survival signaling. These effects were partly reduced in selected studies when exosomal RNA/miRNA content was disrupted or when specific miRNAs were inhibited, supporting a contributory role of miRNA transfer without excluding effects from other exosomal cargo [ 21 – 23 , 64 , 95 ]. Nevertheless, most evidence comes from animal and cell-based studies, and the magnitude and durability of benefit in humans remain unknown. The therapeutic effects of MSC-derived exosomes are probably not explained by a single miRNA. Instead, exosomes contain a complex cargo of miRNAs, proteins, lipids, and other non-coding RNAs that may act simultaneously on apoptosis, angiogenesis, mitochondrial function, inflammation, and extracellular matrix remodeling [ 39 , 87 – 89 , 94 , 95 ]. This pleiotropy may be beneficial in a multicellular injury process such as CYC-induced POI, but it also complicates attribution of causality to specific miRNAs. Clinical translation remains limited. Although regenerative and exosome-based approaches are being explored in ovarian dysfunction, there is currently no validated clinical trial evidence supporting miRNA-targeted therapy specifically for CYC-induced POI [ 87 – 89 , 95 ]. Important unresolved issues include vesicle source, donor variability, manufacturing standards, purity, potency assays, miRNA cargo characterization, dosing, route of administration, ovarian retention, long-term safety, and potential oncologic concerns in cancer survivors. Therefore, exosome-based strategies should be presented as promising but experimental platforms requiring rigorous preclinical validation and well-designed clinical trials. Circulating miRNAs have attracted interest as minimally invasive biomarkers because they are relatively stable in biofluids and may be transported within extracellular vesicles, lipoprotein complexes, or Argonaute-containing ribonucleoprotein particles [ 56 , 96 ]. In the context of ovarian injury, circulating miRNAs may reflect cellular stress, apoptosis, inflammatory signaling, extracellular vesicle release, or tissue remodeling within the ovarian microenvironment. However, circulating miRNAs are not ovary-specific, and circulating profiles may be influenced by systemic inflammation, immune disease activity, chemotherapy indication, renal or hepatic function, and pre-analytical variables [ 56 , 80 , 81 ]. Several miRNAs implicated in granulosa-cell apoptosis, oxidative stress, and ovarian reserve decline have also been evaluated in circulating samples. For example, plasma miR-22-3p has been reported to be decreased in patients with POF and modestly associated with FSH, while miR-21-5p and miR-100-5p have been associated with AMH-defined ovarian reserve categories. Urinary exosomal miR-4516 has also been proposed as a candidate POI biomarker, and follicular-fluid exosomal miRNA profiles have been associated with oocyte quality [ 73 – 76 ]. Nevertheless, most available studies are observational and include heterogeneous populations, and few, if any, longitudinal human studies specifically evaluate women before and after CYC exposure. Therefore, these circulating miRNAs should be interpreted as candidate biomarkers of ovarian reserve decline rather than validated tools for early detection of CYC-induced gonadotoxicity. The main translational appeal of circulating miRNAs is their potential to complement, rather than replace, established ovarian reserve markers such as AMH, antral follicle count, menstrual pattern, and gonadotropin concentrations [ 1 , 13 – 15 , 74 ]. In theory, serial miRNA assessment before, during, and after CYC exposure could help identify molecular changes preceding overt POI. However, this hypothesis remains unproven. At present, there is insufficient evidence to define diagnostic thresholds, optimal sampling time points, or clinically actionable miRNA panels for predicting CYC-induced POI. The integrated mechanistic and translational relationships among CYC exposure, ovarian injury pathways, candidate miRNA networks, circulating biomarker candidates, and emerging miRNA-based therapeutic strategies are summarized in Fig.  2 . Table  2 summarizes circulating and extracellular miRNAs with potential biomarker relevance for ovarian reserve decline or POI. Table  3 summarizes experimental and emerging therapeutic strategies involving miRNAs in chemotherapy-associated ovarian injury models. Fig. 2 Integrated mechanistic and translational model of miRNA involvement in CYC-induced ovarian toxicity. Exposure to CYC generates ovarian injury through oxidative stress, DNA damage, granulosa-cell apoptosis, impaired autophagy, and follicular depletion, ultimately contributing to decreased ovarian reserve. Candidate miRNAs implicated in these processes include miR-15b and miR-122-5p as pro-injury mediators and let-7a and miR-22-3p as protective/restorable miRNAs in direct CYC-related models. Additional miRNAs, such as miR-10a, miR-21, and miR-144-5p, are mainly supported as therapeutic or exosome-delivered protective cargos in chemotherapy-associated ovarian injury models. The figure also highlights the potential translational relevance of circulating or extracellular miRNAs as candidate biomarkers and summarizes emerging miRNA-based therapeutic approaches, including miRNA mimics, antagomirs, and stem cell-derived extracellular vesicles/exosomes Integrated mechanistic and translational model of miRNA involvement in CYC-induced ovarian toxicity. Exposure to CYC generates ovarian injury through oxidative stress, DNA damage, granulosa-cell apoptosis, impaired autophagy, and follicular depletion, ultimately contributing to decreased ovarian reserve. Candidate miRNAs implicated in these processes include miR-15b and miR-122-5p as pro-injury mediators and let-7a and miR-22-3p as protective/restorable miRNAs in direct CYC-related models. Additional miRNAs, such as miR-10a, miR-21, and miR-144-5p, are mainly supported as therapeutic or exosome-delivered protective cargos in chemotherapy-associated ovarian injury models. The figure also highlights the potential translational relevance of circulating or extracellular miRNAs as candidate biomarkers and summarizes emerging miRNA-based therapeutic approaches, including miRNA mimics, antagomirs, and stem cell-derived extracellular vesicles/exosomes Table 2 Circulating and extracellular miRNAs with potential biomarker relevance for ovarian reserve decline, POI, or chemotherapy-associated ovarian injury miRNA or miRNA profile Direction / finding Sample source Study population or model Association with ovarian outcome CYC-specific? Main limitations Ref. miR-22-3p Downregulated in POF Plasma Han Chinese patients with POF and age/BMI matched controls; discovery and validation cohorts Modest discrimination of POF from controls; inversely associated with FSH No Cross-sectional human study; not chemotherapy specific; modest diagnostic performance [ 73 ] miR-21-5p Altered abundance in abnormal ovarian reserve; best AUC for predicting decreased AMH in the study Whole blood/ PAXgene-based miRNA profiling Women undergoing infertility treatment classified by AMH as low, normal, or high ovarian reserve Correlated with AMH-defined ovarian reserve categories No Infertility-clinic population; AMH-based classification; not CYC-exposed; requires external validation. [ 74 ] miR-100-5p Altered abundance; best AUC for predicting increased AMH in the study Whole blood/PAXgene-based miRNA profiling Women undergoing infertility treatment classified by AMH Candidate marker of abnormal ovarian reserve No Not POI-specific; not-chemotherapy-specific; clinical utility not established [ 74 ] AMH-associated circulating miRNA panel Multiple validated miRNAs altered in low-AMH or high-AMH groups; several were downregulated in low-AMH profiles Whole blood/ PAXgene-based miRNA profiling Women undergoing infertility treatment Associated with low AMH ovarian reserve category No Exploratory biomarker study; not CYC-specific; matrix differs from serum/plasma EV studies [ 74 ] miR-4516 Upregulated in POI and Turner associated POI; also increased in ovary of CYC/BUS-induced POI mouse model Urinary exosomes; mouse ovary Human POI cohorts and CYC/BUS-induced POI mice Candidate diagnostic marker of POI; associated with reduced STAT3 protein in mouse ovary Partially; animal CYC/BUS model, but human cohort not CYC-specific Requires validation across etiologies; urinary exosomal source may not be ovary-specific [ 75 ] miR-23a Differentially expressed in POF plasma; functionally pro-apoptotic in GCs Plasma; human GCs POF patients and normal cycling women; granulosa-cell experiments Promotes granulosa-cell apoptosis via XIAP/caspase-3 regulation No Biomarker and functional evidence are not CYC-specific; plasma source may reflect systemic biology; small discovery cohort [ 72 ] miR-23a/miR-27a Previously reported as elevated in plasma of POF patients; promote apoptosis in human GCs Human GCs; prior circulating context GCs from IVF/ET patients Promote apoptosis through SMAD5/FasL-Fas/caspase-3 signaling No Mechanistic granulosa-cell evidence, not a direct circulating CYC biomarker study; not longitudinal [ 66 ] miR-122-5p Increased in ovarian tissue-derived exosomes from POI mice Ovarian tissue-derived exosomes; primary GCs CYC/BUS induced POI mouse model; CYC-damaged primary GCs Promotes granulosa-cell apoptosis by targeting BCL9; miR-122-5p inhibition reduces apoptosis Yes, preclinical CYC/BUS Ovarian exosome model, not validated as circulating human biomarker; tissue-derived exosome signal may not reflect plasma/urine [ 19 ] Follicular-fluid exosomal miRNA profile 47 differentially expressed follicular-fluid exosomal miRNAs between high- and poor-oocyte-quality groups Human follicular fluid exosomes Women undergoing assisted reproduction classified by oocyte quality Candidate markers of oocyte quality and ovarian microenvironment No Oocyte-quality rather than POI/CYC study; mostly exploratory; pathway inference requires functional validation [ 76 ] miR-483-5p Increased in serum of POI patients and in cisplatin-induced POI models Serum; ovarian tissue; GCs/KGN cells POI patients and cisplatin-induced POI experimental models Aggravates cisplatin-induced ovarian injury by targeting FKBP4 No; cisplatin-specific Chemotherapy-related but not CYC-specific; limited generalizability to alkylating agents; small human serum sample [ 69 ] Circulating miRNAs in general Stable extracellular miRNAs may be carried by vesicles or Ago2-containing ribonucleoprotein complexes Plasma/serum Human plasma/serum methodological studies Supports feasibility of circulating miRNA biomarker development and highlights carrier heterogeneity Not applicable Pre-analytical and carrier heterogeneity; circulating signal is not tissue-specific; EV origin cannot be assumed without fractionation [ 56 ] Abbreviations AMH anti-Müllerian hormone, AUC area under the curve, BMI body mass index, BUS busulfan, CYC cyclophosphamide, EV extracellular vesicle, FSH follicle-stimulating hormone, GC granulosa cell, HAMH High AMH, IVF/ET in vitro fertilization/embryo transfer, KGN human granulosa-like tumor cell line, LAMH low AMH, NAMH normal AMH, NTA nanoparticle tracking analysis, POI premature ovarian insufficiency, POF premature ovarian failure, qRT-PCR quantitative reverse transcription polymerase chain reaction, ROC receiver operating characteristic, TEM transmission electron microscopy Circulating and extracellular miRNAs with potential biomarker relevance for ovarian reserve decline, POI, or chemotherapy-associated ovarian injury Abbreviations AMH anti-Müllerian hormone, AUC area under the curve, BMI body mass index, BUS busulfan, CYC cyclophosphamide, EV extracellular vesicle, FSH follicle-stimulating hormone, GC granulosa cell, HAMH High AMH, IVF/ET in vitro fertilization/embryo transfer, KGN human granulosa-like tumor cell line, LAMH low AMH, NAMH normal AMH, NTA nanoparticle tracking analysis, POI premature ovarian insufficiency, POF premature ovarian failure, qRT-PCR quantitative reverse transcription polymerase chain reaction, ROC receiver operating characteristic, TEM transmission electron microscopy Table 3 Experimental and emerging miRNA-based therapeutic strategies in chemotherapy-associated ovarian injury and related POI models Therapeutic approach Target miRNA(s) Evidence context Model and delivery strategy Proposed or validated mechanism Main functional outcome Translational interpretation Ref. let-7a mimic let-7a Direct active CYC-metabolite model PND3 mouse ovaries exposed to 4-HC; liposome-based let-7a mimic delivery Downregulation of cell-death-related gene expression; proposed targets include HMGA2/STAT3-related networks Reduced chemotherapy-induced ovarian apoptosis Candidate protective miRNA; requires in vivo fertility validation [ 27 ] miR-10a-enriched AFSC-derived exosomes or miR-10a delivery miR-10a Chemotherapy-induced ovarian damage model CYC-treated mice and damaged GCs; AFSC-derived exosomes or miRNA delivery Anti-apoptotic effect; proposed regulation of Bim and inflammatory/apoptotic mediators such as IRAK1/TRAF6 Reduced granulosa-cell apoptosis and follicular atresia Promising exosome-mediated protective cargo; preclinical only [ 21 ] miR-21-overexpressing MSCs miR-21 Chemotherapy-induced ovarian damage model CYC-induced POF rats; phosphoramide mustard-damaged GCs; transplantation of miR-21-MSCs PTEN and PDCD4 downregulation; reduced GC apoptosis Increased ovarian weight, follicle counts and E2; reduced FSH and TUNEL-positive GCs Therapeutic cargo evidence; does not prove endogenous miR-21 downregulation by CYC [ 22 ] AFMSC-derived EVs enriched or modified for miR-21 miR-21 Chemotherapy-induced POD model AFMSC-derived EVs; EVs carrying miR-21 mimics/inhibitors PTEN and caspase-3 modulation Restored AMH, follicle counts, estrous cycles and fertility-related outcomes; miR-21 inhibition reduced benefit Cell-free therapeutic candidate; preclinical only [ 23 ] Anti-miR-122-5p / miR-122-5p inhibition miR-122-5p Direct CYC/BUS POI evidence CYC/BUS-induced POI mouse model; CYC-damaged primary ovarian GCs; ovarian tissue-derived exosomes BCL9 preservation and Wnt/β-catenin-related survival signaling Reduced apoptosis of CYC-damaged GCs Candidate antagomir strategy; no human therapeutic evidence [ 19 ] miR-22-3p restoration miR-22-3p Direct CYC POI evidence CYC-induced POF mice; intraovarian lentiviral miR-22-3p or related constructs CMKLR1 suppression; reduced apoptosis Improved E2/AMH/FSH profile, ovarian weight, follicle counts, and apoptosis-related endpoints Candidate protective strategy; delivery and safety unresolved [ 63 ] Targeting c-fos/MALAT1/miR-22-3p/STAT1 network miR-22-3p axis CYC/BUS POI and CYC treated KGN evidence POI mice and CYC-treated KGN cells; sh-c-fos or sh-STAT1 interventions Restoration of miR-22-3p availability and STAT1 suppression Improved ovarian function and granulosa-like cell proliferation/ apoptosis phenotype Mechanistic target network; not yet a miRNA drug strategy [ 24 ] BMSC-derived exosomal miR-144-5p miR-144-5p Direct CYC POF evidence CYC-induced POF rats; CYC-damaged GCs; BMSC-derived exosome transplantation PTEN targeting; reduced granulosa-cell apoptosis Restored estrous cycle, E2/AMH, follicle counts; reduced FSH/LH and apoptosis Strong preclinical exosome-cargo evidence; no clinical validation [ 64 ] hUCMSC-derived exosomes Not single-miRNA-specific Direct CYC POI exosome evidence CYC-induced POI mice; hUCMSC-derived exosome transplantation Hippo pathway regulation and ovarian-cell proliferation Improved hormone levels, follicle number and reproductive outcomes Useful exosome platform evidence, but effect should not be attributed to a specific miRNA cargo unless validated [ 95 ] Abbreviations AFMSC amniotic fluid mesenchymal stem cell, BMSC bone marrow mesenchymal stem cell, CYC cyclophosphamide, EV extracellular vesicle, FSH follicle-stimulating hormone, hUCMSC human umbilical cord mesenchymal stem cell, MSC mesenchymal stem cell, PND3 postnatal day 3, POI premature ovarian insufficiency, POF premature ovarian failure Experimental and emerging miRNA-based therapeutic strategies in chemotherapy-associated ovarian injury and related POI models Abbreviations AFMSC amniotic fluid mesenchymal stem cell, BMSC bone marrow mesenchymal stem cell, CYC cyclophosphamide, EV extracellular vesicle, FSH follicle-stimulating hormone, hUCMSC human umbilical cord mesenchymal stem cell, MSC mesenchymal stem cell, PND3 postnatal day 3, POI premature ovarian insufficiency, POF premature ovarian failure

Rodent models have provided most mechanistic evidence for miRNA-mediated regulation in chemotherapy-associated ovarian injury. In CYC-induced POF mice, miR-15b upregulation contributes to α-Klotho suppression, impaired autophagy, increased oxidative stress, and TGF-β/SMAD activation in GCs [ 20 ]. In CYC/BUS-induced POI mice, ovarian tissue-derived exosomal miR-122-5p is increased and promotes granulosa-cell apoptosis by targeting BCL9 [ 19 ]. In CYC-induced POF mice, miR-22-3p restoration improves ovarian endocrine and histological outcomes through CMKLR1 suppression, whereas related CYC/BUS POI models implicate the c-fos/MALAT1/miR-22-3p/STAT1 axis [ 24 , 63 ]. Therapeutic delivery studies suggest that restoring protective miRNAs can attenuate chemotherapy-associated ovarian injury. let-7a mimic delivery reduces 4-HC-induced ovarian apoptosis in neonatal mouse ovaries [ 27 ]. Amniotic fluid stem cell-derived exosomal miR-10a reduces chemotherapy-induced granulosa-cell apoptosis and follicular atresia [ 21 ]. miR-21-overexpressing MSCs or miR-21-enriched extracellular vesicles improve ovarian structure and function in chemotherapy-induced ovarian damage models by modulating PTEN, PDCD4, and caspase-3-related apoptosis pathways [ 22 , 23 ]. BMSC-derived exosomal miR-144-5p improves ovarian function in CYC-induced POF rats through PTEN inhibition and reduced granulosa-cell apoptosis [ 64 ]. These findings are promising, but they remain preclinical and should not be interpreted as evidence of clinical efficacy in women exposed to CYC [ 19 , 21 – 23 , 64 ]. In vitro models allow direct manipulation of specific miRNAs and targets, but they cannot fully reproduce the multicellular ovarian niche. Mouse ovarian GCs exposed to CYC show increased miR-15b and reduced α-Klotho, with impaired autophagy and oxidative stress responses [ 20 ]. CYC-damaged primary GCs exposed to ovarian tissue-derived exosomes demonstrate the pro-apoptotic effect of miR-122-5p through BCL9 targeting [ 19 ]. KGN granulosa-like cells treated with CYC show reduced miR-22-3p and increased c-fos/MALAT1/STAT1 signaling, supporting a ceRNA-mediated mechanism in granulosa-like cells [ 24 ]. Phosphoramide mustard- or nitrogen mustard-based injury systems have also been used to mimic the ovarian chemotherapy microenvironment and to test miR-21-or miR-10a-mediated anti-apoptotic interventions [ 21 , 22 ]. In contrast, several mechanistic studies are informative but not CYC-specific. miR-23a/27a promote human granulosa-cell apoptosis through XIAP or SMAD5-related mechanisms [ 66 , 72 ]. miR-146a promotes apoptosis through IRAK1/TRAF6 and caspase-related pathways in POF-related granulosa-cell models [ 68 ]. miR-146b-5p attenuates granulosa-cell oxidative injury in HFHS-induced POF models by targeting p38-MAPK14 [ 67 ]. These studies should be used to explain candidate pathways rather than to infer CYC-specific regulation. Human data remain limited and should be interpreted cautiously. Several circulating or extracellular miRNAs have been associated with ovarian reserve, POI, oocyte quality, or chemotherapy-related ovarian injury, but few studies evaluate women longitudinally before and after CYC exposure. miR-22-3p is decreased in plasma from patients with POF and shows modest diagnostic discrimination, whereas miR-21-5p and miR-100-5p have been associated with AMH-defined ovarian reserve categories in women undergoing infertility treatment [ 73 , 74 ]. Urinary exosomal miR-4516 is increased in women with POI and in a CYC/BUS-induced POI mouse model, suggesting possible biomarker relevance, although this marker is not specific to CYC-induced POI [ 75 ]. Follicular fluid exosomal miRNA profiles have also been associated with oocyte quality, supporting the broader concept that extracellular miRNAs may reflect ovarian microenvironmental status [ 76 ]. At present, circulating miRNAs should be considered candidate biomarkers of ovarian reserve decline or ovarian stress, not validated tools for early detection of CYC-induced gonadotoxicity. Established clinical markers such as AMH, AFC, menstrual pattern, and gonadotropins remain the reference framework for ovarian reserve assessment [ 1 , 13 – 15 ]. In women receiving chemotherapy, baseline AMH has demonstrated clinical value for predicting chemotherapy-induced amenorrhea, whereas miRNA panels require longitudinal validation before they can be used for risk prediction or fertility-preservation decision-making [ 74 , 77 ]. Future studies should collect pre-treatment samples, serial post-CYC samples, standardized ovarian reserve outcomes, and harmonized miRNA measurements to determine whether miRNA changes precede, accompany, or merely reflect established ovarian reserve decline [ 78 ]. Because available evidence derives from heterogeneous experimental settings, Table  1 distinguishes direct CYC evidence from broader chemotherapy-related or POI/granulosa-cell mechanistic evidence. This distinction is essential because candidate miRNAs identified in human POI, ovarian aging, cisplatin injury, or non-CYC oxidative-stress models cannot be assumed to represent a conserved CYC-specific signature without direct validation.

Premature ovarian insufficiency (POI) is defined as the loss of ovarian function before the age of 40. Clinically, it is characterized by menstrual irregularity, hypoestrogenism, and persistently elevated gonadotropin levels [ 1 , 2 ]. The syndrome results from depletion or dysfunction of primordial follicles, culminating in infertility and long-term systemic effects such as osteoporosis and cardiovascular/metabolic risk [ 3 , 4 ]. Although the etiology of POI is multifactorial, iatrogenic forms secondary to chemotherapy or radiotherapy represent a growing clinical challenge as cancer survival improves [ 5 ]. Among gonadotoxic drugs, cyclophosphamide (CYC) (a nitrogen mustard alkylating agent) is particularly detrimental to the female gonad. CYC is extensively employed in oncologic protocols and in the management of autoimmune diseases such as systemic lupus erythematosus and vasculitis, as well as in immunosuppressive regimens [ 6 – 9 ]. Its cytotoxic mechanism involves DNA cross-linking and oxidative injury, leading to accelerated follicular atresia and irreversible loss of reproductive potential [ 10 – 12 ]. Clinical data indicate that cumulative alkylating-agent and chemotherapy exposure correlates with diminished ovarian reserve, reduced anti-Müllerian hormone (AMH) levels, and amenorrhea in a clinically relevant proportion of treated premenopausal women, depending on age, regimen, baseline ovarian reserve, and cumulative exposure [ 5 , 13 – 15 ]. Traditionally, the pathogenesis of CYC-induced POI was attributed to the direct destruction of oocytes and granulosa cells (GC) by its metabolites. However, this purely genotoxic model cannot fully explain the inter-individual variability in susceptibility or the partial reversibility observed in some patients. Emerging studies highlight a pivotal contribution of epigenetic regulators, particularly microRNAs (miRNAs), in determining cellular response to ovarian stress and chemotherapy-associated injury [ 9 , 16 – 24 ]. These small non-coding RNAs govern the balance of proliferation, differentiation, and apoptosis within the ovarian microenvironment by controlling transcriptional networks. Alterations in miRNA expression have been documented not only in natural aging and idiopathic POI but also in preclinical models of chemotherapy-associated ovarian damage [ 16 , 19 – 27 ]. Consequently, mapping the miRNA landscape in CYC-induced ovarian toxicity provides mechanistic insight and may identify predictive biomarkers and therapeutic targets. miRNAs are 18–24 nucleotide non-coding RNAs that regulate gene expression post-transcriptionally by guiding the RNA-induced silencing complex (RISC) to complementary sequences in target messenger RNAs, resulting in translational repression and/or mRNA decay [ 17 , 18 , 28 , 29 ]. Because a single miRNA can modulate multiple transcripts, miRNAs function as regulatory nodes that coordinate proliferation, differentiation, apoptosis, steroidogenesis, and cellular stress responses within the ovary [ 25 , 30 , 31 ]. In the ovarian follicle, miRNAs participate in bidirectional communication among oocytes, GCs, theca cells, and the surrounding follicular microenvironment. Experimental disruption of miRNA biogenesis through Dicer deletion or impaired miRNA processing produces infertility, anovulation, impaired folliculogenesis, and luteal defects, supporting the essential role of miRNA networks in ovarian homeostasis [ 28 , 32 , 33 ]. Several conserved miRNAs orchestrate follicular dynamics: ▪ miR-21 blocks apoptosis in mouse periovulatory GCs, supporting an anti-apoptotic role during the peri-ovulatory period; however, its validated targets appear context-dependent and should not be limited to PTEN or FasL in this physiological setting [ 34 ]. ▪ miR-145 suppresses mouse granulosa-cell proliferation by targeting activin receptor IB, whereas other follicular miRNAs regulate granulosa-cell proliferation and differentiation through activin/TGF-β-related pathways [ 35 – 37 ]. ▪ miR-26b promotes granulosa-cell apoptosis during follicular atresia by targeting ATM and modulating DNA-damage signaling [ 38 ]. ▪ miR-181a suppresses mouse granulosa-cell proliferation by downregulating activin receptor type IIA (ACVR2A) and attenuating activin/SMAD signaling [ 37 ]. ▪ The let-7 family contributes to follicular and reproductive regulation, although its specific role depends on the developmental stage, cell type, and experimental model [ 27 , 35 , 39 ]. ▪ miR-21 blocks apoptosis in mouse periovulatory GCs, supporting an anti-apoptotic role during the peri-ovulatory period; however, its validated targets appear context-dependent and should not be limited to PTEN or FasL in this physiological setting [ 34 ]. ▪ miR-145 suppresses mouse granulosa-cell proliferation by targeting activin receptor IB, whereas other follicular miRNAs regulate granulosa-cell proliferation and differentiation through activin/TGF-β-related pathways [ 35 – 37 ]. ▪ miR-26b promotes granulosa-cell apoptosis during follicular atresia by targeting ATM and modulating DNA-damage signaling [ 38 ]. ▪ miR-181a suppresses mouse granulosa-cell proliferation by downregulating activin receptor type IIA (ACVR2A) and attenuating activin/SMAD signaling [ 37 ]. ▪ The let-7 family contributes to follicular and reproductive regulation, although its specific role depends on the developmental stage, cell type, and experimental model [ 27 , 35 , 39 ]. Together, these molecules maintain a balance between follicular recruitment and dormancy. Therefore, chemotherapy-induced disruption of these regulatory networks may accelerate follicular atresia and mimic features of premature ovarian aging [ 35 , 40 , 41 ]. miRNAs also interact with other epigenetic regulators such as DNA methylation and histone modification. For example, the miR-29 family regulates DNA methyltransferase 3 (DNMT3), influencing global methylation patterns during early embryonic development [ 42 ]. More broadly, miRNA activity may be modulated by epigenetic mechanisms, including DNA methylation, histone modifications, RNA-binding proteins, competing endogenous RNAs, and RNA editing [ 43 ]. These reciprocal interactions between miRNAs and other epigenetic regulators provides a plausible mechanism through which CYC-induced oxidative and genotoxic stress may establish persistent pro-apoptotic transcriptional programs in ovarian cells, as supported by recent evidence that CYC alters EZH2/H3K27me3-dependent transcriptional repression in GCs and epigenetic marks in oocytes [ 9 , 43 , 44 ]. CYC is a pro-drug requiring hepatic cytochrome P450 activation to 4-hydroxy-CYC, which spontaneously converts to phosphoramide mustard and acrolein. Phosphoramide mustard forms DNA inter- and intra-strand cross-links that impair replication and transcription, while acrolein generates reactive oxygen species (ROS), lipid peroxidation, and glutathione depletion [ 7 , 8 , 11 , 45 – 51 ]. These metabolites reach the ovary through the bloodstream and may affect not only oocytes and GCs but also stromal fibroblasts, endothelial cells, theca-stromal compartments, and immune-related cells within the ovarian niche. Classically, CYC-induced POI has been explained by direct damage to oocytes and GCs. GCs are particularly relevant because they provide metabolic, paracrine, and endocrine support to the oocyte; their apoptosis can precipitate secondary oocyte loss and follicular atresia [ 9 , 31 , 49 ]. However, CYC-induced ovarian injury should be interpreted as a multicellular process rather than an isolated granulosa-cell or oocyte phenomenon. Stromal fibroblast activation may contribute to cortical fibrosis; endothelial apoptosis may reduce ovarian perfusion and oxygenation; and local inflammatory signaling may perpetuate oxidative stress and follicular niche dysfunction [ 11 , 45 – 47 , 52 ]. At the molecular level, CYC activates p53/BAX/caspase-dependent apoptosis, ROS-mediated NF-kB and MAPK signaling, and impaired DNA-repair responses involving ATM and BRCA-related pathways [ 9 , 38 , 46 , 53 – 55 ]. Current evidence strongly supports apoptosis as a major mechanism of CYC-induced primordial follicle and granulosa-cell injury, whereas premature follicle activation/burnout remains more controversial and model-dependent [ 50 , 54 ]. Persistent stress may also suppress adaptive autophagy through TGF-B/SMAD signaling and alter chromatin states in GCs and oocytes [ 9 , 20 , 44 ]. Accordingly, miRNA alterations detected in whole-ovary tissue, circulating samples, or extracellular vesicles may reflect a composite signal derived from multiple ovarian cell types, rather than a granulosa-cell-specific response. This is especially relevant for circulating miRNAs, because not all extracellular miRNAs are vesicle-associated; a substantial fraction may circulate in protein complexes such as Argonaute2-bound ribonucleoproteins [ 56 ]. This distinction is important when interpreting candidate miRNA signatures across in vitro, animal, and human studies. Recent reviews have addressed related but distinct aspects of this field. Chen et al. summarized the role of miRNAs in granulosa-cell proliferation, apoptosis, steroidogenesis, oxidative stress, and follicular atresia, with emphasis on granulosa-cell biology rather than CYC-specific ovarian toxicity [ 31 ]. Vassilaki et al. reviewed the broader interaction between miRNAs and oxidative stress in human fertility, including male infertility, polycystic ovarian syndrome, POI, and endometriosis, but did not focus on alkylating-agent-induced ovarian injury [ 57 ]. Shao et al. reviewed mesenchymal stem cell-based therapy for POI, including exosomal miRNA cargo, but their emphasis was regenerative therapy rather than the integration of miRNA signatures after CYC exposure/injury [ 58 ]. Therefore, the present review differs from these previous works by specifically integrating in vitro, animal, and human evidence on miRNA-mediated regulation in CYC-induced POI, with attention to CYC genotoxicity, ovarian multicellularity, candidate circulating biomarkers, and the translational limitations of miRNA- and exosome-based interventions. Because the available literature is heterogeneous, this review distinguishes direct CYC evidence from broader chemotherapy-related or POI/granulosa-cell contextual evidence.

The proposed miRNA pattern should be interpreted in the context of substantial methodological heterogeneity across studies. Available evidence integrates immortalized granulosa-like cell lines, primary granulosa-cell cultures, whole-ovary rodent models, extracellular vesicle experiments, and small human observational cohorts [ 19 – 24 , 27 , 64 , 73 – 76 ]. These models provide complementary information but are not biologically interchangeable. Cell-line studies allow mechanistic manipulation of specific miRNAs and targets, but they do not reproduce the full ovarian niche, including oocytes, theca cells, stromal fibroblasts, endothelial cells, immune cells, and extracellular matrix remodeling. Conversely, whole-ovary animal studies better capture tissue-level injury, but miRNA changes measured in total ovarian extracts cannot be attributed to a specific cell population without cell-specific validation [ 9 , 20 , 24 ]. CYC exposure protocols also differ substantially between studies. Some models use single high-dose exposure to induce acute ovarian injury, whereas others use repeated lower-dose schemes to approximate cumulative clinical exposure [ 9 , 19 – 24 , 63 , 64 ]. The timing of tissue collection after CYC administration may further influence miRNA profiles, because early changes may capture follicular depletion, stromal remodeling, fibrosis, or compensatory repair [ 9 , 54 , 70 , 71 ]. Therefore, differences in dose, dosing frequency, route of administration, and sampling time may partly explain the variability in reported miRNA direction and magnitude. Species differences represent another key limitation. Although several miRNAs are conserved between rodents and humans, conservation of the mature miRNA sequence does not necessarily imply conservation of all target sites, downstream messenger RNAs, or ovarian cell-type-specific functions. Human-mouse orthology is particularly relevant when inferring clinical significance from murine CYC models or when extrapolating therapeutic effects of miRNA mimics, antagomirs, or exosome cargo. Accordingly, recurrent miRNAs identified across species should be considered candidate nodes of a conserved injury-response network rather than validated human biomarkers or therapeutic targets [ 43 , 79 ]. Human evidence remains limited by small sample size, cross-sectional design, heterogeneous indications for CYC exposure, differences in age and baseline ovarian reserve, and variability in biological matrices, including serum, plasma, follicular fluid, whole blood, and extracellular vesicle-enriched fractions [ 56 , 73 – 77 ]. These limitations restrict causal inference and preclude the definition of clinically actionable thresholds. Longitudinal human studies with pre-treatment samples, serial post-CYC follow-up, standardized ovarian reserve markers, and harmonized miRNA quantification methods are needed to determine whether miRNA alterations precede accompany, or merely reflect established ovarian reserve decline [ 56 , 74 , 80 , 81 ].