577 PREDICTORS OF FETAL LOSS (FL) FOLLOWING CHORIONIC VILLUS SAMPLING (CVS) ANTHONY ODIBO 1 , DAVID STAMILIO 1 , BARBARA OBERLE 1 , JEFFREY DICKE 1 , DIANA GRAY 1 , GEORGE MACONES 1 , 1 Washington University in St. Louis, St. Louis, Missouri OBJECTIVE: To identify risk factors associated with FL following CVS STUDY DESIGN: Retrospective cohort study including all women undergoing chorionic villus sampling at a single center over a 16-year period. Those having amniocentesis or pregnancy termination were excluded from this analysis. FL was defined as any loss prior to 24 weeks. Univariate and Logistic regression analyses were used to compare pregnancies resulting in fetal loss to those without a loss and to adjust for potential confounders between the groups. Forward and backwards stepwise logistic regression was used to determine the final prediction model for FL. The accuracy of the prediction models was compared using the area under the ROC curves. RESULTS: Among 5614 women who underwent CVS over the study period, follow-up outcome was obtained in 5487 (98%). FL occurred in 108/5487 (1.97%). Variables compared included maternal socio-demographic characteristics, indica- tions for CVS, route (transcervical (TC) versus transabdominal (TA)) number of insertions or aspirations on CVS and complications during CVS. The final predic- tors of FL are shown in the table. The AUC for the final model was 0.57. FL rates did not differ by CVS route (1.9% in TC and 2.6% in TA, p=0.48). CONCLUSION: Significant predictors of fetal loss following CVS include mater- nal race, multiple aspirations/insertions and excessive bleeding at the time of the procedure. The final prediction model was however only modestly accurate in predicting FL. Final prediction model for SFL following CVS Factors Adjusted OR 95% CI P-value Black race 2.1 1.0-4.4 0.047 2 or more aspirations/insertions 2.2 1.3-3.7 0.003 Heavy bleeding during CVS 3.1 1.5-6.4 0.002 0002-9378/$ - see front matter doi:10.1016/j.ajog.2007.10.601 578 CORRELATION OF CVS MORPHOLOGY WITH PREGNANCY OUTCOME AND KARYOTYPIC ABNORMALITIES VICTORIA BELOGOLOVKIN 1 , LAUREN FERRARA 2 , CLAIRE MCCLUNG 2 , LISA EDELMANN 3 , ERIN MOSHIER 4 , KATE JANDL 2 , KEITH A. EDDLEMAN 2 , JOANNE STONE 2 , 1 University of South Florida, Tampa, Florida, 2 Mount Sinai School of Medicine, New York, New York, 3 Mount Sinai School of Medicine, Department of Human Genetics, New York, New York, 4 Mount Sinai School of Medicine, Com- munity and Preventive Medicine, New York, New York OBJECTIVE: To identify morphologic characteristics of chorionic villi as a po- tential predictor of pregnancy outcome and karyotypic abnormality STUDY DESIGN: Patients undergoing CVS were prospectively enrolled, n =187 singletons and n=55 twins. The morphologic characteristics of the villi were recorded and pregnancy outcomes collected. Information on karyotype and maternal and neo- natal characteristics were collected. The villi were evaluated for budding, fluid retention, opacity, diameter and vascularity. The association of villi morphology on the develop- ment of gestational HTN, preeclampsia, GDM and birth weight, as well as the associa- tion with abnormal karyotype, were evaluated using adjusted logistic regression. T-tests were used to compare means. Chi-Square and Fisher=s Exact were used to evaluate differences in proportions were appropriate. RESULTS: Maternal characteristics are presented in Table 1. The presence of villi edema and poor budding was present more often in those with abnormal karyotype with OR 12.5 (3.8, 42.1) and OR 4.0 (1.4, 11.3) respectively. Tri 21 was most common in 33 % of abnormals followed by tri 18 in 16%. Thickest villi diameter was positively associated with birth weight (p=0.07). There was no asso- ciation between villi characteristics and GHTN, preeclampsia or GDM after adjust- ing for potential confounders. CONCLUSION: The morphologic characteristics of villus edema and abnormal budding is associated with increased likelihood of chromosomal abnormality. When these characteristics are present, FISH may be considered for more rapid diagnosis of aneuploidy in these patients. Maternal characteristics Singleton (n = 187) % Twins (n=55) % P Age  SD 37.8  3.7 38.1  4.0 0.7 Race-white 88 79 0.2 Parity 1 56 33 0.01 ART 15 94 0.01 GHTN 7 5 0.7 Prec 6 1.8 0.8 GDM 5 4 0.3 0002-9378/$ - see front matter doi:10.1016/j.ajog.2007.10.602 579 OVERCOMING THE BARRIER OF NON SPECIFICITY OF FETAL CELL ANTIGENS: I MPROVED YIELD OF CIRCULATING FETAL CELLS UTILIZING AUTOMATED MICROSCOPY FOR DETECTION OF ANEUPLOIDY MARK EVANS 1 , ANTTI SEPPO 2 , VERONIKA FRISOVA 3 , YOUNGMIN KIM 2 , TRIANTAFYLLOS TAFAS 2 , KYPROS NICOLAIDES 3 , PETROS TSIPOURAS 2 , MICHAEL W KILPATRICK 2 , 1 Comprehensive Genetics, New York, New York, 2 Ikonisys Inc., New Haven, Connecticut, 3 Harris Birthright Research Centre for Fetal Medicine, King’s College Hospital, London, United Kingdom OBJECTIVE: Fetal cell detection has been hampered by lack of specific fetal antigens. Attempts to get around this roadblock have taken many, largely unsuc- cessful, approaches. Here we attempt to maximize fetal cell yield by automated microscopy combined with dual labeling. STUDY DESIGN: We utilized an automated microscopy system developed to identify and enumerate rare cells. For identification based on fetal hemoglobin expression, fetal cells are verified based on anti-gamma or anti-epsilon antibody signals, and FISH signals for the X and Y chromosomes. For FISH-based scanning, verified fetal cells are identified based on a dual FISH probe labeling approach. Previously we showed that dual labeling reduces the false positive rate below 0.00005% when scanning for rare nuclei. Fetal nuclei are identified based either on the presence of a Y chromosome or aneuploid FISH signals for chromosome 21, and verified at high mag utilizing two FISH probes for the chromosome of interest. RESULTS: Identification based on fetal hemoglobin expression did not allow de- tection of fetal erythroblasts in numbers sufficient for clinical diagnosis. However, FISH-based scanning identified fetal cells in 28 out of 29 maternal samples ( 2 = 46.6, p.001). On average, 2.3 fetal cells per million nucleated maternal cells were detected. Antibody based fetal cell identification FISH based fetal cell identification Cases Cell identifier Fetal cell positive cases Fetal cells/ case Cases Cell identifier Fetal cell positive cases Fetal cells/ case 10 -Hg 1 0.1 (0–1) 11 (1 st trimester) Y chromosome 11 8 (1–20) 12 -Hg 0 0 18 (2 nd trimester) Y chromosome 17 7 (0–17) CONCLUSION: Our data demonstrate that automated microscopy was able to detect fetal cells in greater than 95% of maternal samples, both first and second trimester, utilizing dual FISH probes for the chromosome of interest. This suggests that automated scanning for aneuploid FISH signals could form the basis of a credible clinical test for non-invasive prenatal diagnosis, eliminating the previous roadblock of the need for a fetal cell specific biomarker. 0002-9378/$ - see front matter doi:10.1016/j.ajog.2007.10.603 580 CHANGING THE PARADIGM: BOTH FISH AND KARYOTYPE SHOULD BE ROUTINE FOR PRENATAL DIAGNOSIS MARK EVANS 1 , STEPHANIE ANDRIOLE 1 , EUGENE PERGAMENT 2 , 1 Comprehensive Genetics, New York, New York, 2 Northwestern University, Chi- cago, Illinois OBJECTIVE: To assess the differential impact of classic cytogenetics and FISH for both CVS and amniocentesis (AMN) specimens. STUDY DESIGN: Outcomes of 4942 specimens were compared by type and in- dication from one lab over the past 3 years in which FISH (13,18,21,X,Y) was performed routinely along with prenatal karyotypes. RESULTS: FISH and karyotype results for both CVS and AMN are highly accu- rate but neither are perfect. True positives were seen in 79/2574 (3.1%) CVS and 37/2574 (1.6%) AMN. CVS detection after positive 1st TM screen was 12/133 = 9.0% and 2nd TM screen AMN was 8/427 (1.9%). There were a total of 35 discrep- ant cases with no false positive somatic FISH, and 16 false negative CVS (5 aneu- ploid; 11 structural) (0.32%) which were resolved by karyotype. There were 15 CVS and 4 AMN cases in which the FISH correctly predicted genotype, and the karyo- type did not correctly represent the fetal genotype. CONCLUSION: CVS had a higher yield of abnormalities (both AMA and screen- ing) than AMN. Even with the inherent limitations of only 5 chromosome FISH, there were more cases (19) for both CVS and AMN specimens for which ultimately the FISH was more accurate than either the CVS or AMN karyotype. This compares to 16 cases when the karyotype was more accurate than the FISH. Overall 35 dis- crepant cases (0.7%) required both methods to determine the correct result. Since its introduction 15 years ago, FISH has been viewed as an adjunct to karyotyping as gold standard. Consideration should be given to the proposal that there should now be two parallel investigations (FISH and karyotype) with laboratory and clinical judgment required to achieve optimal results FISH & Karyotype PROC True Neg False Pos FISH corr Kary corr True Pos Total CVS 2466 0 15 14 79 2574 AMN 2325 0 4 2 37 2368 0002-9378/$ - see front matter doi:10.1016/j.ajog.2007.10.604 www.AJOG.org SMFM Abstracts Supplement to DECEMBER 2007 American Journal of Obstetrics & Gynecology S167