Counsel on and interpret prenatal screening for aneuploidy, including NIPT/cell-free DNA

In one line

Aneuploidy screening estimates a probability, never a diagnosis: every screen-positive woman is offered confirmatory invasive testing before any irreversible decision. The positive predictive value of even cell-free DNA rises and falls with the woman's prior risk, so the same "positive" result means very different things in a 40-year-old and a 22-year-old — which is why no single PPV figure can be quoted for it.

Mechanism & pathophysiology

Aneuploidy is a deviation from the euploid 46-chromosome complement, almost always arising from meiotic non-disjunction — failure of homologues (meiosis I) or sister chromatids (meiosis II) to separate, yielding a gamete with 24 or 22 chromosomes. The maternal oocyte, arrested in meiosis I from fetal life until ovulation decades later, accumulates cohesin degradation over time; this is the mechanistic basis of the steep, non-linear rise in trisomy with maternal age, and it is why age alone is the oldest and crudest screening filter.

The clinically relevant aneuploidies behave according to which chromosome carries the extra (or missing) material:

• Trisomy 21 (Down syndrome) — the commonest viable autosomal trisomy (~1 in 700 livebirths overall, far higher with advancing age). Chromosome 21 is gene-poor, which is why the trisomy is survivable; the screening markers it produces are the reference against which every test is calibrated.
• Trisomy 18 (Edwards) and trisomy 13 (Patau) — gene-richer chromosomes, hence overwhelmingly lethal in utero or in early infancy. Their biochemical signature differs from T21 (notably low PAPP-A and low free β-hCG in T18, distinct from the high β-hCG of T21), and they frequently carry major structural anomalies visible on ultrasound, so they are often suspected on the anomaly scan independent of the serum screen.
• Monosomy X (Turner syndrome, 45,X) — usually paternal sex-chromosome loss; most are lost early, the survivors typically presenting with cystic hygroma, hydrops or a raised nuchal translucency rather than a serum-marker pattern.
• Triploidy (69,XXX / 69,XXY) — a whole extra haploid set, digynic (maternal, small placenta, severe early growth restriction) or diandric (paternal, partial molar placenta with markedly raised β-hCG). It is non-viable, and recognising its biochemistry prevents misclassification.

The serum and sonographic markers exploited by screening are downstream consequences of abnormal placentation and fetal development. Nuchal translucency (NT) — the fluid collection at the fetal neck measured at 11–13⁺⁶ weeks — is increased in T21, T18, T13, Turner syndrome and in structurally normal fetuses with cardiac defects, reflecting transient lymphatic/cardiac dysfunction. PAPP-A (pregnancy-associated plasma protein-A) and free β-hCG are placental products: in T21 PAPP-A is low and free β-hCG high; in the trisomies 18 and 13 both tend to be low. The second-trimester quadruple markers — AFP (fetal-origin, low in T21), hCG (high in T21), unconjugated oestriol (uE3) (low in T21) and inhibin A (high in T21) — capture the same disordered feto-placental endocrinology later in gestation.

Cell-free DNA (cfDNA) is a different mechanism entirely, and its name is a half-truth that the counselling must respect. From around 4 weeks, fragments of DNA circulate in maternal plasma. The fraction that is "fetal" is in fact placental — it derives from apoptotic cytotrophoblast, not the fetus directly. That single fact explains every important limitation of the test:

• The proportion of the total cfDNA that is placental is the fetal fraction. Below roughly 4% the assay cannot reliably call a result and returns a "no-call"; fetal fraction falls with high maternal weight (dilution by maternal cfDNA), early gestation and certain aneuploidies (notably T18, T13 and triploidy, where the placenta is small), so a no-call is itself a soft risk signal, not merely a technical nuisance.
• Because the DNA is trophoblastic, a placenta that is genetically discordant from the fetus — confined placental mosaicism — produces a false-positive (or, rarely, false-negative) cfDNA result that the actual fetal karyotype does not share. This is the dominant biological reason cfDNA is a screen, not a diagnosis.
• Other sources of placental or maternal DNA confound the result: a vanishing twin continues to shed trophoblastic DNA for weeks; maternal copy-number variants, mosaicism or an occult malignancy contribute aneuploid-appearing maternal DNA and are a recognised, if rare, cause of an uninterpretable or falsely positive result that occasionally unmasks maternal disease.

Two patterns are measured: counting-based (massively parallel shotgun sequencing or targeted counting of chromosome-specific reads against a reference) and SNP-based methods. The performance figures below are broadly method-independent for the common trisomies.

Assessment

Screening is a structured pathway, not a single test, and the assessment is as much about who the woman is as about the marker values.

• Establish dates and chorionicity first. Every risk algorithm is gestation-dependent (NT and PAPP-A/β-hCG have narrow valid windows); a misdated pregnancy generates a spurious risk. In multiple pregnancy, chorionicity changes both the validity and the interpretation of every method.
• Quantify the prior risk — maternal age, gestation, previous affected pregnancy, parental balanced translocation. This baseline is what the screen modifies; it is also what determines the post-test predictive value of any positive result, and it must be elicited explicitly rather than assumed.
• First-trimester combined test (11–13⁺⁶ weeks): maternal age + NT + PAPP-A + free β-hCG. Reported detection for T21 is of the order of 85–90% at a ~5% false-positive rate. Its advantages are early timing (earlier reassurance or earlier decision), an NT that also flags non-aneuploid structural disease, and the chance to identify T18/T13 by their distinct biochemistry.
• Second-trimester quadruple test (≈14–20 weeks): maternal age + AFP + hCG + uE3 + inhibin A, for women who book late or where first-trimester NT is unavailable — common in the SA public sector. Detection for T21 is lower (~75–80% at a 5% FPR). The quadruple test screens for T21 (and AFP flags open neural-tube defects) but is not designed to give a discrete T18/T13 risk in the way the combined test does.
• Integrated and contingent strategies combine first- and second-trimester information to lift detection and cut the false-positive rate, at the cost of delaying the result. The contingent model — perform a cheap first-line test on everyone, reserve the expensive second-line test (cfDNA) for those whose first-line risk crosses a threshold — is the structure that makes high-performance screening affordable in a constrained system.
• cfDNA / NIPT can be deployed as a primary screen offered to all, or as a contingent second-line test after a high-risk combined or quadruple result. The interpretation hinges on prior risk (below). A no-call result warrants a repeat sample and active counselling, not silent re-bleeding.

Interpreting a result means converting a risk into a plan: a low-risk screen reduces but never abolishes risk (the residual is the false-negative rate); a high-risk screen is an invitation to confirmatory testing, not a diagnosis. The number a woman is told — "1 in 80", "high-chance" — is meaningless to her unless the counselling makes clear what proportion of women with that result actually carry an affected fetus.

Screening in multiple pregnancy and the limits of soft markers

Two assessment situations trip up an otherwise sound plan: twins and the incidental "soft marker".

In multiple pregnancy every method degrades. Serum biochemistry averages two feto-placental units, so a normal twin can mask an affected co-twin and blunt detection. NT is the more useful first-trimester tool because it is fetus-specific and lets each twin carry its own risk, and it doubles as the chorionicity and discordance assessment that governs the whole pregnancy. cfDNA performs respectably for T21 in twins but with a higher no-call rate (the fetal fraction is split, and a low contribution from one placenta can drop the assay below threshold), and it cannot localise an abnormal signal to one twin — so a positive cfDNA in twins still leads to fetus-specific invasive testing. A vanishing twin is a specific cfDNA trap: the demised co-twin's placenta sheds discordant trophoblastic DNA for weeks, producing a false-positive that an early dating scan documenting the loss should pre-empt.

Soft markers — echogenic intracardiac focus, mild pyelectasis, echogenic bowel, a short femur, choroid plexus cysts found on the anomaly scan — are weak likelihood ratios, not diagnoses. Their modern interpretation is that an isolated soft marker in a woman who has had a normal cfDNA result adds essentially nothing and should not trigger invasive testing; the same marker in an unscreened woman, or alongside other markers or a structural anomaly, shifts the prior risk enough to warrant a formal risk recalculation and a discussion of diagnostic testing. Treating every soft marker as an indication for amniocentesis is the over-reaction; ignoring a cluster of them, or one in a high-prior-risk woman, is the under-reaction.

Management

The management of a screening programme is the counselling pathway and the response to each result; structure it immediate → ongoing → long-term across the pregnancy.

Immediate — informed choice before any test. Screening is offered, never imposed. Pre-test counselling must establish that the woman wants the information, distinguish a screening test (a probability) from a diagnostic test (a karyotype), and name what she would do with each possible result — because a woman who would not act on a positive result may reasonably decline screening altogether. This is a consent conversation, and documenting that she was offered both screening and diagnostic testing and could decline either is the medicolegal core of the encounter.

Ongoing — running the chosen pathway.

• A low-risk result: reassure, but state the residual risk and that screening does not replace the anomaly scan.
• A high-risk combined/quadruple result: offer either contingent cfDNA (to refine the risk and avoid many invasive procedures) or direct invasive testing (amniocentesis/CVS) for a definitive karyotype. The choice depends on how high the risk is, the woman's tolerance of a small procedure-related miscarriage risk, and access.
• A high-risk cfDNA result is still a screen: it must be confirmed by invasive testing before any irreversible decision, because even a >99%-sensitive test with a sub-0.1% false-positive rate can have a modest positive predictive value when the condition is rare.
• A cfDNA no-call: repeat the sample once; a persistent no-call, especially with other risk features, raises the index of suspicion (it associates with aneuploidy and with adverse outcomes) and should prompt detailed ultrasound and a discussion of invasive testing rather than indefinite re-bleeding.

Long-term — acting on a confirmed result. A confirmed aneuploidy moves into multidisciplinary counselling: the natural history of the specific condition, the realistic prognosis, referral to clinical genetics and (for survivable conditions) to paediatric and fetal-medicine teams, and — where the woman chooses it and the law permits — termination of pregnancy. In South Africa termination is governed by the Choice on Termination of Pregnancy Act, and a confirmed severe fetal abnormality is a recognised indication beyond the early gestational limits; the counselling must be non-directive, support whichever decision the woman reaches, and arrange continuity of care for a continuing pregnancy as carefully as for a termination. The groundwork on Rhesus status and the small alloimmunisation risk of an invasive procedure is assumed here — anti-D prophylaxis after amniocentesis or CVS in a RhD-negative, non-sensitised woman is covered at rhesus-alloimmunisation and in Rh isoimmunisation basics.

The counselling encounter — what a defensible conversation contains

The screening consultation is the part of this topic most often done badly, and it is examinable as a clinical skill in its own right. A complete pre-test conversation establishes, in plain language, four things: that this is a screen estimating a chance, not a test that finds or excludes the condition; that a low-chance result still carries a small residual risk; that a high-chance result will lead to an offer of a definitive diagnostic test that itself carries a small miscarriage risk; and that she is free to accept or decline at every step. Eliciting what she would do with the information is not idle — a woman who would continue the pregnancy regardless may still want to know (to prepare, to deliver in the right place) or may reasonably decline screening entirely, and both are valid choices to support rather than steer.

The post-result conversation must translate the number. "High-chance, 1 in 50" should be reframed as "roughly one in fifty women with your result has an affected baby, and forty-nine do not — which is why we confirm before any decision." For a positive cfDNA the individualised positive predictive value, computed from her prior risk, is the honest currency. Non-directiveness is the discipline throughout: the clinician's role is to inform and support, not to recommend a decision. Documenting that screening and diagnostic testing were both offered, that limitations were explained, and that the woman's choice was her own is the medicolegal spine of the encounter — and in a system where a missed or mis-counselled aneuploidy is a recurrent source of litigation, that documentation is protective.

Managing the cfDNA no-call

A no-call (test failure for low fetal fraction or an uninterpretable result) is a clinical event, not an administrative one. The first step is to check the modifiable contributors — gestation too early for an adequate fetal fraction, and high maternal weight diluting the placental signal — and to repeat the sample once, ideally a little later in gestation. A second no-call should not trigger endless re-bleeding: persistent low fetal fraction is itself associated with aneuploidy (particularly the small-placenta trisomies 18 and 13 and triploidy) and with other adverse placental outcomes, so the appropriate response is a detailed ultrasound and an explicit discussion of diagnostic testing. Counting a no-call as simply "inconclusive, try again" misses that the failure carries information.

Guidelines compared

The major bodies agree on the principles — screening is optional, cfDNA is the most accurate screen for the common trisomies, and a positive cfDNA needs confirmation — but diverge sharply on how cfDNA is offered, which is the difference that matters in practice.

The fault-line is universal-offer (ACOG/SMFM) versus contingent (FASP/NICE, and SA by necessity). The disagreement is less scientific than economic: in a high-prevalence, well-resourced system a primary cfDNA offer maximises detection and minimises false positives; in a contingent model the cheaper combined/quadruple test filters the population so that the expensive test is spent only where prior risk is already elevated — which, conveniently, is also where its positive predictive value is highest. Recent direction of travel (FASP's evaluative rollout, the WHO-aligned move toward more accurate first-line tests where affordable) is toward wider cfDNA use, but no body has moved to replace diagnostic confirmation of a positive result.

The evidence & the controversy

The defining statistic of this topic is positive predictive value, and the error candidates make is quoting a single number for it. PPV is not a property of the test; it is a property of the test applied to a population with a given prior probability. The same cfDNA assay, with >99% sensitivity and a false-positive rate under 0.1%, returns very different PPVs depending on whom it screens — and two NEJM studies make this concrete. In the NEXT trial, a routine first-trimester population, the PPV of cfDNA for T21 was 80.9% — meaning roughly one in five "positive" results was a false positive even in that setting. In the CARE study, a general, lower-risk obstetric population, the PPV for T21 was only 45.5% — fewer than half of positives were true. Nothing about the assay changed between the two; the prior risk did. The consultant counselling point falls straight out of this: a woman with a positive cfDNA must be told her individualised chance of a truly affected fetus, which her clinician computes from her prior risk — and she must be told that confirmation by invasive testing is the only way to know.

A live controversy is the direct-to-consumer and "expanded" NIPT market, which is reshaping practice faster than evidence. Laboratories increasingly offer cfDNA for microdeletions (22q11.2, 1p36) and for rarer autosomal trisomies. For these rare conditions the prior probability is tiny, so even a very specific test produces a PPV that can fall into the single-digit percentages — a screen that is "positive" but where the great majority of positives are false. Marketing that presents NIPT as near-diagnostic, and the temptation to expand the panel because the blood is already drawn, are exactly where uncritical use causes harm: anxiety, unnecessary invasive procedures, and occasionally a terminated unaffected pregnancy. The defensible position is to screen for what is validated (the common autosomal trisomies, with sex-chromosome and microdeletion screening only after explicit specific consent) and to resist the panel creep.

A second, quieter controversy is whether cfDNA's very success erodes the skills and the safety net of structured screening. cfDNA does not assess fetal anatomy; the 11–13⁺⁶ NT and the mid-trimester anomaly scan find structural disease, multiple pregnancy complications and non-aneuploid pathology that a normal cfDNA result silently misses. A normal NIPT is not a normal pregnancy. There is also the incidental finding that occasionally surfaces maternal malignancy from an uninterpretable, multi-chromosome cfDNA pattern — a genuine clinical event that the consenting conversation should, briefly, foreshadow.

The point all of this converges on is that screening and diagnosis are different acts with different consequences. A screen returns a probability and is harmless to the pregnancy; a diagnostic test (amniocentesis or CVS) returns a karyotype and carries a small procedural risk. The whole architecture — combined or quadruple as a cheap first line, cfDNA as a high-performance refiner, invasive testing as the definitive confirmer — exists so that the small risk of an invasive procedure is spent only on the women whose post-screen probability justifies it, and so that no irreversible decision rests on a probability alone. Where guidelines disagree about the offer, they agree absolutely on this sequence: a positive screen of any kind is confirmed before it is acted upon. The modern miscarriage figures (procedure-related loss around 0.1–0.3% for amniocentesis and ~0.2% for CVS) matter here precisely because over-quoting the old "1 in 100" risk wrongly deters women from the confirmation that the whole pathway is built to deliver.

A newer ethical thread is the incidental maternal finding. An uninterpretable cfDNA result showing multiple chromosomal imbalances across several chromosomes has, in a small but well-documented number of cases, been the first sign of an occult maternal malignancy — most often a lymphoma or a colorectal or breast cancer shedding tumour DNA into the same plasma the assay samples. This is neither common nor a reason to alarm every woman, but it is a recognised event that the pre-test conversation should briefly foreshadow, and a multi-chromosome anomalous pattern with a normal fetal ultrasound should prompt maternal investigation rather than repeat fetal testing. A parallel societal debate concerns early fetal-sex determination: because cfDNA reports fetal sex reliably from the first trimester, it can be used — and is marketed in some settings — for non-medical sex selection. The defensible professional stance, mirrored in the international position statements, is that sex-chromosome and sex reporting carries specific consent and that screening offered for medical indications should not be repurposed as a sex-selection tool; in jurisdictions where sex-selective practice is a concern this is an active regulatory question rather than a settled one.

In South Africa the controversy is one of equity. NIPT is unfunded in the public sector and accessed almost entirely privately or through medical aid for high-risk pregnancies; the dominant public-sector screen remains maternal age, with serum screening confined to district-hospital level and frequently unavailable at clinics. The honest plan names the evidence-based pathway (combined or contingent cfDNA, confirmed invasively) and the deliverable one (age-based selection, anomaly-scan markers, and invasive testing offered where a high prior risk justifies it). A genuinely SA-relevant research direction is droplet digital PCR–based NIPT (ddNIPT), proposed as a cheaper alternative to next-generation-sequencing NIPT that might make accurate aneuploidy screening affordable at scale; it is not yet validated for routine use but is the most concrete equity lever on the horizon. There is a distinct local upside worth holding: NIPT avoids the small invasive-procedure risks that matter more in an HIV-prevalent population, where minimising procedures with any theoretical vertical-transmission concern is advantageous.

Landmark trials & key evidence

A worked arithmetic makes the PPV point unforgettable. Take cfDNA with sensitivity ~99% and specificity ~99.9% (FPR 0.1%). In a population where the prior risk of T21 is 1 in 100 (an older cohort), of 10,000 women ~100 are affected (≈99 detected) and ~9,900 unaffected (≈10 false positives), so PPV ≈ 99 / (99 + 10) ≈ 91%. In a population where the prior risk is 1 in 1,500 (a young cohort), of 15,000 women ~10 are affected (≈10 detected) and ~14,990 unaffected (≈15 false positives), so PPV ≈ 10 / (10 + 15) ≈ 40% — almost identical to CARE's observed 45.5%. Same test, same sensitivity and specificity; the PPV halved because the prior risk fell. This is precisely why "what is the PPV of NIPT?" has no single answer, and why the contingent model — applying cfDNA where prior risk is already raised — gets more reliable positives for less money.

Exam traps & red flags

• Quoting a fixed PPV for NIPT. There is no single PPV: it tracks the woman's prior risk. The defensible answer states the principle and, ideally, computes her individual figure — high in an older or screen-positive woman, much lower in a young low-risk one.
• Calling cfDNA diagnostic. It is a screen. A positive result — however high the sensitivity — is confirmed by amniocentesis or CVS before any irreversible decision. Acting on NIPT alone is the classic error that risks terminating an unaffected pregnancy.
• Misnaming the source of cfDNA. It is placental (trophoblastic), not fetal, DNA — which is why confined placental mosaicism, a vanishing twin and maternal factors cause discordant results. Saying "fetal DNA" without that caveat misses the mechanism of every false positive.
• Ignoring a no-call. A failed cfDNA result is not just a technical hiccup: low fetal fraction associates with high maternal weight, early gestation, and aneuploidy (especially T18/T13/triploidy with their small placentas). Repeat once, then escalate to ultrasound and discuss invasive testing — do not simply re-bleed indefinitely.
• Treating a normal NIPT as a normal pregnancy. cfDNA screens only for the chromosomes tested; it sees no fetal anatomy. The NT and anomaly scans remain essential for structural disease, and a low-risk screen never abolishes residual risk.
• Expanded-panel and microdeletion creep. For rare microdeletions the prior probability is tiny, so the PPV can be single-digit; offering them as routine, or as near-diagnostic, generates anxiety and unnecessary invasive procedures. Validated common trisomies only, with specific consent for anything beyond.
• Over-quoting invasive-procedure loss. The "1 in 100/200" miscarriage figure is outdated; modern meta-analyses put procedure-related loss far lower (amniocentesis ~0.1–0.3%, CVS ~0.2%), and over-quoting it inappropriately deters women from definitive testing.
• Forgetting anti-D. An invasive procedure in a RhD-negative, non-sensitised woman is a sensitising event requiring anti-D prophylaxis.
• Presenting NIPT as the SA standard. It is unfunded in the public sector. The credible plan names the evidence-based pathway and the deliverable one (age-based selection, anomaly-scan markers, invasive testing where prior risk justifies it).

Evidence anchors

• FASTER trial — Malone et al., N Engl J Med 2005 — first- vs second-trimester vs combined T21 screening; combined 85% / quad 81% / integrated 96% detection at 5% FPR
• NEXT trial — Norton et al., N Engl J Med 2015
• CARE study — Bianchi et al., N Engl J Med 2014
• cfDNA screening performance meta-analysis — Gil et al., Ultrasound Obstet Gynecol 2017
• Procedure-related miscarriage after amniocentesis/CVS — Akolekar et al., Ultrasound Obstet Gynecol 2015
• Updated procedure-related miscarriage meta-analysis — Salomon, Sotiriadis, Akolekar et al., Ultrasound Obstet Gynecol 2019
• ACOG Practice Bulletin 226 — Screening for Fetal Chromosomal Abnormalities, Obstet Gynecol 2020
• ISPD position statement on NIPT — Hui et al., Prenat Diagn 2023
• Scoping review: NIPT in South Africa — Labuschagne et al., J Community Genet 2025
• NHS Fetal Anomaly Screening Programme (FASP) handbook — combined test (CRL 45–84 mm; maternal age + NT + PAPP-A + free β-hCG), quadruple test (14⁺² to 20⁺⁰ weeks; AFP, hCG, uE3, inhibin A), higher-chance cut-off ≥1 in 150 at term, contingent NIPT after a higher-chance result.
• The value and role of NIPT in a select South African population — Mnyani, Nicolaou & Bister, SAMJ 2016 (NIPT as contingency screening in high-/intermediate-risk women; abnormal results confirmed by invasive testing).
• South African NDoH/SASOG Clinical Guidelines for Genetics Services (2021) — maternal-serum screening at district-hospital level; maternal age the dominant public-sector risk filter.