Gut 60:138-139 doi:10.1136/gut.2009.198465
  • PostScript
  • Letter

How accurate are the Swansea criteria to diagnose acute fatty liver of pregnancy in predicting hepatic microvesicular steatosis?

  1. G Chandy1
  1. 1Department of Gastrointestinal Sciences, Christian Medical College, Vellore, Tamil Nadu, India
  2. 2Department of Pathology, Christian Medical College, Vellore, Tamil Nadu, India
  1. Correspondence to Dr C E Eapen, Department of Gastrointestinal Sciences, Christian Medical College, Vellore 632004, Tamil Nadu, India

We noted that in two recent reports in Gut, none of 62 patients diagnosed by the Swansea criteria to have acute fatty liver of pregnancy (AFLP) underwent liver biopsy.1 2 We retrospectively assessed the accuracy of the Swansea criteria to predict hepatic microvesicular steatosis in 34 patients with suspected pregnancy-related liver disease who underwent liver biopsy at our centre between 1998 and 2006. These patients tested negative for other causes of acute liver dysfunction such as hepatitis viruses (hepatitis B virus surface antigen (HBsAg), hepatitis C virus (HCV) antibody, immunoglobulin (Ig) M hepatitis A virus (HAV) antibody, IgM hepatitis E virus (HEV) antibody), malarial parasite and sepsis (blood culture). No patient gave a history of ingestion of a potentially hepatotoxic drug. We excluded 10 patients (details in figure 1).

Figure 1

Flowchart of patients with suspected pregnancy-related liver disease who underwent liver biopsy. *Refers to diffuse/perivenular hepatic microvesicular steatosis.

The remaining 24 patients included in this study were at 36 (21–40) weeks gestation (median (range)), 23 (17–29) years old and 71% were primigravida. The interval from the first symptom to presentation to our centre was 5 (1–14) days.

Abnormal variables in Swansea criteria for AFLP at presentation in the 24 study patients were: vomiting (11/21 patients), abdominal pain (3/14), polydipsia/polyuria (1/1), encephalopathy (9/24), hyperbilirubinaemia (24/24), hypoglycaemia (8/24), hyperuricaemia (8/10), leucocytosis (20/23), ascites/bright liver on ultrasonogram (16/22), elevated transaminases (23/24), hyperammonaemia (2/2), renal impairment (17/24), coagulopathy (23/24) and hepatic microvesicular steatosis (17/24). Some variables were not recorded/not tested in all 24 patients (eg, vomiting recorded in only 21; uric acid tested in only 10). Baseline laboratory results in study patients were: serum total bilirubin, 12.7±6.2 mg/dl (mean±SD); alanine aminotransferase (ALT), 119±6.2 IU/l; prothrombin time, 39±30 s; serum creatinine, 1.7±0.9 mg/dl; and MELD (Model for End-Stage Liver Disease) score, 30 (range 13–46).

Liver biopsy was done either immediately postmortem (5/24) or postnatally via the transjugular route (19/24). Biopsies were fixed in formalin and routinely stained for H&E and other special stains. Liver histology, reviewed by a single hepatopathologist, showed diffuse microvesicular steatosis (15 patients), perivenular microvesicular steatosis (2), hepatocanalicular cholestasis (4), marked centrizonal congestion (1), bridging necrosis (1) and non-specific changes (1). The interval between delivery and liver biopsy was <3 days (8 patients), 3–7 days (13 patients) and 8–11 days (3 patients). There was a trend, albeit non-significant, to find diffuse/perivenular microvesicular steatosis in patients who underwent liver biopsy earlier after delivery (seen in 10/12 patients biopsied ≤4 days after delivery and in 7/12 patients biopsied >4 days after delivery (p=0.19)). Three patients who did not have hepatic microvesicular steatosis despite fulfilling Swansea criteria underwent liver biopsy 2, 5 and 8 days after delivery.

The sensitivity and specificity of Swansea criteria vis-à-vis diffuse or perivenular microvesicular steatosis was 100% (95% CI 77% to 100%) and 57% (95% CI 20% to 88%), with positive and negative predictive value of 85% and 100%, respectively. Of four patients not fulfilling the Swansea criteria, none had diffuse or perivenular microvesicular steatosis (figure 1). A median of 8 (range: 7–11) variables were abnormal in 20 patients fulfilling the Swansea criteria. All 20 patients who fulfilled the Swansea criteria (ie, had ≥6 abnormal variables), met these criteria, despite excluding histology. We found that by fulfilling the Swansea criteria, no patient with hepatic microvesicular steatosis would have been missed, even if liver biopsy was not done (negative predictive value of 100%). Only four patients did not fulfil the Swansea criteria—this is a limitation of our report.

Based on the described criteria,1 some of the 20 patients who met the Swansea criteria for AFLP also fulfilled the criteria for HELLP (haemolysis, elevated liver enzymes, low platelets) syndrome (8 patients), partial HELLP syndrome (5 patients) and pre-eclamptic liver dysfunction (2 patients).

The urgent need in a patient with suspected AFLP is early diagnosis, since early termination of pregnancy dramatically improves maternal survival. In this clinical scenario, the Swansea criteria (without liver biopsy) are a good screening tool with 100% negative predictive value for hepatic microvesicular steatosis, thus obviating the need for liver biopsy in clinical management. Studies with larger numbers of patients are warranted to validate our findings.


We wish to thank Professor Elwyn Elias, Liver Unit, Queen Elizabeth Hospital, Edgbaston, Birmingham, UK for his critical review of this manuscript.


  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.


Relevant Articles