Manage obstetric cholestasis and liver disease in pregnancy

In one line

Liver disease in pregnancy splits into pregnancy-specific disorders (intrahepatic cholestasis, the pre-eclampsia/HELLP spectrum, acute fatty liver) and coincidental disease (viral hepatitis, gallstones). The peak total bile acid stratifies cholestasis-related stillbirth risk and times delivery, and acute fatty liver and hepatitis E are true maternal emergencies the itchy-but-well cholestasis patient is not.

This chapter assumes the diagnostic and first-line groundwork in pre-eclampsia & HELLP basics and the Intermediate medical-disorders content; it covers distinguishing the subtypes by mechanism, defending a bile-acid threshold, and running the sick-liver emergency. The shared severe-pre-eclampsia engine (magnesium regimens, BP control, fluid discipline) is developed in pre-eclampsia-early-onset-severe and hypertension-in-pregnancy-antihypertensives.

Assessment

The discriminating question is always which liver disease, and is the woman sick or well? Itch with a normal-looking woman points to intrahepatic cholestasis of pregnancy (ICP); a vomiting, encephalopathic, hypoglycaemic woman points to acute fatty liver of pregnancy (AFLP) or fulminant hepatitis E — a different universe of urgency. Beyond the Intermediate-level clinical pictures, the discriminating skill is classifying the subtypes by mechanism, reading the atypical presentations, and stratifying severity.

Classify by mechanism. The pregnancy-specific liver disorders are three distinct lesions, and naming the lesion predicts the management:

• ICP — a transport/canalicular lesion. Oestrogen and progesterone metabolites impair the bile-salt export pump (BSEP/ABCB11) and the canalicular phospholipid flippase (MDR3/ABCB4); bile acids accumulate in maternal serum and cross to the fetus. The consequence that matters is fetal, not maternal: bile acids are arrhythmogenic and cause acute fetal cardiac events, which is why ICP kills by sudden stillbirth in an otherwise well fetus with a normal CTG the day before — a death that surveillance cannot reliably predict and only timed delivery prevents. There is a familial/genetic subgroup (ABCB4 variants) that presents early, runs severe, and overlaps with progressive familial intrahepatic cholestasis — flag these for hepatology after delivery.
• AFLP — a mitochondrial fatty-acid-oxidation lesion. Microvesicular steatosis from impaired β-oxidation; the established association is a fetus homozygous for a long-chain 3-hydroxyacyl-CoA dehydrogenase (LCHAD) / mitochondrial trifunctional protein defect (the G1528C variant is the common allele) carried by a heterozygous mother, whose liver cannot clear the fetal/placental 3-hydroxy-fatty-acid load. It is a metabolic failure that only resolves once the placenta is removed, and the neonate must be screened because LCHAD deficiency is treatable and lethal if missed. The mother's own genotype does not predict who gets AFLP.
• HELLP — an endothelial/microangiopathic lesion. The same anti-angiogenic endothelial injury as pre-eclampsia with severe features (see pre-eclampsia-early-onset-severe for the sFlt-1/PlGF mechanism), expressed in the hepatic microvasculature: periportal haemorrhage, fibrin deposition, sinusoidal obstruction, and the feared subcapsular haematoma. It sits on the pre-eclampsia spectrum and is managed as such.

Stratify ICP by the peak value, not a single reading. Severity, and therefore stillbirth risk and delivery timing, tracks the peak random (non-fasting) total bile acid: mild 19–39, moderate 40–99, severe ≥100 µmol/L (RCOG GTG 43, 2022). Transaminases are not required and ICP does not need deranged LFTs; a single value can sit in a lower band than the peak, so it is the trajectory and the highest value that grade the disease; and the diagnosis is one of exclusion — atypical itch (early onset, rash, marked jaundice, very high transaminases, or persistence postpartum) mandates a fuller liver screen (hepatitis B/C/E serology, autoimmune screen — anti-mitochondrial/anti-smooth-muscle antibodies — and an ultrasound) because primary biliary cholangitis, autoimmune hepatitis and biliary obstruction all masquerade as "ICP".

The diagnostic threshold is where the guidelines disagree. RCOG diagnoses ICP on itch plus a peak bile acid ≥19 µmol/L; the US Society for Maternal-Fetal Medicine (SMFM Consult #53) is more permissive, treating any bile-acid elevation above the laboratory upper limit of normal (often quoted around 10 µmol/L) in a pruritic woman as diagnostic. The consequence is a larger, milder cohort labelled "ICP" under SMFM. This matters less than it sounds, because both anchor delivery timing on the same ≥100 µmol/L stillbirth threshold — the diagnostic cut-off changes who you label, not who you deliver early.

Read the atypical and overlap presentations.

• AFLP presents in the third trimester with malaise, vomiting and abdominal pain, then hypoglycaemia, coagulopathy, hyperuricaemia, renal impairment and encephalopathy. The Swansea criteria (≥6 of 14 features) support the diagnosis: vomiting, abdominal pain, polydipsia/polyuria, encephalopathy, raised bilirubin, hypoglycaemia, raised urate, leucocytosis, ascites or a bright liver on ultrasound, raised transaminases, raised ammonia, renal impairment, a coagulopathy (prolonged PT), and microvesicular steatosis on biopsy. The criteria are sensitive but not specific (they overlap heavily with pre-eclampsia with severe features), so they support rather than make the diagnosis; biopsy is rarely done because the coagulopathy makes it dangerous and it does not change the management — deliver.
• The overlap. HELLP, AFLP and pre-eclampsia with severe features coexist and blur — many women have features of more than one. Pragmatic discriminators that change action: hypoglycaemia, disproportionate hyperuricaemia, hyperammonaemia, a frank coagulopathy with low fibrinogen and prolonged PT, and profound malaise out of proportion to the BP favour AFLP; haemolysis (LDH, schistocytes) with low platelets and near-normal glucose and clotting favours HELLP; severe hypertension with proteinuria and a relatively preserved liver favours pre-eclampsia. Either way the treatment is delivery — but AFLP additionally needs aggressive glucose and coagulation correction and neonatal LCHAD screening that HELLP does not.
• The mimics that do not resolve with delivery. Thrombotic thrombocytopenic purpura and atypical haemolytic-uraemic syndrome present with microangiopathic haemolysis and thrombocytopenia and are routinely mistaken for HELLP, but carry disproportionate renal failure or neurological signs, often worsen after delivery, and need plasma exchange (TTP) or complement blockade (aHUS) — not delivery alone. A "HELLP" that fails to improve 48–72 h postpartum is one of these until proven otherwise. (Developed further in pre-eclampsia-early-onset-severe.)

Coincidental disease in the SA setting. Screen for hepatitis B universally (SA antenatal practice); hepatitis E is the lethal one in pregnancy (below); right-upper-quadrant pain with normal LFTs in the third trimester is gallstones until proven otherwise (cholelithiasis is commoner in pregnancy because oestrogen raises biliary cholesterol saturation and progesterone slows gallbladder emptying). New jaundice with a cholestatic picture also raises drug-induced liver injury — in the SA setting, anti-tuberculosis therapy and antiretrovirals are the common culprits and must be on every differential.

Management

Work immediate → ongoing → long-term, and keep the sick-liver (AFLP/HELLP/HEV) pathway separate from the itchy-but-well (ICP) pathway. The shared pre-eclampsia engine is in pre-eclampsia-early-onset-severe.

Intrahepatic cholestasis — ongoing surveillance, then timed delivery.

These thresholds are RCOG GTG 43 (2022); the logic is the bile-acid–stillbirth curve from Ovadia 2019 — risk is near-background below 100 µmol/L and rises sharply above it. The judgement calls that sit on top of the table:

• Comorbidity and multiple pregnancy pull timing earlier. A rapidly rising bile acid, severe symptoms, or a comorbidity (twins, GDM, pre-eclampsia) shifts birth earlier. Twins are the important one: the background stillbirth risk is higher and bile acids run higher, so most units lower the threshold for early delivery in twin ICP even below the singleton numbers — but the singleton ≥100 µmol/L evidence does not transfer cleanly, and this is an individualised, counselled decision rather than a fixed gestation.
• What surveillance can and cannot do. CTG and ultrasound do not predict the sudden arrhythmic stillbirth of severe ICP — a normal CTG yesterday is no reassurance. In ICP, timing of delivery is the surveillance, and serial bile acids drive timing. Continuous intrapartum EFM is reasonable in severe disease, but its purpose is intrapartum, not antenatal, safety.
• Symptom control — the named drugs and where they sit post-PITCHES. Ursodeoxycholic acid (UDCA) 500 mg bd (titratable to ~1 g–1.5 g/day) is first-line, primarily for maternal itch and biochemistry — not as a guaranteed perinatal-outcome treatment (PITCHES; see the controversy). Add chlorphenamine for sleep, topical emollients/menthol for itch, and vitamin K (water-soluble menadiol) if the prothrombin time is prolonged or there is steatorrhoea (fat-soluble-vitamin malabsorption raises both maternal and neonatal bleeding risk). For refractory severe ICP unresponsive to UDCA, rifampicin is the recognised second-line add-on (it induces alternative bile-acid detoxification pathways and improved itch and biochemistry in the King's College series), but it does not change the delivery-timing calculus. Cholestyramine is an older option that worsens fat-soluble-vitamin malabsorption (so co-prescribe vitamin K) and is now little used. ICP needs delivery at the right time, not endless drug escalation.

AFLP — resuscitate and deliver, now. An obstetric emergency with no expectant window. Immediate: ABC, correct hypoglycaemia (10% dextrose infusion, frequent glucose checks — this is the single most under-treated derangement and contributes to the encephalopathy), correct coagulopathy (FFP, cryoprecipitate for fibrinogen, vitamin K), treat hyperammonaemia, and involve ICU, anaesthesia and a referral-level hepatology/MFM team early. Definitive: expedite delivery regardless of gestation once the mother is stabilised — the liver recovers only after the pregnancy ends. Mode is obstetric: vaginal if rapidly achievable and the coagulopathy permits, else caesarean with meticulous haemostasis and a low threshold for blood-product cover (these women bleed). Post-delivery: the liver typically recovers over days but can transiently worsen before it improves; manage in high-care, watch for hepatic encephalopathy, hypoglycaemia, AKI and infection, and have a transplant referral pathway in mind for the minority who progress to fulminant failure. Long-term: screen mother and neonate for fatty-acid oxidation (LCHAD/MTP) defects — newborn screening and early dietary management are life-saving for the affected child — and counsel a real recurrence risk in future pregnancies, especially where an LCHAD defect is confirmed.

HELLP is managed as pre-eclampsia with severe features: magnesium sulphate for seizure prophylaxis, control blood pressure, stabilise then deliver (the magnesium regimens, BP agents and fluid discipline are in pre-eclampsia-early-onset-severe and hypertension-in-pregnancy-antihypertensives — not repeated here). The HELLP-specific points: transfuse platelets only for active bleeding or to cover surgery (a count safely above the unit's neuraxial threshold, commonly cited around >50 × 10⁹/L for caesarean and higher for regional anaesthesia, with a falling trend giving pause), watch for the hepatic subcapsular haematoma (sudden shoulder-tip/RUQ pain, shock, a falling haematocrit — image, involve surgery/interventional radiology, do not rupture it at laparotomy), and remember the postpartum nadir of platelets is typically day 1–3, so HELLP can deepen after delivery. Routine high-dose corticosteroids do not improve maternal HELLP outcomes — the Cochrane review (Woudstra 2010) found no effect on maternal death, severe morbidity or perinatal death; the only clear effect was a higher platelet count (dexamethasone > betamethasone), which has not translated into better clinical endpoints. Give steroids for fetal lung maturation where preterm delivery is planned, not to "treat the platelets".

Viral hepatitis in the SA setting. For hepatitis B, the NDoH pathway is PVT (prevention of vertical transmission — the current term, not "PMTCT"). Antiviral prophylaxis is maternal tenofovir (TDF) from around the third trimester (≈28–32 weeks) when the HBV viral load is high (the widely adopted WHO threshold is HBV DNA ≥200,000 IU/mL) or the mother is HBeAg-positive, on top of universal neonatal immunoprophylaxis — hepatitis B immunoglobulin and the birth-dose vaccine within 12 hours, completing the schedule. Maternal antiviral suppression lowers the viral inoculum the neonate meets, and immunoprophylaxis blocks what gets through; both are needed in the high-viral-load mother, and omitting either is a transmission failure. Where HBV DNA quantitation is not readily available at district level, HBeAg status is the practical proxy for "high transmission risk", and these women belong on the referral pathway for tenofovir. Hepatitis E management is supportive in high-care/ICU with early transfer — no specific antiviral, prevention is clean water and sanitation, and the licensed vaccine is not part of routine SA antenatal care. The lethal pregnancy form is genotype 1/2 (epidemic, faecal–oral, water-borne, endemic in resource-limited settings) — the zoonotic genotype 3/4 seen in higher-income settings is not associated with the catastrophic third-trimester maternal outcome, so the clinical weight a positive HEV result carries depends on the epidemiological setting.

The evidence & the controversy

Does ursodeoxycholic acid actually save babies? For two decades UDCA was given to reduce stillbirth. PITCHES (Chappell, 2019) — the largest RCT, 605 women — found no reduction in the composite of perinatal death, preterm birth or neonatal-unit admission (adjusted RR 0.85, 95% CI 0.62–1.15), reframing UDCA from a fetal-protective drug to a maternal symptom drug. UDCA modestly improves itch and biochemistry, is safe, and stays in guidelines — but it does not prevent stillbirth, and it does not replace correctly timed delivery.

The counter-argument is the Ovadia 2021 individual-participant-data meta-analysis, which pooled trials beyond PITCHES and found UDCA associated with reduced spontaneous preterm birth, with an RCT-only signal for stillbirth-plus-preterm-birth — though it could not independently demonstrate a stillbirth reduction (event rates are tiny, so every study is underpowered for that endpoint). The defensible synthesis: UDCA is reasonable, chiefly for symptoms and possibly preterm birth; bile-acid–guided timing of delivery is what actually prevents stillbirth. PITCHES is not "UDCA doesn't work" — it is "UDCA does not do the one thing (prevent the composite fetal harm) we hoped it did", which is a narrower claim than either camp tends to make.

What cut-off drives delivery, and how strong is the number? Ovadia 2019 is the pivotal data: singleton stillbirth prevalence was 0.13% below 40 µmol/L, 0.28% at 40–99, and 3.44% at ≥100 µmol/L — the ≥100 stratum carrying a hazard ratio of 30.5. Moving a woman from the ≥100 band to below it changes absolute stillbirth risk from ~3.44% to ~0.13–0.28% — an absolute risk reduction of roughly 3.2–3.3 percentage points, so the number-needed-to-deliver-early to prevent one stillbirth in severe ICP is on the order of ~30 (1 ÷ 0.032 ≈ 31). For mild disease the calculation runs the other way: below 40 µmol/L the risk is already at background (~0.13%), so delivering those women early has essentially no stillbirth ARR and an NNT in the hundreds-to-thousands — pure iatrogenic prematurity for no fetal gain. The step-change at 100 is exactly why severe ICP is delivered at 35–36 weeks and milder disease is not delivered prematurely. The live debate is whether any pre-term delivery is justified below 100 µmol/L — and over-treating mild ICP is the commoner error.

AFLP, HELLP and the limits of conservatism. Unlike pre-eclampsia, where expectancy buys fetal maturity, AFLP allows no expectant window — delay kills. Modern series still report maternal mortality of a few percent and perinatal mortality well above background; the gains came from earlier recognition and prompt delivery, not from any drug. The same applies to the steroid question in HELLP: the temptation to "buy time" with dexamethasone for the platelets is unsupported (Woudstra 2010 — improved counts, unchanged clinical outcomes), and the only intervention that changes the trajectory is delivery.

Landmark trials & key evidence

The pattern across these studies is consistent: bile-acid level drives the stillbirth risk and the timing decision; UDCA touches symptoms more than hard fetal outcomes.

Screening & prevention

Screening turns on who is screened and what each test characteristic delivers:

• Hepatitis B — universal antenatal HBsAg. A positive HBsAg triggers risk-stratification for vertical transmission: HBeAg status and, where available, HBV DNA quantitation. The threshold for adding maternal tenofovir prophylaxis is HBV DNA ≥200,000 IU/mL or HBeAg-positivity; everyone HBsAg-positive still needs the neonatal immunoglobulin-plus-vaccine package. The screening test (HBsAg) is highly specific but identifies carriage, not transmission risk — so a positive must be followed by the HBeAg/DNA step, not acted on alone.
• ICP "screening" is symptom-triggered, not population-wide. There is no antenatal bile-acid screen of asymptomatic women; bile acids are measured in response to pruritus of normal skin. Bile acids fluctuate and are higher non-fasting, so a single normal value in a clearly itchy woman does not exclude ICP — repeat, and grade on the peak.
• AFLP and the neonate. There is no maternal screening test for AFLP; the screening that matters is neonatal — once AFLP is diagnosed, the newborn enters fatty-acid-oxidation-defect screening (acylcarnitine profile / newborn screen) because the treatable, lethal disease is the baby's LCHAD deficiency.
• Hepatitis C and E are not part of routine SA universal antenatal screening; test on risk factors or clinical suspicion (HCV in HIV-co-infected or injecting-drug-use risk; HEV in the jaundiced, encephalopathic third-trimester woman, especially in an outbreak/endemic setting).

Long-term, postnatal & counselling

• ICP resolves — and that is diagnostic. Pruritus and biochemistry normalise within days to a few weeks postpartum; bile acids and LFTs should be confirmed normal at a postnatal check (commonly ~4–6 weeks). Persistently abnormal LFTs after delivery mean the diagnosis was wrong — investigate for chronic liver disease (PBC, autoimmune hepatitis, viral hepatitis, ABCB4-related cholestasis). Counsel a high recurrence in future pregnancies and a recognised association with combined-oral-contraceptive- and HRT-related cholestasis (oestrogen is the common trigger) — progestogen-only or non-hormonal contraception is the safer default, and a woman who develops cholestasis on the COC should stop it.
• AFLP carries a real recurrence risk, much higher where a fetal LCHAD/MTP defect is confirmed; future pregnancies need early MFM involvement, and the index child needs lifelong metabolic follow-up. Counsel that the maternal liver recovers fully in survivors — there is no chronic maternal liver sequela — but the genetic/metabolic implications for the child and for future pregnancies are permanent.
• HELLP shares pre-eclampsia's long arc: recurrence of a hypertensive disorder in a future pregnancy, eligibility for low-dose aspirin prophylaxis next time, and the doubled long-term maternal cardiovascular risk that makes pre-eclampsia/HELLP a cardiovascular risk marker warranting primary-care follow-up (see pre-eclampsia-early-onset-severe).
• Hepatitis B is a chronic maternal diagnosis: link the woman to hepatology, vaccinate household contacts, and ensure the infant's vaccine schedule and post-vaccination serology are completed.

Exam traps & red flags

• Treating the bile-acid number with UDCA instead of timing delivery. UDCA controls itch; delivery timing (driven by peak bile acid) addresses stillbirth. Promising parents UDCA "prevents stillbirth" is post-PITCHES wrong.
• Over-delivering mild ICP. Below 100 µmol/L stillbirth is near-background; iatrogenic prematurity from delivering every itchy woman at 37 weeks is real harm (the NNT to prevent a stillbirth there runs into the hundreds-plus). Stratify by the peak value, not a single reading.
• Trusting the CTG in ICP. A reassuring CTG yesterday does not prevent the sudden arrhythmic stillbirth of severe ICP — antenatal surveillance does not predict it, timed delivery is the intervention.
• Missing AFLP behind "hyperemesis" or "gastroenteritis." A third-trimester woman vomiting with hypoglycaemia, coagulopathy and hyperuricaemia is AFLP until excluded — check glucose and clotting, do not just give antiemetics. There is no conservative option: stabilise and deliver.
• Forgetting the neonate in AFLP — screen for LCHAD / fatty-acid-oxidation defects; missing it risks the child's life.
• HELLP managed with steroids "to fix the platelets." Corticosteroids raise the count but not maternal/perinatal outcomes (Woudstra 2010) — they are for fetal lung maturity only. Treatment is magnesium, BP control and delivery.
• Mistaking TTP/aHUS for HELLP. Microangiopathy with disproportionate renal failure or neurology, or one that worsens after delivery, is TTP/aHUS — it needs plasma exchange or complement blockade, not delivery, and getting this wrong is fatal.
• Rupturing a subcapsular liver haematoma. Sudden RUQ/shoulder-tip pain with shock in HELLP is a hepatic capsular bleed — image and involve surgery/IR; do not blunder into the capsule.
• Under-rating hepatitis E. In an endemic/outbreak setting, fulminant third-trimester HEV (genotype 1/2) has a maternal case fatality up to 20–25% — escalate to high-care early; it is not "just viral hepatitis."
• Dropping hepatitis B PVT. A high-viral-load (≥200,000 IU/mL) or HBeAg-positive mother needs third-trimester tenofovir and the neonate needs immunoglobulin plus birth-dose vaccine within 12 hours — omitting either is a vertical-transmission failure.
• Calling ICP a chronic liver disease. It resolves within days–weeks postpartum; persistently abnormal LFTs afterwards mean the wrong diagnosis — investigate for underlying liver disease. Recurrence is high — counsel it, and steer away from oestrogen-containing contraception.

Worked viva — how to structure the answer

Examiners give a stem like "a 31-year-old, 35 weeks, two-day history of vomiting and malaise, now drowsy; glucose 2.1 mmol/L, platelets 95, INR 1.9, urate markedly raised, BP 138/88." A high-scoring answer runs:

1. Frame it as a sick liver, not an itchy one — "This is a third-trimester woman who is unwell — encephalopathy, hypoglycaemia and coagulopathy with only mild hypertension point to acute fatty liver of pregnancy rather than ICP or pure HELLP; the Swansea features are accumulating."
2. Resuscitate before you deliver — correct the hypoglycaemia (10% dextrose), correct the coagulopathy (FFP, cryoprecipitate, vitamin K), ICU/anaesthetic/hepatology involvement, and treat as a maternal emergency.
3. Commit to delivery — "There is no expectant window in AFLP; once stabilised I would expedite delivery regardless of gestation — the liver only recovers after the placenta is out — by the quickest safe route the coagulopathy allows, with blood products on standby."
4. Justify from evidence and contrast the differential — name how you separated AFLP (hypoglycaemia, ammonia, coagulopathy) from HELLP (haemolysis, near-normal glucose) and from TTP/aHUS (disproportionate renal failure, worsens after delivery, needs plasma exchange); cite that steroids don't fix HELLP (Woudstra) and that delivery is the only liver-changing step.
5. Anticipate complications — postpartum hepatic and renal deterioration, hypoglycaemia, DIC, PPH on a coagulopathic background, fulminant failure needing transplant referral.
6. Close the loop — neonatal LCHAD/fatty-acid-oxidation screening, recurrence counselling, and early MFM input for the next pregnancy.

For the itchy-but-well stem instead ("34 weeks, intense palmar itch, normal-looking skin, bile acids 140 µmol/L"): frame as severe ICP, grade on the peak (≥100), give UDCA for symptoms while being honest it does not prevent stillbirth, and plan delivery at 35–36 weeks with continuous intrapartum EFM — the timing, not the drug, is the stillbirth-preventing decision.

Evidence anchors

• Ovadia C, et al. Bile acids and adverse perinatal outcomes in ICP — IPD meta-analysis. Lancet 2019
• Chappell LC, et al. PITCHES — ursodeoxycholic acid vs placebo in ICP. Lancet 2019
• Ovadia C, et al. Ursodeoxycholic acid in ICP — systematic review and IPD meta-analysis. Lancet Gastroenterol Hepatol 2021
• Girling J, et al. Intrahepatic cholestasis of pregnancy (RCOG Green-top Guideline No. 43). BJOG 2022
• Society for Maternal-Fetal Medicine (SMFM) Consult Series #53 — Intrahepatic cholestasis of pregnancy. Am J Obstet Gynecol 2021
• Ch'ng CL, et al. Prospective study of liver dysfunction in pregnancy in Southwest Wales (Swansea criteria). Gut 2002
• Woudstra DM, et al. Corticosteroids for HELLP syndrome in pregnancy. Cochrane Database Syst Rev 2010
• WHO — Hepatitis E fact sheet
• National Department of Health, National Integrated Maternal and Perinatal Care Guideline (2024) and Guidelines for Maternity Care in South Africa (knowledgehub.health.gov.za) — SA antenatal hepatitis B screening, PVT and referral context