Research and Health Policy Studies, Tufts-New England Medical Center, Boston, Massachusetts

Transcription

1 Human chorionic gonadotropin administration vs. luteinizing monitoring for intrauterine insemination timing, after administration of clomiphene citrate: a meta-analysis Ioannis P. Kosmas, M.D., a Athina Tatsioni, M.D., Ph.D., b Human Musavi Fatemi, M.D., a Efstratios M. Kolibianakis, M.D., Ph.D., a Herman Tournaye, M.D., Ph.D., a and Paul Devroey, M.D., Ph.D. a a Centre for Reproductive Medicine, Dutch-Speaking Brussels Free University, Brussels, Belgium; and b Institute for Clinical Research and Health Policy Studies, Tufts-New England Medical Center, Boston, Massachusetts Objective: To systematically compare hcg administration as a method for intrauterine insemination (IUI) timing with conservative urinary LH surge detection in infertility treatment. Design: Meta-analysis of prospective and retrospective trials. Setting: Tertiary fertility and IVF center. Patient(s): One thousand four hundred sixty-one patients who received hcg after a clomiphene citrate regimen, compared with 1,162 patients who had an LH surge detection for IUI timing. Intervention(s): Both MEDLINE and Cochrane Collaboration were searched. References of retrieved articles were included in the search. The meta-analysis included all controlled trials examining the effectiveness of hcg administration before IUI on clinical-pregnancy rates in comparison with LH detection. Two independent reviewers performed data extraction. Main Outcome Measure(s): Pregnancy rates. Result(s): Seven studies with 2,623 patients were included in the meta-analysis (1,461 patients received hcg, and 1,162 had LH surge detection). When all studies were combined, patients who received hcg before IUI demonstrated lower clinical-pregnancy rates than did women who had IUI after spontaneous ovulation (odds ratio, 0.74; 95 confidence interval, ). In subgroup analysis of studies that considered ovulatory dysfunction to be the infertility reason, the results favored women who received hcg. In contrast, across studies that reported male factor as the infertility reason, as well as across studies including women with unexplained infertility, results appeared to favor the LH surge detection approach. However, none of those subgroup analyses reached statistical significance. Conclusion(s): Available data do not demonstrate a consistent, clinically important benefit of hcg-induced ovulation compared with spontaneous ovulation for IUI timing. (Fertil Steril 2007;87: by American Society for Reproductive Medicine.) Key Words: Clomiphene citrate, intrauterine insemination, LH surge, hcg administration, male infertility, ovulatory dysfunction, unexplained infertility Intrauterine insemination is a first method of treatment for many causes of infertility, mainly unexplained infertility, male infertility, and ovulatory dysfunction. Insemination follows successful ovulation induction with clomiphene citrate. The clinician can follow two tactics for intrauterine insemination (IUI) timing: hcg administration, with IUI after hours, or LH recording, followed after 24 hours by IUI. Many biological reasons exist for defining which is the best method for insemination timing after clomiphene citrate ovulation induction. Received March 22, 2006; revised September 23, 2006; accepted October 6, Reprint requests: Ioannis P. Kosmas, M.D., Clinical and Research Fellow, Centre for Reproductive Medicine, Laarbeeklaan 101, 1090 Brussels, Belgium (FAX: ; kosmasioannis@gmail.com). The exact time at which ovulation occurs after LH surge begins cannot be known earlier. It varies from 24 to 56 hours. Oocyte-fertilization capacity and sperm lifetime are 1 day and 1.4 days, respectively. Insemination needs to be performed close to ovulation time, and accurate synchronization is compulsory. The LH surge can occur in various follicular sizes (1), and individual follicular maturation adds to the risk of trial failure. Urinary LH recording may present false-negative results when peak LH concentrations are low ( 40 IU/L). When a patient receives hcg, follicular rupture takes place hours after injection. A certain follicle diameter is required for hcg administration. Follicle diameter as a single measurement for ovulatory status may be misleading (2). During the follicular phase, follicles grow at a linear rate, but the eventual size is not predictive for the time of ovulation. However, with hcg administration, there is the /07/$32.00 Fertility and Sterility Vol. 87, No. 3, March 2007 doi: /j.fertnstert Copyright 2007 American Society for Reproductive Medicine, Published by Elsevier Inc. 607

2 theoretical risk of an increase in the release rate of immature oocytes or of luteinized unruptured follicles (3, 4). Versatility of hcg administration and better timing for IUI has led to this method, which widely is accepted for final oocyte maturation. Pelvic ultrasound for follicular-development recording is compulsory with this method. This adds significantly to cost and to the human resources needed for cycle observation. To allow highly accurate usage of the LH surge for monitoring, measurements must occur every 3 hours. The need for accurate timing for IUI may pose difficulties (3). However, at a clinical level, once-daily morning LH measurements are satisfactory, when the follicle approaches sonographic maturity. This meta-analysis reports the results after data synthesis from seven studies with patients who have undergone IUI after clomiphene citrate ovulation induction. This synthesis is essential in understanding whether different IUI timing methods affect pregnancy rates and indicates a framework for choosing the proper method for dealing with different infertility causes. MATERIALS AND METHODS Identification and Eligibility of Relevant Studies We used MEDLINE searches (through February 2006) of various combinations of terms: clomiphene citrate, hcg, LH surge, IUI, ovulation timing, and pregnancy outcome. The search was complemented with perusal of the bibliographies of retrieved research articles and review articles. Studies included in the current meta-analysis were those that evaluated the comparison of hcg administration and LH surge detection for IUI timing after a clomiphene citrate ovulationinduction regimen. Although unpublished studies cannot be evaluated adequately for their design and quality, meeting proceedings were considered on one occasion (5). Both retrospective and prospective studies were included (crossover and randomized), regardless of size. No language limit was set. Data Extraction For each study, information was obtained about the authors, journal, and year of publication; country and years of study enrollment; study design and eligibility (inclusion exclusion) criteria; number of participating centers; age of the subjects (mean or median); power analysis mentioned; and number of cases and controls for each group. Outcome measures for IUI included only clinicalpregnancy rates. Subgroup analysis was performed for patients with male infertility, ovulatory dysfunction, and unexplained infertility. Data extraction was performed independently by two investigators. Meta-Analysis For dichotomous or categorical outcomes, the clinicalpregnancy rates in the hcg-administration group and in the LH-detection group were expressed as an odds ratio, with 95 confidence intervals. This was achieved by using a random effects model meta-analysis. Analyses were performed by using RevMan Analyses (2003; Cochrane Collaboration) and by using SPSS 12.0 (SPSS, Inc., Chicago, IL). RESULTS Eligible Studies Forty-nine abstracts were retrieved and further screened. From these, all articles that attempted to compare hcg administration with urinary LH recording for IUI timing, after a clomiphene citrate regimen, were retrieved for full review (n 9). One study (3) was excluded because it had used as controls women who had undergone monitoring for insemination timing, other than LH measurement. A second study (6) was excluded because it had used as controls women who had undergone insemination after a natural cycle. Seven studies therefore were eligible for the metaanalysis (Fig. 1). Of these seven studies, six recruited women in the United States, and one, in Europe (Netherlands). Five of these studies were published in Fertility and Sterility, and two, in Human Reproduction. Mean age of participating women ranged between 31 and 35 years. Only one study mentioned the ethnic origin of recruited patients. Entered women were enrolled during the previous decade, except in two studies. Four studies were retrospective, two studies were prospective with crossover design, and one was a randomized controlled trial. There were no multicenter studies. Five studies mentioned inclusion and exclusion criteria. Recruitment patterns and eligibility for cases were variable, but typically, women with ovulatory dysfunction and unexplained infertility, as well as couples with male in- FIGURE 1 Flow diagram for meta-analysis: abstract identification and selection. Kosmas. HCG vs. LH for IUI timing, after CC: a meta-analysis. Fertil Steril Kosmas et al. HCG vs. LH for IUI timing, after CC: a meta-analysis Vol. 87, No. 3, March 2007

3 fertility were eligible. Only four studies mentioned power calculations. Recording of LH was performed consistently across studies, by LH detection in urine. Urine LH of 40 to 50 IU/L was considered to be the lower LH detection threshold in the studies mentioned. Only one study (1) confirmed urinary LH kit results for ovulation surge with ultrasound and serum LH measurement. Meta-Analysis: Main Results We analyzed 1,461 patients who were administered hcg, as well as 1,162 conservatively monitored patients who served as controls. In the overall and subgroup analysis, there was no significant between-study heterogeneity for pregnancy rates. Summary estimates found a difference between hcg administration and conservative monitoring (odds ratio, 0.74; 95 confidence interval, ), favoring LH recording. Smaller prospective crossover trials showed more favorable results for LH monitoring when compared with larger retrospective trials. The randomized controlled trial showed no significant difference between the two groups. Summary estimates did not change significantly as data were collected over time (Fig. 2A). Subgroup Analyses for Unexplained and Male Infertility and for Ovulatory Dysfunction Three comparisons included patients with male infertility, unexplained infertility, and ovulatory dysfunction (1, 7, 8). Pooled data for these three specific infertility groups showed contradictory results. In the male-infertility group, LH recording for IUI timing appears to be more useful than hcg administration (odds ratio, 0.66; 95 confidence interval, ; Fig. 2B). The same association is observed in the unexplained-infertility group (odds ratio, 0.79; 95 confidence interval, ; Fig. 2D). However, in ovulatorydysfunction patients (1, 8), hcg administration achieves better pregnancy rates than does LH recording (odds ratio, 2.00; 95 confidence interval, ; Fig. 2C). In all of the three different infertility groups, results did not reach statistical significance. DISCUSSION We found a significant difference in pregnancy rates between the hcg-administration group and the LH-recording group that favored LH ovulation monitoring. This metaanalysis includes seven trials, with 2,481 subjects. Although pooled data for all infertility groups showed that hcg administration achieves lower pregnancy rates than does LH recording before IUI, these results cannot be considered definitive. With regard to the hcg-administration group, we observed a trend for higher pregnancy rates for patients undergoing IUI for ovulatory dysfunction, although this did not reach statistical significance. Opposite trends were observed for the two other subgroups, male factor and unexplained infertility. In these subgroups, no statistical significance was reached either. Because of the randomization and short number of studies included, final conclusions cannot be drawn from these results. In this meta-analysis, there was no between-study heterogeneity for the primary outcome. The results of the clinical-pregnancy rates in this metaanalysis are consistent with those in the studies by Martinez et al. (9), Ling et al. (5), and Lewis et al. (10). This meta-analysis has its limits. It consists mainly of retrospective studies and prospective studies with crossover design. Only one study is a randomized controlled trial. With crossover design, patients receive their treatments simultaneously. Misleading results may arise because if pregnancy is achieved, patients are not exposed to the second treatment. If a patient drops out, this may affect the validity of study design. However, retrospective studies can help interpret the results of randomized trials. Randomized controlled trials with parallel design are the best choice for evaluating infertility treatments (11). These studies need to recruit large numbers of patients and standardize their conduct and reporting. Publication bias results in stronger associations in smaller studies than in larger studies. As a practical indication, the smaller the studies that are conducted in a scientific field, the less likely it is that the research findings are reliable (12). Prospective crossover studies included small numbers of patients, whereas retrospective studies included larger numbers of patients. Although the randomized controlled trial included the number of patients needed to achieve 80 power, larger numbers of patients needed to end in a conclusion. Biological reasons might explain the results of this metaanalysis. Administration of hcg carries the theoretical risk of release of an oocyte from an immature follicle or no release at all, leading to a luteinized, unruptured follicle (3). Coetsier and Dhont (4) observed complete luteinized unruptured follicles in their study. Luteinizing hormone acts as a major paracrine regulator, and when a surge occurs, so does final oocyte maturation (meiosis). Pregnancy rates are better if the LH surge occurs before hcg administration, than when hcg is given before the LH surge (13). The same study reports mean follicular diameter to be significantly smaller in the hcg group when compared with the case of the LH surge group. Selective reporting of outcome data was evident in this meta-analysis. The number of follicles, implantation rates, number of abortions, and number of term pregnancies were not available for all included studies. More rigorous reporting of future clinical research on hcg administration com- Fertility and Sterility 609

4 FIGURE 2 (A) Overall results of hcg administration vs. LH monitoring for IUI timing after a clomiphene citrate regimen and the odds for pregnancy. (B) Subgroup analysis for male infertility after hcg administration vs. LH monitoring for IUI timing in a clomiphene citrate regimen and the odds for pregnancy. (C) Subgroup analysis for ovulatory dysfunction after hcg administration vs. LH monitoring for IUI timing in a clomiphene citrate regimen and the odds for pregnancy. (D) Subgroup analysis for unexplained infertility after hcg administration vs. LH monitoring for IUI timing in a clomiphene citrate regimen and the odds for pregnancy. In all panels, each study is shown as an odds-ratio estimate, with whiskers corresponding to its 95 confidence interval, and studies are ordered according to year of publication. 01 clomiphene + hcg vs clomiphene +Lh overall 01 clomiphene + hcg VS clomiphene + LH Treatment Control OR (fixed) Weight OR (fixed) or sub-category n/n n /N Year Martinez 4/43 9/ [0.11, 1.41] 1991 Agarwal 17/247 29/ [0.32, 1.11] 1995 Deaton 10/182 17/ [0.36, 1.78] 1997 Ling 25/420 28/ [0.34, 1.04] 1997 Zreik 2/27 4/ [0.07, 2.53] 1999 Vlahos 58/468 32/ [0.53, 1.34] 2005 Lewis 23/74 17/ [0.65, 2.83] 2006 Total () [0.57, 0.96] Total events: 139 (Treatment), 136 (Control) Test for heterogeneity: Chi² = 5.30, df = 6 (P = 0.51), I² = 0 Test for overall effect: Z = 2.30 (P = 0.02) 02 CC + hcg VS CC+Lh male factor 01 male factor cc + hcg cc+ LH OR (fixed) Weight OR (fixed) or sub-category n/n n/n Year Deaton 0/65 1/ [0.01, 5.94] 1997 Zreik 1/9 0/ [0.12, 93.83] 1999 Vlahos 28/226 18/ [0.33, 1.21] 2005 Total () [0.35, 1.21] Total events: 29 (cc + hcg), 19 (cc+ LH) Test for heterogeneity: Chi² = 1.31, df = 2 (P = 0.52), I² = 0 Test for overall effect: Z = 1.35 (P = 0.18) 03 cc+hcg VS cc+lh ovulatory dysfunction c ovulatory dysfunction cc+ hcg cc+lh OR (fixed) Weight OR (fixed) or sub-category n/n n/n Year Deaton 4/36 6/ [0.35, 4.99] 1997 Zreik 0/19 0/18 Not estimable 1999 Vlahos 14/57 4/ [0.81, 8.92] 2005 Total () [0.84, 4.77] Total events: 18 (cc+ hcg), 10 (cc+lh) Test for heterogeneity: Chi² = 0.61, df = 1 (P = 0.43), I² = 0 Test for overall effect: Z = 1.56 (P = 0.12) 04 cc+ hcg VS CC+LH unexplained infertility 01 unexplained infertility cc+ hcg cc+lh OR (fixed) Weight OR (fixed) or sub-category n/n n/n Year Deaton 6/81 10/ [0.35, 2.84] 1997 Zreik 2/42 3/ [0.11, 4.31] 1999 Vlahos 6/72 6/ [0.19, 2.06] 2005 Total () [0.38, 1.64] Total events: 14 (cc+ hcg), 19 (cc+lh) Test for heterogeneity: Chi² = 0.36, df = 2 (P = 0.84), I² = 0 Test for overall effect: Z = 0.64 (P = 0.52) Kosmas. HCG vs. LH for IUI timing, after CC: a meta-analysis. Fertil Steril Kosmas et al. HCG vs. LH for IUI timing, after CC: a meta-analysis Vol. 87, No. 3, March 2007

5 pared with LH recording after clomiphene citrate should be encouraged. Clomiphene citrate doses varied between studies from mg/d. Only two studies (1, 8) mentioned individualization of regimen. All studies except one mentioned criteria for hcg administration (14). In four studies (10, 1, 7, 9), mean follicle diameter during hcg administration was 20 mm. In all studies, 10,000 IU of hcg were administered. Whether insemination is timed for 24 or 36 hours after hcg administration may affect pregnancy rates, with the 36-hour choice favored (15). In all studies, for that group, insemination was performed in an interval that varied from 36 to 42 hours. For the LH-monitoring group, insemination was performed after hours. Three studies in the LH group (7, 8, 10) and one study in the hcg group (7) did not mention insemination timing. Only Deaton et al. (8) mentioned that luteal support was not given. All other investigators did not mention luteal support. No studies mentioned sperm sample preparation details. Only Vlahos et al. (1) mentioned total motile sperm inseminated in both groups, and total volume inseminated was mentioned only by Martinez et al. (9) and Agarwal and Buyalos (14). Also, it was not recorded in all studies whether intercourse was allowed. Donor sperm was excluded in the study by Vlahos et al. (1) but was included by Lewis et al. (10). Administration of hcg for IUI timing has been accepted as the standard of care for patients who are undergoing insemination for various infertility causes. A well-proven biological base, clinical predictability, and ease of planning are some of the major advantages of this method. Disadvantages include the cost of the medication and the time-consuming ultrasonographic follicle monitoring, which add to the indirect costs of the method. However, urinary LH measurement provides a costeffective and patient-friendly way of recording follicular development and ovulation after clomiphene citrate administration. Despite all physiological and timing drawbacks that may exist for this method, this meta-analysis shows a trend toward better pregnancy rates in unexplained-infertility and male-infertility patients for the LH-monitoring group. Unfortunately, these positive trends cannot be evaluated fully. When these meta-analysis caveats are taken into account, it cannot be assessed at this stage which is the best method for IUI timing after clomiphene citrate induction. Mode of randomization, small numbers of patients, and flexibility in study designs may have contributed to a falsely significant association. No financial interests were mentioned, although such considerations may have tipped the balance in favor of hcg administration. There is a need for large, parallel-design, randomized trials, to focus on the previous question, that will allow association of pregnancy rates with these two tactics for IUI timing in a clomiphene citrate ovulation-induction protocol. In the meantime, and until enough evidence is accumulated, LH for IUI timing remains an equally effective but more practical approach. This meta-analysis shows that there is not enough evidence to choose hcg administration rather than LH measurements. One may expect higher pregnancy rates for hcg administration timing, but yet there is not enough evidence to sustain this asumption. The LH hormone measurement procedure shows a trend toward better pregnancy rates, at least in unexplained-infertility and male-infertility patients, and daily morning measurements must be considered a good solution for IUI timing. In the search for increasing results, a combination of the two protocols may minimize the biological drawbacks while achieving higher pregnancy rates (16). In a retrospective study, investigators trying to find the best timing method for IUI observed that a combination of hcg plus LH surge achieved a significantly higher clinical-pregnancy rate in CC cycles, when compared with the groups using LH or hcg alone. Further research should identify the ideal interval between LH surge and IUI, thereby defining the scheduling for this method. REFERENCES 1. Vlahos NF, Coker L, Lawler C, Zhao Y, Bankowski B, Wallach EE. Women with ovulatory dysfunction undergoing ovarian stimulation with clomiphene citrate for intrauterine insemination may benefit from administration of human chorionic gonadotropin. Fertil Steril 2005;83: Messinis IE, Templeton A. Urinary oestrogen levels and follicle ultrasound measurements in clomiphene induced cycles with an endogenous luteinizing hormone surge. Br J Obstet Gynaecol 1986;93: Check JH, Peymer M, Zaccardo M. Evaluation of whether using hcg to stimulate oocyte release helps or decreases pregnancy rates following intrauterine insemination. Gynecol Obstet Invest 1994;38: Coetsier T, Dhont M. Complete and partial luteinized unruptured follicle syndrome after ovarian stimulation with clomiphene citrate/human menopausal gonadotrophin/human chorionic gonadotrophin. Hum Reprod 1996;11: Ling PY, Maier D, Thornton K, Nulsen JC. Comparison of pregnancy rates in clomiphene citrate (CC) ovarian stimulation cycle combined with intrauterine insemination (IUI) in which ovulation occurred spontaneously or was induced with human chorionic gonadotropin (hcg). Fertil Steril 1997;68(Suppl 1):S4. 6. Arici A, Byrd W, Bradshaw K, Kutteh WH, Marshburn P, Carr BR. Evaluation of clomiphene citrate and human chorionic gonadotropin treatment: a prospective, randomized, crossover study during intrauterine insemination cycles. Fertil Steril 1994;61: Zreik TG, Garcia-Velasco JA, Habboosh MS, Olive DL, Arici A. Prospective, randomized, crossover study to evaluate the benefit of human chorionic gonadotropin-timed versus urinary luteinizing hormonetimed intrauterine inseminations in clomiphene citrate-stimulated treatment cycles. Fertil Steril 1999;71: Deaton JL, Clark RR, Pittaway DE, Herbst P, Bauguess P. Clomiphene citrate ovulation induction in combination with a timed intrauterine Fertility and Sterility 611

6 insemination: the value of urinary luteinizing hormone versus human chorionic gonadotropin timing. Fertil Steril 1997;68: Martinez AR, Bernadus RE, Voorhorst FJ, Vermeiden JP, Schoemaker J. A controlled study of human chorionic gonadotrophin induced ovulation versus urinary luteinizing hormone surge for timing of intrauterine insemination. Hum Reprod 1991;6: Lewis V, Queenan J, Hoeger K, Stevens J, Guzic D. Clomiphene citrate monitoring for intrauterine insemination: timing a randomized trial. Fertil Steril 2006;85: Daya S. Is there a place for the crossover design in infertility trials? Fertil Steril 1993;59: Ioannidis JP. Why most published research findings are false. PLoS Med 2005;2:e Fuh KW, Wang X, Tai A, Wong I, Norman RJ. Intrauterine insemination: effect of the temporal relationship between the luteinizing hormone surge, human chorionic gonadotrophin administration, and insemination on pregnancy rates. Hum Reprod 1997;12: Agarwal SK, Buyalos RP. Corpus luteum function and pregnancy rates with clomiphene citrate therapy: comparison of human chorionic gonadotrophin-induced versus spontaneous ovulation. Hum Reprod 1995;10: Robb PA, Robins JC, Thomas MA. Timing of hcg administration does not affect pregnancy rates in couples undergoing intrauterine insemination using clomiphene citrate. J Natl Med Assoc 2004;96: Mitwally MF, bdel-razeq S, Casper RF. Human chorionic gonadotropin administration is associated with high pregnancy rates during ovarian stimulation and timed intercourse or intrauterine insemination. Reprod Biol Endocrinol 2004;2: Kosmas et al. HCG vs. LH for IUI timing, after CC: a meta-analysis Vol. 87, No. 3, March 2007