Impact of salpingectomy on the risk and characteristics of ectopic pregnancy after IVF/ICSI in patients with ectopic pregnancy history: A large retrospective cohort study.

Xueyi Jiang: Acquisition of data, analysis and interpretation of data, drafting of the manuscript, statistical analysis. Shutian Jiang: Study concept and design, acquisition of data, analysis and interpretation of data, drafting of the manuscript, study supervision. Hongyuan Gao: Acquisition of data. Jing Ye: Acquisition of data, analysis and interpretation of data. Yanping Kuang: Study concept and design, drafting of the manuscript, study supervision. All authors: Critical revision of the manuscript for important intellectual content.

This study was ethically reviewed by the Shanghai Ninth People's Hospital institution (SH9H‐2022‐TK109‐8) on June 6, 2022. The procedures used in this study adhere to the tenets of the Declaration of Helsinki. Written informed consent was obtained from all participants.

Science and Technology Commission of Shanghai Municipality (22Y21900800); National Natural Science Foundation of China (82201886 and 82271732).

A total of 43 647 eligible patients were included between January 2013 and June 2022. After propensity score matching, there were 3252 participants in the no previous EP history cohort (NEP cohort), 3252 in the one EP history cohort (One‐EP cohort), and 1571 in the 2 or more EP history cohort (> = 2 EP cohort). Among all of the cycles, 394 were fresh and 7681 were frozen. The baseline characteristics of the NEP cohort, One‐EP cohort and > =2 EP cohort are shown in Table  1 . These three cohorts were comparatively balanced after matching (Figure  S2 ), but the duration of infertility, cycles using high‐quality embryos, and the year of treatment remained statistically different. The cycle characteristics and pregnancy outcomes of the first cycles are shown in Table  2 . The rates of overall IVF‐EP were significantly higher in patients with EP history (NEP cohort: 1.3% (43/3252) vs. One‐EP cohort: 2.2% (72/3252) vs. > =2 EP cohort: 2.0% (31/1571), p  = 0.02. All the results below are in the same order). The risk ratio (RR) of One‐EP cohort to NEP cohort was 1.343 (1.059, 1.704), and RR of > = 2 EP cohort to NEP cohort was 1.163 (0.957, 1.412). Similarly, the rates of non‐tubal IVF‐EP were higher in One‐EP cohort and > =2 EP cohort compared to NEP cohort (0.1% (2/3252) vs. 0.5% (15/3252) vs. 0.6% (10/1571), p  = 0.001). The percentage of non‐tubal IVF‐EP to IVF‐EP increased significantly as the number of previous EP increased (4.7% (2/43) vs. 20.8% (15/72) vs. 32.3% (10/31), p  = 0.008). The biochemical pregnancy rate was lower in patients with EP history (55.5% vs. 52.3% vs. 52.7%, p  = 0.02), with a higher singleton birth rate in > = 2 EP cohort (73.2% vs. 74.7% vs. 80.4%, p  = 0.003). No differences in other pregnancy outcomes were detected. Baseline characteristics of all patients. Note : a, b, c refer to the result of Post‐hoc‐comparison of Chi‐square test (Bonferroni test), the same letter means no statistical difference ( p  > 0.05/3); *, p  < 0.05. Cycle characteristics and pregnancy outcomes of the first IVF/ICSI cycles of patients. Note : a, b, c refer to the result of Post‐hoc‐comparison of Chi‐square test (Bonferroni test), the same letter means no statistical difference ( p  > 0.05/3); Non‐tubal ectopic pregnancy / ectopic pregnancy refers to the ratio of non‐tubal ectopic pregnancy to ectopic pregnancy; *, p  < 0.05. Subcohort analysis was designed to examine the effect of tubal status on the incidence of IVF‐EP in patients with EP history (Table  3 ). In One‐EP cohort, there were 1457 patients in the unresected cohort, 1680 in the unilateral salpingectomy cohort, and 115 in the bilateral salpingectomy cohort. There were no statistically significant differences in overall IVF‐EP rate among the three subcohorts (unresected cohort: 2.1% (31/1457) vs. unilateral salpingectomy cohort: 2.3% (38/1680) vs. bilateral salpingectomy cohort: 2.6% (3/115), p  = 0.93). In > = 2 EP cohort, there were 342, 591, and 638 patients in each subcohort. There were also no significant differences in overall IVF‐EP and non‐tubal IVF‐EP rate among the three subcohorts (IVF‐EP: 2.0% (7/342) vs. 2.5% (15/591) vs. 1.4% (9/638), p  = 0.36; non‐tubal IVF‐EP: 0.0% (0/342) vs. 0.0% (1/591) vs. 0.6% (4/638), p  = 0.12). Unilateral salpingectomy and bilateral salpingectomy subcohorts had lower rates of implantation ( p  = 0.02), biochemical pregnancy ( p  = 0.03), clinical pregnancy ( p  = 0.04) and live birth ( p  = 0.04) than unresected subcohort in > = 2 EP cohort. The baseline data for three subcohorts are shown in Table  S1 . Cycle characteristics and pregnancy outcomes of three subcohorts (sorted by tubal status) in patients with a history of EP. Note : a, b refer to the result of Post‐hoc‐comparison of Chi‐square test (Bonferroni test), the same letter means no statistical difference ( p  > 0.05/3); *, p  < 0.05. There was no statistical difference in Number of embryos transferred, FET rate, Cleavage stage transfer rate, Endometrial thickness, Embryo quality, and Year of treatment in both One‐EP cohort and > =2 EP cohort. (see details in Figure  S1 ). The effect of different influencing factors on IVF‐EP is shown in Table  4 . The risk of IVF‐EP increased with increasing number of embryos transferred (aOR [95% CI]: 1.65 [1.26, 2.17]) and number of previous EP (aOR [95% CI] 1.85 [1.42, 2.43]), while it decreased with increasing endometrial thickness (aOR [95% CI]: 0.92 [0.89, 0.96]) and age (aOR [95% CI]: 0.95 [0.93, 0.97]). The transfer of blastocysts (as opposed to cleavage‐stage embryos) reduced the risk of EP in the overall population (aOR [95% CI]: 0.51 [0.35, 0.76]) and in the NEP cohort (aOR [95% CI]: 0.50 [0.32, 0.78]). Influencing factors associated with EP by logistic regression analysis in all IVF/ICSI cycles (before matching). Note : Factors included: age, BMI, primary infertility (refer to secondary infertility), pluriparous, duration of infertility, year of treatment, embryo quality, number of embryos transferred, embryo type, transfer protocol, endometrial thickness, number of previous ectopic pregnancy, fallopian tubal status and fallopian tubal status*number of previous EP. Factors not displayed in the table were not statistically significant in each cohort and the full population (before matching). *, p  < 0.05; Only patients with a history of previous EP participated in the results of the interaction term(Fallopian tubal status*Number of previous EP) as the coefficients could not be calculated when the number of previous EP = 0. Abbreviations: aOR, adjusted odds ratio; CI, confidence interval; EP, ectopic pregnancy. The cutoff value was 27.50, 29.50, 30.50, and 29.50 for each cohort and all patients respectively. The sensitivity/1‐specificity of patient age cutoff value was 0.256/0.132, 0.458/0.273, 0.548/0.364, and 0.438/0.279 respectively; Area under ROC was 0.562 ( p  = 0.160), 0.600 ( p  = 0.004), 0.640 (p = 0.008), and 0.597 ( p  < 0.001) respectively. In addition, because of a potential interaction between tubal status and EP history, fallopian tubal status*number of previous EP was defined as an interaction term and was introduced into the logistic regression analysis. Unilateral and bilateral salpingectomy could both reduce the risk of IVF‐EP when the number of previous EP was also taken into account (aOR [95% CI]: 0.20 [0.14, 0.29]; aOR [95% CI]: 0.21 [0.14, 0.33]). Further investigations concerning the effect of this interaction effect on non‐tubal IVF‐EP in all populations is presented in Table  5 . It was found that both the increased number of previous EP and bilateral salpingectomy would raise the incidence of non‐tubal IVF‐EP. Notably, these findings held true in patients without tubal resection, that is, in patients with unresected tubes, the non‐tubal IVF‐EP rate increased with increasing number of previous EP (aOR [95% CI]: 3.02 [1.04, 8.81]). The impact of fallopian tubal status and number of previous EP on non‐tubal ectopic pregnancy outcomes in all IVF/ICSI cycles (before matching). Note : * p  < 0.05; aOR was adjusted by multifactorial regression, factors included: age, BMI, primary infertility (refer to secondary infertility), pluriparous, duration of infertility, year of treatment, embryo quality, number of embryos transferred, embryo type, transfer protocol, endometrial thickness, number of previous ectopic pregnancy, fallopian tubal status and fallopian tubal status*number of previous EP. Abbreviations: aOR, adjusted odds ratio; EP, ectopic pregnancy; CI, confidence interval.

Our study showed that previous EP history had an obvious impact on the occurrence of IVF‐EP and non‐tubal IVF‐EP, and had no significant influences on other pregnancy outcomes, including the rate of implantation, clinical pregnancy, or live birth. The subcohort analysis combined with a multifactorial binary regression analysis showed that salpingectomy can reduce the risk of IVF‐EP when the number of previous EPs was taken into account, so salpingectomy may be a preferred treatment for EP to avoid recurrent IVF‐EP in patients with EP history. In addition, both EP history and bilateral salpingectomy led to an increased risk of non‐tubal IVF‐EP. Notably, even for patients with bilateral tubes preserved, the non‐tubal IVF‐EP rate was higher in those with EP history; therefore, increased attention needs to be given to the possibility of non‐tubal IVF‐EP in this population. Only one previous study mentioned the impact of tubal status (blocked/removed tube or not) on the risk of tubal EP after IVF in patients with EP history, but the number of tubal EP cases was too small to assess the conclusion. 8 Due to the differences in tubal status after the previous EP, we analyzed the effect of salpingectomy in subcohorts with different numbers of previous EP. Theoretically, since the proportion of recurrent EP patients (> = 2 EP) in the bilateral salpingectomy cohort was higher, the probability of IVF‐EP in this cohort should also be higher. However, our data showed no statistical difference in IVF‐EP rate among the three subcohorts in > = 2 EP cohort, suggesting that bilateral salpingectomy has a protective effect by reducing the incidence of IVF‐EP in this population. Therefore, we might recommend salpingectomy for recurrent EP. 21 This result was confirmed in the subsequent multifactorial binary regression analysis in all population, but no definitive conclusions were obtained in single‐cohort regressions, so caution was warranted. Additionally, the inferior outcome of the remaining pregnancies in the bilateral salpingectomy subsohorts may be attributed to the impact of salpingectomy on ovarian blood supply and subsequent ovarian reserve and follicular development. 22 , 23 , 24 Besides, it was worth noting that there were still several cases of tubal EP in the bilateral salpingectomy subcohort (as shown in Table  S2 ), which should not occur on the same side after salpingectomy. Possible explanations might be that the previous resection was not thoroughly (to the cornu of the uterus) and the embryos implanted in the remaining tube. Prior studies examining the effect of the number of previous EP on IVF‐EP were scarce and contradictory. Spandorfer et al. proposed a positive correlation between the recurrence of IVF‐EP and the number of previous EP 10 ; however, Zhang et al. suggested that women with recurrent EP are at a lower risk of IVF‐EP compared with women with a single EP. 11 The IVF‐EP rate was slightly higher in One‐EP cohort in our study, but there was no statistically significant difference in the rates of IVF‐EP between One‐EP cohort and > =2 EP cohort, which may be due to the confounding effect of differences in tubal status after previous EP. We therefore performed a multifactorial regression, which showed that the number of previous EP and tubal status were both influencing factors for IVF‐EP, and there was an interaction effect between the two variables. After adjusting the interaction effect, we found that compared to patients with no EP history, an increased number of previous EP was associated with a higher IVF‐EP rate, while salpingectomy was also associated with IVF‐EP rate. Since a large proportion of patients with EP history underwent salpingectomy, the failure to exclude this interaction effect in the design of previous studies has led to different conclusions. Considering that a proportion of patients underwent protective salpingectomy due to other problems of the fallopian tubes (see Figure  S3 ), we reran the multifactorial regression after deleting salpingectomy cycles in the NEP cohort and bilateral salpingectomy cycles in the One‐EP cohort (see Table  S3 ), and the results showed that the interaction term was no longer statistically significant. However, the number of previous EP was still a risk factor and the bilateral salpingectomy was still a protective factor against IVF‐EP (compared to unilateral salpingectomy). This is consistent with the conclusions we have now obtained and suggests that our previous results obtained without excluding these two populations are also convincing. Non‐tubal ectopic pregnancies were rare in spontaneous pregnancies, accounting for only 5% of EP, with an incidence of about 0.05% 25 ; while the incidence in IVF has not been reported. In our study, we demonstrated a significantly higher rate of non‐tubal IVF‐EP in women with EP history than in those without EP history. No previous study has discussed in depth the effect of EP history on non‐tubal IVF‐EP. For the first time, we found that the ratio of non‐tubal IVF‐EP to IVF‐EP increased with the number of previous EP, which suggested that the increased risk of non‐tubal IVF‐EP could not simply be attributed to the fact that it was a special type of EP. Interestingly, in patients without tubal resection, an increase in the number of previous EP still increased the probability of non‐tubal EP, suggesting the increased risk was not just due to the inability to have a tubal pregnancy in patients with bilateral salpingectomy. Instead, there may be some underlying factors contributing to the increased risk of non‐tubal EP in patients with recurrent previous EP, such as endometrial receptivity problems, and abnormal signal transduction. 26 , 27 , 28 Therefore, for patients with EP history (even with no tubal resections), more attention should be given to the location of the gestational sac during early pregnancy ultrasound to prevent misdiagnosis of non‐tubal IVF‐EP. 29 , 30 We also tried to find a way to protect high‐risk patients with EP history from recurrent IVF‐EP. Several previous studies have shown that transferring blastocysts can reduce the incidence of EP. 31 , 32 , 33 , 34 However, we found that blastocyst transfer, which reduces the incidence of IVF‐EP in the general population, did not work in patients with EP history, suggesting that recurrent EP may not be an indication for blastocyst transfer. The number of embryos transferred, endothelial thickness, and maternal age may also influence the recurrent IVF‐EP rate and require attention in subsequent clinical practice. Overall, the study had a large sample size and was well‐designed. A total of 43 647 patients were enrolled in our study. To eliminate confounding factors, we included only the first IVF cycle at our center and strictly matched the three cohorts by propensity score matching. Interaction between number of previous EP and tubal status was innovatively taken into account to avoid confounding effects. We innovatively studied tubal status at the time of IVF rather than the management of previous EP as an independent influencing factor of IVF‐EP. This approach can rule out the possibility that some patients with recurrent EP have changes in tubal status after receiving several different treatments or that some patients undergo further surgical tubal treatment before IVF/ICSI. Moreover, tubal status at the time of IVF/ICSI is a more desirable and intuitive variable for clinical assisted reproduction doctors. On the whole, research on the effect of tubal status on the risk of IVF‐EP may in turn provide decisive evidence for gynecology when choosing management for EP. In addition, we did more in‐depth research on non‐tubal EP, not just as a part of EP. Admittedly, several limitations of our study should be discussed. Since the salpingectomy was not performed at our hospital, there may be information bias due to patient reports and medical records, and it was not possible to identify the exact extent of the salpingectomy. Patients with incomplete resections may exist in all salpingectomy subcohorts, and the inability to distinguish such populations resulted in bias. In addition, although matching was performed, some of the baseline variables remained significantly different. Considering that patients with or without EP history differed significantly in tubal status due to previous EP treatment, we did not match the cause of infertility or tubal status for each cohort. Instead, we carried out a multifactorial regression analysis for further verification. Moreover, this was a single‐center retrospective study with possible selection bias as well as population homogeneity, leading to the possibility that it may not be generalizable to the full population and that a multi‐center randomized controlled trial is needed to strengthen the level of evidence.

This study found that previous EP history was associated with a higher risk of IVF‐EP and non‐tubal IVF‐EP. Salpingectomy reduced the risk of overall IVF‐EP in patients with EP history, while bilateral salpingectomy increased the risk of non‐tubal IVF‐EP. Moreover, blastocyst transfer would lower the risk of EP in patients without EP history, but would not take effect in those with EP history.

Ectopic pregnancy (EP) is a serious clinical gynecological emergency, and EP rupture can cause hypovolemic shock, which is the leading cause of death in early pregnancy and accounts for 10%–15% of all maternal deaths. 1 , 2 The etiology of EP is unknown, but several risk factors may be involved, including pelvic inflammatory disease, pelvic surgery, endometriosis, and most notably, assisted reproductive technology (ART) and history of EP. 3 The incidence of EP in ART has been reported to be approximately 2%–8%, which is four times higher than that in spontaneous conception, 4 , 5 and women with EP history have a 4.7‐fold greater risk of later recurrence compared with the general population. 6 These two risk factors often coexist and interact, since patients with EP history are prone to secondary infertility and thus seek ART treatment due to tubal factors. Therefore, the study of EP resulting from IVF (IVF‐EP) in patients with EP history is highly important. Previous studies have shown that patients with EP history have a higher risk of recurrent IVF‐EP, varying from 1.4 to 3.3 times. 7 , 8 , 9 However, these studies only reported the incidence of IVF‐EP in patients with EP history. Clinical strategies to improve the prognosis of this population remain unclear. Given the increased EP rate in subsequent spontaneous pregnancies in patients with EP history, a significant proportion of patients had two or more spontaneous EP. This kind of patients with multiple repeated spontaneous EP may have some underlying differences in terms of their genetic background, physiology and anatomy. Hence it is necessary to differentiate patients with previous EP according to the number of EP, and to focus on their respective IVF outcomes. Few studies have investigated the effect of the number of previous EP on IVF‐EP, with the only two related studies drawing opposite conclusions. 10 , 11 In addition, it has been suggested that different previous EP treatments do not significantly affect the incidence of IVF‐EP, 9 , 10 , 11 , 12 , 13 which is quite different from the results in spontaneous pregnancies. However, the mixed pattern of tubal treatment after multiple previous EP and the interaction between tubal status and the number of previous EP made the study difficult. We therefore narrowed down our study scope to the effect of tubal status on IVF‐EP in cohorts with different numbers of spontaneous EP, which might gather evidence for gynecologists when choosing EP management. Therefore, we investigated the effect of unilateral/bilateral salpingectomy on the risk and characteristics of IVF‐EP in patients with a history of spontaneous EP, so as to provide recommendations for clinical practice in this population.

The authors report no conflict of interest.

This was a retrospective cohort study conducted at the Department of Assisted Reproduction, Shanghai Ninth People's Hospital, Shanghai JiaoTong University School of Medicine. We included women who underwent their first cycle of IVF/ICSI at our institution between January 2013 and June 2022; incomplete documented cycles were excluded, and only the absolute initial cycle of IVF/ICSI was included for all patients. The data was obtained from medical records, which have been described previously. 14 Patients with a history of spontaneous EP were divided into One‐EP cohort and > =2 EP cohort according to the number of previous EP; patients without EP history were classified into the NEP cohort. The history of previous EP was obtained from patient reports and medical records. Due to the large variability of the three cohorts, NEP cohort, One‐EP cohort and > =2 EP cohort were matched according to age, BMI, primary infertility (refer to secondary infertility), pluriparous, duration of infertility, year of treatment, embryo quality, number of embryos transferred, embryo type, transfer protocol, and endometrial thickness by propensity score matching (caliper value = 0.02) at a 2:2:1 ratio without replacement. Infertility was defined as no pregnancy after 1 year of normal sexual activity without contraception, and the duration of treatment was divided every 2–3 years to minimize the impact of a long study duration. To assess the effect of tubal status on IVF‐EP, patients with spontaneous EP history (One‐EP cohort and > =2 EP cohort, separately) were divided into three subcohorts according to their tubal statuses: the unilateral salpingectomy cohort, bilateral salpingectomy cohort, and unresected cohort. The tubal status was obtained from patient reports and medical records. All salpingectomy patients had paper‐based discharge summaries. Salpingostomy patients were categorized into the unresected cohort. The data processing and cohort divisions are shown in Figure  S1 . ART operation was described in detail in previously published articles. 15 , 16 , 17 , 18 , 19 , 20 The primary outcome was the IVF‐EP rate (including heterotopic pregnancy). The secondary outcome indicators included the non‐tubal IVF‐EP rate, implantation rate, biochemical pregnancy rate, clinical pregnancy rate, ongoing pregnancy rate, live birth rate, and miscarriage rate after IVF/ICSI. EP was defined as a pregnancy in which the embryo was implanted outside the uterine cavity. Heterotopic pregnancy was defined as the coexistence of ectopic and intrauterine pregnancies. Non‐tubal EP was a special kind of EP and defined as a cornual pregnancy, interstitial tubal pregnancy, cervical pregnancy, cesarean scar pregnancy, ovarian pregnancy, or abdominal pregnancy. Biochemical pregnancy was defined as blood HCG >25 U/L on the 14th day after embryo transfer. Clinical pregnancy was defined as a gestational sac observed by transvaginal ultrasound 35 days after embryo transfer. Ongoing pregnancy was defined as a pregnancy that persisted for 20 weeks or more. Live birth was defined as the delivery of one or more live births. Miscarriage was defined as spontaneous abortion or intrauterine demise before 24 weeks of gestational age. All outcome indicators were calculated based on the number of cycles. All the statistical analyses were performed with SPSS 24.0. The normality of the quantitative data was tested by the Kolmogorov–Smirnov and Shapiro–Wilk tests. Normally and nonnormally distributed data are expressed as the mean ± SD and median (first quartile, third quartile), respectively. Qualitative data are expressed as n (%). Student's t‐test was used to compare normally distributed independent samples, while the nonparametric Mann–Whitney U test was used to compare abnormally distributed independent samples. Comparisons of rates were made using either the Chi‐squared test or the Fisher exact test. Bonferroni test was performed to do post‐hoc testing of the comparison of three rates between cohorts as well as subcohorts. For all comparisons, p value <0.05 indicated statistical significance ( p value <0.05/3 indicated statistical significance in Bonferroni test). Multivariable regression was performed to exclude the effect of confounding factors remaining after matching on EP as well as to investigate the effect of the interacting factor.

Figure S1. Figure S2.Figure S3. Table S1. Table S2. Table S3.