Article Text
PDF
Abstract
Background Most medical liver biopsies in the UK are now taken in radiology departments using 18 g biopsy needles. Subjectively, the resulting biopsies are narrow and fragile.
Aim To compare the quality of liver biopsy tissue sections obtained from 16 and 18 g biopsy needles.
Method Fifty consecutive routine medical liver biopsies obtained with 16 and 18 g needles, processed identically in the same laboratory, were measured using digital pathology software. We recorded their fragmentation, length, width, area and number of portal tracts.
Results Biopsies obtained with 16 g needles more often resulted in an intact core in tissue sections than those with 18 g needles (71% vs 24%, p<0.001) and were significantly wider (average width of tissue 0.88 vs 0.53 mm, p<0.001). The average total area of tissue per pass was 11.38 mm2 compared with 8.34 mm2 (p<0.001). The number of complete portal tracts per length of biopsy was very variable, but double for 16 vs 18 g biopsies. Routinely taking two passes with the 18 g needle compensated for the reduced area, but the resulting liver in tissue sections was fragmented and distorted.
Conclusions Our results support the routine use of 16 g rather than 18 g biopsy needles for routine ultrasound-guided medical liver biopsies. A second pass should be considered if the first biopsy core is short, especially for investigation of disease stage.
- LIVER
- DIGITAL PATHOLOGY
- DIAGNOSIS
Statistics from Altmetric.com
Introduction
Liver biopsy has a central role in investigation and management for many patients with medical liver disease.1 ,2 Factors affecting the biopsy diagnosis include sampling error, the sample size in relation to the distribution of the pathological features in the liver, the technical quality of the slides, the experience of the pathologist and the opportunity for clinical–pathological correlation. Inherent in all liver biopsies is the need to balance the diagnostic benefit with the risks of the invasive procedure. Introduction of non-invasive tests for disease stage is reducing the requirement for biopsy purely for staging purposes.
Most biopsies for the investigation of medical liver disease are now performed in radiology departments under image guidance, using Tru-cut cutting-type biopsy gun devices, and a recent Royal College of Radiologists (RCR) audit showed that 70% of these used 18 g needles.3 There have been many studies designed to establish what should be considered an ‘adequate’ biopsy size, and there is recognition that biopsies with less than 11 portal tracts underestimate disease stage and grade in chronic hepatitis.4 ,5 Based on these, the American Association for the Study of Liver (AASLD) Position Paper recommends two passes with a 16 g needle.2 However, a systematic review found that this standard was almost never achieved in practice.6 An alternative standard of ≥6 portal tracts for diagnosis (as opposed to reproducible staging) has been proposed.7
In Leeds, because of concern over inadequate biopsy size, our routine protocol for medical liver biopsies was changed from 18 to 16 g needle size in 2005. This increase in biopsy width resulted in significantly larger tissue sections that distort and fragment less than narrower cores, and therefore are preferable for interpretation.8
The aim of this study is to quantify the improved sample yielded by the use of the wider gauge needle to support its recommendation in the Royal College of Pathologists tissue pathways for medical liver biopsies9 by comparing 18 and 16 g biopsies processed concurrently in the same laboratory.
Methods
This study compares the quality of routine tissue sections obtained from medical liver biopsies taken from adult patients under ultrasound guidance with either 16 or 18 g Tru-cut biopsy needles. The histopathology laboratory in Leeds provides the laboratory technical service for both Bradford and Leeds hospitals, providing the opportunity for our study design that controls for variables such as laboratory processing, embedding and tissue sectioning and staining.
Bradford is a large general hospital employing two hepatologists; there were 144 medical liver biopsies in 2011. Leeds is a tertiary hepatology and transplant centre, employing six adult and four paediatric hepatologists; there were 490 medical liver biopsies in 2011.
In Leeds, routine medical liver biopsies in adult patients are taken using one pass of a 16 g needle under ultrasound guidance by consultant or trainee radiologists. In Bradford, 18 g needles are used, usually with two passes. Biopsies are transferred from the needle into a container of formalin and transported to the laboratory. On receipt, the number of pieces and length of fragments are recorded, biopsies wrapped in tissue paper and placed in cassettes for processing. Biopsies are routinely sectioned at 3 microns to give three levels at H&E together with the routine panel of six special stains.
The first 50 consecutive adult medical liver biopsies (excluding transplant biopsies) from each hospital in 2011 were identified from the laboratory computer to match the study design of the RCR audit.3 The radiology records were reviewed for the number of passes, size of needle and operator (consultant or trainee). The histopathology reports were reviewed for macroscopic description of number and length of biopsy cores, age and gender of patient and main medical indication/diagnosis.
Fragmentation of the biopsy between the initial procedure and receipt in histology was recorded as the difference between the number of passes and the number of fragments in the macroscopic description of the biopsy. Fragmentation between receipt in the laboratory and microscopy was recorded as the difference between the number of fragments in the macroscopic description and number of pieces measured in image analysis. This includes real fragmentation due to breakage of cores and apparent fragmentation due to incomplete tissue sections as a result of interruption of tissue within the plane of section.
The best of the three H&E levels was selected for morphometry, which was performed by one observer (JIW). The H&E slides were scanned at 40× magnification with an Aperio AT scanner (Aperio, Vista, California, USA) to produce a whole slide image (virtual slide) with a final resolution of 0.23 microns per pixel. Using Aperio ImageScope software (V.11.1.2, Aperio, Vista, California, USA) each virtual slide was manually annotated using a mouse. Annotations were used to measure the number of tissue fragments, total length of fragments and actual area of tissue on the slide. The maximum width identified as the greatest of at least three diameters taken from what was subjectively the widest point of the biopsy (figure 1). The average biopsy width was calculated (= actual area of tissue/total length of biopsy cores). The calculated maximum possible area was derived from (total length×maximum width) of the cores (figure 2), and the extent to which the actual tissue sections fell short of this maximum (the ratio of actual area:calculated maximum area) was calculated.
Digital images of annotated virtual slides showing total length (yellow line), total measured area (area contained within green line) and maximum diameter (greatest of at least three diameters from subjectively widest point in the biopsy, at high magnification). The biopsies illustrated represent the median area of the 16 g (a) and 18 g (b) groups.
The calculated theoretical maximum area of a longitudinal section through the midpoint of the cylinder of tissue is the length×maximum width (green line). The actual area (see figure 1a) is less than the maximum possible area, depending on the irregularity and fragmentation of the biopsy, in this case, 11.62/(12.94×1.12) or 80.2%.
A different observer (IG) used the same slides to count the number of complete and incomplete portal tracts in the best H&E section.
Statistical comparisons were made using the Mann–Whitney U test and contingency table χ2 test as appropriate, using SPSS V.19 (IBM SSPS Statistics for Windows, V.19.0, IBMCorp).
Results
Of the 100 biopsies, 10 were excluded because of the needle gauge or transjugular route (2 transjugular 19 g biopsies and 8 biopsies with 18 or 14 g needles from Leeds). The remaining 49 Bradford biopsies were all obtained with an 18 g needle; 30 were taken by consultants and 19 by trainees. There were two passes in 32 cases and one pass in 17 cases. The remaining 41 Leeds biopsies were all obtained with a 16 g needle; 8 were taken by consultants and 33 by trainees. All except one had a single pass. We found no difference in biopsy length when comparing specimens obtained by consultant and trainee radiologists.
The patients were well matched for age, gender and indication for liver biopsy and frequency of cirrhosis (table 1).
Age, sex and indication for biopsy
Fragmentation of needle biopsies
The number of macroscopic pieces of biopsy core on receipt relative to the number of passes is shown in table 2. This indicates that biopsy cores were already broken when received in the laboratory in 17/49 (35%) of 18 g biopsies and 7/41 (17%) of 16 g biopsies (p=0.06). The number of microscopic pieces counted on image analysis exceeded the number of macroscopic pieces in 35/49 (71%) of 18 g biopsies and 9/41 (22%) of 16 g biopsies (p<0.001). Overall there was more than one piece of microscopic liver tissue per pass (ie, fragmentation of at least one core) in 37/49 (76%) of 18 g biopsies and 12/41 (29%) of 16 g biopsies (p<0.001), with most of the fragmentation occurring during tissue processing or sectioning for 18 g biopsies. The separate fragments seen in tissue sections may be due to either breakage of the biopsy core or to irregular calibre or failure of flat embedding of the tissue resulting in gaps in the tissue section.
Pieces of tissue per pass of the biopsy needle
Area of tissue in histology sections
Figure 1 shows the annotated virtual slide images of the biopsies with median areas from the 18 and 16 g groups. The total length, maximum width and measured area of tissue per pass of the biopsy needle are shown in table 3. The median length of the core per pass is similar in the two groups (14.10 mm for 16 g and 14.61 mm for 18 g, p=0.25). The median value for maximum width measured was 0.74 mm (range 0.53–0.93 mm) for the 18 g biopsies and 1.04 mm (range 0.85–1.20 mm) for the 16 g biopsies (p<0.001). The median value for the measured area for the 16 g needle was 11.38 mm2 (range 5.12–17.42 mm2). The median area for cases with one pass of the 18 g needle was 8.34 mm2, but for cases with two passes, the median area are was 15.19 mm2, exceeding that of the 16 g needle (figure 1).
Morphometric measurements of liver biopsies: length, maximum width, area and calculated values for average width, and % of theoretical biopsy area achieved in practice
The average biopsy width in tissue sections was calculated from measured area/length. The median width for 18 g biopsies (0.53 mm) is significantly less than for 16 g biopsies (0.87 mm, p<0.001).
Percentage of possible maximum area included in tissue sections
A perfect longitudinal section through the widest part of the biopsy cylinder would have a calculated theoretical maximum area of the length×maximum width (figure 2). The percentage of this achieved in practice (measured area/calculated theoretical maximum area) is also shown in table 3 and was significantly higher for the 16 g biopsies (median 82%) than for the 18 g biopsies (median 73%).
Number of portal tracts
Five biopsies showed late-stage cirrhosis, and it was not possible to count their portal tracts. The number of complete and incomplete portal tracts in the remaining 85 biopsies and per centimetre length of biopsy is shown in table 4. There is a wide variation in the number of portal tracts per length of biopsy, and so the portal tract count cannot be predicted by the biopsy length. The wider biopsies contain nearly twice as many complete portal tracts per centimetre of biopsy.
Portal tracts per length of specimen (data shown per specimen, ie, includes total tissue per case whether one or two passes)
Using the criterion of >10 complete portal tracts per section, only 5/46 (10.8%) of our routine 18 g biopsies (most of which had two passes) and 7/39 (17.1%) of 16 g biopsies would be classed as adequate. To reliably obtain a sample with over 10 complete portal tracts, multiple passes would be needed with either needle.
Based on the number of portal tracts per 10 mm, we have calculated that the number of biopsies that would have met the criterion of ≥6 complete portal tracts in a single 15 mm core is 7/46 (17.9%) for 18 g biopsies and 28/39 (74%) for 16 g biopsies. For two passes, achieving 30 mm of tissue, this would be 70% and 95%, respectively. However, if all portal tracts (complete and incomplete) are included, there would have been ≥6 tracts in 93% of 18 g biopsies and 100% of 16 g biopsies within one 15 mm core biopsy.
Discussion
The purpose of the medical liver biopsy is to make or contribute to the diagnosis when this is not sufficiently clear from non-invasive liver tests. By providing a direct visualisation of the disease process, the pathologist can evaluate how the individual non-invasive test results—surrogate markers for hepatocyte and bile duct injury, inflammation and fibrosis—are integrated within the individual patient's liver disease. The value of the biopsy is limited by sampling error, artefact and observer variation. Sampling error depends on the scale of the abnormality in relation to the size of the biopsy sample—a disorder affecting most hepatocytes will be detected by even the smallest biopsy as long as it contains liver tissue. Larger-scale abnormalities, notably the tissue architecture, require larger samples, and width is more important than length.4 The diagnosis of macronodular cirrhosis with nodules >3 mm diameter will always be difficult to confirm by needle biopsy with width <1 mm. We found that the average width of biopsy sections was only 0.53 mm using a 18 g needle and 0.88 mm for a 16 g needle.
Ours is the first study to measure the tissue area of liver biopsies and has shown that in practice tissue sections contain less than 100% of the predicted area of biopsy length×internal calibre of biopsy needle (73% vs 82% of predicted area for 18 and 16 g needles, respectively). Real biopsies are fragile, fragmented and distorted, especially when the needle is narrow. Most biopsies in the UK are now performed in radiology departments under image guidance using Tru-cut cutting-type biopsy gun devices.3 These deliver a biopsy core that is generally shorter and narrower than the Menghini type of aspiration needles that were in general use when liver biopsies were taken by physicians on the ward. In the triangle of physician requesting, radiologist performing and histopathologist reporting, this reduction in biopsy size may not have been appreciated. Radiologists are not aware of this problem—in their UK audit, a biopsy was considered adequate as long as a histopathology report could be provided, which was the case in 98.3%.3 The biopsy is rarely so small as to preclude any assessment; however, a larger, intact core is clearly preferable to small fragments. Based on our findings, by using a 16 g rather than 18 g needle for routine image-guided medical liver biopsies, the proportion of biopsies seen intact in tissue sections can increase from 24% to 71%. The finding that a 16 g needle provides a substantially larger sample than an 18 g needle, far greater than the ratio of 18:16 would suggest, is initially surprising. The ‘gauge’ system initially developed for the diameter of iron wires has been adapted for stainless steel needle tubing by the International Organization for Standardization.10 The minimum inside diameter is 40% wider for a 16 g needle (1.10 mm) than an 18 g needle (0.79 mm). In tissue sections, the D-shaped cross section of the Tru-cut biopsy core, which cannot be appreciated during embedding, accounts for further narrowing.
There are potential confounding factors in this study. The origins of the biopsies (Leeds or Bradford) were known to the assessors from the accession number. The excluded 18 g biopsies from Leeds were within the range of morphometric results for 18 g biopsies from Bradford (data not shown). Biopsy size and fragmentation did not show an association with experience of operator (consultant or trainee radiologist) or with the main diagnosis. Laboratory factors were controlled for in the study of the design. The significant differences in tissue area, and hence portal tracts, are accounted for by the wider internal diameter of the biopsy needle, while length of biopsy and fragmentation could in theory be affected by the operator or nature of disease, the width of the biopsy specimen could not.
The criteria for an adequate biopsy are affected by the purpose of the biopsy. Previously research has focused on the reproducible assessment of fibrosis stage, largely in the context of antiviral therapy for chronic viral hepatitis.6 The evidence-based recommendation for biopsy size was rarely attained in practice (only 2/147 studies included the relevant information to assess whether this was achieved).6 However, the commonest indication for medical biopsy in the UK is the investigation of abnormal liver function tests (43% of RCR audit compared with 24% for viral hepatitis),3 and alternative non-invasive tests for disease stage in viral hepatitis are reducing this indication for liver biopsy.11 Recommendations should therefore focus on safely obtaining a biopsy core adequate for routine diagnosis. A pragmatic standard of six portal tracts was discussed in a commentary by Scheuer.7 The number of portal tracts per length of biopsy core is widely variable12 and only becomes apparent once the biopsy slides are produced. What length of biopsy core does this translate to? The number of complete portal tracts is doubled in a 16 g needle compared with an 18 g needle. Based on our data and that of Chan et al,11 one good 16 g core resulting in a tissue section of at least 15 mm would contain at least six portal tracts in 95% biopsies, including incomplete tracts. A second pass of 15 mm (the AASLD recommendation)2 would contain ≥6 complete portal tracts in 95% biopsies. The recent RCR audit of UK liver biopsies showed that severe complications are rare for medical liver biopsies (6/2257, 0.26%) compared with targeted tumour biopsies (16/1222, 1.3%, p<0.0008). There was no additional risk for 16 g compared with 18 g needles; the needle gauge was recorded in 16/22 cases with severe complications, and all were 18 g.13
Conclusion
The prevalence of chronic liver disease is rising in the UK. Non-invasive tests for disease stage are increasingly available, and the role of the biopsy is anticipated to shift towards the diagnosis of the cause of liver disease where this is not apparent from non-invasive liver screen tests. The emphasis is on obtaining a sufficient biopsy sample safely. Based on our experience, a single pass of a 16 g needle, consider a second pass if the biopsy core is <15 mm long, and a meticulous laboratory technique, should be the standard. This would significantly improve the diagnostic quality of liver biopsy tissue sections in routine practice.
Take home messages
-
Most liver biopsies for the diagnosis of medical liver disease in the UK are now performed under ultrasound guidance, and there is no formal guidance for radiologists regarding specimen adequacy.
-
16 g biopsy needles provide a biopsy sample that generates tissue sections superior to the more commonly used 18 g needles.
-
One full core with a 16 g needle provides a specimen sufficient for routine diagnosis in most cases, although they may be suboptimal for staging of fibrosis.
References
Footnotes
-
Contributors JIW devised the study in consultation with the other named authors and performed the image analysis. TP collected patient information and contributed to the writing of the paper. IG performed the portal tract counts. MS and RK collected radiology data on the biopsy procedures. AW and DT provided computing expertise and support for image analysis. DT provided statistical advice and performed the statistical analysis. All authors approved the final version of the manuscript.
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.