INTRODUCTION

Sonographically guided chorionic villus sampling (CVS) has been available in the United States since the early 1980s and has offered couples at genetic risk an early and rapid prenatal diagnosis.¹ The procedure, which can be performed as early as 10 weeks of menstrual age, provides preliminary cytogenetic results within 48 hours and final culture results within 7 days. In contrast, genetic amniocentesis is not routinely performed until approximately 16 weeks of menstrual age with an additional 7 to 10 days required to culture the amniotic fluid cells. Fluorescent in situ hybridization (FISH) can be used for both techniques. Thus, the pregnancy is nearly half completed before a definitive diagnosis can be established with amniocentesis. If a significant fetal abnormality is identified, the prospective parents must make a difficult choice of whether to continue or terminate the pregnancy. Postponing this decision until the midtrimester is extremely difficult because fetal movement has been perceived and significant bonding between the parent and fetus has occurred. In addition, the pregnancy is public knowledge, thereby precluding an element of privacy in decision making. If termination is chosen, maternal risks are greater than in the first trimester, with maternal mortality being up to 5 times higher.²

Despite the advantages of CVS, the procedure has struggled to become universally accepted. This has been due predominantly to a perception that the sampling and laboratory procedures are more complex than amniocentesis. In addition, there have been concerns that the procedure may induce fetal limb defects. Recently, however, enthusiasm for CVS has been renewed. First, contemporaneous studies have demonstrated the accuracy of the laboratory results, the reliability of the sampling, and the safety of the procedure if performed after 10 weeks of gestation by experienced operators. Second, studies have established the superior safety to CVS over first trimester amniocentesis.^3-5 Additionally, over the last decade, the complication rates of CVS and midtrimester amniocentesis are comparable due to the reduction of CVS problems.⁶ Third, the recent success of first trimester screening for fetal chromosomal abnormalities provides an impetus for a first trimester diagnostic procedure for fetal karyotype.⁷

CONCEPTS AND INDICATIONS FOR CHORIONIC VILLUS SAMPLING

For years, prenatal diagnosis has relied on the analysis of amniotic fluid fibroblasts as an indirect reflection of the fetal genetic makeup. Similarly, chorionic villi are fetal in origin, and as such are also an appropriate and useful source of tissue for the evaluation of fetal genetic disease. Their cytogenetic, molecular, and biochemical properties reflect those of the fetus. In addition, the villi are partly composed of cytotrophoblast cells, which are an actively dividing source of spontaneous mitoses that can be used to obtain a rapid chromosomal analysis. Finally, villi can be easily obtained without requiring puncture of the chorion or amnion membrane.

With the exception of α-fetoprotein analysis, the indications for CVS are essentially the same as those for amniocentesis. The major indications are listed in Table 27-1. Advanced maternal age (>35 years) is the most common indication, accounting for 90% of procedures.⁸ In addition, parents who have previously had a child with a chromosomal abnormality that may recur are likely to request early invasive testing, as are couples who are carriers of chromosome translocations or autosomal recessive biochemical or molecular diseases. First-trimester prenatal diagnosis is often requested by women who carry sex-linked diseases because of the 50% recurrence risk in male offspring. Recently, screening for trisomies 21 and 18 in the first trimester has become possible by using a combination of biochemical analysis (pregnancy-associated plasma protein A [PAPP-A] and human chorionic gonadotropin [hCG]) and measurement of the fetal nuchal translucency.⁷ If the preliminary work demonstrating almost 90% sensitivity is substantiated, a positive screen could become a major indication for CVS.

HISTORY OF CHORIONIC VILLUS SAMPLING

First-trimester prenatal diagnosis is not a new concept. The ability to sample and analyze villus tissue was demonstrated more than 25 years ago by the Chinese who, in an attempt to develop a technique for fetal sex determination, inserted a thin catheter into the uterus guided only by tactile sensation.⁹ When resistance from the gestational sac was felt, suction was applied and small pieces of villi aspirated. Although this approach seems relatively crude by today’s standards of ultrasonically guided invasive procedures, the diagnostic accuracy and low miscarriage rate demonstrated the feasibility of first-trimester sampling.

In 1968, Hahnemann and Mohr attempted blind transcervical (TC) trophoblast biopsy in 12 patients using a 6-mm-diameter instrument.¹⁰ Although successful tissue culture was obtained, half of these subjects subsequently aborted. In 1973, Kullander and Sandahl used a 5-mm-diameter fiberoptic endocervoscope with biopsy forceps to perform TC CVS in patients requesting pregnancy termination.¹¹ Although tissue culture was successful in approximately half of the cases, two of the subjects subsequently became septic.

In 1974, Hahnemann described further experience with first-trimester prenatal diagnosis using a 2.5-mm hysteroscope and cylindrical biopsy knife.¹² Once again, significant complications, including inadvertent rupture of the amniotic sac, were encountered. By this time, the safety of midtrimester genetic amniocentesis had become well established, and further attempts at first-trimester prenatal diagnosis were temporarily abandoned in the Western hemisphere.

Two technological advances occurred in the early 1980s to allow reintroduction of CVS. The first of these was the development of real-time sonography, making continuous guidance possible. At the same time, sampling instruments were miniaturized and refined. In 1982, Kazy et al reported the first TC CVS performed with real-time sonographic guidance.¹³ That same year, Old reported the first-trimester diagnosis of β-thalassemia major using DNA from chorionic villi obtained by sonographically guided TC aspiration with a 1.5-mm-diameter polyethylene catheter.¹⁴ Using a similar sampling technique, Brambati and Simoni diagnosed trisomy 21 at 11 weeks of gestation.¹⁵

After these preliminary reports, several CVS programs were established in both Europe and the United States, with the outcomes informally reported to a World Health Organization (WHO)-sponsored registry maintained at Jefferson Medical College. This registry and single-center reports were used to estimate the safety of CVS until 1989, when two prospective multicentered studies, one from Canada,¹⁶ and one from the United States,¹⁷ were published and confirmed the safety of the procedure.

CHORIONIC VILLUS SAMPLING: THE PROCEDURE

Procedure-Related Anatomy

Between 9 and 12 weeks after the last menstrual period, the developing gestation does not yet fill the uterine cavity (Figure 27-1). The sac is surrounded by the thick leathery chorionic membranes, within which are both the amniotic cavity and the extraembryonic coelom. The amniotic cavity contains the embryo and is enclosed by the thin, wispy, freely mobile amniotic membrane. The extraembryonic coelom is located between the amniotic and chorionic membranes, contains a tenacious mucoid-like substance, and disappears as the amniotic sac grows toward the chorion and the two membranes become juxtaposed.

Before 9 weeks, chorionic villi cover the entire outer surface of the gestational sac. As growth continues, the developing sac begins to fill the uterine cavity, and most villi regress except at the implantation site, where they are associated with the decidua basalis (see Figure 27-1). Villi in this area rapidly proliferate to form the chorion frondosum, or fetal component of the placenta. Between 9 and 12 weeks of gestation, the villi float freely within the blood of the intervillus space and are only loosely anchored to the underlying decidua basalis.

Sampling Techniques

Sampling by CVS is generally performed between 70 and 91 days after the last menstrual period. This window is chosen to minimize the background spontaneous miscarriage rate that is higher in early pregnancy while still allowing sufficient time for results to be available within the first trimester. The chorion frondosum is easily localized by ultrasound as a hyperechoic homogeneous area by this gestational age (Figure 27-2). In addition, fusion of the amnion and chorion has not yet occurred, thereby decreasing the risk of amnion rupture during the procedure. Sampling significantly earlier in gestation may be associated with an increased risk of fetal abnormalities and should not routinely be done.¹⁸ TC sampling may be more difficult after 12 weeks of menstrual age due to the increasing distance between the cervix and placental site as uterine growth continues.

CVS can be performed by either the TC or the transabdominal (TA) approach (Figure 27-3). The techniques are equally safe and efficacious, and the majority of patients can be sampled by either technique.¹⁹ In most cases, physician or patient preference will dictate which approach is used; however, in approximately 3% to 5% of patients, clinical circumstances will support one approach over the other (Table 27-2), requiring operators to be proficient in both.^19,20 TC CVS is preferred when the placenta is located on the posterior uterine wall, whereas TA sampling is particularly useful when the placenta is implanted in a fundal or high anterior location. TC sampling has the advantage of minimal patient discomfort but is somewhat more difficult to learn.²¹ Both approaches are best performed by using a two-person technique, with one individual performing the sampling and the other guiding the ultrasound. Communication between the sonographer and sampler is imperative, and the best results have come from centers in which a limited number of samplers and sonographers perform CVS.

Transcervical Sampling

TC CVS is performed by using a polyethylene catheter through which a stainless-steel malleable stylet has been inserted. The stylet fits snugly through the catheter and provides sufficient rigidity for adequate passage through the cervix and into the frondosum. The stylet has a rounded, blunt end that protrudes slightly beyond the end of the catheter to prevent sharp edges that may potentially perforate the membranes. The catheter has a luerlock end to accommodate a syringe. The Trophcan catheter (Portex Company, Concord, MA, USA) was previously the one most frequently used in the United States. However, this catheter has recently been removed from the market by the manufacturer, leaving the catheter manufactured by the Cook Company (Spencer, IN, USA) as the only commercially available TC sampling device (Figure 27-4).

Before performing the CVS procedure, ultrasound scanning confirms fetal viability and establishes the area of the chorion frondosum. An approach is mentally mapped that allows catheter placement parallel to the chorionic membrane. Uterine contractions may be present and obstruct or alter the sampling path (Figure 27-5). They may also alter the appearance and location of the placenta by pulling it into unusual locations. When contractions significantly interfere with a proposed sampling path, delaying the procedure for 15 to 30 minutes until they abate is suggested. The presence of large placental lakes should also be noted so that they can be avoided because sampling through these lakes has been associated with increased postprocedure bleeding.²²

The maternal bladder should be sufficiently full to provide an acoustic window through which the vagina, cervix, and uterus can be visualized. Overfilling makes retrieval more difficult by increasing patient discomfort and displacing the uterus out of the pelvis, which extends and fixes the sampling path.

The procedure is performed in the lithotomy position on a standard examination table with foot stirrups. A speculum is inserted, and the vagina and cervix are cleansed with antiseptic solution. The catheter is prepared by slightly curving its distal 3- to 5-cm part with the guidewire in place to allow easy insertion through the cervix. In most cases, only a minimal amount of curvature is required. The cervical canal is then reimaged by ultrasound, and the catheter is introduced through the cervix until loss of resistance at the internal os is felt. Once the sonographer clearly identifies the catheter tip, it is guided by real-time sector scanning to the placental site (Figure 27-6A). The catheter is directed by gently maneuvering the curved periphery of the gestational sac. A greater amount of upward or downward movement of the tip can be accomplished by manipulating the speculum to redirect the angle of approach. Severe bending of the stylet is rarely, if ever, required, but occasionally use of a single-tooth tenaculum on the cervix is needed to alter uterine position.

Insertion of the catheter in the correct tissue plane between the inner uterine wall and gestational sac is critical to safe sampling. Although sonographic guidance is crucial, tactile sensation is equally important. The catheter can be easily advanced if it is in the proper tissue plane, whereas resistance is encountered if it is against the chorionic membrane or uterine wall. A gritty sensation is felt if the catheter is inserted too deeply into the decidua. Slight readjustment of the angle of direction corrects the problem. To ensure an adequate sample, the catheter should be advanced through the full length of the placenta. The guidewire is then removed, and a 20-cc syringe containing approximately 5 cc of a collection medium is attached. The sample is collected by aspiration using negative pressure as the catheter is slowly withdrawn. Slight distortion of the placental surface may be noted sonographically during this process, and larger villus fragments may be visualized as they pass through the catheter lumen.

Transabdominal Chorionic Villus Sampling

Two techniques for TA sampling are presently used. In the single-needle approach a 20-gauge spinal needle is used.²³ Alternatively, some operators perform a double-needle technique that uses an outer guide needle (18-gauge thin wall or a 16- to 17-gauge standard spinal needle) and a smaller sampling needle (20 gauge).²⁴ In general, a 3.5-inch-long needle is sufficient for most patients, but a 5- or 6-inch-long needle should be available for very obese women.

With the single-needle technique, a sampling path is chosen so that the tip of the needle passes within the chorion frondosum parallel to the chorionic membrane. Intervening bowel and bladder must be avoided. The needle tip is first inserted into the myometrium and then redirected parallel to the membrane (Figure 27-7). As with cervical sampling, the needle should be passed through as much villus tissue as possible and remain parallel to the chorionic membrane to avoid inadvertent puncture (Figure 27-6B). Once appropriately placed within the placenta, the stylet is removed, and a syringe containing 5 cc of media is attached. Under continuous suction, four or five to-and-fro passes within the frondosum are made. The needle is then removed from the abdomen while suction is continued. This “vacuuming” technique is required to ensure retrieval of sufficient villus tissue because the diameter of the 20-gauge needle is slightly smaller than that of a TC catheter.

The 2-needle technique uses a slightly larger-gauge spinal needle as a trocar, which is inserted into the myometrium. A thinner (19- to 20-gauge) and longer sampling needle is passed through the trocar into the chorion frondosum. The stylet of the sampling needle is then replaced with a syringe, and sampling is performed as with a single needle.

Both TA sampling approaches appear to be equally safe. The two-needle technique is theoretically less traumatic because the outer trocar remains still during sampling. It also has the advantage of allowing the operator to obtain additional villi by reinserting the sampling needle without requiring a second skin puncture. The single-needle approach is quicker, less uncomfortable, able to retrieve adequate tissue with minimal insertions, and appears to be the technique that has gained widest acceptance. Both techniques have a learning curve, and operator experience does seem to have a bearing on fetal loss rate.²⁵

Confirmation of Adequate Tissue Retrieval

The presence of adequate villus tissue can usually be confirmed by visual inspection of the syringe contents, but occasionally the sample may need to be evaluated under a dissecting microscope. Samples typically contain a mixture of predominantly villi with a small amount of maternally derived decidua. The chorionic villi appear as free-floating, white structures with fluffy, filiform branches (Figure 27-8A). Contaminating decidua tissue has a more amorphous appearance and lacks distinct branches. Although these two tissues can usually be grossly distinguished by virtue of their respective morphology, confirmation under a dissecting microscope is required if there is uncertainty that adequate villi have been retrieved. Microscopically, the villi have a distinctive branched appearance. Their surface is punctuated by small buds consisting of an outer syncytiotrophoblast covering and a core of mitotically active cytotrophoblast cells (Figure 27-8B). Within the center of each villus is the mesenchymal core, through which capillaries carrying fetal blood cells course.

A minimum of 5 mg of villus tissue is required for most genetic analyses. If insufficient villi are present with the initial attempt, a second aspiration may be performed without additional risk.⁸ Pregnancy loss rates increase significantly when more than two insertions are required and may be as high as 10% if three attempts are made.^17,26 Therefore, a third pass should only be attempted if successful retrieval seems certain. Before a third attempt, the anatomic relationships should be reevaluated, interfering contractions should have abated, and consideration should be given to sampling by the alternative route. In most experienced centers, more than 99% of patients can be successfully sampled with two or fewer insertions. In our center, we have not had a failed procedure in our last 15,000 patients.

Patients may resume normal physical activity after CVS, although strenuous exercise should be avoided for 24 hours. Sexual abstinence is recommended for a short period of time to minimize any risk of ascending infection. Patients may have some mild vaginal bleeding after CVS; therefore, they should be counseled about this possibility before sampling.

RISKS ASSOCIATED WITH CHORIONIC VILLUS SAMPLING

Bleeding

Vaginal bleeding is uncommon after TA CVS but is seen in 7% to 10% of patients sampled transcervically. Minimal spotting is a common occurrence and may occur in almost one-third of women sampled by the TC route.¹⁷ In most cases, the bleeding is self-limited, and the pregnancy outcome is excellent. However, a subchorionic hematoma may be visualized immediately after sampling in up to 4% of TC samples.²⁷ The hematoma usually disappears before the 16th week of pregnancy and is only rarely associated with adverse outcome. Of the more than 15,000 CVS procedures performed in our center, we have never needed to terminate a pregnancy or admit a patient for excessive postprocedural bleeding.

Cases of heavy bleeding and resulting hematoma formation occur from accidental placement of the TC catheter into the vascular decidua basalis underlying the chorion frondosum. In extreme cases, the development of the hematoma can actually be seen on ultrasound. In most of these cases, a gritty feeling indicates penetration into the decidual layer. Careful attention to the feel of the catheter and avoidance of unnecessary manipulation can prevent most of these hemorrhagic episodes and minimize this complication.

Infection

Since the initial development of TC CVS, there has been concern that transvaginal passage of an instrument would introduce vaginal flora into the uterus. This possibility was confirmed by cultures that isolated bacteria from up to 30% of catheters used for CVS.^28-30 However, in clinical practice, the incidence of post-CVS chorioamnionitis is low.^16,17,31,32 In a recently published US study of more than 2000 cases of TC CVS, infection was suspected as a possible etiology of pregnancy loss in only 0.3% of cases.¹⁷ Infection after TA CVS also occurs and has been demonstrated, at least in some cases, to be secondary to bowel flora introduced by inadvertent puncture by the sampling needle.

In our own series of more than 15,000 procedures in which prophylactic antibodies are not used, we have not observed any cases of chorioamnionitis requiring uterine evacuation. Our incidence of periabortion chorioamnionitis was 0.08% for both TC and TA sampling; this rate is about the same as that seen in series of spontaneous abortions that have not been sampled.^33,34 At present, because of the clinically low incidence of post-CVS chorioamnionitis, routine pre-CVS vaginal or cervical cultures for any organism other than gonococcus is not indicated.

Early in the development of TC CVS, two life-threatening pelvic infections were reported.^35,36 Each initially presented with a mild prodrome of maternal myalgias and low-grade fever without localized adnexal or uterine tenderness and subsequently led to maternal sepsis. Both occurred early in the respective center’s experience, and in both the same catheter was used for repeat insertions. Since these reports, a practice of using a new sterile catheter for each insertion has been universally adopted, with only exceedingly rare reports of serious infectious complications.

Ruptured Membranes

Acute rupture of the membranes, documented by either obvious gross fluid leakage or a decrease in measurable amniotic fluid on ultrasound evaluation, is a very rare complication of CVS.^17,37 In our own experience, acute rupture of the membranes has not occurred. Experimental attempts to rupture membranes intentionally with a TC catheter have confirmed that the chorion can withstand significant punishment without perforation.

Gross rupture of the membranes days to weeks after the procedure is acknowledged as a possible post-CVS complication. Delayed rupture can result from either mechanical injury to the chorion at the time of sampling with rupture from exposure of the amnion, or chronic irritation or inflammation from a hematoma on low-grade infection, allowing exposure of the amnion to subsequent damage or infection. One group reported a 0.3% incidence of delayed rupture of the membranes after CVS,³² a rate confirmed by Brambati et al.²⁷

Unexplained midtrimester oligohydramnios has been suggested as a rare complication of TC CVS and may occur from delayed chorioamnion rupture with slow leakage of amniotic fluid.³⁷ These cases are frequently associated with postprocedure bleeding and an elevated maternal serum α-fetoprotein (MSAFP). Operator experience will markedly reduce the risk of this complication, probably by decreasing hematoma formation with its potential to serve as either a nidus for a smoldering infection or a chemical irritant of the membranes.

Elevated MSAFP

An acute rise in MSAFP after CVS has been consistently reported, implying a detectable degree of fetal maternal bleeding.^38-40 The elevation is transient, occurs more frequently after TA CVS, and appears to be dependent on the quantity of tissue aspirated.⁴⁰ Some studies have also demonstrated a correlation between the degree of elevation and the incidence of pregnancy loss.⁴¹ Levels will drop to normal ranges by 16 to 18 weeks, which allows neural tube defect (NTD) serum screening to proceed according to usual prenatal protocols.

Rh Isoimmunization

In Rh-negative women, the otherwise negligible fetal maternal bleeding that follows CVS accrues special importance because Rh-positive cells in volumes as low as 0.1 mL have been shown to cause Rh sensitization.⁴² Because all women with even a single pass of a catheter or needle show detectable rises in MSAFP, it seems prudent that all Rh-negative nonsensitized women undergoing CVS receive Rho (D) immunoglobulin subsequent to the procedure.

The potential for a CVS-induced maternal-to-fetal transfusion to worsen already existing Rh immunization has been described, suggesting that a patient who is already sensitized represents a contraindication to the procedure.⁴³

Pregnancy Loss

Multiple reports from individual centers have demonstrated the safety and low pregnancy loss rates after CVS.^8,44-51 In experienced centers, the rate of miscarriage from the time of CVS until 28 weeks of gestation is approximately 2% to 3%.¹⁹ However, to determine the incidence of procedure-induced pregnancy loss, adjustments for the relatively high background loss at this gestational age must be made.

First-trimester spontaneous abortion in women not undergoing CVS is a common event, occurring in one in every six clinically recognized pregnancies.⁵² However, miscarriage rates after ultrasound confirmation of a viable gestation are expected to be less. Simpson et al reported that, when ultrasound confirmation of fetal viability was noted at 8 weeks, 3.2% of 220 women with a mean age of 30 years aborted.⁵³ Christiaens and Stoutenbeek noted a 3.3% fetal loss rate in 274 women with proven fetal viability at 10 weeks.⁵⁴ Because the majority of women undergoing CVS are older than 35 years and the spontaneous miscarriage rate increases with advancing maternal age, this variable must also be considered. Wilson et al found a total fetal loss rate after proven viability by first-trimester ultrasonography of 1.4% in women younger than 30 years, 2.6% in those between 30 and 34 years old, and 4.3% in women older than 35 years.⁵⁵ It appears that the best estimate of the background spontaneous miscarriage rate in a population of women similar to those undergoing CVS is approximately 2% to 3%. Although this rate is similar to the postprocedure loss rate in other centers, a randomized clinical trial is necessary to quantify the procedure-induced risk precisely. Unfortunately, no randomized comparison of sampled with unsampled patients is likely; however, comparisons to amniocentesis have been performed.

Because the background loss rate is higher in the first-trimester than in the second, any comparison of CVS to second-trimester amniocentesis must enroll all patients before the gestational age at which CVS is performed. The total loss rates can then be compared. All losses must be included, whether from a spontaneous miscarriage or an induced termination for abnormal results. This approach eliminates any bias that may occur when comparing procedures performed at significantly different gestational ages, and also takes into account cytogenetically abnormal embryos that miscarry before an amniocentesis, which would be electively terminated after CVS.

The largest demonstrations of data evaluating the relative safety of CVS and amniocentesis come from three recent collaborative reports. In 1989, the Canadian Collaborative CVS-Amniocentesis Clinical Trial Group reported its experience with a prospective, randomized trial comparing TC CVS with second-trimester amniocentesis.¹⁶ During the study period, patients across Canada were only able to undergo CVS in conjunction with the randomized protocol. There were 7.6% fetal losses (spontaneous abortions, induced abortions, and late losses) in the CVS group and 7.0% in the amniocentesis group. Thus, in desired pregnancies, an excess loss rate of 0.6% for CVS over amniocentesis was obtained; this difference was not statistically significant.

Two months after the publication of the Canadian experience, the first American collaborative report appeared.¹⁷ This study was a prospective, although nonrandomized, trial of more than 2200 women who chose either TC CVS or second-trimester amniocentesis. Patients in both groups were recruited in the first-trimester of pregnancy. As in the Canadian study, advanced maternal age was the primary indication for prenatal testing. When the loss rates were adjusted for slight group differences in maternal and gestational ages at enrollment, an excess pregnancy loss rate of 0.8% referable to CVS over amniocentesis was calculated, which was not statistically significant.

Although both North American trials showed no statistical difference in pregnancy loss when CVS was compared with amniocentesis, a prospective, randomized collaborative comparison of more than 3200 pregnancies sponsored by the European MRC Working Party on the Evaluation of CVS demonstrated a 4.6% greater pregnancy loss rate after CVS (95% confidence interval [CI], 1.6%-7.5%).³⁶ This difference reflected more spontaneous deaths before 28 weeks of gestation (2.9%), more terminations of pregnancy for chromosomal anomalies (1.0%), and more neonatal deaths (0.3%) in the CVS group.

The factors responsible for the discrepant results between the European and North American studies remain uncertain, but it is probable that inadequate operator experience with CVS accounted for a large part of this difference. The US trial consisted of 7 centers and the Canadian trial 11 centers, whereas the European trial included 31 sampling sites. There were, on average, 325 cases per center in the US study, 106 in the Canadian study, and only 52 in the European trial. Although no significant change in pregnancy loss rate was demonstrated during the course of the European trial, it appears that the learning curve for both TC and TA CVS may exceed 400 or more cases.^56,57 Operators having performed fewer than 100 cases may have two or three times the postprocedure loss rate of operators who have performed more than 1000 procedures.

The consensus of the recent literature indicates that with experienced operators, the procedural complication rates with CVS and amniocentesis is comparable; however, CVS is more difficult to learn.⁵

There have been similar comparisons between CVS and early amniocentesis, defined as amniocentesis performed before 14 weeks of gestation. In these comparisons of two first-trimester procedures, consideration of gestational age differences is not necessary. Nicolaides et al compared TA CVS with amniocentesis performed between 10 and 13 weeks gestation.⁵⁸ In this prospective comparison, the spontaneous loss rate was significantly higher after early amniocentesis (5.3%) than after CVS (2.3%). Also, a significant increase in the incidence of talipes equinovarus was seen after early amniocentesis. In another recent comparison, Sundberg et al randomized patients to either amniocentesis between 11 and 13 weeks or TA CVS between 10 and 12 weeks.³ Although the initial end point of this trial was intended to be pregnancy loss, the trial was stopped early because of an increased risk of talipes equinovarus in the early amniocentesis group. Although the power of the trial to compare fetal loss rates was limited by the incomplete sample, no significant difference was demonstrated. The amniocentesis loss rate, however, was 0.6% higher. Leakage of amniotic fluid after sampling occurred significantly more frequently after amniocentesis. Overall, the higher loss rates, increased risk of fluid leakage, and subsequent club foot deformity with early amniocentesis suggest that CVS is the preferred technique for first-trimester sampling.

PREGNANCY LOSS: TRANSCERVICAL VERSUS TRANSABDOMINAL CHORIONIC VILLUS SAMPLING

Randomized trials have compared the TC and TA appro­aches.^19,57,59-61 The US collaborative CVS project performed a randomized, prospective study and found no difference in the postprocedure pregnancy loss rates between the two approaches (TC, 2.5%; TA, 2.3%).¹⁹ Equally important was that the overall post-CVS loss rate in the study (2.5%) was 0.8% lower than that in the initial US study, which compared CVS with second-trimester amniocentesis. Because 0.8% was the quantitative difference in loss rates between amniocentesis and CVS in the original study, this finding suggests that, when centers become equivalently experienced, amniocentesis and CVS may have the same risk of pregnancy loss.

Smidt-Jensen et al, pioneers of TA CVS, added additional information to the comparative safety of the procedures.⁶¹ In a prospective, randomized study, they found no difference in pregnancy loss between TA CVS and second-trimester amniocentesis, but the study did demonstrate an increased miscarriage risk for TC CVS, the procedure for which their center was least experienced. Chueh et al, in a retrospective review of more than 9000 CVS procedures, showed that in their center TC CVS had a slightly greater risk of pregnancy loss than TA sampling.⁶² It appears safe to speculate that fetal loss rates between TC and TA sampling will be similar in most centers once equivalent expertise is gained with either approach. Integration of both methods into the program of any single center will offer the most complete, practical, and safe approach to first-trimester diagnosis.

RISK OF FETAL ABNORMALITIES AFTER CHORIONIC VILLUS SAMPLING

It has recently been suggested that CVS may be associated with the occurrence of specific fetal malformations. The first suggestion of this was reported by Firth et al.⁶³ In a series of 539 CVS-exposed pregnancies, they identified five infants with severe limb abnormalities, all of which came from a cohort of 289 pregnancies sampled at 66 days of gestation or less. Four of these infants had the unusual and rare oromandibular-limb hypogenesis syndrome, and the fifth had a terminal transverse limb reduction defect. Oromandibular-limb hypogenesis syndrome occurs with a birth prevalence of 1 per 175,000 live births,⁶⁴ and limb reduction defects occur in 1 per 1690 births.⁶⁵ Therefore, the occurrence of these abnormalities in more than 1% of CVS-sampled cases raised strong suspicion of an association. In this initial report, all of the limb abnormalities followed TA sampling performed between 55 and 66 days of gestation.

Subsequent to this initial report, others added supporting cases to this list. Using the Italian multicenter birth defects registry, Mastroiacovo et al reported, in a case control study, an odds ratio of 11.3 (CI 5.6-2.13) for transverse limb abnormalities after first-trimester CVS.⁶⁶ When stratified by gestational age at sampling, pregnancies sampled before 70 days had a 19.7% increased risk of transverse limb reduction defects, whereas patients sampled later did not demonstrate a significantly increased risk. Other single-center and case control studies, however, have been inconclusive about an association of CVS with limb reduction defects, with the majority demonstrating no increased risk (Table 27-3).

There is support of the notion that CVS may increase the risk of limb defects when sampling is performed before 63 days of gestation. Most notably, Brambati et al, an extremely experienced group who have reported no increased risk of limb defects in patients sampled after 9 weeks, have reported a 1.6% incidence of severe limb reduction defects when patients were sampled at 6 and 7 weeks.⁶⁷ This rate decreased to 0.1% for sampling at 8 to 9 weeks. Hsieh et al, in a report of the Taiwan CVS experience, reported 29 cases of limb reduction defects after CVS from September 1990 until June 1992; four cases had oromandibular-limb hypogenesis syndrome.⁶⁸ There were two remarkable aspects of this report. First, although the gestational age at sampling was not known with certainty in all cases, the majority were performed at less than 63 days after the last menstrual period. Second, very inexperienced community-based operators performed the cases with limb reduction defects, whereas no defects were seen from the major centers. This experience suggests that very early sampling with excessive placental trauma may be etiologic in some reports of post-CVS limb reduction defects.

The question continues to be debated of whether CVS sampling after 70 days has the potential of causing more subtle defects, such as shortening of the distal phalanx or nail hypoplasia.⁶⁹ At present, there are few data to substantiate this concern. On the contrary, most experienced centers performing CVS after 10 weeks have not seen an increase in limb defects of any type. A recent review of more than 200,000 CVS procedures reported to the WHO registry was reported and demonstrated no increase in the overall incidence of limb reduction defects after CVS or in any specific type or pattern of defect.⁷⁰ In a similar review of more than 65,000 procedures performed in 10 of the most experienced centers in the world, no increase in limb reduction defects was identified.⁷¹

Mechanisms by which early CVS could potentially lead to fetal malformations continue to be disputed. Placental thrombosis with subsequent fetal embolization has been raised as a potential etiology but is unlikely because fetal clotting factors appear to be insufficient at this early gestational age. Inadvertent entry into the extraembryonic coelom with resulting amnionic bands has also been raised as a potential mechanism, but this appears unlikely as well because actual bands have not been observed in the majority of the cases. In addition, many of the cases of oromandibular-limb hypogenesis syndrome had internal central nervous system anomalies that cannot be accounted for by fetal entanglement or compression.

Uterine vascular disruption appears to be the most plausible mechanism at present.⁶⁴ In this hypothesis, CVS causes placental injury or vasospasm that subsequently results in underperfusion of the fetal peripheral circulation. After the initial insult, there may be subsequent rupture of the thin-walled vessels of the damaged distal embryonic circulation, leading to further hypoxia, necrosis, and eventual resorption of preexisting limb structures. A similar mechanism leading to limb defects has been demonstrated in animal models after uterine vascular clamping, maternal cocaine exposure, or even simple uterine palpation.^71,72

In a recent report, Quintero et al added additional information about a possible etiology.^73,74 Using TA embryoscopic visualization of the first-trimester embryo, they demonstrated the occurrence of fetal facial, head, and thoracic ecchymotic lesions after traumatically induced detachment of the placenta with subchorionic hematoma formation. No changes in fetal heart rate were seen. Although these lesions consistently appeared after major physical trauma to the placental site, they were not able to be produced by the passage of a standard CVS catheter.

Any theory of CVS-induced limb defects must consider that there are different stages of fetal sensitivity and should demonstrate a correlation between the severity of the defects and the gestational age at sampling. Firth et al recently presented evidence that appears to illustrate that sampling before 9 weeks of gestation induces the most severe and proximally located fetal limb defects.⁷⁵ These severe defects are not seen after later CVS. Alternatively, Froster and Jackson reviewed the severity of the post-CVS limb defects reported to the WHO registry and found no such correlation.⁷⁰

At the present time, patients planning to have CVS can be counseled that there is no increased risk of severe limb defects if CVS is performed after 70 days of gestation.⁷⁶ They should be made aware of the present controversy concerning more subtle defects and reassured that this has not been seen in most experienced centers. If such a risk does exist, the magnitude based on case control studies can be estimated to be no higher than 1 in 3000.⁷⁶ Ideally, centers performing CVS should have aggressive follow-up systems in place and be capable of giving patients information about the rate of congenital abnormalities in their center. Sampling before 10 weeks of gestation should be limited to very exceptional cases, and these patients must be informed of a 1% or higher risk of limb reduction defects.^{77-90,124-137}

PERINATAL RISKS AND IMPACT ON LONG-TERM DEVELOPMENT OF THE INFANT

No increases in preterm labor, premature rupture of the membranes, small-for-gestational-age infants, maternal morbidity, or other obstetric complications have occurred in sampled patients.⁹¹ Although the Canadian collaborative study showed an increased perinatal mortality in CVS sampled patients, with the greatest imbalance being beyond 28 weeks, no obvious recurrent event was identified.¹⁶ To date, no other studies have seen a similar increase in perinatal loss.

Long-term infant follow-up has been performed by Chinese investigators, who evaluated 53 children from their initial placental biopsy experience of the 1970s. All were reported in good health, with normal development and school performance.⁹²

LABORATORY ASPECTS OF CHORIONIC VILLUS SAMPLING

CVS is now considered a reliable method of prenatal diagnosis, but early in its development incorrect results were reported.^93-95 The major sources of these errors included maternal cell contamination and misinterpretation of mosaicism confined to the placenta. Today, genetic evaluation of chorionic villi provides a high degree of success and accuracy, in particular with regard to the diagnosis of common trisomies.^96,97 In 1990, the US collaborative study reported a 99.7% rate of successful cytogenetic diagnosis, with 1.1% of the patients requiring a second diagnostic test, such as amniocentesis or fetal blood analysis to further interpret the results.⁹⁶ In most cases, the additional testing was required to delineate the clinical significance of mosaic or other ambiguous results (76%), and laboratory failure (21%) and maternal cell contamination (3%) also required follow-up testing. Continued experience has almost eliminated maternal cell contamination as a source of clinical errors. In addition, we now have a better understanding of the biology of the placenta so that confined placental mosaicism (CPM) no longer leads to incorrect diagnosis, but provides us with information predictive of pregnancy outcome and can serve as a clue to the presence of uniparental disomy (UPD). Therefore, an understanding of villus morphology and CVS laboratory techniques is required to provide correct clinical interpretation.

Chorionic villi have three major components: (1) an outer layer of hormonally active and invasive syncytiotrophoblast, (2) a middle layer of cytotrophoblast from which syncytiotrophoblast is derived, and (3) an inner mesodermal core containing blood, capillaries for oxygen, and nutrient exchange (Figure 27-8B). After collection, the villi are cleaned of any adherent decidua and then exposed to trypsin to digest and separate the cytotrophoblast from the underlying mesodermal core. The cytotrophoblast has a high mitotic index, with many spontaneous mitoses available for immediate chromosomal analysis. The liquid suspension containing the cytotrophblast is either dropped immediately onto a slide for analysis or may undergo a short incubation.^98-100 This “direct” chromosomal preparation can provide preliminary results within 2 to 3 hours. However, most laboratories now use overnight incubation to improve karyotype quality and thus report results within 2 to 4 days (Figure 27-9). The remaining villus core is placed in tissue culture and is typically ready for harvest and chromosome analysis within 1 week.¹⁰¹ The direct method has the advantage of providing a rapid result and minimizing the decidual contamination, whereas tissue culture is better for interpreting discrepancies between the cytotrophoblast and the actual fetal state. Ideally, both the direct and culture methods should be used because they each evaluate slightly different tissue sources. Abnormalities in either may have clinical implications. However, the direct preparation is labor intensive, adds additional cost, and is not routinely available in some laboratories.

MATERNAL CELL CONTAMINATION

Chorionic villus samples typically contain a mixture of placental villi and maternally derived decidua. Although specimens are thoroughly washed and inspected under a microscope after collection, some maternal cells may remain and grow in the culture. As a result, two cell lines, one fetal and the other maternal, may be identified. In other cases, the maternal cell line may completely overgrow the culture, thereby leading to diagnostic errors including incorrect sex determination^8,102-104 and potentially to false-negative diagnoses, although there are no published reports of the latter. Direct preparations of chorionic villi are generally thought to prevent maternal cell contamination,^100,103 whereas long-term culture has a contamination rate ranging from 1.8% to 4%.¹⁰⁴ Because maternal decidua has a low mitotic index in contrast to cytotrophoblast, it is highly desirable for laboratories to offer a direct chromosomal preparation and a long-term culture on all samples of chorionic villus. Even in culture, the contaminating cells are easily identified as maternal and should not lead to clinical errors. Interestingly, for reasons still uncertain, maternal cell contamination occurs more frequently in specimens retrieved by the TC route.¹⁰⁴

Contamination of samples with significant amounts of maternal decidual tissue is almost always due to small sample size, making selection of appropriate tissue difficult. In experienced centers in which adequate quantities of villi are available, this problem has disappeared. Choosing only whole, clearly typical villus material and discarding any atypical fragments, small pieces, or fragments with adherent decidua will avoid confusion.¹⁰⁵ Therefore, if the initial aspiration is small, a second pass should be performed rather than risk inaccurate results. When proper care is taken and good cooperation and communication exists between the sampler and the laboratory, even small amounts of contaminating maternal tissue can be absent.

Fluorescent in situ hybridization (FISH) for common chromosomal abnormalities can be helpful in reaching a rapid diagnosis (within hours) without the concern for maternal contamination.

CONFINED PLACENTAL MOSAICISM

The second major source of potential diagnostic error associated with CVS is mosaicism confined to the placenta. Although the fetus and placenta have a common ancestry, chorionic villus tissue will not always reflect fetal genotype.^96,106 Although there was concern that this might invalidate CVS as a prenatal diagnostic tool, subsequent investigations have led to a clearer understanding of villus biology so that accurate clinical interpretation is now possible. This understanding has also revealed new information about the etiology of pregnancy loss, discovered a new cause of intrauterine growth retardation, and clarified the basic mechanism of UPD.

Discrepancies between the cytogenetics of the placenta and fetus occur because the cells contributing to the chorionic villi become separate and distinct from those forming the embryo in early development. Specifically, at approximately the 32- to 64-cell stage, only three to four become compartmentalized into the inner cell mass (ICM) to form the embryo, and the remainder become precursors of the extraembryonic tissues.¹⁰⁷ Mosaicism can then occur through two possible mechanisms.¹⁰⁸ An initial meiotic error in one of the gametes can lead to a trisomic conceptus that normally would spontaneously abort. However, if one of the early aneuploid precursor cells loses one of the chromosomes contributing to the trisomic set during subsequent mitotic divisions, the embryo can be “rescued” by reduction of a portion of its cells to disomy. This will result in a mosaic morula, with the percentage of normal cells dependent on the cell division at which rescue occurred. More abnormal cells will be present when correction is delayed to the second or a subsequent cell division. Because the majority of cells in the morula proceed to the trophoblast cell lineage (processed by the direct preparation), it is highly probable that that lineage will continue to contain a significant number of trisomic cells. Alternatively, because only a small number of cells are incorporated into the ICM, involvement of the fetus will depend on the chance distribution of the aneuploid progenitor cells. Involvement of the mesenchymal core of the villus, which also evolves from the ICM, is similarly dependent on this random cell distribution. Noninvolvement of the fetal cell lineage will produce CPM, in which the trophoblast and perhaps the extraembryonic mesoderm will have aneuploid cells but the fetus will be euploid.

Alternatively, mitotic postzygotic errors can produce mosaicism, with the distribution and percentage of aneuploid cells in the morula or blastocyst dependent on the timing of nondisjunction. If mitotic errors occur early in the development of the morula, they may segregate to the ICM and have the same potential to produce an affected fetus as do meiotic errors. Mitotic errors occurring after primary cell differentiation and compartmentalization have been completed lead to cytogenetic abnormalities in only one lineage.

Meiotic rescue can lead to UPD. This occurs because the original trisomic cell contained two chromosomes from one parent and one from the other. After rescue, there is a theoretical one in three chance that the resulting pair of chromosomes came from the same parent, which is called UPD. UPD may have clinical consequences if the chromosomes involved carry imprinted genes in which expression is based on the parent of origin. For example, Prader-Willi syndrome may result from uniparental maternal disomy for chromosome 15. Therefore, a CVS diagnosis of CPM for trisomy 15 may be the initial clue that UPD could be present and lead to an affected child.^109,110 Because of this, all cases in which CVS reveals trisomy 15 (either complete or mosaic) should be evaluated for UPD by subsequent amniotic fluid analysis. In addition to chromosome 15, chromosomes 7, 11, 14, and 22 are felt to be imprinted and require follow-up.¹¹¹

Recently, there has been evidence that CPM (unassociated with UPD) can alter placental function and lead to fetal growth failure or perinatal death.^108,112-117 The exact mechanism by which abnormal cells within the placenta alter fetal growth or lead to fetal death is unknown. However, the effect may be limited to specific chromosomes. For example, CPM for chromosome 16 leads to severe intrauterine growth restriction, prematurity, or perinatal death, with fewer than 30% of pregnancies resulting in normal full-term infants appropriate for gestational age.^118-125

CVS mosaic results require diligent follow-up by amniocentesis or fetal sampling to determine their clinical significance because, in most cases, if the mosaic results are confined to the placenta, fetal development will be normal. However, if the mosaic cell line also involves the fetus, there may be significant phenotypic consequences. Mosaicism occurs in about 1% of all CVS samples,^{97,104,121,122} but it is confirmed in the fetus in only 10% to 40% of these cases. The probability of fetal involvement appears to be related to the tissue source in which the aneuploid cells were detected and the specific chromosome involved.¹¹⁰ Mesenchymal core culture results are more likely than direct preparation to reflect a true fetal mosaicism.

In a recent review, Phillips et al demonstrated that autosomal mosaicism involving common trisomies (ie, 21, 18, and 13) was confirmed in the fetus in 19% of cases, whereas uncommon trisomies involved the fetus in only 3%.¹²³ When sex chromosome mosaicism was found in the placenta, the abnormal cell line was confirmed in the fetus in 16% of cases. When a nonfamilial marker chromosome was involved, it was confirmed in the fetus in more than one-fourth of cases, whereas mosaic polyploidy was confirmed in only 1 of 28 cases. Chromosomal structural abnormalities were confirmed in 8.6% of cases.

When placental mosaicism is discovered, amniocentesis is frequently performed to elucidate the extent of fetal involvement. When mosaicism is limited to the direct preparation only, amniocentesis appears to correlate perfectly with fetal genotype.¹²³ However, when a mosaicism is observed in tissue culture, both false-positive and false-negative amniocentesis results occur. In these cases, amniocentesis will predict the true fetal karyotype in approximately 94% of cases.¹²³ Most importantly, these discrepancies may involve the common autosomal trisomies. There have been three cases reported of mosaic trisomy 21 on villus culture and a normal amniotic fluid analysis, followed by a fetus or newborn with mosaic aneuploidy.⁹⁶

At present, the following clinical recommendations may be used to assist in the evaluation of CVS mosaicism. Analysis of CVS samples should, if possible, include both direct preparation and tissue culture. Although the direct preparation is less likely to be representative of the fetus, its use will minimize the likelihood of maternal cell contamination, and if culture fails, a nonmosaic normal direct preparation result can be considered conclusive, although rare cases of false-negative results for trisomies 21 and 18 have been reported.^124-128 If mosaicism is found on either culture or direct preparation, follow-up amniocentesis should be offered. Under no circumstances should a decision to terminate a pregnancy be based entirely on a CVS mosaic result. For CVS mosaicism involving sex chromosome abnormalities, polyploidy, marker chromosomes, structural rearrangements, and uncommon trisomies, the patient can be reassured if amniocentesis results are euploid and detailed ultrasonographic examination is normal. However, no guarantees should be made and, as described previously, in certain cases testing for UPD will be indicated. If common trisomies 21, 18, and 13 are involved, amniocentesis should be offered, but the patient must be advised of the possibilities of a false-negative result. Follow-up may include detailed ultrasonography, fetal blood sampling, or fetal skin biopsy. At present, the predictive accuracy of these additional tests is uncertain.

BIOCHEMICAL AND DNA PROCEDURES

Most biochemical and molecular diagnoses that can be made from amnionic fluid or cultured amniocytes can also be made from chorionic villi. In many cases, the results will be available more rapidly and more efficiently by using villi because sufficient enzyme or DNA is present in villus samples to allow direct analysis rather than wait for tissue culture. For example, the analysis of Tay-Sachs disease can be performed in less than 30 minutes using fresh villi.¹²⁹

A discussion of individual biochemical or molecular diagnoses is beyond the scope of this chapter and is impractical because techniques are changing so rapidly. A registry of diagnoses performed by CVS is kept and updated through the World Health Organization by Dr Hans Galjaard in Rotterdam, The Netherlands, and a published summary of the early worldwide experience is available.¹³⁰

It cannot be assumed that biochemical or molecular results from villus tissue will always be a true reflection of the fetal state. Recently, misdiagnosis of the peroxisomal disorder, X-linked adrenoleukodystrophy, from cultured villus cells has been reported.¹³¹ In addition, tests requiring determination of DNA methylation status, such as that for fragile X,¹³² are also not always reliable in villus tissue. This does not, however, preclude CVS from making these prenatal diagnoses because other molecular approaches can be used. It does emphasize that all tests on villus tissue must be validated by testing sufficient numbers of affected and unaffected pregnancies before being used clinically. Because of the rarity and unique aspects of most biochemical and molecular disorders, specific diagnoses are usually performed by only a few laboratories. Before performing a CVS, the clinician should contact the center analyzing the tissue so that the details of testing can be discussed.

CHORIONIC VILLUS SAMPLING IN MULTIPLE GESTATIONS

CVS is a safe and effective approach to examining twins. Not only does it provide results early in pregnancy but, if discordancy is discovered, it also allows the medical and psychological difficulties encountered with selective termination to be minimized. However, it can be technically more demanding because it requires an experienced operator and sonographer. The ideal time to perform a twin CVS is similar to that for singletons. Ultrasound initially identifies placental locations, determines chorionicity, and confirms fetal sizes and viability. Sampling of each sac is independently performed by either a TC or TA approach, with separate passes of a new sampling instrument for each attempt. Because no unique marker is available to ensure that the samples have been retrieved from distinct placentas, it is imperative that insertion of the instrument into each frondosum is certain. Longitudinal and transverse scanning planes should be used to ensure proper location. If any doubt exists, a repeat procedure is required, but with increased experience the need for repeat procedure is rare.¹³³

Contamination of one sample with villi from the other sac is possible and occurs most commonly when retrieval is performed near the dividing membrane, or if a needle or catheter is dragged through one frondosum while sampling another. When the chorions appear fused, sampling near the cord insertion sites, with avoidance of the area of confluence of the two placentas, should prevent contamination and ensure sampling of each fetus. A combination of TC and TA sampling can minimize cotwin contamination by ensuring unique sampling paths. For example, if both chorion frondosa are situated along the anterior uterine wall, the lower one can be sampled transcervically and the upper transabdominally without contaminating either sample. Despite meticulous sampling techniques, cotwin contamination occurs in up to 4% of cases¹²⁰ but is rarely of clinical concern. If the cytogenetic laboratory is aware of the presence of a multiple gestation, the presence of both normal and aneuploid cells in the sample will be correctly interpreted. The possibility of contamination is of greater consequence if sampling is being performed for biochemical analysis, where a small amount of contaminating tissue could lead to an incorrect diagnosis. For this reason, instead of pooling a sample, we recommend analyzing individual villi when biochemical studies are to be performed. An experienced sampler and an aware and knowledgeable laboratory are extremely important in such cases.

The need to map the placental location clearly and accurately is mandatory, because the risk of discordant results is higher early in the pregnancy and selective termination may be required. The relative positions of the placentas will remain stable over the time frame during which results are obtained, so that identification of the affected fetus can usually be determined even 2 or 3 weeks after the procedure. However, if there is uncertainty, a repeat CVS with direct villus analysis can confirm the position immediately before the termination.

Procedure-related loss rates after CVS sampling of twins are well studied. In experienced centers, no increased procedure-related loss risks are seen compared with second trimester amniocentesis.^133,134 Table 27-4 presents overall pregnancy loss rates to 28 weeks of 1.6% to 2.8% in sampled twins. When compared with a contemporaneously sampled group of patients choosing either CVS or second trimester amniocentesis, we found no difference in the overall risk of pregnancy loss (2.9% amniocentesis vs 3.2% CVS). There was, however, a slightly increased risk of losing one fetus in the group sampled by amniocentesis.¹³³

The choice of the appropriate technique for sampling twins depends on a number of factors including locally available skill and expertise. In centers in which amniocentesis and CVS are both available, CVS may be the preferred approach because it provides results 1 month sooner, thus providing earlier reassurance. When discordant results are encountered, complications and loss rates are decreased when selective termination is performed before 16 weeks of gestation.¹³⁵

CHORIONIC VILLUS SAMPLING AND MULTIFETAL PREGNANCY REDUCTION

Multifetal pregnancy reduction (MFPR) to improve perinatal outcome is an unfortunate but accepted part of reproductive medicine. As with selective termination, MFPR is most safely performed in the first-trimester.¹³⁶ Therefore, in high-order gestations at increased risk for a genetic abnormality, CVS before a reduction can avoid the potential need for a later selective termination. The CVS can be performed between 10 and 11.5 weeks of gestation, and a rapid karyotype can be available within 24 to 48 hours, after which the MFPR is performed. Villus mesenchymal core tissue culture is also performed, but results are usually not available for 7 to 10 days. Because of the small increased risk of CPM with the direct preparation, awaiting culture results when time allows is suggested. The positions of the sampled fetuses will be the same when the patient returns for the MFPR. In most cases, only the two fetuses most likely to remain after the MFPR are sampled. If an abnormality is identified, an additional fetus can be sampled at the time of the MFPR, and the patient can return if this is also abnormal. Of 745 MFPR reductions performed at our center, 254 had an initial CVS. Abnormal chromosomal results were present in approximately 2.5% of pregnancies, and the abnormal fetus was terminated as part of the reduction procedure. The pregnancy loss rate to 24 weeks of gestation of those having a preceding CVS was 5.5% versus 5.6% in those having only MFPR. This encouraging outcome is similar to that described by Brambati et al¹³⁷ in which the cohort of patients undergoing CVS before MFPR demonstrated no increased risk of pregnancy loss, prematurity, or SGA infants.

SUMMARY

CVS is a safe technique for first-trimester prenatal diagnosis of genetic disorders. Real-time sonography and technologic advances of sampling instruments have been critical in establishing a safe technique for retrieving villus tissue for genetic analysis. Clinical trials suggest that CVS carries a low risk of pregnancy loss, which is comparable to that of second-trimester amniocentesis. An understanding of the laboratory techniques and human embryology is essential in avoiding diagnostic errors related to CPM.

To maximize outcome, CVS should be performed by an experienced team of physicians, ultrasonographers, and genetic laboratory technicians. Before initiating a CVS program, operators should have considerable experience in the placement of the catheter, which can be achieved either in a formal training program with observation of 50 procedures followed by close hands-on supervision of another 100 cases,⁸ or by supervised practice on pregnancies undergoing subsequent abortion. In addition, it is advisable to limit the number of ultrasonographers assigned to assist in this procedure because sampling success is equally dependent on skillful ultrasound guidance. Similar to the physician retrieving the villi, the guiding sonographer should be knowledgeable in the didactics of CVS sampling and should obtain adequate hands-on training before beginning work in this area. This can be achieved in a formal training program or by visiting centers performing this procedure.⁸ In this setting, CVS will continue to be an important, reliable, and safe contributor to prenatal genetic diagnosis.

HISTORICAL ASPECTS

Amniocentesis is the oldest invasive procedure for prenatal diagnosis. It has been in use for more than 100 years, being first employed in the 19th century for the treatment of polyhydramnios.^138-140 Subsequently, amniocentesis was used to perform procedures such as amniography¹⁴¹ and elective terminations of pregnancy,^142,143 and it gained popularity in the last five decades as a diagnostic tool in the management of pregnancies complicated by isoimmunization¹⁴⁴ as well as in cases of genetic defects. Indeed, Fuchs and Riis¹⁴⁵ reported in 1956 prenatal gender determination by analysis of the sex chromatin of cells obtained by amniocentesis. Human cytogenetics emerged as a field in 1966 when Steele and Breg¹⁴⁶ reported the successful determination of a human karyotype from cultured amniotic fluid cells. A year later, the first prenatal diagnosis of a chromosomal abnormality was reported: a balanced translocation.¹⁴⁷ Trisomy 21 was initially diagnosed in 1968 by Valenti et al,¹⁴⁸ and the first congenital metabolic disorder diagnosed by amniotic fluid analysis was adrenogenital syndrome in 1965.¹⁴⁹

INDICATIONS

Table 27-5 presents the most frequent indications for amniocentesis. The most common indication is prenatal diagnosis. Evaluation of fetal lung maturity has been the leading indication for the procedure in the third trimester. The role of amniotic fluid analysis in the diagnosis of intrauterine infection has increased over the last decade and is discussed in detail later in this chapter. Regardless of the indication for amniocentesis, all patients must undergo detailed counseling about the objectives, procedure, and potential complications of amniocentesis. An informed consent should be signed before the procedure.

TECHNICAL ASPECTS OF THE PROCEDURE

Before the amniocentesis, an ultrasound examination is mandatory to determine fetal viability, gestational age, number of fetuses, placental location, adequacy of amniotic fluid volume, and the presence of any abnormality that might impact the performance of the procedure (ie, uterine leiomyoma, fetal anomaly, etc).

Gestational Age

Genetic amniocentesis is usually performed after 15 weeks (most commonly between 16 and 18 weeks of gestation). Before the introduction of ultrasonographic guidance, early studies of amniocentesis indicated a lower success rate in retrieving amniotic fluid when the procedure was performed at gestational ages equal to or less than 15 weeks.^150,151 The maternal risks associated with late second trimester termination of pregnancy and the psychologic trauma of terminating a pregnancy that is both visible and felt led to the search for alternative procedures for earlier prenatal diagnosis, namely CVS and early amniocentesis.^152-172 Early amniocentesis is discussed in detail in the next section. For a discussion of CVS, please refer to Part 1 of this chapter.

Needle Selection

Amniocentesis is performed with a regular spinal needle. It has been recommended that the needle bore should be between 20 and 22 gauge. Larger-bore needles have been associated with an increased incidence of fetal loss.^150,152 Smaller-bore needles prolong the period of time required to obtain amniotic fluid and are more difficult to guide if intraoperative manipulations are required. The standard length of a spinal needle is 8.89 cm (excluding the hub), but longer needles are required for procedures in obese patients. Currently, commercially available needles are optimized for sonographic visualization. Acoustic visualization can be improved by roughening the needle, and Teflon coating has been added to decrease needle friction. Needles with side orifices are also commercially available. This design allows fluid to flow not only through the needle tip but also through the side orifices. This may have some advantages in cases of severe oligohydramnios. Although in vitro studies have shown that the flow rate during aspiration doubles with 22-gauge needles with side holes when compared with standard spinal needles of the same caliber, the use of these needle types is strictly dependent on personal preference until an adequate clinical trial is performed.

Ultrasound

Ultrasound has become an integral part of the amniocentesis procedure. The term sonographically guided technique refers to the use of ultrasound for selecting the site of needle insertion before amniocentesis; then, the needle is inserted blindly into the amniotic cavity. In contrast, the term sonographically monitored technique describes the continuous use of ultrasound throughout the procedure so that the insertion of the needle and the movement of the fetus are under constant surveillance.¹⁷³ The sonographically monitored technique has been shown to reduce the frequency of bloody and dry taps (and multiple needle insertions) in comparison with the sonographically guided technique.¹⁷³ The monitored technique also adds to the ease and expedience of the procedure, improves the patient’s understanding of amniocentesis, and allows the operator to correct for potential difficulties during the procedure, such as tenting of the membranes or uterine contractions.

Operative Technique

Figure 27-10 shows the instruments used to perform an amniocentesis. After the ultrasound examination, asepsis of the skin is performed with wide boundaries (Figure 27-11), and then the field is draped. Needle insertion is performed under sonographic visualization. Several techniques have been described for this purpose, and we have described a technique that continues to be employed in our institution.^173,174 A sterile coupling agent is applied to the skin (Figure 27-12), and a nonsterile coupling agent is applied to the surface of the transducer, which is placed into a sterile glove or a sterile plastic bag (Figures 27-13 and 27-14). A site for needle insertion is selected, while trying to avoid a localized contraction, a uterine leiomyoma, the umbilicus, the placenta, or the umbilical cord. Once the insertion site has been selected, a finger is placed under the transducer (Figure 27-15). Decoupling of the transducer from the skin’s surface produces a shadow that allows the identification of the needle path. Under direct sonographic visualization, the needle is inserted along the side of the transducer, and the tip, which appears as a bright echo, is continuously monitored throughout the procedure (Figure 27-16). The stylet is removed, and a syringe is attached to the needle to aspirate the amniotic fluid (Figure 27-17). Alternatively, an extension tube is attached to the hub of the needle and connected to the syringe (Figure 27-18). The purpose of the plastic tube is to allow the needle to float freely in the amniotic cavity, thereby decreasing the likelihood of fetal injury. This method also prevents operator movements (ie, sneezing) from being transmitted to the needle. The first 0.5 cc of amniotic fluid is discarded to decrease the likelihood of contamination of the fluid with maternal cells. After obtaining the fluid, the stylet is replaced and the needle is removed. Fetal cardiac activity should always be documented with ultrasound after the procedure.

Needle-guiding devices have been introduced as another technical modification for use in invasive prenatal diagnosis. Some investigators have reported reduced risk of amniocentesis with devices that affix to the ultrasound transducer and mechanically guide the needle used to aspirate amniotic fluid. Since this modification can limit the operator’s ability to maneuver the needle, the technique has been further modified to incorporate an articulated needle guide to combine the advantages of the needle-guided procedure and the free-hand technique.

At the conclusion of the procedure, either the syringe or the tubes in which the amniotic fluid will be transported to the laboratory are labeled with the name of the patient. The patient is asked to verify the identification of her own sample. An amniocentesis report should include the ultrasound findings, number of needle insertions, color, and volume of the amniotic fluid retrieved, whether or not the amniocentesis was transplacental, and any other unusual findings or events occurring during the procedure. The patient is instructed to report any signs of infection or vaginal leakage of fluid. There is no evidence to support restriction of normal activity after an uneventful procedure. If the patient is Rh-negative and not sensitized, Rh immunoglobulin is administered.

An anterior placenta does not contraindicate the procedure, but if a transplacental puncture is necessary, preference is given to the thinnest portion of the placenta. The potential role of transplacental needle passage in procedure-related pregnancy loss in early and second trimester amniocenteses has been assessed. A study of 380 cases of amniocenteses before 14.9 weeks at the University of Tennessee found 147 cases of transplacental needle passage (38.7%).¹⁷⁵ The frequency of bloody fluid was higher in cases of transplacental (n = 6) versus nontransplacental (n = 1) needle passage (P < .05). The rate of pregnancy loss, however, did not differ between cases of transplacental (3.6%) and nontransplacental (3.6%) needle passage. These results were similar to those obtained in second trimester amniocentesis. An investigation of 1000 consecutive patients referred for genetic amniocentesis showed a nonsignificant difference in pregnancy loss between 306 cases with transplacental needle passage (1.96%) and 694 cases without transplacental needle passage (1%, P = .23).¹⁷⁶ Similarly, a retrospective review of 2083 second trimester amniocenteses showed no increased risk of pregnancy loss for 476 patients exposed to transplacental needle passage.¹⁷⁷ None of the studies had adequate power to rule out a difference between groups, and the investigators reported that, when transplacental needle passage was unavoidable, the thinnest accessible portion of placenta was traversed. Amniotic fluid contamination with maternal cells is regularly observed in pregnancies with an anterior placenta, whereas it is rare in cases with or posterior placenta. The maternal cells are therefore thought to be artificially introduced into the amniotic fluid sample during the transplacental puncture.

Local anesthesia is not routinely employed at our institution. Although some patients may benefit from its use, a limitation of local anesthesia is that it requires an additional needle puncture. Moreover, Van Schoubroeck and Verhaeghe¹⁷⁸ conducted a randomized clinical trial including 220 women who underwent genetic amniocentesis between 14 and 20 weeks of gestation. Patients were randomized to receive local anesthesia (lidocaine 1%) or no treatment. The authors demonstrated that there was no significant difference in the mean pain score measured by visual analog scale (VAS) between patients that received local anesthesia and those who did not (1.4 ± 1.5 vs 1.3 ± 1.4, respectively). Interestingly, 97% of patients described the procedure as not painful or tolerable, 79% had anticipated the procedure to be more painful, 59% reported the amniocentesis to have a comparable discomfort as venous blood sampling, and 98% of women stated they would undergo an amniocentesis again if indicated. These findings support the view that pain associated with amniocentesis is usually mild. Harris et al¹⁷⁹ reported that 62% of women undergoing genetic amniocentesis did not have pain or the pain was described as mild, while 31% described the pain as discomforting, and only 7% described it as distressing or horrible. The mean intensity of pain was 1.6 ± 1.3 (on a scale of 0-7). Maternal weight, parity, previous surgery, fibroid tumors, depth of needle insertion, and presence or absence of an accompanying person were not associated with pain intensity. In contrast, pain during amniocentesis was associated with increased maternal anxiety, a history of menstrual cramps, a previous amniocentesis, and insertion of the needle in the lower uterus. Indeed, the degree of pain may depend on location of needle insertion. In a recent study,¹⁸⁰ women undergoing midtrimester amniocentesis were asked to complete a VAS after the amniocentesis. Location of needle insertion was defined as “upper third” or “middle third” based on the distance from uterine fundus to the pubic symphysis. The mean VAS score was 2.7 ± 2.1, and perception of pain was significantly less in patients with needle insertion in the upper third as compared with that of those in the middle third (VAS 2.2 vs 3.9, respectively; P = .002). Previous amniocentesis, transplacental amniocentesis, previous abdominal surgery, and operator’s experience were not associated with pain intensity.¹⁸⁰

Volume of Amniotic Fluid

The volume of amniotic fluid drawn at the time of genetic amniocentesis varies from 8 to 45 cc. Most centers retrieve between 20 and 25 cc. This represents approximately 10% of the mean volume of amniotic fluid at 16 weeks of gestation. The effect of amniotic fluid volume on pregnancy loss has been the subject of contradictory reports. Whereas the American Collaborative Study concluded that the volume removed was unrelated to fetal loss,¹⁵² the Canadian Trial suggested an increased prevalence of neonatal complications with volumes greater than 16 cc.¹⁵⁰ The volume of amniotic fluid retrieved in centers performing early amniocenteses (before 15 weeks) has ranged between 8 and 25 cc. The recommended amount of fluid to be withdrawn is 1 mL per week of gestation. The first 0.5 to 1 mL of fluid aspirated should be discarded to decrease the risk of maternal cell contamination. The time required to replace the volume of fluid aspirated has not been established. Although a 3-hour period has been cited as an average replacement time for midtrimester amniocentesis (16-18 weeks), there are no adequate data to support this view.

INTRAOPERATIVE COMPLICATIONS

Membrane Tenting

The term membrane tenting refers to the separation of the chorioamniotic membrane from the anterior uterine wall during needle insertion. The membranes are carried forward by the needle but are not penetrated. The diagnosis is made when the tip of the needle is visualized within a clearly defined pocket and no fluid is obtained. Tenting of the membranes is a frequent cause of amniocentesis failure, and multiple needle insertions can be overcome by twisting or redirecting the needle. If these maneuvers fail, another approach is to advance the needle into the posterior uterine wall, physically displacing the obstructing membrane down the shaft and away from the tip. Occasionally, patients will be referred to a different center when amniocentesis has failed and significant chorioamniotic separation has occurred. Amniocentesis can be deferred until reattachment is documented (1 or 2 weeks), or a transplacental needle insertion can be attempted because membrane tenting is unlikely to occur at this site since the chorioamniotic membranes are more adherent on the surface of the chorionic plate. Dombrowski et al¹⁸¹ reported a modified stylet technique to overcome failed aspiration of amniotic fluid in cases complicated by tenting of amniotic membranes. The technique of advancing a stylet from a 12.7-cm spinal needle through the original 8.9-cm 22-gauge spinal needle yielded successful aspiration of fluid in 21 of 22 cases of tenting of amniotic membranes. This technique can be used to avoid multiple needle insertions or postponement of the procedure when tenting of membranes persists despite rotating or advancing the needle. Johnson et al¹⁸² reported an association between amniotic membrane tenting and abnormal karyotype, specifically trisomy 21. Tenting was significantly more frequent among trisomy 21 fetuses (29%) compared to only 9% of other chromosomally abnormal pregnancies, and only 8% when the karyotype was normal.

Multiple Needle Insertions

The American Collaborative Study¹⁸³ of midtrimester amniocentesis reported an increased frequency of spontaneous abortion and stillbirth after multiple needle insertions (>2). Similarly, the Canadian Trial¹⁵⁰ found an association between multiple needle insertions (>2) and total fetal losses. Other studies examining this relationship, however, have not confirmed these findings. Since the introduction of sonographic monitoring of amniocentesis, the frequency of dry taps and multiple needle insertions has decreased significantly. For example, only 1.9% of all patients in a trial conducted by Tabor et al¹⁸⁴ required more than one needle insertion, and no patient required more than two needle insertions. These investigators could not demonstrate an association between the number of taps and fetal loss. Another study¹⁸⁵ including amniocenteses performed only in the third trimester reported that 9% of patients required more than one needle insertion (1.4% required three or four attempts). The need for multiple attempts was not associated with the level of training of the operator.

Bloody Taps

Bloody amniotic fluid can result from contamination with maternal blood, fetal blood, or both. The prevalence of this complication has also decreased with the use of the sonographically monitored technique. In third trimester amniocenteses, Gordon et al¹⁸⁵ found that the incidence of bloody amniotic fluid was higher if more than one needle stick was necessary (30% vs 8%, P = .001) or if the needle traversed the placenta (40% vs 3.6%, P = .001). Giorlandino et al¹⁸⁶ reported that in repeated amniocenteses performed on 20 women 2 weeks after the first procedure (due to contamination of the first cell cultures), blood-stained amniotic fluid could be observed in 75% of women. The authors found a significantly higher red blood cell count and hemoglobin concentration in the amniotic fluid of women who had previously undergone a transplacental amniocentesis when compared to those who did not have a transplacental procedure, and speculated that bleeding may occur after withdrawing the needle, leaving this contamination unrecognized in most cases. Yet it has been observed that, even if contamination with red blood cells occurs during amniocentesis, centrifugation usually prevents hemolysis and hemoglobin release.

The American Collaborative Study¹⁸³ did not demonstrate an increased risk of fetal loss with bloody taps, whereas most other studies did not comment on this issue (ie, the Canadian and British Collaborative Studies).^150,187 Ron et al¹⁸⁸ specifically addressed the clinical significance of blood-contaminated amniotic fluid. They studied 706 women undergoing midtrimester amniocentesis. The first 2 mL of amniotic fluid was examined for the presence of blood from maternal or fetal origin. The prevalence of bloody taps was 25.5% (180 of 706). Maternal contamination occurred in 84.4% (152 of 180) of cases, while fetal contamination was less frequent (15.6%; 28 of 180). A mixture of fetal and maternal blood was present in eight cases. The incidence of spontaneous abortion was significantly greater in patients with bloody taps than in patients with clear taps (maternal blood contamination: 6.6%; fetal blood contamination: 14.3%; and clear taps: 1.7%). The incidence of pregnancy loss when fetal blood contamination occurred was twice that observed when maternal blood contamination was documented. This difference, however, did not reach statistical significance. The investigators did not provide separate risk figures for microscopic versus gross bloody taps.

Fetomaternal Transfusion

Fetomaternal transfusion can be detected by performing a Kleihauer-Betke test after the procedure or by determinations of maternal serum α-fetoprotein (AFP) before and after the procedure. The frequency of fetomaternal transfusion depends on the test employed to diagnose this complication and also on placental location, since fetomaternal transfusion is more common with anterior placentae. The AFP method is more sensitive than the Kleihauer-Betke test. For example, Lele et al¹⁸⁹ reported that the frequency of fetomaternal transfusion was 1.8% and 7%, as detected with the Kleihauer-Betke test and the AFP method, respectively. It has been suggested, however, that in some cases the rise of AFP concentration in maternal serum after the procedure is due to contamination with amniotic fluid and not fetomaternal transfusion. This is important because fetomaternal transfusion can lead to isoimmunization (see section Isoimmunization) and an increased frequency of pregnancy loss. Mennuti et al¹⁹⁰ reported that spontaneous abortion occurred more frequently in patients with an elevation of maternal serum AFP after amniocentesis than in those with no elevation (14.2% vs 0.98%).

Recently, detection of circulating fetal nucleic acids in maternal blood by use of real-time quantitative polymerase chain reaction (qPCR) has been proposed as a promising tool for noninvasive prenatal diagnosis, as well as for quantifying fetal-maternal hemorrhage after amniocentesis. It has been found that amniocentesis is associated with a significant increase in the fetal DNA concentrations in maternal blood, which may represent transfer of either fetal cells or fetal DNA to the maternal circulation.

Discolored Amniotic Fluid

Brown- and green-stained amniotic fluids are occasionally obtained in midtrimester amniocenteses.^191-194 Whereas brown amniotic fluid is considered an indicator of an intra-amniotic hemorrhage, green fluid has been attributed to meconium staining or an old hemorrhage. Hankins et al¹⁹⁴ detected total and fetal hemoglobin in brown, green, and clear second trimester amniotic fluid samples, with higher concentrations in brown amniotic fluid, and reported that fetal hemoglobin was determined to be 20% to 100% of the total hemoglobin. This evidence suggests that both green and brown amniotic fluids probably represent the occurrence of previous intraamniotic bleeding.¹⁹⁴ Indeed, both green- and brown-stained amniotic fluids are positive for free hemoglobin. Furthermore, they have a similar spectrophotometric pattern that is consistent with the presence of oxyhemoglobin and free hemoglobin.

It has been suggested that the color of the fluid correlates with the amount of hemoglobin present. In vitro experiments, in which amniotic fluid is contaminated with blood, have indicated a sequential color change. Green-colored and brown-colored amniotic fluids were seen after 3 and 7 days of incubation, respectively. This interpretation is consistent with the clinical observation that women with green or brown amniotic fluid have a positive history of vaginal bleeding, but little is known about the etiology of the bleeding episode. Cassell et al¹⁹⁵ reported the recovery of Mycoplasma hominis and Ureaplasma urealyticum in 4 of 33 samples with discolored, second trimester amniotic fluid. Similarly, Gray et al¹⁹⁶ found the amniotic fluid to be discolored in four of eight amniotic fluid samples positive for U urealyticum. It is possible that an intrauterine infection may lead to bleeding, or that bleeding and clot formations provide an adequate nidus for microbial growth. Gomez et al¹⁹⁷ proposed that a subclinical intrauterine infection can cause deciduitis and decidual bleeding, which can be clinically manifested as vaginal bleeding. Indeed, microbial invasion of the amniotic cavity has been implicated as a common etiology of vaginal bleeding, as it was found in 14% of pregnant women with “idiopathic” vaginal bleeding. Moreover, both microbial invasion of the amniotic cavity and vaginal bleeding were associated with subsequent preterm prelabor rupture of membranes (PROMs) and early preterm delivery. It has been proposed that subclinical intrauterine infection can cause deciduitis and decidual bleeding, leading to vaginal bleeding and subsequent adverse pregnancy outcomes.¹⁹⁷ Recently, Vaisbuch et al^198,199 measured amniotic fluid concentrations of total and fetal hemoglobin in normal pregnant women, as well as in patients with spontaneous preterm labor with intact membranes and in those with preterm PROM. The authors found that, in normal pregnancy, the median amniotic fluid total and fetal hemoglobin concentrations at term were significantly higher than that in the midtrimester, and that women at term in labor had higher median hemoglobin concentrations than those not in labor. Among patients with preterm labor and those with preterm PROM, no differences were found in the amniotic fluid concentrations of fetal hemoglobin in the presence or absence of intraamniotic infection/inflammation (IAI). In contrast, the median total hemoglobin concentrations in amniotic fluid were significantly higher in patients with IAI than those without IAI. These observations suggest that the main source of fetal hemoglobin in cases with IAI is of maternal origin, and that the detected fetal hemoglobin may represent the normal percentage of hemoglobin F in the maternal blood. The fact that total hemoglobin was present in the amniotic fluid of normal and complicated pregnancies suggests that detection of hemoglobin in amniotic fluid should not be necessarily considered as an abnormal finding. Of note, no difference was found in the median total hemoglobin concentrations between women with and without transplacental amniocentesis (P = .3).

The existence of free hemoglobin in amniotic fluid and its association with brownish coloration of the amniotic fluid or clinical outcomes has been previously rep­orted,^{191,194,200-202} although there is conflicting evidence regarding the presence of hemoglobin in clear amniotic fluid.^194,200-202 Francoual et al²⁰¹ measured (by colorimetry) total hemoglobin concentration in amniotic fluid obtained by amniocentesis or from the vaginal pool in 78 patients in the late third trimester. A higher mean total hemoglobin concentration was found in “meconium-stained” amniotic fluid than in “clear” amniotic fluid, although the authors reported that hemoglobin was present in most samples of “clear” amniotic fluid.²⁰¹ In contrast, Legge²⁰⁰ examined 208 second trimester amniotic fluid samples for pigmentation (visually and spectrophotometrically) and reported that none of the clear amniotic fluid samples contained abnormal pigmentation, while 14 of the 15 dark brown–colored samples had chemically detectable hemoglobin (14 had hemoglobin A, and 8 also had hemoglobin F).

The possibility that some green-stained amniotic fluid could be due to meconium passage cannot be ruled out entirely. The human fetus produces and can pass meconium before the 20th week of gestation. Although the evidence is not consistent, most studies examining the prognostic significance of dark-stained amniotic fluid have suggested an increased risk of pregnancy loss. The relative risk for spontaneous abortion after retrieval of discolored amniotic fluid has been reported to be 9.9 by Tabor et al.¹⁸⁴ King et al²⁰³ suggested that the prognosis is worse if discolored fluid is associated with an elevated maternal serum AFP determined before the amniocentesis. These findings are at variance with those of Hankins et al,¹⁹⁴ who did not find an increased frequency of poor pregnancy outcome after examining data from 83 patients with dark (n = 6) or green fluid (n = 77) of 1227 women undergoing midtrimester amniocenteses.

AMNIOCENTESIS RISKS

Maternal Risks

Maternal complications are extremely rare. They include perforation of intraabdominal viscera with subsequent intraabdominal infection, bleeding, and blood group sensitization. One case of amniotic fluid embolism has been reported after a third trimester amniocentesis. The patient presented at 32 weeks of gestation with polyhydramnios. After draining 200 cc, the patient developed respiratory distress and disseminated intravascular coagulation. The mother was treated with exchange transfusions and survived, but the fetus died in utero. Severe hemorrhage due to laceration of the inferior epigastric vessels has also been reported after a third trimester amniocentesis. Thorp et al²⁰⁴ reported a maternal death after an uncomplicated second trimester amniocentesis. The patient died from Escherichia coli sepsis and disseminated intravascular coagulation 40 hours after the amniocentesis. The autopsy showed normal-appearing needle entries into the skin and uterus without evidence of bowel adhesion or a needle track through the bowel. The amniotic fluid specimen demonstrated negative leukocyte esterase activity, negative Gram stain for bacteria and white blood cells, and normal glucose and interleukin-6 (IL-6) concentrations. Similarly, Kim et al²⁰⁵ reported a case of a 40-year-old female patient admitted with the diagnosis of fetal death and complaining of fever, chills, progressive dyspnea, hematuria, and abdominal pain 2 days after an amniocentesis was performed at 17 weeks of gestation. The patient developed septic shock and disseminated intravascular coagulation, which was further complicated by an acute myocardial infarction. The maternal blood culture was positive for E coli. The patient survived and recovered approximately 3 weeks after the amniocentesis. Erez et al²⁰⁶ reported two cases of septic abortion diagnosed shortly after uncomplicated amniocenteses performed under sterile technique and continuous ultrasound guidance at 18 and 24 weeks of gestation. The first patient had a history of ulcerative colitis in remission and a total proctocolectomy with an ileal pouch performed 10 years earlier, while the medical history of the second patient was unremarkable. Of interest, both patients developed fever just minutes or a few hours after the amniocentesis. The amniotic fluid cultures were positive for E. coli in the first case, and E. coli as well as Staphylococcus epidermidis in the latter. The authors propose that contamination of the uterine cavity by gastrointestinal flora due to unintended intestinal puncture during needle insertion is a plausible explanation for the first case because of the history of intestinal surgery, and that contamination of the uterine cavity after a sterile amniocentesis technique may also occur by direct inoculation of skin flora, which may be the explanation for the amniotic fluid culture positive for S. epidermidis in the second case.²⁰⁶

Fetal Risks

Fetal complications can be grouped into fetal loss and needle injuries. Fetal loss can be idiopathic or due to direct fetal injury resulting in exsanguination or infection. The term idiopathic refers to unexplained fetal death that occurs during the procedure, and in which postmortem examination yields no demonstrable reason for the demise. In these cases, fetal heart activity is detected before but not after the amniocentesis. A neurogenic mechanism has been postulated, but there is no evidence to support this mechanism.

Fetal Loss

Ager and Oliver²⁰⁷ analyzed the results of 28 different reports of midtrimester amniocenteses published between 1977 and 1985. The total fetal loss rate [(total spontaneous abortions + stillbirths + neonatal deaths)/effective total number of pregnancies] differed considerably across the studies (range: 2.4%-5.2%) [Note: The effective total number of pregnancies: total number of pregnancies – (deaths prior to amniocentesis + total elective abortions + pregnancies lost to follow-up)]. However, the authors concluded that these estimations of risk are questionable because of differences in amniocentesis procedures, patient populations, methods of risk estimation, and problems in the design and analysis of the different studies. Most studies do not have a control group, or the control group was the result of matching. Nonrandomized studies are susceptible to bias. Furthermore, most studies do not have the adequate sample size to detect a significant difference in the risk associated with the procedure.

The traditional rate of pregnancy loss related to amniocentesis recommended by the Centers for Disease Control and Prevention is 0.5%,²⁰⁸ which is different from the one that Tabor et al¹⁸⁴ reported in a randomized controlled trial of genetic amniocentesis in 4606 low-risk women. This study provides the best risk estimation available in the literature. Patients were invited to participate in a study in which they were randomized to have an amniocentesis (study group) or an ultrasound (control group) in the midtrimester, and those at increased risk for chromosomal anomalies, neural tube defects, metabolic disorders, or spontaneous abortions were excluded. Amniocenteses were performed under sonographic guidance with a 20-gauge needle by a group of five physicians.¹⁸⁴ The rate of spontaneous abortion after 16 weeks of gestation was higher in the group that underwent amniocentesis than in the group that had an ultrasound (1.7% vs 0.7%, P < .01). The excess spontaneous abortion rate of 1% (95% confidence interval [CI], 0.3-1.5) corresponds to a relative risk of 2.3 (95% CI, 1.3-4.0). There was a different distribution of the time elapsed between procedure and spontaneous abortion in the study and control groups. The median time interval was 21.5 days (range 5-67) in the study group versus 46.5 days (range 8-70) in the control group. An elevated maternal serum AFP (greater than two multiples of the median for gestational age), perforation of the placenta, and discolored amniotic fluid were identified as risk factors for spontaneous abortion. Their relative risks for fetal loss were 8.3 (95% CI, 2.4-19.8), 2.6 (1.3-5.4), and 9.9 (4.3-22.6), respectively. The researchers pointed out that the 1% increased risk of spontaneous abortion after midtrimester amniocentesis may be an underestimation of the real risk. Termination of pregnancies with fetuses affected with chromosomal abnormalities (identified in the study group and not in the control group) may have artificially reduced the rate of spontaneous abortion of the study group. Contrary to what has been reported in previous studies, no correlation was found between the rate of spontaneous abortion and number of needle insertions, placental site, or experience of the operator. Whereas amniotic fluid leakage occurred more commonly in the study group than in the control group (1.7% vs 0.4%, P < .001), vaginal bleeding occurred with similar frequency (2.4% vs 2.6%). Other obstetric complications such as preterm delivery, spontaneous rupture of membranes, and abruptio placentae had similar prevalence in both groups.

A randomized trial of amniocentesis compared with no amniocentesis to try to define contemporary procedure-related loss rates is not likely to be performed due to feasibility and ethical considerations. Eddleman et al²⁰⁹ attempted to quantify the contemporary procedure-related loss after midtrimester amniocentesis using a database generated from patients who were recruited to the First and Second Trimester Evaluation of Risk (FASTER) for Aneuploidy trial, which included 35,003 patients who either did (study group, n = 3096) or did not undergo midtrimester amniocentesis (control group, n = 31,907). The rate of fetal loss before 24 weeks of gestation was compared between the two groups. The observed difference in crude pregnancy loss rates less than 24 weeks between the groups was 0.06% (1.0% – 0.94%, respectively). This study reveals an amniocentesis-related loss risk of approximately 1 in 1600, which is substantially lower than the traditionally quoted risk of 1 in 200.²⁰⁸ Despite large number of patients included, the nonrandomized design, limited information concerning the needle size used for the procedure, and the experience of clinicians performing the procedures are considered as potential biases of this study. Odibo et al,²¹⁰ in a retrospective cohort study including all pregnant women presenting for prenatal care between 1990 and 2006 at the Department of Obstetrics and Gynecology Washington University, compared the fetal loss rate of women who underwent amniocentesis between 15 and 22 weeks of gestation (n = 11,746) with those who did not have any invasive procedure and had a live fetus documented on ultrasound examination between 15 and 22 weeks (n = 39,811). The fetal loss rate less than 24 weeks attributable to amniocentesis was 0.13% (1 of 769). It is possible that increased quality image of the current ultrasound machines and experience of clinicians performing amniocentesis may account, in part, for the reduction in the complication rate associated to amniocentesis in the last decade.

The Practice Bulletin of the American College of Obstetricians and Gynecologists published in 2007²¹¹ states that early amniocentesis (procedure conducted at less than 15 weeks of gestation) should not be performed because of the higher risk of pregnancy loss and complications compared with traditional amniocentesis (procedure conducted at 15 weeks of gestation or later). It also states that amniocentesis at 15 weeks of gestation or later is a safe procedure. According to the Bulletin, the procedure-related loss rate after midtrimester amniocentesis is less than 1 in 300 to 500.²¹¹

Recently, an excellent review conducted by Seeds²¹² included 29 studies published between 1976 and 2002 that reported at least 1000 midtrimester amniocenteses each, as well as information about the spontaneous pregnancy loss risk after the procedure and the use of ultrasound guidance during amniocentesis. The author focused on the rate of spontaneous pregnancy loss between the procedure and 28 weeks of gestation. Patients were excluded if they were lost to follow-up or had therapeutic abortions, third trimester losses, or perinatal deaths. A total of 68,119 midtrimester genetic amniocenteses were included and divided into two groups according to the performance of amniocentesis with or without ultrasonographic guidance. The first group included 33,975 amniocenteses with ultrasound only before the procedure. The overall risk of pregnancy loss before 28 weeks of gestation was 2.1%. In contrast, 34,144 amniocenteses were performed under the sonographic monitoring technique with a spontaneous pregnancy loss of 1.4%. Further, when the analysis was restricted only to studies that included control patients, the rate of pregnancy loss between amniocentesis with concurrent ultrasound monitoring and control subjects was significantly different (1.7% vs 1.1%, respectively; P < .0001), with a procedure-related rate of fetal loss of 0.6%. Finally, the author reported nine studies that addressed the impact of transplacental amniocentesis on the risk of pregnancy loss. The aggregate rate of spontaneous pregnancy loss was 1.4% (73 of 5203), similar to the overall pregnancy loss rate after ultrasound-monitored amniocentesis and lower than that reported from controlled studies, suggesting that transplacental amniocentesis in the midtrimester does not increase the rate of pregnancy loss.²¹²

Infection

Amniocentesis may lead to intraamniotic infection by introducing microorganisms into the amniotic cavity (ie, contaminated instruments, passage of the needle through contaminated skin, or intraabdominal viscera). Alternatively, if amniocentesis results in the rupture of membranes, ascending infection may occur. Although blood cultures obtained around the time of the procedure have been negative in a small study,²¹³ and antibiotic prophylaxis is not a routine practice, the midtrimester period seems to be particularly vulnerable to microbial invasion as the antibacterial activity of amniotic fluid is at its nadir.²⁰⁷ Some case reports have suggested a temporal relationship between the procedure and clinical chorioamnionitis; however, the prevalence of intraamniotic infection after midtrimester amniocentesis is unknown.

The traditional view is to consider intraamniotic infection as an acute complication of pregnancy. Yet, there is evidence implicating infection as an etiologic factor for pregnancy loss after midtrimester amniocentesis. Evidence from studies of the microbiological state of the amniotic cavity using traditional culture techniques or PCR at the time of genetic amniocentesis in asymptomatic patients suggest that, in some cases, intraamniotic infection may precede the procedure rather than follow it. Cassell et al¹⁹⁵ were the first to report the presence of genital mycoplasmas in 6.6% (4 of 61) of amniotic fluid samples collected by amniocentesis between 16 and 21 weeks. Two patients had positive cultures for M hominis and two for U urealyticum. Patients with M hominis delivered at 34 and 40 weeks without neonatal complications, whereas those with U urealyticum had premature delivery, neonatal sepsis, and neonatal death at 24 and 29 weeks. In a series of 2641 second trimester genetic amniocenteses reported by Gray et al,¹⁹⁶ all of which were cultured for mycoplasmas, 0.37% (9 of 2461) had a positive culture for U urealyticum. One patient was excluded from the analysis because of a subsequent therapeutic abortion. The perinatal outcome of the other eight patients was compared with 86 patients with complete follow-up having genetic amniocenteses during the same study period and negative cultures for mycoplasmas. Among the eight patients who cultured positive for U urealyticum, 75% (6 of 8) had a spontaneous abortion within 4 weeks of the procedure as compared with only 1.2% (1 of 86) of the patients with negative cultures. The other two patients in the positive culture group delivered prematurely at 24 and 30 weeks, and only one of the infants survived. All eight placentas had evidence of chorioamnionitis on histologic examination. Fifty percent (four of eight) of the samples obtained from Ureaplasma-positive patients were discolored, while only 2.3% (2 of 86) of the fluid samples obtained from Ureaplasma-negative patients were discolored. Discolored amniotic fluid was significantly associated with the presence of U urealyticum in amniotic fluid (P < .001) and an adverse perinatal outcome. Horowitz et al²¹⁴ detected U urealyticum in 2.8% (6 of 214) of amniotic fluid samples obtained between 16 and 20 weeks of gestation. The rate of subsequent adverse pregnancy outcome (fetal loss, preterm delivery, and low birth weight) was significantly higher in patients with a positive amniotic fluid culture than in those with a negative culture (50% vs 12%, P = .035). Similarly, Berg et al²¹⁵ found 1.6% (44 of 2718) positive cultures for U urealyticum/M hominis. Among these 44 patients, 35 were treated with oral erythromycin. Midtrimester pregnancy loss was significantly higher in the untreated than in the treated group (44% vs 11%; P = .04).

Fetal Injuries Associated with Amniocentesis

Table 27-6 presents fetal injuries that have been attributed to amniocentesis and whether or not ultrasound was employed during the procedure. The spectrum of lesions ranges from mild skin dimples to fetal death due to exsanguination.^216-242 Although fetal injuries are generally associated with bloody taps, they have also been reported after clear taps. If a fragment of tissue is retrieved, histologic examination is recommended to establish its origin. Fetal injuries have occurred even with the use of a sonographically monitored technique. The most frequent lesion associated with amniocentesis is skin puncture. Although a cause-effect relationship is difficult to establish, needle injuries should be suspected if the shape of the lesion resembles a needle track or a depressed punctiform scar.

Several ocular injuries have been attributed to amniocentesis. Typically, these are unilateral lesions detected shortly after birth. In one case,²³⁴ the newborn had a small and cloudy eye, a coloboma of the upper lid, and a hazy and edematous cornea. The combination of lesions in the eyelid and cornea suggests that the injury occurred before separation of the eyelids. The mother had an amniocentesis at 19 weeks and the first 2 cc of fluid were bloody, but the subsequent 30 cc were clear. In two cases,^219-223 a red and photophobic eye in the newborn period was subsequently associated with the development of an enlarging cystic mass in the anterior chamber of the eye. The cystic lesions evolved over a period of several months and were lined by a stratified squamous epithelium. Similar findings have been reported for two children with unilateral and progressively large epithelial iris cysts occupying nearly half of the pupil. The lesions were diagnosed at 8 months and 5 years of age, respectively, and both mothers had an amniocentesis performed at 18 and 43 weeks. The children had no history of postnatal ocular trauma.²⁴³ Five other children were reported with lesions attributed to ocular perforation during amniocentesis: the first child had an amniocentesis at 16 weeks and was found at birth to have a cystic lesion communicating with the right lateral ventricle, left homonymous hemianopia, and possible damage to the right optic tract; the second child had an amniocentesis at 30 weeks and presented in the neonatal period with a distorted pupil toward the 3 o’clock position and a small tag or iris drawn up toward the full-thickness corneal scar; the third child had an amniocentesis performed at 16 weeks and presented at birth with left exotrophia, limitation of abduction, and microphthalmia; the fourth child had an amniocentesis performed at 15 weeks and was noted at 3.5 years to have a small adherent leukoma near the limbus; the last child in this series had an amniocentesis performed during the second trimester and was found at 5 years of age to have a small chalazion on its right upper lid with a small full-thickness scar near the limbus.²⁴⁴

A porencephalic cyst in a newborn with two subcutaneous nodules in the right and left occipital region (suggesting a needle track) has been reported.²³⁵ An amniocentesis was performed at 18 weeks. In addition, in utero injection of contrast in the ventricular system during the course of amniograms has been reported. Squier et al²⁴⁵ reported five cases of brain injury after midtrimester amniocenteses (16-18 weeks). In the first case, an amniocentesis was performed under ultrasonographic guidance. The tip of the needle was not seen after its insertion and the fetus moved to the target area. The needle was found to be inserted into the fetal brain. Tissue fragments found in the amniotic fluid corresponded to germinal matrix and choroid plexus. Follow-up ultrasounds were considered normal but the mother decided to terminate the pregnancy. The brain was examined and a subarachnoid hemorrhage at the base of the brain and over the left sylvian fissure was observed, in addition to hemorrhage and necrosis of the deep white matter, lentiform nucleus, and internal capsule. The second and third cases corresponded to amniocenteses performed without ultrasonographic guidance. Both infants developed seizures, cerebral palsy, and developmental delay. In the second case, there was atrophy of the right cerebral hemisphere and a hemispherectomy was performed at 8 years of age. The fourth case was a twin pregnancy that had numerous needle insertions at 16 and 17 weeks of gestation under sonographic guidance. One twin had a skin dimple and atrophy of both left cerebral and cerebellar hemispheres, developed cerebral palsy, and died at 7 years of age due to respiratory complications secondary to cerebral palsy. Finally, the fifth case was found to be hypotonic at the time of delivery and had a skin scar in the occipital region. The patient had an amniocentesis performed without ultrasonographic guidance at 16 weeks of gestation. The infant developed cerebellar hypoplasia and cerebral atrophy, and died because of a broncopneumonia at 2 years of age.²⁴⁵

Thoracic lesions associated with amniocentesis include hemothorax, pneumothorax, and fetal cardiac tamponade. In the abdomen, injuries have ranged from laceration of the liver, kidney, and spleen to ileocutaneous fistula with ileal atresia. In one case, a fragment of tissue retrieved during amniocentesis grew small bowel mucosa in culture, confirming intraoperative bowel injury.²³⁷

Limb lesions have included disruption of the patellar tendon, gangrene of one arm (perforation of the subclavian artery), and an arteriovenous fistula between the popliteal artery and vein. Amniocentesis has been implicated by some investigators in the etiology of amniotic band syndrome; however, there is no agreement regarding a cause-effect relationship between this syndrome and amniocentesis. Two cases of hematomas of the umbilical cord have been reported.^219,241 Fetal exsanguination due to vascular puncture has also been reported.

Other Complications

The frequency of orthopedic congenital anomalies was not different in the amniocentesis and control groups in the study conducted by Tabor et al.¹⁸⁴ These results are comparable with the findings of a case control study in which amniocentesis had not been performed more often in mothers of newborns with orthopedic abnormalities than in a control group.²⁴⁶ In contrast, the British¹⁸⁷ and American¹⁸³ collaborative studies reported an increased incidence of orthopedic abnormalities (talipes equinovarus, congenital dislocation of the hip, and metatarsus abductus) in newborns of mothers who underwent amniocentesis.

Respiratory distress syndrome is considered a potential complication of amniocentesis. Indeed, the prevalence of respiratory distress syndrome was higher in neonates born to mothers in the study group than in those born to mothers in the control group (1.1% vs 0.5%, P < .05).¹⁸⁴ A similar finding was reported for neonatal pneumonia (0.7% vs 9.3%, P < .05). These observations are of considerable interest because they are consistent with those of the British Collaborative Study¹⁸⁷ in which there was an increased incidence of respiratory distress (defined as respiratory difficulties requiring oxygen and lasting more than 24 hours) in neonates born to mothers who had undergone amniocentesis in comparison with those in the control group (1.27% [30 of 2370] vs 0.38% [9 of 2402]).¹⁸⁴ In addition, the mean crying vital capacity, a measure of lung volume, was found to be lower in 10 neonates of mothers undergoing midtrimester amniocentesis in comparison with those in a control group. Studies in monkeys (Macaca fascicularis), specifically designed to study the effect of amniocentesis on lung development, have indicated that a reduction in the number of alveoli and in lung volume can occur after amniocentesis at a period equivalent to 14 to 17 weeks of gestation in humans. Thompson et al²⁴⁷ evaluated the prevalence of respiratory distress and lung growth by measuring the functional residual capacity (FRC) in 74 newborns of mothers who had an amniocentesis at 10 to 13 weeks, and in 86 newborns of mothers who had a CVS during the same gestational age interval. Six infants in the CVS group, but none in the amniocentesis group, required admission to the intensive care unit because of respiratory distress (P < .05). The FRC was not different between the two groups. The overall incidence of FRC values below the 2.5th percentile was higher than expected (9%), indicating that both amniocentesis and CVS performed in the first trimester of pregnancy may impair antenatal lung growth. Although other clinical studies have not demonstrated an increased incidence of pulmonary complications after amniocentesis, their design and sample size are not as adequate as those reported by Tabor et al.¹⁸⁴

Leakage of Amniotic Fluid

Transient vaginal leakage of small volumes of amniotic fluid after genetic amniocentesis occurs in approximately 1% to 2% of all cases and resolves spontaneously within 48 hours. Devlieger et al²⁴⁸ reported a prevalence of 4% of vaginal fluid loss in 100 amniocenteses performed between 14 and 20 weeks of gestation. Tabor et al¹⁸⁴ reported amniotic fluid leakage in 1.7% of patients (39 of 2239). In a study that included 3469 early amniocenteses (11.7-14.9 weeks), 28% were performed at 12 weeks, 51% at 13 weeks, and 21% at 14 weeks. The rate of leakage of fluid decreased with advancing gestational age (2.5%, 1.7%, and 1%, respectively).²⁴⁹ Chronic leakage of amniotic fluid is rare. Of the eight cases reviewed by Crane and Rohland,²⁵⁰ preterm delivery before 32 weeks occurred in three cases, while clubfoot deformity was present in two cases. One death occurred in an infant delivered at 31 weeks of gestation with clubfoot deformity and Potter facies. Although there are not enough data to support definitive conclusions, expectant management was associated with successful pregnancy outcome in six of seven patients who experienced amniotic fluid leakage within 24 hours of a genetic amniocentesis in a series of 603 patients reported by Gold et al.²⁵¹ All patients were placed on bed rest, had digital cervical examination prohibited, frequent white blood cell counts with differential analysis, and close maternal surveillance for clinical evidence of chorioamnionitis. Six patients delivered healthy neonates at term, and one patient had an intrauterine fetal demise at 25 weeks. In this case, an unsuccessful amniocentesis was attempted 6 weeks before delivery.

Long-Term Outcome

Midtrimester amniocentesis has been used as a diagnostic procedure for more than 20 years. A Canadian trial²⁵² compared the long-term outcome (7-18 years) of 1297 children whose mothers had midtrimester amniocentesis for advanced maternal age with a group of 3704 controls matched for maternal age, sex, date, and place of birth. With the exception of a higher risk of ABO isoimmunization (9.5% [6 of 1297] vs 0.03% [1 of 3704]; relative risk 3.3; 95% CI, 2.4-4.5), the offspring of women who had midtrimester amniocenteses did not have more disabilities than those who did not (including cerebral palsy, delayed speech, intellectual impairment, hearing deficits, epilepsy, asthma, and limb anomalies).

Isoimmunization

Fetal red blood cells contain the D antigen on their surface and are capable of immunizing the Rh-negative mother after a fetomaternal transfusion in the midtrimester. This event can occur spontaneously during pregnancy or after an amniocentesis. The World Health Organization (WHO) and the ACOG have recommended the administration of the anti-D immunoglobulin G (IgG) to women after midtrimester amniocentesis. There is no agreement on the dose; the WHO recommends 50 μg, whereas the ACOG recommends 300 μg.²⁶⁰ The basis for this recommendation is that midtrimester amniocentesis has been associated with an increased incidence of transplacental hemorrhage, a risk factor for isoimmunization, although the precise risk of isoimmunization after midtrimester amniocentesis has not been well defined. The incidence of Rh isoimmunization in the randomized controlled clinical trial reported by Tabor et al¹⁸⁴ was 0.3% (7 of 370) in the study group and 0.1% (3 of 347) in the control group (anti-D IgG was not administered to Rh-negative patients undergoing amniocentesis). Although this difference is not significant, the number of patients required to detect a difference of 1% between the amniocentesis and the controls group would be 2896 in each group.¹⁸⁴ The analysis of the previous reports suggests that midtrimester amniocentesis is associated with an increased risk of isoimmunization, and the magnitude of the increased risk seems to be approximately 1%.²⁰⁷

Murray et al²⁶¹ provided a comprehensive analysis of the advantages and disadvantages of anti-D IgG administration. The objections that have been raised against the routine use of anti-D IgG are unproven efficacy (isolated case reports have indicated that sensitization can occur after anti-D IgG administration) and unproven long-term safety. Anti-D IgG crosses the placenta and coats Rh-positive fetal red cells. It is unclear if this could have adverse effects. The theoretic risk of augmentation has been suggested. This phenomenon consists of an enhancement of the immune response in the context of small amounts of antibody. Furthermore, the long-term effects of exposing the immunologically “naïve” fetal immune system to human immunoglobulins are unknown. Although there is no irrefutable evidence to support the routine administration of anti-D IgG after midtrimester amniocentesis, it has become the standard practice in the United States.

EARLY AMNIOCENTESIS

Traditionally, genetic amniocentesis has been carried out after the 16th week of gestation. To obtain results earlier, several centers have introduced the performance of “early amniocentesis” at a gestational age lower than 15 weeks. Early amniocentesis is an attractive approach because it provides early cytogenetic diagnosis and amniotic fluid for AFP determinations for the assessment of neural tube defects.^157,161,169 These investigators demonstrated that amniotic fluid AFP obtained between 13 and 15 weeks had a 100% sensitivity for the diagnosis of anencephaly and a 96% sensitivity for the diagnosis of spina bifida when using cutoff values greater than 2.0 and 2.5 multiples of the normal median, respectively.^167-169

Although early amniocentesis has been performed as early as the seventh week of gestation,¹⁶⁹ it is usually performed between 11 and 15 weeks of gestation. The technique for early amniocentesis is basically the same as for midtrimester amniocentesis, with minor differences (1) most investigators have recommended the use of a smaller bore (22-gauge) spinal needle^{152-165,168,169,172}; (2) the fluid should be aspirated slowly to prevent collapse of the amniotic sac¹⁶³; and (3) because the amniotic membrane may not be completely fused with the chorion at this state in pregnancy, membrane tenting is a more frequent complication than with midtrimester amniocentesis.

Transvaginal aspiration of amniotic fluid has been proposed as an alternative to the transabdominal approach for early amniocentesis. Potential advantages include the high resolution of the transvaginal probe and easy access to the amniotic sac. Jorgensen et al¹⁶⁹ attempted the procedure in 36 women between 7 and 12 weeks of gestation. Although amniotic fluid was obtained in all cases and the patients tolerated the procedure well, culture was unsuccessful in six cases (16.7%) because of bacterial or fungal overgrowth. In contrast, culture success was reported in all 96 control samples obtained by the transabdominal route. Shalev et al²⁶³ compared the clinical and laboratory results of first trimester transvaginal amniocentesis with those of transcervical CVS and midtrimester amniocentesis. Transvaginal amniocentesis was performed in 355 women between 10 and 12 weeks of gestation using a 20-cm-long 22-gauge needle. The volume of amniotic fluid retrieved was 1 mL per week of gestation. Three hundred fifty-six consecutive transcervical CVS and 356 consecutive midtrimester transabdominal amniocenteses were matched for maternal age and indication for the procedure and selected as controls. Amniotic fluid was successfully retrieved in 99.7% (355 of 356) of the first trimester amniocenteses and 100% (356 of 356) of the midtrimester amniocenteses. In comparison, CVS was successful in only 97.8% (346 of 356) of the cases (P < .05 vs first trimester and midtrimester amniocenteses, respectively). No significant differences in culture success rates were observed between patients undergoing early transvaginal amniocentesis (97.9% [344 of 355]), midtrimester transabdominal amniocentesis (97.2% [348 of 356]), and CVS (96.5% [335 of 346]). The spontaneous pregnancy loss, however, was significantly higher in patients undergoing early transvaginal amniocentesis than in patients who had midtrimester transabdominal amniocentesis (3.2% [11 of 345] vs 0.9% [3 of 350], P < .05). There were no differences in total pregnancy loss between either group or patients who had CVS (2.9% [10 of 344]).²⁶³

The major concern with early amniocentesis is the increased risk of spontaneous fetal loss in comparison with either CVS or conventional amniocentesis. The Canadian Early and Midtrimester Amniocentesis Trial (CEMAT) Group²⁶⁴ conducted a randomized trial in 4374 patients to assess safety and efficacy of early and midtrimester amniocentesis. The mean gestational age and amniotic fluid volume retrieved at early amniocentesis were 12.1 weeks and 11 cc, respectively, while in patients who underwent midtrimester amniocentesis they were 15.4 weeks and 20 cc, respectively. After excluding intrauterine and neonatal death, the group reported a postprocedure spontaneous loss rate of 2.6% and 0.8% for early and midtrimester amniocentesis, respectively. When only patients who received their procedure within the allocated window were considered (<13 weeks and >15 weeks), the postprocedure spontaneous loss rate for early and midtrimester amniocentesis was 2.7% and 0.5%, respectively.¹⁸² The procedure-related variables that were found to be associated with pregnancy losses after early amniocentesis were: (1) difficult procedures (5.7% vs no difficult procedure: 2.4%, P = .001); (2) amniotic fluid leakage (11.7% vs no amniotic fluid leakage: 2.4%, P = .001); and (3) vaginal bleeding (10.5% vs no bleeding: 2.1%, P = .001). Interestingly, although a trend toward lower rates of pregnancy loss among the most experienced operators and centers was observed, the differences were not statistically significant. This suggests that operator and center experience were not associated with a higher rate of pregnancy loss.

Early amniocentesis was associated with a higher rate of multiple needle insertions (5.4% vs 2.1%, P < .0001) and amniotic fluid leakage before 22 weeks of gestation (3.5% vs 1.7%, P = .0007) than that of midtrimester amniocentesis. The incidence of talipes equinovarus was higher in the early amniocentesis group than in the midtrimester group (1.6% vs 0.1%, P < .001).¹⁸² A possible explanation for talipes is the removal of a relatively large amount of amniotic fluid (11 mL, about 20% of the total amount of amniotic fluid) during early amniocentesis in the CEMAT study. Cytogenetic results from the CEMAT study (including culture success, detection of chromosome abnormalities, frequency of mosaicism, and maternal cell contamination) showed no difference between early and midtrimester amniocentesis. A repeat amniocentesis was required more frequently after early amniocentesis than after midtrimester amniocentesis.²⁵⁸

Table 27-7 displays the overall spontaneous pregnancy loss of early amniocentesis reported in published studies including at least 100 cases.^{129,153,157,158,162,163,165-169,267-275,277-283} The fetal loss rate is related to the gestational age at which the procedure is performed. Table 27-8 shows the data of four different studies in which the fetal loss rate by gestational age at amniocentesis was reported. The overall fetal loss rate was significantly higher when the amniocentesis was performed before 12 weeks of gestation than after this gestational age (8.1% [5 of 62] vs 1.3% [38 of 2950], P < .001).^{153,157,163,167,168} The combined experience of 7 centers for which data are available (Table 27-9) shows an overall loss rate of 1.3% (44 of 3296) within 14 days of the procedure, with a significant increase when amniocentesis is performed before 12 weeks (6.9% [5 of 72] vs 1.2% [39 of 3224], P < .001).^{153,157,160,163,165,167,168,171}