P-112

Clomiphene and intrauterine insemination (IUI)—What is the best wayto time insemination? Vivian Lewis, David S. Guzick. Univ of RochesterSch of Medicine and Dentistry, Rochester, NY.

Objective: Clomiphene citrate and IUI is frequently used to increasefecundity in couples with unexplained infertility. However, studies compar-ing methods of IUI timing are either retrospective or have inconsistentresults. We tested the hypothesis of whether pregnancy rates in womentreated with clomiphene and IUI would be different depending on whetherthe IUI was timed by spontaneous LH surge or hCG administration. Apositive pregnancy test with rising hCG concentrations was the main out-come measure.

Design: Prospective randomized clinical trialMaterials/Methods: Subjects were ovulatory women with at least one

patent tube and a partner or sperm donor with at least 4 million normalmotile sperm per ejaculate. Randomization was performed at the baselineultrasound, after which patients took 100 mg of clomiphene, days 5–9. Thesurge group began testing for LH surge using an over-the-counter kit on day12 and underwent IUI on the day after a positive test. Patients who failed todetect a surge were dropped from the study. The hCG group returned forultrasound starting on day 12 and received hCG when there was at least 1follicle measuring 20 mm mean diameter and when the endometrial thick-ness was at least 8 mm with a trilaminar pattern. If the LH kit becamepositive before ultrasound criteria were met, IUI was scheduled accordingly.Patients remained in the same study group in subsequent months for amaximum of 3 cycles. An intent to treat approach was used in analyzing thedata so that information from dropped cycles could be captured.

Results: Seventy-nine patients have completed 167 cycles with a total of19 pregnancies: 6/86 in the surge group and 13/81 in the hCG group (p �.075,chi-square). There were 4 viable pregnancies among 40 patients in thesurge group and 10 among 39 patients in the hCG group (p � . 054, Fisherexact test). Mean age and proportion of patients with male factor or tubalfactor were not different. There were 15 patients who were dropped from thesurge group and 8 from the hCG group, with failure to detect the LH surgebeing the most common reason.

Conclusions: These data suggest a trend toward higher pregnancy rateswith hCG to trigger ovulation for IUI. These findings could be partly due toa lower chance of cycle cancellation, but could also relate to oocytematuration, corpus luteum function or endometrial readiness for implanta-tion. Clearly, the small sample size precludes justification for routine use ofhCG given the higher cost. Subject accrual is continuing which may confirmor extend these data.

Supported by: Medications provided by Serono Laboratories.

P-113

Induction of ovulation and ovarian cancer: A meta-analysis. SonyaKashyap, David Moher, Micheal Fung Kee Fung. Ctr for ReproductiveMedicine, Weill Medical Coll, Cornell Univ, New York, NY; Univ ofOttawa, Ottawa, ON, Canada.

Objective: To evaluate the current available data regarding induction ofovulation and the incidence of ovarian cancer.

Design: Meta analysisMaterials/Methods: A systematic review and meta analysis has been

conducted. A comprehensive, literature search employing the followingdatabases was done: Premedline, Medline, Cancerlit, Cinahl, Current con-tents and Biological abstracts. The search was not limited by language oryear of publication. All citations were reviewed for relevance. The finalgroup of citations was then reviewed for inclusion/exclusion criteria, qualityassessment and data abstraction. This process has been completed by twoindependent reviewers. Discrepancies between reviewers have been re-solved by consensus. Citations were supplemented by references pulledfrom review articles and studies as well as hand searches (10 years) of 5 keyjournals. We also searched for grey/unpublished literature by consultationwith content experts, reviews of conference proceedings, and we attemptedcontact with authors. Inclusion/Exclusion criteria included: outcome: pri-mary incident ovarian cancer of any histological type; population: womenwho had an explicit and reproducible diagnosis of infertility and these caseswere to be compared with fertile and infertile controls; intervention/expo-sure: induction of ovulation by the following medications: clomiphenecitrate, human menopausal gonadotropin, and gonadotropin releasing ago-

nists. Data analyses was done using Revman 4.1 and Metaview 4.0. Cohortand case control data were analysed separately and infertile controls wereused as the reference group where possible. Both fixed and random effectsmodels were used for reasons identified in the manuscript

Results: Original search revealed 2114 articles, which were reduced to947 when limited to human, adult female. Relevance filter and manualremoval of duplicates reduced the number further to 94. Inclusion/ exclu-sion criteria further reduced these numbers to 38. These also included :Search for unpublished literature revealed 2 studies whose results were atleast partly published subsequently. Hand searching revealed 3 extra stud-ies. When multiple publication occurred only the most recent or completepapers were included. Precision of the search strategy was 1.6% and recallwas 83%. In the final quantitative analysis 3 cohort studies and 7 casecontrol studies were included. Case control data comparing infertile treatedwomen with general population controls:OR 1.52 (1.16, 2.01); Case controldata comparing infertile treated with infertile untreated: OR 1.00 (0.68,1.46); Cohort data comparing infertile treated to infertile untreated : RR0.66 (0.31, 1.41).

Conclusions: These data do not support an increased risk of ovariancancer as a result of induction of ovulation. In fact, fact the results arereassuring if not encouraging. There is a trend towards protection of infertilewomen from ovarian cancer via induction of ovulation.

Supported by: None.

P-114

Tissue sampling technique affects accuracy of karyotype from missedabortions. Ruth Bunker Lathi, Amin A. Milki. Stanford Univ Sch ofMedicine, Stanford, CA.

Objective: A high rate of 46XX results has been reported with cytoge-netic analysis of products of conception (POC) from first trimester missedabortions. This study evaluates whether a careful technique for villi selec-tion by the clinician, prior to specimen submission, will decrease falsenegative results caused by maternal contamination.

Design: Retrospective reviewMaterials/Methods: Since 1998, all patients with a missed abortion in the

senior author’s (AAM) infertility practice were offered cytogenetic testingof the POC obtained by suction currettage. Prior to July 1999(group A n �15), POC were drained of blood then divided into a sample sent forhistopathologic diagnosis and a sample for chromosomal testing. Since July1999(group B n � 40), the technique for choosing a sample for geneticanalysis was changed in an attempt to improve diagnostic accuracy. ThePOC were drained of blood and rinsed in a saline basin then placed in aclean saline basin and carefully examined to identify chorionic villi. Asample of villi was dissected clear from other tissue using forceps andscissors then washed again and sent for chromosomal analysis. Cytogeneticsreports from other physicians using the same lab during this time periodwere used for comparison (group C, n � 59).

Results: The percentage of 46XX results was significantly decreased inthe test group when compared to historical and community controls: 30% vs73% and 56% respectively. The percentage of aneuploid results was sig-nificantly higher in the test group: 60% vs 7% and 36% in the historical andcommunity controls respectively. The mean ages were not significantlydifferent.

Study group

Group A(n � 15)

Group B(n � 40)

Group C(n � 59)

Number 46XX (%)* 11 (73%) 12 (30%) 33 (56%)Number abnormal (%)** 1 (7%) 24 (60%) 21 (36%)Number 46XY 3 (20%) 4 (10%) 5 (8%)Mean Age 37.6 37.5 36.8

* B vs A p � 0.005, B vs C p � 0.014, B vs A � C p � 0.004** B vs A p � 0.0005, B vs C p � 0.02, B vs A � C p � 0.003

Conclusions: Although the cytogenetics laboratory personnel routinelyattempts to isolate villus or fetal material from the submitted sample, a highproportion of 46XX results was seen in both groups where the techniquedescribed in this report was not applied. Thorough separation and cleaningof villi performed prior to sending missed abortion specimens allows

S154 Abstracts Vol. 78, No. 3, Suppl. 1, September 2002