Diagnostic biopsy for hepatocellular carcinoma in cirrhosis: useful, necessary, dangerous, or academic sport?

Hepatocellular carcinoma (HCC) is the eighth most frequent tumour entity worldwide (one million per year). The incidence in the USA and elsewhere has increased, which is probably due to the rise in the incidence of hepatitis C virus infection. A number of cohort studies have shown that surveillance using abdominal ultrasound (US) improves management of HCC developing in cirrhotic patients as it allows early detection and application of potentially curative treatments.¹ Thus far it is not entirely clear whether surveillance increases survival as no randomised controlled trials have been conducted to answer this question. However, as eventually curative treatments such as resection, ablation, and in particular liver transplantation have emerged as effective means of treating HCC as long as it is not too large and too widespread, early diagnosis seems to be a reasonable goal in improving the poor prognosis of this cancer.^2–4 Interestingly, it has recently been shown by comparison of time periods that besides being part of the more recent cohort, tumour staging was the only independent predictor of survival.¹ Thus it seems obvious that detection of early carcinoma in liver cirrhosis would be a reasonable goal. Progress made in recent years regarding imaging, in particular spiral computed tomography and using the increased vascularity of HCC, allowing for better scanning protocols, has increased the detection rate significantly. In addition, magnetic resonance imaging has become the diagnostic procedure of choice in some institutions. However, most experts still advocate liver biopsy.^5–11

In addition, improved US techniques such as contrast enhancement with and without Doppler ultrasonography^12,13 may improve early diagnosis further. However, no adequate comparisons between all competing techniques are available to date. Most studies have used two or a maximum of three modalities and have come to different conclusions.¹⁴ The role of positron emission tomography in different modes is also not clear but it does not seem to be very helpful.¹⁵ Well performed large studies on modalities for early diagnosis of HCC are therefore welcome.

In this issue of Gut, a study from Italy¹⁶ reports on 4375 patients with newly detected cirrhosis and no focal lesions at initial US followed over a period of nine years (see page 1356). Of 4581 patients agreeing to participate, 206 had nodular lesions at initial US. Patients were followed every 4–6 months using α fetoprotein (AFP) and abdominal US. If a lesion was diagnosed and AFP levels were above 400 ng/ml, HCC was assumed and treated accordingly. Lesions associated with an AFP level below 400 ng/ml underwent US guided fine needle biopsy (FNB) performed by an experienced gastroenterologist using fine cutting needles and non-cutting spinal needles. Choice between needles was left to the individual examiner. When histology or cytology indicated HCC, the patient was treated. If biopsy revealed a negative result, computed tomography and, in one centre, contrast enhanced magnetic resonance imaging were performed. A total of 688 new focal lesions of 6–42 mm in diameter were detected. In 294 patients (43%), lesions were ⩽20 mm; 258 of these (87.6%) were diagnosed as having HCC. In those with very small nodules (⩽10 mm), HCC was diagnosed in 69% of cases in contrast with 91% in those with nodules of 11–20 mm. Only 20 of 294 patients had an elevated AFP >400 ng/ml. The remaining 274 underwent biopsy (fig 1 ▶). In 245 patients, a correct diagnosis was made which was confirmed by follow up: 210 patients had HCC, four had a non-Hodgkin lymphoma, and 31 were found to be benign, as judged by follow up. Twenty nine of 274 cases were false negatives. Thus in a screening population, more than 66% of small nodules arising in cirrhotic livers may prove to be HCC and 90% of those can be identified using US guided FNB.¹⁶

Figure 1.

 Overview of the study.¹⁶ AFP, α fetoprotein; US, ultrasound; FNP, fine needle biopsy.

This topic has been discussed previously in this and other journals^17–22 15–20 years ago. Very similar results regarding the sensitivity of AFP and good sensitivity of US guided biopsy have been reported.²¹ A larger series conducted 10 years ago again revealed similar results.²³

Why then do we need new studies and why do we still not know how to proceed? This may be due to the fact that three questions still remain unanswered:

1. How good is the technique and which is the preferred technical modality to perform a biopsy?

2. How dangerous is such a procedure and does it interfere with later treatment?

3. Is a biopsy necessary at all, and does it change the outcome?

EFFICACY OF US GUIDED BIOPSY

When considering the historical review given above and data from the paper published in this issue of Gut,¹⁶ it appears at first glance that US guided biopsy has a very high efficacy in defining HCC. A rate of 10% false negatives is probably difficult to beat by any other technique. However, a recent study following 12 cases of lesions with an initial fine needle aspiration diagnosis of regenerating nodules revealed subsequently that 10 of these patients had a HCC. Thus a negative biopsy result is probably not helpful in suspected HCC, as in any other suspected tumour. If the initial biopsy is negative, repeat biopsy leads to diagnostic gain in one third of patients, in particular if the first biopsy provided a non-diagnostic sample (necrosis) or a false negative result due to well differentiated HCC.²⁴ It is not entirely clear if there is a major difference between cutting and aspiration needles—early studies did not find a difference when an experienced pathologist was involved.²⁵ The positive predictive value of imaging modalities has recently been shown to be rather high.¹¹ However, this is not the case in all studies. A recent study from Japan demonstrated that many nodules detected by ultrasonography could not even be found by computed tomography.⁸ Another study used digital subtraction angiography and magnetic resonance imaging and found that imaging alone was sufficient to diagnose HCC in 58% of well differentiated and 87% of moderately and poorly differentiated nodules in 139 patients with chronic liver disease, where 207 nodules had been found by periodic surveillance.⁹ Imaging alone allowed for diagnosis of HCC in 60% of cases: 55% of patients with nodules of 10 mm in diameter or less required biopsy. With these conflicting results and the data of Caturelli and colleagues,¹⁶ it appears that at present, in experienced hands, US guided biopsy using either cut needles or fine needle aspiration seems to be technically appealing and most effective. However, the approach chosen by the authors does not necessitate biopsy in all patients as those patients with AFP levels >400 ng/ml and those with typical features of HCC in larger nodules using imaging are taken for granted. Only 274 nodules in 4375 patients were biopsied because their size was ⩽20 mm in diameter.

COMPLICATIONS

With regard to the second question, data are also contradictory. Overall, complications of US guided diagnostic and therapeutic procedures on focal liver lesions are rare. In a large series from the same authors, 16 648 guided biopsies and 3035 therapeutic procedures in 13 222 patients were analysed. Overall mortality was 0.06% (none after diagnostic procedures).²⁶ However, the bleeding risk is greater in HCC as these lesions are usually hypervascular. In a series of 159 HCC patients undergoing fine needle aspiration, four (2.5%) bled significantly, one of whom died.²⁷ Thus bleeding has to be considered as a possible complication of biopsy in this tumour entity, in particular with subcapsular localisation of the tumour.

Another “complication” is the risk of mismanagement after interpretative errors. It is sometimes very difficult to distinguish well differentiated HCC from benign regenerative nodules, and criteria used by pathologists need to be very clear.²⁸ With fine needle aspiration in particular, diagnosis may be difficult; aspiration of necrotic material seems to be another limiting factor for correct interpretation.¹⁰

Others consider needle tract seeding of malignant cells an important risk factor. This risk was found to be 5.1% (three patients) in a series of 59 patients.²⁹ It was smaller in a series of 122 patients (1.6% (two patients)).³⁰ No recurrence was observed after local excision in both series. However, in other series the risk of seeding was much lower.³¹ These series were however the result of questionnaires, with all the uncertainties inherent in this approach.

In our view, there are risks present but these are negligible and, in particular, the risk of needle seeding should not preclude biopsy if the result influences treatment choice. The risk of seeding has to be weighted against a 2.5% unnecessary surgery rate when patients are not biopsied.¹¹ An exception may be patients with advanced liver cirrhosis who are awaiting transplantation, independent of a possible tumour. Here tumour seeding could be disastrous and a “wait and see strategy” may be reasonable. However, a biopsy may be helpful because in the case of a tumour, one would have to do an adequate “bridging therapy” until transplantation (for example, chemoembolisation) or consider early transplantation (for example, by a living donor).

NEED FOR BIOPSY

The question of whether biopsy is necessary is also difficult to answer with the data available. Examining again the paper of Caturelli and colleagues¹⁶ it is obvious that in many patients an early diagnosis was made which may allow for different treatment options, such as percutaneous ethanol injection, resection, or liver transplantation. A recent study demonstrated that percutaneous ethanol injection had a similar survival rate as surgery and was actually approximately 60% at five years for patients with small HCC.³² However, the European Association for the Study of Liver currently recommends a “wait and see policy” for small lesions³³ as only half of these are considered to be HCC. The current study contradicts this assumption.³⁴ However, the question of whether obtaining a liver biopsy affects clinical outcome cannot yet be answered with the current information. Early detection of HCC seems to increase the chance of treatment, at least in Hong Kong,³⁵ and also seems to improve prognosis, as recently described in this journal.³⁶ It is doubtful if randomised controlled studies will ever be conducted to answer this question definitively.

In the past, outcome for most patients, regardless of stage, was believed to be poor. However, as transplantation results in a prolonged disease free survival or even cure, as shown by some studies (for example, Mazzaferro and colleagues³⁷), biopsy with needle seeding would actually deny the patient treatment and therefore change the clinical outcome significantly. Thus in our opinion, patients waiting for a transplant should only be biopsied if AFP levels are less than 20 ng/ml, if the lesion cannot be characterised as HCC using other modalities (all of them should be performed), and if the presence of a HCC favourably alters the patient’s candidacy for liver transplantation. If resection appears to be the best option, biopsy may or may not be performed. When palliative treatment is planned, biopsy makes sense in order to avoid unnecessary treatment.

In summary, the study of Caturelli and colleagues¹⁶ provides long sought after and important information on the possible efficacy of US guided biopsy of HCC in experienced hands and will probably define the gold standard of quality of this procedure. However, it does not answer the question of whether biopsy should be performed and, if so, in which patients when HCC is suspected on the basis of a cirrhotic liver. This remains to some extent a bedside decision, which probably should be made after discussion between surgeons, hepatologists, and interventionalists, and will be dependent on the treatment options available for the individual patient.