Chinese-German Journal of Clinical Oncology

   April 2007, Vol. 6, No. 2, P107–P116

DOI  10.1007/s10330-007-0036-1

Pancreatic cancer – Diagnostics: CT, MRI Goetz M. Richter Department Diagnostic Radiology der University Hospital Heidelberg, Im Neuenheimer Feld 110 69120 Heidelberg, Germany Received: 14 February 2007 / Revised: 28 February 2007 / Accepted: 6 March 2007

Introduction The introduction of modern imaging and fast imaging processing has tremendously improved detection and staging of pancreatic cancer. Besides progress in surgical techniques and handling the advances of both computerized tomography (CT) and magnetic resonance imaging (MRI) have remarkably contributed to higher resection rates, better patient selection and clinical management. Both imaging modalities have reached a high reliability in defining local spread and potential vascular involvement. Already 1997 Trede reported higher resection rates of up to 40% as a direct result of high standard MRI [1]. Our own research group could confirm this for CT in various studies applying the so-called hydro-CT technique for the detection and staging of pancreatic cancer [2, 3]. This progress in imaging can be highlighted and summarized with the following key points: 1. The accuracy in tumor detection reaches approximately 95%; 2. The diagnosis of non-resectability is close to 100%; 3. Both CT-angiography and MR-angiography have diagnostic accuracy in defining vascular infiltration higher than classical catheter angiography; 4. Detection rates of metastatic disease of the liver are well above 95%; 5. Dedicated pancreatic centers have developed established radiology and surgery interaction criteria for resectability facilitating standardized reporting. As already mentioned above the significant progress of both major imaging modalities CT and MRI have raised the question which is preferred when. Many studies have tried to show more benefit from one modality versus the other [4–7]. In our own direct comparison study we were able to show that on the basis of accuracy in detection and staging alone neither modality reaches superiority over the other. However, patient comfort, length and Correspondence to: Goetz M. Richter. Email: [email protected]

complexity of examination and, last bot not least, cost of examination are strongly in favour of CT [8]. This has been corroborated by various very recent studies relying on meta-analysis or evidence-based approaches [9–11]. In the following chapters we describe our experiences and concepts on the actual status, progress and potential outlook of radiodiagnosis in pancreatic cancer with special reference to pancreatic adenocarcinoma and to endocrine and cystic tumors which, in summary, is the result of more than one decade of intensive radiological-surgical cooperation and interaction.

Tumor detection and differentiation Ultrasound  US, as well, has undergone tremendous changes and much progress has been made in time and image resolution. Doppler mode and Doppler colour mode have been added. Furthermore, echo-enhancing contrast material has been introduced into pancreatic imaging. The present state-of-art scanners have high resolution 512 channel technology and on-line Doppler and Duplex equipment. Particularly, in combination with tissue harmonic imaging high accuracy rates for tumor detection and differentiation of up to 90% have been reported [12, 13]. With echo-enhanced imaging tumor visualization might be improved because of the highly different tissue perfusion pattern of normal pancreas versus tumor tissue, the latter in case of ductal adenocarcinoma dominated by a relative low perfusion intensity as a result of widespread tumorous desmal component. Furthermore, “indirect” tumor detection might be achieve by US by identifying a double duct sign suspicious of malignant origin within the pancreatic parenchyma. Unfortunately, the major draw back of US in general, is not substantially changed or surpassed by any technologic progress. Obesity or just overweight, which are a pertinent feature of the western population prevents high detection rates on a real world intention-to-diagnose ba-

108

sis. The operator dependent character, furthermore, will never change. In a selected group of patients applying a so-called hydrosonography technique, which includes filling of the stomach with high volumes of still water, we were able to reach detection rates of slightly higher than 80% in pancreatic adenocarcinoma [14]. However, this required an examination time of at least 30 min and a highly trained examiner. Echo-enhanced US has been specifically advocated also for endocrine tumors and the differential diagnosis of cystic tumors. In both entities a characteristic and mostly hypervascular pattern exists by contrast to the often hypovascularized ductal adenocarcinomas as compared to the surrounding tissue. Endocrine tumors are most often hypervascular as or their need for blood supply for endocrine functioning. In the malignant versions, though, this might be less pronounced. Cystadenomas frequently show many vessels along fibrotic strands. Data from prospective studies have demonstrated that based on these imaging criteria, the sensitivity and the specificity of echo-enhanced sonography in diagnosing the degree of differentiation of pancreatic masses are equal to, or greater than, 85% and 90%, respectively [13]. In conclusion, pancreatic tumors have a different vascularization pattern in echo-enhanced ultrasound. These characteristics can be used with high a diagnostic accuracy for differentiation. MRI In the past, MRI was of limited use in the diagnosis of pancreatic disease due to respiratory motion and cardiac pulsation artifacts, bowel peristalsis, and long acquisition times for T1-weighted and T2-weighted spin-echo and turbo spin-echo sequences. However, the lack of ionizing radiation and, nevertheless, the very positive earlyon judgement of a very prominent specialist in pancreatic surgery[1] in the mid nineties have led to an increased interest in surgery for pancreatic cancer. It was based on the introduction of new MR technology with fast sequence and 3-D angiographic imaging protocols (see below) and

www. springerlink. com/content/1613-9089

the add-on possibility of the visualization of the biliary and pancreatic duct by MRCP (magnetic resonance cholangiopancreatography) (Fig. 1). This allowed an examination of the entire area of interest including liver, pancreas, abdomen, retroperitoneal spaces etc., which was summarized as a “all-in-one” or “one-stop-shopping” examination. Particularly, the fast post-contrast (gadolinium-enhanced) angiographic sequences allowed to identify differences in the vascularization pattern similarly to CT and, simultaneously, describing the status of the major vessels, the ruling-out of liver metastasis or major lymph node spread. Hence, the state-of-the-art MRI of pancreatic neoplasms is optimally performed with 1.5 Tesla gradient systems using phased-array body coils to improve the signal-to-noise ratio, optimized with thin slice profiles and small fields of view. Latest MRI scanners are able to combine and manage up to 8 different coils and parallel imaging resulting in faster imaging protocols and much higher time and spatial resolution than just a few years ago. Breath-hold acquisitions are obtained with fast spin echo (FSE) or gradient echo (GRE) sequences. In many studies much emphasis has been put on the kinetics of specific vascular enhancement for tumor identification during early, venous and delayed phase scanning after administration of contrast material (gadolinium-DTPA). Elsayes demonstrated recently that the introduction of 3D gradient recalled echo (GRE) sequence such as volumetric interpolated breath hold examination (VIBE) has dramatically improved MR imaging by providing dynamic enhanced thin-slice images with fat saturation and high signal-to-noise ratio [15]. Post gadolinium-DTPA sequences are obtained in arterial, portal and delayed phases (Fig. 2). Half-Fourier rapid acquisition with relaxation enhancement (RARE) sequence is the sequence technology most frequently used (Fig. 1) for magnetic resonance cholangiopancreatography (MRCP) because ot its highest spatial resolution compared to other extremely T2 weighted projection methods [16]. The HASTE (half-Fourier snap-

Fig.  1  MRCP (RARE sequence) of a ductal adenocarcinoma within the head of the pancreas over a length of approximately 4 cm. Note the dilated pancreatic duct. The biliary duct is normal Fig.  2  Dynamic gadolinium-DTPA enhanced MRI. (a) 15s; (b) 90s after contrast administration. Improved tumor conspicuous during the course of the examination

Chinese-German J Clin Oncol, April 2007, Vol. 6, No. 2

109

Fig.  3  Giant mucinous cystadenoma of the tail of the pancreas. 42 years old female, abdominal symptoms. (a) Coronal RARE sequence. Note the multiple large cysts; (b) Axial HASTE sequence. Upper portion, fine delineation of septae; (c) Axial HASTE sequence. Lower portion, fine delineation of septae; (d) Non-enhanced T1 weighted (GRE) sequence. Note the differences in signal intensity of the cyst fluid. Probably, because of different protein and mucinous content; (e) 3D sequence, T1 and fat suppressed post-gadolinium-DTPA. Arterial phase, very faintly contrasted cyst walls; (f) 3D sequence, T1 and fat suppressed post-gadolinium-DTPA. Venous phase, no signs of infiltratio, splenic vein only slightly displaced

shot turbo spin echo) sequence is such an extremely T2 weighted breath-hold sequence, has the advantage of virtually no motion artifact in a normal breathing patient and can be used for maximum intensity projections (MIP) because of its high spatial resolution. This sequence is extremely helpful for all cystic lesion and fine septal delineations (Fig. 3, 4). Lately, another type of contrast material, Mangatrifodipir (Mn-DPDP, Teslascan), has been advocated for its use in MRI for pancreatic disease. This is a cell-specific contrast material. Its uptake is found with normal hepatocytes and the epithelial cells of the pancreas. Therefore it functions as a negative T1 contrast material for any pancreatic lesion and is used for enhancing the sensitivity of lesion detection [17]. Because of this fairly unspecific nature it is still not widely used. It might bear potential for the future, though, for the identification of very small solid pancreatic lesions [17–19]. Magnetic resonance imaging has also considerable potential in characterizing pancreatic masses. Certain features can be used by the radiologist to establish a definitive diagnosis for most pancreatic tumors including ductal adenocarcinoma, islet cell tumors, solid and papillary epithelial neoplasms, micro- and macrocystic adenoma, and metastases. Recognition of these tumors on imaging is important since it often changes the treatment approach and may obviate the need for surgery. Neuroendocrine tumors (Table 1), by contrast to duc-

tal adenocarcinoma, in general, are well perfused entities, probably, because of their need for fast blood supply for endocrine activity. Because of the pluripotent nature of the cells of origin, they may express any of a number of different polypeptide hormones and which determine their “functional” names, accordingly (see also Table 1). Hyperfunctioning tumors tend to present when the lesion is still small because of symptoms related to the secreted hormone. Nonhyperfunctioning tumors, on the other hand, more often present after the lesion has reached a significant size, with symptoms related to mass effect or metastases. The majority of nonhyperfunctioning tumors are malignant, although the proportion of malignant tumors varies greatly among hyperfunctioning subtypes. Islet cell tumors have an increased prevalence in patients with von Hippel-Lindau disease and multiple endocrine neoplasia type I, in whom multiple lesions and extrapancreatic location are frequent. Islet cell tumors are often difficult to detect and localize on imaging studies due to their small size and variable imaging features. MRI has been advocated for detecting small lesions and metastases because of its optimal contrast resolution and the ability to easily perform dynamic imaging and also in combination with the use of Mangatrifodipir [20]. However, most often in malignant neuroendocrine tumor differentiation, a more cystic appearance becomes more prominent disabling easy detection by the before mentioned perfusion and contrast uptake pattern [21]. Recently, many attempts

110

www. springerlink. com/content/1613-9089

Table  1  Islet cell tumors: hormonal activity, lesion characteristics (modified from [44, 45]) Preferred localization

Preferred appearance

Malignancy

Preferred imaging

Insulinoma

No preference

Solitary, small size

10%

Gastrinoma

Neck/head

Multiple, medium size

60%–90%

Glucagenoma

Body/tail

Multiple, large size

50%–90%

Vipoma

Neck/tail

Multiple, indeterm

40%–70%

Somatostatinoma

Neck/tail

Solitary, indeterm

70%–80%

Non-functioning tumor

No preference

Solitary, large calfications

60%–80%

High resolution FS-T1,T2 MRI Dynamic CE-3D MRI dual phase MDCT High resolution FS-T1,T2 MRI Dynamic CE-3D MRI dual phase MDCT High resolution FS-T1,T2 MRI Dynamic CE-3D MRI dual phase MDCT High resolution FS-T1,T2 MRI Dynamic CE-3D MRI dual phase MDCT High resolution FS-T1,T2 MRI Dynamic CE-3D MRI dual phase MDCT Native CT (calfication) dual phase MDCT High resolution FS-T1,T2 MRI Dynamic CE-3D MRI

have been made to identify the best imaging method both for benign and malignant endocrine tumors. So far, neither MRI nor CT have gained definite superiority [22]. While pro MRI the higher soft tissue contrast has been advocated pro CT the higher spatial resolution and less artifacts was mentioned (Table 1). In routine, probably, the method with which the radiologist is more familiar might be preferred [23]. Cystic tumors of the pancreas (Fig. 3, 4) are less frequent than solid lesions and are often detected incidentally, as many of these lesions are small and asymptomatic. However, they may be associated with pancreatitis or have malignant potential. With advancements in diagnostic imaging, cystic lesions of the pancreas are being detected with increasing frequency. Many lesions can cause a pancreatic cyst, most being non-neoplastic while approximately 10% are cystic tumors, ranging from benign to highly malignant tumors. With increasing experience it is becoming clear that the prevalence of pseudocysts among cystic lesions of the pancreas is lower than usually presumed. Cystic tumors of the pancreas are formed by serous or mucinous structures showing all stages of cellular differentiation. According to the WHO classification, they can be subdivided on the basis of their histological type and biological behavior into benign tumors, borderline tumors, and malignant tumors. Cystic pancreatic tumors can be subdivided into peripheral (serous cystadenomas, mucinous cystic tumors, solid and papillary epithelial neoplasms, cystic islet cell tumors), which do not communicate with the main pancreatic duct, and ductal tumors (mucinous tumor), according to their site of origin and, essentially, do communicate. On the basis of imaging criteria alone, it can be very difficult to differentiate nontumoral cystic lesions from neoplastic ones. Most recent

studies favour MRI over CT attributable to its high sensitivity for fluid content and ductal delineation [20]. However, in large series with surgical and pathologic correlation neither of the two was found superior [24, 25]. MRI findings of cystic tumours are as follows: Serous cystadenoma  The vast majority of these benign tumors demonstrate a polycystic or microcystic pattern consisting of usually more than six cysts that range from a few millimeters up to 2 cm in size and which should be identified with bright signal intensity on T2 weighted images with or without breath hold sequences. A fibrous central scar with or without a characteristic stellate pattern of calcification might be found and a “honeycomb” or “spongy” pattern. Pancreatic ductal dilatation at MRCP or T2 weighted sequences is not a common finding in these tumors. Mucinous cystadenomas (Fig. 3)  They predominantly involve the body and tail of the pancreas, and, although they do not communicate with the pancreatic duct, they can cause partial pancreatic ductal obstruction which is typically demonstrated at MRCP and T2 weighted imaging Also the multilocular macrocystic character should be identified with MRI. Unfortunately, the cysts might occasionally contain debris or hemorrhage thereby significantly altering the signal intensities typically generated with T2 and T1 weighted image sequences (Fig. 3). When increased signal intensity is found both at T1and T2 weighted images hemorrhagic components might be responsible. Another typical feature, “eggshell” calcification is, usually missed at MRI (see respective chapter on CT). IPMN (intraductal papillary mucinous neoplasms, Fig. 4)  Main duct, branch duct (side-branch), or mixed

Chinese-German J Clin Oncol, April 2007, Vol. 6, No. 2

111

Fig.  4  Malignant duct type IPMN, 63 year male. (a) Coronal HASTE sequence. Main duct dilatation 16 mm. Liver cyst; (b) Coronal HASTE sequence. Main duct dilatation until tail region, several papillary projections; (c) Axial HASTE sequence. Preampullary solid nodule; (d) 3D GRE sequence post gadolinium-DTPA. Note the solid intraductal nodule at the neck Fig.  5  Very small ductal adenocarcinoma in the head with subsequent duct dilatation. Note the small hypoattenuating tumor mass during venous phase imaging. (a) and (b) contiguous 3 mm slices Table  2  Characteristics of IPMN (modified from [34]) Morphology

Relevance for malignancy

Main duct involved Marked dilatation of main duct (> 13 mm) Diffuse or multifocal involvement

Main duct type alone or combined type IPMN high rates of malignancy Marked dilatation of main duct (7–15 mm) high rates of malignancy Involvement of more than one pancreatic segment (uncinate process, head, body, tail) high rates of malignancy In both case high rates of malignancy (nodule size > 3 and up to 10 mm) In branch duct type or combined type IPMN large masses (> 2 cm) high rates of malignancy Sign of local invasion, highly suspicious for malignancy High predictive value for malignancy

Large mural node or solid mass Size of the tumor Bile duct dilatation Calcified intraluminal contents

IPMNs, depending on the site and extent of involvement have been identified (Table 2). Side-branch IPMN or a mixed IPMN (in which a side-branch tumor extends to the main pancreatic duct) can have the MRI features of a complex pancreatic cyst, making clear-cut distinction from a mucinous cystic neoplasm difficult. Identification of a septated cyst that communicates with the main pancreatic duct is highly suggestive of a side-branch or mixed IPMN. This should be well visualized, like in the other cystic lesions, on MRCP and high resolution T2 weighted imaging. In main duct IPMN, which is characterized by a much higher malignancy rate [26], MRI might be helpful in the discrimination between malignant and benign. The presence of a dilated main pancreatic duct (> 13 mm), mural nodules (= papillary projections), thickened wall and peripancreatic haziness may be used as independent

predictive signs of malignancy when identified at high resolution T2 weighted imaging or 3D gradient recalled echo imaging (Fig. 4). The latter should provide imaging of the vascularization kinetic while identifying the main duct dilatation in its length and dimension and the papillar projections by its contrast uptake during dynamic scanning (Fig. 4). In patients with the main duct type, the main pancreatic duct, usually is significantly narrower when associated with benign as compared to malignant tumors. Irie and colleagues in a series of 34 patients found that all malignant tumors showed diffuse main pancreatic duct dilatation, whereas all benign tumors showed segmental dilatation. Among patients with branch duct type, the cyst was smaller when it was a benign rather than malignant tumor. All but one malignant tumor showed mild associated main pancreatic duct dilatation, whereas

112

www. springerlink. com/content/1613-9089

Fig.  6  T3 ductal adenocarcinoma. 16 MDCT. Blurred tumor margins and lack of hypoattenuation at CT resulting from previous papillotomy and stent insertion and subsequent local inflammation (peripapillary)

benign tumors were not associated with main pancreatic duct dilatation. Filling defects suggested malignancy, although half of the malignant tumors had no filling defects (Table 2) [27]. Computerized tomography Multidetector CT technology takes advantage of sufficiently reduced image acquisition time for multiphasic thin slice acquisitions of the pancreas and liver in a single breath-hold with superior contrast bolus utilization. Recently, 64 row multidetector CT has become more and more available which enables a tremendous improvement both in spatial and time resolution. Generally, dual phase imaging in arterial and portal venous phases is used for evaluating suspected pancreatic neoplasms and depicting peripancreatic/peritumoural vascular anatomy. For dual phase imaging in arterial and portal venous phases for evaluating pancreatic neoplasms, 100–130 mL of non-ionic contrast media (≥ 300 mg Iodine/mL) is injected at 3–5 mL/s and images are acquired following delays of approximately 15–30 s for arterial and 50–70 s for portal venous phases, respectively. Contiguous thin sections are obtained allowing overlapping reconstruction. Arterial phase CT imaging is used, furthermore for 3D reconstructions to generate virtual CT angiographic images (CTA). Multiplanar three-dimensional reconstruction techniques including volume rendering, maximum intensity projection and shaded surface display provide comprehensive information about the relationships and possible involvement of vascular structures in pancreatic neoplasms, and the degree and level of dilatation of pancreatic and biliary ducts (see also below). The surrogate for tumor identification in CT is comparable identical to gadolinium-DTPA-enhanced MRI. Because of the predominantly desmoplastic tissue architecture of ductal adenocarcinoma the tumor mass shows hypoattenuation compared to the normal pancreatic tissue and is identified as an ill-defined hypodense lesion within the pan-

creas (Fig. 5). Serial kinetic studies have focused on the timing to reach highest lesion to parenchyma contrast [28]. Maximal enhancement of the normal pancreatic parenchyma occurs during the pancreatic parenchymal phase. Maximum tumor-to-parenchyma attenuation differences during a (earlier) parenchymal phase and (later) portal venous phase are equivalent but greater than that during the arterial phase and so is tumor conspicuity. However, as this is related to ductal adenocarcinoma only, there is agreement that imaging during arterial phase is required to allow as much tumor detection capabilities as possible with special regard to all hypervascular lesion as already outlined above in the chapter on MRI for tumor detection and differentiation. Our group has shown that the so-called hydro-CT technique based on stomach distension by > 1l still water combined with pharmacologic intestinal paralysis reaches a approximately 95% detection rate in ductal adenocarcinoma of the pancreas [2, 3, 8]. This compares very well with a very recent meta-analyis by Bipat and colleagues who were able to analyze 68 articles and found 91% sensitivity for tumor detection with CT (contrast enhanced imaging). In this study CT was superior to the other imaging modalities (US, MRI) [9]. Two diagnostic drawbacks of tumor detection by CT, though, are noteworthy: (a) In patients who had undergone papillotomy and subsequent stent placement for relief of obstructive jaundice the peripapillary inflammatory reaction and subsequent increased hypervascularity might extend deep into the parenchyma of the pancreatic head (Fig. 6) thereby “obscuring” the tumor surrogate of hypoattenation during dynamic scanning [3]. In real world decision and diagnosis making this might be more an academic problem as the tumor diagnosis would be readily made through the ERCP finding of a double duct sign (Fig. 6); (b) In patients with chronic head pancreatitis and, furthermore, in duodenal ulcerous disease penetrating to the pancreatic head tumor mimikry might entail the risk of false positive tumor diagnosis [3]. As already mentioned

Chinese-German J Clin Oncol, April 2007, Vol. 6, No. 2

113

Fig. 7  Benign serous cystadenoma. 57 year old female. (a) 64 MDCT. Arterial phase, lower portion of the tumor in the uncinate process. Note the fine small cysts, no signs of vascular infiltration; (b) 64 MDCT. Arterial phase, upper portion of the tumor in the uncinate process. Cyst size aroun 5–10 mm small, no signs of vascular infiltration; (c) 64 MDCT. Venous phase same slice (a); (d) 64 MDCT. Venous phase same slice (b)

Fig.  8  Small malignant mucinous cystadenoma, 61 year old female. (a) 64 MDCT. Arterial phase. Note the two peripheral calcifications in the hypoattenuation lesion in the head of the pancreas. 10 mm diameter; (b) Same slice as (a) venous phase; (c) Virtual 3D angiography, reconstructed from 0.6 mm slices. No major vessel infiltration

in the chapter on MRI several researches have stressed the role of MRI in detecting very small tumors (< 2 cm in diameter). However, MDCT with 16 and more detectors rows provides enough high spatial and time resolution (Fig. 5) when combined with scanning protocols optimally trimmed and timed for highest lesion conspicuity that sensitivity levels in the detection of small pancreatic masses are close to 80% and the specificity reaches 100% [29, 30] . In islet cell tumors, like in ductal adenocarcinoma, tumor detection and differentiation is not inferior to MRI [22] . The imaging surrogates are identical. CT canning during arterial phase readily depicts the hypervascular nature of the tumors. In malignant and little or non-functioning variants the limitations in lesion conspicuity are identical to MRI [31, 32]. Non-functioning tumor might lack hypervascularity and highly malignant tumors might show even cystic appearance when necrosis is predominant. As specific types prefer specific localizations in the pancreas the radiologist needs to be familiar with this and direct the attention and imaging protocol accordingly. In cystic tumors of the pancreas (Fig. 7, 8), as outlined in the previous chapter, MRI has played a dominant role during the last years because of its ability to depict fine septae and fluid content in various protein concentrations. With increased spatial resolution and progress in tissue contrast with MDCT and particularly as provided by the latest generation of 64 detector row CT new studies have emerged recently showing high diagnostic yield

in serous or mucinous cystic adenomas. On unenhanced CT, serous adenomas appear as hypodense masses that frequently show central calcification. Contrast enhancement of the septations results in a honeycombing appearance due to the presence of tiny cysts (Fig. 7). In addition, a characteristic central fibrotic scar may be appreciated on contrast enhanced CT scan. Mucinous cystic neoplasms (Fig. 8) appear as round to ovoid, externally smooth, near-water-density cystic lesions. Amorphous calcification, septations and solid excrescences may be seen. Contrast enhanced CT demonstrates the enhancement of cystic walls and thin, straight or curvilinear septations. Particularly the capability of CT to depict the fine “eggshell” (compare Fig. 8, no eggshell type) type of calcifications has been mentioned as a powerful diagnostic tool in discriminating benign from malignant which, in mucinous cystic neoplasm is highly important because of the much higher likelihood of malignancy as compared to serous cystic adenomas [24, 33]. Similarly like in islet cell tumors, with the advent of the latest scanner generation the diagnostic accuracy of CT in IMPN has reached the level of MRI as recently impressively outlined by Chiu and colleagues [26] and in a pictorial overview by Kawamoto and colleagues [34]. Again, the same surrogates for tumor detection and differential diagnosis apply as for MRI. Newer developments in CT–for example, high-resolution multislice helical imaging–in combination with postprocessing techniques such as curved 2D reformations can provide details, particularly,

114

Fig.  9  Course during neoadjuvant therapy of a primary T4 ductal adenocarcinoma (48 Gy, Gemcitabine) and liver metastasis. (a) baseline; (b) control after 2 months, clear tumor regression locally and in the liver

in imaging studies of IPMN, including the presence of a ductal communication, similar to the findings on magnetic resonance cholangiopancreatography. The presence of mural nodules (papillary projections) and a segmental or diffuse dilatation of the main pancreatic duct greater than 15 mm in diameter has been reported as highly indicative of malignancy, the latter being highly specific for main duct IPMN [26, 35]. The usefulness of curved 2D reformatting algorithms was underlined recently in a study by Sahani and colleagues. They compared the accuracy of MDCT (16 row detector system) with that of MRCP for characterization of IPMN as malignant and used the pathologic examination as reference standard. They found an excellent correlation between both modalities. Cyst communication was seen in 20 and 21 of 24 branch pancreatic duct IPMNs with CT and MRCP, respectively. Sensitivity, specificity, and accuracy for detection of malignancy were 70%, 87%, and 76% (CT) and 70%, 92%, and 80% (MRCP) [36].

Staging and resectability Complete staging of pancreas tumors requires the correct determination of a variety of morphological parameters. This is achieved in very different ways and efforts (operator and technical equipment) by the different imaging techniques. There is still no imaging method (!) which reliably depicts peritoneal carcinosis directly and which detects micrometastases in the liver, or can distinguish between specific and unspecific enlargement of peripancreatic lymph nodes 1 to 1.5 cm in size. In clinical reality, however, such drawbacks play just a minor role. Small and easily resectable tumors seldom display one or more of the negative factors described. Conversely, far advanced carcinomas are usually inoperable due to their local vascular or perineural infiltration, which in turn should be detectable by state-of-the-art imaging, thus rendering existing and possibly undetected peritoneal infiltration or liver metastasis less significant. It might estimated that a percentage of approximately 1%

www. springerlink. com/content/1613-9089

–3% of patients with T3 stages might fall into that category. One further unsolved problem in staging concerns the standardization of generally accepted criteria for vessel infiltration of the superior mesenteric vein, the portal vein and the peripancreatic visceral vessels, although this is of fundamental significance for intraoperative handling (resection vs. bypass, radicality, etc.). What makes this even more difficult is that confusion often exists regarding terminology for resectability. In the sense of positive radicality, this cannot alone mean that the tumor simply can be removed, but rather that an R0 situation is attainable. Another point is the possibility of complete microscopic assessment, which might nor be guaranteed, when the tumor has been peeled or directly sliced away of the major vessel walls and potentially remaining tumor clusters do not necessarily become microscopically detectable. In the process of correct and and valuable radiologic decision making the relationship of the tumor spread and adjacent arterial and venous walls must be described according to length, width and extent in such a way that the surgically required level of radicality is clearly attained. It is, ideally, most expedient when the surgeon and the radiologist, together, at a work station of the corresponding modality, identify the tumor spread in such a way that an optimal solution in terms of surgical tactics might be achieved. Recently this has become even more important, since large studies have shown benefit from neoadjuvant therapies for patients with primarily unresectable adenocarcinom of the pancreas [37, 38]. Sonography For staging and evaluation US even with the latest scanner generation plays only a subordinate role in relation to the other imaging techniques, having even less significance than in tumor detection alone. Although the depiction of the relation of tumor to visceral vessels by means of the duplex technique as well as echo-enhancing administration of contrast medium is theoretically much better using modern instruments than was the case a few years ago [13], imaging quality is generally not sufficient for clarification of the problematic vessel infiltration zones already described above. This was corroborated in a comparison study of various imaging modalities by Bipat and colleagues who found a statistically less sensitivity of US versus MDCT: For determining resectability, sensitivities of helical CT, conventional CT, MRI, and US were 81%, 82%, 82% and 83% and specificities were 82%, 76%, 78%, and 63% respectively. Specificity of US was significantly lower compared with helical CT (P = 0.011) [9]. As however complete occlusion of the main peripancreatic vessels can usually be recognized reliably, confirmation of inoperability in the case of very advanced tumors might be provided. Further limitations in usefulness for staging

Chinese-German J Clin Oncol, April 2007, Vol. 6, No. 2

arise from the much poorer accuracy in the diagnosis of metastases compared to the other imaging techniques. Lastly, the already above mentioned increased operator’s time and effort required also holds true for tumor staging. Magnetic resonance imaging Since the first positive reports concerning the value of magnetic resonance imaging in the staging of pancreatic carcinoma, many studies have been presented emphasizing the outstanding or important role of the procedure [39]. Arguments in favor of the role of magnetic resonance imaging concern the following imaging possibilities: representation of peripancreatic arteries and veins in time-limit released thin section 3D sequences with contrast medium enhancement (MR-A), detection of retroperitoneal tumor infiltration using fat suppression sequences, ductal visualization by MRCP and finally the high potency in terms of differential diagnosis of liver lesions [40]. This evidence is qualitatively completely correct. It is based in large part on “blinded reading” studies with ROC analyses. A serious flaw in many such studies is however their, usually retrospective approach and the high number of surgically explored and proven tumor, rendering the probability of error regarding unresectability nonexistent. True comparison studies as to its value compared to computer tomography with equivalent modern technology are very rare. In our own direct comparison study, as already mentioned in the introduction chapter, we specifically failed to show superiority of MRI over state-of-the-art CT. However, patient comfort, length and complexity of examination and, last bot not least, cost of examination are strongly in favour of CT [8]. In a very recent prospective study comparing 16 row MDCT and a 1.5 Tesla TIM scanner (Siemens), again we could not confirm any advantage of MRI versus CT (unpublished data). Reports that the MR-A can reach the diagnostic accuracy of conventional angiography are questionable. This is impossible due to principle natural laws of physics such as local or contrast resolution. As with sonography, however, complete vessel occlusion can be detected clearly. Computerized tomography As already discussed in depth in the previous chapter CT remains the work horse in the diagnostic work-up for resection of pancreatic neoplasm [9, 11, 19, 41]. And, in particular, with the advent of 64 detector row scanners, CT has taken a huge, if not revolutionary, step into the future of medical imaging. Earlier generations of multidetector CT scanners (enabling the acquisition of up to 16 sections) could achieve faster scanning speeds or higher z-axis resolution at traditional helical CT speeds. Extreme multidetector CT scanners enabling the acquisition of 64 or more sections can produce isotropic spatial

115

resolution while simultaneously delivering exceptional temporal resolution with excellent z-axis coverage (4–8 cm/sec) from which all aspects of imaging of pancreatic neoplasm benefit. Axial anatomic viewing, the traditional mode of CT image interpretation, was not established by choice or scientific consensus but rather by the necessity of CT acquisition geometry. Extreme multidetector CT scanners have changed this by enabling coronal, sagittal, and nonorthogonal multiplanar image viewing. Therefore, the same dynamic adjustment of the viewing orientation to optimize interpretation accuracy as in MRI is possible. This, in summary, applies very significantly for precise vessel infiltration both of the venous and arterial system adjacent to the pancreas and crucial for resection [42] . While axial source images depict the involvement of the superior mesenteric vein, sagittal reformats are best for demonstrating superior mesenteric artery involvement. Coronal reformats help in demonstrating local extension to the stomach and duodenum. Standard oral contrast media for pre-operative assessment of pancreatic neoplasm and peripancreatic vasculature is unnecessary. Instead, a negative oral contrast agent such as water provides optimum visualization of the duodenal ampulla and duodenal–pancreatic interface, aiding detection of duodenal/paraduodenal tumour invasion and terminal bile duct calculus as we and others have shown in various studies [2, 3, 8, 43]. The overall accuracy in assessment of resectability and unresectability is approximately 95%.

References 1. Trede M, Rumstadt B, Wendl K, et al. Ultrafast magnetic resonance imaging improves the staging of pancreatic tumors. Ann Surg, 1997, 226: 393–405. 2. Richter GM, Simon C, Hoffmann V, et al. Hydrospiral CT of the pancreas in thin section technique. Radiologe, 1996, 36: 397–405. 3. Richter GM, Wunsch C, Schneider B, et al. Hydro-CT in detection and staging of pancreatic carcinoma. Radiologe, 1998, 38: 279–286. 4. Nishiharu T, Yamashita Y, Abe Y, et al. Local extension of pancreatic carcinoma: assessment with thin-section helical CT versus with breath-hold fast MR imaging – ROC analysis. Radiology, 1999, 212: 445–452. 5. Miller FH, Rini NJ, Keppke AL. MRI of adenocarcinoma of the pancreas. AJR Am J Roentgenol, 2006, 187: 365–374. 6. Pamuklar E, Semelka RC. MR imaging of the pancreas. Magn Reson Imaging Clin N Am, 2005, 13: 313–330. 7. Saisho H, Yamaguchi T. Diagnostic imaging for pancreatic cancer: computed tomography, magnetic resonance imaging, and positron emission tomography. Pancreas, 2004, 28: 273–278. 8. Grenacher L, Klauss M, Dukic L, et al. Diagnosis and staging of pancreatic carcinoma: MRI versus multislice-CT – a prospective study. Rofo, 2004, 176: 1624–1633. 9. Bipat S, Phoa SS, van Delden OM, et al. Ultrasonography, computed tomography and magnetic resonance imaging for diagnosis and de-

116

10. 11. 12.

13. 14. 15.

16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

www. springerlink. com/content/1613-9089 termining resectability of pancreatic adenocarcinoma: a meta-analysis. J Comput Assist Tomogr, 2005, 29: 438–445. Katz MH, Savides TJ, Moossa AR, et al. An evidence-based approach to the diagnosis and staging of pancreatic cancer. Pancreatology, 2005, 5: 576–590. Michl P, Pauls S, Gress TM. Evidence-based diagnosis and staging of pancreatic cancer. Best Pract Res Clin Gastroenterol, 2006, 20: 227–251. Rickes S, Unkrodt K, Neye H, et al. Differentiation of pancreatic tumours by conventional ultrasound, unenhanced and echo-enhanced power Doppler sonography. Scand J Gastroenterol, 2002, 37: 1313– 1320. Rickes S, Monkemuller K, Malfertheiner P. Contrast-enhanced ultrasound in the diagnosis of pancreatic tumors. Jop, 2006, 7: 584–592. Simon C, Hoffmann V, Richter GM, et al. Hydrosonography of the pancreas. Initial results of a pilot study. Radiologe, 1996, 36: 389–396. Elsayes KM, Narra VR, Abou El Abbass HA, et al. Pancreatic tumors: diagnostic patterns by 3D gradient-echo post contrast magnetic resonance imaging with pathologic correlation. Curr Probl Diagn Radiol, 2006, 35: 125–139. Irie H, Honda H, Tajima T, et al. Optimal MR cholangiopancreatographic sequence and its clinical application. Radiology, 1998, 206: 379–387. Zanello A, Nicoletti R, Brambilla P, et al. Magnetic resonance with manganese-DPDP (mangafodipir) of focal solid pancreatic lesions. Radiol Med (Torino), 2004, 108: 194–207. Romijn MG, Stoker J, van Eijck CH, et al. MRI with mangafodipir trisodium in the detection and staging of pancreatic cancer. J Magn Reson Imaging, 2000, 12: 261–268. Kalra MK, Maher MM, Mueller PR, et al. State-of-the-art imaging of pancreatic neoplasms. Br J Radiol, 2003, 76: 857–865. Schima W. MRI of the pancreas: tumours and tumour-simulating processes. Cancer Imaging, 2006, 6: 199–203. Herwick S, Miller FH, Keppke AL. MRI of islet cell tumors of the pancreas. AJR Am J Roentgenol, 2006, 187: 472–480. Reznek RH. CT/MRI of neuroendocrine tumours. Cancer Imaging, 2006, 6: 163–177. Noone TC, Hosey J, Firat Z, et al. Imaging and localization of islet-cell tumours of the pancreas on CT and MRI. Best Pract Res Clin Endocrinol Metab, 2005, 19: 195–211. Allen PJ, D’Angelica M, Gonen M, et al. A selective approach to the resection of cystic lesions of the pancreas: results from 539 consecutive patients. Ann Surg, 2006, 244: 572–582. Visser BC, Muthusamay VR, Mulvihill SJ, et al. Diagnostic imaging of cystic pancreatic neoplasms. Surg Oncol, 2004, 13: 27–39. Chiu SS, Lim JH, Lee WJ, et al. Intraductal papillary mucinous tumour of the pancreas: differentiation of malignancy and benignancy by CT. Clin Radiol, 2006, 61: 776–783. Irie H, Honda H, Aibe H, et al. MR cholangiopancreatographic differentiation of benign and malignant intraductal mucin-producing tumors of the pancreas. AJR Am J Roentgenol, 2000, 174: 1403–1408.

28. McNulty NJ, Francis IR, Platt JF, et al. Multi-detector row helical CT of the pancreas: effect of contrast-enhanced multiphasic imaging on enhancement of the pancreas, peripancreatic vasculature, and pancreatic adenocarcinoma. Radiology, 2001, 220: 97–102. 29. Bronstein YL, Loyer EM, Kaur H, et al. Detection of small pancreatic tumors with multiphasic helical CT. AJR Am J Roentgenol, 2004, 182: 619–623. 30. Shimizu Y, Yasui K, Matsueda K, et al. Small carcinoma of the pancreas is curable: new computed tomography finding, pathological study and postoperative results from a single institute. J Gastroenterol Hepatol, 2005, 20: 1591–1594. 31. Horton KM, Hruban RH, Yeo C, et al. Multi-detector row CT of pancreatic islet cell tumors. Radiographics, 2006, 26: 453–464. 32. Rappeport ED, Hansen CP, Kjaer A, et al. Multidetector computed tomography and neuroendocrine pancreaticoduodenal tumors. Acta Radiol, 2006, 47: 248–256. 33. Spinelli KS, Fromwiller TE, Daniel RA, et al. Cystic pancreatic neoplasms: observe or operate. Ann Surg, 2004, 239: 651–657. 34. Kawamoto S, Horton KM, Lawler LP, et al. Intraductal papillary mucinous neoplasm of the pancreas: can benign lesions be differentiated from malignant lesions with multidetector CT? Radiographics, 2005, 25: 1451–1468. 35. Brugge WR, Lauwers GY, Sahani D, et al. Cystic neoplasms of the pancreas. N Engl J Med, 2004, 3514: 1218–1226. 36. Sahani DV, Kadavigere R, Blake M, et al. Intraductal papillary mucinous neoplasm of pancreas: multi-detector row CT with 2D curved reformations – correlation with MRCP. Radiology, 2006, 238: 560– 569. 37. Beger HG, Poch B, Schwarz M, et al. Pancreatic cancer. The relative importance of neoadjuvant therapy. Chirurg, 2003, 74: 202–207. 38. Mornex F, Girard N, Delpero JR, et al. Radiochemotherapy in the management of pancreatic cancer – part I: neoadjuvant treatment. Semin Radiat Oncol, 2005, 15: 226–234. 39. Bluemke DA, Fishman EK. CT and MR evaluation of pancreatic cancer. Surg Oncol Clin N Am, 1998, 7: 103–124. 40. Ishiguchi T, Ota T, Naganawa S, et al. CT and MR imaging of pancreatic cancer. Hepatogastroenterology, 2001, 48: 923–927. 41. Li H, Zeng MS, Zhou KR, et al. Pancreatic adenocarcinoma: signs of vascular invasion determined by multi-detector row CT. Br J Radiol, 2006, 79: 880–887. 42. Fukushima H, Itoh S, Takada A, et al. Diagnostic value of curved multiplanar reformatted images in multislice CT for the detection of resectable pancreatic ductal adenocarcinoma. Eur Radiol, 2006, 16: 1709–1718. 43. Winter TC, Ager JD, Nghiem HV, et al. Upper gastrointestinal tract and abdomen: water as an orally administered contrast agent for helical CT. Radiology, 1996, 201: 365–370. 44. Mansour JC, Chen H. Pancreatic endocrine tumors. J Surg Res, 2004, 120: 139–161. 45. Mulkeen AL, Yoo PS, Cha C. Less common neoplasms of the pancreas. World J Gastroenterol, 2006, 12: 3180–3185.