Article Routine Chlamydia antibody testing is of limited use in subfertile women with anovulation

Transcription

1 RBMOnline - Vol 14. No Reproductive BioMedicine Online; on web 5 February 2007 Article Routine Chlamydia antibody testing is of limited use in subfertile women with anovulation Olga van Tetering was born on 16 February She started her medical studies in 1996 at the University of Leuven, Belgium. Two years later she switched to Maastricht University, and in 2003 she graduated cum laude. After graduating she took up a post at the department of Obstetrics and Gynaecology in Twee Steden Hospital, Tilburg and Maxima Medical Centre, Veldhoven. She published several articles, supervised by Dr BWJ Mol, during this appointment. She is currently completing her residency in the Intensive Care Unit at Amphia Hospital, Breda, The Netherlands. Olga van Tetering EAA van Tetering 1,7, P Bourdrez 2, CAM Koks 1, FMV Delemarre 3, HJLA Ruis 4, HJHM van Dessel 5, BWJ Mol 1,6 1 Department of Obstetrics and Gynaecology, Máxima Medical Center, PO Box 7777, 5500MB, Veldhoven, The Netherlands; 2 Department of Obstetrics and Gynaecology, VieCuri Medical Center, Venlo, The Netherlands; 3 Department of Obstetrics and Gynaecology, Elkerliek Hospital, Helmond, The Netherlands; 4 Stichting Geertgen, The Netherlands; 5 Department of Obstetrics and Gynaecology, Twee Steden Hospital, Tilburg, The Netherlands; 6 Center for Reproductive Medicine, Academic Medical Center, Amsterdam, The Netherlands 7 Correspondence: Tel: +31 (0) ; Fax: +31 (0) ; ovantetering@gmail.com Abstract The Chlamydia antibody titre (CAT) is a test used to identify subfertile couples at increased risk for tubal pathology. The usefulness of the routine performance of CAT was evaluated in a multicentre prospective cohort study, in women without regular ovulation. Consecutive couples presenting with subfertility due to an irregular menstrual cycle or amenorrhoea were included. A total of 711 women were studied, all of whom underwent CAT. Tubal status was verified in 190 of these women. Two-sided tubal pathology was found in 5% of these women, and one-sided occlusion in 10%. Of all the women in the study group, 33 (4.6%) had an abnormal CAT, of which 21 underwent further tubal testing. Tubal pathology was found in two (10%) of these 21 patients. The sensitivity and specificity of CAT were respectively 20% and 89%. Correction for verification bias increased the specificity to 96% with a drop of the sensitivity to 9%. In subfertile couples with anovulation, the performance of CAT is not useful. It is proposed that testing for tubal disease in these women is delayed until treatment with clomiphene citrate has failed. Keywords: anovulation, Chlamydia antibody titre, subfertility, tubal pathology Introduction 322 Absence of regular ovulation is an important cause of subfertility. Anovulation can easily be identified, as women without regular ovulation report oligo- or amenorrhoea. The majority of women without regular ovulation have a normogonadotrophic oligoamenorrhoeic status (WHO group 2) (Rowe et al., 1993). In these women, ovulation induction with clomiphene citrate (CC) is the treatment of first choice (Rowe et al., 1993). The success rate of this treatment is relatively high, with ovulation rates up to 80%, resulting in conception in about half of the couples (Imani et al., 1998; 1999). At present, most guidelines for subfertile couples recommend tubal patency testing in couples presenting with subfertility to exclude tubal pathology (Nederlandse vereniging voor obstetrie en gynaecologie (NVOG), 2004a,b; National Institute for Clinical Excellence (NICE), 2004). The Fallopian tubes can be assessed by hysterosalpingography (HSG), transvaginal hydrolaparoscopy (THL) or laparoscopy with dye testing. The use of Chlamydia antibody titre (CAT) has been suggested as a non-invasive alternative first-line strategy (Dabekausen et al., 1994). In a meta-analysis reporting on patients from the general 2007 Published by Reproductive Healthcare Ltd, Duck End Farm, Dry Drayton, Cambridge CB3 8DB, UK

2 subfertile population, the accuracy of this test for the diagnosis of tubal occlusion has been reported to be equal to that of HSG (Mol et al., 1997a). In the recommendation of tests for tubal disease, the pretest probabilities for the presence of tubal pathology are not taken into account (National Institute for Clinical Excellence (NICE), 2004). However, it is probable that differences in pretest probability might affect the value of a diagnostic strategy. One could postulate that in the presence of anovulation (an explanation for infertility per se), the presence of other causes of subfertility, such as tubal disease, may be less likely. In this study, the usefulness of testing for tubal disease among the initial work-up for women suffering from subfertility due to anovulation was evaluated. Materials and methods This study was performed in five hospitals in The Netherlands [Máxima Medical Center, VieCuri Medical Center, Elkerliek Hospital, Stichting Geertgen, Twee Steden Hospital]. These centres prospectively enter all data for couples presenting with subfertility in a similar database programme. In all these centres, a CAT is performed in all subfertile couples. Between January 2002 and December 2004, all couples presenting with subfertility based on anovulation were included in the study. The study was limited to women suffering from normogonadotrophic normooestogenic anovulation. Besides information on the menstrual cycle, a detailed history was performed, including information on age, duration of subfertility, history of salpingitis, previous subfertility treatment and past abdominal operations. All these potential prognostic factors were used in the analysis. The CAT was determined with enzyme-linked immunosorbent assay (ELISA) (ANI, Labsystems, Finland). A positive CAT result was defined as an ELISA measurement of 1 or more. In cases of a negative CAT result, women were treated with CC. When a positive CAT result was determined, the status of the tubes was assessed using either HSG or laparoscopy with dye tests. In one hospital (MMC Veldhoven), the procedure of first choice to assess tubal structures and tubal patency was THL. The prevalence of an abnormal CAT result was evaluated. Subsequently, the result of the CAT was tabulated against the final tubal status as assessed with HSG, THL or laparoscopy, which was considered to be the reference. Patients were categorized as having two-sided tubal pathology, one-sided tubal pathology, or no tubal pathology. The sensitivity was then calculated, defined as the number of patients with an abnormal CAT result among those with tubal pathology, and specificity, defined as the number of patients with a normal CAT among those without tubal pathology. In this calculation, two-sided tubal pathology was considered to be abnormal, whereas onesided tubal pathology was considered to be normal. In this initial calculation, women who did not have a HSG, THL or a laparoscopy were not included in the analysis. As a substantial number of couples either conceived prior to HSG, THL or laparoscopy or did not undergo such a test, verification of tubal status was not available in all patients. Therefore, the initial calculations of sensitivity and specificity are likely to be inflated by verification bias (Begg and Greenes, 1983). To correct for this bias, it was assumed that the women in whom tubal pathology was not assessed had the same distribution of tubal pathology as the women who had the same CAT result and in whom tubal status was verified. After these patients had been added to a 2 2 table of expected results, sensitivity and specificity were recalculated. Alternatively, one could consider those patients who conceived, with either an ongoing pregnancy or a non-viable intrauterine pregnancy, as having healthy tubes. The remaining patients, i.e. those without tubal testing and without a pregnancy, were then added to the 2 2 table according to the distribution of tubal disease in those with a positive CAT result and those with a negative test result. After this second correction for verification bias, sensitivity and specificity were recalculated. Finally, the consequences of the use of CAT were compared with no testing for tubal disease and ovulation induction in all women, and with testing for tubal disease with HSG, laparoscopy or THL in all patients prior to ovulation induction. To do so, it was assumed that the pregnancy rate after ovulation induction was 50% in cases where tubal disease was absent, whereas the pregnancy rate would be zero in cases of doublesided tubal disease. Results A total of 711 couples were included in this study. The mean duration of subfertility was 1.3 years and the median age of all females was 29.5 years. Among the 711 couples, 513 (72%) were suffering from primary subfertility and 160 women (22.5%) had been treated for subfertility in the past. Four women reported a history of salpingitis, whereas in 125 patents it was unknown whether they had a previous salpingitis. In the four women reporting a previous salpingitis, one had twosided tubal pathology, one had one-sided tubal pathology, one patient was without tubal pathology, and the tubal status was not assessed in the fourth patient. Among these four women, only the one without tubal pathology had a positive CAT. There were 22 women who had a history of abdominal surgery, of whom two had two-sided tubal pathology. As shown in Table 1 and in Figure 1, the CAT was elevated in 33 of 711 patients (4.6%). In 21 of these 33 patients, tubal pathology was assessed. Despite a positive CAT, in 12 other patients no tubal testing was performed for various reasons: in seven patients pregnancy occurred before tubal testing could be performed, one patient was referred for intracytoplasmic sperm injection (ICSI) treatment, one patient was lost to follow up, one patient refused further tubal testing, one patient had no wish for a child and one patient decided to stop further investigation. In the 21 patients who underwent further testing, double-sided tubal pathology was found in two patients, whereas one-sided tubal pathology was never found. The positive predictive value of a CAT was 2/21 (10%) for any type of tubal pathology. Out of 678 patients with a negative CAT, 169 (25%) eventually underwent tubal testing. In these patients, one-sided tubal pathology was found in 19 patients (11%) and an additional eight had two-sided tubal pathology (4.7%). The sensitivity of CAT was 20% for a specificity of 89%. 323

3 Table 1. Tubal status in relation to Chlamydia antibody titre (CAT) status. Two-sided One-sided Normal No tubal No tubal Total (%) tubal tubal tubes testing, testing, pathology pathology pregnant not pregnant No. CAT positive (4.6) No. CAT negative (95.4) Total Sensitivity: 2/10 (20%); Specificity: 161/180 (89%). 324 Figure 1. Flow chart of results of CAT and tubal testing. (CAT = Chlamydia antibody titre)

4 Table 2. Tables constructed taking verification bias into account. In (a), patients who had no tubal testing and also those who conceived prior to tubal testing are distributed among the categories according to the prevalence of these diseases among Chlamydia antibody titre (CAT) positives and negatives, respectively. In (b), patients who conceived were considered to have no tubal pathology, whereas those who did not conceive were distributed among the categories in a similar way as in (a). (a) 1 (b) 2 Two-sided One-sided Normal Two-sided One-sided Normal tubal tubal tubes tubal tubal tubes pathology pathology pathology pathology No. CAT positive No. CAT negative Sensitivity: 3/35 (9%); Specificity: 646/676 (96%). 2 Sensitivity: 2/25 (8%); Specificity: 655 /686(96%). Table 3. Comparison of three different strategies with respect to tubal testing in 200 fictitious women with ovulation disorders. CAT Diagnostic Treated Pregnancy measurement laparoscopy, with CC THL or HSG Treatment with clomiphene citrate Laparoscopy, THL or HSG for all Laparoscopy, THL or HSG based on CAT result CAT: Chlamydia antibody titre; CC: clomiphene citrate; HSG: hysterosalpingography; THL: transvaginal hydrolaparoscopy. As the definitive tubal status was assessed in only 190 of 711 patients, immediate calculation of test indices sensitivity and specificity will be affected by verification bias. However, if it is assumed that the distribution of tubal pathology is equal among the patients in whom tubal status was verified and those in which it was not, a 2 2 table of expected results can be constructed. Table 2) shows a 2 2 table after correction for verification bias. As for 21 patients with a positive CAT, who had a tubal test, there were 12 who did not have a tubal test, the first row of Table 1 has to be multiplied by a factor of 1.6 (33/21). Similarly for each patient with a negative CAT who had a tubal test, there were three patients without a tubal test. The second row of Table 1 has therefore to be multiplied by a factor of four. The sensitivity of CAT to detect two-sided tubal pathology can now be calculated to be 3/35 (9%) for a specificity of 646/676 (96%). In Table 2a, it was assumed patients who conceived without having had a HSG or laparoscopy had unconfirmed tubal status. However, one could argue that in women conceiving an intrauterine pregnancy, no relevant tubal pathology exists. Patients who conceived were therefore considered as having no tubal pathology, whereas those who did not conceive were distributed among the categories two-sided tubal pathology, one-sided tubal pathology, and no tubal pathology in a similar way as in Table 2a. The expected results are shown in Table 2b. The sensitivity is then estimated to be 2/25 (8%) versus a specificity of 655/686 (96%). If we apply these rates to a fictitious population of 200 women, 10 of them would have double-sided tubal pathology. Table 3 shows that ovulation induction would result in the treatment of 200 women, of whom 95 would conceive (i.e. 50% of the 190 women without double-sided tubal pathology). Immediate invasive tubal testing would result in the prevention of treatment with ovulation induction in 10 women, however at the expense of 200 invasive procedures. Risk stratification with CAT would reduce the number of diagnostic procedures to be performed to 22, but as a consequence only two out of 10 women with tubal disease could be detected. 325

5 326 Discussion In this study, it was found that, in a large cohort of subfertile women with an ovulation disorder, the prevalence of two-sided tubal pathology was 5%, whereas 10% had one-sided tubal pathology. Of the 711 women, 33 (4.6%) had an abnormal CAT, of whom 21 underwent further tubal testing. Tubal pathology was found in two of these 21 (10%) patients. The sensitivity of CAT was 20% for a specificity of 89%. Adjustment for verification bias did not alter these results substantially. The fact that only a minority of the patients had a definite assessment of their tubal status affects our findings. Ideally, all women would have had a visual assessment of the Fallopian tubes immediately after testing for CAT. Moreover, assessment of the tubal status was sometimes done with laparoscopy and sometimes with HSG. Although it would be preferable to have a uniform reference standard, we want to stress that the literature on the subject to date has similar design flaws. In a meta-analysis assessing the accuracy of CAT, almost half of the studies used a combined reference standard, i.e. HSG and laparoscopy (Mol et al., 1997a). In that meta-analysis, studies using a combination of HSG and laparoscopy did not report a different sensitivity and specificity as compared with studies using laparoscopy alone. That study reported on 711 patients who had a CAT test, of whom 190 underwent tubal testing, and 197 conceived without IVF, thus proving that they had no tubal pathology that was of clinical relevance. In the only meta-analysis on the subject, 27 studies were included, but only four studies reported on >190 patients, and the largest studies included 256 patients (Mol et al., 1997a). However, larger studies, in which the diagnosis is verified in all patients, are in our opinion unlikely to be feasible, leaving the present approach as the most practical one. The problem is that almost all studies on this subject suffer from such verification bias, because CAT, while under evaluation, is already applied to indicate who will have HSG or laparoscopy. As this paper presents reports according to the new Standards for Reporting of Diagnostic Accuracy (STARD) criteria that are adopted by the major scientific journals in the world, the results can be appraised by the reader (Bossuyt and Reitsma, 2003). Theoretically, one could consider a third hypothesis, in which all the non-pregnant and untested women were classified as having two-sided tubal pathology. This would imply that the number of women with two-sided tubal disease would increase to an unrealistically high 207, of whom only nine would have been detected. The sensitivity would drop to 9/207 = 4%, rendering CAT even less useful than estimated with other assumptions. Thus, under all assumptions, this paper shows that the value of CAT is limited. A second weakness of this study is the lack of blinding. The CAT result was likely to be known to the doctor who was performing the HSG, laparoscopy or THL. Despite this lack of blinding, we found a very poor sensitivity and specificity of CAT in women with ovulatory problems. As lack of blinding tends to overestimate the test accuracy, we do not feel that this bias affects the interpretation of our results (Urman et al., 2005). In the calculations, two-sided tubal pathology was considered to be abnormal, whereas one-sided tubal pathology was considered to be normal. Although one could argue that onesided pathology is not normal, in our opinion anovulatory patients with one-sided tubal disease should be treated with ovulation induction, and the fact that one-sided tubal pathology is present does not change this policy. In fact, only double-sided tubal pathology leads to a change in clinical management. Among the patients who underwent tubal testing, the prevalence of tubal pathology was found to be low, i.e. 5% for double-sided tubal pathology and 10% for one-sided tubal pathology. These rates are in agreement with previous studies (Mol et al., 1997b) whereas other studies have reported even lower prevalences of tubal disease among patients with anovulation (Bayram et al., 2004). The combination of these low prevalences with the test characteristics that we found imply that the predictive value of a positive CAT is low. From a pre-test probability of two-sided tubal pathology of 5%, the probability only increased to 10% when the CAT was positive, with the probability in those who had a negative CAT being just under 5%. In our opinion, it is questionable whether a HSG, THL or laparoscopy is justified in these women. Several studies (Imani et al., 1998; 1999) report 80% ovulation in normogonadotrophic oligoamenorrhoeic women treated with CC, and finally 50% of these women conceive. Improvement of lifestyle might even increase these numbers (Moran et al., 2006). The flip side of the question would be that ovulation-induction is performed in a patient with blocked tubes. Although the calculations presented here show that the probability of a patient with anovulation having tubal pathology is low, it should nevertheless be communicated to the patient, who should be informed about the low chance of tubal pathology (Van der Steeg et al., 2006). In view of our findings, a possible alternative strategy for the testing of all patients could be to delay tubal testing for CCresistant patients until a laparoscopy has to be performed for laparoscopic electrocauterization of the ovaries (Bayram et al., 2004). In those patients who react favourably to CC treatment, one could perform assessment of the tubal status if pregnancy is not achieved within six to 12 cycles. Based on the present data, we feel that performance of the CAT is not useful in subfertile couples in which the woman is suffering from an ovulation disorder. We propose that testing for tubal disease in these women is delayed until treatment with CC has failed. References Bayram N, van Wely M, Kaaijk EM et al Using an electrocautery strategy or recombinant follicle stimulating hormone to induce ovulation in polycystic ovary syndrome: randomised controlled trial. British Medical Journal 328, Begg CB, Greenes RA 1983 Assessment of diagnostic tests when disease verification is subject to selection bias. Biometrics 39, Bossuyt PMM, Reitsma JB 2003 The STARD initiative. Lancet 361, 71. Dabekausen YAJM, Evers JLH, Land JA et al Chlamydia trachomatis antibody testing is more accurate than hysterosalpingography in predicting tubal factor infertility. Fertility and Sterility 61, Imani B, Eijkemans MJ, te Velde ER et al Predictors of chances to conceive in ovulatory patients during clomphene citrate

6 induction of ovulation in normogonadotropic oligoamenorrheic infertility. Journal of Clinical Endocrinology and Metabolism 84, Imani B, Eijkemans MJ, te Velde ER et al Predictors of patients remaining anovulatory during clomphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. Journal of Clinical Endocrinology and Metabolism 83, Mol BWJ, Dijkman B, Wertheim P et al. 1997a The accuracy of serum chlamydial antibodies in the diagnosis of tubal pathology: a metaanalysis. Fertility and Sterility 67, Mol BWJ, Swart P, Bossuyt PMM et al. 1997b Is hysterosalpingography an important tool in predicting fertility outcome? Fertility and Sterility 67, Moran LJ, Brinkworth G, Noakes M, Norman RJ 2006 Effects of lifestyle modification in polycystic ovarian syndrome. Reproductive BioMedicine Online 12, National Institute for Clinical Excellence (NICE) 2004 Fertility Guidelines CG11. Nederlandse vereniging voor obstetrie en gynaecologie (NVOG) 2004a Richtlijn 01, Oriënterend Fertiliteits-onderzoek (OFO), November [Dutch Association of Obstetrics and Gynaecology, Guideline 01, Basic Subfertility Work-up, November 2004.] Nederlandse vereniging voor obstetrie en gynaecologie (NVOG) 2004b Richtlijn 02, Anovulatie en kinderwens, November [Dutch Association of Obstetrics and Gynaecology, Guideline 02, Anovulation and childbearing wishes, November 2004.] www. nvog.nl. Rowe PJ, Comhaire FH, Hargreave TB, Mellows HJ (eds) 1993 WHO manual for the standardized investigation and diagnosis of the infertile couple. Cambridge: The Press Syndicate of the University of Cambridge. Urman B, Yakin K, Balaban B 2005 Recurrent implantation failure in assisted reproduction: how to counsel and manage. A. General considerations and treatment options that may benefit the couple. Reproductive BioMedicine Online 11, Van der Steeg JW, Steures P, Eijkemans MJ et al Which factors play a role in clinical decision-making in subfertility? Reproductive BioMedicine Online 12, Received 25 May 2006; refereed 26 July 2006; accepted 13 December