Liver biopsy: blind or guided?

BMJ 1994; 309 doi: http://dx.doi.org/10.1136/bmj.309.6967.1455 (Published 03 December 1994) Cite this as: BMJ 1994;309:1455
  1. Guy Vautier,
  2. Brian Scott,
  3. David Jenkins

    Benefits of guided biopsy are clear only for focal lesions

    Despite advances in imaging techniques and serological investigations percutaneous needle biopsy of the liver is still important in accurately diagnosing hepatic disease. The basic technique, described by Sherlock, has changed little over the past 50 years.1 It is simple, cheap, and relatively safe and can be carried out at the bedside. In the past few years, however, ultrasonography has been increasingly used to guide the biopsy needle. A recent large survey of consultant gastroenterologists showed that 1 in 8 always used ultrasonography guidance for biopsies.2 Some consultants now believe that ultrasonographically guided biopsies are so much safer that blind biopsy can no longer be defended. Before this policy is adopted uncritically, however, it is important to examine the current evidence concerning safety, diagnostic yield, and cost.

    Percutaneous liver biopsy has a mortality of 0.01%-0.1%.3 4 Death is usually due to bleeding or to biliary peritonitis as a result of puncture from the gall bladder. The incidence of bleeding is probably proportional to the incidence of formation of haematomas, which is not affected by the use of ultrasonographic guidance.5 Although, intuitively, guided biopsy might be expected to reduce the risk of puncturing the gall bladder, no randomised controlled trial has been large enough to show reduced mortality with ultrasonography. Identifying deaths related to procedures is not easy. There is an overall mortality of 19% among patients within three months of biopsy, but most deaths are due to underlying disease.6 Retrospective reviews may therefore fail to give a true indication of the risks of the procedure.

    The 1991 national audit of liver biopsies reviewed 1504 biopsies, of which a third were guided by ultrasonography.6 Two deaths definitely related to the procedure occurred, one each from bleeding and from biliary peritonitis. Both biopsies were carried out without ultrasonographic guidance. Surprisingly, postmortem examinations were not performed, but the second death might have been avoided had ultrasonography been used. Data were also collected on pain and bleeding after the biopsy. Pain was experienced by 25% of the patients who had nonguided biopsies and 22% of the patients who had guided biopsies. Serious bleeding occurred in 1.6% of non-guided cases and 2.5% of guided biopsies. These differences were not significant.

    The largest single controlled trial is that of Papini and colleagues, who randomised 240 patients to guided or non-guided biopsy.7 They reported one complication (bleeding into the abdominal cavity) in the group that had guided biopsy and seven in the group that had non-guided biopsy. Four of the complications in the group that had non-guided biopsy, however, were asymptomatic and were disclosed only by follow up ultrasonography. The other problems were transient early hypotension in two patients and an ileus that spontaneously resolved in another.

    Diagnostic yield was also assessed in the National Audit in 1991. Where ultrasonography before biopsy showed one or more focal lesions non-guided biopsy was successful in confirming the final diagnosis in only one third of patients, whereas guided biopsy confirmed the diagnosis in nearly two thirds of patients. The audit also suggested that if the clinical diagnosis before the biopsy was of cancer there was a greater chance of verifying this with a guided biopsy even if there was no focal lesion. For non-malignant diffuse disease there was no difference between the two procedures in the ability to confirm diagnoses.

    Cost and convenience must also be considered. Guided biopsies need greater resources, both of equipment and of trained staff. The biopsy is usually done in a radiology department, which means that the patient would be waiting to return to a ward without being observed during the time when at least 60% of complications occur.3 Doctors in some centres identify the optimal site of puncture by ultrasonography but perform the biopsy in the ward.

    What recommendations can be made? In patients with diffuse non-malignant disease guided biopsy has no diagnostic advantage and there is no firm evidence that the procedure is safer. When malignancy is suspected before the biopsy is performed a guided biopsy should be considered. When a focal lesion has already been shown the biopsy should be guided. The ideal biopsy may be one that is performed in the ward by the gastroenterologist using ultrasonographic guidance. For most patients this is currently not an option owing to the lack of ultrasound machines and trained clinicians.

    To establish firmer guidelines a randomised controlled trial of guided versus non-guided biopsy in patients with diffuse disease might be considered. Since the mortality is so low, however, a large number of biopsies—we estimate 10 000—would be needed to give sufficient statistical power. This is probably not feasible. Our recommended alternative is a national scheme for reporting mortality and morbidity after liver biopsy, perhaps as part of the national confidential inquiry into perioperative deaths.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.