Pancreatic Cancer: Radiologic Imaging

Clinical Overview

Each year in the United States, more than 40,000 patients are diagnosed with pancreatic cancer and a nearly equal number will die from their disease. Pancreatic cancer is now the fourth most common cause of cancer death in the United States. In recent years, there have been impressive gains in our knowledge of both the underlying risk factors and the molecular biology of this disease. In addition, refinements in surgical management have led to considerable reduction in overall postoperative mortality. Nevertheless, there are few long-term survivors of pancreatic cancer because of the preclinical dissemination of this disease via lymphatic and perineural spread.

The only treatment that has been shown to prolong survival is surgical resection with negative margins (R0 resection); with this treatment, a mean survival of 24 months can be anticipated, compared to 12 months for patients with disease deemed unresectable. It is important to stress, however, that even surgery with negative margins fails to result in long-term survival: only 3% to 16% of patients undergoing R0 resection are alive after 5 years. There are many reasons for this; one often-cited factor is the propensity of pancreatic carcinoma to spread via the celiac and mesenteric neural plexi. Perineural invasion often results in dissemination of tumor before the onset of clinical symptoms. Standard surgical resection will fail to eliminate perineural spread of ductal adenocarcinoma, and this most likely contributes to the overall poor prognosis of the disease.

Despite the dismal outlook for patients with pancreatic carcinoma, radiologic imaging plays a critical role in selecting either medical or surgical therapy for patients, depending on the likelihood of resectability. Accurate preoperative staging is therefore a crucial step in identifying the small number of patients who may potentially benefit from pancreaticoduodenectomy. Conversely, identifying patients with either distant or locally advanced disease is another important goal of imaging so that these patients are not subjected to a fruitless major surgical intervention. Finally, it is essential to distinguish a pancreatic adenocarcinoma from an ampullary tumor, given the far better prognosis associated with the latter.

Although a variety of imaging techniques, such as multidetector computed tomography (MDCT), magnetic resonance imaging (MRI), endoscopic ultrasound (EUS), and fluorodeoxyglucose positron emission tomography (FDG-PET) have all been used in the initial diagnostic assessment of pancreatic lesions, in general, MDCT is the mainstay of diagnosis. This is due in large part to the high spatial resolution of its three-dimensional (3D) data sets that are indispensable in determining the local extent of tumor infiltration.

MDCT and MRI Technique

Patients with pancreatic carcinomas frequently present with vague symptoms that are often misdiagnosed clinically as other intra-abdominal abnormalities. Not infrequently, a pancreatic mass may be detected on a screening abdominal ultrasound or CT scan of the abdomen. It is important to emphasize that, in many instances, these screening studies are not adequate to assess local invasion. Therefore, a dedicated pancreatic-protocol MDCT study, which will improve the accuracy of local staging by obtaining the highest resolution possible, is indicated. A dedicated pancreatic protocol MDCT has four critical components, and attention to detail is essential to achieving optimal results. These components include (1) neutral oral contrast administration to distend the stomach and duodenal sweep, (2) intravenous contrast injection via a rapid bolus, (3) optimized biphasic MDCT scan acquisition, and (4) postprocessing with a combination of two-dimensional (2D) and 3D volumetric image displays to highlight extrapancreatic extension along blood vessels and peripancreatic tissues.

To begin the pancreatic protocol, the patient ingests 750 to 1000 mL of a neutral oral contrast, such as water, to distend the stomach and duodenum immediately before scan acquisition. Gastroduodenal distension allows better depiction of invasion of these structures and also highlights the region of the ampulla of Vater. It is important to not miss an underlying ampullary carcinoma or misconstrue it as a ducal adenocarcinoma because of the far better prognosis associated with ampullary tumors ( Fig. 1 ) . Ampullary carcinomas, particularly of the gastrointestinal type, may be associated with a 40% 5-year survival, which is substantially greater than survival for patients with ductal adenocarcinoma.

After the ingestion of neutral contrast, an intravenous bolus is administered with a rapid 4- to 5-mL/s injection of 150 mL of nonionic contrast via power injector. This rapid bolus causes intense enhancement of the normal pancreas, which will highlight the differences in vascular perfusion between the typically hypovascular ductal adenocarcinoma and the normal pancreatic parenchyma ( Figs. 2 and 3 ) . A late arterial-phase acquisition, typically 35 seconds from the onset of intravenous contrast injection, is optimal to obtain the initial breath-held images of the pancreas and upper abdomen. Late arterial-phase images improve conspicuity of underlying pancreatic lesions and also demonstrate hypervascular lesions, such as neuroendocrine tumors and their hypervascular liver metastasis ( Fig. 4 ) . The late arterial phase also optimally enhances the splanchnic vasculature, which enables detection of subtle perivascular infiltration of the underlying tumor ( Fig. 5 ) . It is important that the entire pancreas and upper abdomen are scanned with a breath-held acquisition, as late arterial-phase images may demonstrate flow phenomena associated with subtle liver metastasis that will not be evident on portal-venous phase images alone. These phenomena include thin peripheral enhancement around subtle hypodense metastasis and perfusion abnormalities such as a transient hepatic attenuation defect associated with hepatic metastasis.

After the late arterial-phase acquisition, an additional scan is acquired during the portal-venous phase, typically 60 to 70 seconds after the onset of intravenous injection. These venous-phase images may more optimally demonstrate small hypodense liver metastasis in some patients and improve overall visualization of a portal venous system, which will allow for assessment of venous encasement or obstruction ( Fig. 6 ) . Often, peripancreatic and perisplenic varices are optimally seen in this phase, which is a clue to underlying splenic or portal vein occlusion or both.

MRI has proven to be comparable to MDCT in accuracy of staging ductal adenocarcinoma. Although it has slightly less spatial resolution than MDCT, the excellent contrast resolution of MRI makes it an attractive alternative to MDCT when patients are allergic to iodinated contrast. Unlike noncontrast CT, MRI without intravenous contrast may yield useful images of the pancreas, and thus MRI is very helpful in patients with renal insufficiency ( Fig. 7 ) . Its superior contrast resolution may also be of particular value in assessing neuroendocrine tumors and subtle liver metastasis. Optimal MRI technique involves a combination of pulse sequences, including fat-suppressed T1-weighted imaging, T2-weighted images, and gadolinium-enhanced gradient-echo images, to obtain breath-held late-arterial and portal-venous phase acquisitions through the upper abdomen and pancreas ( Fig. 8 ) . Diffusion-weighted imaging is a relatively new MRI technique that may be particularly valuable in the detection of liver metastasis because of restricted diffusion of these solid lesions. In addition, newer hepatic-specific contrast agents may prove to be a valuable contribution to the overall assessment of liver metastasis.

One considerable advantage of MDCT is that the thin collimation scans (0.6 mm) result in isotropic data sets with resolution comparable in all three planes. These high-resolution data sets enable the performance of unique imaging displays in both 2D and 3D that greatly facilitate depiction of extrapancreatic spread of carcinoma. Curved-planar reformations depict, in two dimensions, a tubular structure coursing through the 3D data set. Curved-planar reformations of the pancreatic and common bile ducts highlight subtle dilatation from underlying obstructing lesions (see Fig. 9 ). hese types of displays have proven to be highly effective in visualizing tumor spread along blood vessels ( Fig. 10 ) . 3D imaging, particularly with volume-rendered displays, has been gaining widespread utilization with pancreatic-protocol MDCT ( Fig. 11 ) . These are often used in an interactive display as the interpreting radiologist scrolls through a 3D volume data set to perceive subtle peripancreatic soft-tissue infiltration. Ray-casting techniques, such as maximum-intensity projections, allow depiction of peripancreatic vasculature that highlights areas of venous encasement and peripancreatic varices, as well as arterial anomalies and areas of narrowing (see Fig. 5 ). Minimum-intensity displays are often very useful in the depiction of low-attenuation structures such as the pancreatic and common bile ducts and cystic pancreatic masses ( Fig. 12 ) .

Images of fluid-filled structures, such as the pancreatic and common bile ducts, used for MR cholangiopancreatography (MRCP) are created with heavily T2-weighted sequences and are some of the most important displays of pancreatic MRI (see Fig. 7 ). These displays often highlight subtle areas of ductal narrowing and are particularly useful for evaluating pancreatic cystic lesions. Because MRI lacks ionizing radiation, it is often the technique of choice in follow-up examinations for patients with suspected cystic lesions of low malignant potential.