Introduction

Prenatal genetic diagnosis has evolved over 45 years since its introduction in the late 1960s. Initially, the invasive procedure amniocentesis allowed diagnosis of major chromosomal abnormalities by amniotic fluid cell analysis. Amniocentesis was offered only to women of advanced (35 years) maternal age at delivery. Later, other invasive procedures (e.g. chorionic villus sampling [CVS]) allowed diagnosis earlier in pregnancy. Concurrently, non-invasive methods using maternal serum analytes were developed, permitting younger women at increased risk for aneuploidy to be identified and offered an invasive procedure. Multiple non-invasive protocols are now offered, and others are still being developed (e.g. cell-free fetal DNA). These protocols are increasingly used by providers and patients alike, the net effect being fewer invasive procedures are now carried out. It is, therefore, proper to wonder whether invasive procedures (e.g. amniocentesis and CVS) are becoming obsolete except in the context of follow up before non-invasive screening.

In this chapter, we first consider the procedure-related risks of amniocentesis and CVS, both often overestimated. We then consider detection rates predicted by non-invasive first- and second-trimester maternal serum analyte and fetal ultrasound screening, neither yet as high as that with universal invasive procedures. Finally, we discuss potential advantages of an expanded diagnostic armamentarium: array comparative genome hybridisation (CGH), which will require chorionic villi and amniotic fluid cells and, hence, an invasive procedure.

Safety of amniocentesis

For decades, the risk of pregnancy loss after amniocentesis at 15–22 weeks was stated to be 1 in 200, the citation being a 1976 National Institute of Child Health and Human Development (NICHD)collaborative study. This study showed a 0.5% arithmetic increase in miscarriage in women undergoing amniocentesis ( n = 1040) compared with a control group ( n = 992). No statistically significant difference, however, was reported. Although never proved to be statistically different, the familiar one in 200 procedure-related risk remained. Irrespective of this, risk of pregnancy loss has clearly decreased since the initial introduction of amniocentesis. For example, in the early collaborative studies cited, ultrasonography was not available and, by today’s standards, not for another 2 decades. At present, the risk of procedure-related losses after amniocentesis in singleton pregnancies is considered in experienced hands to be about 1 in 400 or less. Risks logically should be increased further when amniocentesis is carried out in twins. Even this, however, is difficult to prove.

Maternal risks are low, with symptomatic amnionitis occurring only rarely (0.1%). Minor maternal complications, such as transient vaginal spotting or minimal amniotic fluid leakage, occur in 1% or less of cases, but these complications are almost always self-limited in nature. Rare complications include intra-abdominal viscus injury or haemorrhage. The most serious is fulminant sepsis ( Escherichia coli or Clostridia ), resulting in maternal mortality in extraordinarily rare cases.

In contrast to relative safety of amniocentesis at 15–22 weeks, the American College of Obstetricians and Gynecologists (ACOG) stated unequivocally that early amniocentesis (before 13–14 weeks’ gestation) should not be carried out for genetic indications. The basis for this guideline is large studies showing untoward results, including higher rates of pregnancy loss, talipes equinovarus , and amniotic fluid leakage.

Safety of amniocentesis

For decades, the risk of pregnancy loss after amniocentesis at 15–22 weeks was stated to be 1 in 200, the citation being a 1976 National Institute of Child Health and Human Development (NICHD)collaborative study. This study showed a 0.5% arithmetic increase in miscarriage in women undergoing amniocentesis ( n = 1040) compared with a control group ( n = 992). No statistically significant difference, however, was reported. Although never proved to be statistically different, the familiar one in 200 procedure-related risk remained. Irrespective of this, risk of pregnancy loss has clearly decreased since the initial introduction of amniocentesis. For example, in the early collaborative studies cited, ultrasonography was not available and, by today’s standards, not for another 2 decades. At present, the risk of procedure-related losses after amniocentesis in singleton pregnancies is considered in experienced hands to be about 1 in 400 or less. Risks logically should be increased further when amniocentesis is carried out in twins. Even this, however, is difficult to prove.

Maternal risks are low, with symptomatic amnionitis occurring only rarely (0.1%). Minor maternal complications, such as transient vaginal spotting or minimal amniotic fluid leakage, occur in 1% or less of cases, but these complications are almost always self-limited in nature. Rare complications include intra-abdominal viscus injury or haemorrhage. The most serious is fulminant sepsis ( Escherichia coli or Clostridia ), resulting in maternal mortality in extraordinarily rare cases.

In contrast to relative safety of amniocentesis at 15–22 weeks, the American College of Obstetricians and Gynecologists (ACOG) stated unequivocally that early amniocentesis (before 13–14 weeks’ gestation) should not be carried out for genetic indications. The basis for this guideline is large studies showing untoward results, including higher rates of pregnancy loss, talipes equinovarus , and amniotic fluid leakage.

Safety of chorionic villus sampling

Chorionic villus sampling allows diagnosis in the first trimester of pregnancy. Thus, if desired, pregnancy termination can be carried out at an earlier stage in gestation, when it is much safer for the mother. The maternal death rate is 1 per 100,000 early in pregnancy compared with seven to 10 per 100,000 in mid-pregnancy. Early diagnosis also makes it safer to carry out selective fetal reduction in multiple gestations, a relatively facile procedure in the first trimester but less so in the second trimester. Early termination also protects patient privacy. Chorionic villi analysis and amniotic fluid cell analysis offer the same information concerning chromosomal status, enzyme levels, and gene mutations. The exception is that CVS is not useful for those few assays requiring amniotic fluid liquor, namely alpha-feto protein for diagnosis of neural-tube defects.

Given the decided advantage of diagnosis earlier in gestation, more widespread use of CVS would have been expected. Chorionic villus sampling, however, is not so easily mastered as amniocentesis and, in some centres, CVS has proven difficult to implement. Concern has been raised in many of those circles that undue risks exist in CVS compared with amniocentesis. This is actually not true in experienced hands.

In 1989, the US Collaborative Clinical Comparison of Chorionic Villus Sampling and Amniocentesis study reported that the pregnancy losses rate after CVS was no different than rates after second-trimester amniocentesis. Randomised-controlled trials also found no difference between trans-cervical CVS and trans-abdominal CVS. In the second phase of the NICHD collaborative CVS study, reported in 1992, 1194 women were randomised to trans-cervical CVS and 1929 to trans-abdominal CVS. Loss rates in cytogenetically normal pregnancies up to 28 weeks were 2.5% and 2.3%, respectively. The overall loss rate (i.e. background plus procedure-related) during the randomised trial was 0.8% lower than rates observed during the mid-1980s. This probably reflects increasing operator experience as well as availability of both transcervical and transabdominal approaches. Chorionic villus sampling is operator-dependent, apparently more so than mid-trimester amniocentesis.

A 2003 Cochrane review assessing comparative safety among trans-abdominal CVS, trans-cervical CVS, early amniocentesis and second-trimester amniocentesis concluded that both second-trimester amniocentesis and trans-abdominal CVS are both safer than trans-cervical CVS and early amniocentesis. In the US collaborative study, the investigators were highly experienced, and they suggest that, in some hands, trans-cervical CVS is more difficult to master than trans-abdominal CVS.

Atypical results were found in a single, early, multi-centre, randomised study in the UK, which compared second-trimester amniocentesis and CVS. The primary outcome was completed pregnancies, with 4.4% fewer completed pregnancies in the CVS cohort, reflecting unintended and intended pregnancy terminations. Operators in this study, however, were unavoidably less experienced than those in the US trial, because the sole requirement for participation in the UK study was 30 ‘practice’ CVS procedures. Greater experience is necessary to acquire expertise in CVS. Another important factor in the UK study was that the indication for CVS was an increased nuchal translucency measurement, a cystic hygroma, or any other anomaly; thus, the risk of miscarriage was increased.

Controversy has arisen concerning the risk of limb-reduction defects after CVS. Limb-reduction deformities (LRD) and oromandibular-limb hypogenesis after transcervical CVS were reported in the UK, US, and Taiwan. The US Centers for Disease Control and Prevention conducted a case-control study of 131 infants with non-syndromic limb deficiency, identified in seven population-based birth defects surveillance programmes. The study compared 131 controls having other birth defects, matched to cases by infant year of birth, mother’s age, race and state of residence. The odds ratio for all limb deficiencies after CVS during 8–12 weeks’ gestation was 1.7, not significantly increased given the 95% confidence interval 0.4 to 6.3. A significant association, however, was found in subgroup analysis for one type of LRD: transverse digital deficiency (OR, 6.4; 95% CI 1.1 to 38.6).

The upper limit of increased risk for transverse LRD has sometimes been stated to be 1 in 3000, but it is arguable whether risk is greater than background unless carried out before 9 weeks. When CVS is carried out before 9 completed weeks, the risk for limb-reduction defects may be increased, but, at this stage of gestation, CVS is now carried out only in exceptional cases (e.g. Orthodox Jewish communities where termination is not possible after 40 days embryogenesis [54 days gestation]). The consensus is that LRDis not a major concern when CVS is carried out by experienced individuals at 10–13 weeks’ of gestation.

In experienced hands, CVS is safe in multiple gestations. In the US Collaborative NICHD study, the total loss rate of chromosomally normal fetuses (e.g. spontaneous abortions, stillbirths, neonatal deaths) was 5%, only slightly higher than the 4% absolute rate observed in singleton pregnancies.

Chromosomal abnormalities detectable using aminocentesis or chorionic villus sampling

With invasive tests such as amniocentesis or CVS, every chromosomal disorder that can be identified by karyotype after birth (postnatal) can also be detectable in utero (prenatal). Resolution by a 600 band karyotype typically identifies a 5 Mb (5 million base pairs) aberration. Thus, any pregnant woman could, if desired, undergo an invasive procedure to exclude aneuploidy, or technically deletions or duplications 5 Mb or greater (resolutions of 500–600 band karyotype). Indications that historically justified prenatal cytogenetic diagnosis included (1) advanced maternal age, resulting in increased risk for autosomal trisomy; (2) parental chromosome rearrangements (e.g. balanced translocation); (3) previous pregnancy with autosomal trisomy; (4) abnormal fetal ultrasound findings on a fetus during the current pregnancy: fetal structural anomalies, intrauterine growth retardation, decreased or increased amniotic fluid volume abnormalities; and (5) increased risk calculated from non-invasive screening results (nuchal translucency and maternal serum analytes).

Prenatal cytogenetic studies (screening or invasive procedures) are most commonly pursued because of advanced maternal age, historically 35 years at delivery. The incidence of trisomy 21 is 1 in 800 live births in the USA: the risk of other aneuploidy is about twice that. Trisomy 13, trisomy 18, 47,XXX, and 47,XXY also increase with advanced age ( Table 1 ).

Risk figures shown in Table 1 are applicable only for liveborn infants. Prevalence of chromosomal abnormalities when amniocentesis is carried out is higher than that at birth by about 30%. Thus, in non-invasive screening protocols, the likelihood of Down’s syndrome is usually considered to be increased at 1 out of 270 (screen positivity) because that is the mid-trimester risk for a 35-year-old woman. Compared with amniocentesis, prevalence of aneuploidy at CVS is a further increased by 50%. That the frequency of chromosomal abnormalities is lower in liveborn infants than in first- or second-trimester fetuses reflects the disproportionate likelihood that fetuses with chromosomal abnormalities will be lost spontaneously. That is, some aneuploid fetuses would have died spontaneously in utero had iatrogenic intervention not occurred in the form of prenatal diagnosis and pregnancy termination. Of relevance, 8–13% of stillborn infants show chromosomal abnormalities, more if the stillborn has overt malformations.

The ACOG last updated their recommendations on prenatal cytogenetic screening in January 2007. Previous standards recommend offering invasive chromosomal diagnosis to all women who, at their expected delivery date, will be 35 years or older and have singleton pregnancies. During those years, a procedure need not necessarily be offered to younger women. The choice of age 35 years was, however, always arbitrary, and was selected when risk figures were available only in 5-year intervals (i.e. 30–34 years, 35–39 years, 40–44 years), as discussed by Simpson. In twin gestations, it was subsequently recommended that an invasive procedure be offered at age 32 or 33 years. The rationale was that, given most twins being dizygotic, the likelihood that either one of the two would be aneuploid is the sum of the individual risks. At age 33 years, the sum of risks for either liveborn being trisomic was one in 347, comparable to the singleton risk at age 35 years (one in 375). The same reasoning applies to all chromosomal aneuploidies: one in 176 compared with one in 202.

Current (2007) ACOG guidelines state unequivocally that neither age 35 years nor any specific age should be used as a threshold for invasive or non-invasive screening: ‘All women, regardless of age, should have the option of invasive testing’ (ACOG). The guidelines specifically elaborate that ‘patients informed of the risks, especially those at increased risk of having an aneuploid fetus, may elect to have diagnostic testing without first having screening.’ Younger women may elect an invasive procedure because they wish to achieve the near 100% detection, possible only with an invasive procedure; detection by an invasive procedure exceeds by 10–15% that of any non-invasive screening protocol.

Detection of microdeletions and microduplications using comparative genomic hybridisation in chromosomal microarrays

Compared with karyotypes, greater ability to detect aneuploidies and smaller deletions or duplications is achieved using chromosomal microarrays. The CGH protocol allows comprehensive analysis of the entire genome at greater resolution than banded karyotypes. The principle is based on ability of single-stranded DNA to anneal (hybridise) obligatorily with its complementary single standard DNA. Hybridisation applies whether DNA is derived from the same individual or from different individuals. Normal (control) DNA can be labelled with a fluorochome of one color (e.g. green); test (patient) DNA is then labelled with a fluorochome of a different color (e.g. red). If both test and control DNA are denatured (single-stranded), test DNA hybridises to the control DNA. If equal amounts of control and test DNA are present, one would expect the colour of hybridised mixture to be yellow. If the test DNA were present in excess (e.g. trisomy), the relevant chromosome or chromosomal region would show more of the colour used to connote test (patient) DNA (e.g. red in the above example). For duplications or deficiencies of chromosomal regions, the difference would be evident only in a portion of the chromosome. For trisomy, the colour difference would apply to the entire chromosome.

Conventional karyotypes can detect an alteration of 5–10 million base pairs (5Mb). Chromosomal microarrays detect alterations as small as 200 kilobases (200,000) in size or, if desired, even 20–50 bases. The size of the copy number variant (CNV) capable of being detected is determined by vendor platform, but detection encompasses the entire genome. A caveat is that the smaller the CNV detected, the less likely a pathogenic effect; however, even small CNVs can be significant if involving a deleted gene (e.g. a deleted dominant gene). On the other hand, at the 50–200 base pair level, CNVs (duplications or deletions) are often polymorphisms, inherited from a clinically normal parent and without clinical significance. Increasing number of CNVs are known to have pathogenic consequences and can be detected by array CGH but not by traditional karyotype. In aggregate, these occur in perhaps 1–2% of fetuses of women over age 35 years or women having a fetus with an ultrasonographic detected anomaly. Even greater clinical value is evident in evaluating children with developmental delay, 5–7% of whom have CNV variants. Array CGH is now considered a standard part of postnatal evaluation in that group.

The recently reported NICHD prenatal cytogenetic array study involved 4401 women having these indications: 46% maternal age; 26% fetal ultrasound anomaly; 18% increased risk by maternal serum analyte screening; and 9% others. All 316 autosomal trisomies and all 57 sex chromosome aneuploidies detected by karyotypes were also detected by array CGH. Microarrays were also designed to interrogate 84 chromosomal regions that were too small to be detected by conventional karyotypes, as well as 43 centromeric and 41 telomeric regions. Those microdeletions or microduplications predicted in advance to be detected (e.g. Di George Syndrome) were in fact. In this sample, 5.8% of fetuses with a normal karyotype and fetal anomalies had a microdeletion or microduplication, as did 1.7% of those where indication was maternal age or increased risk in serum analyte screening. Clinical difficulty arose when an additional 156 (4.1%) CNVs of uncertain clinical significance were detected. A Clinical Advisory Committee independent of laboratory directors considered 62 to be benign based on extant data and potential status, thus not warranting disclosure to parents. Of the remaining 94, 61 were considered to be potentially significant and, for a variety of reasons, required disclosure to parents. Disclosure was sometimes considered warranted because pregnancy termination might be an option, sometimes because follow-up ultrasound seemed of value and not likely to be pursued without disclosure, and sometimes on the basis of relevance to a family history of structural abnormalities. The manner in which disclosure and counselling is conveyed is crucial.

A special advantage of array CGH is that it can be carried out on uncultured DNA, providing information on circumstances (e.g. stillbirths) in which the cell cultures necessary to obtain a karyotype are not successful. Array CGH have a few disadvantages compared with the traditional karyotype. Chromosonal microarray analysis cannot distinguish a balanced translocation from a genetically normal, non-rearranged, chromosomal complement. Low-level mosaicism is difficult to detect and, in some platforms, triploidy cannot be excluded.