FERTILITY AND STERILITY VOL. 71, NO. 2, FEBRUARY 1999 Copyright 1999 American Society for Reproductive Medicine Published by Elsevier Science Inc. Printed on acid-free paper in U.S.A. Risk factors for ectopic pregnancy in assisted reproduction Annika Strandell, M.D., Jane Thorburn, M.D., Ph.D., and Lars Hamberger, M.D., Ph.D. Department of Obstetrics and Gynecology, Sahlgrenska University Hospital, Göteborg University, Göteborg, Sweden Received May 19, 1998; revised and accepted September 12, 1998. Supported by grants from Göteborg Medical Society, the Hjalmar Svensson Foundation, and the Ordensällskapet W:6 Society, Göteborg, Sweden. Presented at the 10th World Congress on In Vitro Fertilization and Assisted Reproduction, Vancouver, British Columbia, Canada, May 24 28, 1997. Reprint requests: Annika Strandell, M.D., Department of Obstetrics and Gynecology, Sahlgrenska University Hospital, S-413 45 Göteborg, Sweden (FAX: 46-31-82-92-48; E-mail: annika.strandell@medfak.gu.se). 0015-0282/99/$20.00 PII S0015-0282(98)00441-5 Objective: To identify risk factors for ectopic pregnancy (EP) after IVF-ET. Design: Retrospective cohort study. Setting: In vitro fertilization clinic at a university hospital. Patient(s): A total of 725 women who conceived after IVF were studied with regard to background factors, indications for IVF, and factors related to the IVF procedure through review of their medical charts. The rate of EP was 4%, corresponding with 29 EPs, of which 2 were heterotopic. Intervention(s): None. Main Outcome Measure(s): Pregnancy outcome, defined as intrauterine pregnancy or EP. Background factors, indications for IVF, and factors related to the IVF procedure were analyzed for possible correlation with the outcome of EP. Risk factors for EP were identified by logistic regression analysis. Result(s): Tubal factor infertility, various previous abdominal surgeries, previous EP or pelvic infection, presence of a hydrosalpinx or fibroid, and type of transfer catheter used showed a positive correlation with EP as outcome. Logistic regression analysis identified two factors with predictive power: tubal factor infertility and previous myomectomy. Conclusion(s): Tubal factor infertility was the most prominent risk factor for EP after IVF. Previous myomectomy appeared to be another important risk factor, but this is a new finding that needs to be confirmed by further study. (Fertil Steril 1999;71:282 6. 1999 by American Society for Reproductive Medicine.) Key Words: Ectopic pregnancy, in vitro fertilization, logistic regression, myomectomy, risk factors, tubal factor infertility Ectopic pregnancy (EP) is a well-known risk of IVF. The incidence of EP after IVF is generally 5% (1), but in selected groups with tubal factor infertility, it is as high as 11% (2). In the western world, the incidence of EP is approximately 2% in the general population (3 5) but as high as 20% (6, 7) in patients who have undergone tubal surgery. Risk factors for EP after natural conception (e.g., tubal surgery, previous EP, pelvic inflammatory disease, and infertility) have been well described (8, 9). Patients who undergo IVF often have a high number of these risk factors and also are at high risk for EP after natural conception; however, it is not clear that these risk factors have the same effect after treatment with IVF. In addition to traditional risk factors, other factors related to both technical and qualitative aspects of the IVF procedure (e.g., stimulation protocols, endometrial and ovarian response, embryo quality, transfer technique, number of embryos transferred, and use of luteal support) must be taken into consideration. Another well-known consequence of IVF is an increased risk of heterotopic pregnancy (10). For this problem, early detection and management is of the utmost importance. It is important to identify risk factors to provide patients with adequate information, to rule out an EP or to diagnose and treat an EP early, and possibly, to develop preventive strategies. The aim of the present study was to identify risk factors for EP among patients who conceive through IVF. MATERIALS AND METHODS From 1991 1995, 3,019 ETs were performed at the IVF Unit, Department of Obstetrics and Gynecology, Sahlgrenska University 282
Hospital, Göteborg, Sweden. The first cycle that resulted in a pregnancy for each patient was studied (n 739). Fourteen biochemical pregnancies were excluded. Among the remaining 725 pregnancies, 29 were ectopic, of which 2 were heterotopic, yielding an incidence of EP of 4%. A total of 696 pregnancies were intrauterine, of which 22.3% ended in miscarriage. The patients medical records were reviewed. Approval from our institutional review board was not necessary because of the retrospective design of the study. Patients who conceived were studied with regard to background factors, indications for IVF, and factors related to the IVF procedure (Table 1). The women underwent down-regulation with a GnRH agonist (buserelin acetate, Suprefact; Hoechst AG, Frankfurt, Germany) starting in the follicular phase at a dosage of 1.2 mg intranasally for 3 5 weeks and subsequently were stimulated with gonadotropins (hmg, Pergonal; Serono, Geneva, Switzerland and/or FSH, Fertinorm HP; Serono), which were prescribed individually (dosages usually varied from 150 300 U/d). The cycles were monitored with serum E 2 analyses and ultrasound examinations. Ovulation was induced with 10,000 U of hcg (Profasi; Serono) when three follicles had reached a diameter of 18 mm. The oocytes were collected and fertilized according to routine procedures with either conventional IVF or intracytoplasmic sperm injection (ICSI). The embryos were evaluated for fragmentation on a four-grade scale, and only grades I II generally were used for transfers (11). Embryo transfer took place 2 3 days after oocyte retrieval. At the beginning of the study, a maximum of three embryos were transferred as a general policy. During the last 2 years of the study, only two embryos were transferred routinely. There also was a shift in catheter use. The Frydman TDT catheter (SET T.D.T.; Prodimed; Frydman TDT, Neuilly-en-Thelle, France) was used most frequently at the beginning of the study, whereas the Edwards-Wallace embryo replacement catheter (Simcare Manufacturing Ltd.; Wallace, Colchester, United Kingdom) was used most frequently during the last year of the study. For luteal support, hcg (Profasi; Serono) was administered subcutaneously; in patients at risk for hyperstimulation, progesterone-in-oil ex tempore was administered intramuscularly. Transfers of frozen and thawed embryos were performed in natural cycles. Pregnancies were checked with ultrasound examinations at least twice; the first examination was performed 4 weeks after ET. In case of pathologic findings, individual follow-up was undertaken. Cycles that resulted in an EP (n 29) were compared with cycles that resulted in an intrauterine pregnancy (n 696). Preselection of the variables listed in Table 1 was undertaken with the use of bivariate analysis. A t-test was TABLE 1 Distribution of factors analyzed for possible correlation with risk of ectopic pregnancy in 725 women who conceived after IVF. Variable Finding Age* (y) 32.5 (22 40) Duration of infertility* (y) 6.5 (1 20) Background factor Obstetric history Previous EP 177 (24) Previous delivery 109 (15) Previous abortion 211 (29) Previous salpingitis 123 (23) Endometriosis 137 (19) Previous abdominal surgery 438 (61) Salpingectomy Unilateral 112 (15) Bilateral 25 (3) Total 137 (19) Ovarian surgery 106 (15) Myomectomy 20 (3) Fertility surgery 272 (38) Sterilization 9 (1) Presence of hydrosalpinx Unilateral 48 (7) Bilateral 28 (4) Total 76 (11) Presence of fibroid 38 (5) Indication for IVF treatment Tubal factor 387 (54) Endometriosis 43 (6) Unexplained 115 (16) Male factor 133 (18) Hormonal 21 (3) Mixed 24 (3) Other 2 ( 1) Factors related to the IVF procedure Standard IVF/ICSI 551/174 (76/24) Fresh transfers/frozen thawed transfers 659/66 (91/9) Endometrial maturation Thickness 8 mm 615/9 (99/1) Triple layer 531/67 (89/11) No. of oocytes retrieved* 12.6 (1 32) Embryo quality (grade I II/III IV) 688/29 (96/4) Transfer day 2/3 in fresh cycles 633/26 (96/4) No. of embryos transferred* 2.4 (1 4) Transfer catheter used (Frydman TDT/ Edwards-Wallace) 657/47 (93/7) Luteal support used (hcg/progesterone/both) 367/235/41 (57/37/6) Note: EP ectopic pregnancy, hcg human chorionic gonadotropin; ICSI intracytoplasmic sperm injection. * Values are means, with range in parentheses. Values are actual no. of patients, with percentage in parentheses. used for comparison of means, and Fisher s exact test was used for comparison of proportions between groups. A P value of.05 was considered statistically significant, but variables with a P value of.10 were chosen for inclusion in the subsequent regression analysis. FERTILITY & STERILITY 283
TABLE 2 Comparison of rates of ectopic pregnancy in patients with and without infertility risk factors. Risk factor No. of women with EP/no. with indicated risk factor (%) No. of women with EP/no. without indicated risk factor (%) P value* Tubal factor infertility 27/387 (7.0) 2/338 (0.6).000 Standard IVF versus ICSI 29/551 (5.3) 0/174.001 Previous ectopic pregnancy 14/177 (7.9) 15/548 (2.7).004 Previous fertility surgery 18/272 (6.6) 11/451 (2.4).010 Presence of hydrosalpinx 7/76 (9.2) 22/648 (3.4).025 Previous abdominal surgery 24/438 (5.5) 5/284 (1.8).033 Previous myomectomy 3/20 (15.0) 26/705 (3.7).041 Previous salpingitis 8/123 (6.5) 10/406 (2.5).043 Presence of fibroid at transfer 4/38 (10.5) 25/684 (3.7).060 Explained versus unexplained infertility 28/610 (4.6) 1/115 (0.9).069 Unilateral salpingectomy 8/112 (7.1) 21/611 (3.4).071 Absence of progesterone as luteal support 17/364 (4.7) 8/269 (3.0).309 Transfer day 2 versus day 3 26/633 (4.1) 0/26.647 Transfer catheter Frydman TDT versus Edwards-Wallace 26/657 (4.0) 1/47 (2.1).716 Note: EP ectopic pregnancy; ICSI intracytoplasmic sperm injection. * Determined by Fisher s exact test. Stepwise logistic regression analysis was used to identify prognostic variables among the preselected variables. For model discrimination, the c statistic was calculated (12). The c statistic can be interpreted as the proportion of pairs of cases with different observed outcomes in which the model results in a higher probability for the cases with the event than for the cases without the event. The c statistic ranges in value from 0.5 1.0. A value of 0.5 means that the model is no better than a random guess for assigning cases to groups. A value of 1.0 means that the model always assigns higher probabilities to cases with the event than to cases without the event. Statistical analyses were performed with the SPSS software system (SPSS Inc., Chicago, IL). Attributable risks for prognostic variables were calculated. RESULTS The most frequent site of an EP was in the ampullar part of the tube (21/29). Two EPs were located in the isthmic part and 1 in the pouch of Douglas. Three intramural pregnancies occurred in patients who previously had undergone salpingectomy. Among the 29 EPs that occurred, 2 were heterotopic. Both the affected patients were treated with salpingectomy; one of the remaining intrauterine pregnancies progressed to full term and the other ended in a miscarriage. Two patients did not undergo laparoscopy, and the exact sites of their EPs were therefore unclear. Among the 29 patients with EPs, 14 had had a previous EP. Background factors, indications for IVF, and factors related to the IVF procedure (Table 1) were analyzed for possible correlation with the risk of EP. Their frequency distribution, mean, and range, when applicable, also are shown in Table 1. Factors that correlated with the risk of EP or were of theoretic interest as a risk factor are listed in Table 2. Of those factors that correlated with the risk of EP, two were selected by stepwise logistic regression analysis as having predictive power: tubal factor infertility and previous myomectomy (Table 3). The probability of EP can be calculated according to the following equation: L 1n(P/(1 P)) b b 1 tubal factor infertility b 2 previous myomectomy. The values of the coefficients b, b 1, and b 2 can be seen in Table 3. The variables of tubal factor infertility and previous myomectomy are coded as one or zero. If a patient has tubal factor infertility and has undergone myomectomy, the following equation can be used: L 5.23 2.56 1 1.69 1 0.98 With the use of this equation, L 1n(P/(1 P)), the probability of EP can be estimated at 27.2%. The four different combinations of the predictive factors and their probabilities are shown in Table 4, which can be used to calculate the risk of EP for every patient. To measure the ability of the model to discriminate between an EP and an intrauterine pregnancy, the c statistic was calculated and attained a value of 0.73. Among the 387 cycles performed in patients with tubal factor infertility, the risk of EP attributable to tubal factor infertility was 91%. In the group in which both risk factors were present (n 10), the attributable risk increased to 98%. In the total population of 725 cycles, the risk of EP attributable to both tubal factor infertility and previous myomectomy was 85%. 284 Strandell et al. Risk of ectopic pregnancy after IVF Vol. 71, No. 2, February 1999
TABLE 3 Results of stepwise logistic regression analysis to assess the risk of subsequent ectopic pregnancy. Variable Regression coefficient b* Among the factors that correlated with EP but were not found to have predictive power in the multivariate analysis, previous salpingectomy was studied further. A higher rate of EP was observed among women who had undergone unilateral salpingectomy (8/112; 7.1%) than among women who had undergone bilateral salpingectomy (1/25; 4%) or who had not undergone salpingectomy (20/568, 3.5%). With regard to the day of transfer in fresh cycles, no EPs occurred after transfers made on day 3 after oocyte retrieval. DISCUSSION Standard error se(b) P value Constant 5.23 0.72 Tubal factor infertility (b 1 ) 2.56 0.74.0005 Previous myomectomy (b 2 ) 1.69 0.70.015 * Used for calculation of the odds ratio. Used for calculation of confidence intervals. The 4% incidence of EP found in the present study is similar to that found in other reports (6, 13, 14). The rate of heterotopic pregnancy was lower (0.3%) than that reported by other investigators (1%) (10, 15). Although the patient sample is too small to allow any definite conclusions to be drawn, the low numbers of embryos transferred could be part of the explanation. Previous studies of risk factors for EP after IVF primarily have used bivariate correlation analyses, which do not take into consideration interactions and confounding factors. The clinical importance of each factor therefore can be underestimated or overestimated for a given individual. The advantage of using logistic regression analysis is that every factor is evaluated and weighted in relation to the other. Tubal factor infertility was identified as the main risk TABLE 4 Probability of ectopic pregnancy with the use of two predictive factors identified by logistic regression analysis. Tubal factor infertility Previous myomectomy L Probability of ectopic pregnancy* Observed proportion Yes Yes 0.98 27.2% 30.0% (3/10) Yes No 2.67 6.4% 6.4% (24/377) No Yes 3.55 2.8% 0 (0/10) No No 5.24 0.5% 0.6% (2/328) * Calculated according to the logistic regression model: L 1n(P/(1 P)) b b 1 tubal infertility b 2 previous myomectomy. factor for EP after IVF in this study. Several other reports (2, 13, 16) also have demonstrated a positive correlation between tubal factor infertility and EP. In these studies, the investigators reported rates of EP in the subgroup of women with tubal factor infertility similar to or somewhat higher (7% 11%) than our rate of EP (7%). In agreement with our results, a correlation between pelvic inflammatory disease and prior reconstructive tubal surgery and subsequent EP has been described previously (14, 17). It is obvious that risk factors that have a positive correlation with EP, such as previous fertility surgery, abdominal surgery, or EP, and pelvic inflammatory disease and the presence of a hydrosalpinx, coexist with tubal factor infertility. Consequently, it is not surprising that the logistic model in our study rejected all but tubal factor infertility as predictive. Although the rate of EP in patients with tubal factor infertility is higher after tubal surgery than after IVF (6, 7), it is in the same range after laparoscopic surgery and after IVF (6.5%) (18). In 1976, Steptoe and Edwards (19) suggested that there was no risk of EP when the oviducts were completely absent. We found, however, that all three intramural pregnancies in our study occurred in patients who previously had undergone salpingectomy. This finding indicates that in the case of EP, there is a risk that the pregnancy will be in a location that is more difficult to manage. It has been suggested that unilateral salpingectomy may enhance the risk of EP on the contralateral side (20); our findings tend to support this possibility (rate of EP in patients with unilateral salpingectomy 7.1%; rate of EP patients without salpingectomy or with bilateral salpingectomy 3.4%). In contrast with tubal factor infertility, male factor infertility (designated as ICSI in Table 2) was found to have a protective effect. In our study, ICSI corresponded with male factor infertility in 94% of the patients (21). Since the ICSI technique was introduced, the proportion of IVF cycles performed for male factor infertility has been increasing, and this should lead to a further decrease in the overall rate of EP after IVF. The influence of fibroids on IVF outcome has not been evaluated previously. Any pelvic surgery can contribute to tubal factor infertility, but all patients with a previous myomectomy and tubal factor infertility in our series had tubal factor infertility diagnosed before the myomectomy. It is still unknown whether there is a causal relation between previous myomectomy and an increased risk of EP after IVF, but the myometrial scar may disturb the normal uterine contractility or prevent an intrauterine implantation. Further, the positive correlation between the presence of a fibroid at ET and the subsequent development of an EP might strengthen that theory. According to the calculated probabilities of EP in Table 4, when previous myomectomy was the only risk factor FERTILITY & STERILITY 285
present, the risk of EP (2.8%) was not increased compared with that of the entire study population (4%). It was the combination of the two predictive factors that yielded the high risk of almost 30%. However, our identification of previous myomectomy as a risk factor for EP after IVF could be a statistical type I error; further studies are necessary to rule out this possibility. Factors related to the technical and qualitative aspects of the IVF procedure have been investigated in several studies, and no convincing evidence of any correlation has been found. In agreement with other reports, we also did not find a correlation between EP and any of the following factors: number of oocytes retrieved, embryo quality, number of embryos transferred, and use of luteal support (13, 14, 16, 17). In addition, we analyzed fresh versus frozen ET, endometrial maturation, day of ET, and type of transfer catheter used and did not detect significant correlation. It is of note that no EPs occurred after day 3 ET, and only a single EP occurred after the use of a soft catheter. If progesterone protects against EP, as has been hypothesized, this should be detectable in regimens that include progesterone for early luteal support and/or ET on day 3 or later, when progesterone levels are higher. Our data, which show a slight reduction in the incidence of EP with the use of progesterone, can neither support nor reject this theory. From a clinical point of view, it seems plausible that aspects of the transfer technique, such as injection pressure, injected volume, and location of the catheter tip at the time of embryo placement, could have an effect on the tubal implantation rate. This hypothesis was supported by Yovich et al. (22), who reported that a technique that involved deep fundal transfer was associated with a higher risk of EP than a technique that involved midcavity embryo placement. Another study that supports this hypothesis was reported by Knutzen et al. (23), who performed a mock transfer and demonstrated tubal reflux in 38.2% of the patients. It also was demonstrated that the injected volume might have an effect on the site of implantation in a study in which a larger volume increased the risk of EP (14). The results of this study indicate that tubal factor infertility is the most important risk factor for EP after IVF. Previous myomectomy might be an additive risk factor that has not been described previously, but this finding needs to be confirmed by expanded studies, which are ongoing. References 1. World collaborative report. Proceedings of the Xth World Congress on In Vitro Fertilization and Assisted Reproduction. J Assisted Reprod Genet 1997;14. 2. Dubuisson J, Aubriot F, Mathieu L, Foulot H, Mandelbrot L, Bouquet de Jolinière J. Risk factors for ectopic pregnancy in 556 pregnancies after in vitro fertilization: implications for preventive management. Fertil Steril 1991;56:686 90. 3. Ectopic pregnancy: United States. MMWR Morb Mortal Wkly Rep 1990;39:401 4. 4. Coste J, Job-Spira N, Aublet-Cuvelier B, Germain E, Glowaczower E, Fernandez H, et al. Incidence of ectopic pregnancy. First result of a population-based register in France. Hum Reprod 1994;9:742 5. 5. Thorburn J. Is the epidemic of ectopic pregnancy over in Sweden? A report on present incidence and mortality. Lakartidningen 1995;92: 4701 6. 6. Tomazevic T, Ribic-Pucelj M. Ectopic pregnancy following the treatment of tubal infertility. J Reprod Med 1992;37:611 4. 7. Strandell A, Bryman I, Janson PO, Thorburn J. Background factors and scoring systems in relation to pregnancy outcome after fertility surgery. Acta Obstet Gynecol Scand 1995;74:281 7. 8. Thorburn J, Berntsson C, Philipsson M, Lindblom B. Background factors of ectopic pregnancy. Frequency distribution in a case-control study. Eur J Obstet Gynecol Reprod Biol 1986;23:321 31. 9. Tuomivaara L, Kauppila A. Ectopic pregnancy: a case-control study of aetiological risk factors. Arch Gynecol Obstet 1988;243:5 11. 10. Tal J, Haddad S, Gordon N, Timor-Tritsch I. Heterotopic pregnancy after ovulation induction and assisted reproductive technologies: a literature review from 1971 to 1993. Fertil Steril 1996;66:1 12. 11. Veech LL. Oocyte assessment and biological performance. Ann N Y Acad Sci 1988;541:259 74. 12. Hanley D, McNeil B. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29 36. 13. Verhulst G, Camus M, Bollen N, Van Steiterghem A, Devroey P. Analysis of risk factors with regard to the occurrence of ectopic pregnancy after medically assisted procreation. Hum Reprod 1993;8: 1284 7. 14. Marcus S, Brinsden P. Analysis of the incidence and risk factors associated with ectopic pregnancy following in-vitro fertilization and embryo transfer. Hum Reprod 1995;10:199 203. 15. Svare J, Norup P, Grove Thomsen S, Hornnes P, Maigaard S, Helm P, et al. Heterotopic pregnancies after in vitro fertilization and embryo transfer a Danish survey. Hum Reprod 1993;8:116 8. 16. Herman A, Ron-El R, Golan A, Weinraub Z, Bukovsky I, Caspi E. The role of tubal pathology and other parameters in ectopic pregnancies occurring in in vitro fertilization and embryo transfer. Fertil Steril 1990;54:864 8. 17. Zouves C, Erenus M, Gomel V. Tubal ectopic pregnancy after in vitro fertilization and embryo transfer: a role for proximal occlusion or salpingectomy after failed distal tubal surgery? Fertil Steril 1991;56: 691 5. 18. Audibert F, Hédon B, Arnal F, Humeau C, Boulot P, Bachelard B, et al. Therapeutic strategies in tubal infertility with distal pathology. Hum Reprod 1991;6:1439 42. 19. Steptoe PC, Edwards RG. Reimplantation of a human embryo with subsequent tubal pregnancy. Lancet 1976;1:880 2. 20. Cohen J, Mayaux MJ, Guihard-Moscato ML, Schwartz D. In-vitro fertilization and embryo transfer: a collaborative study of 1163 pregnancies on the incidence and risk factors of ectopic pregnancies. Hum Reprod 1986;1:255 8. 21. Wennerholm UB. Obstetric and prenatal outcome of pregnancies following intracytoplasmic sperm injection. Hum Reprod 1998;11: 1113 9. 22. Yovich JL, Turner S, Murphy A. Embryo transfer technique as a cause of ectopic pregnancies in in vitro fertilization. Fertil Steril 1985;44: 318 21. 23. Knutzen V, Stratton CJ, Shee G, McNamee PI, Huang TT, Soto-Albors C. Mock embryo transfer in early luteal phase, the cycle before in vitro fertilization and embryo transfer: a descriptive study. Fertil Steril 1992;57:156 62. 286 Strandell et al. Risk of ectopic pregnancy after IVF Vol. 71, No. 2, February 1999