Prenatal diagnosis by chorionic villus sampling (CVS; 10–12 menstrual weeks) or amniocentesis (15–18 menstrual weeks) is chosen by many women at increased risk of fetal genetic abnormality, despite the low, but real, risk of pregnancy loss each procedure carries.1–5 In contrast, women at lower risk increasingly use screening procedures that can inform them of an increased risk of genetic abnormality as early as 11–14 weeks of gestation, when both CVS and amniocentesis can be offered for a definitive diagnosis. Several observational studies of amniocentesis claim safety at 13–14 weeks that is comparable with traditional performance at 16–18 weeks,6–8 and a similar statement applies to “late” CVS9,10; however, no formal evaluation has been undertaken.
This study evaluated the relative safety and efficacy of these procedures in a randomized trial of “late” CVS and “early” amniocentesis, each performed at a comparable late first-trimester gestational age. The results should provide important information for women and their obstetricians facing the choice of a diagnostic procedure in this gestational period after first-trimester screening or pre-existing increased genetic risk.
MATERIALS AND METHODS
The protocol was approved by the Institutional Review Boards at the 14 participating clinical centers. Inclusion criteria included women seeking invasive procedures for advanced maternal age (> 34 years of age at enrollment) or a previous fetal trisomy or women having a positive first-trimester noninvasive aneuploidy screen. Gestational age was determined by ultrasound. If a menstrual dates/ultrasound discrepancy was greater than 8 days, a subsequent scan demonstrating normal interval growth became an inclusion requirement.
Exclusion criteria were multifetal pregnancy, other indications for invasive prenatal diagnosis (eg, familial chromosomal rearrangements, inherited enzyme disorders), and serious maternal illnesses (eg, insulin-dependent diabetes, severe hypertension, human immunodeficiency virus), bleeding equivalent to menstruation, intrauterine device in situ, oligohydramnios, or a recognized fetal abnormality.
Data from an earlier study5 identified a combined 2.7% rate of fetal loss and delivery before 28 weeks after transabdominal CVS in women with cytogenetically normal pregnancies. To detect a 50% increase in the early amniocentesis group to 4%, an initial sample size of 6,200 cytogenetically normal pregnancies was estimated (α = .05 2-sided; β = .20). We were unable to reach our sample size goal within the funding period, in part because of an obligatory narrowing of the gestational age range for study eligibility (described below).
The randomization sequence, stratified by clinical center, was generated by the Data Coordinating Center using the urn method,11 which assures unpredictability and balance in treatment assignment.
Initially the procedures were to be performed between 11 and 14 weeks (77–104 days). However, protocol revisions first eliminated week 11 (before recruitment began) and subsequently week 12 after reports indicated an increased risk of talipes equinovarus after early amniocentesis in those weeks.12,13 Thus, more than 90% of the procedures were performed at 13–14 weeks (91–104 days).
Immediately before randomization, eligible patients underwent ultrasound evaluation by the procedure operators to confirm fetal viability and to assure that uterine and placental position and amniotic fluid volume made both early amniocentesis and late CVS procedures feasible. If so, subjects were randomized to 1 of the 2 study procedures by telephone through an interactive voice response computer-based system.
To participate in the trial, operators were required to have completed at least 25 amniocenteses and 25 transabdominal CVS between 77 and 104 days of gestation. Thirty-two operators were certified to perform procedures at the 14 clinical centers. Continuous transabdominal ultrasound guidance was used. Two sampling passes were allowed; a second procedure, if required, could only be performed 7 days after the first attempt. Early amniocentesis was performed with a 22-gauge spinal needle, withdrawing 1 mL of fluid for each week of gestation. Transplacental procedures were permitted. Late CVS was accomplished with either a single (19 or 20 gauge) or double-needle technique (18, 20 gauge), with the larger “guide” needle introduced to the margin of the chorion followed by the sample needle passing through the guide needle into the villi.14
Standard protocols at North American centers included metaphase analysis of 15 cells, with detailed chromosome analysis of at least 2 cells. At the Danish center, 10 cells were counted, with detailed analysis of 3 cells. At all centers, cells were derived from at least 5 colonies from in situ cultures, 2 flask cultures, or a combination of culture and direct preparations, provided no more than half of the cells were derived from direct preparation. Sample adequacy was determined by each laboratory; specimens with insufficient material for diagnosis were classified as failed procedures. Laboratory failure was defined as an acceptable sample that failed to yield a cytogenetic diagnosis.
The primary outcome for the trial was a composite of fetal loss or preterm delivery before 28 weeks (196 days) among cytogenetically normal pregnancies. Preterm delivery before 28 weeks was included as part of the primary outcome because its relationship to premature membrane rupture could represent a complication of invasive diagnostic procedures. Fetal loss included all spontaneous abortions, induced terminations and stillbirths at less than 28 weeks (a cutoff point that correlates with a historical perception of a point in gestation beyond which early invasive procedures are unlikely to affect pregnancy outcome).
Secondary outcome measures included all fetal loss and neonatal death, oligohydramnios, gestational age at delivery, intrauterine growth restriction, respiratory distress syndrome, limb reduction defects, talipes equinovarus, and other congenital malformations. Accuracy measures included failure to obtain a sample, failure to obtain a laboratory diagnosis, and cytogenetic concordance.
Eligibility and baseline demographic data were collected before randomization and observations regarding sampling technique and immediate complications were recorded at the time of the procedure. Follow-up data were obtained by patient phone contact 14 to 21 days after sampling. Information on newborn infants was obtained both from the obstetrician and the mother through telephone contact within 4 months of delivery. Neonatal hospital records were sought if complications required clarification or if respiratory problems were recorded in infants born at term. Orthopedic surgeons evaluated all suspected cases of talipes equinovarus in the newborn period.
Interim analyses were conducted twice to allow the Data and Safety Monitoring Committee to assess primary outcome, using the Lan and DeMets method15 with a uniform spending function to adjust the type I error for multiple looks at the data. Safety issues were monitored by the Committee on a more frequent basis. Pregnancies were analyzed in the group to which they were randomized. Continuous variables were evaluated between groups using the Wilcoxon rank-sum test, whereas categorical data were compared by the Fisher exact or χ2 tests, as appropriate. Nominal 2-sided P values are reported here for the primary analysis, in addition to secondary and baseline comparisons. SAS statistical software (SAS Institute Inc, Cary, NC) was used for the analysis. The Data Coordinating Center audited protocol adherence, data quality, and loss to follow-up on a continuous basis.
The principal funding source was the U.S. National Institute of Child Health and Human Development (NICHD), which approved the design of the trial. Neither the NICHD nor the Centre for Evaluation and Health Technology Assessment of the Danish National Board of Health (which funded continuation of enrollment in 2001) had any role in data collection, analysis, or interpretation of the data.
From January 1997 to December 2001, 6,370 women were screened for eligibility (Figure 1). Of these, 3,803 eligible women consented to participation, and 3,775 were randomized. Before the eligibility criteria changed, 3 subjects (0.1%) were mistakenly randomized during week 11, and 331 (8.8%) were randomized during week 12. The remaining 2,446 (64.8%) and 995 (26.3%) were randomized during weeks 13 and 14, respectively. Eighty-seven percent were randomized at Rigshospitalet, Denmark, 7% at 11 U.S. centers, and 6% at 2 Canadian centers. Despite intensive efforts to improve recruitment at U.S. centers, there remained a significant lack of patient response. Entrenched procedure preferences among patients and their caregivers (referring physicians and center staff alike) appeared to underlie this lack of response.
Baseline demography and selected risk factors for adverse pregnancy outcome are provided in Table 1. The 2 groups were well balanced with respect to these factors. However, a slightly higher proportion of late CVS cases had a previous miscarriage or previous preterm birth among the multigravid subset (44.6% versus 39.7%; P = .004). More than 99% of women met the advanced maternal age criterion; 12 nonmaternal age women had a previous trisomic pregnancy and 14 a positive first trimester screening test.
Late CVS procedures were performed as assigned in 99.6% of cases (Table 2). Three women were crossed over to an early amniocentesis (2 for anatomical reasons, 1 at patient request), and 4 women had neither study procedure performed (1 transcervical CVS procedure, 2 patient refusals, and 1 fetal death noted after randomization). In the early amniocentesis group, 99.8% had the assigned procedure. Two women requested a late CVS procedure after randomization, and 1 late CVS procedure was performed by mistake. There were 31 (0.8%) protocol deviations; 19 involved the inadvertent inclusion of ineligible patients, and in 12 cases there was a failure to follow protocol after randomization. Of the 31 protocol deviations, 18 occurred in the late CVS group and 13 in the early amniocentesis group.
Procedure and cytogenetic characteristics are summarized in Table 2. There were 72 pregnancies (1.9%) with a final diagnosis of cytogenetic abnormality, 34 in villous and 38 in amniotic cells (Figure 1). Included were 67 aneuploids, 4 mosaics, and 1 structural abnormality. Sixty-six of these pregnancies were terminated, 2 were lost spontaneously before 20 weeks, and 4 were carried to term. Early in the trial, 2 diagnostic errors were reported in the amniocentesis group. In one, maternal cell contamination was reported with a normal male diagnosis. After an ultrasonography at 20 weeks, further testing confirmed a normal female. In another, no maternal cell contamination was reported, and a normal female diagnosis was made. After a male birth with cytogenetic confirmation, the investigator concluded that the error arose from maternal cell contamination of the sample.
The mean days to final harvest were 6.3 for chorionic villi and 10.3 for amniocytes (P < .001). It is notable that, early in gestation, performance was better for the villi samples than for amniotic fluid samples. Mean days to harvest of CVS cultures were 5.9 in weeks 11–12, 6.3 in week 13, and 6.5 in week 14. In contrast, for cultures from early amniocentesis, mean days to harvest were 12.2 at 11–12 weeks, 10.3 in week 13, and 9.9 in week 14.
Pregnancy outcome data were available for 3,770 (99.9%) of all randomized subjects, of which 3,698 were cytogenetically normal (Figure 1). All pregnancy outcome results presented are analyzed using this cytogenetically normal cohort with patients analyzed in the groups to which they were assigned. Analyses by procedure actually performed yielded similar results. Rates of the primary study outcome were similar in both groups (Table 3) and when stratified by gestational age at sampling. In week 13, the primary outcome rate was 2% (24/1,216) in the CVS group and 2.4% (28/1,177) in the early amniocentesis group (P = .497); in week 14, it was 1.6% (8/501) in the CVS group and 2.1% (10/483) in the early amniocentesis group (P = .579). All of these rates (excluding survivors born before 28 weeks) are marginally higher than expected for traditional 15–18 week amniocentesis. Although only 2 of the 32 operators performed more than an average of 4 of each procedure per month within this trial, we did not detect a difference in pregnancy loss rate or any adverse outcome between the centers or operators with greater or lesser procedural activity. The primary outcome rate for high volume operators was 2% and for low volume operators 2.2% (P = .78).
Twenty-one induced abortions were performed in the cytogenetically normal cohort: 10 after early amniocentesis and 11 after late CVS. Five of the 10 abortions in the early amniocentesis group (amniotic band disruption and leaking of fluid) appeared directly related to the procedure, compared with only 1 of the 11 terminations among late CVS cases (premature rupture of membranes). When these additional 6 “procedure-related” terminations were combined with spontaneous losses before 20 weeks, as an aggregate clinical end point, the week 13 comparison approached significance (relative risk [RR] 2.07; 95% confidence interval [CI] 0.97, 4.40; P = .054), with an approximate doubling of the combined risk of spontaneous loss plus procedure-related termination in the early amniocentesis group (Table 4).
Pregnancy complications are recorded in Table 5. Cramping after the procedure was more frequent after CVS (P < .001). Amniotic fluid leakage at less than 20 weeks of gestation occurred more often after early amniocentesis (P < .001). Significantly higher rates of gestational hypertension or preeclampsia were observed after late CVS (P = .005), but there was no difference in placental abruption.
A 4-fold increase in the rate of talipes equinovarus was observed in association with early amniocentesis: 3 cases after late CVS (0.16%) compared with 12 cases after early amniocentesis (0.66%; P = .017). All cases required surgery, except for 1 case in each group with a planned program of intensive physical therapy and splinting. Most of the neonates affected with talipes equinovarus in the early amniocentesis cohort followed a procedure in week 13 (RR 4.65; 95% CI 1.01, 21.5). When fluid leakage was observed by a patient in the amniocentesis group, the talipes equinovarus rate was 3.2% compared with 0.43% when no leakage was reported (P = .002). Significant leakage confirmed by a physician was observed in 3 of the 12 talipes cases. The talipes equinovarus rate with confirmed significant leakage was 17.7% versus only 0.51% with no confirmed leakage (P < .001).
The difference between groups in the rate of talipes equinovarus also held when calcaneo valgus, talipes varus, and twisting/distortion of the feet were included: 6 (0.3%) in the late CVS group versus 16 (0.9%) in the early amniocentesis group (P = .03, Table 6). The prevalence of other orthopedic problems, such as hip subluxations or upper-limb abnormalities, as well as other newborn anomalies was otherwise similar in the 2 groups. No limb reduction defects were observed. Table 6 summarizes the newborn outcome variables, which were uniformly similar between late CVS and early amniocentesis infants except as noted above.
This large randomized study compares CVS and amniocentesis performed at the same gestational age; therefore, background occurrence of fetal loss, fetal malformation, and other adverse outcomes are the same in both groups, allowing postprocedure outcomes to be directly compared. The gestational age addressed in the study is the one in which large numbers of women will be informed of their elevated risk of fetal trisomy as a result of increased use of early trisomy risk screening. The results of this trial are therefore of significant interest to such women and their caregivers in providing information to help with their choice of confirmatory invasive procedure at 13 and 14 weeks, even though we were not able to achieve full enrollment. Nevertheless, we had more than 50% power to detect a 50% increase in the primary outcome and nearly 95% power to detect a doubling of the outcome.
The results show that amniocentesis at 13 weeks carries a significantly increased risk of talipes equinovarus in a cytogenetically normal cohort (0.76%) compared with CVS (0.16%) at the same gestational age. In the general population, talipes equinovarus occurs with a frequency of approximately 0.1–0.3%.17 Combining our observations with those in the literature produces a gestational age continuum of excess risk for talipes equinovarus after early amniocentesis, with the risk of this anomaly being inversely related to the gestational age at sampling (Table 7). In the Canadian Early and Mid-Trimester Amniocentesis Trial study,13 the rate of talipes equinovarus was 1.6% (28/1,727) in the 11+0–12+6 week amniocentesis group compared with 0.1% at 15+0–16+6 weeks (P < .001). Sundberg et al12 found an excess risk of talipes equinovarus after early amniocentesis up to 88 days (12+4 weeks) compared with CVS at 10–12 weeks. Our study demonstrates an increased risk at 13+0 to 13+6 weeks, and 2 recent observational studies18,19 also support this inverse relationship of gestational age and risk.
The increased risk of talipes equinovarus is almost certainly related to a reduction in amniotic fluid subsequent to the procedure.20 Although we did not routinely obtain ultrasound follow-up to assess the residual volume of amniotic fluid after early amniocentesis, the likelihood of talipes equinovarus was significantly increased when leakage was reported and confirmed by a physician (P < .001), or even when it was just reported by the patient without confirmation (P < .002). Confirmed leakage has been correlated with increased risk of talipes equinovarus in other studies as well.12,13 The fact that talipes can also occur without clinically apparent leakage (as happened in 7 of our 12 cases) suggests that amniotic fluid may leak out of the amnion after amniocentesis but stay within the chorion, remaining clinically silent from the patient's perspective. This theory would preferentially apply to early amniocentesis because chorion-amnion fusion has not occurred uniformly until 14 weeks of gestation in many of these pregnancies.
The primary outcome parameter for this study was defined as pregnancy loss or delivery before 28 weeks of gestation in cytogenetically normal pregnancies. At 2.1% for late CVS and 2.3% for early amniocentesis, these relatively low postprocedure loss rates may be secondary to the considerable experience of operators performing the sampling, in addition to very strict prerandomization ultrasound review to remove potentially abnormal fetuses and those with discrepant dating, as shown in Figure 1 (194/6,370; 3%). However, a more useful estimate of the risk of the invasive procedure than the primary outcome rates may be the combination of spontaneous losses before 20 weeks with procedure-related terminations (premature rupture of membranes, severe fluid leakage/oligohydramnios, and amniotic band disruption). In these 2 categories combined, there were 16 pregnancy losses in the late CVS group and 27 in the early amniocentesis group (RR 1.74; 95% CI 0.94, 3.22; P = .073; Table 4). For week 13 only, there were 10 losses in the late CVS group and 20 in the early amniocentesis group (RR 2.07; 95% CI 0.97, 4.40; P = .054). Although these numbers fail to reach statistical significance, they suggest that the procedure-related fetal loss rate may be slightly higher after early amniocentesis than after late CVS. This is in accordance with the findings of the Canadian Early and Mid-Trimester Amniocentesis Trial study,13 which found an increase in fetal loss rate after early amniocentesis in the 11+0–12+6 gestational weeks. Nicolaides et al21 also found a significantly increased spontaneous loss rate (intrauterine or neonatal death) after early amniocentesis compared with late CVS. Sundberg et al,12 however, did not find any significant loss difference between the groups, but this may be explained by small numbers because the study was stopped early for ethical reasons after the increased risk of talipes equinovarus became apparent for the first time. We were unable to recruit the planned number of subjects within the study timeframe, thus a difference in the early, unintended loss rate between procedures may have been missed.
Apart from the amniotic fluid leakage discussed above, an increase in gestational hypertension and/or preeclampsia after late CVS was observed (5.4% compared with 3.5% in the early amniocentesis group). This finding justifies further analysis and consideration of potential mechanisms that would explain it (Silver RK, Wilson RD, Philip J, Thom EA, Zachary JM, Mohide P, et al. unpublished data). No other significant pregnancy complications were identified, and no differences were observed in the rate of respiratory problems or of hemangiomas between the 2 procedures.
In conclusion, we found a tendency toward increased early pregnancy loss associated with amniotic fluid leakage following amniocentesis in week 13 and a statistically significant increase in the risk of talipes equinovarus associated with amniocentesis in week 13. We did not have adequate data to demonstrate or exclude the same effect in week 14. On the basis of these results, we consider transabdominal chorionic villus sampling the method of choice for invasive prenatal diagnosis before 14 weeks and 0 days gestation.
1. Tabor A, Philip J, Madsen M, Bang J, Obel EB, Norgaard-Pedersen B. Randomized controlled trial of genetic amniocentesis in 4,606 low-risk women. Lancet 1986;1:1287–93.
2. Multicentre randomised clinical trial of chorion villus sampling and amniocentesis. First report. Canadian Collaborative CVS-Amniocentesis Clinical Trial Group. Lancet 1989;1:1–6.
3. Smidt-Jensen S, Permin M, Philip J, Lundsteen C, Zachary JM, Fowler SE, et al. Randomised comparison of amniocentesis and transabdominal and transcervical chorionic villus sampling. Lancet 1992;340:1237–44.
4. Rhoads GG, Jackson LG, Schlesselman SE, de la Cruz FF, Desnick RJ, Golbus MS, et al. The safety and efficiency of chorionic villus sampling for early prenatal diagnosis of cytogenetic abnormalities. N Engl J Med 1989;320:609–17.
5. Jackson LG, Zachary JM, Fowler SE, Desnick RJ, Golbus MS, Ledbetter DH, et al. A randomized comparison of transcervical and transabdominal chorionic villus sampling. N Engl J Med 1992;327:594–8.
6. Crandall BF, Kulch P, Tabsh K. Risk assessment of amniocentesis between 11 and 15 weeks: comparison to later amniocentesis controls. Prenat Diagn 1994;14:913–19.
7. Henry GP, Miller WA. Early amniocentesis. J Reprod Med 1992;37:396–402.
8. Sandstrom M, Stryker JM, Frigoletto FD, Morton CC. Early amniocentesis for chromosome analysis: a preliminary experience. Am J Hum Genet 1987;41:A285.
9. Holzgreve W, Miny P, Basaran S, Fuhrmann W, Beller FK. Safety of placental biopsy in the second and third trimesters. N Engl J Med 1987;317:1159.
10. Brambati B, Tului L, Camurri L, Guercilena S. Early second trimester (13–20 weeks) transabdominal chorionic villus sampling (TA-CVS): a safe and alternative method for both high and low risk populations. Prenat Diagn 2002;22:907–13.
11. Wei LJ, Lachin JM. Properties of the urn randomization in clinical trials [published erratum appears in Controlled Clin Trials 1989;10:following 126]. Controlled Clin Trials 1988;9:345–64.
12. Sundberg K, Bang J, Smidt-Jensen S, Brocks V, Lundsteen C, Parner J, et al. Randomised study of the risk of fetal loss related to early amniocentesis versus chorionic villus sampling. Lancet 1997;350:697–703.
13. Randomised trial to assess safety and fetal outcome of early and midtrimester amniocentesis. The Canadian Early and Mid-Trimester Amniocentesis Trial (CEMAT) Group. Lancet 1998;351:242–7.
14. Smidt-Jensen S, Hahnemann N. Transabdominal fine needle biopsy from chorionic villi in the first trimester. Prenat Diagn 1984;4:163–9.
15. Lan KKG, DeMets CL. Discrete sequential boundaries for clinical trials. Biometrika 1983;70:659–63.
16. Alexander GR, Kogan MD, Himes JH. 1994–1996 U. S. singleton birth weight percentiles for gestational age by race, Hispanic origin, and gender. Matern Child Health J 1999;3:225–31.
17. Lochmiller CL, Johnston D, Scott A, Risman N, Hecht JT. Genetic epidemiology study of idiopathic talipes equinovarus. Am J Med Genet 1998;79:90–6.
18. Nikkila A, Valentin L, Thelin A, Jorgensen C. Early amniocentesis and congenital foot deformities. Fetal Diagn Ther 2002;17:129–32.
19. Yoon G, Chernos J, Sibbald B, Lowry RB, Connors G, Simrose R, et al. Association between congenital foot anomalies and gestational age at amniocentesis. Prenat Diagn 2001;21:1137–41.
20. Potter EL. Bilateral renal agenesis. J Pediatr 1946;29:68.
21. Nicolaides K, Brizot Mde L, Patel F, Snijders R. Comparison of chorionic villus sampling and amniocentesis for fetal karyotyping at 10–13 weeks’ gestation [published erratum appears in Lancet 1994;344:830]. Lancet 1994;344:435–9.
In addition to the authors, the members of The Randomized Trial of Early Amniocentesis and TransAbdominal CVS (EATA) investigative group are as follows: Rigshospitalet, Genetic Center: W. Keller, A. Meyer, M. Vad, B. Binzer, A. G. Sidenius, S. Lindstrøm, S. Flint, K. Sundberg, and L. Sperling; Chromosome Laboratory: C. Lundsteen, A. M. Lind; Hvidovre Hospital: S. Smidt-Jensen; Drexel University College of Medicine: G. Davis, M. DiVito, and M. McGee; Baylor College of Medicine: R. Carpenter, J. Dungan, and A. Burke; Northwestern University Medical School: N. A. Ginsberg, C. Dougherty, and K. DeMarco; Fetal Diagnostic Center, Evanston Hospital of Northwestern University Medical School: K. Blum, E. Leeth, and J. Weimer; Cedars-Sinai Medical Center: D. E. Carlson (deceased), J. Williams, D. Krakow, C. A. Walla, W. Herbert, K. Wendt, and N. Greene; Magee-Womens Hospital: W. A. Hogge, E. Smith, and K. Ventura; UCLA Center for the Health Sciences: S. Beverly; University of Tennessee, Memphis: S. Elias, O. Phillips, L. Seely, and P. King; Wayne State University: M. Evans, D. Duquette, E. Krivchenia, and P. Devers; Yale University: J. Copel, R. Bahado-Singh, M. DiMaio, and S. Turk; McMaster University Medical Center: M. L. Beecroft, G. White, N. Brown, M. Huggins, and V. Freeman; BC Women's Hospital: S. Soanes; Prenatal Diagnosis of Northern California Medical Group: K. Kahl and M. Palmer; University of Maryland, Baltimore: K. Frayer; Mt Sinai School of Medicine, New York: R. Desnick, K. Eddleman, J. Stone, R. Zinberg, and J. Robinowitz; The George Washington University Biostatistics Center: B. Fisher, K. Poydence, P. Van, and N. Boone; The National Institute of Child Health and Human Development: F. de la Cruz and J. Hanson; Danish Centre for Evaluation and Health Technology Assessment, National Board of Health, Denmark: F. Børlum Kristensen.