Invasive Prenatal Genetic Techniques

Advancing capabilities in the detection of genetic abnormalities and monitoring fetal development have enabled the prenatal diagnosis of a wide variety of congenital and acquired conditions. It is now possible to determine whether a fetus is affected with a specific genetic or acquired disorder rather than relying on probability assessment and statistical analyses to assess fetal risk. This chapter reviews the techniques used for invasive prenatal diagnosis and their safety, efficacy, and applications.

Many clinical and historical factors can lead to prenatal diagnosis. Regardless of the specific indication, invasive prenatal testing should be performed with detailed genetic counseling performed before the procedure. Such counseling is needed to obtain critical information from the patient to assess properly fetal risk; to review personal and family histories that have an impact on this risk; to describe the risks, benefits, and limitations of the procedure(s) to be offered; and provide empathetic support to women who are considering such testing. Genetic counseling should be provided in a nondirective fashion, thus providing women with the necessary information (written and verbal) to arrive at prenatal diagnostic decisions but without coercing the woman, either overtly or covertly, into making a particular decision.

The commonest indication for consideration of invasive prenatal testing is an increased risk for fetal chromosome abnormalities. Such an increased risk can occur as result of advanced maternal age (35 years old or older at estimated date of delivery), a positive maternal analyte screen, or previous child with a chromosome abnormality or a parental chromosome rearrangement. Other indications for consideration of invasive prenatal testing include an increased risk for a detectable mendelian disorder (earlier birth of an affected child or family member, positive screening outcome), structural fetal anomalies detected by ultrasound, increased risk for polygenic/multifactorial disorder [neural tube defect]); positive screen, abnormal ultrasound, family history) or exposure to teratogens that increase the risk for detectable abnormalities (e.g., valproic acid and neural tube defects). In some cases, the indication for invasive testing may designate the type of procedure available for diagnosis. For example, in cases of increased risk for fetal neural tube defects, chorionic villus sampling is not an appropriate test as this technique cannot obtain amniotic fluid for the α-fetoprotein (AFP) and acetylcholinesterase analyses.

Traditional Amniocentesis: Gestation of 15 Weeks or Longer

Amniocentesis, the aspiration of amniotic fluid, has traditionally been performed at and after 15 to 17 weeks of gestation. At this stage of gestation, the volume of amniotic fluid is about 200 mL. The ratio of viable to nonviable cells in the amniotic fluid is relatively high,¹ thus allowing timely culture and diagnosis of fetal cytogenetic abnormalities and providing women the option of pregnancy termination should an abnormality be determined.

Amniocentesis is routinely performed in an outpatient facility. An ultrasound examination should be done immediately before the procedure to evaluate fetal number and viability, perform fetal biometric measurements to confirm gestational age, establish placental location, and estimate amniotic fluid volume. In our hospitals, a fetal anatomic survey to screen for major anomalies is standard practice. In addition, ultrasonography may be useful in discovering maternal gynecologic conditions (e.g., leiomyomata) that could influence the technique and timing of the amniocentesis.

After the preoperative ultrasound examination has been completed, a needle insertion site is chosen. We prefer to insert the needle in the midline fundal area of the uterus, but this is not always the site at which the optimal pocket of amniotic fluid is located. Not infrequently, a lower uterine segment or lateral approach is required. If possible, we avoid the placenta; however, transplacental amniocentesis has not been shown to reduce the safety of amniocentesis.^2,3,4 If tapping the optimal pocket of fluid requires traversing the placenta, we select the thinnest portion of the placenta possible through which the needle can be directed. The umbilical cord insertion site should be identified and avoided. The maternal bowel and bladder should also be located, because these should likewise be avoided. A local subcutaneous anesthetic (e.g., 2 to 3 mL of 1% lidocaine) may be used, but we usually find this unnecessary. After the maternal skin is cleaned with an iodine-based solution, sterile drapes are placed around the needle insertion site to help maintain an aseptic field. We prefer a 22-gauge spinal needle and recommend no larger than a 20-gauge needle.

Ultrasonographic monitoring with continuous visualization of the needle should be used throughout the procedure. Ultrasound gel is applied adjacent to the insertion site, and a real-time ultrasound transducer is held in position such that the ultrasound beam is directed parallel to the planned needle track. Needle insertion should be performed with one smooth continuous motion until the needle tip is within the amniotic cavity (Fig. 1). Some practitioners locate the needle tip within the subcutaneous fat and assess the distance from tip to amniotic cavity as well as the appropriate needle angle required to successfully enter the amnion and avoid maternal structures and the fetus.

The first several milliliters of amniotic fluid are aspirated into a syringe. These first few milliliters are theoretically most likely to contain maternal cells; therefore, this initial sample is usually discarded or set aside for an AFP assay. For a second-trimester amniocentesis performed at 14 to 20 weeks inclusive, 20 to 30 mL of amniotic fluid is usually aspirated. After obtaining the amniotic fluid, the specimen should be clearly labeled and transported at ambient temperature to the laboratory.

The amniotic fluid aspirated is bloody in approximately 1% to 2% of amniocenteses. The blood, which is almost always maternal in origin, does not adversely affect amniotic cell growth. Indeed, the performance of a transplacental amniocentesis may increase the risk for a bloody tap; this has not been shown to have an adverse affect on the safety or accuracy of the amniocentesis. By contrast, brown or dark red or wine-colored amniotic fluid is associated with an increased likelihood of adverse pregnancy outcome. This color indicates earlier intra-amniotic bleeding, with hemoglobin breakdown products accounting for the pigment. Pregnancy loss eventually occurs in about one third of such cases.⁵ If the abnormally colored fluid is also characterized by an elevated AFP level, the risk for adverse perinatal outcome (fetal death, anencephaly, spontaneous abortion, or fetal abnormality) is usually further increased. Green amniotic fluid, presumably due to meconium staining, is apparently not associated with poor pregnancy outcome.^6,7,8 By contrast, brown amniotic fluid has been associated with an increased risk of fetal aneuploidy.⁸

The propriety of administering Rh immunoglobulin (RhIG) to prevent Rh immunization in unsensitized women with Rh-positive fetuses remains controversial although widely done. Fetomaternal transfusion by disruption of the fetoplacental circulation logically might have an immunizing effect; however, the magnitude of the risk has not been determined. The task is difficult because one must consider such variables as ABD compatibility, number of needle insertions, placental location, and the amount of fetal blood transfused into the maternal circulation. Early investigators considered that prophylactic RhIG should not be administered following genetic amniocentesis,⁹ but almost all now advocate its routine use.^10,11,12,13 The dose to be administered is controversial. The American College of Obstetricians and Gynecologists recommends that 300 μg of RhIG be administered for an exposure of 30 mL of fetal blood.¹⁴ In the United Kingdom, the recommended dose of RhIG is 50 μg before 20 weeks of gestation and 100 μg thereafter.¹⁵ We routinely administer 300 μg of RhIG following genetic amniocentesis, regardless of whether the needle may have traversed the placenta.

Following amniocentesis, fetal heart motion should be documented by ultrasonographic visualization. The patient is observed briefly following the procedure and is instructed to report any vaginal fluid loss or bleeding, uterine cramping, or fever. Reasonably normal activities may be resumed following the procedure; however, we recommend that strenuous exercise (e.g., jogging or aerobic exercises) and coitus be avoided for 1 day.

Multiple Gestations

In multiple gestations, amniocentesis can usually be performed on all fetuses, provided amniotic fluid volume is adequate.¹⁶ Following aspiration of amniotic fluid from the first sac, 2 to 3 mL of indigo carmine, diluted 1:10 in bacteriostatic water, is injected before the needle is taken out. A second amniocentesis is then performed. The second needle is then inserted into the sac of the second fetus, preferably determined after visualizing the membranes separating the two sacs. Aspiration of clear fluid confirms that the second sac has truly been entered. In experienced hands, amniocentesis is performed successfully in more than 95% of twin pregnancies with ostensibly no increased risks over that of amniocentesis in singleton pregnancies.^16,17 Anderson and Goldberg¹⁷ observed a postprocedure twin-loss rate of 3.57% up to 28 weeks, a rate interpreted as not increased over the sum of background twin-loss rate plus the loss rate associated with singleton amniocentesis. Triplets and higher-order multiple gestations can be managed similarly by sequentially injecting dye into successive sacs. If the aspirated fluid remains clear, the clinician can be reassured that a new amniotic sac has been entered. However, the overall safety of amniocentesis in triplets and higher-order multiple gestations have not yet been formally determined.

Other techniques for sampling multiple gestations have been reported, including a single puncture technique and an air bubble infusion procedure. Despite advances in ultrasonography, we have concerns that single-puncture techniques could lead to cross-contamination between sacs, resulting in diagnostic inaccuracies. In addition, we believe it unwise to instill any substance into the amniotic cavity unless the instillation is performed in a strict, aseptic fashion and serves to improve safety or accuracy considerably and consistently. More recently, some operators have not used dye instillation to distinguish separate sacs, instead relying on ultrasonographic visualization alone.^18,19,20 In view of the considerable positive experience with dye instillation techniques during the past 20 years with regard to safety and accuracy, we still use the dye instillation procedure.

Safety

Any procedure that involves passing a device into an organ, especially the pregnant uterus, entails risk; amniocentesis is no exception. Indeed, amniocentesis carries potential danger to both mother and fetus. Maternal risks are quite low, with symptomatic amnionitis occurring only rarely. Minor maternal complications such as transient vaginal spotting and minimal amniotic fluid leakage occur in 1% or fewer cases, but these are almost always self-limited. Other very rare complications include intra-abdominal organic injury or hemorrhage.

The safety of traditional amniocentesis has been addressed by several large collaborative studies. The US National Institute of Child Health and Human Development²¹ conducted the first major prospective study of genetic amniocentesis that comprised 1,040 study subjects and 992 matched controls. Of the 1,040 women undergoing amniocentesis, 950 (91.3%) had the procedure performed for cytogenetic analysis and 90 (8.7%) to evaluate for the possible presence of an inborn error of metabolism. Of all women who underwent amniocentesis, 3.5% experienced fetal loss between the time of the procedure and delivery compared with 3.2% of controls; the slight difference was not statistically significant and disappeared completely when corrected for maternal age. In Canada, a collaborative group conducted a cohort study but did not include a concurrent control group.^22,23 Analysis was based on 1,223 amniocenteses performed during 1,020 pregnancies in 900 women. The pregnancy loss rate was 3.2%, a frequency similar to that reported in the US collaborative study.

A later British collaborative study found that the rate of fetal loss following amniocentesis was significantly greater than in controls (2.6% versus 1%).²⁴ In the British study, however, one common indication for amniocentesis was elevated maternal serum AFP (MSAFP) now itself recognized as a factor associated with fetal loss and adverse perinatal outcome. Analysis after excluding subjects undergoing amniocentesis for that indication lowered the loss rates between study subject and control groups to less than 1%, albeit still a significant difference.²⁵

None of the collaborative studies cited above was conducted with high-quality ultrasonography as defined by today's standards, nor was concurrent ultrasonography universally applied. More relevant recent data are from a Danish randomized controlled study of amniocentesis that involved 4,606 women aged between 25 and 34 years who were without known risk factors for fetal genetic abnormalities.²⁶ Women with three or more previous spontaneous abortions, diabetes mellitus, multiple gestation, uterine anomalies, or intrauterine contraceptive devices were excluded. Maternal age, social group, smoking history, number of previous induced and spontaneous abortions, stillbirths, livebirths, and low-birthweight infants were comparable in the study and control groups, as was gestational age at time of entry into the study. Amniocentesis was performed under real-time ultrasound guidance with a 20-gauge needle by experienced operators. Followup evaluation was available for all but 3 women. The spontaneous abortion rate after 16 weeks was 1.7% in amniocentesis patients compared with 0.7% in controls (p < .01), with a 2.6-fold relative risk of spontaneously abortion if the placenta was traversed. Transplacental needle passage is a risk factor that has more recently been shown to not increase the risk for postprocedure loss (see previous discussion). The frequency of postural malformations in the infants in the two groups did not differ. However, respiratory distress syndrome was diagnosed more often (relative risk [RR], 2.1) in the study group, and more infants were treated for pneumonia (RR, 2.5). A more recent assessment of the safety of second-trimester amniocentesis can be found in the CEMAT study that compared conventional second-trimester amniocentesis to early amniocentesis.²⁷ In this multicenter trial performed by experienced operators using concurrent ultrasonography, the total loss rate in the second trimester amniocentesis cohort was 5.9%.

In a study from British Columbia, Baird and colleagues²⁸ considered the question of whether children delivered of women who had midtrimester amniocentesis can be identified by a population-based database of congenital anomalies and disabilities at a different rate from that of matched controls (i.e., children of women who had not undergone amniocentesis). The authors studied 1,296 cases (651 boys and 645 girls) and 3,704 matched controls (1,867 boys and 1,837 girls) among live births (1972 to 1983) from the Health Surveillance Registry with data collected to 1990 to allow followup of 7- to 18-years' duration. Cases were children of mothers who had midtrimester amniocentesis for advanced maternal age (35 years or older) and whose examination results were normal for chromosomal disorders and neural tube defects. When possible, three controls per case were matched for age of mother, gender, date of birth, and state of health from provincial birth records. In all 128 (9.9%) of the cases and 308 (8.3%) of the controls were registered (RR, 1.23); this RR was not significantly different from 1. The likelihood of having disabilities was examined for cases compared with controls, and no difference was found except for an increased ABO isoimmunization associated with amniocentesis. Overall, this study provides reassuring data for patients considering midtrimester amniocentesis with respect to long-term outcome.

In conclusion, we believe it wise to continue to counsel that the risk of pregnancy loss secondary to amniocentesis is 0.5% over baseline, or perhaps slightly less at centers with experienced operators. At our centers, serious maternal complications and fetal injuries are stated to be “remote” risks.

Early Amniocentesis: Gestation of 14 Weeks or Less

With the advent of high-resolution ultrasound equipment, some physicians began offering amniocentesis before 15 weeks of gestation. Some programs not offering chorionic villus sampling (CVS) viewed early amniocentesis as an attractive alternative for those women who desired prenatal diagnosis before the time in pregnancy when traditional amniocentesis is performed (i.e., 15 or more weeks of gestation). In other medical centers, early amniocentesis was explored to lessen the inconvenience of patients' having to be rescheduled if they came in for CVS and were determined to be beyond 12 weeks but under 15 weeks of gestation.

Several programs, including those at our own centers, reported experiences suggesting early amniocentesis was a promising technique. In a series of 936 amniocenteses at 12.8 weeks of gestation or less reported by Hanson and colleagues,²⁹ loss rates were 0.7% (7 of 936) within 2 weeks of amniocentesis, with an additional 2.2% before 28 weeks and an additional 0.5% stillbirths or neonatal deaths. Total losses (32 of 936, or 3.4%) were considered comparable with the 2.1% to 3.2% in ultrasonographically normal pregnancies not undergoing a procedure; however, lack of corrections for maternal age and gestational age render comparisons less than exact. Other series reported include those conducted by Elejalde,³⁰ Penso,³¹ Stripparo,³² Hackett,³³ Assel,³⁴ Djalali,³⁵ Henry,³⁶ Yang,³⁷ Eiben,³⁸ and in all cases, by their respective coworkers, and by Kerber and Held.³⁹ Our group, then at the University of Tennessee, Memphis,⁴⁰ compared our initial experience with 250 early amniocenteses (14 weeks or less) to that of our first 250 cases of transabdominal CVS (9.5 to 12.9 weeks), finding loss rates for early amniocentesis and transabdominal CVS to be 3.8% and 2.1%, respectively. Our group also reported early amniocentesis in 6 twin gestations (mean 11.9 weeks; range 10.5 to 13.6 weeks), using a similar dye injection technique as described for traditional amniocentesis (see earlier discussion).⁴¹ We successfully tapped both amniotic sacs in each of 6 cases (5 requiring two needle insertions, 1 requiring three needle insertions); all cultures yielded normal cytogenetic results, and all six pregnancies resulted in the delivery of health infants.

However, these studies represented observational reports of amniocentesis with most procedures being performed at or after 13 weeks of gestation. More rigorous comparative studies evaluating the safety and efficacy of early amniocentesis have, in general, failed to corroborate the generally favorable outcomes reported in the observational studies. Nicolaides and colleagues⁴² reported a comparison of amniocentesis and CVS at 10 to 13 weeks of gestation. Early amniocentesis was performed in 731 patients (493 by choice and 238 by randomization) and CVS in 570 (320 by choice and 250 by randomization). Both procedures were performed by transabdominal ultrasound-guided insertion of a 20-gauge needle. The spontaneous loss (intrauterine or neonatal death) was significantly higher after early amniocentesis (total group mean, 5.3%; 95% confidence interval [CI] 3.8 to 7.2; randomized subgroup mean, 5.9%; CI, 3.3 to 9.7) than after CVS (total group mean, 2.3%, CI 1.2 to 3.9; randomized subgroup, mean, 1.2%; CI, 0.3 to 3.5). Subsequently, Nicolaides and colleagues⁴³ reported further findings of this study. Again, postprocedure loss rates were significantly higher in the total early amniocentesis group (4.9%) and the randomized early amniocentesis subgroup (5.8%) than in the total transabdominal CVS group (2.1%) and the randomized transabdominal CVS subgroup (1.8%). The frequency of talipes equinovarus was higher in the early amniocentesis group, but this difference did not attain statistical significance.

Sundberg and colleagues⁴⁴ performed a similar type of randomized trial comparing early amniocentesis and CVS. In this study, the rate of talipes equinovarus increased significantly in the early amniocentesis group, even though investigators used a filter system to reduce the amount of fluid removed. This finding led the authors to terminate their study for safety concerns prematurely.

The results of the Canadian Early Amniocentesis versus Mid-Trimester Amniocentesis Trial,^27,45 a multicenter, randomized trial of early amniocentesis and conventional second-trimester amniocentesis have been most revealing. This multicenter trial randomized 4,374 women into an early amniocentesis cohort (n = 2,183) and a conventional mid-trimester amniocentesis cohort (n = 2,185). In the early amniocentesis cohort, 1,916 women (87.8%) underwent amniocentesis before 13 weeks of gestation.

Loss rates were 7.6% for the early amniocentesis cohort and 5.9% for the midtrimester cohort (p = .012). Talipes equinovarus occurred in 1.3% of infants delivered of women in the early amniocentesis group compared to 0.1% in the midtrimester cohort (p = .0001). In addition, postprocedure amniotic fluid leakage occurred more frequently in the early amniocentesis group (3.5%) than in the midtrimester group (1.7%; p = .0007). Failed procedures, multiple needle insertions, and culture failures also occurred more frequently in the early amniocentesis group.^27,45

In view of these recent studies, the ongoing United States multicenter trial evaluating the safety and efficacy of early amniocentesis with the EATA (Early Amniocentesis-Transabdominal CVS) trial ceased offering early amniocentesis to women who had sustained less than 14 weeks of gestation. These more recent prospective studies strongly suggest that second trimester amniocentesis and first-trimester chorionic villus sampling should remain, for the foreseeable future, as the invasive procedures of choice for the detection of fetal cytogenetic and Mendelian abnormalities.

Because amniocentesis is most commonly performed in the midsecond trimester (15 to 16 weeks), fetal diagnosis cannot usually be established before 17 to 18 weeks of gestation. A technique that could be performed during the first trimester would be highly desirable to reduce the psychological stress of awaiting results until midpregnancy and to allow a safer method of pregnancy termination, should an abnormality be detected. CVS is such a technique.

Techniques for Chorionic Villus Sampling

TRANSCERVICAL CHRONIC VILLUS SAMPLING.

Transcervical CVS is now usually performed at 10 to 12 completed gestational weeks. Absolute contraindications to transcervical CVS include active cervical or vaginal pathology (e.g., herpetic, chlamydial, or gonorrheal infection) or maternal blood group sensitization. Relative contraindications include leiomyoma obstructing the cervical canal, bleeding from the vagina within 2 weeks of planned CVS, and a markedly retroverted, retroflexed uterus.⁴⁶ Before CVS, fetal viability and normal fetal growth must be confirmed by ultrasound. The procedure is performed with a device that consists of a plastic cannula enclosing a metal obturator extending just beyond the catheter tip; the diameter of most catheters is approximately 1.5 mm.

At our institutions, CVS is performed in an ultrasound suite. The patient is positioned in the dorsal lithotomy position. The vagina is cleaned with an iodine preparation, and the perineum draped with sterile towels. After insertion of a sterile vaginal speculum, placement of a tenaculum on the anterior lip of the cervix may occasionally be needed to help correct uterine anteflexion or retroflexion. The CVS catheter is introduced transcervically under simultaneous ultrasonographic visualization, with optimal catheter placement being parallel to the long access of the placenta (Fig. 2). The obturator is then withdrawn and the catheter Luer-lok attached to a 20- or 30-mL syringe. Chorionic villi are then aspirated by multiple, rapid aspirations of the syringe plunger to 20 to 30 mL negative pressure. The catheter is withdrawn under continuous maximum negative pressure. Although adequate sample is contained in at least 5 mg of villi, a sample of 10 to 25 mg is preferable.

Following the procedure, fetal heart activity is verified by ultrasonography. Patients are monitored for any untoward effects for approximately 30 minutes. Unsensitized Rhnegative patients are given RhIg. Maternal serum AFP screening for fetal neural tube defects is recommended at 16 to 18 weeks gestation.

TRANSABDOMINAL CHORIONIC VILLUS SAMPLING.

Transabdominal CVS applicable to evaluate pregnancies at the same gestational age as transcervical CVS at 10 to 12 weeks; however, this procedure also can be performed later in pregnancy,^47,48,49 particularly when fetal abnormalities are visualized at ultrasound and a rapid diagnosis may influence pregnancy management. Placentas especially amenable to the transabdominal approach include those located in the fundus or those located anteriorly in an anteflexed uterus. Transabdominal CVS is also an option in certain circumstances when transcervical sampling is contraindicated (e.g., active herpes or cervical lesions).

Following selection of a needle insertion site based on ultrasound findings, the overlying skin is infiltrated with a local anesthetic, cleaned with an iodine preparation, and draped with sterile towels. As performed at our centers, a 19-gauge spinal needle is inserted percutaneously through the maternal abdominal wall and myometrium under continuous ultrasound guidance. The tip is then guided into the long axis of the placenta (Fig. 3). The needle stylet is withdrawn, and next a syringe housed in an aspiration device is connected to the Luerlock of the needle. Chorionic villi are obtained by repeated (15 to 20) rapid aspirations of the syringe plunger to 20-mL negative pressure. The needle is then withdrawn under continuous negative pressure. The same postoperative protocol is used as for transcervical CVS.

Most physicians performing transabdominal CVS employ the aforementioned “freehand” technique. Alternatively, some operators still use a guide-needle or double-needle system device, which punctures the uterine wall once but permits multiple attempts at villus aspiration.⁵⁰

TRANSVAGINAL CHORIONIC VILLUS SAMPLING.

Although not commonly necessary, transvaginal CVS may be the only first-trimester option in women having a retroverted retroflexed uterus with a posterior placenta. Patients are prepared in a fashion similar to that used in transcervical CVS. We use a 35-cm 18-gauge aspiration needle to obtain chorionic villi. The wall of the vagina posterior to the cervix is infiltrated with a local anesthetic. Transabdominal ultrasound is used for needle guidance through the vaginal mucosa and myometrium into the placenta. After the needle is within the placenta, the needle stylet is removed and villi are aspirated in a similar fashion as with transabdominal CVS (Fig. 4).⁵¹

Safety of Chorionic Villus Sampling

PREGNANCY LOSS FOLLOWING CHORIONIC VILLUS SAMPLING.

The US Cooperative Clinical Comparison of Chorionic Villus Sampling and Amniocentesis study and the Canadian Collaborative CVS-Amniocentesis Trial Group study initially reported the pregnancy loss rate CVS to be 0.8% and 0.6% higher than traditional amniocentesis, respectively.^52,53 Neither figure was statistically significant. In the US study, 2,278 women self-selected transcervical CVS; 671 women similarly recruited in the first trimester selected amniocentesis. Randomization did not prove possible. The Canadian study was randomized, with 1,391 subjects assigned to transcervical CVS and 1,396 to amniocentesis. Variables shown to influence fetal loss rates adversely included fundal location of the placenta, number of catheter passages, small sample size, and prior bleeding during the current pregnancy.^52,54 Almost all the above invariably reflect technical difficulty. Other obstetric complications (e.g., intrauterine growth restriction [IUGR], placental abruption, premature delivery) did not exceed those in women not undergoing CVS.

Transcervical CVS and transabdominal CVS appear to be equally safe procedures for first-trimester diagnosis. In a later US NICHD collaborative study, in which 1,844 patients were randomized to transcervical CVS and 1,929 patients randomized to transabdominal CVS, the loss rates of cytogenetically normal pregnancies through 28 weeks were 2.5% and 2.3%, respectively.⁵⁵ Of note, the overall loss rate (i.e., background plus procedure related) following CVS decreased by about 0.8% during this 1988 to 1990 randomization trial, in comparison with rates observed during the transcervical versus amniocentesis self-selection study (1985 to 1987). This decrease in procedure-related loss rate probably reflects increasing operator experience, as well as availability of both transcervical and transabdominal approaches. In a small Italian randomized trial, Brambati and partners⁵⁶ also found no difference between transabdominal and transcervical CVS. By contrast, in a randomized comparison of amniocentesis, transabdominal CVS, and transcervical CVS in Denmark, Smidt-Jensen and associates⁵⁷ found similar fetal loss rates after transabdominal CVS and amniocentesis but a significantly increased loss rate associated with transcervical CVS. These results are not surprising, however, given that the Danish experience with transabdominal CVS is far greater than with transcervical CVS.

The one major study that differs substantively from the US, Canadian, and Italian collaborative trials is the Medical Research Council (MRC) Study.⁵⁸ In this multicenter randomized comparison between first-trimester CVS performed in whatever fashion deemed suitable by the obstetrician and second-trimester amniocentesis, the ultimate variable measured was completed pregnancies. The 4.4% fewer completed pregnancies in the CVS cohort reflected both unintended and intended pregnancy terminations. The latter, in turn, probably reflected inexperience in cytogenetic interpretation, given some terminations seemingly arguable in retrospect (i.e., confined placental mosaicism). Experience with CVS by the MRC study operators was also considerably less than in the US operators. For example, the only requirement for participation in the MRC study was 30 “practice” CVS procedures. Some centers also contributed few cases.

No formal attempts to assess the safety of transvaginal CVS have as yet been attempted. In our own experience,⁵¹ as well as that of Sidranski and colleagues,⁵⁹ neither major complications nor obvious excessive fetal loss rates have been observed.

We conclude that clinical judgment and patient individualization in choosing the optimal approach for CVS increases safety. For example, some difficult transcervical CVS procedures (e.g., those involving fundal placentas) should be avoided in favor of a more straightforward transabdominal approach. The converse is also true.

CHORIONIC VILLUS SAMPLING IN MULTIFETAL PREGNANCY.

CVS in multifetal pregnancies has become more widely used with increasing use of fetal reduction resulting from assisted reproductive technologies. Many women requiring assisted reproductive technologies to become pregnant are also at increased risk for fetal chromosome or Mendelian abnormalities. CVS prior to fetal reduction allows for the detection of chromosomal or DNA abnormalities, thus allowing for the reduction of only abnormal gestations.^60,61,62

Few data exist for safety of CVS in multiple gestations. In a major US study involving four centers, the total loss rate of chromosomally normal fetuses (spontaneous abortions, stillborns, neonatal deaths) was 5.0%,⁶³ only slightly higher than the 4.0% observed for singleton pregnancies.⁵² We now use CVS in multiple gestations when both placentas can reliably be sampled as determined by ultrasonography. When placental locations preclude the reasonable procurement of separate and distinct samples, amniocentesis is preferred.

Limb Reduction Deformities (LRD)

Evaluation of the safety of CVS has recently shifted focus from concerns about the risk of fetal loss to its being the possible cause for congenital abnormalities. In 1991, Firth and colleagues^64,65 reported that 5 of 289 (1.7%, or 17 in 1,000) infants exposed to CVS at 56 to 66 days of gestation (i.e., 42 to 50 days after fertilization) had severe limb-reduction deformities (LRD). Of the 5 infants, 4 had oromandibular limb hypogenesis (all transabdominal CVS); the fifth had a terminal transverse limb reduction alone (transcervical CVS). Subsequently, a number of reports followed, both supporting and refuting such an association.^{66,67,68,69,70,71} In the United States, Burton and partners⁷² reported a second cluster among 394 infants whose mothers had undergone CVS. Thirteen (3.3%) had major congenital abnormalities, including four with transverse LRD (10 of 1,000 or 1%). All four LRDs were transverse distal defects involving hypoplasia or absence of the fingers and toes. Of these cases, 3 followed transcervical sampling, using a device that, in the hands of the reporting physicians, was associated with an 11% fetal loss rate.

Teratogenic mechanisms whereby CVS might cause LRD have been hypothesized.⁷³ These include hypoperfusion due to fetomaternal hemorrhage or pressor substances released by disturbance of villi or the chorion; embolization of chorionic villus material or maternal clots into the fetal circulation; and amniotic puncture and limb entrapment in exocoelic gel.

The potential association of LRD with CVS has been explored through various registries. Based on data from the Italian Multicenter Birth Defects Registry, Mastroiacovo and Batto⁷⁴ reported that 8 cases of oromandibular-limb hypogenesis complex were entered into the registry from January 1988 through December 1991, 4 of which had been exposed to CVS, compared with 36 exposed subjects among 8,445 controls. There were 166 cases of transverse limb defects alone, 4 having been exposed to CVS, compared with 36 cases among 8,445 controls. A 1994 update of this study, based on 11 CVS-exposed cases, continued to indicate an association with transverse limb defects.⁷⁵ The highest risk was associated with procedures performed at less than 70 days gestation (odds ratio [OR], 23.2; CI, 1.31 to 41.0); a lower, but still increased risk with procedures at 70 to 76 days (OR, 17.1; CI, 6.7 to 44.0); over 44 days there were no exposed cases and the risk interpreted as considerably lower. By contrast, analysis of other European registries in aggregate involving more than 600,000 births showed that only 4 of 336 cases (1.2%) with limb reduction abnormalities had been exposed to CVS, compared with 78 of 11,883 (0.66%) cases with other malformations (OR, 1.8; CI, 0.7 to 5.0).⁷⁶

Firth and coworkers⁷⁷ summarized LRD in 75 infants exposed to CVS by combining their own cases with those reported in the literature. The median gestational age at CVS ranged from 56 (range, 49 to 65) postmenstrual days for the most severe defects to 72 (range, 51 to 98) for the least severe. They concluded that there was a correlation between the severity of the defects and the duration of gestation when CVS was performed.

In 1994, the US Centers for Disease Control (CDC) held an open forum in Atlanta where data were presented from a US multistate case control study in an effort to quantify the risk of LDR associated with CVS.⁷⁸ Case subjects were 131 infants with nonsyndromic limb deficiency from seven population-based birth defects surveillance programs born to women 34 years or older from 1988 through 1992. Controls were 131 infants with other birth defects matched to case subjects by the infant's year of birth, mother's age, race, and state of residence. Overall, the OR for limb deficiency after CVS from 8 to 12 weeks gestation was 1.7 (CI, 0.4 to 6.3). However, when analyzed for specific anatomic subtypes, there was an association for transverse digital deficiency (OR, 6.4; CI, 1.1 to 38.6). They concluded that the absolute risk for such defects was approximately 1 in 3,000. It should be noted that such transverse digital deficiencies occurred in only seven exposed cases.

Finally, the World Health Organization Committee on Chronic Villus Sampling recently analyzed data collected through an international voluntary registry.^79,80 In total, 77 cases with LRD among 138,996 infants born after CVS were reported from 63 registering centers. Pattern analysis of the types of limb defects and overall frequencies of specific LRD were compared with a background population study from British Columbia, Canada.⁸¹ The pattern of defects showed the upper limb to be affected in 65%, 13% in the lower limb, and 23% in both upper and lower limbs compared with frequencies in the general population of 68%, 23%, and 9%, respectively. Transverse limb defects occurred in 41% of infants in the cohort exposed to CVS, compared with 43% in the general population and longitudinal limb deficiencies were found in 59% of cases, compared with 57% in the general population. It was concluded that the pattern analysis of the types of limb defects and calculation of overall incidences failed to find a difference between the CVS and background populations.

A more recent review of CVS safety sponsored by the World Health Organization⁸² evaluated the clinical outcomes of 216,381 CVS cases performed worldwide. In this review, 115 LRDs were observed with an incidence of 1 in 1,881, similar to the rate of 1 in 1,642 observed in the general population. The distribution of LRD (upper limbs 71.3%, lower limbs 11.3%, both limbs 17.4%) was also similar to that observed in the general population. The authors concluded the data demonstrate that “… CVS carries no increased risk for fetal loss or congenital malformation, including limb reduction defect …” compared with conventional mid-trimester amniocentesis.

CVS remains the only established method for first trimester prenatal diagnosis and carries minimal, if any, increased risk for adverse pregnancy outcomes when performed by experienced operators. To maximize the safety of CVS, we avoid performing CVS before 10 weeks gestation except in cases of profound risk for fetal abnormalities (e.g., fetal hydrops). We continue to counsel patients about the LRD controversy and inform them that the absolute risk at 10 to 12 weeks is believed to be very low, approximately 1 in 2,000, a rate similar to that observed in the general population.⁸² We also stress that this issue be placed in proper perspective and weighed against the substantial advantages that first trimester prenatal diagnosis that CVS offers.

Access to the fetal circulation was initially accomplished by fetoscopy, a method of directly visualizing the fetus, umbilical cord, and chorionic surface of the placenta, using endoscopic instruments.^83,84 Fetoscopy for this purpose has now been replaced by ultrasound-directed percutaneous umbilical blood sampling (PUBS), also termed cordocentesis or funipuncture.

Fetal blood chromosome analysis has been used to help clarify purported chromosome mosaicism detected in cultured amniotic fluid cells⁸⁵ or chorionic villi. Rapid assessment of fetal chromosome complement has been accomplished by “direct” cytogenetic analysis of uncultured nucleated blood cells.⁸⁶ Short-term fetal lymphocyte cultures can provide a cytogenetic result within 72 hours; direct analysis of spontaneously dividing fetal cells (probably nucleated red blood cells) can provide a karyotype result within 24 hours. This proves particularly useful for patients presenting late in the second trimester, when results from amniocentesis would be available only after pregnancy termination would no longer be available. Also, in cases of fetal structural abnormalities or intrauterine growth retardation (IUGR) presenting in the third trimester, rapid results may prove useful for decision making concerning the mode of delivery.^87,88 More recently, fluorescent in situ hybridization (FISH) with chromosome-specific DNA probes has also been used for rapid prenatal diagnosis of aneuploidy using nucleated fetal blood cells from umbilical cord blood, as well as amniotic fluid cells.⁸⁹

Fetal blood sampling is used in the prenatal evaluation of many fetal hematologic abnormalities.⁹⁰ Daffos⁹¹ reported normal hematologic values for second-trimester fetuses, and Forestier and separate colleagues^92,93 reported normal blood chemistry values for second-trimester fetuses. Fetal blood hematocrit can be directly measured to assess hemolysis resulting from Rh or other blood antigen isoimmunization states.⁹⁴ Before this, obstetricians had to rely on indirect evidence of fetal hemolysis, such as maternal antibody titers, past obstetric history, ultrasound findings, and spectrophotometry of bilirubin in amniotic fluid; the need for subsequent transfusions was based on somewhat arbitrary guidelines. Now the decisions about who, when, how much, and how often to transfuse can be made more rationally on the basis of actual fetal blood analyses such as hemoglobin level, hematocrit level, blood group, direct antibody titer, and reticulocyte count. Fetal hemoglobin can be directly evaluated to diagnose sickle cell disease, α- or β-thalassemia or other hemoglobinopathies,⁸³ although these disorders are now usually diagnosed by DNA analysis of chorionic villi or amniotic fluid cells. Fetal blood sampling can be used to assess both the quantity of platelets and the quality of function.⁹⁵ The maternal PLA2 (platelet antigen) alloimmunization PUBS is not only useful for diagnosis, but access to the fetal circulation also allows therapeutic alternatives including in utero platelet transfusion or maternal immunotherapy with gammaglobulin or steroids.⁹⁶

Fetal blood has been used for the diagnosis of blood factor abnormalities such as hemophilia A, hemophilia B, or von Willebrand disease.^93,97 In addition to hematologic studies, fetal blood samples can be used to diagnose autosomal recessive or X-linked immunologic deficiencies, including severe combined immunodeficiency (SCID), ChédiakHigashi syndrome, Wiskott-Aldrich syndrome, and chronic granulomatous disease.^98,99,100

Recovery of fetal blood permits assessment of viral, bacterial, or parasitic infections of the fetus. Detection of fetal viral or parasitic infection is usually made on the basis of maternal antibody titers or ultrasound-directed fetal structural abnormalities. Serum studies of fetal blood allow for quantification of antibody titers.^{101, 102} In addition to antibody studies, PUBS can be used for direct analysis of viral, bacterial, and parasitic infections by culture of fetal blood.^102,103,104

Although many indications for detecting fetal abnormalities that previously required fetal blood sampling are now performed by amniocentesis or CVS using DNA analysis,^89,105,106 the continuing value of fetal blood sampling lies not only with those few remaining diagnostic indications but also with providing the potential for pharmacotherapy. For example, fetal arrhythmias have been treated with direct administration of antiarrhythmic medications and fetal paralysis may be induced to facilitate invasive procedures such as fetal transfusions or for magnetic resonance imaging (MRI).¹⁰⁷

Technique

The technique most commonly used for fetal blood sampling is ultrasound-directed PUBS. The procedure can be safely undertaken from 18 weeks onward, although successful procedures have been reported as early as 12 weeks.^108,109 PUBS is usually performed as an outpatient procedure. Maternal sedation is usually unnecessary, but oral benzodiazepine 1 or 2 hours before the procedure may be of benefit to the anxious patient and usually results in a transient decrease in fetal movement that can facilitate the procedure. A preliminary ultrasonographic examination of the fetus is performed to assess fetal viability, placental and umbilical cord location, fetal or placental anomalies, and fetal position. A suitable site for needle insertion is then selected; the skin over this site is anesthetized with 5 mL of 1% lidocaine. A sterile field is established; the skin is cleansed with an iodine-based solution and sterile drapes applied. The ultrasound transducer is placed on the abdomen away from the sterile insertion site, but at a location that permits visualization of the complete path of the needle from skin to fetal blood vessel.

Several potential sampling sites are available. Owing to its fixed position, the placental cord root is usually the site of choice whenever it is clearly visible (Fig. 5). Alternatively, free loops of cord or the intrahepatic vein are possibilities.^90,110,111

Following percutaneous insertion of the spinal needle into the fetal blood vessel under direct ultrasound guidance, a small amount of blood is aspirated. The presence of fetal blood in this initial sample is confirmed using a model ZBI Coulter counter and channelizer to differentiate fetal from maternal blood on the basis of erythrocyte volume. The amount of blood aspirated for diagnosis depends on the indication for PUBS; rarely does the procedure require more than 5 mL.

On completion of the fetal blood sampling, the spinal needle is withdrawn and an ultrasound examination is performed to evaluate fetal status. The woman and her fetus are monitored for about 1 hour following PUBS. At our centers, all women at risk of Rh isoimmunization receive 300 μg of RhIg following the procedure.

Safety

Fetal blood sampling appears to be a relatively safe procedure when performed by experienced physicians, but it does carry more risk than CVS or amniocentesis. Maternal complications are rare but include amnionitis and transplacental hemorrhage.^97,112 Data from large perinatal centers estimate the risk of in utero death or spontaneous abortion to be 3% or less following PUBS.^{91,101,104,113,114,115,116} Collaborative data from 14 North American centers, sampling 1,600 patients at varying gestational ages and for various indications, revealed an uncorrected fetal loss rate of 1.6%.¹¹⁷

A more recent assessment of fetal blood sampling loss risk was performed by Antsaklis and colleagues,¹¹⁸ who reviewed their clinical outcomes based on indication for fetal blood sampling. They divided their cohort into five main indication subgroups and found highly significant differences in procedure-related loss rates among the five groups. The highest loss rates were observed in the two groups characterized by fetal abnormalities and growth restriction, thus demonstrating that indication for fetal blood sampling has a major impact on risk for the procedure. Unfortunately, no studies directly comparing loss rates in control and treated groups have yet been published.

Prenatal diagnosis of severe hereditary skin diseases (genodermatoses) used to require fetal skin biopsy for ultrastructural or immunohistochemical analysis. Such genodermatoses that have been amenable to diagnosis by fetal skin sampling have included anhidrotic ectodermal dysplasia,¹¹⁹ bullous congenital ichthyosiform dystrophia (epidermolytic hyperkeratosis),^120,121,122 epidermolysis bullosa dystrophia (Hallopean-Siemens),¹²³ harlequin ichthyosis,¹²⁴ hypohidrotic ectodermal dysplasia,¹²⁵ epidermolysis bullosa lethalis,^126,127 nonbullous ichthyosiform erythroderma,¹²⁸ and Sjögren-Larsson syndrome.¹²⁹

Most of these conditions can now be diagnosed prenatally with DNA analysis of chorionic villi or amniotic fluid cells.¹³⁰ Indeed, one of the commoner lethal genodermatoses, Herlitz junctional epidermolysis bullosa, is now amenable to prenatal diagnosis by DNA analysis.¹³¹ However, prenatal diagnosis of harlequin ichthyosis still requires fetal skin sampling, although Akiyama and colleagues¹³² have reported that the diagnosis can now be made earlier than 21 to 22 weeks of gestation.

On such occasions, fetal skin sampling is still required and is best performed at 17 to 20 weeks of gestation. The procedure was initially performed under direct visualization using fetoscopy, a procedure associated with a total fetal loss rate of between 4% to 7%,^133,134 or a procedure-related loss rate of perhaps 2%. From a safety standpoint, it seemed reasonable to assume that the smaller the caliber of the instrument introduced into the uterus, the safer the procedure would be. For this reason, we began to use ultrasound-directed biopsy forceps for fetal skin sampling in the 1980s.¹²⁷ We reported the clinical outcomes of 17 ultrasound-guided fetal skin sampling procedures.¹²⁷ In 5 cases, a fetal skin disorder was diagnosed, and these pregnancies were terminated. In the remaining 12 cases, all infants were delivered without complications at 37 weeks of gestation or later. Despite the use of biopsy forceps that take smaller fetal skin samples than the biopsy forceps used for fetoscopically directed fetal skin sampling, superficial scarring lesions have occasionally been noted.

Prenatal diagnosis of Becker-Duchenne muscular dystrophy is usually possible by DNA analysis. However, in a few families, meiotic recombination or homozygosity of multiple restriction fragment length polymorphisms (RFLPs) precludes a DNA-based diagnosis.^135,136 In these cases, fetal muscle biopsies can be used for immunohistochemical analysis with a fluorescent antidystrophin antibody. Males with Duchenne muscular dystrophy lack dystrophin and with Becker muscular dystrophy lack or demonstrate a variable dystrophin pattern,¹³⁷ whereas muscle biopsies from unaffected male fetuses show normal amounts of dystrophin.

Fetal muscle sampling is usually performed at around 18 weeks of gestation using a technique analogous to fetal skin sampling or fetal liver sampling. Muscle biopsies are taken from the gluteal region by directing a Klear Kut (Baxter, Los Angeles, CA) kidney forceps gun toward the fetal buttock. As with fetal skin and liver biopsy procedures, only a small number of fetal muscle sampling procedures have been reported.^{135, 138, 139} Thus, definitive statements cannot be made concerning the safety of this procedure.

Coelocentesis involves the transcervical or transabdominal aspiration of fluid from the extraembryonic coelom. This procedure has been suggested as a possible method for early first-trimester prenatal diagnosis. Ross and colleagues¹⁴⁰ were successful in aspirating fluid from 48 of 50 (96%) pregnancies between 6 and 10 weeks of gestation. Although the authors were unable to obtain metaphase analyses from these specimens, subsequent work by Cruger and colleagues^141,142 demonstrated success in culturing and karyotyping. In addition, Makrydimas and colleagues¹⁴³ reported successful prenatal detection of beta-thalassemia by coelocentesis.

However, Ross and colleagues¹⁴⁰ reported that short-term pregnancy loss rates associated with coelocentesis were markedly increased. In a comparative study of women undergoing pregnancy termination, 25% of women spontaneously aborted before their termination compared with 5% of women not undergoing coelocentesis (p < .01). These high loss rates suggest that coelocentesis may not be appropriate as a routine prenatal diagnostic test.

Of interest is the work of Santolaya-Forgas and colleagues.¹⁴⁴ This group has investigated coelecentesis in baboons to determine its potential applicability for fetal therapy. In one series, coelocentesis was performed in 9 baboons; fluid aspiration (1 to 5 mL) was successful in 8 cases and 7 of the 8 pregnancies were continuing 140 days after the procedure.¹⁴⁴ Coelocentesis was subsequently performed in 6 other baboons; assessment of the extracoelomic fluid osmometry and electrolyte composition in these 6 samples found that the chorion laeve behaves as a semipermeable membrane at 40 days' gestation,¹⁴⁵ suggesting that it may be useful for maternal-fetal transfer of some substances that could be used for fetal therapy.

Embryoscopy is a relatively new and investigational technique that permits direct visualization of the fetus as early as the first trimester.^146,147 Initially, a rigid fiberoptic endoscope was passed transcervically into the extracoelomic cavity, permitting inspection of fetal anatomic structures; fetal blood sampling was also feasible by this method.¹⁴⁷ However, improvements and advancements in fiberoptic technology have led to the performance of thin-gauge transabdominal and transcervical embryoscopy,^148,149 allowing visualization as early as 4 weeks after conception.^150,151

Initial procedures were performed only on women who had elected pregnancy termination; however, embryoscopy has since been performed on continuing pregnancies.^146,149 Ville and colleagues¹⁴⁹ reported a procedure-related loss rate of 12% when the procedure was performed in the first trimester.

Further studies of the safety, accuracy, and applications of this new modality will be needed before embryoscopy is used as a routine prenatal diagnostic tool. However, the ability to access the embryonic circulation may have important application for therapeutic interventions such as drug, gene, and cell therapy.

Invasive prenatal testing remains a mainstay of obstetric practice in the United States. When provided within a framework of nondirective genetic counseling and supportive follow-up, invasive testing can provide critical information to women and couples concerning pregnancy outcomes. However, effective genetic counseling and follow-up as well as extensive clinician training and procedure evaluation is needed to best provide prenatal diagnosis to women at increased risk for detectable fetal abnormalities. It is hoped that further advances will facilitate access to the fetus and help develop safer invasive procedures. Current work at our and other centers is geared to develop noninvasive methods to screen, and eventually to diagnose, fetal abnormalities that are currently amenable to diagnosis by invasive methods. Indeed, the work evaluating the applicability of fetal cells in maternal blood holds great promise for such non-invasive screening and diagnosis. Until that and other technologies become integrated into clinical care, invasive prenatal testing will remain an important part of obstetrical care.