Evaluate and manage thrombocytopenia and haematological disorders in pregnancy

In one line

Thrombocytopenia complicates 7–12% of pregnancies; the task is not to count platelets but to separate benign gestational thrombocytopenia from the dangerous mimics — immune thrombocytopenia (ITP), pre-eclampsia/HELLP, and the thrombotic microangiopathies (TTP, atypical HUS, acute fatty liver) — because a wrong label drives the wrong delivery decision.

This chapter assumes the platelet-physiology and pre-eclampsia groundwork in hypertension in pregnancy and pre-eclampsia & HELLP basics; the consultant-level content is the discriminators, the subtype-specific regimens, and the judgement calls. The question is not what a falling platelet count is but what it means when the smear, the gestation and the renal function point in different directions.

Assessment

The diagnosis is a pattern-recognition and exclusion problem, not a count. Four mechanistic buckets generate almost every case, and each has a signature read off the bloods before the label is written.

1. Dilution / accelerated clearance (gestational thrombocytopenia, GT). Plasma-volume expansion plus mildly increased platelet turnover at the placental bed. 75–80% of all cases; a third-trimester, asymptomatic, >80 × 10⁹/L phenomenon that resolves by 1–2 months postpartum. It has no first-trimester form and never causes neonatal thrombocytopenia.

2. Immune destruction (ITP, HIV-related, drug-induced, SLE/APS). Antiplatelet (or, in HIV, immune-complex and molecular-mimicry) clearance by the reticuloendothelial system. The mechanism that matters: the antibody crosses the placenta, so unlike GT it can — uncommonly — drop the neonatal count. ITP is the commonest cause of a count <50 × 10⁹/L before 20 weeks.

3. Consumption (pre-eclampsia/HELLP, DIC, abruption, sepsis, TTP/aHUS). Platelets are consumed in a microangiopathic or coagulopathic process. The signature is that thrombocytopenia travels with other organ injury — haemolysis, transaminitis, renal failure, fibrinogen consumption — not in isolation.

4. Underproduction (marrow failure, B12/folate deficiency, advanced HIV, infiltration). Rare but the one bucket a single blood film and reticulocyte count will catch — look for a leuco-erythroblastic film, macrocytosis, or pancytopenia.

The discriminators that decide the diagnosis:

• Timing and trajectory. A count <100 × 10⁹/L in the first or early second trimester, or a steadily falling count, points away from GT and toward ITP or an evolving consumptive process. GT has no early-pregnancy phenotype; if you see one, it is not GT.
• The number itself, read against the gestation. Below 70 × 10⁹/L GT becomes statistically improbable — in dedicated series the GT nadir rarely falls below this. Below 50 × 10⁹/L actively hunt an alternative diagnosis. Isolated thrombocytopenia ≠ benign.
• The smear is the single most informative cheap test. Schistocytes flag a microangiopathy (TTP/HUS/HELLP); platelet clumping flags EDTA-induced pseudo-thrombocytopenia (repeat in citrate or by a manual count before you commit a patient to a workup); a leuco-erythroblastic picture flags marrow infiltration; hypersegmented neutrophils + macrocytosis flag B12/folate. Add LDH, haptoglobin, reticulocytes, creatinine, transaminases, urate, a coagulation profile and fibrinogen.
• The SA-specific filter. In an antenatal population with HIV seroprevalence above a quarter, HIV is a leading remediable cause — via immune destruction (which can occur at a preserved CD4 count and may be the presenting feature) and, in advanced disease, marrow suppression. Test and stage HIV before labelling any thrombocytopenia "gestational" or "immune", and remember that ART-related and TB-treatment–related cytopenias add a drug axis.
• No diagnostic test confirms ITP or GT — both are diagnoses of exclusion. Bone marrow and antiplatelet-antibody assays are not required to diagnose ITP in pregnancy, and a normal marrow does not exclude it.

The atypical and the deceptive

The straightforward picture is the asymptomatic third-trimester woman with a count of 110. The harder ones do not read cleanly:

• The GT/ITP overlap zone (50–80 × 10⁹/L, third trimester, no prior counts). This is the genuinely hard call, and it is usually unresolvable in real time. When pre- and post-pregnancy counts are later checked, roughly one in three women labelled GT actually meet ITP criteria — so the safest consultant move in the overlap zone is to manage the count, document the uncertainty, and book a postpartum recheck at 6–8 weeks, which retrospectively makes the diagnosis (normalised → GT; persistently low → ITP). An experimental discriminator exists — serum thrombopoietin runs far higher in ITP than GT — but it is not an NHLS routine and does not change acute management.
• Severe isolated thrombocytopenia (<20–30 × 10⁹/L) with a normal smear and normal organs. This is ITP until proven otherwise; GT essentially never does this, and a consumptive process would derange something else.
• Thrombocytopenia that worsens after delivery. HELLP and acute fatty liver improve once the placenta is out; a count that keeps falling on day 2–3 postpartum, especially with rising creatinine or LDH, is TTP or atypical HUS revealing itself — the placenta was never the driver.
• "HELLP" that will not resolve. Persistent haemolysis and thrombocytopenia beyond ~72 h postpartum should reopen the microangiopathy differential rather than be accepted as slow HELLP recovery.

Severity stratification and the high-stakes split

Stratify by count band (driving bleeding and procedural risk) crossed with organ involvement (driving the diagnosis and delivery decision). A count of 60 with normal organs and a stable trend is a different patient from a count of 60 with rising LDH and creatinine, even though the number is identical.

The single highest-stakes split is HELLP vs TTP vs aHUS vs acute fatty liver (AFLP), because management diverges sharply:

HELLP and AFLP improve with delivery; TTP and aHUS do not — delivering a TTP patient instead of starting plasma exchange is a classic, fatal error. Three mechanism→consequence links carry this table: (1) TTP is an ADAMTS13 problem, so the deficiency is severe (≤10%) and the haemolysis is the most florid — LDH is highest and schistocytes most numerous; (2) aHUS is a complement-regulation problem centred on the renal endothelium, so renal failure dominates and is disproportionate to the BP and liver; (3) AFLP is a hepatic mitochondrial/fatty-acid-oxidation problem, so it uniquely couples thrombocytopenia with hypoglycaemia and true coagulopathy (low fibrinogen, prolonged PT) — the combination that should never be called "HELLP".

A practical bedside rule when the pre-eclampsia history is convincing: if BP and transaminases dominate and the count recovers after delivery, it was HELLP; if renal failure or coagulopathy is out of proportion to the hypertension, or the picture worsens postpartum, escalate down the TMA/AFLP path the same day.

Management

Management runs immediate → ongoing → long-term, and is subtype-specific — treat the diagnosis, not the number.

Immediate (any cause). Resuscitate active bleeding; correct coagulopathy; transfuse platelets only for active haemorrhage, DIC, or to cover surgery — never to "treat a number" in stable ITP. The arithmetic explains why: one apheresis unit (or a pool of 4–6) raises an average adult by only ~20–40 × 10⁹/L and in ITP those platelets are antibody-coated and cleared within hours, so you buy a transient, expensive bump and no durable count. In a microangiopathy (TTP/aHUS) platelet transfusion may even feed the thrombotic process and is reserved for life-threatening bleeding only. If a microangiopathy is suspected, escalate to a regional/tertiary haematology–obstetric team the same day — the diagnostic clock and the treatment clock are the same clock.

ITP — the regimen, and how the agents differ

Treat only for bleeding or platelets <20–30 × 10⁹/L, or to reach a delivery-safe count near term. The two first-line agents are not interchangeable — they differ in speed, durability, placental transfer and side-effect profile, and the choice turns on which of those axes matters for this woman:

Two deliberate departures from non-pregnant practice: the steroid dose is halved (the non-pregnant ASH default is ~1 mg/kg), and prednisolone is chosen over dexamethasone because dexamethasone crosses the placenta largely intact (it is the fetal-lung-maturation steroid) and is reserved for fetal indications — prednisolone is ~90% metabolised by placental 11β-hydroxysteroid dehydrogenase, so the fetus is spared.

Second line / refractory (tertiary, individualised): combine IVIG + steroid; azathioprine (pregnancy-compatible immunosuppressant); rituximab (effective but causes neonatal B-cell depletion and blunted vaccine responses — coordinate paediatric follow-up). The international consensus permits TPO-receptor agonists (eltrombopag, romiplostim) in the third trimester for refractory, bleeding ITP failing first-line therapy — but the obstetric evidence is case-series only, so this is an individualised tertiary decision, not a routine step. IV anti-D is an option only in an Rh(D)-positive, non-splenectomised woman and is generally pushed to third line because of the maternal/fetal haemolysis risk. Splenectomy (ideally second trimester, laparoscopic, after steroids and IVIG have failed) is now rarely needed but remains the definitive option for truly refractory bleeding disease. Mycophenolate and cyclophosphamide are teratogenic and contraindicated.

Delivery-count targets. Aim ≥50 × 10⁹/L for caesarean and ≥30 × 10⁹/L for an uncomplicated vaginal birth; start prednisolone or IVIG at ~36 weeks if below target (IVIG if you need the count up fast for a scheduled birth). Mode of delivery follows obstetric indication, not the platelet count — there is no evidence caesarean reduces neonatal haemorrhage, and the maternal count does not predict the neonatal count. Avoid the manoeuvres that injure a potentially thrombocytopenic fetus: fetal scalp electrodes, fetal blood sampling, ventouse, and rotational/mid-cavity forceps.

Neuraxial anaesthesia — the number, and what sits behind it

The SOAP interdisciplinary consensus supports neuraxial block at ≥70 × 10⁹/L with a stable count, normal coagulation and no antiplatelet/anticoagulant drugs; 50–70 × 10⁹/L is an individualised risk–benefit decision; below 50 × 10⁹/L, generally avoid. The numbers behind the threshold are the modelled upper 95% confidence bounds for spinal–epidural haematoma: roughly 0.19% at 70–100 × 10⁹/L, 2.6% at 50–69 × 10⁹/L, and ~9% below 50 × 10⁹/L in this obstetric population. The risk is not a step function at 70 — it climbs steeply through the 50s — which is exactly why a count of 68 on a rising trend with normal coagulation can be reasonable while 68 and falling with HELLP physiology is not. The number is necessary but not sufficient; trajectory and coagulation complete the decision.

TTP — the emergency that delivery does not fix

Daily therapeutic plasma exchange (TPE) is the emergency intervention and must start on clinical suspicion — do not wait for the ADAMTS13 result (turnaround can be days; the mortality of untreated TTP is the reason). Add corticosteroids; add rituximab for confirmed acquired (antibody-mediated) TTP to suppress the autoantibody; and, in specialist centres, caplacizumab (an anti-von-Willebrand-factor nanobody) to accelerate platelet recovery and reduce recurrence. The pregnancy nuance: a first-ever presentation of TTP in pregnancy may be congenital (Upshaw–Schulman) — ADAMTS13 severely low but no inhibitor antibody — because the late-pregnancy surge in von Willebrand factor (to ~200–300% of baseline) unmasks a previously silent enzyme deficiency. Congenital TTP is treated with plasma infusion (or recombinant ADAMTS13), not immunosuppression, and recurs predictably in every subsequent pregnancy unless given prophylactic plasma — so getting the antibody result changes both this pregnancy's drug choice and the next pregnancy's plan.

aHUS and AFLP

aHUS (complement-mediated, renal-dominant, often postpartum-triggered) does not respond reliably to plasma exchange and is treated with the terminal-complement inhibitor eculizumab (with meningococcal vaccination/prophylaxis); suspect it when renal failure dominates and TPE is not working. AFLP is delivery-plus-supportive: expedite birth, correct hypoglycaemia and coagulopathy (FFP, cryoprecipitate, glucose), and support the liver — it is a different disease from HELLP wearing the same coat.

Long-term/postpartum. Confirm GT by documented resolution at 6–8 weeks (this is what retrospectively distinguishes it from ITP); counsel ITP recurrence in future pregnancies and the need for early-pregnancy counts. Arrange neonatal cord and serial platelet counts for any antibody-mediated maternal disease — the neonatal nadir is at days 2–5, after the cord count, so a reassuring cord platelet does not end surveillance. When the neonatal thrombocytopenia is severe or the maternal count was normal, think neonatal alloimmune thrombocytopenia (NAIT) rather than ITP (see below) and apply the fetomaternal-antibody logic developed for Rh & red-cell alloimmunisation.

Neonatal alloimmune thrombocytopenia (NAIT) — the platelet analogue of Rh disease

NAIT is fetomaternal incompatibility for a human platelet antigen (HPA-1a accounts for ~75–80% of cases), with maternal alloantibody crossing the placenta to destroy fetal platelets — the exact mechanistic parallel to red-cell alloimmunisation, but for platelets. Two features make it dangerous and distinct from ITP: it can be severe enough to cause fetal/neonatal intracranial haemorrhage, and — unlike Rh disease and unlike ITP, where the mother's own count is low — the mother's platelet count is normal and up to roughly half of cases present in the first affected pregnancy (no prior sensitisation needed, because exposure is to a paternally-derived antigen). It is therefore typically diagnosed after an unexpectedly thrombocytopenic or intracranially-bled neonate, or in a pregnancy following an affected sibling. Antenatal management of a subsequent at-risk pregnancy is maternal IVIG ~1 g/kg weekly from around 20–24 weeks, ± corticosteroids, with referral to fetal-medicine; the diagnosis is confirmed by HPA typing of both parents and maternal anti-HPA antibodies. A healthy mother with a bled or profoundly thrombocytopenic baby is NAIT, not maternal ITP, and the management is given to the mother in the next pregnancy, not the current neonate alone.

The evidence & the controversy

The most relevant controversy is that the major ITP guidelines explicitly exclude pregnancy: the ASH 2019 guideline states it does not address pregnancy, so SA practice leans on the Provan international consensus (2019) and the obstetric/anaesthetic literature. The "treat at 30" threshold is the non-pregnant ASH figure extrapolated, not graded for the pregnant patient.

The neuraxial threshold is genuinely contested. The historical "100 × 10⁹/L for epidural" dogma is obsolete; the SOAP consensus reset the floor to 70 × 10⁹/L, but acknowledges the haematoma-risk estimate is imprecise because the event is so rare no trial can power it — the defensible numbers are modelled upper confidence bounds (≈0.19% at 70–100k), not measured rates. A 65–70 individualised decision rests on trajectory, coagulation, and absence of other risk factors, not on a single cut-off.

A live first-line-therapy debate is whether to intensify upfront. The FLIGHT trial showed adding mycophenolate to steroids improved response and reduced refractoriness in non-pregnant ITP — but mycophenolate is teratogenic and contraindicated in pregnancy, so FLIGHT cannot be transplanted to the antenatal patient: a strong RCT whose population excludes the pregnant woman.

The caplacizumab evidence, with the arithmetic. HERCULES randomised 145 patients with acquired TTP to caplacizumab or placebo on a background of plasma exchange. The composite of TTP-related death, recurrence, or major thromboembolic event during the treatment period was 12% with caplacizumab vs 49% with placebo. Worked through: absolute risk reduction ≈ 49% − 12% = 37%, so the NNT ≈ 1 / 0.37 ≈ 2.7, i.e. ~3 patients treated to avoid one composite event — a striking effect size, though driven heavily by reduced recurrence/exacerbation rather than mortality, and the agent does not replace plasma exchange or immunosuppression. In SA the constraint is access and cost: plasma exchange + steroids ± rituximab remains the deliverable backbone, with caplacizumab a specialist add-on where available.

Finally, in the SA setting the most important "controversy" is overdiagnosis — labelling HIV- or pre-eclampsia-related thrombocytopenia as GT/ITP. Sebitloane's SA cohort found most antenatal thrombocytopenia, even in untreated HIV, was gestational, but the corollary holds: you cannot make that call without excluding HIV and pre-eclampsia first, and the one-in-three GT-labels-that-were-really-ITP finding shows how often the easy label is wrong.

Landmark trials & key evidence

Worked viva — how to structure the answer

A stem like "a 28-year-old G2P1 at 33 weeks, no antenatal complications, FBC done for an unrelated bruise shows platelets 24 × 10⁹/L, normal smear, normal organs, HIV-negative." A high-scoring answer runs:

1. Frame the count against the gestation. "A count of 24 with a normal smear, normal organs and a third-trimester presentation is too low and too isolated for gestational thrombocytopenia, which rarely goes below 70; this is immune thrombocytopenia until proven otherwise. I would confirm it is real (rule out EDTA clumping by repeating in citrate) and exclude HIV, pre-eclampsia and drugs."
2. Decide whether to treat, and why. "She is below 30 and bleeding, so I would treat. With seven weeks to term I have time, so I would start prednisolone 0.25–0.5 mg/kg/day — deliberately below the non-pregnant dose, and prednisolone not dexamethasone, which crosses the placenta. If I needed a fast rise — bleeding or imminent delivery — I would use IVIG 1 g/kg, which works in 1–3 days."
3. Plan the delivery on the count and the obstetrics. "I aim for ≥50 for a caesarean and ≥30 for a vaginal birth, escalating with IVIG near 36 weeks if needed. Mode of delivery is by obstetric indication — caesarean does not protect the baby's platelets — and I avoid scalp electrodes, fetal blood sampling, ventouse and rotational forceps."
4. Anaesthetic plan. "For neuraxial block I want ≥70 and a stable count with normal coagulation; between 50 and 70 it is an individualised decision; below 50 I avoid it."
5. Neonatal plan. "Cord platelets and serial neonatal counts to the day-2–5 nadir, because maternal antibody can cross; if the baby is severely affected with a normal maternal count I would reconsider NAIT."
6. Close the loop. "Postpartum I recheck her count at 6–8 weeks to confirm the diagnosis, and counsel recurrence and early counts in a future pregnancy."

The same scaffold flips for the microangiopathy stem: frame the organ pattern → name HELLP vs TTP vs aHUS vs AFLP → state that delivery fixes two and plasma exchange/eculizumab fixes the others → get ADAMTS13 and read the smear → do not delay TPE for the result.

Exam traps & red flags

• Delivering a TTP patient instead of starting plasma exchange — TTP and aHUS do not resolve with delivery; HELLP and AFLP do. Get the ADAMTS13 and read the smear, but start TPE on suspicion.
• A count that falls after delivery — HELLP/AFLP recover postpartum; worsening thrombocytopenia with disproportionate renal failure is TTP/aHUS unmasking itself.
• Quoting "100 × 10⁹/L for epidural" — outdated; the defensible floor is 70 × 10⁹/L (individualise 50–70), and the risk climbs steeply, not stepwise, below it.
• Choosing caesarean to "protect the baby's platelets" — mode of delivery is by obstetric indication; the maternal count does not predict the neonatal count, and severe neonatal thrombocytopenia is rare in gestational thrombocytopenia.
• Transfusing platelets in stable ITP to reach a number — one unit raises an adult only ~20–40 × 10⁹/L and the platelets are cleared within hours; use steroids/IVIG, reserving transfusion for active bleeding or surgical cover.
• Dexamethasone for maternal ITP — it crosses the placenta; use prednisolone unless a fetal indication exists.
• Mycophenolate in pregnancy — teratogenic; never an antenatal ITP agent despite FLIGHT.
• Calling it GT without excluding HIV and pre-eclampsia — in SA the commonest missed remediable causes; and one in three "GT" labels are really ITP on later counts, so document and recheck at 6–8 weeks.
• Missing AFLP — hypoglycaemia + coagulopathy + thrombocytopenia is acute fatty liver until proven otherwise; deliver and support, do not call it slow HELLP.
• A healthy mother with a profoundly thrombocytopenic or intracranially-bled baby — think NAIT, not maternal ITP; the management belongs to the next pregnancy (maternal IVIG from 20–24 weeks).
• A first-ever TTP in pregnancy — consider congenital (Upshaw–Schulman): ADAMTS13 very low, no antibody — treat with plasma infusion, not immunosuppression, and plan prophylaxis next pregnancy.

Evidence anchors

• Provan et al. Updated international consensus report on the investigation and management of primary immune thrombocytopenia. Blood Adv 2019;3(22):3780–3817.
• Neunert et al. American Society of Hematology 2019 guidelines for immune thrombocytopenia. Blood Adv 2019;3(23):3829–3866.
• Bauer et al. SOAP Interdisciplinary Consensus Statement on Neuraxial Procedures in Obstetric Patients With Thrombocytopenia. Anesth Analg 2021;132(6):1531–1544.
• Scully et al. Caplacizumab Treatment for Acquired Thrombotic Thrombocytopenic Purpura (HERCULES). N Engl J Med 2019;380:335–346.
• Cheng et al. Eltrombopag for management of chronic immune thrombocytopenia (RAISE). Lancet 2011;377:393–402.
• Bradbury et al. Mycophenolate Mofetil for First-Line Treatment of Immune Thrombocytopenia (FLIGHT). N Engl J Med 2021;385:885–895.
• Sebitloane HM. Thrombocytopenia during pregnancy in women with HIV infection receiving no treatment. S Afr Med J 2016;106(2).
• Diagnosis and management of thrombocytopenia in pregnancy. Blood Res 2022.
• NDoH Guidelines for Maternity Care in South Africa / National Integrated Maternal and Perinatal Care Guideline (2024) — district→regional→tertiary referral and FBC at booking. (No stable per-section DOI.)