Counsel for and perform invasive prenatal diagnosis — amniocentesis, chorionic villus sampling and fetal blood sampling

In one line

Invasive prenatal diagnosis is the deliberate sampling of feto-placental tissue — amniocentesis from 15+0 weeks, chorionic villus sampling from 11+0 to 13+6 weeks, fetal blood sampling later — to obtain a diagnostic genetic or haematological answer that screening cannot give; the consultant skill is no longer the needle but the counselling: who genuinely needs it now that the procedure-related miscarriage risk is roughly 0.1–0.3% rather than the 1% historically quoted, and what the right laboratory test is for the question being asked.

Mechanism & pathophysiology

Each technique samples a different tissue, and what it samples determines both the genetic answer it can give and its characteristic failure mode.

Amniocentesis aspirates amniotic fluid, which contains amniocytes — desquamated fetal cells from skin, the respiratory and urinary tracts, and the amnion. These are genuinely fetal in origin, so amniocentesis is the reference standard against which the others are judged for accuracy: a result reflects the fetal karyotype directly. The trade-off is timing. Before about 15 weeks the amnion has not fully fused with the chorion and the cell yield is poor, so the fluid is sparse and the membranes tent away from the needle — the reason early amniocentesis was abandoned (it raised both loss and talipes, the lesson behind the 15-week floor).

Chorionic villus sampling aspirates trophoblast and mesenchymal core from the developing placenta (chorion frondosum). Its great advantage is gestation: a diagnosis in the first trimester, when termination is simpler, safer and more private. Its characteristic pitfall is biological — the placenta is not always genetically identical to the fetus. The early embryo segregates into trophoblast and inner cell mass within the first few cell divisions, and a post-zygotic mitotic error can produce an aneuploid cell line confined to the placenta while the fetus is normal (or, less often, the reverse). This is confined placental mosaicism, found in roughly 1–2% of pregnancies sampled by CVS, and it is the dominant explanation when CVS mosaicism is later not confirmed: about 86% of mosaic CVS results are confined to the placenta rather than true fetal mosaicism. Direct (cytotrophoblast) preparations reflect this layer most; the cultured mesenchymal core correlates better with the fetus, which is why a discrepant or mosaic CVS result is resolved by a follow-up amniocentesis, not by repeating the CVS.

Fetal blood sampling (cordocentesis, usually at the placental cord insertion) draws fetal blood directly. It bypasses the placental-mosaicism problem and, uniquely, gives real-time fetal haematology, blood gas, platelet count and the ability to treat in the same sitting — the needle that diagnoses fetal anaemia is the needle that transfuses it.

The laboratory method matters as much as the sample, and the two have decoupled over the last decade.

• QF-PCR (quantitative fluorescent PCR) reads short-tandem-repeat dosage on chromosomes 21, 18, 13, X and Y, giving a rapid aneuploidy result in under 48 hours with near-complete concordance with karyotype for those chromosomes. It is the standard first-line rapid test on amniotic fluid or villi where the question is "is this one of the common trisomies?" — it does not see other or structural abnormalities.
• Full karyotype (cultured cells) remains the only method that detects balanced rearrangements (reciprocal translocations, inversions) and triploidy, which is exactly why it cannot be discarded.
• Chromosomal microarray scans the whole genome for copy-number gains and losses at far higher resolution than karyotype, on uncultured cells, and is the test of choice when there is a fetal structural anomaly — but it is blind to balanced rearrangements and triploidy (no net copy-number change) and will turn up variants of uncertain significance that demand careful counselling.
• FISH gives an interphase aneuploidy answer on uncultured cells in hours, now largely supplanted by QF-PCR for cost.
• Beyond the chromosome, single-gene testing (targeted mutation analysis, exome sequencing) is used when the indication is a Mendelian disorder in the family.

The practical synthesis: choose the sample for the gestation and the indication, then choose the laboratory test for the question — QF-PCR for a high-risk aneuploidy screen result, microarray when an anomaly is on the scan, karyotype when a balanced translocation must not be missed.

Two further mechanistic points separate the consultant from the registrar. First, fetal blood sampling can also diagnose fetal infection and immune status directly — a fetal full blood count, reticulocytes and direct Coombs in alloimmunisation, or fetal-blood PCR/IgM for an intrauterine infection (parvovirus B19, cytomegalovirus) when amniotic-fluid PCR is equivocal — because it samples the fetal compartment itself rather than a surrogate. Second, the same uterine breach that lets the operator into the fetal circulation lets fetal red cells out into the maternal circulation: a feto-maternal haemorrhage is the unifying mechanism behind both the rhesus-sensitisation risk (why anti-D is mandatory) and the blood-borne-virus inoculation risk (why maternal viraemia must be controlled first). The needle is bidirectional, and almost every complication of invasive testing — loss, sensitisation, vertical transmission — flows from that single fact.

Assessment

The decision to put a needle into the uterus is a counselling decision first. Establish why a diagnostic test is being considered, because the indication sets both the urgency and the right tissue and assay.

• Define the indication precisely. The common drivers are: a high-risk combined or quadruple screen; a high-chance cfDNA/NIPT result; a fetal structural anomaly or soft-marker constellation on ultrasound; a parental balanced translocation or a known single-gene disorder with a defined recurrence risk; and, for fetal blood sampling specifically, suspected fetal anaemia (red-cell alloimmunisation, parvovirus B19), thrombocytopenia, or fetal infection requiring direct sampling.
• Distinguish screening from diagnosis explicitly in the counselling. A high-chance NIPT result is not a diagnosis. The NEXT trial put a number on why this matters: in a routine first-trimester population the positive predictive value of cfDNA for trisomy 21 was 80.9% — excellent for a screen, but it still means roughly one in five "high-chance" results is a false positive, and the PPV falls further for the rarer trisomies and in younger, lower-prior-risk women. A woman must understand that an invasive test is what confirms before any irreversible decision is taken.
• Pre-procedure work-up. Confirm viability, gestation and number on ultrasound; document placental site and the safest needle trajectory; check maternal rhesus group and antibody screen; review blood-borne virus status — HIV, hepatitis B, hepatitis C — because these change whether and when you proceed. In a multiple pregnancy, map and label each sac before sampling (chorionicity, membrane geometry, a reliable way to tell the twins apart) — a mislabelled result in twins is a catastrophic error.
• Counsel the actual numbers. Quote the procedure-related miscarriage risk honestly: on the best contemporary meta-analytic data it is around 0.1–0.3% above background for both amniocentesis and CVS in skilled hands — an order of magnitude lower than the "1% / 2%" still in old consent forms. Counsel the test's limitations (what QF-PCR will and will not see; the chance of a mosaic or uncertain result), the small failure/repeat rate, and the turnaround time.
• Interpretation is part of assessment. A normal QF-PCR does not exclude a microdeletion; a "low-level mosaic" on CVS usually means confined placental mosaicism rather than an affected fetus and is resolved by amniocentesis; a variant of uncertain significance on microarray is a counselling problem, not a result to be acted on alone.

The judgement layer — who actually needs a needle now

Screening has changed the question from "which invasive test?" to "does this woman need an invasive test at all?", and the consultant call turns on a few distinctions.

• A high-chance NIPT for a common trisomy in an otherwise well pregnancy is best confirmed by the least morbid diagnostic route. Because the rapid QF-PCR answer for trisomy 21/18/13 is all that is clinically needed, amniocentesis after 15 weeks (or CVS in the first trimester if the timeline favours an earlier decision) with QF-PCR is appropriate, reserving a full microarray for when the scan is abnormal.
• A fetal structural anomaly on ultrasound changes the assay, not just the indication: this fetus needs microarray, because karyotype alone will miss the ~6% of clinically significant copy-number changes that an anomaly predicts. The sample is chosen by gestation — CVS if the anomaly declares early, amniocentesis later.
• A known parental balanced translocation or a defined single-gene disorder is a targeted test: send the specific karyotype/FISH for the translocation, or the family's known mutation, and do not let microarray's blindness to balanced rearrangements give false reassurance.
• Fetal anaemia, thrombocytopenia or an infection question the indirect tests cannot settle is the residual indication for fetal blood sampling — and its loss rate keeps it a last resort, used when middle-cerebral-artery Doppler, amniotic-fluid PCR and the rest have taken the question as far as they can.

The corollary is that a confidently low-chance NIPT in a structurally normal fetus is usually the end of the road, not a prelude to amniocentesis — over-testing the reassured woman exposes her to a small but real procedure risk for no diagnostic gain.

Management

Structure the procedure as immediate (the technique and timing) → ongoing (the laboratory pathway and the index pregnancy) → long-term (counselling and recurrence).

Immediate — technique and timing

• Amniocentesis is performed transabdominally under continuous ultrasound guidance with a 20–22G needle, aiming for a clear pool of fluid clear of the fetus and, ideally, the placenta; the first 1–2 mL (risk of maternal-cell contamination) is discarded and ~15–20 mL aspirated. The single most important rule is the 15+0-week floor — earlier amniocentesis increases both fetal loss and talipes equinovarus and is not done.
• CVS is performed transabdominally or transcervically, by needle aspiration or biopsy forceps, again under continuous ultrasound, sampling the chorion frondosum. The route is chosen by placental position and operator experience; outcomes are comparable in trained hands. The hard floor is 10+0 weeks — CVS before 10 weeks is associated with transverse limb-reduction (and oromandibular) defects, a vascular-disruption phenomenon whose frequency and severity rise the earlier the procedure (the CDC put the limb-deficiency risk at roughly 1/3,000–1/1,000 overall, concentrated in very early sampling). Performing CVS at 11–13 weeks, never before 10, removes essentially all of that excess.
• Fetal blood sampling is a tertiary-unit procedure: a 20–22G needle to the umbilical vein at the cord insertion under ultrasound guidance, confirming a fetal (not maternal) sample (e.g. by red-cell indices or a Kleihauer/i-stat check). It carries the highest procedure-related loss of the three — historically ~1–2%, and higher (up to ~4%) before 20 weeks — so it is reserved for questions the safer tests cannot answer, principally fetal anaemia, where the same needle delivers an intrauterine transfusion (covered in rhesus-alloimmunisation).

Ongoing — the laboratory pathway and protecting the index pregnancy

• Match the assay to the question. Send QF-PCR for a rapid common-trisomy answer; add microarray when there is a structural anomaly (it finds a clinically significant copy-number change in about 6% of anomalous fetuses with a normal karyotype, and ~1.7% with a normal scan); retain or add karyotype when a parental balanced translocation must be excluded or triploidy is possible, because microarray cannot see either. Send targeted single-gene testing when the indication is a known Mendelian disorder.
• Resolve mosaic and discordant results correctly. A mosaic or unexpected CVS result is most often confined placental mosaicism; confirm with amniocentesis before any decision. Do not act on a single mosaic CVS.
• Anti-D for rhesus-negative, non-sensitised women. Every invasive procedure is a potential sensitising event, so a non-sensitised Rh-negative woman should receive anti-D immunoglobulin after amniocentesis, CVS or fetal blood sampling, ideally within 72 hours. In South Africa this is a genuine resource constraint, not a formality: anti-D supply is intermittent across the public sector and the SASOG guidance explicitly addresses administration during shortages — the consultant point is to plan anti-D before booking the procedure, not discover its absence afterwards. (If the woman is already alloimmunised, anti-D is pointless — check the antibody screen first.)
• The blood-borne-virus interface — the SA-defining issue. A needle traversing the placenta can inoculate maternal blood into the fetal compartment, so maternal viraemia is the dominant procedure-modifying factor here.
  • HIV: invasive testing in an HIV-positive woman raises a real vertical-transmission concern, but the determinant is viral load, not serostatus. With combination antiretroviral therapy and a suppressed/undetectable viral load, contemporary series report no vertical transmission after amniocentesis, so the management is to optimise ART and defer the procedure until the viral load is suppressed wherever the clinical timeline allows, rather than to refuse testing. This sits inside the SA prevention-of-vertical-transmission (PVT) framework — suppression first, then proceed, avoiding the placenta on the needle path where possible. An unbooked or unsuppressed woman who needs urgent invasive testing is a multidisciplinary decision, weighing the genetic indication against the transmission risk.
  • Hepatitis B/C: similar logic — review viral load and antigen status and individualise; high maternal viraemia raises transmission risk.
• After the procedure: brief rest, advise on the small risk of leaking, bleeding, cramping or infection and when to return, and arrange the result and its counselling.

The multiple pregnancy is a category of its own. Before any needle goes in, the operator must record chorionicity, the precise position of each sac and a durable way to identify the fetuses (the membrane, the cord insertions, fetal sex if discordant, a labelled diagram) — because a result attributed to the wrong twin is among the worst errors in fetal medicine, and discordant results are exactly what invasive testing in twins exists to find. The procedure-related loss is higher than in singletons (RCOG quotes around 1% for twins), each sac is sampled separately with care not to cross the dividing membrane, and in a monochorionic pair a single sample is usually taken because the twins share a genome — unless discordant growth or anomaly suggests otherwise. If a serious abnormality is confined to one twin, the counselling extends to selective termination, a tertiary fetal-medicine decision that depends entirely on the labelling being right.

Long-term — counselling and recurrence

The result is the start of the consultation, not its end. A confirmed aneuploidy or serious structural diagnosis triggers non-directive counselling about continuing the pregnancy with planned care versus termination, with referral to fetal-medicine, genetics and the relevant paediatric specialty; a normal diagnostic result reassures but does not promise a normal baby (it excludes the conditions tested, not all of them). A confirmed parental balanced translocation reframes future pregnancies and brings in recurrence-risk counselling and preimplantation genetic testing options. Document the karyotype/microarray/single-gene result, the counselling given, and the agreed plan.

A specific consent pitfall belongs here: the variant of uncertain significance. Microarray, by scanning the whole genome, will sometimes return a copy-number change of unknown clinical meaning — a finding that cannot be confidently called benign or pathogenic. Women must be warned of this possibility before the test, because an uncertain result in a wanted pregnancy generates real distress, and parental testing (to see whether the variant is inherited from an unaffected parent) is often needed to interpret it. The same applies, more sharply, to prenatal exome sequencing, increasingly offered when an anomalous fetus has a normal microarray: it raises the diagnostic yield but multiplies uncertain and incidental findings, and is a specialist, consented undertaking rather than a routine add-on.

Guidelines compared

The bodies agree on the hard floors (15 weeks for amniocentesis, 10 weeks for CVS) and the principle that microarray is the right test for an anomalous fetus. Where they diverge is how the loss risk is framed: ACOG's bulletin-era figures (≈1/769 and 1/455) and RCOG's "below 0.5%" both predate or sit conservatively against the Akolekar meta-analytic 0.1–0.3%, and the contemporary, defensible counselling number is the lower one — quoting "1%" today overstates the harm and may wrongly deter a woman from a test she needs. The change worth flagging is precisely this downward revision of the quoted risk over the last decade.

The evidence & the controversy

The single most consequential shift is the collapse of the quoted miscarriage risk, and getting it right is now a consent-and-medicolegal issue. The Akolekar 2015 meta-analysis — restricted to studies of more than 1,000 procedures to limit small-study bias — found a procedure-related loss of only 0.11% for amniocentesis and 0.22% for CVS above background, an order of magnitude below the figures still printed on many consent forms. The 2019 update went further: when the invasive-test and control groups were matched for the same underlying chromosomal-risk profile (the fairer comparison, because women having invasive tests are higher-risk to begin with), the procedure-attributable risk was 0.12% for amniocentesis and effectively zero (−0.11%) for CVS. Much of the historical "excess loss" was the background risk of a higher-risk population, not the needle. The defensible position is to counsel the low contemporary figure while being honest that it is operator- and volume-dependent — these numbers describe skilled, high-throughput units, not an occasional operator.

The second shift is that screening has eaten the indications. cfDNA/NIPT has cut the number of invasive tests sharply, because a low-chance NIPT now reassures the large population previously sent for amniocentesis on age or a borderline serum screen. But NIPT has not abolished invasive diagnosis, and the consultant error is to treat a high-chance NIPT as a diagnosis: the NEXT trial's PPV of ~81% for trisomy 21 means a fifth of high-chance results are false positives, the PPV is worse for the rarer findings, and NIPT does not see the structural and balanced abnormalities that a microarray or karyotype on a diagnostic sample will. The contemporary algorithm is therefore more selective but not zero — NIPT triages, invasive testing confirms and broadens. The live controversy is the expansion of NIPT to genome-wide microdeletion screening with low positive predictive values, which risks generating anxious women and unnecessary invasive procedures chasing findings of doubtful significance — a screening-overreach debate that bears directly on how invasive testing is requested.

Running alongside this is a live laboratory controversy: how far to let sequencing into prenatal diagnosis. Prenatal exome sequencing on an invasive sample raises the diagnostic yield in structurally anomalous fetuses with a normal microarray, but at the cost of variants of uncertain significance, incidental adult-onset findings and turnaround pressure against the termination window — so it is offered selectively, in a counselled fetal-medicine setting, not as a default. The parallel debate on the screening side — expanding cfDNA to genome-wide microdeletions with poor positive predictive values — feeds straight back into invasive testing: a low-PPV screen result that prompts an amniocentesis to chase a finding of doubtful significance is iatrogenic anxiety, and recognising it as such is part of using these tools well.

The third, more uncomfortable, current thread is access and equity. In a well-resourced unit the question is which test for which indication; in much of the SA public sector the prior question is whether the woman can reach a centre that performs CVS or fetal blood sampling at all, whether anti-D will be in stock, and whether a confirmed serious diagnosis can be acted on within the legal and logistical window for termination. The evidence base (loss rates, NIPT performance) is generated in high-volume Northern centres and must be applied to a service where a single tertiary fetal-medicine unit may serve a province. Naming that gap — and building the referral pathway around it — is part of the answer, not a footnote to it.

Landmark trials & key evidence

A worked piece of arithmetic for the consent conversation: at a procedure-related loss of 0.11% (Akolekar), the number of amniocenteses performed for one procedure-attributable miscarriage is roughly 1 / 0.0011 ≈ 900 — which is why counselling "1 in 100" (the old figure) materially misstates the harm and can wrongly tip a woman against a test she needs. The contemporary, evidence-based number is closer to 1 in 900 for a skilled high-volume operator, and the honest caveat is that the figure is operator- and volume-dependent.

Exam traps & red flags

• Quoting the obsolete loss rate. Counselling "1% for amniocentesis, 2% for CVS" is no longer defensible; the meta-analytic procedure-related risk is ~0.1–0.3%, and over-quoting the harm can wrongly deter a needed test.
• CVS before 10 weeks. Sampling before 10+0 weeks risks transverse limb-reduction and oromandibular defects (vascular disruption). The window is 11+0–13+6; never before 10.
• Amniocentesis before 15 weeks. Early amniocentesis raises loss and talipes and was abandoned — 15+0 is the floor.
• Treating a high-chance NIPT as a diagnosis. PPV is ~81% for trisomy 21 and lower for rarer findings; acting irreversibly on a screen without diagnostic confirmation is a serious error.
• Acting on a single mosaic CVS result. Most CVS mosaicism is confined placental mosaicism (~86% of mosaic results) with a normal fetus — confirm with amniocentesis before counselling, never with a repeat CVS.
• Ordering the wrong assay. Karyotype misses microdeletions/duplications (use microarray for an anomalous fetus); microarray misses balanced translocations and triploidy (keep karyotype where those matter); QF-PCR sees only chr 21/18/13/X/Y.
• Forgetting anti-D. Every invasive procedure can sensitise — give anti-D to a non-sensitised Rh-negative woman within 72 hours (and check she is not already alloimmunised first, where it is futile). In SA, plan the supply before booking.
• Ignoring the viral load in an HIV-positive woman. The risk is driven by viraemia, not serostatus — defer until ART-suppressed where possible rather than refusing the test; avoid the placenta on the needle path.
• Doing fetal blood sampling for a question the safer tests can answer. Cordocentesis carries the highest loss (~1–2%, more before 20 weeks) — reserve it for fetal anaemia/thrombocytopenia and direct sampling needs, not routine karyotyping.
• Mislabelling in twins. A diagnostic result attributed to the wrong twin is catastrophic — map and label the sacs before the needle.

Evidence anchors

• Tabor et al. — RCT of genetic amniocentesis in 4606 low-risk women (origin of the "1%" loss figure), Lancet 1986
• Akolekar et al. — procedure-related miscarriage risk of amniocentesis and CVS, meta-analysis, Ultrasound Obstet Gynecol 2015
• Salomon/Akolekar et al. — updated meta-analysis, Ultrasound Obstet Gynecol 2019
• Norton et al. — NEXT trial, cfDNA vs standard screening, N Engl J Med 2015
• Wapner et al. — chromosomal microarray vs karyotyping for prenatal diagnosis, N Engl J Med 2012
• CDC/MMWR — CVS and amniocentesis prenatal counselling recommendations, 1995
• Botto et al. — CVS, transverse digital deficiencies and gestational age, Am J Med Genet 1996
• RCOG Green-top Guideline No. 8 (2021) — Amniocentesis and Chorionic Villus Sampling: amniocentesis not before 15+0 weeks; CVS 11+0–13+6, not before 10+0; additional miscarriage risk below 0.5% in skilled hands; review blood-borne-virus status before sampling.
• ACOG Practice Bulletin No. 162 (2016) — Prenatal Diagnostic Testing for Genetic Disorders: loss rates ~1/769 (amniocentesis) and 1/455 (CVS); chromosomal microarray recommended when invasive testing is done for a structural anomaly.
• SASOG Rh Disease and Alloimmunisation guidance — anti-D administration after potentially sensitising events, including during stock shortages in the SA public sector.
• SA HIV Consolidated Guidelines (PVT framework) — optimise ART and viral-load suppression before, or in planning, invasive testing in HIV-positive women.