First follicular fluid cf-mtDNA sequencing unveils ovarian stimulation-induced mutations impacting oocyte quality and IVF success

Abstract

Purpose

Cell-free mitochondrial DNA (cf-mtDNA) has emerged as a promising non-invasive method to predict embryo implantation potential. This study presents the first sequencing of cf-mtDNA in follicular fluid, aiming to explore the impact of ovarian stimulation treatment on the accumulation of mtDNA mutations and assess their implications for IVF outcomes.

Methods

We selected 24 women aged 27 to 35 from the Assisted Reproduction Unit based on inclusion criteria requiring them to be undergoing their first IVF cycle without known infertility issues. For each participant, blood samples were collected prior to ovarian stimulation, and follicular fluid samples were obtained at the time of oocyte retrieval. Long-PCR amplification and next-generation sequencing were used to identify mtDNA mutations.

Results

Ovarian stimulation-induced mtDNA mutations in follicular fluid were identified primarily in the D-loop region of the mitochondrial genome. These mutations were correlated with low oocyte counts and poor embryo quality, ultimately reducing IVF success rates. The accumulation of these mutations followed a time-dependent pattern, with longer stimulation durations resulting in higher mutation rates and lower pregnancy outcomes. Prolonged stimulations may exacerbate oxidative stress, negatively impacting oocyte quality.

Conclusions

By introducing the first sequencing of follicular fluid cf-mtDNA, this study establishes a direct link between ovarian stimulation-induced mtDNA mutations and reduced IVF efficacy, highlighting them as non-invasive biomarkers for predicting IVF outcomes. Future research may focus on optimizing stimulation protocols or incorporating antioxidants to minimize mtDNA mutations, improving oocyte quality and overall IVF success rates for patients.

Introduction

Mitochondrial dysfunction and mitochondrial DNA in human reproductive disorders

Reproductive disorders such as polycystic ovary syndrome (PCOS), endometriosis, and aging-associated factors are known to reduce in vitro fertilization (IVF) success rates and are linked to mitochondrial dysfunction. Mitochondrial dysfunction can lead to structural anomalies and mitochondrial DNA (mtDNA) damage in granulosa cells and oocytes, impairing their functionality at molecular, cellular, and systemic levels [1–4].

Found within the mitochondria's matrix, mtDNA consists of multiple copies of a 16.6 kb closed circular double-stranded chain, replicates independently of nuclear DNA, and has its own repair mechanisms. It encodes for 37 genes and contains a ~ 1200 bp non-coding region, known as displacement loop (D-loop), which contains promoters for mtDNA replication and transcription. Specifically, mtDNA is essential for mitochondrial growth and duplication, energy production, and calcium regulation and signaling, all of which impact oocyte maturation and future embryo development [5–7].

Several studies indicate that abnormal mtDNA copy numbers in oocytes, culture media, and follicular fluid are associated with inefficient maturation and poor implantation potential [8–11]. In PCOS, lower mtDNA copy numbers in oocytes correlate with compromised quality. Similar findings in endometriosis patients suggest mitochondrial impairment's role in infertility. Aging also shows a negative correlation between maternal age and mitochondrial activity, leading to poor ATP production, spindle abnormalities, and increased aneuploidies, reducing fertility [12–14]. These studies collectively emphasize the critical role of mtDNA quantity and integrity in proper mitochondrial function and determining oocyte quality and reproductive outcomes across various conditions.

New biomarkers for oocyte quality in IVF: follicular fluid analysis

In IVF, traditional biomarkers are based on morphological parameters like size, shape, and fragmentation, but these assessments may introduce a subjective bias. Consequently, there is a need for new approaches by identifying biomarkers to enhance oocyte quality and embryo selection, which is crucial to improve success rates [10, 15, 16].

The follicular fluid microenvironment plays a critical role in oocyte meiosis resumption and protection from oxidative stress. Changes in this environment can impact oocyte quality and embryo potential [17]. Typically discarded during IVF, follicular fluid is now seen as a valuable source for non-invasive biomarkers that assess oocyte quality and predict embryo implantation potential. Follicular fluid provides essential nutrients and biomolecules necessary for folliculogenesis and meiosis. It regulates signaling mediated by pituitary hormones, facilitating the transfer of gonadotropins like follicle-stimulating hormone (FSH) and luteinizing hormone (LH), which stimulate granulosa cells to produce estradiol, crucial for ovarian follicle growth and maturation [15, 18–20].

Research has focused on identifying biomarkers in follicular fluid, including hormones, lipids, metabolites, and cfDNA. Cell-free mitochondrial DNA (cf-mtDNA) from granulosa cells and oocytes is released into the follicular fluid, potentially through apoptosis. While the role of cf-mtDNA in oocyte maturation is unclear, its quantification is a promising biomarker in IVF. However, studies characterizing the cf-mtDNA genome sequence as a measure of mitochondrial health are still lacking [15, 16, 21–23].

Building on previous research showing that ovarian stimulation in IVF promotes follicle development and oocyte maturation while also inducing oxidative stress in the follicular fluid [17], we recognize that this stress can affect granulosa cells and oocytes due to an imbalance between reactive oxygen species (ROS) and mitochondrial antioxidants. The follicular microenvironment plays a crucial role in oocyte potential, and disruptions in this environment can compromise oocyte quality. In this study, we hypothesize that mtDNA, which is prone to mutations caused by ROS, could potentially compromise granulosa cells and oocyte development, negatively impacting female fertility [1, 2, 17, 24].

Our objective is to explore the mitochondrial genome by directly sequencing cf-mtDNA in follicular fluid, which reflects physiological and developmental processes in the follicle [16, 22]; and to determine if ovarian stimulation induces mtDNA mutations in follicular fluid and their implications on oocyte status and embryo developmental potential. We also aim to assess cf-mtDNA mutations as potential biomarkers to enhance oocyte quality assessment and ultimately improve IVF success rates.

Materials and methods

Patient recruitment, in vitro fertilization, and sample gathering

The study was approved by the Ethics Committee of the Balearic Islands (CEI-IB) with project number IB 4125/20, receiving a favorable report on May 20, 2020. A subset of patients treated at the Assisted Reproduction Unit of the Son Espases University Hospital was selected. All enrolled patients were informed of the purpose of the research and provided signed informed consent before participation. We recruited 24 women aged between 27 to 35 years who met the inclusion criteria, which specified an age range of 18 to 35 years, no prior infertility diagnoses (including PCOS, endometriosis, cancer, or other known diseases), no previous IVF cycles, standard hormonal parameters, and a normal ovarian reserve (assessed by antral follicle count via ultrasonography and/or anti-Müllerian hormone (AMH) levels). IVF cycles were initiated at the time of enrollment, with no delays exceeding one year. Blood samples, collected prior to ovarian stimulation, served as control specimens.

Treatments were personalized for each woman based on initial parameters and evolution. The ovarian response to stimulation was monitored by quantifying serum E2 levels and by ultrasound assessment of follicular and endometrial growth. All patients had received antagonist treatment and ovarian stimulation with recombinant human FSH (Gonal-F; Merck Serono) or human menopausal gonadotropin (hMG-HP; Menopur; Ferring), both of which are known to be equally effective [25]. The average dosage was 2149 ± 143 IU for FSH and 1863 ± 402 IU for hMG. Some FSH treatments were also combined with LH. Ovulation was triggered by a single 250 μg injection of human chorionic gonadotropin (hCG) (Ovitrelle, Merck Serono, Lyon, France), when at least three follicles reached the diameter of 17 mm or more on ultrasound examination. After 9 to 14 days of antagonist protocol, oocyte retrieval was performed by transvaginal ultrasound-guided aspiration 36 h after hCG administration, and all follicles were aspirated. A minimum of 4 mL of follicular fluid sample per woman was collected during the follicular puncture. Hemolyzed follicular fluid samples were discarded. Non-hemolyzed samples were centrifuged at 3000 × g, 4 °C for 15 min to precipitate possible cells in the medium. The supernatant was collected and centrifuged at 16,000 × g for 10 min and stored at − 80 °C.

In parallel, a routine sperm analysis was performed on either partners or sperm donors following the guidelines and criteria of the World Health Organization (WHO) laboratory manual for the examination and processing of human semen [26]. After recovery, oocytes went through a standard IVF/ICSI procedure and were only fertilized with good-quality sperm from healthy males, to minimize the impact of the male factor. Embryo quality was assessed according to the guidelines established by the Association for the Study of Biology of Reproduction (ASEBIR) [27]. After transfer, the remaining good-quality embryos (grades A and B) were cryopreserved.

mtDNA extraction, amplification, and NGS sequencing

Blood DNA was extracted using the Wizard® Genomic DNA Purification Kit out of the buffy coat fraction that was separated from the whole blood [28, 29]. Follicular fluid cfDNA was extracted using QIAamp® Circulating Nucleic Acid Kit with a starting volume of 4 mL of follicular fluid [30, 31]. DNA quantification was performed using the QuantiFluor® dsDNA system for both blood and follicular fluid.

The mitochondrial genome was amplified following the long-range PCR protocol established by Guo et al. in 2013 using the TaKaRa LA Taq™ PCR kit for blood and follicular fluid samples [32].

The sequencing library was generated from the amplified mtDNA using the Nextera XT DNA Library kit and Nextera XT DNA Library Prep Kit v2 Set A from Illumina. Libraries were quantified with QuantiFluor®, normalized to 4 nM and pooled equally. A high-depth sequencing was performed using the MiSeq instrument with a MiSeq v2 Reagent Kit, using a 2 × 251 cycle run and 11% of PhiX control.

Variant interpretation and analysis

Sequencing analysis and variant identification were performed using the Geneyx data analysis platform, with reference to the revised Cambridge Reference Sequence (rCRS) mitochondrial genome [33, 34]. Variants in blood and follicular fluid were identified and annotated. Low-quality variants and those with insufficient coverage were excluded, as were variants in the D310 region and positions 16,183–16,189, due to known intra-individual variation at these sites [35, 36]. A 1% heteroplasmy threshold was applied to exclude sequencing artifacts and minimize potential biases.

An mtDNA mutation was defined as a variant present in the follicular fluid but absent in the corresponding blood sample from the same woman. Variants with differing heteroplasmy levels between tissues were excluded from the analysis due to mitochondrial segregation events [37–39]. All identified mutations were visually validated using the Integrative Genomics Viewer tool [40]. Out of 24 participants, two had incomplete sample collections, and one follicular fluid sample was excluded due to sequencing failure, resulting in a final analysis cohort of 21 women.

Statistical analysis

All statistical analyses were performed using RStudio, utilizing the STATS v4.3.2 and MASS 7.3.60.0 packages. The statistical analysis used to determine the effectiveness and influence of the different treatments was Fisher’s exact test for the pregnancy rates, and the Kruskal–Wallis test for the number of mutations. Statistical analysis of the IVF process and outcomes used the Kruskal–Wallis test followed by a Dunn’s post hoc test. The same tests evaluated the impact of treatment duration on mtDNA mutations and oocyte retrieval. A Poisson regression analyzed the higher incidence of variants exclusively found in follicular fluid compared to blood, with an overdispersion test confirming the model's adequacy. An independent two-sample t-test compared heteroplasmy levels between follicular fluid and blood variants. An exact binomial test assessed mitochondrial mutation distribution. Pearson’s correlation coefficient evaluated the correlation between mtDNA mutations in follicular fluid and initial oocyte count. A Wilcoxon test, with continuity correction adjustment, assessed the significance of embryo quality based on the number of mutations per sample.

Results

In vitro fertilization treatments and outcome

IVF treatments and outcomes were evaluated in a cohort of women with an average age of 32.7 ± 0.5 years, ranging from 27 to 35 years old. Data on ovarian stimulation protocols, IVF evolution, embryo quality, and outcomes were systematically collected (Online Resource 1). Ovarian reserve was evaluated using ultrasonography and/or serum AMH measurements. Among those assessed by ultrasonography, the antral follicle count per ovary ranged from 4 to 11, with a mean total count of 14.6 ± 1.2. For patients with available AMH data, values ranged from 0.76 to 1.85 ng/mL, thereby providing a comprehensive overview of ovarian reserve across the cohort, which was accepted as normal in all cases. A correlation analysis further indicated that age was independent of the number of cf-mtDNA mutations identified (Online Resource 2). Additionally, all participants had no known infertility issues or other clinical conditions that might influence IVF success.

Although the different stimulation protocols—FSH, HMG, FSH and HMG, and FSH and LH—are recognized as equally effective [25], first, we conducted an analysis to verify that there were no significant differences in pregnancy rates or the number of mutations identified. The results confirmed no significant differences among the treatment groups (p = 0.197 and p = 0.956, respectively). It is important to note that individual variations in dosages, driven by the evolution of each patient’s stimulation protocol and differences in drug metabolism, limited our ability to perform a more precise correlation analysis between dosage and outcomes.

The oocyte maturation rate was 79.1%, slightly higher than typical rates of 55–75% [41, 42], reflecting the efficiency of the stimulation protocol. The first cycle achieved a pregnancy rate of 41.7% (10 out of 24 women), while two additional successful pregnancies were obtained in subsequent cycles, resulting in a cumulative clinical pregnancy rate of 50% (12 out of 24 women).

Influence of the initial oocyte count and evolution on IVF outcome

The number of retrieved oocytes significantly impacts IVF success [43, 44]. Figure 1 shows the initial oocyte count, mature oocyte (MII) proportion, and fertilization rates. Women were grouped by IVF outcome: pregnant, non-pregnant, and without viable embryos. Pregnant women had a higher average initial oocyte count (12.4) compared to non-pregnant (7.6) and women without viable embryos (6.3) (p-value = 0.038 and 0.027, respectively).

Fig. 1.

Oocyte and embryo counts association with IVF pregnancy outcomes. Number of oocytes/embryos per woman in different stages (initial oocytes, mature oocytes, and fertilized embryos) of the IVF process depending on the outcome: pregnant, non-pregnant, and without viable embryos. *p-value < 0.05

The higher number of oocytes associated with a successful pregnancy was consistent through all IVF stages compared to unsuccessful IVF cycles, despite being less significant (p-value > 0.05) in the MII and fertilized embryos stage (Fig. 1).

Incidence of mtDNA variants follicular fluid and blood

To assess ovarian stimulation's effect on mtDNA mutations, we compared variants in blood and follicular fluid. Blood samples served as a reference due to the infeasibility of pre-treatment follicular fluid collection. Homoplasmic pathogenic variants were not detected. For each variant, we recorded relevant details such as location and locus, variant nomenclature (HGVS), percentage of heteroplasmy, and population frequency (Online Resource 3).

A total of 66 variants were identified exclusively in follicular fluid, with an average of 3.24 variants per woman. Among the 21 women sequenced, 20 (95.2%) exhibited at least one variant. Additionally, the heteroplasmy levels of these variants ranged from 1.03 to 12.03%, with a mean level of 3.10%. These quantitative results provide a detailed overview of the mitochondrial mutation profile present in the follicular fluid samples.

Without access to follicular fluid prior to ovarian stimulation as a reference point, we cannot assure that the totality of those variants found exclusively in follicular fluid reflects novel mutations in oocytes and granulosa cells as a consequence of the treatment. Given that ovarian tissue is genetically homogeneous under normal conditions with minimal heteroplasmy [45], the presence of multiple variants in follicular fluid suggests the accumulation of mtDNA mutations. From this point forward, these variants will be considered as mtDNA mutations in follicular fluid.

Location and distribution of mutations within the mitochondrial genome

The location of mtDNA mutations is crucial for understanding susceptibility to redox processes. The distribution of mutations can indicate which genes or regions are more prone to be mutated and its implications for the correct mitochondrial function and segregation.

We observed a 3.14-fold increase in mutations in the non-coding region, or D-loop, compared to the coding ones (p-value = 8.04·10⁻⁴). If we focus on the hypervariable regions (HVR) within the D-loop, we find a significant 4.65-fold increase in the number of mutations (p-value = 1.62·10⁻⁵). From the three different HVRs (HVR I, HVR II, and HVR III), the HVR II had the most mutations, with 7, followed by HVR I and HVR III with 4 and 3, respectively (Table 1; Fig. 2). No other regions had significant differences in the mutation rates.

Table 1.

Number of mutations in different mitochondrial regions and relative fold increase. The regions displayed are: coding, non-coding, non-coding hypervariable region (HVR), and whole mitochondrial genome. The region length, the number of mutations found per base-pair and the relative fold increase (coding region as reference) are also specified

Region	Number of mutations	Region length (bp)	Mutation per kbp	Fold increase
Coding	53	~ 15,369	3.45	1
Non-coding	13	~ 1200	10.83	3.14 **
Non-coding (HVR)	13	810	16.05	4.65 ***
Whole mtDNA	66	16,569	3.98	-

** p-value < 0.01; *** p-value < 0.001

Fig. 2.

Circular representation of the human mtDNA genome. Each yellow star represents a mutation identified in our study and its specific location. Note that the HVR region is zoomed in

mtDNA alterations on initial oocyte count and embryo quality

Out of 66 mtDNA mutations identified, 51 had not been previously described, 8 were benign, 6 as of uncertain significance (VUS), and one pathogenic (3243 A > G, rs199474657). The pathogenic variant was found in a patient with 1.61% heteroplasmy. Noteworthy, this patient had a reduced number of oocytes and an unsuccessful pregnancy.

A higher number of mutations per mitochondrion, a phenomenon which is related and well-documented in aging, can induce a lower number of oocytes [46]. Our results suggest a trend (p-value = 0.081) towards an association between a higher number of mtDNA mutations in follicular fluid and a lower number of oocytes retrieved (Fig. 3).

Fig. 3.

Correlation between mtDNA mutations in follicular fluid and initial oocyte count. Each dot represents a single woman, and the solid line displays the trendline of the distribution. ^a Outlier sample removed from the statistical analysis

To evaluate the impact of mtDNA mutations found in follicular fluid on embryo quality, we categorized participants by the number of good-quality embryos (grades A and B). We found a trend of fewer mtDNA mutations in those with multiple good-quality embryos (2.55 vs. 3.80, p-value = 0.357) (Fig. 4A). When applying a more refined quality distinction (women with grade A embryos versus non-grade A embryos), we observed a more pronounced difference. Women with grade A embryos had significantly fewer mutations than the group with non-grade A embryos (2.5 vs. 6.33, p-value = 0.026) (Fig. 4B).

Fig. 4.

Impact of mtDNA mutations in follicular fluid on embryo quality. A Participants categorized by the production of good quality embryos: more than one good quality embryo (n = 11) vs. none or single good quality embryo (n = 10). Grade A and grade B are considered good quality embryos. B. Categorized by participants with at least one grade A embryo (n = 14) versus those without any grade A embryos (grades B, C, and D; n = 3). *p-value < 0.05

mtDNA mutations and oocyte count in varying treatment duration

The ovarian stimulation treatment during IVF can induce mutations in the mtDNA, with implications for the quantity and quality of the oocytes [47]. We then classified treatments into short (9–10 days), medium (11–12 days), and long (13–14 days) durations. Although not statistically significant, we observed a linear increasing tendency in the number of mtDNA mutations, where short treatments had fewer mutations (2.58) than medium (3.60) and long (4.25) treatments (p-value = 0.448) (Fig. 5A). Regarding the number of oocytes retrieved, we observed a linear decrease where longer treatments showed a significant reduction in the number of oocytes compared to short treatments (11.5 in short, 9.0 in medium, 5.8 in long, p-value = 0.15) (Fig. 5B). Noteworthy, no successful pregnancies were achieved in the long treatment group.

Fig. 5.

mtDNA mutations and initial oocyte count association with ovarian stimulation treatment duration. Ovarian stimulation treatment periods were categorized into short (9–10 days; n = 13), medium (11–12 days; n = 7), and long (13–14 days; n = 4). A Number of mtDNA mutations in follicular fluid per woman after different treatments periods. B Number of initial oocytes gathered per woman after different treatments periods. *p-value < 0.05.

Discussion

Initial oocyte count is a significant factor for the IVF success rate

A higher number of oocytes retrieved generally increases the likelihood of success in IVF treatment, providing more chances for fertilization and viable embryo development [43, 44]. However, conditions like PCOS can result in lower fertilization rates and poor IVF outcomes despite higher oocyte counts. Factors such as age, ovarian reserve, and response to stimulation medications also affect the number and quality of oocytes retrieved. Women with diminished ovarian reserve or atypical responses (hypo-responders with < 4 oocytes or hyper-responders with ≥ 15 oocytes) may yield fewer good-quality oocytes, adversely affecting fertilization and IVF success rates [48, 49]. Regarding women’s age, it is widely studied that chromosomal aberrations and mutations increase exponentially in women aged above 35, which implicates less responsiveness to the treatment and an overall lower number of fertilized embryos. However, this issue is known not to be a problem or a factor in women below 35, as we have verified, since their status remains similar across the age range of our patients (27 to 35).

One way to measure the success of IVF is by looking at the clinical pregnancy rate (CPR). Although IVF has improved in the last decade, mostly thanks to advances in the cryopreservation process and preimplantation genetic testing, the success rates are still modest. Generally, CPRs remain below 45% for women under 35 and drop below 20% for those over 40 [44, 50]. In this study of 24 women under 35, the average cumulative CPR was 50%, slightly higher than the average for standard IVF techniques. Interestingly, if we focus on women with more than 10 oocytes retrieved, the CPR increased to 80%. These findings align with previous studies, highlighting the importance of oocyte count in achieving good-quality embryos and successful pregnancies.

Follicular fluid mtDNA damage after the IVF treatment

Recent studies propose non-invasive strategies and biomarkers to predict oocyte and embryo quality in addition to the morphological assessment. For mitochondria, cumulus cell mtDNA copy number and mtDNA count in blastocyst-stage culture media have been suggested as biomarkers, though the latter carries a high contamination risk with foreign DNA [10, 11, 51, 52].

During IVF, hormonal stimulation increases oxidative stress, which can impact oocytes, granulosa cells, and other supporting cells. This stress may contribute to mtDNA mutations, detectable in follicular fluid cf-mtDNA, offering insights into mitochondrial function and metabolic activity [11, 22, 51]. Due to ethical and methodological constraints, we used blood as a reference mitochondrial genome, assuming minimal baseline differences between these tissues.

Unlike blood, which undergoes constant cellular turnover and is prone to low-level heteroplasmic variants, ovarian tissue, that is where follicles develop, remains genetically stable with minimal heteroplasmy under normal conditions [39, 45]. As the first study to sequence cf-mtDNA from follicular fluid, the absence of prior sequencing data makes it challenging to establish baseline mutation rates or directly compare our findings. However, the significantly high number of variants observed in follicular fluid is unlikely to reflect baseline variability and instead suggests mutations acquired in response to ovarian stimulation. The accumulation of mutations in stimulated cycles, in contrast to the typically low heteroplasmy observed in ovarian tissue, suggests that oxidative stress during ovarian stimulation induces mutational events in the follicular microenvironment containing the granulosa and theca cells that may impact oocyte and embryo quality. These mutations could serve as potential biomarkers for IVF outcomes, offering a new perspective on the effects of ovarian stimulation at the mitochondrial level.

The mitochondrial genome displays a non-random distribution of mtDNA variants

The mitochondrial genome is highly variable in sequence and organization. The mitochondrial genome’s regions have uneven susceptibility to mutations, with the HVR being more prone to mutations. This region, which includes the replication origin and many gene promoters, likely has higher mutation rates due to lower selection pressures and accessibility during replication and transcription. Our findings align with previous research, showing higher mutation rates in the HVR and its neighboring regions compared to the rest of the mitochondrial genome [45, 53]. mtDNA mutation rates are 10 to 20 times higher than nuclear DNA, and within the mitochondrial genome, the HVR mutation rate is four to five times higher. Although the HVR is not a coding region, mutations here can affect mtDNA copy number, impacting transcription initiation and regulation. The follicular microenvironment is very susceptible to changes, and cells within the follicle with altered mtDNA copy number could contribute to poorer oocyte maturation. Although we do not know if oocytes are directly affected by these mutations, it is also known that oocytes with fewer mtDNA copies tend to have reduced maturation and poorer IVF outcomes [53, 54].

Acquired mtDNA mutations can impair oocyte maturation, even at low heteroplasmy levels

Mitochondrial pathologies linked to heteroplasmy often affect enzymes or proteins needing precise regulation, with a 20% heteroplasmy threshold typically used to predict functional consequences. In our study, all identified mutations were below this threshold, yet a high number of low-level heteroplasmy mutations were observed in the follicular fluid [55]. Our results suggest an impact of these acquired mtDNA mutations, even at low heteroplasmy levels, which can impair oocyte maturation.

The pathogenic variant 3243 A > G, known to alter RNA^Leu (UUR) transcription, was found in one of our samples. The woman carrying this variant was healthy and had no remarkable family or medical history. This variant, common in Caucasians, is linked to conditions like MELAS syndrome at high heteroplasmy levels and milder issues like diabetes and hearing loss at lower levels [56]. Despite the low heteroplasmy, 3243 A > G could negatively impact mitochondrial function of the affected cells, contributing to low oocyte number and unsuccessful pregnancy seen in this patient [56–58].

Inefficient oxidative phosphorylation and poor ATP production can lead to abnormal embryo development. Thus, minor variations in oocytes and granulosa cells may affect these pathways [51]. Our results suggest a trend where more mtDNA mutations, even at low heteroplasmy levels, correlate with fewer oocytes retrieved after IVF treatment. Although not statistically significant, this trend supports the hypothesis that mtDNA mutations negatively impact oocyte count and quality.

Besides the number of oocytes, women with good-quality embryos (grades A and B) tend to have fewer mtDNA mutations in follicular fluid. This correlation is statistically significant for women with grade A embryos, indicating a negative impact of mtDNA mutations on oocyte quality and IVF outcomes. Detecting mtDNA mutations in follicular fluid could be crucial for predicting IVF outcomes.

Prolonged IVF treatments may increase mtDNA damage without enhanced oocyte count and quality

IVF has seen significant improvements in embryo vitrification and selection recently. However, ovarian stimulation treatments have remained largely unchanged for the past 20 years [59, 60]. With the growing demand for IVF and a lack of new, safer treatments, concerns have arisen about the adverse effects of prolonged hormone treatments on both women and oocyte/embryo quality. These treatments can lead to insulin resistance, inflammatory responses, and ovarian hyperstimulation syndrome, which can predispose women to adverse outcomes compared to typical pregnancies. Specifically, these treatments can disrupt cell homeostasis by increasing ROS accumulation, thereby affecting mitochondrial integrity and function [1, 4, 61].

Our study found that longer stimulation protocols lead to the accumulation of mtDNA mutations in follicular fluid, likely due to mtDNA damage in cells within the follicle. Interestingly, no differences in heteroplasmy levels were found between short, medium, and long treatments. According to the bottleneck theory, longer IVF treatments promote more mutations, but the heteroplasmy percentage depends on the mtDNA expansion process [38, 51].

Longer treatments increased mtDNA mutations by 65% compared to shorter treatments, combined with a decrease in the number of oocytes and poorer pregnancy rates. We speculate that hormone treatments over 12 days could harm reproductive outcomes and not improve conception chances due to increased ROS and mtDNA mutations.

To mitigate mtDNA mutations during extended IVF protocols, a combined approach involving hormonal treatments and antioxidant supplementation has been suggested to enhance embryo quality and implantation rates, potentially improving overall IVF success [62, 63]. Further research involving larger cohorts and incorporating functional analyses is warranted. In particular, comparing mitochondrial mutation rates between IVF cycles with and without antioxidant supplementation could provide critical insights into the efficacy of this approach.

Conclusions

This pilot study marks a significant advancement in reproductive medicine by being the first to successfully sequence cf-mtDNA from follicular fluid. Our findings suggest that the follicular microenvironment of women undergoing ovarian stimulation, which is subject to increasing oxidative stress, promotes the emergence of mtDNA damage. We demonstrate a direct correlation between the accumulation of mtDNA mutations and reduced oocyte quality and embryo viability. Importantly, this is the first evidence linking mtDNA integrity in follicular fluid directly with oocyte quality, highlighting cf-mtDNA as a potential biomarker for assessing the viability of oocytes in IVF treatments.

The observed association between the duration of ovarian stimulation and an increase in mtDNA mutations suggests that shorter treatment protocols and the integration of antioxidant therapies may enhance oocyte integrity by mitigating oxidative damage to mtDNA. These strategies could significantly improve the success rates of IVF by ensuring the obtention and selection of high-quality oocytes.

Future research should focus on validating these findings in a broader cohort to verify the reliability of cf-mtDNA as a biomarker. Additionally, understanding the underlying mechanisms contributing to mtDNA damage caused by ovarian stimulation treatments can provide crucial insights for optimizing reproductive strategies and ensuring the overall success of assisted reproductive techniques.

Abbreviations

ASEBIR

Association for the Study of Biology of Reproduction

cf-mtDNA

Cell-free mitochondrial DNA

CPR

Clinical pregnancy rate

FSH

Follicle-stimulating hormone

hCG

Human chorionic gonadotropin

HGVS

Human Genome Variation Society

HVR

Hypervariable region

IGV

Integrative genomics viewer

LH

Luteinizing hormone

MII oocytes

Meiosis II oocytes

mtDNA

Mitochondrial DNA

OXPHOS

Oxidative phosphorylation

PCOS

Polycystic ovary syndrome

rCRS

Revised Cambridge Reference Sequence

ROS

Reactive oxygen species

VUS

Variant of unknown significance

WHO

World Health Organization

Author contribution

Laura Peralta, Laura Torres-Juan, and Antonia Picornell were involved in the study design, concept, and funding. Laura Peralta and R.B. participated in patient recruitment and IVF procedures. Jessica Hernandez-Rodriguez, Marc Ventayol-Guirado, Maria Victoria Llull-Alberti, Joana Florit, and Joana Francesca Ferragut took part in the experimental execution, data analysis, and interpretation. Marc Ventayol-Guirado wrote the manuscript. Laura Peralta, Laura Torres-Juan, Damian Heine-Suñer, Jessica Hernandez-Rodriguez, Iciar Martinez, Joana Francesca Ferragut, Antonia Picornell, and Marc Ventayol-Guirado commented on and approved the final version of the article.

Funding

This work was supported by the Hospital Universitari Son Espases (Palma, Balearic Islands, Spain) grant “Proyecto PILOTO2019/PERALTA.”

Data availability

The sequencing data (raw.fastq files) has been deposited in the Sequence Read Archive (SRA) from NCBI and will be available upon publication under the submission number SUB14879848; the provisional link to access is the following: https://dataview.ncbi.nlm.nih.gov/object/PRJNA1193537?reviewer=3d5c3cjq11u87ocete93muc2s3. The patient’s data will be shared on reasonable request to the corresponding author.

Declarations

Ethics approval and consent to participate

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of the Balearic Islands (CEI-IB) with project number IB 4125/20, with a favorable report on the 20 th of May 2020.

Consent to participate

Informed consent was obtained from all individual participants prior to the inclusion in the study.