{"paper_id":"b44a2301-79ea-47ab-bed0-23b1d5617b87","body_text":"Comparison of Pregnancy Outcomes and Vaginal Microbiota in Endometriosis Patients Undergoing Frozen Embryo Transfer\nUsing Letrozole Combined HMG Versus Hormone Replacement Therapy with GnRH-a Pretreatment\nJie Zhang, Lei Dai, Chunyan Jiang, Yuxin Zhao, Xiang Ma, Yugui Cui, Jiayin Liu\nCite this article as:\nJie Zhang, Lei Dai, Chunyan Jiang, Yuxin Zhao, Xiang Ma, Yugui Cui, Jiayin Liu. Comparison of Pregnancy Outcomes and Vaginal\nMicrobiota in Endometriosis Patients Undergoing Frozen Embryo Transfer Using Letrozole Combined HMG Versus Hormone Replacement\nTherapy with GnRH-a Pretreatment[J]. Journal of Biomedical Research, In press. doi: 10.7555/JBR.39.20250205\nView online: https://doi.org/10.7555/JBR.39.20250205\nArticles you may be interested in\nHistone lactylation promotes cell proliferation, migration and invasion through targeting HMGB1 in endometriosis\nThe Journal of Biomedical Research. 2023, 37(6): 470   https://doi.org/10.7555/JBR.37.20230095\nExtradural contralateral S1 nerve root transfer for spastic lower limb paralysis\nThe Journal of Biomedical Research. 2023, 37(5): 394   https://doi.org/10.7555/JBR.37.20230068\nWashed microbiota transplantation stopped the deterioration of amyotrophic lateral sclerosis: The first case report and narrative review\nThe Journal of Biomedical Research. 2023, 37(1): 69   https://doi.org/10.7555/JBR.36.20220088\nGut microbiota links with cognitive impairment in amyotrophic lateral sclerosis: A multi-omics study\nThe Journal of Biomedical Research. 2023, 37(2): 125   https://doi.org/10.7555/JBR.36.20220198\nhUCMSC-derived extracellular vesicles relieve cisplatin-induced granulosa cell apoptosis in mice by transferring anti-apoptotic miRNAs\nThe Journal of Biomedical Research. 2025, 39(1): 36   https://doi.org/10.7555/JBR.37.20230310\nSystemic lupus erythematosus therapeutic strategy: From immunotherapy to gut microbiota modulation\nThe Journal of Biomedical Research. 2024, 38(6): 531   https://doi.org/10.7555/JBR.38.20240009\nUnproofed\n\nComparison  of  Pregnancy  Outcomes  and  Vaginal  Microbiota\nin Endometriosis Patients Undergoing Frozen Embryo Transfer\nUsing  Letrozole  Combined  HMG  Versus  Hormone\nReplacement Therapy with GnRH-a Pretreatment\nJie Zhang, Lei Dai, Chunyan Jiang, Yuxin Zhao, Xiang Ma, Yugui Cui✉, Jiayin Liu✉\nState Key Laboratory of Reproductive Medicine and Offspring Health, Clinical Center of Reproductive Medicine, The\nFirst Affiliated Hospital with Nanjing Medical University, Nanjing, Jiangsu 210029, China.\nAbstract\nThis  study  investigated  differences  in  reproductive  outcomes  and  vaginal  microbiota  profiles  between  two\nendometrial preparation protocols—letrozole (LE) combined with human menopausal gonadotropin (HMG) and\nhormone  replacement  therapy  (HRT)  with  GnRH-a  pretreatment —in  women  with  endometriosis  (EMs)\nundergoing frozen embryo transfer (FET). Following 1∶1 propensity score matching, a total of 770 FET cycles\nwere analyzed. No statistically significant differences were observed in live birth rates or clinical pregnancy rates\nbetween the two groups. However, the LE + HMG group showed a lower miscarriage trend (13.7% vs. 19.8%, P =\n0.070)  and  significantly  fewer  cesarean  deliveries  (64.9% vs.  75.4%,  P  =  0.020)  and  hypertensive  disorders  of\npregnancy  (4.8% vs.  10.1%,  P  =  0.039).  Recent  evidence  suggests  that  GnRH-a  treatment  may  disrupt\nreproductive tract microbiota. Given ethical constraints on endometrial sampling during FET, vaginal microbiota\nwas  used  as  a  surrogate  to  explore  microbial  differences  between  protocols.  In  the  prospective  arm,  vaginal\nsamples from 55 women in the LE + HMG group and 50 in the GnRH-a HRT group were analyzed using 16S\nrRNA  sequencing  and  droplet  digital  PCR.  While  no  significant  differences  were  observed  in Lactobacillus or\nGardnerella abundance,  the  GnRH-a  HRT  group  exhibited  enrichment  of  potential  pathogens,  such  as\nEscherichia-Shigella and Staphylococcus. In conclusion, although both protocols achieved comparable live birth\noutcomes,  the  LE  +  HMG  regimen  was  associated  with  fewer  obstetric  complications  and  a  more  favorable\nvaginal microbiota profile compared to GnRH-a HRT.\nKeywords: Endometriosis, Letrozole, Gonadotropin-releasing hormone agonist, Frozen embryo transfer, Vaginal\nMicrobiota\n \nIntroduction Endometriosis  (EMs)  is  a  common,  chronic,\nestrogen-dependent  and  inflammatory  disease\n✉Corresponding authors: Jiayin Liu and Yugui Cui, Clinical Cen-\nter  of  Reproductive  Medicine,  the  First  Affiliated  Hospital  with\nNanjing Medical  University,  300  Guangzhou  Road,  Nanjing,  Ji-\nangsu  210029,  China.  E-mails: jyliu_nj@126.com (Liu)  and\ncuiygnj@njmu.edu.cn (Cui).\nReceived:  10  May  2025; Revised:  28  May  2025; Accepted:  03\nJune 2025; Published online: 04 June 2025\nCLC number: R71, Document code: A\nThe authors reported no conflict of interests.\nThis is an open access article under the Creative Commons Attribu-\ntion (CC BY 4.0) license, which permits others to distribute, remix,\nadapt and build upon this work, for commercial use, provided the\noriginal work is properly cited.\n \nAvailable online at www.jbr-pub.org.cn\nOpen Access at PubMed Central\nJournal of Biomedical Research, 2025 39(0): 1–13  \n \nOriginal Article\n©  2025 by Journal of Biomedical Research. https://doi.org/10.7555/JBR.39.20250205\nUnproofed\n\ncharacterized  by  the  presence  of  endometrium-like\ntissue outside the uterus. It is estimated to impact 10-\n15% of  reproductive  women[1],  with  approximately\none-third  experiencing  infertility[2].  Many  of  these\nwomen  require  in  vitro  fertilization  (IVF)  to  achieve\npregnancy[3].  In  recent  years,  the  global  adoption  of\nfrozen-thawed  embryo  transfer  (FET)  has  expanded\nswiftly,  largely  due  to  advancements  in  vitrification\ntechniques and blastocyst culture.\nA key determinant of FET success is the adequacy\nof endometrial preparation. As an estrogen-dependent\ncondition,  EMs  is  characterized  by  abnormal\naromatase overexpression in both ectopic lesions and\nthe  eutopic  endometrium,  leading  to  a  disrupted\nestrogenic  microenvironment[4].  This  imbalance  may\ninterfere  with  embryo-endometrium  communication,\naffecting  pregnancy  outcomes[5].  This  unique\npathological  feature  requires  tailored  endometrial\npreparation strategies for EMs patients. However, the\nEuropean  Society  of  Human  Reproduction  and\nEmbryology  (ESHRE)  guidelines  offer  no  specific\nrecommendations for endometrial preparation in EMs\npatients[3],  making  this  an  important  area  for\ninvestigation.\nGnRH-a (gonadotropin-releasing hormone agonists)\nhas  been  shown  to  suppress  local  inflammation  and\nreduce  oxidative  stress,  thereby  improving\nendometrial  receptivity  in  EMs  patients[6].\nConsequently,  the  hormone  replacement  therapy\n(HRT)  with  GnRH-a  pretreatment  protocol  has\nbecome  a  common  choice  for  FET  in  EMs  patients.\nHowever,  the  latest  ESHRE  guidelines  no  longer\nrecommend  the  use  of  GnRH-a  prior  to  assisted\nreproductive  technologies  (ART)  in  EMs  patients[3].\nEmerging  evidence  suggests  that  GnRH-a  HRT  does\nnot offer significant advantages in improving fertility\noutcomes  when  compared  to  HRT  alone  or  natural\ncycle  (NC)  protocols[7-8].  It  is  well  known  that  both\nHRT  and  GnRH-a  HRT  protocols  involve  excessive\nsupplementation of estradiol and progesterone, which\nmight  raise  the  risk  of  thromboembolic  events[9].\nMoreover,  lack  of  corpus  luteum  (CL)  formation  in\nthese cycles has been associated with a higher risk of\nadverse maternal and perinatal outcomes [10]. Although\nNC  protocols  are  considered  safer  and  more\nphysiological,  they  lack  flexibility  in  scheduling,\nrequiring  frequent  ovulation  monitoring  and  facing\nhigher  cycle  cancellation  rates.  These  limitations\nunderscore the need to explore alternative endometrial\npreparation  strategies  that  are  both  effective  and\npatient-friendly  for  women  with  EMs  undergoing\nFET.\nLetrozole  (LE),  classified  as  a  third-generation\naromatase inhibitor, functions by suppressing estrogen\nsynthesis  and  facilitating  follicular  development\nthrough  negative  feedback  on  the  hypothalamic-\npituitary axis[11-12]. Importantly, LE-induced ovulation\nresults  in  the  formation  of  a  healthy  CL,  which\nreduces  the  risk  of  hypertensive  disorders  of\npregnancy (HDP)[13]. Up to now, LE ovarian induction\nhas been increasingly used for endometrial preparation\nin  FET,  especially  for  women  with  polycystic  ovary\nsyndrome[14] and  anovulation[15].  Beyond  its\nreproductive  applications,  LE  has  also  demonstrated\nefficacy in alleviating EMs-related pain and reducing\ndisease  recurrence  in  both  premenopausal  and\npostmenopausal  populations[16-17].  Emerging  evidence\nsuggests  that  LE  may  benefit  the  endometrium  of\nwomen  with  EMs  by  suppressing  the  estrogen-\ninflammatory  axis[18],  and  enhancing  integrin  ανβ3\nexpression,  which  could  improve  endometrial\nreceptivity  and  implantation  rates[19-20].  Despite  its\npotential  advantages,  few  studies  have  evaluated  the\nefficiency  of  LE-based  ovarian  induction  in  FET\ncycles for women with EMs.\nRecent  microbiota  research  has  revealed  that\ntreatment  with  GnRH-a  for  3  to  6  months  in  EMs\npatients for gynecological symptom management may\nled  to  a  marked  decline  in Lactobacillaceae and\nincreased  levels  of Streptococcaceae,\nStaphylococcaceae,  and Enterobacteriaceae in\nendometrial  samples,  suggesting  a  potential\nassociation  between  GnRH-a  use  and  subclinical\nintrauterine  infections[21].  Despite  these  findings,  the\nimpact  of  various  endometrial  preparation  regimens\non  the  reproductive  tract  microbiota  during  FET\ncycles remains poorly understood. Importantly, direct\nsampling of the endometrial microbiota on the day of\nembryo transfer is not feasible due to both ethical and\nclinical  constraints —performing  an  endometrial\nbiopsy  at  this  critical  time  may  damage  the\nendometrium,  reduce  the  likelihood  of  implantation,\nand  raise  ethical  concerns  regarding  unnecessary\nharm.  Interestingly,  previous  studies  have\ndemonstrated  a  continuous  gradient  in  microbial\ncomposition  extending  from  the  vagina  to  the  pelvic\ncavity[22].  Moreover,  animal  experiments  have  shown\nthat  transplantation  of  vaginal  microbiota  from\npatients  with  chronic  endometritis  into  rats  activates\nthe  endometrial  TLR4/NF-κB  pathway[23],  indicating\nthat disturbances in the vaginal microbiota may reflect\nor  induce  similar  changes  in  the  upper  reproductive\ntract.  Therefore,  in  the  absence  of  direct  endometrial\nsampling,  analyzing  vaginal  microbiota  provides  a\nrepresentative  and  ethically  acceptable  surrogate  for\nassessing microbial impacts during FET.\n2 Zhang J et al. J Biomed Res, 2025, 39(0)\nUnproofed\n\nAccordingly,  this  study  was  designed  to  evaluate\nand  compare  pregnancy  and  perinatal  outcomes,  as\nwell  as  vaginal  microbiota  characteristics,  in  women\nwith EMs undergoing FET using either LE + HMG or\nGnRH-a  HRT  protocols.  By  integrating  clinical\nefficacy  and  microbial  profiling,  our  findings  aim  to\nprovide  evidence-based  insights  for  optimizing\nendometrial  preparation  strategies  in  this  unique\npatient population. \nMaterials and methods\n \nStudy design and sample collection\nThis  retrospective  cohort  study  included  women\nwith  EMs  who  underwent  either  LE  +  HMG  or\nGnRH-a HRT FET cycles between January 2016 and\nDecember  2023.  Inclusion  criteria  were  as  follows:\n(i) The first three FET cycles per patient; (ii) Female\nage  under  43  years;  (iii)  A  single  blastocyst  transfer\nper  cycle.  Exclusion  criteria  included:  (i)  History  of\nrecurrent spontaneous abortion; (ii) Congenital uterine\nmalformations;  (iii)  Use  of  preimplantation  genetic\ntesting;  (iv)  Loss  to  follow-up  or  incomplete  data;\n(v) Multiple pregnancies.\nIn  addition,  a  prospective  study  component\ninvolved vaginal sample collection from EMs patients\nundergoing FET with either the LE + HMG (n = 55)\nor  GnRH-a  HRT  (n  =  50)  protocol  between  January\nand  June  2024.  The  inclusion  criteria  for  the\nmicrobiota  cohort  included:  (i)  diagnosis  of\nendometriosis;  (ii)  female  age  under  43  years.\nHowever,  embryo  quality  was  not  restricted,  as  the\nprimary aim of this component was to investigate how\ntwo  endometrial  preparation  protocols  affect  the\ncomposition  of  the  vaginal  microbiota,  without\naddressing  potential  associations  between  microbiota\nalterations and pregnancy outcomes at this stage. The\nexclusion  criteria  were  consistent  with  the\nretrospective  cohort  but  included  additional\nmicrobiota-specific  considerations:  patients  were\nexcluded  if  they  had  used  antibiotics  or  vaginal\nprobiotics within one month prior to sampling, or had\nacute  reproductive  tract  infections,  diabetes  mellitus,\nor autoimmune diseases.\nVaginal  secretions  were  collected  prior  to  embryo\ntransfer  on  the  day  of  the  procedure,  before  any\nsurgical manipulation. A sterile speculum was used to\nexpose the cervix, and two sterile cotton swabs were\nused  to  obtain  secretions  from  the  upper  third  of  the\nvaginal wall. One swab was used for 16S rRNA gene\nsequencing  and  the  other  for  droplet  digital  PCR\n(ddPCR) analysis. Importantly, this microbiota cohort\nwas  independent  of  the  retrospective  cohort.  In  total,\n210 vaginal samples (two per patient) were collected.\nThe  study  was  approved  by  the  ethics  committee  of\nthe  First  Affiliated  Hospital  of  Nanjing  Medical\nUniversity (No. 2023-SR-325). \nDiagnosis of EMs and adenomyosis\nEMs  was  diagnosed  either  through  surgical\nmethods  (laparoscopy  or  laparotomy)  or  based  on\ntransvaginal  ultrasound  findings  consistent  with\novarian  endometriotic  cysts.  The  diagnosis  of\nendometriotic  cysts  via  ultrasound  had  to  be\ndocumented in at least two separate menstrual cycles.\nAdenomyosis  was  diagnosed  based  on  imaging\ncriteria  using  transvaginal  ultrasound,  with\nassessments  conducted  by  at  least  two  highly  skilled\nradiologists.  The  diagnosis  was  established  when\npatients  presented  with  clinical  symptoms  such  as\nhypermenorrhea  or  dysmenorrhea  and  exhibited  at\nleast  two  of  the  following  ultrasound  features,  as\ndefined  by  the  Morphological  Uterus  Sonographic\nAssessment (MUSA) criteria [24]. \nEndometrial preparation\nIn  the  LE  +  HMG  group,  LE  (Hengrui,\nLianyungang, Jiangsu, China) was administered orally\nat a daily dose of 2.5 mg starting on the 4th day of the\nmenstrual cycle for 5 consecutive days. Additionally,\n75  IU  of  human  menopausal  Gonadotropin  (Lizhu,\nZhuhai,  Guangdong,  China)  were  given  every  other\nday. Follicle monitoring began on the 12th day of the\nmenstrual  cycle  and  continued  until  the  follicle\ndiameter  exceeded  18  mm.  Subsequently,  5,000-\n10,000  IU  of  urinary  human  chorionic  gonadotropin\n(Lizhu,  Zhuhai,  Guangdong,  China)  was  injected.\nFollowing  triggering,  oral  dydrogesterone\n(Duphaston; Abbott Laboratories, Chicago, IL, United\nStates) was prescribed at a dose of 10 mg twice daily\nfor luteal phase support. On the 6th day after trigger,\nblastocyst transfer was performed, and Duphaston was\ncontinued until the 10th week of pregnancy (Fig.1A).\nIn  the  GnRH-a  HRT  group,  long-acting  GnRH-a\n(Diphereline, 3.75 mg; Ipsen Pharma Biotech, Signes,\nFrance)  was  administered  by  intramuscular  injection\non menstrual cycle days 1-2. After 30 days, estradiol\nvalerate  (Progynova;  Bayer,  Leverkusen,  North\nRhine-Westphalia, Germany) was prescribed orally at\na  daily  dose  of  4-6  mg  to  stimulate  endometrial\nproliferation  until  the  endometrial  thickness  (EM)\nreached ≥8  mm.  Luteal  support  consisted  of  oral\ndydrogesterone  (10  mg  twice  daily)  combined  with\nvaginal progesterone gel (Crinone, 90 mg once daily;\nMerck  Serono,  Darmstadt,  Hesse,  Germany).\nBlastocyst  transfer  was  scheduled  five  days  after\nPregnancy and Microbiota in Letrozole vs GnRH-a Protocols 3\nUnproofed\n\ninitiating progesterone support. Upon confirmation of\nclinical pregnancy, estradiol and vaginal progesterone\nwere tapered off by gestational weeks 7-8, while oral\ndydrogesterone  was  maintained  until  10  weeks\n(Fig.1B). \nBlastocyst morphological evaluation\nEmbryos  were  cultured  to  the  blastocyst  stage,\ntypically achieved on day 5 or day 6 post-fertilization.\nMorphological  assessment  was  performed  according\nto  the  Gardner  and  Schoolcraft  classification  system.\nEmbryos with a grade of 3BC or higher were deemed\nsuitable  for  cryopreservation.  Prior  to  frozen  embryo\ntransfer,  thawed  blastocysts  were  re-evaluated  for\nstructural  integrity  and  developmental  quality.  High-\nquality blastocysts were defined as those graded AA,\nAB, BA, or BB with sufficient expansion, while those\nclassified as AC, CA, BC, CB, or CC were considered\nof  lower  quality,  despite  meeting  the  minimum\nexpansion criterion of grade 3. \nData collection and outcome measures\nClinical characteristics for this study were obtained\nfrom  the  institution's  electronic  database.  Maternal\nand  fetal  outcome  data  were  collected  through\ntelephone interviews with parents one to three months\nafter  the  expected  delivery  date  and  recorded  in  the\nelectronic  medical  records  by  trained  nurses.  The\nprimary  outcome  was  the  live  birth  rate.  Secondary\noutcomes  included  EM  on  the  transfer  day,\nbiochemical  pregnancy  rate,  clinical  pregnancy  rate,\nmiscarriage rate and perinatal outcomes. \n16S rRNA sequencing\nMicrobial genomic DNA was isolated from vaginal\nsamples  using  the  FastPure  Stool  DNA  Isolation  Kit\n(Vazyme, China). DNA concentration was quantified\nwith  a  NanoDrop  2000  spectrophotometer  (Thermo\nFisher Scientific, USA), and purity was confirmed by\nan  A260/A280  ratio  exceeding  1.8.  PCR  was\nconducted  using  primer  pair  338F  (ACTCCT\nACGGGAGGCAGCA)  and  806R  (GGACTACH\nVGGGTWTCTAAT).  The  amplification  protocol\nconsisted of  an  initial  denaturation  at  95  ℃ for\n3  minutes,  followed  by  27  cycles  of  denaturation  at\n95 ℃ for 30 seconds, annealing at 55 ℃ for 30 seconds,\nand extension  at  72  ℃ for  30  seconds,  with  a  final\nelongation step at 72 ℃ for 10 minutes. PCR reactions\nwere  carried  out  using  a  T100  Thermal  Cycler  (Bio-\nRad,  USA).  Sequencing  was  conducted  on  the\nIllumina  NextSeq  2000  PE300  platform.  Raw  data\nwere processed using fastp 0.19.6 software to remove\nlow-quality  sequences  (length < 50  bp).  Sequences\nwere  then  merged  using  FLASH  1.2.11  software.\nChimeric sequences were filtered out. Alpha and beta\ndiversity,  along  with  LEfSe  (linear  discriminant\nanalysis effect size), were analyzed using QIIME 2. \nDroplet digital PCR quantification\nMicrobial  genomic  DNA  was  extracted  using  the\nMagicPure®  32  Microbiome  DNA  Isolation  Kit\n(Fullgene  Biotech,  China).  Lactobacillus  species  and\nGardnerella  vaginalis were  detected  based  on  16S\nrRNA  gene  sequences.  Target  sequences  for  these\nspecies  were  downloaded  from  the  NCBI  database,\nand  primers  and  probes  were  designed  and  validated\nfor  specificity  using  the  NCBI  BLAST  tool.  The\nprimers  and  probes  used  for Lactobacillus were:\nforward  primer  338F  (AGAGGAGAGTGGAACT\nCCA), reverse primer 806R (CTCCCAACACTTAGC\n \nHMG75 IU every other day\nLetrozole 2.5 mg/d\nBlastocyst transfer\n1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 …… 10th week\n10th week\nFD > 18mm\n HCG 5 000-10 000 IU\nGnRH-a 3.75 mg \n1 2 3 4 5 …… 28 29 … 38 39 40 41 42 43 44 45 46 47 48 ……\nEM≥8 mm\nOral E2 4-6 mg/d\nVaginal P 90 mg/d\nFollicle monitoring\nBlastocyst transfer\nOral P 10 mg bid\nOral P 10 mg bid\nA\nB\n \n \nFig.  1   Schematic  representation  of  protocols. A:  Letrozole  +  HMG  protocol.  B:  GnRH-a  HRT  protocol.  Abbreviations:  FD,  follicle\ndiameter; EM, endometrial thickness; E2, estradiol; P, progesterone.\n4 Zhang J et al. J Biomed Res, 2025, 39(0)\nUnproofed\n\nACT),  and  probe  5'-FAM-CTGAGGCTCGAAAG\nCATGGGTAG-BHQ1-3';  for Gardnerella  vaginalis:\nforward primer F (GGTGAGTAATGCGTGACCAA),\nreverse primer R (GCCTACAAGCTGATAGGACG),\nand  probe  P  (5'-HEX-AATAGCTCTTGGAAACG\nGGTGG-BHQ1-3').  The  ddPCR  was  performed  with\nan initial denaturation at 95 ℃ for 5 minutes, followed\nby 40  cycles  of  95  ℃ for 15  seconds  and  58  ℃ for\n25 seconds.\nFluorescent signals from each droplet were detected\nusing the AccuONE Pro chip reader (Zhenuo Biotech,\nChina).  Fluorescence  intensity  was  categorized  by  a\nthreshold as \"1\" (positive) or \"0\" (negative). The total\ncopy number of the target gene was calculated using a\nPoisson distribution model. \nStatistical analyses\nIn  the  retrospective  study,  propensity  score\nmatching  (PSM)  was  used  to  adjust  for  imbalanced\ncovariates,  including  maternal  age,  infertility  type,\nBMI,  type  of  ART,  serum  AMH  level,  blastocyst\nquality,  associated  endometrioma,  associated\nadenomyosis,  and  prior  use  of  GnRH-a  within  3\nmonths.  The  propensity  scores  were  estimated  using\nlogistic  regression,  and  LE  +  HMG  cycles  were\nmatched  with  GnRH-a  HRT  cycles  in  a  1∶1  ratio\nwith a 0.05 caliper to ensure comparability, using the\nnearest neighbor method.\nTo determine independent predictors of live birth in\nEMs  patients  undergoing  FET,  both  univariate  and\nmultivariate  logistic  regression  analyses  were\nconducted  after  PSM.  Prior  to  inclusion  in  the\nmultivariate  model,  all  variables  were  screened  for\nmulticollinearity.  A  backward  stepwise  elimination\nmethod  was  applied  to  identify  significant\nindependent variables.\nAdditionally, subgroup analyses were undertaken in\nspecific  subpopulations.  Given  that  adenomyosis  is  a\nsignificant comorbidity of EMs, the study population\nwas stratified into EMs with or without adenomyosis.\nWithin  each  subgroup,  maternal  age,  infertility  type,\nBMI,  type  of  ART,  serum  AMH  level,  blastocyst\nquality,  associated  endometrioma  and  prior  use  of\nGnRH-a  within  3  months  were  matched  between  the\ntwo groups.\nThe  consistency  between  ddPCR  and  16S  rRNA\nsequencing  results  was  assessed  using  Bland-Altman\nplots  and  intra-class  correlation  coefficients  (ICC).\nBland-Altman  analyses  were  performed  using\nMedCalc software (version 15.6).\nAll  statistical  procedures  were  conducted  using\nSPSS version 26.0 (IBM Corp., USA) and R software\nversion  4.4.1.  For  continuous  variables,  Student’s  t-\ntest  was  applied  when  data  followed  a  normal\ndistribution, while the Mann–Whitney U-test was used\nfor  non-normally  distributed  variables.  Categorical\ndata were analyzed using either the chi-square test or\nFisher’s  exact  test,  as  appropriate.  A  two-sided  P-\nvalue < 0.05 was considered statistically significant. \nResults\n \nBaseline characteristics\nA total of 3,235 cycles from patients with EMs who\nunderwent either LE + HMG or GnRH-a HRT cycles\nwere  screened.  Of  these,  1,156  cycles  from  948\npatients (8 patients had 3 cycles, 192 had 2 cycles, and\n748  had  1  cycle)  were  included.  Exclusion  reasons\nwere  shown  in Fig.2 Among  the  1,156  cycles,  403\nwere  from  the  LE  +  HMG  group  and  753  from  the\nGnRH-a HRT group. After PSM at a 1∶1 ratio, 385\nmatched cycles remained in each group.\nBefore  PSM,  serum  AMH  level,  antral  follicle\ncount (AFC), associated endometrioma, adenomyosis,\nprior use of GnRH-a within 3 months, and blastocyst\nquality  were  different  between  the  groups  (Table  1).\nAfter PSM, no significant differences were found. \nPregnancy and obstetric outcomes\nAs  summarized  in Table  1,  following  PSM,  live\nbirth  rate  (54.0% vs.  53.8%,  P  =  0.942),  clinical\npregnancy rate (62.6% vs. 67.0%, P = 0.200) and EM\non the day of embryo transfer (9.97 mm vs. 9.86 mm,\nP = 0.356) were similar between the LE + HMG and\nGnRH-a HRT groups. Notably, the LE + HMG group\ndemonstrated  a  significantly  lower  biochemical\npregnancy  rate  (67.0% vs.  75.6%,  P  =  0.009)  and  a\ntrend  toward  a  reduced  miscarriage  rate  (13.7% vs.\n19.8%,  P  =  0.070).  Moreover,  patients  in  the  LE  +\nHMG group experienced significantly fewer cesarean\ndeliveries  (64.9% vs.  75.4%,  P  =  0.020)  and  a\nlower incidence of HDP (4.8% vs. 10.1%, P = 0.039)\n(Table 2). Table 3 showed that duration of infertility,\nserum AMH level, associated adenomyosis, transfer of\ngood-quality blastocysts, and EM on transfer day were\nindependent  predictors  for  live  birth  in  EMs  patients\nundergoing FET. \nSubgroup analysis\nIn  the  subgroup  analysis  of  women  with  EMs\nwithout adenomyosis, after PSM, 338 matched cycles\nwere included in each group (Supplemental Table 1).\nThe  GnRH-a  HRT  group  had  a  higher  biochemical\npregnancy  rate  (76.3% vs.  69.2%,  P  =  0.038),  while\nthe  LE  +  HMG  group  had  a  lower  miscarriage  rate\n(10.9% vs. 17.8%, P = 0.037). In the subgroup of EMs\nPregnancy and Microbiota in Letrozole vs GnRH-a Protocols 5\nUnproofed\n\npatients  with  adenomyosis,  after  PSM,  38  matched\ncycles from each group were included (Supplemental\nTable 2). Conversely, the GnRH-a HRT group had a\nhigher live birth rate (42.1% vs. 21.1%, P = 0.048). \n16S rRNA sequencing and OTU analysis\nNo  significant  baseline  differences  were  observed\nbetween  the  two  groups  (Table  4).  After  quality\ncontrol  and  merging,  6,097,496  optimized  sequences\nwere retained for analysis. A total of 352 operational\ntaxonomic  units  (OTUs)  were  identified  across  all\nsamples,  with  207  OTUs  shared  between  the  two\ngroups. The LE + HMG group had 58 unique OTUs,\nwhile the GnRH-a HRT group had 87 unique OTUs,\nrepresenting  16.48% and  24.72% of  the  total  OTUs,\nrespectively. Venn diagram analysis (Fig.3) shows the\noverlap between the groups. \nSpecies diversity analysis\nAlpha  diversity  analysis  revealed  significant\ndifferences  in  the  Ace  index  (P  =  0.039)  and\nsequencing depth (Coverage, P = 0.017) between the\ngroups  (Fig.4A).  However,  no  significant  differences\nwere observed in the Shannon and Simpson diversity\nindices  or  species  evenness  (Pielou_e,  P  =  0.717).\nBeta diversity analysis, based on unweighted UniFrac\ndistances, indicated no significant differences between\nthe  groups  (PCoA,  P  =  0.059;  NMDS,  P  =  0.089)\n(Fig.4B). \nSpecies composition and differential analysis\nAt the genus level, both groups were dominated by\nLactobacillus (75.97% vs.  75.54%), Gardnerella\n(7.14% vs.  6.77%),  and Streptococcus (6.57% vs.\n6.98%) (Fig.5A). However, in genera with abundance\ngreater  than  0.01%,  levels  of Escherichia-Shigella\n(1.17% vs.  0.06%,  P < 0.01), Limosilactobacillus\n(0.63% vs.  0.10%,  P < 0.01),  and Staphylococcus\n(0.33% vs. 0.22%, P < 0.01) were significantly higher\nin the GnRH-a HRT group than the LE + HMG group\n(Fig.5B). \nLEfSe differential analysis\nLEfSe  analysis  (LDA > 2)  revealed  that Pseudo-\nmonadota  taxa,  including Gammaproteobacteria,\nEnterobacterales, Enterobacteriaceae,  and Esche-\nrichia-Shigella,  were  significantly  enriched  in  the\nGnRH-a  HRT  group  (Fig.  5C-D).  Additionally,\nStaphylococcaceae and Staphylococcus were  more\nabundant  in  this  group.  In  contrast,  the  LE  +  HMG\ngroup  exhibited  significantly  higher  levels  of\nBacillaceae and Bacillales. \nDetection  of Lactobacillus and Gardnerella by\nddPCR\nLactobacillus and Gardnerella are among the most\nprevalent bacterial genera in the vaginal microbiota. In\nthis  study,  we  explored  the  potential  application  of\nddPCR  in  reproductive  medicine  by  targeting  these\ntwo  genera.  To  evaluate  the  consistency  between\nddPCR  and  16S  rRNA  sequencing,  we  log-\ntransformed  the  copy  numbers  obtained  from  ddPCR\nand the relative abundances from 16S rRNA data for\nboth Lactobacillus and Gardnerella.  The  ICC  was\n0.875  (95% CI:  0.815 - 0.916),  indicating  excellent\n \nEndometriosis women underwent FET using letrozole+HMG\nor GnRH-a HRT without PGT from 2016 to 2023\n(n=3 235)\nExclusion cycles (n=2 079)\n· FET cycle serial number>3 (n=621)\n· Day-3/4 frozen embryo transfer (n=1315)\n· Age at FET>42 years (n=61)\n· Congenital uterine malformations (n=15)\n· Recurrent spontaneous abortion (n=9)\n· Multiple gestations (n=15)\n· Data missing (n=43)\nEndometriosis women underwent frozen-thawed blastocyst\ntransfers using letrozole+HMG or GnRH-a HRT\n(n=1 156)\nLetrozle+HMG group\n(n=403)\nLetrozole+HMG group\n(n=385)\nGnRH-a HRT group\n(n=753)\nGnRH-a HRT group\n(n=385)\nPSM 1: 1\n \n \nFig. 2   Flowchart of patient inclusion and exclusion criteria. Abbreviations: FET, frozen embryo transfer; HMG, human menopausal\ngonadotropin; GnRH-a, gonadotropin-releasing hormone agonist; HRT, hormone replacement therapy; PGT, preimplantation genetic testing;\nPSM, propensity score matching.\n6 Zhang J et al. J Biomed Res, 2025, 39(0)\nUnproofed\n\nagreement (ICC ≥ 0.75), which was further supported\nby Bland-Altman analysis (Fig. 6A).\nFor Lactobacillus,  the  median  copy  number\ndetected  by  ddPCR  was  1.03×108 (IQR:  6.28×106 -\n3.97×108)  in  the  GnRH-a  HRT  group  and  1.90×108\n(IQR: 1.78×107 - 7.35×108) in the LE + HMG group.\nThe  difference  between  the  two  groups  was  not\nstatistically significant (P = 0.280) (Fig. 6B), a result\nconsistent  with  16S  rRNA  sequencing  data  (P  =\n0.601) (Fig. 6C).\nFor Gardnerella, ddPCR detected positive samples\nin  73.3% of  cases,  compared  to  58.1% detected  by\n16S  rRNA  sequencing,  suggesting  that  ddPCR  may\noffer  greater  sensitivity.  The  median Gardnerella\n \nTable 1   Baseline characteristics and assisted reproductive pregnancy outcomes between the two groups before and after PSM\nVariables\nBefore PSM After PSM\nLE + HMG\n(n=403)\nGnRH-a HRT\n(n=753) P-value LE + HMG\n(n=385)\nGnRH-a HRT\n(n=385) P-value\nMaternal age at FET (y) 30.4±3.6 30.8±3.3 0.053 30.5±3.6 30.7±3.3 0 .426\nPaternal age at FET (y) 31.3±3.8 31.7±4.0 0.150 32.0±3.9 31.7±4.1 0 .389\nDuration of infertility (y) 3.6±2.7 3.3±2.5 0.067 3.6±2.6 3.5±2.6 0 .530\nBMI (kg/m2) 21.7±2.7 21.5±2.8 0.257 21.6±2.7 21.7±2.9 0 .871\nAMH (ng/mL) 6.0±4.2 5.2±3.9 0.001 5.8±4.0 5.7±3.9 0 .825\nBasal FSH level (IU/L) 6.9 (5.8-8.1) 7.1 (5.8-8.5) 0.104 6.9 (5.9-8.2) 6.9 (5.7-8.3) 0 .848\nAFC 16.0±6.1 14.6±6.2 <0.001 15.9±6.2 15.7±6.3 0 .686\nType of infertility, %(n) 0.551 0 .530\n　Primary 70.7 (285) 72.4 (545) 70.9 (273) 68.8 (265)\n　Secondary 29.3 (118) 27.6 (208) 29.1 (112) 31.2 (120)\nAssociated endometrioma, %(n) 52.1 (210) 70.0 (527) <0.001 54.3 (209) 53.0 (204) 0 .718\nAssociated adenomyosis, %(n) 10.2 (41) 17.9 (135) <0.001 10.7 (41) 11.4 (44) 0 .730\nOS protocol, %(n) 0.047 0 .099\n　Agonist protocol 65.3 (263) 67.9 (511) 66.5 (256) 64.7 (249)\n　Antagonist protocol 28.3 (114) 28.8 (217) 27.0 (104) 31.7 (122)\n　Other protocols 6.5 (26) 3.3 (25) 6.5 (25) 3.7 (14)\nType of ART, %(n) 0.478 0 .531\n　IVF 80.7 (325) 82.3 (620) 80.5 (310) 78.7 (303)\n　ICSI 19.4 (78) 17.7 (133) 19.5 (75) 21.3 (82)\nNo. of oocytes retrieved 11.9±4.7 11.1±4.4 0.003 11.7±4.6 11.6±4.5 0 .807\nNo. of 2PN 9.8±4.2 9.2±4.0 0.024 9.6±4.1 9.8±4.1 0 .604\nFertilization rate 0.8±0.2 0.8±0.2 0.335 0.8±0.2 0.8±0.2 0 .189\nViable embryos 8.9±3.9 8.3 (3.6) 0.011 8.8±3.8 8.8±3.8 0 .901\nBlastocyst formation rate 0.6±0.3 0.6±0.2 0.081 0.6±0.3 0.6±0.2 0 .279\nPrior GnRH-a within 3 months, %(n) 17.9 (72) 28.8 (217) <0.001 18.7 (72) 20.0 (77) 0 .648\nGood-quality blastocyst transfer, %(n) 65.0 (262) 57.9 (436) 0.018 63.4 (244) 62.6 (241) 0 .823\nEM on the transfer day (mm) 10.0±1.7 9.9±1.6 0.822 10.0±1.7 9.9±1.6 0 .356\nBiochemical pregnancy rate, %(n/N) 67.7 (273/403) 73.3 (552/753) 0.046 67.0 (258/385) 75.6 (291/385) 0 .009\nClinical pregnancy rate, %(n/N) 63.3 (255/403) 73.3 (552/753) 0.403 62.6 (241/385) 67.0 (258/385) 0 .200\nLive birth rate, %(n/N) 55.1 (222/403) 53.5 (403/753) 0.610 54.0 (208/385) 53.8 (207/385) 0 .942\nMiscarriage rate, %(n/N) 12.9 (33/255) 18.6 (92/495) 0.049 13.7 (33/241) 19.8 (52/258) 0 .070\nAbbreviations: LE, letrozole; HMG, human menopausal gonadotropin; FET, frozen embryo transfer; GnRH-a, gonadotropin-releasing hormone agonist; HRT, hormone\nreplacement therapy; BMI, body mass index; AMH, anti-Mullerian hormone; FSH, follicle-stimulating hormone; AFC, antral follicle count; OS, ovarian stimulation;\nART,  assisted  reproductive  technique;  IVF,  in  vitro  fertilization;  ICSI,  intracytoplasmic  sperm  injection;  2PN,  two  pronuclei;  EM,  endometrial  thickness.  PSM,\npropensity score matching.\nPregnancy and Microbiota in Letrozole vs GnRH-a Protocols 7\nUnproofed\n\n \nTable 2   Perinatal outcomes between the two groups before and after PSM\nOutcomes\nBefore PSM After PSM\nLE + HMG\n(n=222)\nGnRH-a HRT\n(n=403) P-value LE + HMG\n(n=208)\nGnRH-a HRT\n(n=207) P-value\nGestational age(weeks) 38 (38-39) 39 (38-39) 0.459 38 (38-39) 39 (38-39) 0 .185\nBirth weight (g) 3 410.4±511.3 3 412.3±503.2 0.966 3 404.9±518.7 3 435.7±523.6 0 .547\nDelivery mode, %(n) <0.001 0 .020\n　Vaginal birth 35.6 (79) 21.6 (87) 35.1 (73) 24.6 (51)\n　Caesarean section 64.4 (143) 78.4 (316) 64.1 (135) 75.4 (156)\nNewborn sex, %(n) 0.849 0 .404\n　Female 41.9 (93) 42.7 (172) 41.4 (86) 45.4 (94)\n　Male 58.1 (129) 57.3 (231) 58.7 (122) 54.6 (113)\nLBW, %(n) 3.2 (7) 4.2 (17) 0.507 3.4 (7) 5.3 (11) 0 .330\nMacrosomia, %(n) 9.9 (22) 9.9 (40) 0.995 10.1 (21) 11.6 (24) 0 .624\nLGA, %(n) 2.7 (6) 3.7 (15) 0.499 2.9 (6) 5.8 (12) 0 .145\nSGA, %(n) 17.6 (39) 17.9 (72) 0.926 16.4 (34) 18.8 (39) 0 .505\nPTB, %(n) 9.0 (20) 7.7 (31) 0.565 9.1 (19) 7.3 (15) 0 .483\nGDM, %(n) 10.8 (24) 10.4 (42) 0.880 11.1 (23) 13.0 (27) 0 .534\nHDP, %(n) 4.5 (10) 9.2 (37) 0.034 4.8 (10) 10.1 (21) 0 .039\nPlacenta previa, %(n) 3.2 (7) 5.5 (22) 0.190 3.4 (7) 3.4 (7) 0 .993\nAbbreviations: LE, letrozole; HMG, human menopausal gonadotropin; GnRH-a, gonadotropin-releasing hormone agonist; HRT, hormone replacement therapy; LBW,\nlow birthweight; LGA, large for gestational age; SGA, small for gestational age; PTB, preterm birth; GDM, gestational diabetes mellitus; HDP, hypertensive disorders\nof pregnancy; PSM, propensity score matching.\n \nTable 3   Univariate and multivariate logistic regression analysis of the live birth after PSM\nVariables\nUnivariate analysis Multivariate analysis\nOR (95% CI) P-value OR (95% CI) P-value\nMaternal age at FET (years) 0.94 (0.90, 0.98) 0 .002\nPaternal age at FET (years) 0.98 (0.95, 1.02) 0 .359\nDuration of infertility (years) 0.95 (0.90, 1.01) 0 .087 0.94 (0.88, 0.99) 0.019\nBMI (kg/m2) 1.01 (0.96, 1.06) 0 .680\nAMH (ng/mL) 1.07 (1.03, 1.11) <0 .001 1.06 (1.02, 1.10) 0.003\nBasal FSH level (IU/L) 1.00 (0.99, 1.02) 0 .564\nAFC 1.02 (1.00, 1.05) 0 .055\nType of infertility (Secondary vs. Primary) 0.88 (0.65, 1.20) 0 .427\nAssociated endometrioma (Yes vs. No) 0.87 (0.66, 1.16) 0 .339\nAssociated adenomyosis (Yes vs. No) 0.38 (0.24, 0.61) <0 .001 0.44 (0.27, 0.72) 0.001\nPrior use of GnRH-a within 3 months (Yes vs. No) 0.84 (0.59, 1.20) 0 .332\nEndometrial preparation protocol (GnRH-a vs. LE + HMG) 0.99 (0.75, 1.31) 0 .942\nGood-quality blastocyst transfer (Yes vs. No) 1.95 (1.45, 2.62) <0 .001 1.71 (1.26, 2.33) 0.001\nEM on the transfer day (mm) 1.12 (1.02, 1.22) 0 .015 1.11 (1.03, 1.12) 0.013\nAbbreviations: LE, letrozole; HMG, human menopausal gonadotropin; FET, frozen embryo transfer; GnRH-a, gonadotropin-releasing hormone agonist; BMI, body\nmass  index;  AMH,  anti-Mullerian  hormone;  FSH,  follicle-stimulating  hormone;  AFC,  antral  follicle  count;  EM,  endometrial  thickness;  PSM,  propensity  score\nmatching.\n8 Zhang J et al. J Biomed Res, 2025, 39(0)\nUnproofed\n\ncopy number in the GnRH-a HRT group was 6.63×103\n(IQR: 0 - 5.75×105), while in the LE + HMG group it\nwas 2.81×104 (IQR: 0 - 3.94×105), with no significant\nintergroup  difference  (P  =  0.393)  (Fig.  6D).  This\nfinding  was  also  consistent  with  16S  rRNA\nsequencing  results,  which  showed  similar  relative\nabundances  of Gardnerella between  the  two  groups\n(P = 0.429) (Fig. 6E). \nDiscussion\nTo our knowledge, this is the first large-scale study\ncomparing pregnancy and perinatal outcomes between\nthe LE + HMG and GnRH-a HRT groups in women\nwith EMs. To date, only two studies have attempted to\nevaluate different endometrial preparation protocols in\nthis  population.  One  study  compared  GnRH-a  HRT,\nHRT,  and  NC  protocols  but  did  not  include  any  LE-\nbased regimens[7]. The other study incorporated LE +\nHMG  in  its  comparison  alongside  GnRH-a  HRT,\nHRT,  and  NC  protocols[8];  however,  it  involved  only\n42  LE  +  HMG  cycles,  substantially  limiting  its\nstatistical  power  and  the  generalizability  of  its\nfindings.\nThe  LE  +  HMG  group  showed  lower  rates  of\nmiscarriage,  cesarean  delivery,  and  HDP.  These\nfindings  may  be  partly  explained  by  the  presence  or\nabsence of the CL in different endometrial preparation\nprotocols.  In  conventional  HRT  cycles,  with  or\nwithout  GnRH-a  pretreatment,  the  hypothalamic–\npituitary–ovarian  axis  is  suppressed,  resulting  in  the\nabsence  of  a  functional  CL.  The  CL  plays  a  pivotal\nrole  in  early  pregnancy  by  producing  not  only\nestradiol and progesterone but also key vasoactive and\nangiogenic  factors  such  as  relaxin  and  vascular\nendothelial  growth  factor  (VEGF).  These  hormones\nare  essential  for  embryo  implantation,  endometrial\ndecidualization,  and  proper  placental  development[25].\nA  deficiency  in  these  factors  may  impair  vascular\nremodeling  and  placental  formation,  potentially\ncontributing  to  abnormal  implantation  and  an\nincreased  risk  of  miscarriage.  Moreover,  relaxin  is\ninvolved  in  cardiovascular  adaptations  during\npregnancy;  its  absence  has  been  implicated  in  the\npathophysiology  of  preeclampsia  and  other\nhypertensive  complications[15].  The  LE  +  HMG\nprotocol,  by  preserving  ovulation  and  CL  function\nthrough mild ovarian stimulation, ensures endogenous\nproduction  of  these  crucial  hormones.  This  may\nunderlie the observed reduction in HDP incidence and\ncesarean  section  rates  in  the  LE  group.  Nonetheless,\nfurther  investigation  is  warranted  to  elucidate  the\nassociation  between  HRT  protocols  and  obstetric\ncomplications, such as miscarriage and HDP.\nBlastocyst  quality  emerged  as  an  independent\npredictor  of  live  birth  in  our  analysis.  Accumulating\nevidence  suggests  that  impaired  endometrial\nreceptivity may not be the principal factor underlying\nimplantation  failure  in  ART,  even  in  patients  with\nEMs[26]. Rather, embryo quality appears to play a more\ncritical  role  in  determining  successful  pregnancy\noutcomes[27].  Additionally,  the  presence  of\nadenomyosis  was  independently  associated  with\nreduced  live  birth  rates  among  women  with  EMs,\nemphasizing  the  importance  of  considering\nadenomyosis  as  a  significant  confounding  factor.  Its\ndetrimental  effect  on  reproductive  outcomes  may\nsurpass that of EMs alone[28]. In our subgroup analysis\ninvolving  patients  diagnosed  with  both  EMs  and\nadenomyosis,  those  undergoing  GnRH-a  HRT\ndemonstrated  a  significantly  higher  live  birth  rate\ncompared  to  those  treated  with  the  LE  +  HMG\nprotocol.  This  finding  suggests  that  GnRH-a\n \nTable  4   Baseline  characteristics  of  endometriosis  patients\nfor investigation of the vaginal microbiota\nVariables\nLE+\nHMG\n(n=55)\nGnRH-a\nHRT\n(n=50)\nP-value\nMaternal age at FET (y) 33.1±3.8 32.8±3.9 0.705\nDuration of infertility (y) 3.0±2.7 3.3±2.7 0.544\nBMI (kg/m2) 21.8±2.1 22.3±3.0 0.312\nAMH (ng/mL) 4.7±4.1 3.8±2.5 0.213\nBasal FSH level (IU/L) 7.0±2.2 7.7±2.1 0.172\nAFC 13.7±5.7 14.2±6.8 0.679\nPrimary infertility, %(n) 65.5 (36) 52.0 (26) 0.161\nAssociated endometrioma, %(n) 47.3 (26) 58.0 (29) 0.329\nAssociated adenomyosis, %(n) 14.6 (8) 20.0 (10) 0.605\nAbbreviations: LE, letrozole; HMG, human menopausal gonadotropin; FET,\nfrozen  embryo  transfer;  GnRH-a,  gonadotropin-releasing  hormone  agonist;\nHRT,  hormone  replacement  therapy;  BMI,  body  mass  index;  AMH,  anti-\nMullerian  hormone;  FSH,  follicle-stimulating  hormone;  AFC,  antral  follicle\ncount.\n \nGnRH-a\n87\n24.72%\n207\n58.81%\n58\n16.48% LE\n \n \nFig.  3   Venn  diagram showing  shared  and  unique  operational\ntaxonomic  units  (OTUs)  between  the  GnRH-a  HRT  group\n(GnRHa) and the letrozole + HMG group (LE).\nPregnancy and Microbiota in Letrozole vs GnRH-a Protocols 9\nUnproofed\n\npretreatment may be particularly beneficial for women\nwith adenomyosis, aligning with previous studies that\nsupport  the  use  of  GnRH-a  to  improve  reproductive\noutcomes in this subgroup[29].\nThis study also systematically evaluated the impact\nof  the  two  endometrial  preparation  protocols  on  the\nstructure of the vaginal microbiota and the abundance\nof  key  bacterial  taxa.  Notably,  the  GnRH-a  HRT\nprotocol was associated with an increased prevalence\nof  potentially  pathogenic  bacteria,  such  as\nEscherichia-Shigella and Staphylococcus.  This  may\nbe attributable to the hypoestrogenic state induced by\nGnRH-a  treatment.  Estrogen  is  known  to  play  a\ncrucial  role  in  maintaining  mucosal  immunity  by\nmodulating  the  expression  of  antimicrobial  peptides\n(AMPs),  such  as  defensins  and  secretory  leukocyte\nprotease inhibitors, within the reproductive tract [30-31].\nA reduction in estrogen levels may lead to decreased\nAMP expression, thereby compromising local defense\nmechanisms  and  facilitating  colonization  by\nopportunistic pathogens in the vaginal environment.\nWhile  16S  rRNA  sequencing  is  widely  used  for\nqualitative  analysis  and  to  determine  microbial\ndiversity  and  relative  abundance,  it  has  limitations\nsuch as lower resolution, reduced detection efficiency\nfor  certain  genera,  and  longer  testing  periods\n \nA\nB\n2.0 1.1 120\n110 0.65\n0.5 GnRH-a\nLE\nGnRH-a\nLE\n−0.4\n0.4\n1.8 1.0 110\n100 0.60\n0.4\n−0.3\n0.2\n1.6 0.9 100\n90 0.55\n0.3\n−0.2\n0\n1.4 0.8\n90\n80 0.50\n0.2\n−0.1\n−0.2\n1.2 0.7\n80\n70 0.45\n0.1\n0\nNMDS1\nNMDS2\n−0.4\n1.0 0.6\n70\n60\n0.40\n0\n0.1\n−0.6\n0.8 0.5\n60\n50\n0.35\n−0.1\n0.2\n0.6\n0.4\n50\n40\n0.30\n−0.2\n0.3\n0.4\n0.3\n40\n30\n0.25\n−0.3\n0.4\n0.2\n0.2\n30\n20\n0.20\n−0.4\n0.5−0.4 −0.3 −0.2 −0.1 0 0.1 0.2 0.3 0.4\n0\n0.1\n1.000 1\n1.00e+0\n1\n1.00e+0\n0.999 9\n1.00e+0\n0.999 8\n1.00e+0\n0.999 7\n1.00e+0\n0.999 6\n1.00e+0\n0.999 5\n9.99e−1\n0.999 4\n20\n10\n0.15\n−0.5\n−0.2\n10\n0\n0.10\n0\n−10\n0.05\n0\n−0.05\n−10\nShannon index\nGnRH-a LE GnRH-a LE GnRH-a LE\nGnRH-a LE GnRH-a LE GnRHa LE\nShannon index of OTU level Chao index of OTU level\nPielou_e index of OTU level\nCoverage index of OTU level\nSimpson index of OTU level\nAce index of OTU level\nP=0.669 6\nSimpson index\nP=0.688 4\nAce index\n*\nP=0.039 13\nChao index\nP=0.659 9\nPCoA on OTU level\nPC2(13.09%)\nPC1(19.19%)\nR2=0.017 2, P=0.059\nNMDS on OTU level\nstress: 0.075, R2=0.020 4, P=0.089\nCoverage index\nP=0.017 06\nPielou_e index\nP=0.717\n \n \nFig. 4   Comparison of vaginal microbiota diversity between GnRH-a HRT (GnRHa) and letrozole + HMG (LE) groups. A: Alpha\ndiversity indices (Shannon, Simpson, ACE, Chao, Coverage, and Pielou_e). B: Beta diversity analysis by Principal Coordinates Analysis\n(PCoA) and Non-metric Multidimensional Scaling (NMDS). P-values < 0.05 were considered statistically significant.\n10 Zhang J et al. J Biomed Res, 2025, 39(0)\nUnproofed\n\n(typically  5-7  days).  In  contrast,  ddPCR  offers  high\nsensitivity,  specificity,  and  rapid  diagnostic\ncapabilities  (within  3  hours).  In  our  study,  we\nsuccessfully  established  a  ddPCR  assay  for  the\ndetection  of Lactobacillus and Gardnerella by\ndesigning  specific  primers  and  optimizing  reaction\nconditions.  The  ddPCR  results  showed  excellent\nconcordance  with  16S  rRNA  sequencing,  further\nvalidating  its  clinical  applicability.  Notably,  ddPCR\nexhibited  superior  sensitivity  in  detecting\nGardnerella,  a  low-abundance  but  potentially\npathogenic bacterium that may be underrepresented in\nsequencing-based  analyses.  This  highlights  the\nadvantage of ddPCR in precisely identifying clinically\nrelevant microorganisms within the reproductive tract.\nTaken  together,  these  findings  support  a  two-step\nmicrobial  detection  strategy:  initial  screening  using\n16S  rRNA  sequencing  to  identify  microbiota  shifts\nassociated  with  reproductive  outcomes,  followed  by\ntargeted  ddPCR  analysis  for  rapid  and  accurate\npathogen detection.\nThis  study  has  several  limitations  that  should  be\nacknowledged.  First,  it  was  a  single-center,\nretrospective  analysis,  which  may  be  subject  to\ninherent  selection  biases.  Prospective,  multicenter\nrandomized  controlled  trials  (RCTs)  are  needed  to\nvalidate  our  findings.  Second,  NC  and  pure  HRT\nprotocols  were  not  included  due  to  their  limited\napplication  at  our  center,  which  may  restrict  the\ngeneralizability  of  the  results.  Third,  some  patients\nunderwent  multiple  FET  cycles,  which  could\nintroduce  intra-patient  variability  and  potential\nconfounding.  Additionally,  perinatal  outcomes  were\ncollected through telephone interviews, which may be\nless  accurate  than  those  obtained  from  standardized\nmedical record reviews.\nFor  the  microbiota  analysis,  we  intended  to\ncompare the vaginal microbiota between live birth and\nnon-live  birth  groups.  However,  after  matching  for\nkey  confounders  (e.g.,  age,  blastocyst  quality,\nadenomyosis),  only  11  samples  remained  in  each\ngroup.  Preliminary  results  showed  no  significant\ndifferences,  and  the  small  sample  size  limited\ninterpretability; thus, detailed data were not presented.\nFurther studies with larger sample sizes are needed to\nclarify  the  potential  relationship  between  vaginal\nmicrobiota and reproductive outcomes.\nFinally,  the  ddPCR  analysis  in  this  study  was\nlimited to Lactobacillus and Gardnerella, and did not\nencompass  other  potentially  pathogenic  taxa  such  as\nEscherichia, Shigella, Staphylococcus, Streptococcus,\nand Enterococcus.\nIn  conclusion,  our  findings  suggest  that  live  birth\nrates  were  comparable  between  the  LE  +  HMG  and\n \nA\nC\nB\nD\nTop 10 genus\nLactobacillus\nGardnerella\nStreptococcus\nBifidobacterium\nFannyhessea\nPrevotella\nEscherichia-Shigella\nEnterococcus\nLimosilactobacillus\nAerococcus\nRelative abundance on genus level\n100\nGnRH-a LE\n80\n60\n40\n20\n0\nEscherichia-Shigella\nLimosilactobacillus\nStaphylococcus\nLawsonella\nNovosphingobium\nBlautia\nRuminococcus\nSutterella\nSegatella\nWeizmannia\nBar plot on genus level\n95% Cl P value\nProportions (%) Difference between proportions (%)\n0 0.5 0 1.0 1.5 1 2 3\nGnRH-a\nLE\n0.002 409 0\n0.000 911 6\n0.000 292 7\n0.027 080 0\n0.031 820 0\n0.043 650 0\n0.020 210 0\n0.015 420 0\n0.022 570 0\n0.008 694 0\na: p_Pseudomonadota\nb: c_Alphaproteobacteria\nc: c_Gammaproteobacteria\nd: o_Bacillales\ne: o_Enterobacterales\nf: o_Lysobacterales\ng: o_Staphylococcales\nh: f_Bacillaceae\ni: f_Enterobacteriaceae\nj: f_Lysobacteraceae\nk: f_Staphylococcaceae\n1: g_Escherichia-Shigella\nm: g_Lawsonella\nn: g_Limosilactobacillus\no: g_Staphylococcus\np_Pseudomonadota\nf_Enterobacteriaceae\nc_Gammaproteobacteria\no_Enterobacterales\ng_Escherichia-Shigella\ng_Limosilactobacillus\ng_Staphylococcus\nf_Staphylococcaceae\no_Staphylococcales\nc_Alphaproteobacteria\nf_Bacillaceae\no_Bacillales\nLEfSe bar\nGnRH-a\nLE\n0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0\nLDA SCORE\np_Pseudomonadota\n \n \nFig. 5   Comparison of vaginal microbiota composition between GnRH-a HRT (GnRHa) and letrozole + HMG (LE) groups. A: Bar\nplots depicting the relative abundance of the top 10 genera at the genus level. B: Genus-level microbial taxa showing statistically significant\ndifferences in abundance between the two groups. C: Cladogram generated by LEfSe analysis. Each concentric circle represents a taxonomic\nlevel, from phylum (center) to class, order, family, and genus (outer layers). D: Linear Discriminant Analysis (LDA) score plot showing the\neffect size of taxa with significant intergroup differences.\nPregnancy and Microbiota in Letrozole vs GnRH-a Protocols 11\nUnproofed\n\nGnRH-a  HRT  protocols  in  women  with  EMs\nundergoing FET. However, the LE + HMG group was\nassociated  with  a  lower  incidence  of  pregnancy\ncomplications.  The  GnRH-a  HRT  protocol  may  be\nlinked  to  an  increased  risk  of  colonization  by\npathogenic  vaginal  bacteria,  underscoring  the\nimportance  of  monitoring  microbial  changes  and\nconsidering  appropriate  interventions  in  patients\nreceiving  this  protocol.  Moreover,  ddPCR\ndemonstrated promising clinical utility as a rapid and\nprecise  tool  for  the  detection  of  key  vaginal\nmicroorganisms,  supporting  its  potential  role  in\nclinical practice. \nFundings\nThis  work  was  supported  by  Key  Program  of\nNational  Nature  and  Science  Foundation  of  China\n(81730041),  the  National  Key  Research  and\nDevelopment Program of China (2021YFC2700404). \nAcknowledgments\nWe would like to thank Dr. Jin Liu for his valuable\nguidance on the statistical analysis in this study.\nReferences\n Macer  ML,  Taylor  HS. Endometriosis  and  infertility:  a\nreview  of  the  pathogenesis  and  treatment  of  endometriosis-\nassociated infertility[J]. Obstet Gynecol Clin North Am, 2012,\n39(4): 535–549.\n[1]\n Georgiou  EX,  Melo  P,  Baker  PE,  et  al. Long-term  GnRH\nagonist  therapy  before  in  vitro  fertilisation  (IVF)  for\nimproving  fertility  outcomes  in  women  with\nendometriosis[J]. Cochrane  Database  Syst  Rev, 2019,\n2019(11): CD013240.\n[2]\n Becker  CM,  Bokor  A,  Heikinheimo  O,  et  al. ESHRE\nguideline:  endometriosis[J]. Hum  Reprod  Open, 2022,\n2022(2): hoac009.\n[3]\n Ferrero  S,  Remorgida  V,  Maganza  C,  et  al. Aromatase  and\nendometriosis:  estrogens  play  a  role[J]. Ann  N  Y  Acad  Sci,\n2014, 1317(1): 17–23.\n[4]\n Brosens  J,  Verhoeven  H,  Campo  R,  et  al. High  endometrial\naromatase  P450  mRNA  expression  is  associated  with  poor\nIVF outcome[J]. Hum Reprod, 2004, 19(2): 352–356.\n[5]\n \n8\n−6\n1.3\n6\n−4\n1.2\n4\n−2\n1.1\n2\n0\n1.0\n0\n2\n0.9Proportion of sequences\n−2\n4\n0.8\n−4\n6\n0.7\n0.6\n0.5\n0.4\n0.3\n0.2\n0.1\n0\n1.3\n1.2\n1.1\n1.0\n0.9 Proportion of sequences\n0.8\n0.7\n0.6\n0.5\n0.4\n0.3\n0.2\n0.1\n0\n+1.96 SD\nmean\n0.4\n−1.96 SD\nddPCR vs. 16S rRNA\nLog10-Gardnerella/lactobacillus GnRH-a\n4×109\n3×109\n2×109\n1×109\n0\n−1×109\n6×108\n4×108\n2×108\n0\n−2×108\nLE\nGnRH-a LE GnRH-a LE\nGnRH-a LE\nns\nns\nNS\nNS\n3.0\n−2.3\nA B\nD E\nC\n \n \nFig. 6   ddPCR and 16S rRNA sequencing in quantifying core vaginal microbiota. A: Bland–Altman plot demonstrating that most data\npoints fall within the 95% confidence interval, indicating strong concordance between the two methods. B: Detection of Lactobacillus by\nddPCR. C: Detection of Lactobacillus by 16S rRNA sequencing. D: Detection of Gardnerella by ddPCR. E: Detection of Gardnerella by\n16S rRNA sequencing. P-values < 0.05 were considered statistically significant.\n12 Zhang J et al. J Biomed Res, 2025, 39(0)\nUnproofed\n\n Tamura H, Takasaki A, Nakamura Y, et al. A pilot study to\nsearch  possible  mechanisms  of  ultralong  gonadotropin-\nreleasing  hormone  agonist  therapy  in  IVF-ET  patients  with\nendometriosis[J]. J Ovarian Res, 2014, 7(1): 100.\n[6]\n Guo Y, Fang Z, Yu L, et al. Which endometrial preparation\nprotocol  provides  better  pregnancy  and  perinatal  outcomes\nfor  endometriosis  patients  in  frozen-thawed  embryo  transfer\ncycles? A retrospective study on 1413 patients[J]. J Ovarian\nRes, 2023, 16(1): 7.\n[7]\n Yang  J,  Wen  Y,  Li  D,  et  al. Retrospective  analysis  of  the\nendometrial preparation protocols for frozen-thawed embryo\ntransfer cycles in women with endometriosis[J]. Reprod Biol\nEndocrinol, 2023, 21(1): 83.\n[8]\n Vinogradova  Y,  Coupland  C,  Hippisley-Cox  J. Use  of\nhormone  replacement  therapy  and  risk  of  venous\nthromboembolism:  nested  case-control  studies  using  the\nQResearch and CPRD databases[J]. BMJ, 2019, 364: k4810.\n[9]\n Singh B, Reschke L, Segars J, et al. Frozen-thawed embryo\ntransfer:  the  potential  importance  of  the  corpus  luteum  in\npreventing  obstetrical  complications[J]. Fertil  Steril, 2020,\n113(2): 252–257.\n[10]\n Palomba S. Aromatase inhibitors for ovulation induction[J]. J\nClin Endocrinol Metab, 2015, 100(5): 1742–1747.\n[11]\n Jirge PR, Patil RS. Comparison of endocrine and ultrasound\nprofiles  during  ovulation  induction  with  clomiphene  citrate\nand letrozole in ovulatory volunteer women[J]. Fertil Steril,\n2010, 93(1): 174–183.\n[12]\n Zhang  J,  Wei  M,  Bian  X,  et  al.  Letrozole-induced  frozen\nembryo  transfer  cycles  are  associated  with  a  lower  risk  of\nhypertensive  disorders  of  pregnancy  among  women  with\npolycystic  ovary  syndrome[J]. Am  J  Obstet  Gynecol,  2021,\n225(1): 59. e1–59. e9.\n[13]\n Wang X, Li Y, Tan H, et al. Letrozole-stimulated endometrial\npreparation  protocol  is  a  superior  alternative  to  hormone\nreplacement  treatment  for  frozen  embryo  transfer  in  women\nwith  polycystic  ovary  syndrome,  a  cohort  study[J]. Reprod\nBiol Endocrinol, 2023, 21(1): 101.\n[14]\n Sharon-Weiner  M,  Farladansky-Gershnabel  S,  Schreiber  H,\net  al. Clinical  pregnancy  rates  among  anovulatory  and\noligoovulatory  women  after  letrozole  versus  hormone\nreplacement  therapy  in  frozen-thawed  embryo  transfer\ncycles[J]. Hum Fertil, 2023, 26(1): 107–114.\n[15]\n Lall  Seal  S,  Kamilya  G,  Mukherji  J,  et  al.  Aromatase\ninhibitors in recurrent ovarian endometriomas: report of five\ncases with literature review[J]. Fertil Steril, 2011, 95(1): 291.\ne15–291. e18.\n[16]\n Sasson IE, Taylor HS. Aromatase inhibitor for treatment of a\nrecurrent abdominal wall endometrioma in a postmenopausal\nwoman[J]. Fertil Steril, 2009, 92(3): 1170. e1–1170. e4.\n[17]\n Attar E, Bulun SE. Aromatase and other steroidogenic genes\nin  endometriosis:  translational  aspects[J]. Hum  Reprod\nUpdate, 2006, 12(1): 49–56.\n[18]\n Miller  PB,  Parnell  BA,  Bushnell  G,  et  al. Endometrial\nreceptivity  defects  during  IVF  cycles  with  and  without\nletrozole[J]. Hum Reprod, 2012, 27(3): 881–888.\n[19]\n Brosens  J,  Verhoeven  H,  Campo  R,  et  al.  High  endometrial\naromatase  P450  mRNA  expression  is  associated  with  poor\nIVF outcome[J]. Hum Reprod, 2004, 19(2): 352–356.\n[20]\n Khan  KN,  Fujishita  A,  Masumoto  H,  et  al. Molecular\ndetection  of  intrauterine  microbial  colonization  in  women\nwith  endometriosis[J]. Eur  J  Obstet  Gynecol  Reprod  Biol,\n2016, 199: 69–75.\n[21]\n Chen  C,  Song  X,  Wei  W,  et  al. The  microbiota  continuum\nalong the female reproductive tract and its relation to uterine-\nrelated diseases[J]. Nat Commun, 2017, 8(1): 875.\n[22]\n Wang  J,  Li  Z,  Ma  X,  et  al. Translocation  of  vaginal\nmicrobiota  is  involved  in  impairment  and  protection  of\nuterine health[J]. Nat Commun, 2021, 12(1): 4191.\n[23]\n Van  den  Bosch  T,  Dueholm  M,  Leone  FPG,  et  al. Terms,\ndefinitions  and  measurements  to  describe  sonographic\nfeatures  of  myometrium  and  uterine  masses:  a  consensus\nopinion  from  the  Morphological  Uterus  Sonographic\nAssessment  (MUSA)  group[J]. Ultrasound  Obstet  Gynecol,\n2015, 46(3): 284–298.\n[24]\n Pereira MM, Mainigi M, Strauss III JF. Secretory products of\nthe corpus luteum and preeclampsia[J]. Hum Reprod Update,\n2021, 27(4): 651–672.\n[25]\n Pirtea P, de Ziegler D, Ayoubi JM. Endometrial receptivity in\nadenomyosis  and/or  endometriosis[J]. Fertil  Steril, 2023,\n119(5): 741–745.\n[26]\n Casalechi M, Reschini M, Palermo MC, et al. Is endometrial\nreceptivity  affected  in  women  with  endometriosis?  Results\nfrom  a  matched  pair  case-control  study  of  assisted\nreproductive  technology  treatments[J]. Reprod  Biomed\nOnline, 2023, 47(6): 103414.\n[27]\n Vercellini  P,  Viganò  P,  Bandini  V,  et  al. Association  of\nendometriosis  and  adenomyosis  with  pregnancy  and\ninfertility[J]. Fertil Steril, 2023, 119(5): 727–740.\n[28]\n Younes  G,  Tulandi  T.  Effects  of  adenomyosis  on  in  vitro\nfertilization  treatment  outcomes:  a  meta-analysis[J]. Fertil\nSteril, 2017, 108(3): 483–490. e3.\n[29]\n Valore EV, Park CH, Quayle AJ, et al. Human beta-defensin-\n1:  an  antimicrobial  peptide  of  urogenital  tissues[J]. J  Clin\nInvest, 1998, 101(8): 1633–1642.\n[30]\n Vodstrcil  LA,  Hocking  JS,  Law  M,  et  al. Hormonal\ncontraception  is  associated  with  a  reduced  risk  of  bacterial\nvaginosis:  a  systematic  review  and  meta-analysis[J]. PLoS\nOne, 2013, 8(9): e73055.\n[31]\nPregnancy and Microbiota in Letrozole vs GnRH-a Protocols 13\nUnproofed","source_license":"CC0","license_restricted":false}