Intrauterine infusion of oil-based medium in women with recurrent implantation failure before frozen embryo transfer: a randomized clinical trial.

After a preliminary evaluation of 358 patients with RIF, 96 patients did not meet the enrollment criteria, and 38 patients refused to participate. The 224 enrolled patients were randomized: 112 were allocated to the oil-based medium group, and 112 were allocated to the control group. Twelve patients in the oil-based medium group were subsequently excluded after randomization and intervention: one discontinued follow-up, five did not receive FET within 120 days, one declined to receive FET, one did not receive FET due to an endometrial thickness < 7 mm, one received a fresh embryo transfer to replace FET, and three achieved a natural pregnancy. Among the three patients who experienced a spontaneous pregnancy within 85–113 days after oil-based medium infusion, two patients gave live birth, and one patient had a spontaneous miscarriage at 9 weeks of pregnancy. Five women were excluded after randomization in the control group: two discontinued follow-ups, two did not receive FET due to an endometrial thickness of < 7 mm, and one canceled FET for personal reasons (Fig.  1 ). The demographic characteristics of the individuals in the two groups in the ITT population are presented in Table  1 . The two groups were compared in terms of their characteristics, including age, maternal BMI, infertility duration, infertility type, cause of infertility, and number of embryo transfer cycles (Table  1 ). The number of transferred embryos, number of high-quality transferred embryos, cryopreservation stage, endometrial preparation protocol of FET, and endometrial thickness were similar between the groups based on per-protocol analyses (Table  2 ). To assess the impact of excluding three patients with natural pregnancies on the results, we conducted the following sensitivity analysis 1. ITT: This analysis included the three patients in the treatment group who did not undergo embryo transfer (considered as "non-live birth"), the LBRs in the treatment group and the control group were 34.8% and 25.9%, respectively (RR = 1.345, 95% CI 0.899–2.012, P  = 0.146). 2. Best/Worst Case Analysis: Best case (considering the three were successful cases): The pregnancy rate in the treatment group was 37.5% (rate ratio = 1.448, 95% CI 0.977–2.148, P  = 0.062). Worst case (considering the three as unsuccessful): The results were consistent with the ITT analysis. 3. Comparison of Subgroup Characteristics: The age and duration of infertility of the three patients with natural pregnancies were not significantly different from those of the other patients in the treatment group ( P  > 0.05). Among the 112 patients who underwent intrauterine infusion, none experienced adverse reactions such as allergies, embolisms, fever, or infections. Fig. 1 The study flowchart Table 1 Characteristics of the ITT study population Variables Oil-based medium Control P value (n = 112) (n = 112) Age (years) 34 (31,38) 35 (31,38) 0.989 Maternal BMI (Kg/m 2 ) 21.80 (19.66, 22.90) 20.26 (19.48, 22.57) 0.081 Infertility duration (years) 3 (2,6) 5 (2,7) 0.087 Infertility type   Primary infertility, n (%) 43(38.4) 34(30.4) 0.205   Secondly infertility, n (%) 69(61.6) 78(69.6) Infertility causes   Ovulation factors, n (%) 5(4.5) 9(8.0)   Tubal or pelvic factors, n (%) 88(78.5) 80(71.4)   Male factors, n (%) 9(8.0) 12(10.7) 0.718   Endometriosis, n (%) 7(6.3) 7(6.3)   Unexplained, n (%) 3(2.7) 4(3.6)   Embryo transfer cycles (no) 4(3,4) 4(3,5) 0.057   LBR* 39(34.8) 29(25.9) 0.146 BMI Body mass index; ITT Intention-to-treat; LBR*, Live birth rate, this analysis included three patients in the treatment group who achieved natural pregnancy without undergoing frozen-thawed embryo transfer (categorized as "non-live birth"), resulting in a rate ratio of 1.345 (95% confidence intervals 0.899–2.012). Continuous variables, including age, maternal BMI, infertility duration, and embryo transfer cycles, are presented as median (interquartile range) due to non-normal distribution. Categorical variables (infertility type, infertility causes, and LBR) are expressed as counts (percentages). No significant intergroup differences were observed in the ITT population Table 2 Embryo transfer characteristics and endometrial parameters in RIF patients undergoing FET Variables Oil-based medium (n = 100) Control (n = 107) P value Number of transferred embryos   1, n (%) 45(45.0) 50(46.7)   ≥ 2, n (%) 55(55.0) 57(53.3) 0.439 Percent of high-quality transferred embryos (%) 100(50,100) 100(50,100) 0.443 Stage of cryopreservation   Cleavage stage, n (%) 58(58.0) 52(48.6)   Blastocysts, n (%) 42(42.0) 55(51.4) 0.176  Endometrium thickness (mm) 9(8,11) 9(8,11) 0.912 Endometrial preparation protocol   Natural /Mild stimulation protocol, n (%) 8(8.0) 20(18.7)   HRT, n (%) 46(46.0) 40(37.4) 0.069   GnRHa-HRT, n (%) 46(46.0) 47(43.9) RIF Recurrent implantation failure; FET Frozen-thawed embryo transfer; HRT Hormone replacement therapy; GnRHa-HRT Gonadotropin-releasing hormone agonist-pretreated HRT. Continuous variables (percent of high-quality embryos and endometrial thickness) are presented as median (interquartile range) for non-normally distributed data. Categorical variables, including number of transferred embryos, cryopreservation stage, and endometrial preparation protocol, are reported as n (%). No significant differences were observed in embryo transfer characteristics or endometrial parameters between groups among RIF patients undergoing FET The study flowchart Characteristics of the ITT study population BMI Body mass index; ITT Intention-to-treat; LBR*, Live birth rate, this analysis included three patients in the treatment group who achieved natural pregnancy without undergoing frozen-thawed embryo transfer (categorized as "non-live birth"), resulting in a rate ratio of 1.345 (95% confidence intervals 0.899–2.012). Continuous variables, including age, maternal BMI, infertility duration, and embryo transfer cycles, are presented as median (interquartile range) due to non-normal distribution. Categorical variables (infertility type, infertility causes, and LBR) are expressed as counts (percentages). No significant intergroup differences were observed in the ITT population Embryo transfer characteristics and endometrial parameters in RIF patients undergoing FET RIF Recurrent implantation failure; FET Frozen-thawed embryo transfer; HRT Hormone replacement therapy; GnRHa-HRT Gonadotropin-releasing hormone agonist-pretreated HRT. Continuous variables (percent of high-quality embryos and endometrial thickness) are presented as median (interquartile range) for non-normally distributed data. Categorical variables, including number of transferred embryos, cryopreservation stage, and endometrial preparation protocol, are reported as n (%). No significant differences were observed in embryo transfer characteristics or endometrial parameters between groups among RIF patients undergoing FET All randomized patients who underwent FET had complete pregnancy outcomes. There was no statistically significant difference in the LBR (39.0% vs. 27.1%, RR 1.439, 95% CI 0.969–2.138, P  = 0.069), OPR (40% vs. 27.1%, RR 1.476, 95% CI 0.997–2.186 , P  = 0.049), early miscarriage (16% vs. 10.3%, RR1.556, 95% CI 0.759–3.190, P  = 0.222) between the two groups. However, The CPR was significantly higher in patients who received the oil-based medium infusion than in the control group (56% vs. 37.4%, RR 1.498, 95% CI 1.109–2.033, P  = 0.007), and the implantation rate also increased (median50% [IQR 0–100%] vs. median 0% [IQR 0–50%]), P  = 0.010 (Table  3 ). Table 3 Pregnancy outcome for RIF patients undergoing FET Variables Oil-based medium Control Rate ratio /z/t P- Value (uncontrolled) Significant (Adjust) (n = 100) (n = 107) (95% CI) Live birth, n (%) 38(39.0) 29(27.1) 1.439 (0.969–2.138) 0.069 No Clinical pregnancy, n (%) 56(56.0) 40(37.4) 1.498 (1.109–2.023) 0.007 Yes* Ongoing pregnancy, n (%) 40(40.0) 29(27.1) 1.476 (0.997–2.186) 0.049 No Early miscarriage, n (%) 16(16.0) 11(10.3) 1.556 (0.759–3190) 0.222 No Implantation rate (%) 50(0,100) 0(0,50)  − 2.528(0.010–0.014) 0.010 Yes* Newborn weight(g) 2940 ± 345 3145 ± 388  − 2.453(0.023–0.026) 0.458 No RIF Recurrent implantation failure; FET Frozen-thawed embryo transfer. Data: Continuous variables with a normal distribution are presented as the mean ± standard deviations, and a t -test was used; n (%) for categorical variables (Chi-square/Fisher's exact tests); median (IQR) for non-normal continuous variables (Mann–Whitney U). Primary outcome α = 0.05; secondary outcomes α = 0.013 (Bonferroni-corrected). *Significant differences in clinical pregnancy and implantation rates post-adjustment Pregnancy outcome for RIF patients undergoing FET RIF Recurrent implantation failure; FET Frozen-thawed embryo transfer. Data: Continuous variables with a normal distribution are presented as the mean ± standard deviations, and a t -test was used; n (%) for categorical variables (Chi-square/Fisher's exact tests); median (IQR) for non-normal continuous variables (Mann–Whitney U). Primary outcome α = 0.05; secondary outcomes α = 0.013 (Bonferroni-corrected). *Significant differences in clinical pregnancy and implantation rates post-adjustment Furthermore, we categorized the patients into three groups according to age: those younger than 35 years, those aged 35 to 39 years, and those older than 40 years. We compared the pregnancy outcomes among the patients in the three groups. The results indicated that there was no statistically significant difference in the LBR, among those aged < 35 years (46.2% vs. 32%, RR 1.442, 95% CI 0.875–2.337, P  = 0.143), 35–39 years (27.3% vs. 23.07%, RR 1.064, 95% CI 0.491–2.302, P  = 0.876), or ≥ 40 years (40% vs. 16.7%, RR 2.400, 95% CI 0.720–8.006, P  = 1.134, and there was also no statistically significant difference in OPR and miscarriage rate between the two groups in each age group. However, the CPR and early miscarriage rate in the oil-based intervention group was higher than that in the control group in the group of patients aged 35–39 years old, with CPR (54.5% vs. 30.7%, RR 1.773, 95% CI 1.008–3.117, P  = 0.041), miscarriage rate (24.2% vs. 5.13%, RR 4.727, 95% CI 1.078–3.117, P  = 0.036) (Table  4 ). Table 4 Comparison of pregnancy outcomes among different ages Variables Oil-based medium (n = 100) Control Rate ratio/Z P value (n = 107) (95% CI) Aged < 35 years, n (%) 52(52.0) 50(46.7) Live birth, n (%) 24(46.2) 16(32.0) 1.442(0.875–2.337) 0.143 Ongoing pregnancy, n (%) 24(46.2) 16(32.0) 1.442(0.875–2.337) 0.143 Clinical pregnancy, n (% 30(57.7) 21(42.0) 1.374(0.920–2.050) 0.113 Early miscarriage, n (%) 6(11.5) 5(10.0) 1.154(0.376–3.542) 0.802 Implantation rate (%) 50(0,100) 0(0,62.5)  − 1.673(0.091–0.103) 0.094 Aged 35–39 years, n (%) 33(33.0) 39(36.5) Live birth, n (%) 9(27.3) 9(23.07) 1.064(0.491–2.302) 0.876 Ongoing pregnancy, n (%) 10(30.3) 10(25.6) 1.182(0.562–2.487) 0.66 Clinical pregnancy, n (% 18(54.5) 12(30.7) 1.773(1.008–3.117) 0.041* Early miscarriage, n (%) 8(24.2) 2(5.13) 4.727(1.078–20.736) 0.036* Implantation rate (%) 50(0,50) 0(0,50) -1.918(0.031–0.038) 0.055 Aged ≥ 40 years, n (%) 15(15.0) 18(16.8) Live birth, n (%) 6(40.0) 3(16.7) 2.400(0.720–8.006) 0.134 Ongoing pregnancy, n (%) 6(40.0) 3(16.7) 2.400(0.720–8.006) 0.134 Clinical pregnancy, n (% 8(53.3) 7(38.9) 1.371(0.649–2.897) 0.407 Early miscarriage, n (%) 2(13.3) 3(16.7) 1.041(0.781–1.385) 0.79 Implantation rate (%) 33.33(0,50) 0(0,62.5) -0.438(0.681–0.699) 0.661 Data presented as n (%) for categorical variables (analyzed by χ 2 /Fisher's exact tests) and median (IQR) for non-normally distributed continuous variables (Mann–Whitney U test). The oil-based medium group demonstrated significantly higher clinical pregnancy and early miscarriage rates than controls, specifically in the 35–39-year age subgroup. No differences were observed in patients aged < 35 years or ≥ 40 years Comparison of pregnancy outcomes among different ages Data presented as n (%) for categorical variables (analyzed by χ 2 /Fisher's exact tests) and median (IQR) for non-normally distributed continuous variables (Mann–Whitney U test). The oil-based medium group demonstrated significantly higher clinical pregnancy and early miscarriage rates than controls, specifically in the 35–39-year age subgroup. No differences were observed in patients aged < 35 years or ≥ 40 years A comparison of pregnancy outcomes among different endometrial preparation protocols revealed that in the subgroup analysis of the GnRHa-HRT-FET protocols, the CPR and implantation rates in the oil-based contrast group were significantly higher than those in the control group, including the CPR (71.7% vs. 44.7%, RR 1.606, 95% CI 1.113–2.316, P  = 0.008), and the implantation rate (median 50% [IQR 0–100%] vs. median 0% [IQR 0–100%], P  = 0.048), respectively. Yet, there were no differences in disparity in the LBR, OPR, or miscarriage rate. Furthermore, HRT, natural cycle/ mild stimulation protocol demonstrated comparable clinical outcomes across all measured parameters (Table  5 ). Table 5 Comparison of pregnancy outcomes among different endometrial preparation protocols Variables Oil-based medium (n = 100) Control Rate ratio/Z P value (n = 107) (95% CI) GnRHa-HRT 46(46.0) 47(43.9)   Live birth, n (%) 24(52.2) 18(38.3) 1.362(0.875–2.337) 0.179   Ongoing pregnancy, n (%) 24(52.2) 18(38.3) 1.362(0.875–2.337) 0.179   Clinical pregnancy, n (% 33(71.7) 21(44.7) 1.606(1.113–2.316) 0.008*   Early miscarriage, n (%) 9(19.6) 3(6.4) 3.065(0.885–10.611) 0.058   Implantation rate (%) 50(0,100) 0(0,100) -1.981(0.044–0.053) 0.048* HRT 46(46) 40(37.4)   Live birth, n (%) 12(26.1) 9(22.5) 1.159(0.546–2.463) 0.699   Ongoing pregnancy, n (%) 13(28.3) 9(22.5) 1.256(0.601–2.624) 0.541   Clinical pregnancy, n (% 18(39.1) 13(32.5) 1.204(0.678–2.137) 0.523   Early miscarriage, n (%) 5(10.9) 4(10.0) 1.087(0.313–3.774) 0.895   Implantation rate (%) 0(0,50) 0(0,50) -1.412(0.249–0.266) 0.254 Natural /Mild stimulation protocol 8(8.0) 20(18.7)   Live birth, n (%) 3(37.5) 2(10.0) 3.750(0.765–18.394) 0.086   Ongoing pregnancy, n (%) 3(37.5) 2(10.0) 3.750(0.765–18.394) 0.086   Clinical pregnancy, n (% 5(62.5) 6(30.0) 2.083(0.883–4.914) 0.112   Early miscarriage, n (%) 2(25.0) 3(15.0) 1.667(0.340–8.175) 0.533   Implantation rate (%) 41.66(0, 87.5) 0(0,50) -1.160(0.233–0.249) 0.246 HRT Hormone replacement therapy; GnRHa-HRT Gonadotropin-releasing hormone agonist-pretreated HRT; 95% CI 95% confidence interval. Data expressed as n (%) for categorical variables (χ 2 /Fisher's exact tests) and median (IQR) for non-normally distributed continuous variables (Mann–Whitney U test). In the GnRHa-HRT subgroup, the oil-based medium group exhibited significantly higher clinical pregnancy and implantation rates compared to controls ( P  < 0.05). No significant differences were observed between groups in HRT or Natural/Mild endometrial preparation protocols Comparison of pregnancy outcomes among different endometrial preparation protocols HRT Hormone replacement therapy; GnRHa-HRT Gonadotropin-releasing hormone agonist-pretreated HRT; 95% CI 95% confidence interval. Data expressed as n (%) for categorical variables (χ 2 /Fisher's exact tests) and median (IQR) for non-normally distributed continuous variables (Mann–Whitney U test). In the GnRHa-HRT subgroup, the oil-based medium group exhibited significantly higher clinical pregnancy and implantation rates compared to controls ( P  < 0.05). No significant differences were observed between groups in HRT or Natural/Mild endometrial preparation protocols GnRHa-HRT-FET and HRT-FET cycles were the main endometrial preparation protocols in the oil-based medium infusion group (46% vs. 46%, respectively), and a subgroup analysis was performed to compare pregnancy outcomes between the two cycles. The LBR (52.2% vs.26.1%, RR 1.894, 95% CI 1.035–3.466, P  = 0.023), CPR (71.7% vs.39.1%, rate ratio1.983, 95% CI 1.198–3.283, P  = 0.002), OPR (52.2% vs. 28.3%, RR1.705, 95% CI 1.001–3.008, P  = 0.047), the implantation rate (median 50% [IQR 0–100%] vs. median 50% [IQR 0–50%], P  = 0.021), and endometrium thickness (9.83 ± 1.85 mm vs. 8.92 ± 2.02 mm, P  = 0.027) were significantly higher among the GnRHa-HRT-FET cycles than among the HRT-FET cycles. There were no statistically significant differences in age, infertility causes, high-quality transferred embryos, time from infusion to FET, or early miscarriage rate between the two protocols (Table  6 ). Table 6 Pregnancy outcomes of the GnRHa-HRT and HRT-protocols in the oil-based medium group Variables GnRHa-HRT HRT t/Z/χ 2 /rate ratio P value (n = 46) (n = 46) (95% CI) Age (years) 34.43 ± 4.29 34.28 ± 4.49  − 2.302 0.868 Maternal BMI (Kg/m 2 ) 21.65 ± 2.53 21.91 ± 2.32  − 0.231 0.614 Infertility causes   Ovulation factors, n (%) 2(4.3) 2(4.3)   Tubal or pelvic factors, n (%) 33(71.8) 39(84.9)   Male factors, n (%) 5(10.9) 2(4.3) 2.848 0.584   Endometriosis, n (%) 4(8.7) 2(4.3)   Unexplained, n (%) 2(4.3) 1(2.2)   Time from infusion to FET (days) 79.62 ± 21.99 74.10 ± 23.26  − 2.345 0.756   Endometrium thickness (mm) 9.83 ± 1.85 8.92 ± 2.02  − 2.413 0.027*   High-quality transferred embryos (%) 50(50,100) 100(50,100)  − 1.852 0.064   Embryo transfer cycles (no) 4(3,4) 4(3,4)  − 0.204 0.838 Pregnancy outcomes   Clinical pregnancy, n (%) 33(71.7) 18(39.1) 1.983 (1.198–3.283) 0.002*   Ongoing pregnancy, n (%) 24(52.2)) 13(28.3) 1.705 (1.001–3.008) 0.047*   Live birth, n (%) 24(52.2) 12(26.1) 1.894 (1.035–3.466) 0.023*   Early miscarriage, n (%) 9(19.6) 5(10.9) 4.773(0.619–36.82) 0.087   Implantation rate (%) 50(0,100) 0(0,50) -2.302(0.009–0.013) 0.021* HRT Hormone replacement therapy; GnRHa-HRT Gonadotropin-releasing hormone agonist-pretreated HRT; 95% CI 95% confidence interval. Normally distributed continuous variables: mean ± SD (analyzed by t -test).Non-normally distributed variables: median (IQR) (Mann–Whitney U test).Categorical variables: n (%) (χ 2 /Fisher's exact tests). Compared to HRT, the GnRHa-HRT protocol demonstrated significantly greater endometrial thickness and higher rates of clinical pregnancy, live birth, ongoing pregnancy, and implantation, * P  < 0.05 Pregnancy outcomes of the GnRHa-HRT and HRT-protocols in the oil-based medium group HRT Hormone replacement therapy; GnRHa-HRT Gonadotropin-releasing hormone agonist-pretreated HRT; 95% CI 95% confidence interval. Normally distributed continuous variables: mean ± SD (analyzed by t -test).Non-normally distributed variables: median (IQR) (Mann–Whitney U test).Categorical variables: n (%) (χ 2 /Fisher's exact tests). Compared to HRT, the GnRHa-HRT protocol demonstrated significantly greater endometrial thickness and higher rates of clinical pregnancy, live birth, ongoing pregnancy, and implantation, * P  < 0.05

Intrauterine infusion with an oil-based medium improved the CPR and implantation rate concerning FET prognosis in patients with unknown RIF. However, there is no evidence that it can increase the LBR. This study provides new ideas for the individualized treatment of RIF patients; nonetheless, cautious selection is needed, and more studies are necessary to clarify its role and mechanism.

This multicenter prospective RCT revealed that women with unexplained RIF who underwent oil-based medium infusion had higher CPR in FET. The LBR in the oil-based medium group was 12% higher than that in the control group; however, no statistically significant difference was detected between the two groups. Additionally, there was no disparity in the OPR or miscarriage rate between the two groups. The subgroup analysis on the basis of age indicated that the CPR in the oil-based medium group of those 35–39 years old was significantly higher than that in the control group. The subgroup analysis of the FET endometrial preparation protocol demonstrated that the LBR in the GnRHa-FET group was greater than that in the HRT group. The results of Reilly did not support the idea that an oil-based contrast medium improved LBR to fresh embryo transfer [ 12 ]. Our results revealed that no improvement in the LBR was observed in the FET with uterine perfusion of oil-based medium in patients with RIF, which was in line with their results. Contrary to that, we discovered a slight enhancement in CPR and implantation rates following oil perfusion, accompanied by an increase in the miscarriage rate, particularly in the age group of 35–39, furthermore, their patient received Lipiodol by HSG before IVF treatment within 6 months, and we infused an oil-based medium in a lower dage without fluoroscopic X-ray screening and performed FET at 31 and 120 days after infusion. Iodized oil is likely to mainly have a local regulatory effect in the endometrium. Studies regarding the iodine-uterine effects in animals have indicated that large doses of iodine might be toxic, while smaller doses could be beneficial for creating a uterine environment favorable for normal reproduction [ 9 ]. A previous RCT by Dreyer, K revealed the fertility effects of oil-based medium contrast also continuously functioned during several menstrual cycles and inferred that it was a direct endometrial effect [ 1 ]. The effects of oil-based medium on the endometrium include the improvement of chronic endometritis, immune changes such as alterations in endometrial dendritic cells [ 9 ], reduced expression of osteopontin [ 3 ], and upregulated expression of endometrial alphavbeta3 integri (αvβ3) the window of implantation [ 11 ], and αvβ3 had the best predictive value for endometrial receptivity among biomarkers in the uterine fluid, and play an important role in the endometrial phenotype change that occurs during the secretory phase, the first stage of implantation [ 16 , 17 ]. we hypothesize that the improvement in the CPR might be attributed to the shorter interval between uterine perfusion and embryo transfer, as well as the low dosage of oil-based medium. Furthermore, in the subgroup analysis based on age, we discovered that in the 35–39 age group, both the CPR and miscarriage rates of RIF patients in the iodized oil group were higher than those in the control group. This might be the complex and diverse pathological factors of RIF, especially the quality and developmental potential of transplanted embryos. This is also one of the reasons why, although the clinical pregnancy rate has increased, there is no significant difference in the LBR and OPR between the two groups. Interestingly, in this trial, the rates of clinical pregnancy, ongoing pregnancy, and LBR in the GnRHa–HRT protocol group were significantly higher than those in the HRT protocol group after oil contrast infusion. Some studies have reported fertility-enhancing GnRHa-HRT-FET cycles in women with RIF. According to Steiner, the CPR and LBR were higher among infertile women who had failed two blastocyst transfers and received a GnRHa plus letrozole pretreatment FET than among women with a GnRHa-HRT-FET or women without pretreatment FET. This result suggests that amelioration of endometrial receptivity or a remedy for latent endometriosis might be one of the reasons [ 18 ]. GnRHa-HRT protocols improved the CPR, OPR, and LBR in FET cycles for infertile women of advanced reproductive age with idiopathic RIF. Long-acting GnRHa may influence the regulation of endometrial genetic alterations and affect pregnancy outcomes [ 19 ]. Additionally, oil-based contrast has shown an immunomodulatory advantage in endometriosis-related infertility that might influence the implantation window increasing the number of uterine natural killer cells [ 3 ]. In this trial, 78.6% of the patients had pelvic fallopian tube factors and might have potential endometriosis. GnRHa downregulation may have improved the endometrial receptivity of patients with RIF and endometriosis or latent endometriosis, leading to an increased pregnancy outcome in the oil-based contrast group. Therefore, intrauterine infusion with an oil-based medium plus the GnRHa-HRT-FET cycle may synergistically affect women with RIF and increase endometrial receptivity. While the findings indicate that the CPR improved following the administration of oil, and an increase in the LBR of approximately 12%; however, this improvement also lacked sufficient statistical support to conclusively demonstrate its efficacy. This cautious stance arises from several factors, including the lack of definitive evidence supporting its widespread use and the variability in individual patient responses. Furthermore, given the complexity of RIF and the multitude of contributing factors—ranging from anatomical abnormalities to immunological issues—it is crucial to carefully select candidates who may benefit most from iodized oil treatment. Furthermore, we need to focus on the safety and side effects of oil-based contrast agents. In terms of the safety of intrauterine infusion and flushing with an oil-based contrast agent, in a systematic review, Roest reported that intravasation with an HSG contrast agent was the most common complication, with an occurrence rate of 2.7% [ 20 ]. In recent years, potential iodine excess and the far-reaching impact of oil contrast agents in HSG on the iodine status of mothers and offspring have attracted an increasing amount of attention [ 21 , 22 ]. A multicenter RCT across 27 hospitals revealed that among 557 women assigned to HSG with oil contrast, no cases of subsequent allergic reactions or fat embolism were observed [ 1 ]. In this trial, we infused an oil-based contrast agent at a dosage of 5 mL, equivalent to the uterine cavity volume. Compared with routine HSG, intrauterine infusion with oil-based contrast may be more convenient and without the potential adverse effects of X-rays. In this study, we did not observe any adverse reactions or congenital anomalies in FET patients. There was no statistically significant difference in the weight of the offspring between the two groups, suggesting that a 5 ml perfusion volume might be relatively safe for both the mothers and the offspring. However, we did not conduct any follow-up on the thyroid function of the mother or the offspring. The potential impact of iodized oil on the thyroid function of both the mothers and offspring still cannot be ignored [ 20 ] and should be chosen with caution, especially for high-risk patients with abnormal thyroid function and organic lesions. The ESHRE working group on RIF recommends considering RIF as a secondary phenomenon of ART, which provides a means of individualizing its recognition to each specific patient context [ 6 ]. The diagnosis of RIFs on the basis of the Coughlan [ 14 ]. We focus on the population with RIFs based on the cumulative number of embryo transfer failures. The inclusion and exclusion criteria exclude the influencing factors of RIF, and there was no difference in baseline data between the two groups, however, the large age span of patients may affect pregnancy outcomes. The ESHRE working group suggests that to attain a cumulative implantation/pregnancy rate exceeding 60%, patients under 35 years of age (female) require three transplantation cycles, those aged 35–39 need four cycles, and those over 40 need six cycles. The RIF definition attaches significance to the patient's age and should concurrently take into account the variance in the number of transplantation cycles for different age brackets. In our study, at least one high-quality embryo was transplanted in each cycle. After all patients were examined, the cumulative implantation/pregnancy rate was above 60%. In reference to the consensus of the ESHEG working group [ 6 ], we performed subgroup analyses according to patient age to analyze the pregnancy outcomes of RIF patients. Our research results indicate that the oil-based medium group failed to increase the LBR in all the age groups. We need to pay attention to the impact of the individualized conditions of patients with RIF on pregnancy outcomes in different age groups. This study has several limitations. Since the control group underwent no intervention, our clinical trial design was not blinded. The lack of blinding in this trial may introduce performance and detection biases. For instance, clinicians' heightened attention to the intervention group could lead to differential co-interventions, while participants' awareness of their allocation might influence subjective outcome reporting [ 23 ]. While double-blinding would be ideal, practical constraints prevented its implementation. In this trial, the intervention (intrauterine oil infusion) inherently differs from no infusion in procedural sensation, making sham controls ethically questionable due to unnecessary uterine instrumentation; To mitigate these risks, we employed standardized protocols for both intervention delivery and outcome assessment, embryo transfer techniques and luteal support were strictly standardized across centers. Although this was a multicenter clinical study, the sample size was small, and the distribution of patients with RIF was unbalanced across centers. However, the LBR was the objective primary outcome in this trial. The demographic characteristics, previous transplant failure, endometrial preparation, luteal support plans, and embryo quality were similar between the two groups according to the ITT analysis, which reduced the influence of a lack of blinding and fewer and unbalanced sample sizes. Additionally, three women randomized to oil-based medium infusion had a natural pregnancy after the infusion but before FET. Thus, these patients were excluded from per-protocol analyses because they did not undergo FET. All the results of the sensitivity analysis were consistent with those of the original analysis in terms of direction, and the statistical significance essentially did not change, indicating that the research conclusions were not sensitive to the data exclusion. However, they achieved spontaneous pregnancy within 85–113 days after infusion, and these results were relevant to an oil-based medium but were not well demonstrated and analyzed. In conclusion, although the CPR and implantation rates have increased, at present, there is no evidence to substantiate a significant increase in LBR. Oil-based medium has been a traditional contrast for HSG for several decades, however, intrauterine infusion without HSG for clinical treatment, such as identifying suitable populations and developing individualized treatment plans, requires further investigation. Moreover, there is a critical need for more comprehensive investigations into the underlying mechanisms and larger-scale RCTs.

In recent years, oil-based contrast media in hysterosalpingography (HSG) have been recommended for improving fertility in infertile patients and those with endometriosis [ 1 – 4 ]. Although it is believed to clear debris from fallopian tubes that might effectively impede fertility, oil-based contrast also has immunobiological impacts on the endometrium and pelvic peritoneum, which could contribute to increased fertility [ 5 ]. However, the effects of oil-based contrast agents on women with other reproductive disorders, such as recurrent implantation failure (RIF), merit further investigation. RIF is a clinical phenomenon characterized by repeated failure of implantation. RIF describes the scenario in which the transfer of embryos that are considered to be viable has failed to result in a positive pregnancy test sufficiently often in a specific patient to warrant consideration for further investigations and/or interventions [ 6 ]. Intrauterine infusion therapeutics, including human chorionic gonadotropin (HCG), peripheral blood monocytes, and platelet-rich plasma, may improve endometrial receptivity by directly affecting the endometrium and uterine microenvironment. Nevertheless, therapeutic options for RIF remain challenging owing to the increasing incidence and diversity of etiology. Further studies on individualized therapeutic measures are needed [ 7 ]. Recently, several randomized clinical trials (RCTs) reported the reproductive effects of oil-based contrast, and 2-year follow-up results from an RCT revealed that oil-based contrast flushing was effective at improving fertility and pregnancy outcomes in women with endometriosis and pure unexplained infertility [ 8 ]. A multicenter RCT from 27 hospitals revealed that ongoing pregnancy and live birth rates (LBRs) were higher in patients who underwent HSG with an oil-based contrast medium [ 1 ]. According to the literature, oil-based contrast has multiple effects on the endometrium and pelvic microenvironment, including immunomodulatory properties, inhibiting the expression of CD138, CD38, and NF-KB in the endometrium, regulating the TH1/TH2 axis, modulating dendritic cell and regulatory T cell profiles, and increasing the thickness of the endometrium [ 5 , 9 – 11 ], which collectively underscore its significant exploratory value for the treatment of RIF. The etiologies and pathogenic factors of RIF are diverse. The 2023 consensus of the European Society of Human Reproduction and Embryology (ESHRE) recommends that when patients are suspected of having RIF, hysteroscopy and other examinations should be performed. Great attention should be paid to the role of chronic endometritis in the pathogenesis of RIF, and active anti-inflammatory treatment is necessary. The role of immune factors is also a future research direction for RIF [ 6 ]. At present, only a limited number of studies have analyzed the role of oil-based medium in the population with RIF. In 2019, Reilly, Shelley J reported LBR in the pre-IVF Lipiodol group from pregnancies within 6 months were 11/33 (33%) and 12/37 (32%) in these respective groups (RR 1.03; 95% CI, 0.53 to 2.0), there is no evidence indicating that lipiodol could increase the LBR of fresh cycles in RIF in fresh in vitro fertilization (IVF) cycles[ 12 ], a prospective study based on propensity score matching from Zhonghong Zeng et reported that in the frozen embryo transfer (FET) cycles of patients with RIF, the LBR was significantly lower in the lipiodol flush group, whereas no significant difference was observed in the fresh embryo transfer cycles[ 13 ]. However, there are currently few prospective RCTs on the efficacy of oil-based contrast flushing in FET patients with RIF. This multicenter RCT aimed to assess the effect of intrauterine oil-based contrast infusion on FET prognosis and pregnancy outcomes among patients with unknown RIF. Given the multifaceted regulatory mechanisms of oil-based contrast media on the endometrium and the complex pathophysiology underlying RIF, this study aims to provide novel insights into enhancing the pregnancy outcomes of FET in cases of RIF and developing individualized treatment regimens. This multicenter prospective RCT involved patients with RIF who underwent FET at the First Affiliated Hospital of Guangxi Medical University, Liuzhou Maternity and Child Healthcare Hospital, Affiliated Hospital of Guilin Medical University, Guigang City People's Hospital, Maternity and Child Health Hospital of Guangxi Zhuang Autonomous Region, and Affiliated Hospital of Youjiang between November 2019 and December 2022. This study was approved by the Ethics Committee of the First Affiliated Hospital of Guangxi Medical University and registered at the Chinese Clinical Trial Registry Center ( http://www.chictr.org.cn/index.html , no ChiCTR1900024273) on July 4th, 2019. RIF was diagnosed in accordance with the following criteria reported by Coughlan: women under the age of 40 years who failed to have a clinical pregnancy under transfer, with at least three high-quality embryos in a minimum of three fresh or frozen cycles [ 14 ]. The inclusion criteria for this study were as follows: 1. women aged 20–40 years with spontaneous menstrual cycles; 2. underwent three or more transfer cycles, cumulative transplantation of at least four high-quality cleavage stage embryos or three high-quality blastocysts without clinical pregnancy, and high-quality frozen embryos were still available for implantation; and 3. transferred blastocysts with a grade ≥ 3BB, or transferred cleavage-stage embryos reached a morphological grade of 622 or better according to the Gardner scoring criteria [ 15 ]. The exclusion criteria were as follows, known factors that affect embryo implantation, such as severe uterine adhesions, the septum uterus, and uterine leiomyoma; 2. couples with chromosome abnormalities or males with very weak sperm; 3. patients with hyperthyroidism or hypersensitivity to iodine; 4. severe heart and lung disease, uncontrolled hypertension, diabetes, and obesity (body mass index, [BMI] ≥ 28); 5. serious pregnancy-related diseases, such as luteal and thyroid dysfunction; 6. any conditions that were unstable or unlikely to comply with pregnancy; or 7. inability to be treated with this regimen. Randomization was conducted using computer-generated number sequences by a doctor at the Reproductive Medicine Center of the First Affiliated Hospital of Guangxi Medical University. Patients with RIF were recruited into two groups at a ratio of 1:1. The two groups were the oil-based medium group, which received an intrauterine infusion of 5 ml of ethiodized poppy seed oil (H20160011, Hengrui, China) within 3–7 days after the cessation of menstrual bleeding via a uterine catheter, and the control group which did not receive any intervention. FET was performed during the next menstrual cycle after intrauterine infusion in the experimental group and was completed within 120 days. The endometrial preparation protocols used in this trial included hormone replacement therapy (HRT)-FET cycle, a gonadotropin-releasing hormone agonist (GnRHa) pretreatment HRT(GnRHa-HRT)-FET cycle, and a natural /mild stimulation FET cycle. In natural and mild stimulation FET cycles, thawing transfers with cleavage-stage or blastocyst-stage embryos are performed 4 or 6 days after ovulation. Patients who underwent HRT-FET cycles and GnRHa-HRT-FET cycles were given sequential administrations of estrogen and progesterone, respectively, and 3.75 mg of GnRHa (Diphereline, Ipsen Pty Ltd, France) was injected on the second day of their menstrual cycle. Subsequently, 4–6 mg of estradiol was administered on days 2 and 4 of the menstrual cycle or day 28 after GnRHa administration. The estradiol dosage was increased to 8 mg/day, if necessary, based on endometrial thickness monitoring using vaginal ultrasound. Women receiving progesterone received a daily intramuscular injection of 60 mg of progesterone (XianJu Pharma, China) and daily vaginal administration of 400 mg of progesterone in a soft capsule (Utrogestan, Laboratoires Besins International, France). When the endometrium has a thickness of ≥ 7 mm and a clear triple-line appearance, cleavage- and blastocyst-stage frozen embryos were transferred at 4 and 6 days after progesterone administration, respectively, and progesterone was administered until a pregnancy test was performed. Once pregnancy was achieved, luteal support was maintained until the 10th to 12th weeks of gestation. The primary outcome of our trial was the LBR. The secondary outcomes were implantation rate, ongoing pregnancy rate (OPR), clinical pregnancy rate (CPR), and early miscarriage rates. A biochemical pregnancy was confirmed when the serum β-HCG level was > 25 IU/L by a pregnancy test at 14 days after FET. Clinical pregnancy was defined when at least one gestational sac in the uterine cavity was observed by ultrasound at 4 weeks after FET. Live birth was defined as at least one live-born baby after 24 weeks of gestation. The implantation rate refers to the number of gestational sacs in a clinical pregnancy divided by the total number of transferred embryos. Ongoing pregnancy refers to pregnancy maintenance until ≥ 20 weeks after gestation. Early miscarriage refers to a miscarriage of a clinical pregnancy before 12 weeks. Statistical analysis was performed via the intention-to-treat (ITT) principle, which included all randomized patients enrolled in this study. Subsequently, a per-protocol analysis was performed. The sample size was determined via power analysis, and the sample size was 11.0. At the Center of Reproductive Medicine of the First Affiliated Hospital of Guangxi Medical University, the average LBR in women with RIF in 2018 was 28% per FET, and we assumed an 18% increase after intrauterine infusion with the oil-based medium. The minimum number of each arm was 97 at 80% power, the alpha error was 0.05, and the dropout rate was 10%. Statistical analyses were conducted via the Statistical Package for the Social Sciences 22.0. The Shapiro–Wilk test was used to determine the distribution type of the variables. Continuous variables with a normal distribution are presented as the mean ± standard deviations and a t -test was used. Nonnormally distributed variables are presented as the ideal median (interquartile range IQR 25, 75), and the Mann‒Whitney U test was used. The chi-square test was performed for categorical variables. rate ratio (RR) and 95% confidence intervals were measured for pregnancy outcomes. P  < 0.05 was considered statistically significant. To control for type I error inflation due to multiple comparisons, primary outcome analysis was maintained at α = 0.05; the Bonferroni corrections were used to test the remaining outcomes at α = 0.05/4 = 0.013. Adjusted P -values are reported alongside raw values.