Efficacy of the combination of antimicrobial therapy with immunomodulators in the management of immunological female infertility due to anti-sperm antibodies associated with bacterial vaginosis during unexplained infertility: double-blind factorial randomised controlled trial

Abstract

Background

Unexplained infertility (UI) is a subject of major concern in reproduction and can be associated with bacterial vaginosis (BV) and/or antisperm antibodies (ASA), yet therapeutic strategies for these conditions in resource-limited settings are poorly defined. This study aimed to evaluate the efficacy of immunomodulators compared with placebo following antimicrobial therapy among infertile women with UI linked to ASA and BV.

Methods

This double-blind, 2 × 3 factorial randomised controlled trial was conducted at a tertiary referral hospital in Kisangani, Democratic Republic of the Congo. 123 women with UI, positive ASA, and BV were randomised to receive either metronidazole alone or combined antimicrobial therapy (metronidazole, clotrimazole, and clindamycin) for Factor 1, followed by prednisolone, zinc acetate, or placebo (paracetamol 100 mg) for Factor 2. The primary outcome was time to clinical pregnancy within six months, analysed in the modified intention-to-treat population as a time-to-event endpoint using Kaplan–Meier methods, the log-rank test, and Cox proportional hazards models. The study protocol was initially approved by the provincial health district ethics committee (701/FBL/DPS/TSHOPO/SEC/0173/2023). The trial has been retrospectively registered with the Pan African Clinical Trials Registry on 08 October 2024 under the identifier PACTR202410672612184.

Results

Among 123 participants, immunomodulators significantly increased clinical pregnancy rates compared with placebo (38.55% vs. 12.5%; aHR = 3.43 [1.34–8.81]; p = 0.010). Adjusted analyses showed significantly higher pregnancy rates for both zinc acetate (40.48%; aHR 3.46 [1.27–9.40]; p = 0.012) and prednisolone (36.59%; aHR 2.75 [1.01–7.68]; p = 0.027) compared with placebo, with no significant difference between the two active agents (p = 0.623). Combined antimicrobial therapy was superior to metronidazole alone in achieving therapeutic BV cure at visit 2 (58.06% vs. 37.7%; RR 1.52 [1.04–2.22]; p = 0.023) and, at the margins of the factorial design, was associated with a higher cumulative conception rate (aHR 2.20 [1.09–4.43]; p = 0.027). The Factor 1 × Factor 2 interaction was not statistically significant (likelihood-ratio p = 0.59). While the overall incidence of adverse events was comparable between zinc (16.67%) and prednisolone (21.95%) (p = 0.541), zinc acetate was associated with significantly higher patient-reported tolerance (p = 0.007).

Conclusion

Sequential treatment involving combined antimicrobial therapy followed by immunomodulation significantly improves clinical pregnancy rates in women with UI associated with BV and ASA. Zinc acetate offers an effective and better-tolerated alternative to prednisolone for this indication.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12865-026-00875-z.

Introduction

Infertility is a reproductive system disorder characterised by the inability to achieve a clinical pregnancy after at least twelve months of regular, unprotected sexual intercourse. It is listed among the most common chronic health conditions, regardless of age [1, 2]. The capacity to make an accurate diagnosis of infertility is contingent on the extent and quality of the examinations conducted on the couple. The term unexplained infertility (UI) is a diagnosis of exclusion of known pathologies. It is assigned to couples who are unable to conceive despite regular unprotected intercourse and who do not meet the diagnostic criteria for male infertility, oligo/anovulatory infertility, or anatomical problems such as fallopian tube obstruction, endometriosis, uterine cavity abnormalities, or cervical or vaginal obstruction. This diagnosis is made in nearly one-third of heterosexual couples who consult tertiary fertility centres [3–5].

UI can be primary or secondary, with underlying factors that vary by type [3]. The pathophysiology of UI is heterogeneous. The prevailing models are multifactorial, in which subtle abnormalities, whether isolated or concurrent, reduce the probability of conception per cycle without yielding a straightforward diagnostic phenotype [6, 7]. Several studies have identified risk factors associated with UI [8–10]. Two paths emerge in understanding unexplained infertility: the role of the genital microbiota, particularly bacterial vaginosis (BV), and immunological mechanisms, including antisperm antibodies (ASA). Meta-analyses and reviews have shown an association between BV, vaginal dysbiosis, and reproductive failures, especially in In Vitro Fertilisation (IVF) [11–16]. In fact, infections of the female genital tract, such as BV, have been shown to create an environment conducive to isoimmunization. The bacteria linked to this dysbiosis are known to produce enzymes like sialidases, which can break down the epithelial glycocalyx and the protective mucins of the vaginal mucosa, breaking down the physical and immunological barriers, thereby allowing direct exposure of sperm antigens to the host’s antigen-presenting cells in an inflammatory context [17, 18]. A breakdown of natural sperm tolerance in women can cause sensitisation, leading to sperm rejection or failure of implantation [19–22]. Molecular mimicry is another key theoretical concept that can help to explain how common bacterial infections can trigger an immune response against sperm. When a woman triggers an immune response against a vaginosis pathogen, the antibodies she produces may mistakenly target sperm due to their structural similarity [18, 23]. The antibodies produced following this immunisation against sperm act at several critical levels of the reproductive cascade [19, 21, 22].

In managing unexplained infertility, while general therapeutic strategies rather than targeted ones should be preferred and encouraged [3], an increasing number of experts believe that treatment of infertile couples with UI requires personalised approaches. Several key variables, including age, infertility history, treatment history, and costs and risks, must be considered when choosing an appropriate treatment plan [24, 25]. In the presence of BV and ASA, it is therefore necessary to treat these conditions to improve conception chances in couples with UI.

While the management of BV is standardised [26, 27], the treatment of ASAs remains controversial [20]. Treatments for BV typically involve metronidazole or clindamycin, with recurrence rates of 20 to 50% within three months of treatment [26–28]. Before the advent of assisted reproductive technologies for immunological infertility, various initiatives were implemented to treat immunological infertility through different methods, including washing or proteolysis of sperm antibodies, immunosuppressive therapy, and other strategies based on ejaculation frequency. Regarding immunosuppressive treatment, there is considerable debate about the choice of agents, such as corticosteroids or other immunomodulators, particularly for conception, ASAs clearance, or the optimal strategy to reduce corticosteroid side effects [20, 29–31]. Clinical evidence on ASA efficacy remains heterogeneous, of low level, and largely unstandardised [31–34]. The advent of Intracytoplasmic Sperm Injection (ICSI), which bypasses almost all effects of ASAs, has rendered these previous therapies redundant, except for intrauterine insemination (IUI) [33, 35]. However, ICSI and even IUI pose issues of availability in many developing countries and accessibility for populations where most patients are generally poor [36].

In light of the high prevalence of UI (27.57%), ASA (43.4%) and BV (53.54%), as well as the significant correlation of ASA (φ coefficient = 0.50) and BV (φ coefficient = 0.38) with UI in Kisangani [10, 37], a city located in Africa’s infertility belt [38] and where access to medically assisted reproduction techniques is currently non-existent, there is an urgent need for accessible therapeutic strategies. The primary objective of this study was to assess the efficacy of immunomodulators in comparison with placebo following antimicrobial therapy among infertile women with UI.

Methods

Study design and setting

Design overview

This double-blind, 2 × 3 factorial randomised controlled trial was conducted to evaluate two independent therapeutic factors in women with UI linked to the simultaneous presence of ASA and BV. The first factor (Factor 1) involved antimicrobial treatment for BV, with two levels. The second factor (Factor 2) was an immunomodulatory strategy with three levels. Combining these factors orthogonally resulted in six parallel treatment groups, forming the full factorial matrix (see Table 1). In line with the Consolidated Standards of Reporting Trials (CONSORT) extension for factorial randomised trials [39, 40], the trial is reported as a full factorial design in which every combination of factor levels is represented, and each participant receives exactly one combination. The study protocol was approved by the provincial health district ethics committee, named Comité provincial d’Ethique de la Division provinciale de la Santé (701/FBL/DPS/TSHOPO/SEC/0173/2023). The trial has been retrospectively registered with the Pan African Clinical Trials Registry on 08 October 2024 under the identifier PACTR202410672612184.

Table 1.

Six treatment groups generated by the 2 × 3 factorial design

	Prednisolone	Zinc Acetate	Placebo	TOTAL
Metronidazole ovule	20	21	20	61
Combined ovule	21	21	20	62
TOTAL	41	42	40	123

Trial setting and period

The trial was conducted in the department of obstetrics and gynaecology at the Hôpital du Cinquantenaire de Kisangani (HCKis), a tertiary provincial referral hospital in Kisangani, the Democratic Republic of the Congo. The recruitment, randomisation, biological sampling, ancillary investigations and clinical follow-up were conducted between 2 August 2023 and 31 July 2024. HCKis was selected because it served as the referral hospital for women seeking medical assistance for infertility during a free gynaecological consultation campaign organised from 2 August to 2 October 2023 within 7 health facilities in Kisangani.

Rationale for the factorial design and interaction hypothesis

A factorial design was adopted for two complementary reasons: first, a single trial allowed two clinically and biologically independent interventions to be evaluated concurrently in the same population, thereby maximising the information obtained per enrolled participant compared with two separate trials. Second, a biologically plausible interaction was pre-specified as a mechanistic hypothesis: restoration of a eubiotic vaginal microbiota may potentiate the effect of systemic immunomodulation by reducing chronic antigenic stimulation, which is thought to sustain ASA production. No pharmacological antagonism was anticipated between the antimicrobial and immunomodulatory interventions. Factor 1 comprised two intravaginal antimicrobial regimens (7 consecutive days followed by one ovule every 3 days for 1 month), while Factor 2 consisted of three 12-day oral regimens administered during the peri-ovulatory window, following an intervening menstruation after 7-day antimicrobial treatment.

Timing of allocation to the two factors

Participants were randomised to the two factors at different times. The randomisation to factor 1 occurred at the enrolment visit (day 0), immediately after eligibility was confirmed and informed consent was obtained. Randomisation to factor 2 occurred on day 1 of the menstrual cycle, after the 7-day antimicrobial course was completed, and confirmation of no medical contraindications to the immunomodulator. The flow of participants between these two randomisation points is illustrated in Fig. 1.

Fig. 1.

Randomisation procedure

Study participants

Eligibility criteria

The eligibility criteria were identical for both factors of the factorial design.

Inclusion criteria

Participants were eligible if they met all of the following inclusion criteria:

• age between 18 and 40 years inclusive;

• UI according to international guidance [6], defined by documented regular menstruation, normal ovarian reserve test (10–15 antral follicle counts in both ovaries during the first four days of the cycle), bilateral patent fallopian tubes and normal uterine cavity confirmed by hysterosalpingography, absence of significant uterine pathology, a normal hormonal assessment and a patterner’s semen analysis within World Health Organization 2021 reference values. Hormones assessed included antimüllerian hormone (AMH), prolactin, progesterone, and thyroid hormones (T3 and T4) [10, 37];

• ASA positivity confirmed by a two-step diagnostic strategy combining a rapid serum ASA IgA/IgG test (HIGHTOP Biotech, China) and, when positive, an indirect mixed antiglobulin reaction (MAR) test (SpermMAr^®; FertiPro N.V., Beernem, Belgium) using cervical mucus. Their partners also underwent serological ASA evaluation with normal results;

• BV confirmed by the concurrent presence of at least three Amsel criteria and a Nugent score ≥ 7 [10, 37];

• Written informed consent from the woman prior to any trial-specific procedure, with the partner’s agreement to provide a semen sample for the MAR test.

Exclusion criteria

Women were excluded from the study if they had any of the following conditions:

• an active sexually transmitted infection (gonorrhoea, chlamydia, syphilis, or trichomoniasis);

• concomitant vulvovaginal candidiasis or a non-BV vaginitis;

• a serious systemic disease likely to interfere with fertility or with trial conduct (uncontrolled diabetes, advanced hepatic disease, active systemic lupus erythematosus, severe renal impairment);

• antibiotic use within the 14 days preceding enrolment;

• current pregnancy or lactation;

• known hypersensitivity to any trial drug (metronidazole, clotrimazole, clindamycin, prednisolone, zinc, or paracetamol);

• use of any immunomodulatory medication in the preceding three months;

• major gynaecological disease reducing the likelihood of conception (endometriosis, cavity-distorting fibroids, bilateral tubal obstruction, or polycystic ovary syndrome (PCOS). The updated Rotterdam criteria served as the standard for diagnosing PCOS [41];

• a major psychiatric disorder impairing the capacity to consent or to adhere to the protocol;

• a refusal of the partner to participate in procedures requiring a semen sample.

Recruitment

Participants were recruited from women consulting at HCKis for infertility, including those referred through the free gynaecological consultation campaign. Baseline diagnostic investigations required to confirm UI were performed according to the department’s standard diagnostic workup prior to enrolment.

Sample size

The sample size calculation is reported separately for each main comparison, with explicit indication of whether an interaction was assumed.

Main comparison (Factor 2-immunomodulation)

The trial was a priori powered to assess the superiority of each active immunomodulatory arm (prednisolone or zinc) over active placebo in the cumulative clinical pregnancy rate at 6 months. The following standard formula was used to calculate the number of participants required per group [42]:

Where p0 represents the expected conception rate in the placebo group, p1 the expected conception rate in the active group, p̄ the average of the two proportions defined by (p0 + p1) / 2, z(1 − α/2) the critical value corresponding to the two-sided alpha risk, and z(1 − β) the critical value corresponding to the desired statistical power.

The expected conception rate at 6 months in the placebo group (p0) was set at 0.10, which corresponds to a population with severe infertility and a low probability of spontaneous conception [43]. The expected conception rate in the active treatment groups of prednisolone and zinc (p1) was estimated at 0.40. The minimally clinically relevant absolute difference to detect was therefore 0.30. The alpha risk was set at 0.05 for a two-sided test, and the desired statistical power was 80%. With p0 = 0.10 and p1 = 0.40, the average proportion p̄ was 0.25. The critical values used were 1.96 for an alpha risk of 0.05 and 0.84 for a power of 80%.

Applying this formula yielded a theoretical sample size of approximately 31.3 participants per study arm. With an estimated 10% follow-up, the minimum per-study-arm sample size is 34.43 participants, resulting in a total of 103 cases across the three arms. A total of 123 participants were ultimately enrolled (approximately 41 per Factor 2 arm and 60–63 per Factor 1 arm), providing approximately 20–21 participants per cell of the 2 × 3 factorial matrix. (see Fig. 1).

Main effect of factor 1 (BV treatment)

As transparently recommended [40], the sample-size calculation was not independently powered for the main effect of Factor 1. The comparison of the two Factor 1 levels nevertheless benefited from the full trial sample (approximately 60–63 participants per level), allowing the marginal main effect of Factor 1 on the primary outcome and on BV cure endpoints to be estimated with accompanying 95% confidence intervals.

Factor 1 x factor 2 interaction

No quantitative interaction effect was assumed in the primary sample-size calculation. Consistent with the CONSORT factorial extension [40], the trial was powered to detect main effects under a no qualitative interaction working hypothesis (factorial at-the-margins approach). The interaction analysis was therefore interpreted as hypothesis-generating, with pre-specified effect sizes and precision reported rather than relying on the interaction p-value [40].

Interventions and factorial structure

Patients who were potentially eligible signed informed consent forms before any study-specific procedures. These included a gynecologic examination to measure pH and collect samples for the Amsel criteria [44], the Nugent score [45], and the indirect mixed antiglobulin reaction test (indirect MAR test) [46], a pregnancy test, and a pelvic ultrasound. To ensure drug quality, a randomly selected batch sample was tested at the laboratory for analysis and control of medicines and foodstuffs of the Faculty of Pharmaceutical Sciences at the University of Kinshasa (see Appendix 1). The investigative team comprised eight individuals, including a principal investigator, an Information Technology (IT) specialist, a pharmacist, three laboratory technicians and two nurses. The principal investigator was responsible for supervising activities, enrolling participants, and administering follow-up. The IT specialist was responsible for the randomisation process. The pharmacist was responsible for assembling the treatment kits. Laboratory technicians handled the specific tests required for the study. Nurses were responsible for administering the treatment kit and monitoring participants by sending appointment reminders. They were also responsible for advising and educating participants on the importance of adhering to treatment, keeping appointments, the consequences of high-risk sexual behaviour, and personal hygiene measures.

Factor 1- BV treatment with 2 levels

Factor 1 comprised two intravaginal antimicrobial regimens, each administered once daily at bedtime for 7 consecutive days, followed by the administration of one intravaginal ovule every three days for a duration of one month and was allocated in a 1:1 ratio.

The level 1 consisted of Ovogyl^® ovule (New Cesamex SARL, Kinshasa, DRC) containing metronidazole 500 mg, and the level 2 consisted of Vagiklin forte^® ovule (Biomatrix Healthcare PVT LMT, Gujanat, India), a combined intravaginal ovule containing metronidazole 100 mg, clotrimazole 100 mg, and clindamycin phosphate. The combined ovule was selected to broaden coverage of BV-associated dysbiosis while reducing the risk of secondary vulvovaginal candidiasis.

Factor 2 - immunomodulatory strategy with 3 levels

Factor 2 comprised 3 oral 12-day regimens administered during the peri-ovulatory window and allocated in a 1:1:1 ratio.

Level 1 consisted of prednisolone 5 mg tablets (Caisa Pharma International, Kinshasa, DRC), with 40 mg administered in the morning and 40 mg in the evening from days 1 to 10 of the cycle, followed by a single 5 mg dose in the morning on days 11 and 12. Level 2 consisted of Cezinc 50^® tablets (Aura Lifecare Private LMT, Gujarat, India), a 50 mg zinc acetate tablet taken twice daily for 12 days, 30 min before meals. Level 3 consisted of paracetamol 100 mg tablets (IDA Foundation, Amsterdam, Netherlands), administered twice daily for 12 days, and served as an active placebo. This choice was made because this low dose of paracetamol (acetaminophen) avoids interference with the immunological endpoints. The 100 mg unit dose is 5 to 10 fold below the minimum analgesic/antipyretic threshold in adults (500–1000 mg) [47] and is therefore devoid of clinically relevant pharmacological activity. Paracetamol’s immunomodulatory effects on humoral and T-cell-dependent antibody responses are strictly dose-dependent and have been documented only at full therapeutic (≥ 325 mg per dose, up to 4 g/day) or supratherapeutic doses [48, 49]. A randomised controlled trial in 474 vaccinated adults found no difference in specific antibody titres between 162.5 mg, 325 mg, and an inert placebo [49]. Mechanistically, paracetamol acts predominantly through central COX-2/COX-3 inhibition and endocannabinoid modulation via its AM404 metabolite, with negligible peripheral effect at sub-therapeutic doses [50], and does not modulate the B-cell activation, isotype switching, or antigen presentation pathways underlying ASA production [51]. This pharmacological profile is well characterised. Paracetamol is recognised as a predominantly centrally acting analgesic and antipyretic agent with only weak anti-inflammatory activity [52], and even at a full 1000 mg dose its cyclooxygenase inhibition is functionally COX-2-selective while sparing the COX-1-dependent peripheral pathways governing platelet and gastric prostaglandin synthesis [53]. At the sub-therapeutic 100 mg unit dose used here, a clinically relevant modulation of the immune mechanisms governing antisperm-antibody production or endometrial receptivity is therefore not expected. Furthermore, were this active placebo to exert any residual anti-inflammatory effect, it would be expected to favour the placebo arm, thereby biasing the immunomodulator-versus-placebo comparison toward the null and rendering the corresponding estimates conservative. The inclusion of two mechanistically distinct active immunomodulators (zinc and prednisolone) that share a common comparator further protects the design from attributing any observed difference to the placebo itself. Unlike non-steroidal anti-inflammatory drugs, paracetamol has not been associated with impaired female fertility at occasional low doses, ensuring an optimal safety profile in women of reproductive age enrolled in a fertility trial [54].

The 6 treatment groups of the 2 × 3 factorial matrix

The six treatment groups arising from the 2 × 3 factorial design are summarised in Table 1.

Follow-up

For the Factor 1-BV treatment, participants were required to attend three follow-up visits after BV treatment: between the eighth and tenth day after the start of medication (visit 1), between the 24th and 30th day after visit 1 (visit 2), and between the 75th and 105th day after visit 1(visit 3, three-month check-up). Vaginal samples were collected and adverse events recorded at visits 1 and 2. At visit 3, only vaginal samples were collected. Additionally, during each visit, participants were advised on intimate hygiene measures, which included the recommendation that the intimate area should be washed with only water and not inside the vagina. Furthermore, it was advised that the partner should wash their penis before any sexual intercourse.

For the Factor-2 Immunomodulatory strategy, six-monthly visits were scheduled at the beginning of each cycle or on the expected date of menstruation, if menstruation was delayed. To ensure the protocol’s rigour, a maximum variance of ± 4 days was acceptable for each monthly visit. At each visit, clinical parameters and adverse events were recorded and, if required, a pregnancy test and a pelvic ultrasound were performed.

Adherence support without unmasking

A structured adherence-support programme was integrated into the protocol without disrupting the double-blind design. Before treatment, each participant received a standardised counselling session from a trained clinician independent of the randomisation process. The session covered UI, BV, immunological infertility, basics of BV and immunomodulatory treatments, medication instructions, treatment duration, and adherence importance. Side effects were described neutrally to prevent unblinding. During follow-up, adherence was monitored via brief, standardised interviews at each visit. Participants were advised against self-medication, traditional remedies, or unprescribed medications. Adherence was objectively assessed throughout using pill counts by an independent, blinded nurse.

Discontinuation criteria

Participants were withdrawn from the trial if they experienced a severe adverse event, developed drug intolerance that precluded continuation of treatment, initiated external fertility treatment, or were lost to follow-up after three unsuccessful attempts to contact over a 30-day period. Data collected up to the time of withdrawal were retained for the intention-to-treat analysis.

Randomisation, allocation concealment and blinding

Sequence generation

Two randomisation sequences were computer-generated by an independent statistician using Random Allocation Software version 2.0 (Saghaei, Department of Anaesthesia, Isfahan University of Medical Sciences, Iran). Permuted block randomisation was used with a 1:1 allocation ratio for Factor 1 and a 1:1:1 allocation ratio for Factor 2. No stratification variables were used. Because the two randomisations occurred at different time points in the same participants, the second randomisation was not stratified by the first.

Allocation concealment

Allocation was concealed using sequentially numbered, opaque, sealed, and tamper-evident envelopes prepared by a third-party pharmacist not involved in participant enrolment, follow-up, or outcome assessment. Each enrolment number corresponded to a neutral treatment code, ensuring full anonymisation of the allocation. A single sequential enrolment number assigned at the day-0 visit indexed both allocations: the two independently generated sequences were pre-mapped by the pharmacist to that same number and prepared as two separately sealed, sequentially numbered envelopes per participant, one for each factor. The Factor 1 envelope was opened at day 0, whereas the Factor 2 envelope, bearing the same enrolment number, remained sealed until day 1 of the subsequent cycle, so that the two allocations were linked solely through this concealed identifier without unblinding either. Envelopes were kept under lock and key and opened strictly in enrolment order. Upon opening, the dispensing staff member issued a pre-conditioned treatment kit identified solely by its code.

Blinding

The trial was carried out using a double-blind design. Treatment kits were identical in appearance, shape and labelling, differing only by their identification code. Participants and the investigation team remained blinded to treatment allocation throughout the trial. The code-to-treatment correspondence was accessible only to the independent statistician responsible for the randomisation. Emergency unblinding was permitted only in cases of absolute medical necessity, following a documented procedure recording the clinical justification, the date, the time, and the identity of the person performing the unblinding.

Outcomes

The primary analysis of fertility endpoints was conducted on the modified intention-to-treat (mITT) population, defined as all participants randomised to Factor 2 who received at least one dose of the allocated immunomodulatory treatment, analysed in their originally allocated group irrespective of subsequent treatment modifications. A per-protocol (PP) analysis was pre-specified as a sensitivity analysis and excluded participants who used prohibited co-medication, had missing data on essential endpoints, or had major non-adherence. Safety analyses were conducted on all participants who received at least one dose of immunomodulator, analysed according to the treatment actually received.

Primary outcome

The primary outcome was the time to clinical pregnancy within six months of the immunomodulatory randomisation, analysed as a time-to-event endpoint; participants who had not conceived were censored at month six. The cumulative incidence of clinical pregnancy at six months, estimated by the Kaplan–Meier method, is reported as a complementary descriptive measure. Clinical pregnancy was defined by a positive serum β-human chorionic gonadotropin (β-hCG) assay followed by transvaginal ultrasound visualisation of an intrauterine gestational sac with embryonic cardiac activity at 6–7 weeks of amenorrhoea, as confirmed by a qualified clinician blinded to treatment allocation. Biochemical pregnancies, ectopic pregnancies, and pregnancies lost before ultrasound confirmation of embryonic cardiac activity were not counted as events for the primary outcome. In the pre-specified primary analysis, the effect of factor-2 on time to clinical pregnancy was estimated using a Cox proportional hazards model fitted to the mITT population, yielding hazard ratios with 95% confidence intervals adjusted for age, duration and type of infertility, BMI and MAR titre to assess the robustness of the findings to the observed baseline distribution of this characteristic. Kaplan-Meier curves and the log-rank test were used to compare arms.

Multiplicity was addressed according to the factorial structure of the trial. Because Factor 1 and Factor 2 addressed orthogonal, pre-specified questions, no adjustment was applied across factors. Within the Factor 2 family of immunomodulator-versus-placebo comparisons, the family-wise error rate was controlled using the Holm procedure, and the Bonferroni and Benjamini–Hochberg (false-discovery-rate) procedures were additionally computed as sensitivity analyses (Table 4). The six cell-specific combination contrasts were regarded as secondary and exploratory and are reported descriptively, without confirmatory inference.

Table 4.

Sensitivity of the pre-specified and exploratory treatment comparisons to multiplicity adjustment

Comparison	Unadjusted p	Holm	Bonferroni	BH (FDR)
Primary family - Factor 2 (immunomodulator vs. placebo)
Zinc vs. placebo	0.012	0.024	0.024	0.024
Prednisolone vs. placebo	0.027	0.027	0.054	0.027
Exploratory family - combination cells (vs. Placebo + Metronidazole)
Zinc + Combined	0.016	0.080	0.080	0.043
Prednisolone + Combined	0.017	0.080	0.085	0.043
Zinc + Metronidazole	0.185	0.555	0.925	0.308
Prednisolone + Metronidazole	0.430	0.860	1.000	0.537
Placebo + Combined	0.623	0.860	1.000	0.623

Secondary outcomes

Time to conception

Time to conception was defined as the interval between immunomodulatory randomisation and the estimated date of conception, calculated from the date of the first ultrasound scan or a positive β-hCG test, whichever occurred first, and was performed before the confirmatory ultrasound scan. Participants who had not conceived by day 180 (six-month follow-up), who were excluded, or who were lost to follow-up were censored from the analysis.

The clinical cure rate of BV

Clinical cure was assessed according to the strict Amsel criteria. The patient was declared in clinical recovery if the signs and symptoms initially observed at inclusion had been fully resolved at the protocol-defined assessment visit [55]. The clinical cure was defined in this study as the simultaneous presence of the following three criteria: absence of abnormal vaginal discharge, a negative Whiff test, and absence of clue cells on direct microscopy. If a single criterion remained positive, the participant was classified as not cured. The standard Amsel criteria define BV as the presence of at least 3 criteria, whereas 2 or fewer negative criteria indicate BV negative [44].

The microbiological cure rate of BV

Microbiological cure was determined by the Nugent score. The microbiological cure was defined as the absence of any pathogens targeted in BV, as determined by the reference methods outlined in the Nugent score [45], at the end of the treatment. The Nugent score uses a 0 to 10 scale, with 0 to 3 indicating normal flora, 4 to 6 indicating intermediate flora, and 7 to 10 indicating BV.

The therapeutic cure rate of BV

The therapeutic cure was observed in participants who demonstrated both clinical and bacteriological cure at the same assessment point. This is a combined criterion used in infectious disease to evaluate the overall effectiveness of a treatment [56].

Safety, including causality assessed with the imputability classification of the World Health Organisation-Uppsala Monitoring Centre (WHO-UMC) scale [57, 58] and the incidence and severity of adverse events graded per the Common Terminology Criteria for Adverse Events (CTCAE) [59].

Data collection

Data were captured on paper Case Report Forms (CRFs) completed at each visit and subsequently entered into a dedicated Microsoft Excel spreadsheet with cross-validation. The principal investigator supervised the entire data-management process. Source documents comprised clinical consultation records, laboratory registers, and the electronic medical record. Each participant was assigned a unique alphanumeric code to preserve confidentiality. Protocol deviations were systematically documented and reviewed for their potential impact on the primary and secondary endpoints.

Statistical analysis

Data collected prospectively using paper case report forms (CRFs) were entered into a dedicated Excel spreadsheet, cross-validated, and analysed using R 4.5.0 and Epi Info™ 7.2.2.6.

Baseline data were summarised using frequencies, proportions, means, and standard deviations.

To compare proportions, Pearson’s chi-square test at a significance level of p < 0.05 was used. If the conditions for applying Pearson’s chi-square test were not met, Fisher’s exact test at a significance level of p < 0.05 was used. The relative risk (RR) and its 95% confidence interval (CI) were calculated to assess the strength of the association between categorical variables. The pre-specified primary analysis of the primary endpoint was a time-to-event analysis of time to conception. The extent of the effect of treatments on the occurrence of conception was estimated by calculating Hazard Ratios (HRs) with their 95% confidence intervals. These ratios were obtained using a Cox proportional hazards regression model, which quantifies the instantaneous risk of conception across the study arms. The regression was then adjusted for confounding factors such as age, body mass index (BMI) and indirect MAR test percentage to obtain adjusted Hazard Ratios (HRa). To address the pre-specified synergy hypothesis, the Factor 1 × Factor 2 interaction was formally tested by adding the corresponding product term to the adjusted Cox model and comparing the models with and without the interaction term using a likelihood-ratio test; in the absence of a significant interaction, main effects were estimated at the margins of the factorial design. The proportional hazards assumption was checked graphically using survival curves. The comparison of survival curves across therapeutic arms was performed using two complementary tests to ensure a comprehensive analysis of treatment effects. The Log-rank test was preferred to evaluate the overall difference between groups over the entire follow-up period, giving equal weight to each observed event. Simultaneously, the Wilcoxon test was used for its greater sensitivity to early events, allowing detection of any divergence in efficacy during the early treatment phase. This dual approach helped to address the limitations of each test, particularly in cases of non-proportional hazards over time [60]. Participants who did not conceive by the sixth-month follow-up, were excluded, or were lost to follow-up were censored. To compare the proportions of adverse effects across different grades and arms, the Wilcoxon-Mann-Whitney U test was used.

Results

Baseline characteristics of participants at randomisation

Baseline characteristics of participants are presented in Table 2. A total of 123 participants were randomised into 6 groups. The comparability of the groups at inclusion was assessed for type of infertility, alcohol consumption, smoking, duration of infertility, BMI, Nugent score and ASA titer. The baseline characteristics were generally balanced, indicating the effectiveness of the block randomisation process.

Table 2.

Baseline profile of participants allocated to different factorial treatment arms

	Prednisolone and Metronidazole N = 20	Zinc and Metronidazole N = 21	Placebo and Metronidazole N = 20	Prednisolone and Combined treatment N = 21	Zinc and Combined treatment N = 21	Placebo and Combined treatment N = 20	Total N = 123
	n(%)	n(%)	n(%)	n(%)	n(%)	n(%)	n(%)
Mean age ± SD	32.8 ± 5.6	34.1 ± 4.9	33.9 ± 5.5	32.2 ± 6.0	33.0 ± 6.3	32.4 ± 6.3	32.9 ± 5.7
Type of infertility
Primary	5(25.0)	8(38.1)	1(5.0)	2(9.5)	6(28.6)	7(35.0)	29(23.6)
Secondary	15(75.0)	13(61.9)	19(95.0)	19(90.5)	15(71.4)	13(65.0)	94(76.4)
Alcohol consumption	12(60.0)	14(66.7)	14(70.0)	10(47.6)	12(57.1)	10(50.0)	72(58.5)
Smoking	2(10.0)	1(4.8)	2(10.0)	3(14.3)	1(4.8)	2(10.0)	11(8.9)
Mean duration of infertility ± SD (years)	4.9 ± 4.7	4.7 ± 5.3	4.9 ± 3.6	3.8 ± 2.9	2.9 ± 2.4	4.5 ± 4.1	4.3 ± 3.9
BMI (Mean ± SD)	26.7 ± 5.0	25.7 ± 5.3	23.9 ± 5.4	26.3 ± 5.4	27.9 ± 4.4	27.3 ± 5.0	26.3 ± 5.1
Nugent score (Mean ± SD)	8.0 ± 0.5	8.0 ± 0.5	7.9 ± 0.6	8.1 ± 0.7	7.8 ± 0.6	8.0 ± 0.6	8.0 ± 0.6
Inderect MAR test % (Mean ± SD)	56.4 ± 12.7	48.9 ± 9.4	51.3 ± 9.9	51.3 ± 9.9	50.4 ± 12.9	48.7 ± 7.5	51.1 ± 11.0

Outcomes

Conception rate

The primary endpoint of the study was the cumulative clinical conception rate at 6 months (Table 3).

Table 3.

Comparison of conception rate in different arms treatment using the Cox proportional hazards model adjusted by age, BMI and indirect MAR test titre in the modified Intent-to-Treat population

	Conception n(%)	No conception n(%)	p-value	HR [95% CI]	HRa [95% CI]
Arms comparison
Immunomodulators vs. placebo
Immunomodulators	32(38.55)	51(61.45)	0.010	3,41[1.33-8,77]	3.43[1.34-8,81]
Zinc	17(40.48)	25(59.52)	0.012	3.58[1.32–9.73]	3.46[1.27–9.40]
Prednisolone	15(36.59)	26(63.41)	0.027	3.13[1.13–8.61]	2.75[1.01–7.68]
Placebo	5(12.50)	35(87.50)		1
Comparison of Immunomodulators
Zinc	17(40.48)	25(59.52)	0.623	1.19[0.59–2.38]
Prednisolone	15(36.59)	26(63.41)		1
6 factorial Groups
Zinc_combined treatment	11(52.38)	10(47.62)	0.016	6.40[1.41–29.01]	7.33[1.48–36.29]
Zinc_Metronidazole	6.26(28.57)	15(71.43)	0.185	2.94[0.59–14.59]
Prednisolone_combined treatment	11(52.38)	10(47.32)	0.017	6.26[1.38–28.30]	5.57[1.15-29.0]
Prednisolone-Metronidazole	4(20.00)	16(80.00)	0.430	1.97[0.36–10.78]
Placebo_Combined treatment	3(15.00)	17(85.00)	0.623	1.57[0.26–9.37]
Placebo-Metronidazole	2(10.00)	18(90.00)	1
BV treatment
Combined antimicrobial vs. metronidazole	25(40.33)	37(59.67	0.027	2.32 [1.17–4.62]	2.20 [1.09–4.43]
Factor 1 × Factor 2 interaction§	NA	NA	0.55	1.60 [0.23–11.14]	1.84 [0.26–12.93]

NA not applicable

§likelihood-ratio χ² = 1.05, 2 df, p = 0.59 for the full 2 × 3 design

The intention-to-treat analysis demonstrated a statistically significant superiority of immunomodulators over placebo. Among the total population receiving an immunomodulator, 32/83 women conceived (38.55%), compared with only 5/40 in the placebo group (12.50%), with a p-value of 0.010. The HR for immunomodulators as a group was 3.41 (95% CI [1.33–8.77]). After adjusting for confounding variables, the aHR was 3.43 (95% CI [1.34–8.81]).

Detailed analysis by the active treatment arm revealed similar performance for zinc and prednisolone. The conception rate in the zinc group was 40.48% (17/42, p = 0.012 vs. placebo) with an aHR of 3.46 (95% CI [1.27–9.40]). In the prednisolone group, the rate was 36.59% (15/41, p = 0.027 vs. placebo) with an aHR of 2.75 (95% CI [1.01–7.68]). A direct comparison between zinc and prednisolone showed no significant difference in their effects (HR 1.19; 95% CI [0.59–2.38]; p = 0.623).

The analysis of the conception rate across factorial groups used the Placebo-Metronidazole arm as the reference. The analysis revealed that participants in the Zinc-combined treatment arm were significantly more likely to achieve conception (HR = 6.40 [1.41–29.01], p = 0.016), an effect that remained robust after adjustment for baseline covariates (aHR = 7.33 [1.48–36.29]). Similarly, the Prednisolone-combined treatment arm showed a significant increase in conception rates compared to the reference group (HR = 6.26 [1.38–28.30], p = 0.017; aHR = 5.57 [1.15-29.0]). No statistically significant differences were observed for the remaining groups, including Zinc-Metronidazole (HR = 2.94 [0.59–14.59], p = 0.185), Prednisolone-Metronidazole (HR = 1.97 [0.36–10.78], p = 0.43), and Placebo-combined treatment (HR = 0.63 [0.11–3.82], p = 0.623). The estimates of crude and adjusted Hazard Ratios were consistent across all treatment arms in which adjustment was performed.

The pre-specified Factor 1 × Factor 2 interaction was formally tested and was not statistically significant in the adjusted Cox model (likelihood-ratio χ² = 1.05, 2 df, p = 0.59; interaction HR 1.84 [0.26–12.93]). In the absence of a detectable interaction, the marginal main effects were estimated at the margins of the factorial design and showed that combined antimicrobial therapy versus metronidazole alone was associated with a higher cumulative conception rate (aHR 2.20 [1.09–4.43]; p = 0.027).

The sensitivity of these comparisons to multiplicity adjustment is detailed in Table 4. The pre-specified primary contrast of any immunomodulator versus placebo was a single comparison (p = 0.010) and therefore required no correction. Within the Factor 2 family, the superiority of zinc over placebo was insensitive to the procedure applied, remaining significant under the Holm, Bonferroni and Benjamini–Hochberg corrections (adjusted p = 0.024 in each case), whereas the superiority of prednisolone was preserved under the Holm and Benjamini–Hochberg procedures (adjusted p = 0.027) but fell at the boundary of significance under the more conservative Bonferroni correction (p = 0.054). In the exploratory six-cell family, the two combination cells that were nominally significant before adjustment (zinc + combined and prednisolone + combined) no longer retained family-wise significance (Holm-adjusted p = 0.08; Benjamini-Hochberg p = 0.043), and the remaining cell-specific contrasts were non-significant under every procedure.

The Kaplan-Meier survival curves revealed distinct patterns of event distribution across the six factorial arms, with an early divergence in the experimental groups compared with the control arms. Statistical evaluation of these curves confirmed significant differences in conception probability over time. The Log-Rank test demonstrated an overall significant difference between the treatment groups across the entire follow-up period [statistic = 15.78; p = 0.007]. The early occurrence of events was particularly evident in the Zinc-based arms, specifically within the Zinc-Combined treatment and Zinc-Metronidazole groups, which showed a more rapid decline in infertility probability during the initial phases of the study. This observation was statistically supported by the Wilcoxon test, which, by assigning greater weight to early events, corroborated the findings of the Log-Rank analysis [statistic = 15.44; p = 0.008] (Fig. 2).

Fig. 2.

Comparison of the survival curve of the immunomodulatory treatment with that of the placebo

Test	Statistic	Degree of freedom	P-Value
Log-Rank Statistic	15.78	4	0.007
Wilcoxon	15.44	4	0.008

BV cure rate

The efficacy of the treatment for BV (Phase 1) was assessed using several cure indicators, including clinical, biological, and therapeutic (Table 5).

Table 5.

Comparison of BV clinical, biological, and therapeutic cure rate in different arms of treatment in the modified Intent-to-Treat and per protocol population

	N° of cured patients/N° with non-missing data (%)		p-value	RR [95% CI]
	Combined treatment	Metronidazole
Clinical cure rate
Visit 1	52/62(83.87)	51/61 (83.61)	0.968	1.01[0.62–1.63]
Visit 2	44/62(70.97)	31/61(50.82)	0.022	1.51[1.06–2.14]
Biological cure rate
Visit 1	39/62(62.90)	25/61 (40.98)	0.014	1.56[1.08–2.26]
Visit 2	39/62(62.90)	25/61 (40.98)	0.014	1.56[1.08–2.26]
Therapeutic cure rate
Visit 1	38/62(61.29)	25/61(40.32)	0.024	1.51[1.05–2.18]
Visit 2	36/62(58.06)	23/61 (37.70)	0.023	1.52[1.04–2.22]
Visit 3 clinical cure rate
Intention-to-treat	31/62(50.00)	16/61(26.23)	0.006	1.73[1.12–2.69]
Per protocol	31/59(52.54)	16/52(30.77)	0.021	1.65[1.05–2.59]
Standard Amsel criteria cure rate (Intention-to-treat)
Visit 1	57/62(91,94)	57/61(93.44)	0,748	0.88[0.41–1.88]
Visit 2	56/62(90.32)	47/61(77.05)	0.046	1.53[1.07–2.19]
Visit 3	47/59(79.66)	32/52(61.54)	0,035	1.54[1.05–2.25]

At Visit 1, clinical cure rates were high and comparable between the two arms: 83.87% for the combination treatment versus 83.61% for metronidazole alone (p = 0.968). A significant biological superiority was observed in the combination treatment arm, with a biological cure rate of 62.9% versus 40.98% with metronidazole, and an RR of 1.59 (95% CI [1.08–2.26]; p = 0.014). The therapeutic cure rate was also significantly higher with combination therapy (61.29% vs. 40.32%; p = 0.024), with an RR of 1.51 [1.05–2.18].

At Visit 2, the clinical cure rate was significantly higher in the combined treatment arm (70.97%) than in the metronidazole arm (50.82%), with an RR of 1.51 (95% CI [1.06–2.14]; p = 0.022). The biological cure rate was higher in the combination group (63.93% vs. 40.32%, p = 0.009), with an RR of 1.56 [1.08–2.26]. Therapeutic cure was significantly higher in the combination treatment group (58.06% vs. 37.7%, p = 0.023) with an RR of 1.52 [1.04–2.22].

At Visit 3, clinical cure showed a major difference in favour of combination therapy. In the mITT analysis, the cure rate was 50% for the combination treatment versus 26.33% for metronidazole (RR = 1.73 [1.12–2.69]; p = 0.002). The per-protocol analysis confirmed these results, with a rate of 52.54% versus 30.77% (p = 0.021), yielding an RR of 1.65 [1.05–2.59].

The analysis of clinical care using the standard Amsel criteria showed no difference between the 2 arms at visit 1 (91.94% vs. 93.44%; p = 0.748). Combination therapy became significantly more efficacious at visits 2 (90.32% vs. 77.05%; p = 0.046) and 3 (79.66% vs. 61.54%; p = 0.035), with respective RRs of 1.53 [1.07–2.19] and 1.54 [1.05–2.25].

Safety and tolerance of treatments

The participants’ self-reports of treatment tolerance showed a statistically significant difference between the prednisolone and zinc arms (p-value = 0.007, Wilcoxon–Mann–Whitney U test) (Fig. 3). The group receiving zinc had a significantly higher tolerance profile, with “very good tolerance” reported by 59.52% of participants, compared with 34.15% in the prednisolone group. Conversely, the lowest tolerance levels were more pronounced in the prednisolone arm, where “moderate tolerance” was reported by 26.27% of subjects (compared to only 4.76% for zinc) and “poor tolerance” was reported by 9.76% (compared to 7.14% for zinc).

Fig. 3.

Comparison of tolerance to immunomodulators, as reported by the participants

The analysis of amputability of adverse effects across the various treatment groups targeting ASAs reveals no significant overall difference in the incidence of AEs between the zinc and prednisolone arms (p = 0.541) (Table 6). The overall incidence of AEs was 21.95% (9/41) in the prednisolone group and 16.67% (7/42) in the zinc group. Moderate adverse events occurred in 14.63% (6/41) of the prednisolone group compared with 7.14% (3/42) in the zinc group (p = 0.229). Simple adverse events were observed in 7.32% (3/41) and 9.52% (4/42) of participants in the prednisolone and zinc groups, respectively (p = 0.513).

Table 6.

Amputability of adverse effects observed across the various treatment groups targeting ASAs and BV treatments

Immunomodulators	AEs present	No AEs	P value	RR [95% CI]
	n(%)	n(%)
Prednisolone	9(21.95)	32(78.05)		-
Zinc	7(16.67)	35(83.33)	0.541	1.31[0.54–3.20]
Moderate adverse effects
Prednisolone	6(14.63)	35(85.37)
Zinc	3(7.14)	39(92.86)	0.229	0.91[0.78–1.07]
Simple adverse effects
Prednisolone	3(7.32)	38(92.68)
Zinc	4(9.52)	38(90.48)	0.513	1.02[0.89–1.16]
BV treatment
Metronidazole	13(21.31)	48(78.69)		-
Combination treatment	9(14.52)	53(85.48)	0.32	1.08 [0.92–1.28]
Moderate adverse effects
Metronidazole	12(19.67)	49(80.33)		-
Combination treatment	8(12.90)	54(87.10)	0.31	1.08[0.92–1.26]
Simple adverse effects
Metronidazole	10(16.36)	51(83.61)		-
Traitement combiné	12(19.35)	50(80.65)	0.66	0.84[0.39–1.81]

The participants’ self-reports of tolerance to BV treatments showed no statistically significant difference between the groups (p-value = 0.560, Wilcoxon–Mann–Whitney U test) (Fig. 4).

Fig. 4.

Comparison of tolerance to treatments for BV, as reported by the participants

The analysis of the amputability of the adverse effects of medications across the various treatment groups targeting BV revealed no statistically significant difference in overall adverse event incidence between the metronidazole and combination therapy groups (p = 0.32). Adverse events were reported in 21.31% (13/61) of participants treated with metronidazole and 14.52% (9/62) in the combination therapy group. Moderate adverse events occurred in 19.67% (12/61) and 12.9% (8/62) of participants, respectively (p = 0.31). Simple adverse events were observed in 16.36% (10/61) of the metronidazole group and 19.35% (12/62) of the combination therapy group (p = 0.66) (Table 6).

Discussion

Outcomes

The analysis of the results of this double-blind, randomised, factorial clinical trial offers key insights into the management of immunological infertility caused by anti-sperm antibodies, an area in which clinical certainty is scarce, and protocols often remain empirical [20]. Historically, treatment of ASAs in women relied on condom use (to reduce the antigenic load) or intrauterine insemination (IUI) to bypass antibodies in cervical mucus [20, 34]. However, IUI does not address ASAs present in follicular fluid or the oviduct, which may impair gamete fusion [61–63]. The immunomodulatory approach evaluated here proposes a more comprehensive solution that may act across the entire genital tract.

The results of this study demonstrated that targeted immunomodulatory therapy significantly increases clinical pregnancy rates in women with UI associated with anti-sperm antibodies and BV. The finding that 38.55% of women receiving either zinc or prednisolone achieved clinical pregnancy, compared with only 12.50% in the placebo group (aHR 3.43), underscores the critical role of immune factors in reproductive failure. The markedly higher conception rate in the cells combining an immunomodulator with combined antimicrobial therapy (aHR 7.33 for zinc-combined and 5.57 for prednisolone-combined) is consistent with two independent and approximately multiplicative main effects rather than with a statistically demonstrated synergy: the formal Factor 1 × Factor 2 interaction was non-significant (likelihood-ratio p = 0.59), and the trial was powered to detect main effects at the margins, not interaction. While this cell-level pattern is biologically compatible with potentiation of immunomodulation by a restored vaginal eubiosis, it remains a hypothesis to be confirmed in an adequately powered trial.

This clinical effect is of a magnitude rarely observed in trials of unexplained infertility, where conventional interventions (such as administering aspirin or corticosteroids to a non-targeted population) generally yield only marginal gains and provide no evidence of improved clinical outcomes, even with medically assisted reproduction techniques [20, 32, 33, 64]. The success of prednisolone in this cohort (36.59% conception rate) aligns with historical evidence suggesting that corticosteroids can enhance fertility by reducing levels of circulating immobilising and agglutinating antibodies [65]. While some studies in male immunological infertility have questioned the efficacy of low-dose steroids [66, 67], our results support the effectiveness of cyclic high-dose protocols, which have previously achieved cumulative conception rates of approximately 30% [68]. The use of prednisolone in our trial echoes the work of Räsänen et al. [29], who tested this agent in men with ASAs. Their study showed that a 20 mg/day dose led to a significant reduction in antibodies in some participants but was ineffective in those with the highest antibody loads. Our trial suggests that, in women, the effect of higher-dose prednisolone may be more complex than a simple reduction in antibody titres, including an improvement in endometrial receptivity through modulation of uterine natural killer (NK) cells, inhibition of pro-inflammatory gene transcription (IL-1, IL-6, TNF-alpha) and promotion of anti-inflammatory genes (IL-10) [20, 56, 69–71]. The 2012 Cochrane meta-analysis by Boomsma et al. [58] on peri-implantation glucocorticoid use found no clear benefit in the general population but highlighted the need for data on subgroups with autoantibodies. Our findings provide this missing evidence, supporting steroid use when a biological immune conflict, such as the presence of ASAs, is suspected. Prednisolone may create a calmer environment, aiding fertilisation and early implantation, explaining the high conception rate.

A major finding of this study is the comparable efficacy of zinc acetate (40.48% conception rate) to prednisolone. Our findings indicate that zinc could be a more effective and better-tolerated alternative to traditional corticosteroids for regulating immune responses in infertile women. The improved tolerance profile of zinc, with 59.52% of participants reporting very good tolerance compared to 34.15% in the prednisolone group, further reinforces its potential clinical usefulness. The near equivalence in conception rates suggests that the subtle mechanisms regulating innate and homeostatic immunity activated by zinc are, in this specific context, as powerful as the genomic immunosuppression induced by glucocorticoids [69, 72]. While the role of zinc in male fertility is well established (improving sperm concentration and motility) [73, 74], its use as an immunomodulator in female infertility is less well documented. Omu et al. [75] reported that zinc supplementation in men with asthenozoospermia reduced ASA titres and normalised cytokine levels (decreased TNF-α and increased IL-4). The results of this study suggest that these properties are relevant to the female reproductive system, supporting the efficacy of zinc in conception. Zinc functions as a homeostatic regulator of immune activity, acting at several key points. It inhibits phosphodiesterases (PDEs), leading to increased levels of cyclic AMP (cAMP) and cyclic GMP (cGMP). This rise helps stabilise cell membranes and reduces the release of pro-inflammatory cytokines from monocytes and macrophages. Zinc is also an essential cofactor for the protein A20, which negatively regulates the TLR4 pathway. During Gram-negative infections (such as certain VB species), A20 de-ubiquitinates critical signalling molecules, including TRAF6, thereby stopping NF-kappaB activation and the inflammatory response. A proper balance between regulatory T cells (Treg cells), which promote tolerance, and Th17 cells, which encourage immunity, is crucial. Zinc deficiency skews this balance toward a more aggressive Th17 response. Supplementing zinc helps restore Treg levels, fostering an immune environment that supports sperm and embryo survival. Additionally, zinc promotes epithelial cell growth and tight junction protein production, strengthening the vaginal mucosa after VB-related damage and reducing exposure of sperm antigens to the immune system [73, 74]. Studies have shown that immune suppression occurs at oral zinc intake of 100 mg/day, due to IFN-α production and inhibition of T-cell function at high doses [71].

For BV treatment, the utilisation of vaginal ovules was balanced between the two groups, both for self-reports of AEs (p = 0.32) and for amputability (p = 0.56). While initial clinical cure rates at Visit 1 were comparable between groups (83.87% vs. 83.61%, p = 0.968), combination therapy demonstrated immediate biological superiority over conventional treatment with metronidazole alone (RR = 1.59, p = 0.014) and a significantly higher therapeutic cure rate (61.29% vs. 40.32%, p = 0.024). This therapeutic advantage became increasingly pronounced over time. By Visit 2, the combination significantly outperformed metronidazole in both clinical (RR = 1.51, p = 0.022) and biological (p = 0.009) outcomes. At the final assessment (Visit 3), the clinical cure rate for the combination arm remained nearly twofold higher than that of the metronidazole group (50% vs. 26.33%, RR = 1.73, p = 0.002). Although standard Amsel criteria did not differentiate the regimens at Visit 1 (p = 0.748), combination therapy provided more durable efficacy, maintaining superior clearance through Visits 2 (p = 0.046) and 3 (p = 0.035). While some trials found that adding vaginal clindamycin to oral metronidazole did not reduce long-term BV recurrence [76], our results indicate that a combined local approach may improve initial cure rates and mid-term stability. This is consistent with network meta-analyses suggesting that combined therapies and multi-phase treatments can increase clinical cure rates in BV [77]. These observations are crucial and demonstrate that metronidazole masks clinical symptoms without normalising the underlying microbiological parameters. This clinico-biological discrepancy suggests that the standard treatment fails to restore true eubiosis, leaving a sub-inflammatory environment that is detrimental to sperm survival [17]. Studies have shown the low efficacy of metronidazole alone, and some have suggested that women may benefit from additional biofilm-disrupting and/or pathobiont-targeting treatments [26–28].

Survival analysis confirmed significant differences in the probability of conception across the treatment groups (Log-Rank p = 0.007). Zinc-based regimens, specifically the Zinc-combined treatment and Zinc-metronidazole groups, showed a more rapid initial decline in infertility. This early divergence was corroborated by the Wilcoxon test (p = 0.008), which emphasises early events and indicates a robust and consistent treatment effect throughout the follow-up period. The mechanistic link between BV clearance and ASA reduction is likely rooted in the local immune environment. Chronic vaginal infections and recurrent dysbiosis have been hypothesised to trigger ASA production through immunological cross-reactivity between microbial antigens and sperm cells. Some pathogenic bacteria share common antigens with spermatozoa, potentially leading to polyclonal B-cell activation [78]. Anaerobic bacteria causing BV release toxins and enzymes, such as sialidases, that degrade sialic acids that protect sperm from cervical mucus, exposing sperm epitopes. These unmasking triggers the production of anti-sperm IgG and IgA antibodies [17, 79].

Results of this study demonstrate that a factorial approach targeting vaginal dysbiosis and modulating the immune response leads to a marked improvement in conception rates. Both interventions exerted independent main effects on conception, and the formal Factor 1 × Factor 2 interaction was not statistically significant (likelihood-ratio p = 0.59). By achieving a more robust therapeutic cure of BV, the combined antimicrobial therapy may have reduced the underlying antigenic stimulus; whether this biologically plausible mechanism potentiates the systemic effects of zinc or prednisolone remains a hypothesis to be confirmed in an adequately powered trial.

Limitations

This study has several limitations. Firstly, the trial was powered for the marginal main effects. Its analyses of the antimicrobial-immunomodulatory interaction are exploratory. The design is underpowered for combination effects, interactions, and safety contrasts, with the null results reflecting insufficient power rather than equivalence. After controlling the family-wise error rate with the Holm procedure, the advantage of zinc and prednisolone over placebo survived correction, whereas the combination estimates did not and are hypothesis-generating. This stability held under Bonferroni and Benjamini-Hochberg, confirming immunomodulation as the trial’s principal finding. Secondly, clinical pregnancy, though validated, remains a surrogate for live birth, which the short follow-up in a resource-limited setting could not ascertain. Future trials should extend to live-birth and pregnancy-loss outcomes. Thirdly, ASA titres were not re-measured, so the proposed chain, bacterial vaginosis clearance lowering antigenic stimulation, ASA decline, then conception, rests on inference. Serial ASA measurement and microbiome profiling are needed to demonstrate mediation. Reduced ASA, improved endometrial receptivity and attenuated inflammation remain inseparable explanations, especially for pleiotropic zinc. Finally, strict ASA-plus-BV entry criteria and a single resource-limited centre without assisted reproduction yield a highly selected population, meaning that findings need to be confirmed in broader multicentre settings.

Conclusion

This trial demonstrates that a sequential strategy, combined antimicrobial therapy followed by immunomodulation, can significantly improve fertility in ASA-positive patients with bacterial vaginosis. Combined antimicrobial therapy outperformed metronidazole alone in restoring vaginal eubiosis and was associated with a higher marginal conception rate (aHR 2.20), while immunomodulation independently improved conception (aHR 3.43), consistent with a link between the vaginal microbiome and immunological infertility. Zinc’s non-inferiority to prednisolone and its better tolerability make it a safe, effective, and inexpensive option for regulating innate immunity. Clinicians should view immunological infertility as a mutable mucosal environment, where restoring the vaginal ecosystem with antibacterial treatment followed by zinc modulation offers a promising therapy. This approach reduces reliance on invasive ART and provides hope for couples with unexplained infertility.

Abbreviations

AMH

Antimülleran hormone

ART

Assisted reproductive technology

ASA

Antisperm Antibody

ASRM

American Society for Reproductive Medicine

b-hCG

Beta-human chorionic gonadotrophin

BV

Bacterial vaginosis

CLSI

Clinical and Laboratory Standards Institute

COKIS

Clinique Orchidées de Kisangani

CONSORT

Consolidated Standards of Reporting Trials

DRC

Democratic Republic of the Congo

ESHRE

European Society of Human Reproduction and Embryology

lgA

Immunoglubulin A

lgG

Immunoglubulin G

IT

Information Technology

IUI

Intrauterine insemination

ICSI

Intracytoplasmic Sperm Injection

IVF

In Vitro Fertilisation

LMT

Limited

MAR

Mixed antiglobulin reaction

mITT

Modified intention-to-treat

NA

Not applicable

PP

Per-protocol

PCOS

Polycystic ovary syndrome

RDT

Rapid Diagnostic Test

STIs

Sexually transmitted infections

UI

Unexplained infertility

WHO

World Health Organization

Authors’ contributions

M.A.M.A, J.N.K, S.B.A, J.J.J.S and K.B conceived the protocol. S.B.A, K.B and J.N.K validated the protocol. M.A.M.A, L.B.B, S.M.M and B.A.I collected data. M.A.M.A, S.B.A, N.L.O and K.B supervised data collection and treatment. M.A.M.A and N.L.O wrote the manuscript. M.A.M.A, N.L.O, J.J.J.S, S.B.A, J.N.K, S.M.M, B.A.I and K.B checked the manuscript. K.B supervised all the processes. All authors approved the final version.

Funding

This research was funded by Gardiens de Vies, a local non-governmental organisation.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The study was approved by the Provincial Ethics Committee of the Tshopo Provincial Health Division (701/FBL/DPS/TSHOPO/SEC/0173/2023) and the National Health Ethics Committee (531/CNES/BN/PMMF/2024). It was conducted in accordance with the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use Guideline for Good Clinical Practice and the Declaration of Helsinki [80]. The trial has been retrospectively registered with the Pan African Clinical Trials Registry on 08 October 2024 under the identifier PACTR202410672612184||http://www.pactr.org/. Confidentiality was ensured by maintaining anonymity throughout data collection, processing and analysis. The study's objectives and procedures were explained to the patients beforehand.

Consent for publication

All participants signed an informed consent form before enrolling in the study. Participants could withdraw at any time, without justification and without affecting the quality of care. The principal investigator could withdraw a participant upon the participant’s explicit request. Prior to revealing the results, each patient was required to sign a consent form authorising the publication of de-identified data. Consent for the publication of identifiable images or other personal or clinical details of participants that might compromise anonymity is not relevant for this study.

Competing interests

The authors declare no competing interests.