Of all the GI malignancies, pancreatic adenocarcinoma, as shown in the images below, is the second most common cause of death from cancer. In clinical practice, pancreatic cancer is synonymous with pancreatic ductal adenocarcinoma, which constitutes 90% of all primary malignant tumors arising from the pancreatic gland. Radiologic characteristics of pancreatic adenocarcinoma are seen in the images below.
Tumors may arise from pancreatic ducts (99%) or from acinar cells (1%). More than 90% of pancreatic cancers appear in the late stage of disease; this observation emphasizes the role of radiology in early detection and determination of resectability of the tumor. The role of diagnostic imaging is to demonstrate the tumor and its relationship to surrounding vasculature, and the results determine the possibility of curative resection.
The diagnosis of pancreatic cancer is rarely made at an early stage. This is one of the main reasons for failing to achieve a cure in most patients.
See also Pancreatic Adenocarcinoma Imaging: What You Need to Know, a Critical Images slideshow, for more information on the imaging studies to use to identify and evaluate this disease.
There is much debate concerning the sensitivity and specificity of imaging investigations in the diagnosis and staging of pancreatic carcinoma.
Multisection computed tomography (CT) scanning is generally accepted to be the first line of investigation in a patient with suspected pancreatic cancer. The best imaging technique is determined by local availability and expertise, but this will nearly always be spiral CT (ideally multisection CT). The reasons for this preference include its wide availability, speed, thin sections, optimal enhancement, high spatial resolution, and consistently good images.
The importance of good CT technique cannot be overemphasized, and the key elements are the following: oral water as negative intraluminal contrast, 120-150 mL of iodinated contrast material intravenously administered at a rate of 3-4 mL/s, and scanning with thin (2- to 3-mm) collimation during pancreatic parenchymal phase (at 25-35 s) with the liver phase obtained at 60-70 s. [1, 2]
If the patient is clinically jaundiced and when biliary ductal dilatation is demonstrated on ultrasonographic (US) examination, endoscopic retrograde cholangiopancreatography (ERCP) is the next investigation of choice with a view to a drainage procedure. ERCP reliably demonstrates the point of obstruction. 
US is often the initial test in symptomatic patients. US is used for diagnosis rather than staging, although liver metastasis and ascites may be seen. Significant technical improvements in US have occurred. It may be used for problem solving in thin patients. Portal venous involvement may be more apparent on sonograms than on CT and/or MRI images, and liver lesions can be characterized as cystic or solid.
Magnetic resonance imaging (MRI) has improved considerably in recent years, and it continues to evolve. Studies comparing CT and MRI found that detection and assessment of resectability to be similar with both modalities.  MRI takes longer, it costs more, it is more complex, and it is limited by artifacts. The current role of MRI is probably problem solving. That is if the mass is not demonstrable with CT and US, MRI could be used to evaluate the pancreas in obstructive jaundice. MRI is also helpful in evaluating and characterizing liver lesions in patients with pancreatic cancer.
In the detection and staging of small tumors, endoscopic US (EUS) can be reliable when it is performed by experienced imagers. Previous studies have demonstrated a higher sensitivity and specificity with EUS than with other modalities, but these results probably reflect the use of suboptimal CT and MR techniques. Evidence suggests that EUS is similar to CT in diagnosis and staging of pancreatic cancer. EUS requires special endoscopic skills and expertise, and it is less readily available worldwide.
EUS-guided fine-needle aspiration (FNA) is safe and effective, especially for pancreatic head masses. EUS-guided FNA has sensitivity and specificity similar to that of CT-guided FNA cytology (FNAC).
Kamisawa et al found that diffusion-weighted MRI (DWI) can be used to differentiate autoimmune pancreatitis (AIP) from pancreatic cancer. In a study of 13 patients with AIP and 40 patients with pancreatic cancer, pancreatic cancers were detected as high-signal intensity areas, which were diffuse, solitary, or multiple in patients with AIP, but solitary in patients with pancreatic cancer. Pancreatic cancer more often had a nodular shape, while AIP more often had a longitudinal shape. Apparent diffusion coefficient (ADC) values were significantly lower in AIP than in pancreatic cancer, and an optimal ADC cutoff value of 1.075 x 10-3 mm2/s could be used to distinguish AIP from pancreatic cancer. 
In a study by Zaheer et al, findings found to be helpful for diagnosing AIP, versus pancreatic adenocarcinoma (PA), were diffuse enlargement, parenchymal atrophy, and the absence of pancreatic duct dilatation and focal mass. Findings helpful for diagnosing PA were focal mass and pancreatic ductal dilatation. The authors noted that misdiagnosis of PA in patients with AIP was due to focal mass, pancreatic duct dilatation, and pancreatic atrophy, whereas misdiagnosis of AIP in patients with PA was due to absence of atrophy, presence of diffuse enlargement, and peripancreatic halo. 
According to a study by Takakura et al, DWI and multidetector-row CT were found to be equivalent in distinguishing pancreatic cancer in high-risk patients with main pancreatic duct dilation. The accuracy with DWI was 84% compared to 86% in this small study group. The use of combined MR cholangiopancreatogrpahy and DWI allowed the authors to establish a high-risk population and detect tumors without a contrast medium. 
Limitations of techniques
The detection of a mass on imaging is nonspecific and 5-15% of pancreatic resections show benign pathology.
Transabdominal US (TAUS) has a relatively poor sensitivity, and its results are not satisfactory for assessment in approximately 20% of patients because of a poor acoustic window due to bowel gas.
MRI is sensitive in the detection and staging of pancreatic cancer with sensitivity and specificity similar to that of multisection CT. MRI involves expensive equipment and meticulous attention to the image technique. Other technical limitations are movement artifacts due to bowel peristalsis and breathing. Because the high sensitivity and specificity of MRI in detection and staging small tumors has not been achieved consistently and universally, debate continues about the superiority of MRI over CT.
Multisection CT should be used first in the detection of pancreatic adenocarcinoma. When CT findings are negative, MRI or EUS should be applied for detection and for the assessment of resectability. Although conventional angiography is obsolete in primary staging, it is occasionally required to assess peripancreatic vessels before surgery. Modern multislice CT scanners are capable of excellent depiction of arterial and venous branches. The role of MR angiography (MRA) in the assessment of mesenteric vessels prior to surgery is not firmly established, though some encouraging study results are reported.
Plain radiographs have no role in establishing a firm diagnosis of pancreatic carcinoma. Pancreatic calcifications may be seen concurrently in approximately 2% of patients who have chronic pancreatitis complicated by pancreatic carcinoma.
Upper GI barium studies may reveal an extrinsic impression of the mass on the posteroinferior aspect of the antrum of the stomach. This is known as antral pad sign. The medial margin of the descending duodenum may be pulled medially at the level of the ampulla, forming a reversed-3 appearance. This is known as Frostberg 3 sign. Infiltration of the duodenal mucosa may cause a spiculated appearance with irregularity and thickening of the duodenal mucosa. The changes also may represent a desmoplastic response to malignant disease. A nodular mass with an ampullary carcinoma may be observed.
Barium enema studies may demonstrate loss of normal haustral pattern from haustral padding along the transverse colon. The studies may show infiltration of the colon with an irregular or serrated contour to the bowel margin along the transverse colon, up to the level of splenic flexure. Tethering of colonic or small bowel margins resulting in asymmetry may occur from intraperitoneal seeding of pancreatic carcinoma.
Degree of confidence
In the presence of jaundice, barium studies have reasonably good specificity. The disease is usually advanced by the time the mass produces the characteristic signs on barium studies.
False-positive results may occur from a pseudocyst or other mass lesions producing similar appearances on the duodenal C-loop. Small masses may produce false-negative results.
Features suggestive of underlying pancreatic cancer include the following: alterations in morphology of the gland with abnormalities of CT attenuation values, obliteration of peripancreatic fat, loss of sharp margins with surrounding structures, involvement of adjacent vessels and regional lymph nodes, pancreatic ductal dilatation, pancreatic atrophy, and obstruction of the common bile duct (CBD). See the images below. [8, 9, 10, 11, 12, 13, 14, 15]
Abnormal morphology of the gland, such as a change in size, shape, or attenuation values, may include focal or diffuse enlargement, a focal lobulated eccentric mass, or decreased attenuation of the mass, respectively.
Macari et al determined that at portal venous phase dual-source dual-energy CT, pancreatic malignant-tumor conspicuity is greater at 80 kVp than with 120-kVp acquisition simulated with a weighted-average acquisition. The mean difference in attenuation for pancreatic tumors and adjacent normal pancreas was 83.27 +/- 29.56 (SD) HU at 80 kVp and 49.40 +/- 23.00 HU at weighted-average 120 kVp. At 80 kVp, contrast-to-noise ratio was significantly higher, as was duct visualization. 
A change in size is usually focal, and focal enlargement is seen in about 96% of patients with pancreatic adenocarcinoma. The size is an unreliable indicator of tumor, as a normally sized pancreatic head is consistent with a carcinoma of pancreas when atrophy of the body and tail is observed. This feature may be seen in pancreatic carcinoma in as many as 20% of patients. Focal enlargement also can occur in benign disease; thus, it is a nonspecific finding. Diffuse enlargement is less common and usually suggests pancreatitis.
By the time the mass has grown to produce a focal enlargement, the mass has often progressed to an inoperable stage. Some small tumors may cause biliary ductal obstruction and appear early. A change in the shape of the gland in the absence of enlargement is a more important sign and may suggest underlying tumor. Demonstration of fatty interstices within the mass suggests that the focal lobulation is of normal pancreas. If the fatty interstices are absent and if the mass is completely solid, it is more likely to be abnormal, and a biopsy is recommended.
The normal pancreas has an attenuation value of 30-50 HU. A central zone of decreased attenuation occurs in 83% of patients. The margins of the low-attenuating mass usually are poorly defined and correspond to a hypovascular scirrhous tumor. Pancreatic tumor also may undergo central necrosis to produce a low density, and the tumor then simulates a small pseudocyst.
Needle biopsy is occasionally needed to differentiate necrotic tumor from pseudocyst as a result of focal pancreatitis. The pancreatic tumors are hypovascular and are best demonstrated with the intravenous administration of contrast material and by acquiring images across the mass in the parenchymal arterial phase. The mass is seen as a low-attenuating lesion in the brightly enhancing surrounding parenchyma.
Ductal dilatation occurs in 58% of patients. Among patients with ductal dilatation, 75% have dilation of both the pancreatic ducts and the biliary ducts. Pancreatic ductal dilatation proximal to the obstructing tumor is detected in approximately 88% of pancreatic head tumors and 60% of pancreatic body neoplasms. The duct size in pancreatic cancer is 5-10 mm, and the duct is either smooth or beaded.
The pancreatic duct is dilated to more than 50% of the anteroposterior diameter of the gland in pancreatic cancer from atrophy of the gland. In chronic pancreatitis, duct dilatation is less than 50% of the anteroposterior (AP) diameter. Loss of normal peripancreatic fat-plane attenuation is suggestive of extension of tumor beyond the margins of the gland with invasion. The peripancreatic fat shows an increase in attenuation. Extension to involve peripancreatic fat and surrounding structures is observed on CT scans in 92% of patients.
Vascular encasement usually determines unresectability and is seen on CT as narrowing, displacement or obliteration of the vessel lumen by surrounding tumor. Collateral venous circulation may be observed from venous occlusion with contrast-enhanced vessels around the stomach and splenic hilum. Arterial encasement is usually well demonstrated by good-quality CT scans, and angiography is unnecessary.
The arterial involvement, in descending order of frequency, is as follows: superior mesenteric, splenic, celiac, hepatic, gastroduodenal, and left renal. Spread to surrounding organs may involve the spleen, stomach, duodenum, splenic flexure of the colon, transverse mesocolon, porta hepatis, kidney, and spine. Local, posterior tumoral extension into the porta hepatis is seen in approximately 68% of patients. The presence of ascites indicates peritoneal metastatic disease with implants. Ascites is seen in 13% of patients with pancreatic cancer. The peritoneal deposits are poorly demonstrated by means of CT.
Regional lymph-node metastasis has been reported to vary from 38-65%. Metastasis to liver is the most common in pancreatic cancer, occurring in approximately 17-55%. The CBD is displaced anteriorly and medially when the pancreatic mass causes distal ductal obstruction. Intrahepatic ductal dilatation and gall bladder dilatation can be demonstrated.
Degree of confidence
At present, CT is the most widely used and most sensitive test for an evaluation of the pancreas for pancreatic carcinoma. Dynamic CT has a detection rate of approximately 99%. Multisection CT should be the first-line study for detecting this tumor and for evaluating its resectability.
According to a study by Raman et al, MDCT can accurately stage patients with pancreatic cancer, but its accuracy in excluding distant metastatic disease depreciates over time. The authors concluded from their findings that patients should undergo a repeat MDCT within 25 days of any planned definitive operative intervention for pancreatic cancer to avoid unexpectedly finding metastatic disease at surgery. 
Cysts or focal pancreatitis can occasionally cause problems in diagnosis, and it can produce false-positive and false-negative results.
Magnetic Resonance Imaging
The role of MRI in the management of pancreatic adenocarcinoma has yet to be firmly established. Compared with other modalities, MRI appears to be more valuable for staging the extent and spread of pancreatic carcinoma than for tumor detection of lesions smaller than 2 cm. The ability of MRI to demonstrate pancreatic adenocarcinoma largely depends on the demonstration of deformity of the gland, as reflected in its size, shape, contour, and signal intensity characteristics. [18, 19, 20, 21]
The criteria to suspect a mass are similar to those applied with CT, as discussed above. Rarely, nonenhanced MRI reveals a carcinoma of the pancreas before it deforms the gland. However, when such a feature is encountered, the dilemma to distinguish the focal abnormality from focal pancreatitis becomes challenging.
An alteration in signal characteristics is less specific for tumor because the tissue relaxation times between pancreatic cancer, pancreatitis, and controls can overlap significantly. The mean T1 relaxation time of normal pancreas is 507 ms ± 98, and the T1 relaxation time of pancreatic tumor is about 660 ms ± 115. The T2 relaxation time of normal pancreas is 59 ms ± 9, and the T2 relaxation time of pancreatic tumor is 67 ms ± 29.
The normal pancreas is of low signal intensity on T1-weighted images and of intermediate signal on T2-weighted images, with a variable amount of fat in the gland parenchyma.
Newer techniques to obtain images using breath-hold techniques and advances in body coil technology and faster techniques have made it possible to acquire images with excellent spatial resolution.
T1-weighted fat-suppressed spin-echo and single–breath-hold gradient-echo fast low-angle shot (FLASH) sequences with gadolinium enhancement are valuable for tumor detection. The mass is demonstrated as a low-intensity lesion within a homogeneously enhancing normal pancreatic gland. Intravenous Gd-enhanced FLASH images obtained 10 s after contrast enhancement has proven to be more sensitive in demonstrating tumor than other techniques.
Gadolinium-based contrast agents have been linked to the development of nephrogenic systemic fibrosis (NSF) or nephrogenic fibrosing dermopathy (NFD). The disease has occurred in patients with moderate to end-stage renal disease after being given a gadolinium-based contrast agent to enhance MRI or MRA scans. NSF/NFD is a debilitating and sometimes fatal disease. Characteristics include red or dark patches on the skin; burning, itching, swelling, hardening, and tightening of the skin; yellow spots on the whites of the eyes; joint stiffness with trouble moving or straightening the arms, hands, legs, or feet; pain deep in the hip bones or ribs; and muscle weakness.
Magnetic resonance cholangiopancreatography (MRCP) is as sensitive as ERCP and may prevent inappropriate explorations of the pancreatic and bile ducts in patients with suspected pancreatic carcinoma in whom interventional endoscopic therapy is unlikely. The sensitivity of MRCP in a study of 124 patients was 84% with a specificity of 97% for pancreatic cancer. The findings are complimentary to those of ERCP and percutaneous transhepatic cholangiography (PTC).
Degree of confidence
It has been difficult to prove consistent results using MRI in demonstrating the tumor and its resectability. The degree of confidence with MRI is less than that with CT at the moment because of the wide variability in MRI techniques and its limitations from motion artifacts. Recent studies have demonstrated results confirming greater reliability with MRI performed by using meticulous technique.
The lesion may have a variable appearance on US. It may be hypoechoic, isoechoic, or hyperechoic to the normal pancreas. Pancreatic ductal dilatation and biliary ductal dilatation are easily demonstrated in patients with a tumor in the head of pancreas that causes an obstruction. [22, 23]
Lymphadenopathy, the relation of the tumor to peripancreatic vessels, and the tumor margins are demonstrated less reliably with US than with other modalities. The mass appears as an irregular hypoechoic mass that infiltrates a bright pancreatic parenchyma.
Degree of confidence
The degree of confidence may be improved by using EUS in the detection of tumors smaller than 2 cm.
US equipment has improved considerably, and this is likely to have reflected on the sensitivity for detecting pancreatic masses. TAUS examination is still less sensitive than other modalities in the detection of pancreatic malignancy smaller than 2 cm. It has a sensitivity of 70% and a specificity of 95% for the diagnosis of pancreatic malignancy.
The specificity of EUS for differentiating benign from malignant lesions using US appearance alone remains unsatisfactory. EUS has a high sensitivity and specificity for pancreatic cancer, with an overall staging accuracy higher than 80%. The possibility of performing EUS-guided FNA significantly improves both diagnostic and staging capability of EUS. EUS-guided FNA is safe with a morbidity of less than 2%.
In a review of 63 patients, the assessment of tumor resectability with EUS was compared with an assessment with MRI. The sensitivity for EUS for resectability was 61% and that of MRI was 73%. When EUS and MRI were used together, the sensitivity was 89% for resectability.
The detection of a pancreatic tumor and distinguishing its appearances from those of other focal pancreatic diseases has remained a challenging diagnostic problem.
Positron emission tomography (PET) is based on functional changes in the pancreatic cancer cells caused by enhanced glucose utilization as in any other malignant tissue. With 2-[fluorine 18]-fluoro-2-deoxy-D-glucose (FDG), PET can be used to identify pancreatic cancer and differentiate it from chronic pancreatitis with a sensitivity of 85-98% and a specificity of 53-93%. [24, 25]
PET maps the metabolic activity at a molecular level; therefore, the uptake of FDG by neoplastic tissue is also dependent on factors such as tissue oxygenation, regional blood flow, and a peritumoral inflammatory reaction.
PET is also useful in staging and determination of resectability of the tumor at the time of initial diagnosis. PET also has been shown to be an effective tool in the follow-up care of patients with pancreatic cancer. In more than 50% of patients in one study, additional information using PET influences the therapeutic procedure. [26, 27, 28]
Degree of confidence
In general, the sensitivity of PET is high in the detection of lesions more than a centimeter in diameter. Early detection because of increased metabolic activity in cancer usually precedes the structural changes detected with US, MRI, or CT.
PET has shown some promise in the detection of tumors, with high sensitivity and specificity. PET is dependent on tumor stage. However, further clinical trials are required to demonstrate the limitations of PET in the assessment of early pancreatic cancer.
PET is not an absolute technique as it fails to demonstrate pancreatic adenocarcinoma smaller than 1 cm. False-positive results may occur when focal pancreatitis is associated with early pancreatic carcinoma.
Angiography is an invasive procedure that demands considerable operator skill and high-quality radiographic technique. Selective arteriograms obtained with an injection of iodinated contrast through the celiac axis and superior mesenteric artery with some magnification techniques may be required to demonstrate detail. [29, 15]
Pancreatic carcinoma is relatively avascular and associated with neovascularity in 50% of patients. Pancreatic malignancy usually demonstrates arterial encasement of peripancreatic vessels or, actually, of vessels within the pancreas. The vessels involved are the following, in descending order of frequency: superior mesenteric artery (33%), splenic artery (14%), celiac artery (11%), hepatic artery (11%), gastroduodenal artery (3%), and left renal artery (0.6%).
When the disease is advanced, venous occlusions and venous encasement with collateral vessels may be observed. Superior mesenteric vein encasement by tumor is seen in 23%, and the splenic vein is encased by tumor in 15%, with portal vein infiltration in 4%.
Complete occlusion of the splenic vein is seen in 34%, and complete occlusion of the superior mesenteric vein is seen in 10%. Pancreatic carcinoma can be distinguished from pancreatitis. The demonstration of hypervascularity with the typical beaded changes of alternating narrowing with dilatation of internal pancreatic vessels is a feature of pancreatitis.
The mesenteric circulation has been evaluated by using MRA and compared with conventional angiography. Excellent agreement was seen between MRA and conventional angiography. Gd-enhanced MRA is useful in the evaluation of proximal mesenteric arteries and in the evaluation of portal hypertension. Conventional angiography is needed for the evaluation of intrahepatic arteries and branches of the superior mesenteric artery.
Degree of confidence
Pancreatic carcinoma is a hypovascular lesion; therefore, angiography has rightly been replaced as the method of choice for evaluation of pancreatic parenchymal disease.
Helical CT angiography shows useful information about the peripancreatic vessels in patients with pancreatic carcinoma. In a study of 84 patients, the negative predictive value of a resectable tumor was 96% for helical CT angiography and axial helical CT compared with 70% for helical CT alone. The addition of helical CT angiography improves the radiologist's ability to predict the resectability of pancreatic tumors.
Angiography has an accuracy of only 70% in making a specific diagnosis of pancreatic carcinoma.