Hydrosalpinx and IVF outcome: cumulative results after salpingectomy in a randomized controlled trial*

Human Reproduction Vol.16, No.11 pp. 2403 2410, 2001 Hydrosalpinx and IVF outcome: cumulative results after salpingectomy in a randomized controlled trial* A.Strandell 1,4, A.Lindhard 2, U.Waldenström 3 and J.Thorburn 1 1 Department of Obstetrics and Gynaecology, Sahlgrenska University Hospital, S-413 45 Göteborg, Sweden, 2 The Fertility Clinic, Copenhagen University Hospital, Rigshospitalet, Blegdamsvej 9, DK-2100 Copenhagen, Denmark and 3 IVF Center Falun, Falu Hospital, S-791 82 Falun, Sweden 4 To whom correspondence should be addressed. E-mail: annika.strandell@medfak.gu.se BACKGROUND: A randomized controlled trial of salpingectomy prior to IVF in patients with hydrosalpinges has been conducted in Scandinavia. The results from the first transfer cycle have been published and clearly demonstrated an improved pregnancy outcome after salpingectomy had been performed in patients with hydrosalpinges large enough to be visible on ultrasound. The present article is aimed at analysing the effect of salpingectomy on cumulative birth rate, including all individual transfer cycles. METHODS AND RESULTS: A total of 186 women underwent 452 cycles. Among the 77 women randomized to no surgical intervention, 24 underwent salpingectomy after one or two failed cycles. Cumulative results were analysed by Cox regression, taking into account the number of cycles per patient and the presence of a salpingectomy after a previous transfer. Salpingectomy implied a significant increase in birth rate (hazard ratio 2.1, 95% CI 1.6 3.6, P 0.014). Within the subgroup of patients with ultrasound-visible hydrosalpinges, the birth rate was even higher (hazard ratio 3.8, 95% CI 1.5 9.2, P 0.004). Implantation rate was significantly higher in patients who had undergone salpingectomy (27.2% versus 20.2, P 0.03) and, in the subgroup of patients with ultrasound-visible hydrosalpinges, the difference was even larger (30.3% versus 17.1%, P 0.003). CONCLUSIONS: The results of the cumulative cycles strengthen the recommendation for a laparoscopic salpingectomy prior to IVF in patients with ultrasound-visible hydrosalpinges. Key words: cumulative pregnancy rates/hydrosalpinx/ivf/randomized controlled trial/salpingectomy Introduction Several retrospective reports have demonstrated impaired pregnancy outcomes after IVF in patients with hydrosalpinges. The results have been further compiled in meta-analyses that clearly demonstrated a significant reduction in the probability of pregnancy as well as a significant increase in the probability of spontaneous abortion in the presence of hydrosalpinx (Zeyneloglu et al., 1998; Camus et al., 1999). Tubal factor infertility has been a major cause for treatment since the introduction of IVF, but patients with hydrosalpinges have been identified as a subgroup with significantly impaired pregnancy outcomes compared with patients suffering from *The study group for the multicentre trial and responsible investigator at each site were as follows. Sweden: Annika Strandell, Dept Ob/ Gyn, Sahlgrenska University Hospital, Göteborg; Urban Waldenström, IVF Center Falun, Falu Hospital, Falun; Niklas Simberg, Dept Ob/ Gyn, Uppsala University Hospital, Uppsala; Leif Hägglund, International Fertility Center, Malmö; György Csemiczky, Dept Ob/ Gyn, Karolinska Hospital, Stockholm. Denmark: Anette Lindhard, The Fertility Clinic, Rigshospitalet, University Hospital, Copenhagen. Norway: Vidar von Düring, Dept Ob/Gyn, Trondheim; Kari Flo, Dept Ob/Gyn, University of Tromsø. Iceland: GuPmundur Arason, Dept Ob/Gyn, National University Hospital, Reykjavik. other types of tubal damage (Andersen et al., 1994; Kassabji et al., 1994; Strandell et al., 1994; Vandromme et al., 1995; Fleming and Hull, 1996; Ng et al., 1997). One of the main explanations for the hydrosalpinx-related impairment suggests that the leakage of fluid into the uterine cavity creates an unfavourable endometrial environment for implantation and could also affect embryo development. According to this theory, surgical intervention with salpingectomy should remove any effect of the hydrosalpingeal fluid and restore pregnancy rates. Several retrospective studies have compared the outcome of IVF, with and without salpingectomy, in patients carrying hydrosalpinges (Kassabji et al., 1994; Vandromme et al., 1995; Shelton et al., 1996; Meyer et al., 1997; Murray et al., 1998; Ejdrup Bredkjaer et al., 1999). All have demonstrated a positive effect of salpingectomy on pregnancy rates, although the retrospective designs do not provide evidence for the method s effectiveness. A small randomized pilot study indicated a positive effect of salpingectomy on implantation rates but without statistical significance (Déchaud et al., 1998). A multicentre randomized controlled trial with salpingectomy as the main intervention has been conducted within Scandinavia with the aim to test the hypothesis that removal European Society of Human Reproduction and Embryology 2403

A.Strandell et al. of the hydrosalpinx prior to IVF treatment would improve pregnancy rates. The results from the first cycle (Strandell et al., 1999) clearly showed a benefit of salpingectomy in patients with hydrosalpinges that were large enough to be viewed sonographically (birth rates: 40.0 versus 17.5%, P 0.040) and, in particular, if present bilaterally (55.0 versus 15.8%, P 0.019). The aim of the present study was to examine whether salpingectomy was beneficial in terms of increased birth rates on a cumulative basis, taking into account all transfer cycles. Material and methods A detailed presentation of the study protocol, including a power analysis and the randomization procedures adopted, has been given in the previous publication, in which the first cycle was analysed (Strandell et al., 1999). The most important remarks are represented below. Nine Nordic IVF centres participated in the study with approval from each centre s ethics committee. Study design The physician who referred the patient to IVF made the diagnosis of hydrosalpinx. Hysterosalpingography and/or laparoscopy were used as diagnostic methods. At each centre, patients under the age of 39 years with uni- or bilateral hydrosalpinges were randomized to laparoscopic salpingectomy or no intervention, in a ratio 3:2, prior to their first IVF treatment. Patients in the intervention group underwent a laparoscopic unilateral (n 40) or bilateral (n 63) salpingectomy, depending on whether one or two hydrosalpinges were present. Alternative procedures, like a proximal ligation and fenestration due to the presence of extensive adhesions (n 6), or fertility-promoting procedures like salpingostomy (n 1) or adhesiolysis (n 1), were performed in eight cases. Patients in the non-intervention group started their IVF treatment directly according to routine practices. The stimulation procedures at the individual IVF centres were similar and included a long protocol. A maximum of two embryos were routinely transferred. Two hundred and four patients were recruited of which 185 underwent at least one embryo transfer. The results of the first transfer cycles were analysed as a separate study (Strandell et al., 1999). At that time, one patient in the salpingectomy group had experienced two cycles of poor embryo development and no embryo transfer, but her third cycle succeeded and she was thus included in the cumulative analysis of transfer cycles, giving a sample size of 186 patients. Subsequently, all fresh and frozen thawed transfers were analysed on a cumulative per patient basis. After one or two failed cycles, patients in the control group were permitted to undergo salpingectomy. The legislation and routine practice at the individual centres allowed for three stimulated cycles and subsequent transfers of frozen thawed embryos. The follow-up of all the 204 randomized patients has been completed. Group allocation One hundred and sixteen patients were randomized to salpingectomy and 88 patients to no intervention. Salpingectomy was not performed in five patients randomized to surgery before entering the IVF programme. Two of them underwent surgery after their first transfer cycle one patient due to ectopic pregnancy and the three remaining patients never underwent the surgery at all. In the non-intervention group, one patient had a salpingectomy performed due to infection after her first oocyte retrieval but before transfer, which was postponed and conducted later as a transfer of frozen thawed embryos. The 2404 study protocol restricted the randomization allocation to the first cycle and allowed for a subsequent salpingectomy in the non-intervention group if failure occurred. Among the 77 patients undergoing transfer in the randomization group of non-intervention, 24 patients had in fact undergone a salpingectomy after one or two failed cycles. Of these 24 cases, four were due to ectopic pregnancies, while the remaining surgeries were performed upon the patient s expressed request. The salpingectomies performed were unilateral in nine and bilateral in 15 of these 24 patients. None of the patients have undergone additional cycles after a full-term pregnancy. Figure 1 summarizes the patient flow. Pregnancy definitions Only clinical pregnancies, verified as a gestational sac and visible on ultrasound, were considered for analysis. Implantation rates were calculated as the number of gestational sacs divided by the number of embryos transferred. An ectopic implantation was considered as one implanted embryo. The primary endpoint was birth calculated per patient, based on all her cycles. Secondary endpoints were implantation and pregnancy. Subgroup analysis Patients with bilateral and/or ultrasound-visible hydrosalpinges were included in subgroups defined at the start of the study. The ultrasound examination was performed before the start of ovarian stimulation. The ultrasound data were not available for all patients due to local routines, such as distance monitoring. Data were compared with surgery reports and histopathological examination. The tube was assumed to be visible on ultrasound in cases where ultrasound data were missing and where it was documented that the tube was dilated with a diameter measuring 20 mm (n 17). Complementary groups to the defined subgroups were analysed regarding pregnancy outcome. The complementary groups included patients with a unilateral hydrosalpinx, patients with hydrosalpinges that were non-visible on ultrasound, and patients with a non-visible unilateral hydrosalpinx. Statistical methods The data were subjected both to an intention-to-treat analysis and to an analysis of actual treatment. The intention-to-treat analysis was performed, taking into consideration only the randomization allocation, regardless of subsequent surgical intervention. Fisher s exact test was used for comparison of dichotomous variables between groups. Mantel Haenszel s χ 2 -test was used for test of linear association in ordered contingency tables. As described above, in a large proportion of cycles, the randomization group was not equivalent to the actual treatment group. In order to account for all transfer cycles, regardless of randomization allocation, a statistical method was used which would consider actual treatment without exclusions. The statistical analysis of the primary outcome (birth) was performed with the Cox regression model, using the time-dependent co-variates of treatment group, age, number of good quality embryos transferred and number of transfer cycles before surgery. The cycle number was thus used as a discrete time variable. The Cox regression model was not suitable for analysing the secondary outcome (clinical pregnancy) as this is not a terminal event, since a pregnancy can occur more than once per individual. Instead, clinical pregnancy was analysed by a multivariate logistic regression model for longitudinal data, using the method of generalized estimation equations (Diggle et al., 1994). This method gives robust estimates adjusted for the dependence within each woman. The independent variables included that of treatment group (salpingectomy or no surgical intervention), number of transfer cycles in each treatment

Cumulative results after salpingectomy of hydrosalpinges before IVF Figure 1. Flow chart of 204 randomized patients. group, age and mean number of good quality embryos. For comparison of the continuous variables between groups, mean individual values were calculated to which the Wilcoxon two-sample test was applied. The secondary endpoint, implantation, was compared in the treatment groups of salpingectomy and non-intervention. In the non-intervention group, cycles performed after non-randomized surgery were excluded to avoid inclusion of the same patient in both groups. Two-tailed statistical tests were conducted at the 5% level of significance. The Bonferroni Holm method was used to correct for multiple comparisons (Holm, 1979). Results One hundred and eighty-six patients underwent 452 transfer cycles. Salpingectomy was performed in 103 patients who subsequently underwent a mean of 2.3 embryo transfers (range 1 6); 56 patients in the treatment control group without surgical intervention underwent 2.2 (range 1 6) embryo transfers. In addition, 27 patients were treated in both treatment groups, 1.3 transfers (range 1 3) before (n 26) and 1.9 (range 1 4) after (n 27) surgery. One patient had ovarian stimulation before surgery and her only transfer after salpingectomy. The presented numbers relate to actual treatment groups and not to randomization groups. Demographic data Patient characteristics were similar with respect to age and obstetric history in the two randomization groups (Strandell et al., 1999). The cycle characteristics for the two treatment groups are presented in Table I. No significant differences were detected. Reproductive outcome Intention-to-treat analysis Pregnancy and birth rates based on an intention-to-treat analysis are presented in Table II. The cumulative results (birth rates of 56.5 versus 55.1%) were similar in both randomization groups, considering that one-third (25/78) of the patients in the non-intervention group underwent salpingectomy after failure in the non-surgery group. Actual treatment analyses The effects of salpingectomy on the subsequent chances of birth, based on cumulative cycles, analysed by Cox regression model, are presented in Table III. The chance of birth was doubled after salpingectomy (hazard ratio 2.1, 95% CI 1.6 3.6, P 0.014). The increases in birth rates were even more explicitly expressed in the subgroups of patients with ultrasound-visible hydrosalpinges (hazard ratio 3.8, 95% CI 1.5 9.2, P 0.004). The effects of salpingectomy on clinical pregnancy are presented in Table IV. An increase in pregnancy rates subsequent to salpingectomy was demonstrated both within the total study population (OR 1.7, 95% CI 1.01 2.8, P 0.046) and among patients with ultrasound-visible hydrosalpinges (OR 2.8, 95% CI 1.3 6.1, P 0.01). The highest order of multiple birth was twins, which occurred in 12 out of 71 (16.9%) births in treatment cycles 2405

A.Strandell et al. Table I. Characteristics of all transfer cycles and distribution of types of hydrosalpinges in the two treatment groups Laparoscopic salpingectomy No intervention before IVF (n 130) a (n 82) Mean (SD) Range Mean (SD) Range Age 32.8 (3.5) 24.4 39.6 32.5 (3.8) 23.3 39.1 Units of gonadotrophin 2328 b (944) 750 5850 2201 (771) 500 4875 Duration of ovarian stimulation (days) 11.3 b (2.1) 5 20 11.4 (2.6) 5 25 No. of retrieved oocytes 10.3 b (5.4) 2 34 10.6 (5.4) 1 26 No. of fertilized and cleaved oocytes 6.8 b (4.1) 1 27 7.1 (4.4) 1 23 No. of transfers 2.3 (1.3) 1 8 1.9 (1.1) 1 6 with fresh embryos 1.8 (0.9) 0 5 1.7 (0.8) 1 4 with frozen thawed embryos 0.4 (0.8) 0 4 0.2 (0.6) 0 3 No. of embryos/transfer 2.0 (0.3) 1 3 2.0 (0.3) 1 3 No. of high quality embryos/transfer c 1.6 (0.7) 0 3 1.7 (0.6) 0 3 n % n % Bilateral hydrosalpinges 75 57.7 34 41.5 Ultrasound-visible hydrosalpinges 66 50.8 42 51.2 Bilateral ultrasound-visible hydrosalpinges 41 31.5 19 23.2 Mean refers to the mean of each individual s mean, range refers to range in separate cycles. n denotes number of patients, 26 patients had cycles performed in both groups. a The 130 patients in the salpingectomy group is the sum of 103 patients with salpingectomy before any transfer one patient with salpingectomy after oocyte retrieval but before transfer 26 patients who also had transfers in the no-intervention group. b n 129, one patient had only a transfer of frozen thawed embryos. c A high quality embryo was characterized by 20% fragmentation of the blastomeres. Table II. Intention-to-treat analysis. Cumulative pregnancy rates in 204 patients allocated to the two randomization groups Laparoscopic salpingectomy No intervention before IVF No. of patients not following randomization 7 25 allocation through the entire study Per included patient: 116 88 No. of pregnancies/no. of pregnant women 82 a /73 (62.9) 57 b /50 (56.8) (cumulative rate per woman, %) Cumulative birth rate, n (%) 62 b (53.4) 44 (50.0) Per started cycle: 110 82 No. of pregnancies/no. of pregnant women 80 c /71 (64.6) 56/49 (59.8) (cumulative rate per woman, %) Cumulative birth rate n (%) 61 c (55.5) 43 (52.4) Per transfer cycle: 108 78 No. of pregnancies/no. of pregnant women 80 c /71 (65.7) 56/49 (62.8) (cumulative rate per woman, %) Total no. of spontaneous abortions/no. of aborting 15/13 (13/68 19.1) 10/10 (10/47 21.3) women (abortion rate per woman with IUP) No. of ectopic pregnancies (rate/woman) 4 (3.7) 4 (5.1) Cumulative birth rate, n (%) 61 c (56.5) 43 (55.1) a Includes two spontaneous pregnancies without any IVF treatment. b Includes one spontaneous pregnancy without any IVF treatment. c Includes two spontaneous pregnancies after finalized IVF treatment. IUP intrauterine pregnancy: women with only ectopic pregnancies are excluded. There were no significant differences. following salpingectomy and in 10 out of 30 (33.3%) births following cycles without surgical intervention. Spontaneous abortions occurred in 17 out of 89 intrauterine pregnancies (crude rate 19.1%) in cycles following salpingectomy and in eight out of 37 (21.6%) in cycles without surgical intervention. Three patients had an ectopic implantation despite a previous 2406 bilateral (n 2) or unilateral (n 1) salpingectomy, and five ectopic pregnancies occurred in patients with persistent hydrosalpinges. Approximately one-third (25/78) of the patients randomized to no surgical intervention underwent a salpingectomy after one or two failed cycles. Their cycle characteristics and types

Cumulative results after salpingectomy of hydrosalpinges before IVF Table III. Treatment analysis: effect of salpingectomy on birth rate as primary outcome Population No. of No. of Birth rate patients cycles Hazard 95% CI P Bonferroni Holm ratio adjusted P Total study group 186 452 2.1 1.6 3.6 0.014 0.028 Subgroups of patients with: Bilateral hydrosalpinges 97 246 2.4 1.02 5.7 0.044 0.044 Ultrasound-visible 93 218 3.8 1.5 9.2 0.004 0.015 hydrosalpinges Bilateral ultrasound- 52 122 6.0 1.5 24.3 0.012 0.037 visible hydrosalpinges Cumulative results of 452 transfer cycles in 186 patients, analysed by Cox regression model, analysing the effect of salpingectomy, adjusted for age, number of high quality embryos transferred and number of transfer cycles before surgery. A hazard ratio 1.0 implies a positive effect of salpingectomy. Table IV. Treatment analysis: effect of salpingectomy on clinical pregnancy Population No. of No. of Clinical pregnancy patients cycles Odds 95% CI P Bonferroni Holm ratio adjusted P Total study group 186 452 1.7 1.01 2.8 0.0464 NS (0.0598) Subgroups of patients with: Bilateral hydrosalpinges 97 246 2.3 1.1 4.9 0.0299 NS (0.0598) Ultrasound-visible 93 218 2.8 1.3 6.1 0.0104 0.0312 hydrosalpinges Bilateral ultrasound- 52 122 6.9 1.8 26.8 0.0050 0.0200 visible hydrosalpinges Cumulative results of 452 transfer cycles in 186 patients, analysed by logistic regression using the method of generalized estimation equations, adjusted for age, number of high quality embryos transferred and number of transfer cycles in each treatment group. An odds ratio 1.0 implies a positive effect of salpingectomy. NS not significant. of hydrosalpinges as well as pregnancy outcome are compared in Table V with those of patients who did not undergo the surgery at all. Implantation rates were significantly higher in salpingectomized patients, both in the total study group (27.2% versus 20.2, P 0.03) and in the subgroups, where the differences were even more explicitly expressed (Table VI). The groups complementary to the defined subgroups did not express any significant improvement in birth rates after salpingectomy [patients with unilateral hydrosalpinx (n 89; hazard ratio 1.5, 95% CI 0.7 3.5, P 0.3, not significant), patients with hydrosalpinges not visible on ultrasound (n 66; hazard ratio 0.9, 95% CI 0.4 2.1, P 0.7), patients with neither bilateral nor ultrasound-visible hydrosalpinges (n 34) (hazard ratio 0.7, 95% CI 0.2 2.4, P 0.5)]. Discussion Results from the first cycle in the Scandinavian randomized controlled trial on salpingectomy prior to IVF showed that a surgical intervention increased implantation rates in patients with bilateral hydrosalpinges and demonstrated a clear benefit of salpingectomy in patients with ultrasound-visible hydrosalp- inges (Strandell et al., 1999). We concluded that a preventive salpingectomy can be recommended to patients with hydrosalpinges large enough to be visible on ultrasound and in particular to those with bilateral affection. We claimed that it would be even more valuable to identify a subset of patients who would benefit the most from surgery, and we hoped that the cumulative data, reflecting a more accurate clinical reality, would give us more precise guidelines regarding preventive salpingectomy. With the strength of additional cycles, we have now proven that implantation is significantly impaired in the presence of hydrosalpinges. We have also shown that the chance of birth is improved after salpingectomy in all patients with hydrosalpinges, as demonstrated by statistical significance. Does this result imply a general recommendation of preventive salpingectomy to all such patients? A previous report of retrospective design has suggested that patients with ultrasound-visible hydrosalpinges carry the worst prognosis compared with all other tubal factor infertility (de Wit et al., 1998). Also, studies in which ultrasound has been used for the diagnosis of hydrosalpinx tend to express larger differences in pregnancy rates between hydrosalpinx patients and controls (Andersen et al., 1994; Katz et al., 1996; Van Voorhis et al., 1998). From the results of the present 2407

A.Strandell et al. Table V. Characteristics of all transfer cycles, distribution of types of hydrosalpinges and pregnancy outcome within the randomization group no intervention before IVF No surgery ever (n 53) Salpingectomy after failed cycle(s) (n 24) a Meaan (SD) Range Before surgery All cycles Mean (SD) Range Mean (SD) Range Age 32.1 (3.8) 24.8 39.1 32.8 (3.6) 23.3 38.7 33.4 (3.5) 24.8 39.3 Units of gonadotrophin 2181 (760) 1147 4875 2269 (853) 500 3750 2437 (988) 625 4625 Duration of ovarian stimulation (days) 11.6 (2.8) 6 25 11.2 (2.5) 5 17 11.2 (1.9) 5 15 No. of retrieved oocytes 10.9 (4.8) 1 23 9.7 (6.4) 2 26 8.9 (5.1) 2 20 No. of fertilized and 7.1 (3.5) 1 15 6.6 (5.7) 1 23 6.1 (4.2) 1 15 cleaved oocytes No. of transfers 2.2 (1.2) 1 6 1.3 (0.6) 1 3 3.2 (1.1) 2 7 with fresh embryos 1.8 (0.9) 1 4 1.3 (0.4) 1 2 2.7 (0.6) 2 4 with frozen thawed embryos 0.4 (0.7) 0 3 0.04 (0.2) 0 1 0.5 (0.9) 0 3 No. of embryos/transfer transfer 2.0 (0.3) 1 3 2.0 (0.5) 1 3 1.9 (0.4) 1 3 No. of high quality 1.8 (0.6) 0 3 1.7 (0.7) 0 3 1.5 (0.6) 0 3 embryos/transfer b Bilateral hydrosalpinges, n (%) 20 (37.7) 12 (50.0) Ultrasound-visible hydrosalpinges, n (%) 25 (47.2) 13 (54.2) Bilateral ultrasound-visible hydrosalpinges, n (%) 10 (18.9) 8 (33.3) No. of pregnancies/no. of pregnant 37/34 (64.2) 4/4 (16.7) 18/14 (58.3) women (cumulative rate/woman, %) Cumulative birth rate, n (%) 30 (56.6) 0 12 (50.0) Mean refers to mean of each individual s mean, range refers to range in separate cycles. n denotes number of patients. a One patient with stimulation before and her only transfer after salpingectomy was excluded. b A high quality embryo was characterized by 20% fragmentation of the blastomeres. Table VI. Mean implantation rates (IR) (of each individual s mean implantation rate) in all transfer cycles in the total study population and in subgroups of patients with bilateral and ultrasound-visible hydrosalpinges in the two treatment groups Study group Laparoscopic No intervention P Bonferroni Holm salpingectomy a before IVF adjusted P n Mean (SD) n Mean (SD) IR IR Total study population 104 27.2 (29.1) 82 20.2 (28.6) 0.030 0.030 Bilateral hydrosalpinges 63 29.4 (31.5) 34 13.4 (23.4) 0.004 0.008 Ultrasound-visible hydrosalpinx 51 30.3 (29.4) 42 17.1 (30.4) 0.003 0.009 Bilateral and ultrasound-visible 33 34.2 (32.0) 19 10.5 (26.8) 0.001 0.004 Cycles after non-randomized surgery are excluded. n denotes the number of included patients. a 104 patients in the salpingectomy group is the sum of 103 patients with salpingectomy before any transfer one patient with sapingectomy after oocyte retrieval but before transfer. study, it is obvious that patients with hydrosalpinges large enough to be visible on ultrasound benefit the most from a salpingectomy. None of the groups complementary to the subgroups demonstrated any effect of salpingectomy on birth rates. It was concluded that the positive effect of salpingectomy in the total study population and in patients with bilateral hydrosalpinges was due to the presence of patients with ultrasound-visible hydrosalpinges within these groups. Furthermore, the theories of how hydrosalpinges exert a negative effect on the implantation rate deal mainly with the hydrosalpingeal fluid, supporting the identification of patients 2408 with ultrasound-visible hydrosalpinges as being a poor prognosis group. The suggested mechanisms have focused on the embryotoxicity of the fluid (Mukherjee et al., 1996; Beyler et al., 1997; Murray et al., 1997; Rawe et al., 1997; Sachdev et al., 1997; Granot et al., 1998; Koong et al., 1998; Strandell et al., 1998; Roberts et al., 1999; Strandell 2000), or the mechanical leakage of the fluid into the uterine cavity, causing endometrial alterations, hostile to embryo implantation and development (Meyer et al., 1997; Cohen et al., 1999), or simply a mechanical wash-out of embryos (Andersen et al., 1996; Sharara 1999).

Cumulative results after salpingectomy of hydrosalpinges before IVF There has been debate over the risk of unnecessary removal of Fallopian tubes (Puttemans et al., 2000). There are patients referred for IVF who have not had their tubes properly evaluated, and a distally occluded tube may be found to be suitable for functional surgery. In cases of uncertainty, a diagnostic distal salpingostomy for evaluation of the mucosa, prior to any salpingectomy, is of value. However, in the present study histopathological evaluation of the removed tubes confirmed that the occurrence of a healthy mucosa is very rare (Strandell and Lindhard, 2000), and only one woman had a therapeutic salpingostomy performed. As long as a salpingostomy does not re-occlude, it is likely to have a positive effect on IVF outcome, besides the chance of a spontaneous conception and the risk of an ectopic pregnancy, although there is no prospective study to support this speculation. Other treatment options, such as transvaginal aspiration of the hydrosalpingeal fluid have been examined in retrospective studies but with contradictory results (Sowter et al., 1997; Van Voorhis et al., 1998). Hitherto, salpingectomy is the only treatment of hydrosalpinges that has been evaluated in a prospective randomized trial. It is also the only suggested treatment without any need for repeated intervention. There has also been concern about whether the ovarian function could be at risk of being disturbed by a surgical procedure like salpingectomy. Previous studies on ovarian performance after unilateral salpingectomy due to ectopic pregnancy have not demonstrated any overall effect on the ovarian response to gonadotrophins (Lass et al., 1998; Dar et al., 2000), although the study results were contradictory when separate sides were compared. A separate analysis within the present study, of patients undergoing prophylactic salpingectomy of which the majority was a bilateral procedure, showed that there was no significant reduction in ovarian performance (Strandell et al., 2001). The intention-to-treat analysis displayed equally high birth rates in both randomization groups. Obviously, with one-third of the patients in the non-intervention group having finally undergone the salpingectomy at a later occasion, the cumulative intention-to-treat analysis no longer reflected the surgical intervention. However, it was reassuring to find that the patients reached a full-term pregnancy to the same extent, whether they had undergone a salpingectomy prior to their first IVF or after one or two failed cycles, albeit more cycles were required in the latter case and at a higher expense. The present study has proven the hypothesis that removal of a hydrosalpinx before undergoing IVF treatment improves birth rates is efficacious, and supports the theory that hydrosalpingeal fluid has a causative role in impaired pregnancy outcome. Based on the present results, the following clinical decision can therefore be made: patients with hydrosalpinges, uni- or bilateral and large enough to be seen as fluid-filled by ultrasound before initiating ovarian stimulation, can be recommended for a laparoscopic salpingectomy in order to improve their chance of achieving a birth after IVF treatment. Acknowledgements We are most grateful to the Scandinavian IVF centres participating in the study and the responsible investigator at each site for conducting the patient recruitment and data collection. The study was supported by grants from the Göteborg Medical Society, the Hjalmar Svensson Foundation and the society Ordensällskapet W:6. Nils-Gunnar Pehrson is acknowledged for valuable statistical advice. References Andersen, A.N., Yue, Z., Meng, F.J. et al. (1994) Low implantation rate after in-vitro fertilization in patients with hydrosalpinges diagnosed by ultrasonography. Hum. Reprod., 9, 1935 1938. Andersen, A.N., Lindhard, A., Loft, A. et al. (1996) The infertile patient with hydrosalpinges IVF with or without salpingectomy. Hum. Reprod., 11, 2081 2084. Beyler, S.A., James, K.P., Fritz, M.A. et al. (1997) Hydrosalpingeal fluid inhibits in-vitro embryonic development in a murine model. Hum. Reprod., 12, 2724 2728. Camus, E., Poncelet, C., Goffinet, F. et al. (1999) Pregnancy rates after IVF in cases of tubal infertility with and without hydrosalpinx: meta-analysis of published comparative studies. Hum. Reprod., 14, 1243 1249. Cohen, M.A., Lindheim, S.R. and Sauer, M.V. (1999) Hydrosalpinges adversely affect implantation in donor oocute cycles. Hum. Reprod., 14, 1087 1089. Dar, P., Sachs, G.S., Strassburger, D. et al. (2000) Ovarian function before and after salpingectomy in artificial reproductive technology patients. Hum. Reprod., 15, 142 144. Déchaud, H., Daurès, J.P., Arnal, F. et al. (1998) Does previous salpingectomy improve implantation and pregnancy rates in patients with severe tubal factor infertility who are undergoing in vitro fertilization? A pilot prospective randomized study. Fertil. Steril., 69, 1020 1025. dewit, W., Gowrising, C.J., Kuik, D.J. et al. (1998) Only hydrosalpinges visible on ultrasound are associated with reduced implantation and pregnancy rates after in-vitro fertilization. Hum. Reprod., 13, 1696 1701. Diggle, P.J., Liang, K.Y. and Zeger, S.L. (1994) The analysis of longitudinal data. Oxford University Press, New York, pp. 151 153. Ejdrup Bredkjaer, H., Ziebe, S., Hamid, B. et al. (1999) Delivery rates after in-vitro fertilization following bilateral salpingectomy due to hydrosalpinges: a case control study. Hum. Reprod., 14, 101 105. Fleming, C. and Hull, M.G.R. (1996) Impaired implantation after in vitro fertilization treatment associated with hydrosalpinx. Br. J. Obstet. Gynaecol., 103, 268 272. Granot, I., Dekel, N., Segal, I. et al. (1998) Is hydrosalpinx fluid cytotoxic? Hum. Reprod., 13, 1620 1624. Holm, S. (1979) A simple sequentially rejective multiple test procedure. Scand. J. Statist., 6, 65 70. Kassabji, M., Sims, J.A., Butler, L. et al. (1994) Reduced pregnancy outcome in patients with unilateral or bilateral hydrosalpinx after in vitro fertilization. Eur. J. Obstet. Gynecol. Reprod. Biol., 56, 129 132. Katz, E. Akman, M.A., Damewood, M.D. et al. (1996) Deleterious effect of the presence of hydrosalpinx on implantation and pregnancy rates with in vitro fertilization. Fertil. Steril., 66, 122 125. Koong, M.K., Jun, J.H., Song, S.J. et al. (1998) A second look at the embryotoxicity of hydrosalpingeal fluid: an in-vitro assessment in a murine model. Hum. Reprod., 13, 2852 2856. Lass, A., Ellenbogen, A., Croucher, C. et al. (1998) Effect of salpingectomy on ovarian response to superovulation in an in vitro fertilization embryo transfer program. Fertil. Steril., 70, 1035 1038. Meyer, W.R., Castelbaum, A.J., Somkuti, S. et al. (1997) Hydrosalpinges adversely affect markers of endometrial receptivity. Hum. Reprod., 12, 1393 1398. Mukherjee, T., Copperman, A.B., McCaffrey, C. et al. (1996) Hydrosalpinx fluid has embryotoxic effects on murine embryogenesis: a case for prophylactic salpingectomy. Fertil. Steril., 66, 851 853. Murray, C.A., Clarke, H.J., Tulandi, T. et al. (1997) Inhibitory effect of human hydrosalpingeal fluid on mouse preimplantation embryonic development is significantly reduced by the addition of lactate. Hum. Reprod., 12, 2504 2507. Murray, D.L., Sagoskin, A.W., Widra, E.A. et al. (1998) The adverse effect of hydrosalpinges on in vitro fertilization pregnancy rates and the benefit of surgical correction. Fertil. Steril., 69, 41 45. Ng, E.H., Yeung, W.S. and Ho, P.C. (1997) The presence of hydrosalpinx may not adversely affect the implantation and pregnancy rates in in vitro fertilization treatment. J. Assist. Reprod. Genet., 14, 508 512. Puttemans, P.J., Campo, R., Gordts, S. and Brosens, I.A. (2000) Hydrosalpinx and ART. Hydrosalpinx functional surgery or salpingectomy? Hum. Reprod., 15, 1427 1430. 2409

A.Strandell et al. Rawe, V.J., Liu, J., Shaffer, S. et al. (1997) Effect of human hydrosalpinx fluid on murine embryo development and implantation. Fertil. Steril., 68, 668 670. Roberts, J.F., Clarke, H.J., Tulandi, T. and Tan, S.L. (1999) Effects of hydrosalpingeal fluid on murine embryo development and implantation. J. Assist. Reprod. Genet., 16, 421 424. Sachdev, R., Kemmann, E., Bohrer, M.K. et al. (1997) Detrimental effect of hydrosalpinx fluid on the development and blastulation of mouse embryos in vitro. Fertil. Steril., 68, 531 533. Sharara, F.I. (1999) The role of hydrosalpinx in IVF: simply mechanical? Hum. Reprod., 14, 577 578. Shelton, K.E., Butler, L., Toner, J.P. et al. (1996) Salpingectomy improves the pregnancy rate in in-vitro fertilization patients with hydrosalpinx. Hum. Reprod., 11, 523 525. Sowter, M.C., Akande, V.A., Williams, J.A. et al. (1997) Is the outcome of in-vitro fertilization and embryo transfer treatment improved by spontaneous or surgical drainage of a hydrosalpinx? Hum. Reprod., 12, 2147 2150. Strandell, A. (2000) The influence of hydrosalpinx on IVF and embryo transfer: a review. Hum. Reprod. Update, 6, 387 395. Strandell, A. and Lindhard, A. (2000) Hydrosalpinx and ART. Salpingectomy prior to IVF can be recommended to a well-defined subgroup of patients. Hum. Reprod., 15, 2072 2074. Strandell, A., Waldenström, U., Nilsson, L. et al. (1994) Hydrosalpinx reduces in-vitro fertilization/embryo transfer rates. Hum. Reprod., 9, 861 863. Strandell, A., Sjögren, A., Bentin-Ley, U. et al. (1998) Hydrosalpinx fluid does not adversely affect the normal development of human embryos and implantation in vitro. Hum. Reprod., 13, 2921 2925. Strandell, A., Lindhard, A., Waldenström, U. et al. (1999) Hydrosalpinx and IVF outcome: a prospective, randomized multicentre trial in Scandinavia on salpingectomy prior to IVF. Hum. Reprod., 14, 2762 2769. Strandell, A., Lindhard, A., Waldenström, U. et al. (2001) Salpingectomy prior to IVF does not impair the ovarian response. Hum. Reprod., 16, 1135 1139. Van Voorhis, B.J., Sparks, A.E., Syrop, C.H. et al. (1998) Ultrasound-guided aspiration of hydrosalpinges is associated with improved pregnancy and implantation rates after in-vitro fertilization cycles. Hum. Reprod., 13, 736 739. Vandromme, J., Chasse, E., Lejeune, B. et al. (1995) Hydrosalpinges in invitro fertilization: an unfavourable prognostic feature. Hum. Reprod., 10, 576 579. Zeyneloglu, H.B., Arici, A. and Olive, D.L. (1998) Adverse effects of hydrosalpinx on pregnancy rates after in vitro fertilization embryo transfer. Fertil. Steril., 70, 492 499. Received on February 6, 2001; accepted on August 9, 2001 2410