Pre-liver transplant biopsy in hepatocellular carcinoma: a potential criterion for exclusion from transplantation?

Abstract

Background In cirrhotic patients with hepatocellular carcinoma (HCC), poor differentiation in pre-liver transplantation (LT) biopsy of the largest tumour is used as a criterion for exclusion from LT in some centres. The potential role of pre-LT biopsy at one centre was explored.

Methods A prospective database of patients undergoing orthotopic LT for radiologically diagnosed HCC at St James's University Hospital, Leeds during 2006–2011 was analysed.

Results A total of 60 predominantly male (85.0%) patients with viral hepatitis were identified. There were discrepancies between radiological and histopathological findings with respect to the number of tumours identified (in 27 patients, 45.0%) and their size (in 63 tumours, 64.3%). In four (6.7%) patients, the largest lesion, which would theoretically have been targeted for biopsy, was not the largest in the explant. Nine (31.0%) patients with multifocal HCC had tumours of differing grades. In two (6.9%) patients, the largest tumour was well differentiated, but smaller tumours in the explant were poorly differentiated. In one patient, the largest lesion was benign and smaller invasive tumours were confirmed histologically.

Conclusions The need to optimize selection for LT in HCC remains. In the present series, the largest tumour was not always representative of overall tumour burden or biological aggression and its potential use to exclude patients from LT is questionable.

Introduction

Liver cancer is the sixth most common cancer worldwide and remains the third most common cause of cancer-related death.¹ Hepatocellular carcinoma (HCC) accounts for 90% of primary hepatic malignancies and its incidence in the UK has doubled over the last 30 years, with the majority of HCCs arising on a background of chronic liver disease.^2,3 Liver resection is effective in patients with early cirrhosis and preserved liver function, but liver transplantation (LT) remains the treatment of choice in patients with more severe underlying cirrhosis, portal hypertension or multifocal disease.^4–7 In these patients the introduction of selection criteria, such as those generated in Milan in 1996, aimed to optimize individual patient outcomes and rationalize the use of limited organ resources. This resulted in dramatic improvements in outcomes.⁸ Today, survival rates in patients transplanted within and indeed beyond these criteria now match those in patients undergoing transplantation for cirrhosis alone.^8–12 Despite this, although HCC now accounts for 25% of all indications for LT in Europe, only 5% of patients with HCC actually undergo LT.^13,14 The increasing incidence of HCC, ever-improving management approaches for underlying causative liver diseases, shortage of cadaveric donor organs and the advent of live-donor LT have led to calls for the expansion of selection criteria.¹⁵ Conversely, given the limited supply of organs, others have cautioned that such an expansion should not be to the detriment of non-HCC recipients who remain on waiting lists.¹⁶ There is a need to reassess existing criteria and identify the patients who are likely to benefit most from receiving a transplant.

Despite significant advances in non-invasive imaging techniques, discrepancy between preoperative radiological assessment of lesions and explant pathological examination remains. Misdiagnosis still occurs in up to 20% of patients and staging errors in 20–30% highlight the challenges of accurately characterizing lesions pre-LT.^15,17

The role of preoperative biopsy in HCC remains a source of controversy in the transplant community. In conventional oncological practice, biopsies provide histological confirmation of malignancy and are requisite before starting treatment. Increasingly, in the era of molecular diagnostics this can inform and personalize treatment and prognostication. Conversely, in HCC the use of pretreatment biopsy is reserved for a minority of complex clinical cases or diagnostic dilemmas.¹⁸ However, some specialist centres currently advocate the use of pre-LT biopsy as part of a staging and selection process. Once HCC has been diagnosed, based on established criteria on cross-sectional imaging with or without elevation of alpha-fetoprotein (AFP), the patient undergoes a biopsy.^19,20 Findings of high-grade or poor differentiation in a single HCC, or the largest tumour in multifocal disease, are regarded as markers of aggressive underlying tumour biology and poor prognosis and are used to exclude patients from transplantation and consign them to non-curative treatments. It is proposed that tumour grade on pre-LT biopsy predicts the biological behaviour of tumours, adding significant prognostic power to static morphological selection criteria.

Biopsy of the largest tumour may provide evidence of aggressive tumour biology. Indeed, studies have confirmed that tumour size is positively correlated with vascular invasion, recurrence and survival.¹⁸ Therefore, the dominant tumour in multifocal disease may represent an accessible and potentially representative target. However, if a biopsy fails to demonstrate poor differentiation, the possibility that other areas of the sampled lesion and smaller synchronous tumours may show poor differentiation remains. The undetected presence of poor differentiation may have significant prognostic consequences.

This study was conducted using a prospectively maintained database at a tertiary referral centre in the UK to explore the potential role of pre-LT biopsy as a selection or exclusion criterion for LT.

Materials and methods

A retrospective analysis of a prospectively maintained departmental database of all patients with a pre-LT radiological diagnosis of HCC undergoing LT at St James's University Hospital, Leeds, UK, was conducted. Patients with a radiological diagnosis of HCC at the time of listing and diagnosed with end-stage liver disease whilst on the waiting list were included. Consecutive patients transplanted during 2006–2011 were analysed. The database was cross-referenced with UK Transplant Service (UKTS) records to ensure completeness. Demographic data, selection criteria parameters and clinical factors including aetiology, timing of imaging and treatment were recorded. One patient was excluded because fibrolamellar HCC was diagnosed on histological examination of the explant.

Pre-LT diagnoses of HCC were made in accordance with the European Association for the Study of the Liver (EASL) and the American Association for the Study of Liver Diseases (AASLD) guidelines. All patients underwent multidisciplinary team assessment and standardized diagnostic dual contrast-enhanced magnetic resonance imaging (DCE-MRI) with gadolinium and superparamagnetic iron oxide, performed and reported by specialist hepaticopancreaticobiliary (HPB) radiologists.^21–24 If subsequent DCE-MRI scans were undertaken following diagnosis, the scan closest to the LT was used to stage the disease. During 2006–2008, the unit's selection policy was in accordance with the Milan Criteria; from 2008 onwards, the unit adopted the UK HCC transplantation criteria.^8,25

Routine histopathological examination of all explanted livers was undertaken and the reports were correlated with DCE-MRI assessments. Lesions on DCE-MRI were identified on the basis of their size and location in the explant. Histopathological examination and reporting were undertaken by two specialist HPB pathologists. Explanted livers were sliced at a thickness of 5 mm and subjected to visual inspection. Areas identified as dysplastic or neoplastic by DCE-MRI, large nodules and samples representative of diffuse liver disease were identified at macroscopic examination and submitted for histological evaluation. Tumours were extensively sampled in order to obtain representative samples for grading. Histopathological reports included data on number of tumours, maximum diameter of all tumours measured on the gross specimen, distribution, macrovascular or microvascular invasion and histological grade based on the degree of tumour differentiation.²⁶ Total tumour volume was calculated as the sum of individual tumour volumes, which is calculated as 4/3 × π × r³. At least one block was sampled from each tumour and more blocks were sampled from larger tumours. Tumour grade was determined by the poorest degree of differentiation found within a lesion. Tumours were staged according to the Union for International Cancer Control (UICC)/American Joint Committee on Cancer (AJCC) staging system for HCC.²⁷

Accurate radiological staging was defined as agreement with histopathological examination on tumour number and size (cumulative tumour diameter within 5 mm). Under-staging was defined as a higher tumour number or size (>5 mm larger in cumulative diameter), whereas over-staging was defined as a lower tumour number or size (>5 mm smaller in cumulative diameter). The interval between index diagnostic imaging and transplantation was also recorded in order to interrogate the potential contribution of disease progression prior to LT.

Once listed for LT, 18 (30.0%) patients received additional treatment in the form of transarterial chemoembolization and five (8.3%) patients underwent radiofrequency ablation (RFA) according to standard procedures within the department to restrict tumour progression. One patient underwent a pre-LT biopsy.

Immunosuppressive regimens after transplantation were prescribed in accordance with departmental protocols (either cyclosporin or tacrolimus with azathioprine or mycophenolate). Patients did not receive routine adjuvant chemotherapy after transplantation and were followed up regularly for tumour recurrence by computed tomography (CT) scanning. Patients at risk for hepatitis B virus recurrence were treated with HBIg and oral antiviral agents. The median length of follow-up was 24 months (range: 6–65 months).

Endpoints

Tumour recurrence was identified by regular CT scanning. Time to recurrence from diagnostic radiological investigation was recorded in months, along with the date and cause of death. All-cause mortality is reported because it is the most clinically relevant outcome and avoids underestimation of the impact of HCC. For patients still alive, follow-up ended at the last clinic appointment (censored).

Statistics

All statistical analyses were performed using spss Version 18.0 (SPSS, Inc., Chicago, IL, USA). Student's t-test, Mann–Whitney U-test and related-samples Wilcoxon signed rank tests were used as appropriate. The Kaplan–Meier technique was used to calculate survival. Log-rank (Mantel–Cox) and Cox proportional hazards models were used for univariate and multivariate analyses, respectively. Variables with P-values of <0.2 were entered into the model for multivariate analysis. A P-value of <0.05 was considered to indicate statistical significance.

Results

Patient characteristics and tumour aetiology are shown in Table 1, demonstrating a representative UK cohort with a male preponderance and predominantly viral underlying aetiology.

Table 1.

Clinicopathological characteristics of the study cohort

Characteristic	Demographic data	Radiological data	Histopathological data
Age, years, median (range)	58 (27–73)
Gender, n (%)
Male	51 (85.0%)
Female	9 (15.0%)
Aetiology of cirrhosis, n (%)
Hepatitis C virus	34 (56.7%)
Hepatitis B virus	2 (3.3%)
Alcoholic liver disease	11 (18.3%)
Cryptogenic	2 (3.3%)
Haemochromatosis	4 (6.7%)
Others	7 (11.7%)
MELD score, median (range)	17.5 (7–29)
AFP, ng/ml, mean (range)	169.64 (27–4585.3)
Time to transplantation, days, median (range)	52.5 (5–339)
Preoperative bridging treatments, n (%)
Chemoembolization	18 (30.0%)
Radiofrequency ablation	5 (8.3%)
Milan Criteria, in/out		49/11	44/16
UCSF Criteria, in/out		58/2	49/11
Total tumour volumea, n (%)
<115 cm3		42 (70.0%)
>115 cm3		18 (30.0%)
Tumour number, n (%)
1		34 (56.7%)	31 (51.6%)
2		15 (25.0%)	13 (21.7%)
3		5 (8.3%)	6 (10.0%)
>3		6 (10.0%)	10 (16.7%)
Tumour size, n (%)		Range: 0.5–6 cm	Range: 0.3–5.5 cm
<1 cm		33 (29.5%)	32 (24.8%)
1–3 cm		62 (55.4%)	68 (52.7%)
3–5 cm		14 (12.5%)	26 (20.2%)
>5 cm		3 (2.6%)	3 (2.3%)
Tumour distribution, n (%)
Unilobar		54 (90.0%)	54 (90.0%)
Bilobar		6 (10.0%)	6 (10.0%)
Vascular invasion, n (%)
Macroscopic		0	1 (1.7%)
Microscopic		N/A	11 (18.3%)
None		60 (100%)	48 (80.0%)
TNM stagingb, n (%)
I		N/A	29 (49.2%)
II		N/A	26 (44.1%)
IIIa		N/A	4 (6.7%)
IIIb		N/A	0
IIIc		N/A	0
IV		N/A	0
Differentiation, n (%)
Good		N/A	39 (30.2%)
Moderate		N/A	66 (51.2%)
Poor		N/A	24 (18.6%)

^aCalculated as the sum of the volume of all tumours [(4/3) × π × r³ where r is the maximum radius of each tumour in cm].

^bUnion for International Cancer Control/American Joint Committee on Cancer staging system for hepatocellular carcinoma.

MELD, Model for End-stage Liver Disease; AFP, alpha-fetoprotein; UCSF Criteria, University of California San Francisco Criteria (single tumour < 6.5 cm or up to three tumours with a largest lesion of ≤4.5 cm and total tumour diameter of ≤8.0 cm); TNM, tumour–node–metastasis; N/A, not applicable.

Radiological assessment

The median time between preoperative DCE-MRI and LT was 52.5 days (range: 5–339 days). Based on preoperative radiological assessment, 34 (56.7%) patients had a single HCC and 26 (43.3%) had multifocal disease. The median number of tumours was one (range: 1–6) and the median tumour size was 16 mm (range: 5–60 mm). The majority (98.4%) of tumours measured <50 mm. The cumulative tumour diameter was <80 mm in the majority of patients (n = 56, 93.3%) and tumour distribution was predominantly unilobar (90.0% versus 10.0%). Most patients fulfilled the Milan Criteria (n = 49, 81.7%).

Histopathological assessment

Explant histopathological examination demonstrated that all patients had cirrhosis in the background liver tissue. Overall, the tumour burden in the explants was similar to that in radiological assessment. A total of 31 (51.7%) patients had a single HCC and 29 (48.3%) had multiple tumours. The median number of tumours was one (range: 1–9) (P = 0.819 versus radiological assessment) (Fig. 1). Subgroup analysis revealed a trend towards an increase in the proportion of patients with more than three tumours on pathological examination compared with radiological assessment [n = 9 (15.0%) versus n = 6 (10.0%); P = 0.180]. This resulted in an increase in the number of patients transplanted in whom disease in the explant fell outwith the Milan Criteria [n = 19 (31.7%) versus n = 11 (18.3%); P = 0.037]. As disease progression during the interval between DCE-MRI and LT may contribute to this increase, the delay between scanning and LT was analysed. There was a significantly longer interval between DCE-MRI and LT in these upstaged patients (111.6 days versus 61.3 days; P = 0.016).

Figure 1.

Tumour burden. (a) Number and (b) size of hepatocellular carcinomas detected by radiological and histological examination. Data are given as median ± interquartile range (IQR). Outliers denoted by ○ = values ≥ third quartile plus 1.5 times the IQR and * = values ≥ third quartile plus three times the IQR. Radiological versus histological assessment: P > 0.05, Mann–Whitney U-test

On an individual patient basis, in almost half of patients (n = 27, 45.0%) a discrepancy emerged between the total number of tumours identified in preoperative radiological imaging and that found in explant histopathological examination. The number of tumours was underestimated by DCE-MRI in 17 (28.3%) patients and overestimated in 10 (16.7%).

Overall, median tumour size was 15 mm (range: 3–55 mm) (P = 0.679 versus radiological assessment) (Fig. 1). The proportion of patients with a cumulative tumour diameter of <80 mm was similar to that derived from radiological examinations (n = 54, 90.0%; P = 0.666 versus radiological assessment). However, when individual tumours were considered, size discrepancies emerged in almost two thirds of all HCCs (n = 63, 64.3%). Radiological assessment overestimated lesion size in 34.2% of lesions and underestimated it in 30.1%. Most discrepancies occurred in lesions at the extreme ends of the range of sizes; this may partially reflect both the detection limits of MRI for small tumours and ability to accurately measure large lesions histologically. Nevertheless, in four (6.7%) patients, the largest lesion identified radiologically was not the largest lesion identified histologically (Fig. 2).

Figure 2.

Radiological versus pathological identification and assessment of tumours. (a) Number of individual hepatocellular carcinomas (HCCs) detected by radiological and histological examination on a per patient basis. (b) Size of individual HCCs detected by radiological and histological examination on a per lesion basis. All lesions identified on dual contrast-enhanced magnetic resonance imaging and in explants are displayed. Tumours that underwent bridging therapy are not displayed

A total of 129 tumours were identified in explants with a range of grades of differentiation: 39 (30.2%) were well differentiated; 66 (51.2%) were moderately differentiated, and 24 (18.6%) were poorly differentiated. Of 29 patients with multifocal HCC, one third (n = 9, 31.0%) had tumours of differing histological grades. In two patients (6.9%), the largest tumour, which would have been targeted for preoperative biopsy, was shown to be well differentiated HCC, whereas other smaller tumours in the explant were poorly differentiated HCCs. One of these patients had primary biliary cirrhosis with three HCCs, of which one measured 50 mm and was well differentiated, one measured 15 mm and was moderately differentiated, and one measured 18 mm and was poorly differentiated. The other patients had alcoholic liver disease with one 30-mm, well-differentiated HCC and one 8-mm, poorly differentiated HCC.

Of those lesions specifically diagnosed as HCC on pre-LT imaging, 98 were identified and available for histological examination. In five patients with single HCCs, on pre-LT DCE-MRI, more than one HCC was found in the explant. Only the tumour identified preoperatively was considered for this part of the study. Some HCCs diagnosed preoperatively could not be identified in the explants (n = 3) and some were completely destroyed by bridging therapy and were excluded (n = 6). Of the remaining 89 lesions, 77 (86.5%) were confirmed histologically as HCC and the remainder were identified as regenerative nodules (n = 6, 6.7%), dysplastic nodules (n = 4, 4.5%) or indeterminate despite extensive histological examination (n = 2, 2.2%).

In the present series, one patient underwent a diagnostic preoperative ultrasound-guided biopsy of a suspicious lesion. This patient had hepatitis C virus (HCV) infection and three lesions on DCE-MRI, two of which were thought to be highly suspicious for HCC. The largest (36 mm) was biopsied to reveal a small sample of a well-differentiated hepatocellular lesion. Within the limitation of the small sample size, this lesion was thought to be a dysplastic nodule with no definite evidence of HCC. Post-LT, this lesion was identified in the explant and showed features of a dysplastic nodule. However, two other lesions in the explant that were not biopsied pre-LT were found to be well-differentiated HCCs.

Downstaging

In the group of patients undergoing bridging therapy, as expected, more patients were found to exceed the Milan Criteria [five of 23 (21.7%) versus six of 37 (16.2%) not undergoing bridging treatment]. Three (8.7%) of the treated patients were subsequently downstaged to within the Milan Criteria on pathological assessment. Conversely, in three (8.7%) patients, disease was initially identified as within the Milan Criteria but subsequently progressed beyond the Milan Criteria despite bridging therapy. In the present series, median waiting time to receipt of a graft was slightly shorter in patients undergoing bridging therapy (37 days versus 63 days; P = 0.036).

Patient outcomes

In the present series, 1- and 3-year absolute overall survival rates were 83.1% and 81.0%, respectively (Fig. 3). Two patients experienced recurrence; in one of these patients extrahepatic pelvic metastatic disease occurred within 13 months of LT. This patient had HCV cirrhosis and four HCCs identified on DCE-MRI as beyond the Milan Criteria and underwent RFA therapy prior to LT. Pre-LT AFP was 3372 ng/ml. Pathological assessment confirmed nine viable HCCs in the explant and classed the disease as stage IIIa. There was no microvascular invasion and all tumours were moderately differentiated. A second patient died from pulmonary metastatic disease within 8 months of LT. This patient was also infected with HCV and had a single HCC within the Milan Criteria. Pre-LT AFP in this patient was normal at 11 ng/ml. The explant showed stage II disease with three HCCs, two of which, including the radiologically identified largest tumour, were poorly differentiated. No microvascular invasion was identified.

Figure 3.

Kaplan–Meier survival curves in hepatocellular carcinoma (HCC). (a) Overall survival was 54.23 months [95% confidence interval (CI) 47.78–60.68]. (b) Univariate analysis demonstrated that overall survival was significantly reduced if the largest HCC was poorly differentiated (P = 0.027). (c) Only microvascular invasion (MVI) was identified as an independent predictor in multivariate analysis (hazard ratio = 5.96, 95% CI 1.36–26.39; P = 0.019)

Interestingly, both of the patients with well-differentiated largest tumours, but other smaller poorly differentiated tumours in the liver remained alive without recurrence after 27 months and 55 months of follow-up, respectively.

Prognostic factors

Univariate analysis revealed that the presence of poor differentiation in the largest lesion, the presence of poor differentiation in any lesion in the explant, microvascular invasion and HCC recurrence were significant prognostic markers for survival (P < 0.05). However, following multivariate analysis, microvascular invasion was the only independently significant variable for survival [hazard ratio (HR) = 5.959, 95% confidence interval (CI) 1.35–26.39)] (Table 2). This loss of significance is most likely to be attributable to the small number of terminal events in this series and the significant correlation between microvascular invasion and poor tumour differentiation (χ² = 5.963, P = 0.015). The presence of poor differentiation in any lesion in the explant, the presence of poor differentiation in any smaller lesion (i.e. the non-largest HCC) and pre-transplant AFP of >400 ng/ml were significant univariate predictors of HCC recurrence in this series (P < 0.05).

Table 2.

Cox proportional hazards model for multivariate recurrence and survival analysis

Multivariate analysis
	B	SE	Wald	d.f.	P-value	Exp(B)	95% CI for Exp(B)
Recurrence
Poor differentiation of any HCC	10.078	232.630	0.002	1	0.965	23811.103	0.000–24630
Sex	5.240	35.732	0.022	1	0.883	188.616	0.000–4907
MELD score	−0.225	0.204	1.219	1	0.270	0.798	0.535–1.191
Differentiation of smaller HCC	7.658	35.689	0.046	1	0.830	2116.489	0.000–5056
Alpha-fetoproteina	0.855	1.682	0.259	1	0.611	2.352	0.087–63.538
Survival
Poor differentiation of any HCC	0.747	1.275	0.343	1	0.558	2.110	0.173–25.699
Poor differentiation of largest HCC	−0.546	1.363	0.161	1	0.689	0.579	0.040–8.378
Microvascular invasion	1.785	0.759	5.528	1	0.019	5.959	1.346–26.387
Sex	0.737	0.881	0.700	1	0.403	2.090	0.372–11.750
Recurrence	1.745	1.023	2.911	1	0.088	5.726	0.771–42.514
MELD score	−0.008	0.058	0.019	1	0.891	0.992	0.886–1.111

^aAlpha-fetoprotein was stratified as > 400 ng/ml versus < 400 ng/ml.

SE, standard error; 95% CI, 95% confidence interval; HCC, hepatocellular carcinoma; Poor differentiation of any HCC, the presence of at least one poorly differentiated HCC in the explants; Poor differentiation of smaller HCCs, the presence of poor differentiation in any of the non-largest HCCs; MELD, Model for End-stage Liver Disease.

Discussion

The demand for donor organs continues to far outstrip supply and HCC is one of the fastest-growing indications for LT worldwide.²⁷ This has led to mounting pressure to ensure the optimal use of available grafts and calls for the modification of HCC LT selection criteria and treatment protocols. The ability to reliably assess tumour biology pre-LT would allow accurate prognostication and inform treatment decisions. Beyond the direct impact on recipients, more effective allocation strategies would minimize indirect detrimental impact on those patients without HCC who remain on LT waiting lists, thereby improving collective outcomes. Current radiological staging has limitations and strategies to enhance its utility are required. Evidence on the prognostic value of poor tumour differentiation continues to accumulate.^20,28–33 This makes the use of a pre-LT staging biopsy to select and prioritize patients on the basis of tumour grade an attractive strategy that has been adopted by several groups.^19,20

The present study highlights important issues surrounding pre-LT staging biopsy that require careful consideration. In two (8.7%) patients with multifocal disease, biopsy of the largest tumour would theoretically have revealed falsely reassuring histological grading. In these patients, the largest lesions diagnosed radiologically as HCC were found to be well-differentiated tumours on histopathological examination of explanted material. However, crucially both patients had other smaller, poorly differentiated HCCs within the liver. Proponents of staging biopsy deem the presence of poor differentiation to confer a prognosis poor enough to justify the patient's exclusion from LT. Even if this principle is accepted, the present data make it clear that reliably identifying the largest lesion and obtaining a representative diagnostic biopsy from it presents significant challenges.

In the present study, explants from patients with multifocal HCC showed individual tumours of differing grades in 34.6% of cases. This represents a significant fraction of these patients and, indeed, 15.0% of all patients undergoing LT for HCC. Given this, a potential for misleading pre-LT biopsy certainly exists. Furthermore, the definition of the largest pre-LT lesions can also be problematic. In the present series, discrepancies between radiologically and histologically defined largest lesions emerged in 7.7% of cases. Sampling error, and inter- and intra-observer variability in histological assessment of tumours will also influence the reliability of staging biopsy.

One patient in this series underwent a pre-LT biopsy which revealed a dysplastic nodule, subsequently shown to be a well-differentiated HCC in the explant. This patient had end-stage HCV cirrhosis and although this patient's treatment strategy was not altered by the biopsy findings, this particular case highlights the limitations and technical challenges of histopathological examination of biopsy specimens. Non-representative sampling of heterogeneous tumours is a significant problem. A recent review estimates the specificity and positive predictive value of diagnostic liver biopsy for suspected HCC in cirrhotic patients to be 100%, with a correspondingly infinitely high positive likelihood ratio (LHR) of >10 indicating its ability to rule in the diagnosis.^19,34,35 By contrast, biopsies typically achieve sensitivity of 86–90%. This falls to 83% for lesions of <10 mm in size and to 66% if a smaller needle (21–22-gauge) is used. Negative predictive values vary (13–75%) and the negative LHR is 0.07–0.34 (values of < 0.1 provide strong evidence of a test's ability to rule out a diagnosis). Furthermore, it is estimated that despite image guidance, which is dependent upon the size and location of a tumour, 2–11% of biopsy specimens will be inadequate for diagnostic examination by the pathologist.³⁴ One previous study specifically examined the ability of pre-LT needle core biopsies of HCC to predict the grade on formal histopathological examination of explanted specimens.³⁶ Its authors reported a relatively poor predictive ability [κ-statistic = 0.38, receiver operating characteristic (ROC) area under the curve (AUC) = 0.74] with a correspondingly poor ability to predict the presence of microvascular invasion and outcome compared with formal grading of the tumour in the explanted liver.³⁶

Proponents of biopsying argue that high-quality data from centres that regularly carry out biopsies prove it to be safe and effective with very low rates of needle track seeding.^19,37 At the Barcelona Clinic, rates of needle track seeding are 0.01% and evidence for the safety of laparoscopic biopsy accumulates.³⁸ In a meta-analysis of eight studies conducted in 2008, the authors concluded that the overall incidence of seeding was 2.7% (95% CI 1.8–4.0) or 0.9% per year, with all occurrences treated successfully without impact on operability or survival.^39–47 Strategies to improve the safety of biopsy and reduce seeding have been proposed, such as the use of a coaxial biopsy needle with an introducer, which would allow for multiple passes of the needle with the protective introducer remaining in place, although these have not been rigorously tested.⁴⁵

The increasing use of bridging therapy may also influence the utility of pre-LT biopsy. Although predominantly used as an initial tool in treatment selection protocols before any bridging therapy, there may be occasions when patients who have undergone these treatments require assessment or reassessment for LT. The response to initial bridging or downstaging therapy is itself proposed as a marker of tumour biology, prognosis and a selection parameter and may become more important with the increased use of downstaging therapy.⁴⁸ The effects of these treatments will clearly influence the ability of pre-LT biopsy to accurately stage such lesions. In the present study, the majority of those treated with bridging therapy demonstrated at least some viable HCC present in the explants, although some had been completely destroyed. Whether those persisting neoplastic regions would have theoretically been sampled in a biopsy remains unknown.

The present study utilized formal histological examinations of explant specimens, often supplemented with specialized immunostaining techniques, to speculate on the potential findings of theoretically obtainable small needle core biopsies. As this and previous studies have highlighted, the representative nature of such biopsy samples remains controversial.^36,37,49 Emerging molecular techniques may enhance examinations of biopsy specimens and indeed become helpful in risk stratification. However, they remain far from routine in clinical practice and do not remove the error inherent in the sampling of heterogeneous tumours.⁵⁰

In patients with a well-differentiated largest tumour and smaller coexisting poorly differentiated tumours, biopsy may still have a potential role. Sampling of multiple lesions would allow more accurate staging, but the significantly increased risks are likely to negate any added benefit, whatever modality of biopsy is used. In the future, other novel radiological assessments could be used to target the biologically most aggressive lesion for biopsy. Some evidence supporting the role of positron emission tomograpghy (PET)-CT for non-invasive staging has been reported, but this requires further study and validation.^51–53 The targeted biopsy of lesions likely to be poorly differentiated according to PET-CT criteria could be one potential approach. Non-invasive dynamic and molecular imaging techniques to assess tumour biology are also under development and may circumvent the need for biopsy altogether.^54,55

In conclusion, the present results suggest there is currently a limited role for routine pre-LT HCC staging biopsy at this particular centre. In multifocal disease, the largest lesion was not always representative of tumour burden or biological aggression and its potential use to select or exclude patients from curative treatment is questionable. Equally, the ability to accurately identify and obtain representative biopsy samples from the largest lesion can be difficult with currently available techniques. The accuracy of static radiological staging remains suboptimal and there is certainly a need for alternative strategies to improve treatment and selection. There are still unanswered questions surrounding staging biopsy which require further study, possibly with the use of more advanced statistical modelling techniques.

Conflicts of interest

None declared.