The role of preoperative biopsy in pancreatic cancer

Abstract

Background

Cyto-histological diagnosis of pancreatic pathology in the management of suspected pancreatic malignancy is re-evaluated in the light of evolving trends in management and tissue sampling.

Methods

The literature on cyto-histological diagnosis of pancreatic pathology was reviewed over the period 1977–2003.

Results and conclusion

Endoluminal techniques of tissue sampling carry less risk of tumour seeding and are the sampling methods of choice for potentially resectable lesions if a tissue diagnosis will alter therapy. Endosonar-guided biopsy in expert hands appears to be the most reliable tissue sampling technique. Transcoelomic sampling methods should be reserved for those with un-resectable disease. All methods of cyto-histological diagnosis have high specificity but have a low negative predictive value. Therefore, potentially resectable lesions that are highly suspicious for malignancy should be explored even if pre-operative histology or cytology is negative.

Introduction

The inaccessible position of the pancreas and the difficulty in distinguishing inflammatory changes from malignancy make it difficult to establish a preoperative diagnosis of pancreatic cancer. Previously the high mortality rate associated with pancreato-duodenectomy meant that surgeons were reluctant to proceed to operation without such a diagnosis. Obtaining histological confirmation of cancer before operation became an important issue in management algorithms. However, advances in imaging techniques have greatly improved the capability of radiology to diagnose pancreatic cancer. This development coupled with the fact that pancreato-duodenectomy can be performed with a low mortality rate, albeit with an appreciable morbidity rate, has resulted in less emphasis being placed on the establishment of a preoperative diagnosis ¹,²,³,⁴. Certain considerations make a preoperative tissue diagnosis necessary. These include the use of neo-adjuvant therapy, and situations of diagnostic doubt involving benign strictures or focal areas of pancreatitis, in which either no operation or a different operation would be performed. When non-operative palliation is being considered, a tissue diagnosis is necessary for counselling. In specific instances, a favourable histology report may prompt an operation that would not have been considered in the setting of pancreatic adenocarcinoma. In a small but important subset of patients, treatable conditions such as lymphoma or tuberculosis may be distinguished from adenocarcinoma by preoperative biopsy ⁵,⁶. A variety of techniques can be used to obtain tissue and cellular material for diagnosis.

Histological techniques involve intact tissue specimens and allow assessment of both tissue architecture and cell morphology. Cytological techniques rely on individual exfoliated or aspirated cells. Anisonucleosis, large nuclei and nuclear moulding, nuclear membrane irregularity, nuclear crowding and nuclear enlargement are important cytologic features of pancreatic adenocarcinoma ⁷,⁸.

However, other features such as necrosis, chromatin clearing, mitosis, macronucleoli and hyperchromasia may be mimicked by reactive changes, leading to a false positive diagnosis. For a technique to be widely adopted, it must be sensitive and specific as well as safe and easy to perform. Meaningful interpretation of cytology specimens is dependent on the technique, the site of the lesion, the skill of the clinician performing the biopsy and the cytologist interpreting the specimen. The techniques of obtaining cytology specimens include aspiration of pancreatic or duodenal juice at ERCP, mechanical exfoliation of pancreatic cells by brushing techniques and fine needle aspiration (FNA) of the lesion. This review outlines these methods, their indications and reliability, and attempts to clarify their role in the modern management of patients with pancreatic cancer.

Exfoliative and brush cytology

Pancreatic cells exfoliate into pancreatic and duodenal secretions and provide material for cytological assessment. However, these cells must erode either the ductal system or duodenum to reach the point of sampling, where they are subject to the cytolytic effect of duodenal enzymes. Therefore these cells are often unsuitable for cytological assessment and may display artefactual changes such as cytoplasmic eosinophilia and nuclear crenation ⁹. Brush cytology attempts to mechanically exfoliate cells from the ductal epithelium into the pancreatic juice and is intended to increase cellular yield.

Several authors have reported their results with exfoliative and/or brush cytology ¹⁰,¹¹,¹²,¹³,¹⁴,¹⁵,¹⁶,¹⁷,¹⁸; Table 1 summarizes these results ¹⁰,¹¹,¹²,¹³,¹⁴,¹⁵,¹⁷,¹⁸,¹⁹. Generally the results have been disappointing, with sensitivities ranging from 50% to 70%. The results of exfoliative cytology are particularly poor, with Kurzawinski and colleagues reporting a sensitivity of 33% for this technique ¹²; when they combined exfoliative and brush cytology they improved the sensitivity to 69%. The interpretation of results is made difficult by poorly defined case selection, with several studies including both peri-ampullary and pancreatic tumours. Furthermore, the number of attempts needed to establish a diagnosis is often unclear. Technical modifications have been described, such as the report using a device consisting of a 10F dilator attached to a pad of Velcro, with semi-rigid, mushroom-shaped bristles in a small series of 15 patients ¹⁹. These authors obtained a 100% sensitivity rate for brush cytology using this modified device. However, this result has not been reproduced by any other authors over the last decade. While a positive result absolutely confirms the diagnosis of malignancy the problem of the false negative result is significant. There are other concerns with brush cytology. It is associated with a definite morbidity. In one study pancreatitis developed in 11% and 21% of patients post biliary and pancreatic brush sampling, respectively ¹⁴. Furthermore, brush cytology can only be performed at ERCP and modern non-invasive imaging techniques such as MRCP are likely to restrict ERCP to therapy rather than diagnosis ²⁰,²¹.

Table 1. Results of aspiration and brush cytology techniques.

Author	Year	n	Modality	Sensitivity	Specificity	PPV	NPV
Venu [15]	1990	53	Brush cytology	70%	100%	100%	44%
Kurzawinski [12]	1993	47	Exfoliative cytology	33%	100%.	100%	16%
Kurzawinski [12]	1993	46	Exfoliative and brush cytology	69%	100%	100%	36%
Ferrari Jr [17]	1994	74	Brush cytology	56.2%	100%	100%	51.2%
Mansfield [13]	1997	54	Brush/exfoliative*	54%	100%	100%	8%
Parasher [19]	1998	15	Brush cytology	100%	100%	100%	100%
Vandervoort [14]	1999	134	Brush cytology (bile duct)	50%	89%	89%	50.8%
Vandervoort [14]	1999	134	Brush cytology (pancreatic duct)	58.3%	100%	100%	64%
Glasbrenner [10]	1999	78	Brush cytology	56.1%	90.5%	94.1%	43.2%
Macken [18]	2000	106	Brush cytology	63%	96%	92%	64%
Govil [11]	2002	278	Brush cytology	68%	100%	100%	76%

PPV, positive predictive value; NPV, negative predictive value.

*This study used a variety of methods to obtain cytology; these included standard exfoliative and brush cytology as well as cytology obtained from the stent retriever and the retrieved stents.

New developments in cytology

New biochemical and molecular techniques may improve the sensitivity of cytology. A histochemical adjunct to the differentiation of pancreatitis from pancreatic carcinoma is the oxygen insensitivity assay of glucose 6-phosphate dehydrogenase (G6PDH) activity. Malignant cells in the presence of 100% oxygen reduce a specific tetrazolium salt to produce formazan. This reaction is inhibited in non-malignant cells by oxygen free radicals. Van Driel and co-workers demonstrated oxygen insensitivity of this enzyme in brush cytology material of pre-malignant pancreatic lesions and pancreatic carcinomas, achieving 100% sensitivity and specificity in 11 cases of pancreatitis and 28 cases of carcinoma ²².

The use of digital image analysis (DIA) to assess brush cytology specimens from pancreato-biliary strictures has been reported ²³. The authors de-stained and re-stained with Feulgen dye the cytology specimens from a variety of biliary and pancreatic strictures. Nuclear images were quantified for DNA content. Aneuploidy was assumed to indicate malignancy. The ploidy results of the subsequent DNA histograms were compared with the established diagnosis. DIA had a sensitivity of 85% for the diagnosis of malignancy.

Mutation of codon 12 of the K-Ras oncogene is implicated in pancreatic carcinogenesis. However, the mutation was identified in only 50% of resected pancreatic adenocarcinomas in a recent study ²⁴. Furthermore, the mutation can be detected in patients without any pancreatic pathology and in patients with benign pancreatic pathology ²⁴,²⁵,²⁶,²⁷,²⁸,²⁹,³⁰,³¹,³². A number of techniques may improve our ability to detect the mutation. These include mutant allele-specific amplification to detect the mutation in duodenal aspirates and the polymerase chain reaction (PCR) to detect the mutation in deoxyribonucleic acid extracted from serum ³⁰,³³. Although these molecular techniques may enhance our ability to detect the mutation, the presence of the mutation remains a non-specific finding of uncertain significance ²⁶.

Endoscopic transpapillary biopsy techniques

Biopsy techniques obtain tissue for histological assessment rather than individual cells for cytological assessment. Various devices have been developed to obtain histology during ERCP ³⁴,³⁵,³⁶,³⁷. Kubota and colleagues used biopsy forceps to acquire transpapillary tissue in 43 patients with pancreatic and biliary strictures ³⁴. Tissue sampling from the strictures was successful in 95.3% of patients, without complications. All specimens were adequate for evaluation. Sensitivity was 88% for cholangiocarcinoma and 71% for pancreatic carcinoma. There were no false positive results. Recent authors have used transpapillary biopsy in combination with brush cytology and more recently with intra-pancreatic ductal endosonar to improve accuracy ³⁵. Table 2 summarises the studies that have used transpapillary biopsy in combination with other modalities to diagnose pancreatic cancer. Transpapillary biopsy techniques have a sensitivity ranging from 46% to just under 80% for the diagnosis of pancreatic malignancy and cannot be regarded as a reliable diagnostic tool for the definitive diagnosis of pancreatic cancer.

Table 2. Results of comparative and combination studies using transpapillary biopsy techniques.

Author	Year	n	Modality	Sensitivity	Specificity	PPV	NPV
Kubota [34]	1993	43	Forceps biopsy	71% (pancreatic cancer)	100%	100%	60%
Schoefl [36]	1997	106	Forceps biopsy	64.9%	100%	100%	69.2%
Schoefl [36]	1997	63	Brush Cytology	46.7%	100%	100%	61.9%
Schoefl [36]	1997	48	Forceps + brush	70.4%	100%	100%	71.4%
Pugliese [37]	1995	52	Forceps biopsy	53%	100%	100%	48%
Pugliese [37]	1995	94	Brush cytology	54%	100%	100%	50%
Pugliese [37]	1995	52	Forceps + brush	61%	100%	100%	53%
Domagk [35]*	2002	50	Forceps	52%	100%	100%	29%
Domagk [35]†	2002	30 (pancreatic cancer)	Intra-ductal ultrasound	90%	83%	84%	89%

PPV, positive predictive value; NPV, negative predictive value.

*The pathology reports were unable to classify the origin of the malignancy on the basis of tissue obtained by transpapillary forceps biopsy.

†Intraductal ultrasound was capable of classifying the origin of the biliary tract malignancy: pancreatic (30), bile duct (17) and gallbladder (3).

Percutaneous biopsy

Modern imaging allows accurate percutaneous placement of biopsy needles. This led to reports of guided FNA in the work-up of pancreatic cancer in the late 1970s ³⁸,³⁹. Since these initial small cohorts the technique has been widely adopted. However, the sensitivity of the technique remains unsatisfactory, with modern series seldom achieving sensitivity rates higher than 90% ⁴⁰,⁴¹,⁴². The more recent reports are presented in Table 3. There are a number of limitations with percutaneous FNA. The clinician cannot be sure that the pathological lesion has been sampled accurately. It is a real concern that smaller lesions, which are potentially resectable, may be the most difficult to sample accurately. Serious complications of percutaneous FNA occur but may not always be reported. Bile leak and peritonitis, infection, septicaemia and fatal pancreatitis have all been described ⁴³,⁴⁴,⁴⁵,⁴⁶. The incidence of these complications may be under-reported ⁴⁷.

Table 3. Results of percutaneous FNA and EUS-FNA.

Author	Year	n	Modality	Sensitivity	Specificity	PPV	NPV
Ekberg [40]	1988	79	Ultrasound-guided FNA	85%	100%	100%	55%
Kocjan [42]*	1989	62	Ultrasound-guided FNA	86%	100%	75%	62%
Di Stasi [41]	1998	510	Ultrasound-guided FNA	84%	100%	100%	61%
Bhutani [66]	1997	47	EUS-guided FNA	64%	100%	100%	16%
Williams [65]	1999	144	EUS-guided FNA	82%	100%	100%	51%
Gress [67]	2001	102	EUS-guided FNA	93%	100%	100%	91%
Harewood [68]	2002	185	EUS-guided FNA	94%	71%	96%	60%

PPV, positive predictive value; NPV, negative predictive value.

*The authors excluded inadequate and atypical specimens.

The potential for seeding of malignant cells by these biopsy techniques is a cause for concern. Implantation of tumour along the percutaneous tract has been reported. This is generally regarded as a rare complication ⁴⁸,⁴⁹. The potential for transcoelomic seeding appears to be more significant. Warshaw presented the first evidence that this is a real problem ⁵⁰,⁵¹. Peritoneal washings in 40 patients with pancreatic adenocarcinoma were submitted for cytological assessment, all of which had been deemed to be resectable on imaging. In 30% of patients malignant cells were found in the peritoneal washings. In patients who had previously undergone percutaneous FNA, 75% had positive cytology. In the patients not submitted to FNA only 19% had positive cytology. Warshaw speculated that percutaneous FNA implanted tumour cells into the peritoneum, so converting a potentially curative situation into one of disseminated malignancy. He went on to advocate operating without attempting to biopsy the tumour. Warshaw's concerns about the significance of malignant peritoneal washings are supported by the results of Nakatsuka and colleagues, who showed that positive peritoneal cytology is associated with poorly differentiated pancreatic cancer with significantly worse survival than patients with negative peritoneal cytology ⁵².

However, Yachida and colleagues feel that the significance of positive peritoneal cytology is uncertain ⁵³. They examined peritoneal washings from 134 patients who had been subjected to pancreatic resection for adenocarcinoma. Positive cytology was found in only 14%. Of the 19 patients with positive cytology, 16 had no visible peritoneal disease and the remaining 3 had minimal visible peritoneal disease. There was no difference in survival between those with negative cytology and those with positive cytology and no peritoneal disease. They felt that positive cytology was a rare phenomenon in resectable pancreatic cancer, was of uncertain significance and should not exclude resection.

The MD Anderson group feel that the potential of transcoelomic seeding of malignant cells by percutaneous FNA is overstated ⁵⁴. They prospectively obtained peritoneal washings from 60 patients with potentially resectable pancreatic cancer. Although percutaneous FNA had been performed prior to referral in 82%, only four patients (7%) had positive cytology. Of this group three patients had been subjected to percutaneous FNA and one had not. All of these patients subsequently developed metastatic disease. These authors agreed with Warshaw that positive cytology is a marker of advanced disease; however, they did not agree that percutaneous FNA is causally implicated in the development of peritoneal seeding.

Recommendations as to the role of percutaneous FNA

Percutaneous FNA is as safe as any other technique of pancreatic tissue sampling and has an established role in confirming the diagnosis of pancreatic cancer when non-operative management is being considered. Although the evidence as to the relationship between percutaneous FNA and coelomic seeding is conflicting and awaits further clarification, the possibility of sampling error and the dissemination of localised disease continues to limit the role of percutaneous FNA in potentially resectable pancreatic carcinoma. The emergence of adjuvant and neo-adjuvant therapy in pancreatic cancer and the requirement to confirm a diagnosis before embarking on such treatment has once again pushed the issue of a tissue diagnosis into centre stage ⁵⁵,⁵⁶,⁵⁷,⁵⁸,⁵⁹,⁶⁰. The possibility of patients with pancreatic cancer surviving for longer periods means that the issue of peritoneal seeding is more important than was previously the case when survival was limited to <18 months. Endoscopic ultrasound (EUS) is an emerging technology that may lead to re-appraisal of the place of pancreatic cytology.

EUS-guided FNA

EUS allows a high-frequency ultrasound probe to be placed in close proximity to the pancreas, eliminating interference due to overlying bowel gas. It also allows higher frequencies to be used, resulting in improved resolution of the images of the pancreas and surrounding lymphatic and vascular structures ⁵⁹. However, EUS is not as widely available as spiral CT and requires extensive experience and knowledge of pancreatic sonographic anatomy. EUS appears to be as good as spiral CT scan in the detection of tumours >3 cm and in the assessment of venous invasion and lymph node involvement. Its main advantage is the assessment of small tumours (<3 cm), which may be missed on CT scan. EUS enables the clinician to sample a localized lesion accurately, so enabling a confident tissue diagnosis to be made ⁶⁰,⁶¹,⁶².

Chang and colleagues have evaluated EUS-FNA in the work-up of patients with suspected pancreatic carcinoma ⁶³,⁶⁴. They subjected 44 patients to EUS-guided FNA of pancreatic lesions and associated lymph nodes; most lesions were in the pancreatic head. Adequate specimens were obtained in 94% of the pancreatic lesions and 100% of lymph nodes. EUS-guided FNA had a sensitivity of 92%, specificity of 100%, and diagnostic accuracy of 95% for pancreatic lesions. When inadequate specimens were considered, sensitivity was only 83%, specificity 80% and diagnostic accuracy 88%. However, EUS-guided FNA impacted positively on patient management and avoided surgical exploration in 27%. It avoided further investigations in 57% of patients.

Williams and colleagues reviewed their experience with EUS-FNA in 333 patients, of whom 144 had pancreatic lesions ⁶⁵. EUS-FNA had a sensitivity of 84%, specificity of 96% and accuracy of 85% for the diagnosis of malignancy. EUS-FNA had superior accuracy and specificity for diagnosing malignancy in lymph nodes and only one complication occurred. Bhutani and colleagues achieved a sensitivity of 64%, specificity of 100% and positive predictive value of 100% for the diagnosis of pancreatico-biliary malignancy in 47 patients ⁶⁶. Nine cases (19%) were reported as suspicious for but not confirmatory of malignancy, which reduced their sensitivity rate. Cancer was confirmed at operation in eight of these patients. Gress and colleagues investigated the efficacy of EUS-FNA biopsy of suspected pancreatic malignancies in patients with suspicious lesions but negative cytology ⁶⁷. There were four false negative results and minor complications in three patients. EUS-FNA was considered to complement other techniques such as CT-guided FNA or cytology obtained at ERCP. Harewood and Wiersema used EUS-FNA in patients with suspected pancreatic cancer who had previous negative cytology from CT-guided FNA or ERCP ⁶⁸. In 58 patients with negative CT-FNA, EUS-FNA had 90% sensitivity for malignancy, 50% specificity for benign disease and 84% accuracy. In 36 patients with negative brushings, results for EUS-FNA were sensitivity 94%, specificity 67% and accuracy 92%. There were minimal complications.

EUS-FNA appears to be safe and reliable. It may eliminate the risk of diffuse peritoneal seeding and allows accurate placement of the sampling needle. EUS-FNA has also been reported to obtain better quality aspirates than CT-FNA, so allowing for more detailed cytological assessment ⁶⁹. It is useful in distinguishing small areas of focal pancreatitis from carcinoma and can avoid operation in specific patients. However, EUS-FNA requires skilled endoscopists to achieve meaningful results and the fear of missing a small resectable tumour continues to cause many surgeons to err on the side of caution and proceed to operation ⁷⁰.

Other pancreatic malignancies

Cystic neoplasms and neuro-endocrine tumours (NET) of the pancreas are rare cancers and present specific problems as regards cytological diagnosis. However, they have a better prognosis than pancreatic adenocarcinoma. This makes a tissue diagnosis important, as a cytology/histology report suggesting a cystic or neuro-endocrine malignancy might favour a surgical approach even if previous imaging seemed to indicate an irresectable situation.

Neuro-endocrine tumours

NETs account for <1.5% of pancreatic neoplasms, and cytological experience with these lesions is limited. It is difficult to distinguish NETs from islet cell hyperplasia, a feature of chronic pancreatitis. The differential diagnosis of NET is wide and includes well-differentiated adenocarcinoma, acinar cell carcinoma, metastatic small cell carcinoma and lymphoma. The presence of a cellular aspirate of discohesive and clustered cells exhibiting uniformly round, eccentric nuclei with'salt and pepper' chromatin pattern and minimal granular amphophilic cytoplasm are highly suggestive of a NET ⁷¹. A rosette formation in the absence of tumour diathesis, as well as poorly cellular lesions of small cell clusters admixed with benign exocrine pancreatic cells, favours islet cell hyperplasia. Single, plasmacytoid cells with finely granular chromatin distribution have been reported as being typical of cytology from islet cell tumours with mitoses and cellular necrosis being typical of malignant islet cell tumours ⁶⁹. FNA was able to successfully distinguish NET from pancreatic adenocarcinoma in a small series of six patients ⁷². Immunohistochemical staining was used in five patients and electron microscopy in the remaining case. Five specimens were obtained by percutaneous CT-guided aspiration and one specimen was obtained intra-operatively.

Cystic neoplasms

Cystic neoplasms of the pancreas include serous cystadenoma and the rarely reported serous cyst adenocarcinoma, benign, borderline and malignant mucinous cystadenomas/ cystadenocarcinomas, intraductal papillary-mucinous neoplasms and solid pseudopapillary cystic neoplasms. These tumours pose a diagnostic dilemma requiring differentiation from each other and from adenocarcinoma, and from benign entities such as pseudocysts, retention, congenital, endometrial and parasitic cysts ⁷³,⁷⁴,⁷⁵,⁷⁶,⁷⁷. EUS can accurately demonstrate the internal structures of cystic neoplasms such as mural nodules, so facilitating the diagnosis ⁷⁸. Cytology and biochemical fluid analysis must be interpreted in conjunction with clinical history and imaging. Both intraductal papillary mucinous neoplasms and mucinous cystic neoplasms have a mucinous background, allowing differentiation from serous lesions but not from each other. The sensitivity of FNA is low in malignant lesions, impaired by the focal nature of the malignant transformation. The cytology of serous cystadenoma may be diagnostic provided that cells are adequately represented and preserved. Typically there is a paucity of uniform cuboidal cells with round normochromatic nuclei. A mucinous or inflammatory background is absent, allowing differentiation from mucinous neoplasms and pseudocysts, respectively. However, these features are usually only evident in half the cases. It may therefore be necessary to rely on a number of imaging features, such as microcystic spaces, a central scar and sunburst calcification and hypervascularity to diagnose a cystic neoplasm.

Biochemical analysis of cyst fluid may be helpful ⁷⁹,⁸⁰. Pseudocyst fluid has a low viscosity and has low concentrations of carcinoembryonic antigen (CEA) and mucin-like carcinoma associated antigen (MCA), with elevated amylase. Mucinous lesions are generally viscid with high CEA levels, distinguishing them from serous lesions. Higher values of CEA, CA 125 and CA 15-3 are typical of mucinous carcinomas. Serous cystadenomas have low viscosity, usually low levels of pancreatic enzymes and low levels of CEA, NB70/K, CA 72-4 and CA 15.3. The cyst fluid values of CA 125 and CEA are usually higher in mucinous cystadenocarcinoma than in similar benign and borderline lesions but this is variable.

When assessing a potential cystic neoplasm the clinical presentation, radiological appearance and biochemical and cytological findings must all be considered in conjunction. Most reported cohorts are small. In a report from Germany of eight patients with cystic neoplasma, four patients had benign serous cystadenoma, one a mucinous cystadenoma and three had mucinous cystadenocarcinomas ⁸⁰. Tumour markers were non-specific and CA 19-9 was elevated in two patients, only one of whom had cystadenocarcinoma. FNA helped distinguish between serous and mucinous cystadenoma in four of six cases.

Lympho-epithelial cysts are rare true pancreatic cysts lined by squamous epithelium and surrounded by mature lymphoid tissue, which arise in middle-aged men, and are usually asymptomatic ⁸¹,⁸². Excision is generally curative. FNA may be diagnostic and demonstrates numerous anucleated squamous cells on a background of keratinous and amorphous debris. A few benign nucleated squamous cells and plate-like cholesterol crystals may also be present.

Miscellaneous tumours

Solid pseudopapillary tumours of the pancreas account for 1% of pancreatic tumours, and typically occur in young women ⁸³. They are smooth and encapsulated, appear to have a good prognosis once enucleated and have distinctive cytology permitting diagnosis on FNA. Their cytological appearance is characterised by papillary fronds of regular cells surrounding vessels, with their grooved nuclei separated from the vessel by a thin rim of eosinophilic cytoplasm. Hyaline globules may be present.

Pancreatic adenosquamous carcinoma is an unusual histological subtype of NET of the pancreas with a better clinical course than adenocarcinoma. The tumour demonstrates both a glandular and a squamous component, and can be diagnosed by FNA. The glandular component often reacts with CEA ⁸⁴

Conclusion

As the mortality rate of pancreatic resection has steadily decreased, the reluctance of surgeons to proceed to operation without histological confirmation of pancreatic cancer has tended to disappear. The low sensitivity of brushing techniques and the decline of diagnostic ERCP, as well as ongoing concerns about the safety of percutaneous FNA, have tended to diminish the importance of biopsy. Many experienced pancreatic surgeons feel comfortable to proceed to pancreatic resection without biopsy if the history, preoperative and intra-operative findings are all consistent with a malignant diagnosis. Tissue diagnosis is generally considered necessary when non-operative palliation is being proposed. Biopsy is necessary for counselling and for recognition of rare tumours that might merit operation, which would not have been contemplated for adenocarcinoma, or appropriate systemic therapy. FNA may be appropriate in the setting of NETs and cystic malignancies. However, the interpretation of cytology may be difficult and must be combined with a detailed clinical, radiological and biochemical assessment of these lesions. Two new developments, i.e. the advent of EUS and neo-adjuvant therapy, may challenge these assumptions. The growing use of adjuvant and neo-adjuvant therapy may necessitate a firm tissue diagnosis. The development of EUS-guided FNA allows accurate needle placement with an apparent reduced risk of peritoneal seeding of tumour cells – eliminating oncological concerns about percutaneous biopsy. Nonetheless, there still remains a degree of uncertainty associated with all of these techniques. The negative predictive value is variable and not high enough for the certain exclusion of malignancy. Therefore in the presence of a potentially resectable lesion which is highly suspicious of malignancy the clinician should consider operation even if preoperative histology or cytology is negative.