Материал: Advanced Imaging of the Abdomen - Jovitas Skucas

more delayed images. Normal pancreas achieves maximum MR enhancement during the arterial phase, and images at this time generally are best at maximizing the intensity differences between normal pancreas and a carcinoma. This technique is most useful for small cancers that do not produce a contour abnormality. A hyperintense ring often surrounds the cancer postcontrast, probably representing compressed adjacent pancreatic parenchyma and surrounding inflammation.

The MRCP results are similar to ERCP in detecting pancreatic cancer, but an advantage of MRCP is that it can be combined with other MRI to provide additional information. Thus in 52 patients with suspected pancreatic cancer, the accuracy of MRCP alone was 80% and of ERCP 85%, but combined MRI and MRCP achieved an accuracy of 88% (78).

An MRCP detected an obstruction or irregular stenosis in 96% of 129 patients with a pancreatic carcinoma, while a majority of inflammatory pancreatic tumors had a nonobstructed duct (79). Nevertheless, MR differentiation between a pancreatic carcinoma and chronic pancreatitis is not possible in a minority of patients.

Scintigraphy

No ideal radiopharmaceutical agent is available for pancreatic imaging. Some techniques, however, are promising. Thallium-201-chloride single photon emission computed tomography (SPECT) shows positive uptake in pancreatic cancers. Uptake of this agent is also evident in some patients with chronic pancreatitis.

Focal FDG-PET uptake is seen in pancreatic carcinomas, with several studies achieving sensitivities and specificities in the 80% range. Typically, FDG-PET is considered positive if pancreatic activity exceeds background and negative if activity is less than or equal to background activity. Unfortunately, focal FDG-PET uptake is also found with benign tumors and pancreatic inflammation. Also, the accuracy of FDG-PET scintigraphy is dependent on serum glucose levels.

Iodine 123 vasoactive intestinal peptide (VIP) binds to pancreatic tumor cells and has an affinity for pancreatic cancer. In one study primary pancreatic tumors were visualized by this test in 90% of patients if disease was con-

ADVANCED IMAGING OF THE ABDOMEN

fined to the pancreas and in 32% of patients if regional or liver metastases were present (80).

Percutaneous Transhepatic

Cholangiography

Percutaneous transhepatic cholangiography (PTC) usually can suggest a carcinoma in the head of the pancreas. In most institutions, however, ERCP is the procedure performed. In a setting of a failed ERCP, percutaneous cholangiography is a viable alternative.

Endoscopic Retrograde

Cholangiopancreatography

Endoscopic retrograde cholangiopancreatography not only provides imaging of both the pancreatic and biliary ducts, but also in a setting of suspected malignancy ERCP can obtain brush cytology and thus detect carcinoma in situ and marked atypia. The role of screening high-risk patients with cytology is yet to be established.

Biopsy

If CT or US suggests a pancreatic tumor, these modalities are useful in guiding fine-needle aspiration. Accuracy of a percutaneous pancreatic biopsy has ranged up to 90%, although it is lower in most studies. A positive biopsy or cytology is generally diagnostic of a pancreatic carcinoma. A negative biopsy, however, does not exclude a cancer. In a setting of imaging suspicion for a cancer, some studies report about half of patients with an atypical cell or negative biopsy are eventually confirmed as malignant. This points to some of the problems encountered with biopsies.

Groove carcinoma. A carcinoma originating in the groove between the duodenum and pancreatic head is called a groove carcinoma by some authors. Whether all of these are indeed pancreatic in origin or whether some originate from the papilla of Vater or even distal common bile duct is conjecture.

Imaging reveals sheet-like tumor within this groove (81). These tumors tend to be hypointense on T1-weighted images and hyperintense on T2-weighted MR images. Most are

PANCREAS

hypovascular on early arterial phase and show delayed enhancement during a late phase. Narrowing of the intrapancreatic common bile duct is a common feature. Peripancreatic and adjacent duodenal wall invasion is also common at initial presentation, making resection difficult. The imaging findings resemble those of duodenal carcinoma, metastasis to pancreaticoduodenal lymph nodes or even groove pancreatitis.

Staging/Resectability

Although a Union Internationale Contre le Cancer tumor, node, metastasis (TNM) staging system exists for pancreatic cancer (Table 9.4), the staging of resected cancers is probably of

Table 9.4. Tumor, node, metastasis (TNM) staging of exocrine pancreatic tumors

Source: From the AJCC Cancer Staging Manual, 6th edition (2002), published by Springer-Verlag, New York, NY, used with permission of the American Joint Committee on Cancer (AJCC), Chicago, IL.

little clinical relevance; some might consider such an approach too pessimistic given the gradual improvement, albeit slow, in postoperative survival. The current emphasis is primarily on detecting unresectability. In general, imaging is considerably more accurate in detecting unresectability than in establishing that a pancreatic carcinoma is indeed resectable. In analyzing the data one should keep in mind that resectability is defined differently by various surgeons and the definition has become more liberal with time. Also, with pancreatic cancers resectability is not the same as cure.

Even in a setting of widespread lymph nodes metastases, the lymph nodes often are normal in size. Similarly, enlarged nodes in some patients simply represent associated inflammation. The issue is actually more complex; thus an immunohistochemical antiepithelial monoclonal antibody test revealed tumor cells in 43% of lymph nodes that histopathologically did not contain tumor (82); of note is that 75% of patients staged pN0 had these cells. Using multivariate Cox’s regression analysis, the authors found that the presence of these cells in otherwise histologically tumor-free lymph nodes represents an independent prognostic factor for both reduced relapse-free survival and overall survival, and suggest that staging include immunohistochemical testing (82); patients with minimal residual tumor could then be selected for additional therapy.

Another complicating factor is that pancreatic adenocarcinoma cells frequently bypass initial draining lymph nodes and become established in secondary sites, and small tumors are not uncommonly associated with remote metastases. Cancers of the body and tail are prone to peritoneal and hematogenous spread more often than cancers of the head of the pancreas. Ascites in a patient with known pancreatic cancer generally implies peritoneal spread even in the face of normal imaging.

Arterial occlusion is generally a relatively late finding.

Comparison Studies

In patients with known or suspected pancreatic carcinoma, introduction of helical CT led to a greater sensitivity in establishing resectability with this modality than with MRI, although further advances in MR since then suggest that

MR is superior. A study in 1999 found that MRI was significantly better than CT in assessing resectability (83). Compared with CT, mangafodipir enhanced MR appears superior in evaluating resectability and provides better detection of small metastases (84). Useful MR sequences include FSE, fat-suppression, T1weighted breath-hold gradient-echo FLASH and preand postcontrast. Nevertheless, current multislice CT achieves a diagnosis of unresectability with a sensitivity of about 95% and specificity of 85% (85), impressive results.

In detecting vascular involvement, CT appears superior to endoscopic US. Postcontrast thin-section CT appears superior to breath-hold postcontrast MRI in detecting peripancreatic vein, portal vein, or peripancreatic arterial invasion (86). Color Doppler US detection of portal vein invasion appears to be similar to that seen with CT and angiography.

Computed Tomography

Traditionally, angiography was performed to assess vascular encasement, although currently CT is preferred. More recently, CT arterial portography has been suggested as being superior even to angiography in detecting encasement but is not widely employed.

Preoperative optimized thin slice CT achieves a sensitivity of about 90% in predicting unresectability and a similar or better sensitivity and specificity in assessing resectability. The current major limitation of CT consists of undetected small metastases to the liver and peritoneum (87). CT arterial portography has been suggested as being superior even to angiography in detecting encasement but is not widely employed.

Postcontrast early-phase and late-phase CTs study different entities. Arterial-phase CT detects arterial encasement, while portal phase is useful in determining portal vein and superior mesenteric vein involvement. Retroperitoneal invasion is better evaluated on earlyphase CT. Reconstructed 3D arterial phase data are helpful if arterial encasement is suspected. At times arterial invasion is seen simply as vessel wall thickening. Most major vascular involvement involves the portal vein and superior mesenteric vein, although narrowing can be due to either simply extrinsic impression or direct tumor invasion. Using thin-section con-

ADVANCED IMAGING OF THE ABDOMEN

trast-enhanced CT, if a fat plane or pancreatic parenchymal tissue can be identified between a tumor and adjacent vessels, the tumor is resectable without venous resection being necessary; complicating matters, some veins adjacent to the pancreas are not surrounded by fat and their encasement is difficult to detect. In general, upstream dilation of small peripancreatic veins implies unresectability. Tumor encasement of adjacent vessels implies vascular invasion and generally connotes an inability to resect with negative surgical margins. The degree of circumferential peripancreatic vein encasement by a tumor, as shown by thinsection pancreatic-phase CT, aids in predicting which patients have unresectable tumors (88). Another sign suggesting unresectability is a tethered, teardrop-shaped superior mesenteric vein (teardrop sign) (89); together with other findings, the presence of a teardrop-shaped superior mesenteric vein increases sensitivity in detecting unresectability. Splanchnic venous invasion results in a delay in contrast washout from the small bowel; these patients have increased small bowel wall enhancement and a decrease in portal vein enhancement compared to those without invasion.

Multiplanar and 3D volume-rendered CT techniques show promise in detecting vascular involvement, although considerable intraand interobserver variability in tumor grading exists, and these techniques are best used together with axial images (90).

Infiltration of adjacent organs or adenopathy implies unresectability, keeping in mind that even an optimal CT technique is inherently limited in detecting lymphatic involvement unless a node is enlarged or contains a contour abnormality. Computed tomography detects only about half of nodal involvement, while detecting liver metastases in 75% of patients (91).

Ultrasonography

Endoscopic US aids in staging these tumors. Endoscopic US suggests portal vein invasion if an irregular venous wall outline, loss of wall interface and proximity of tumor are detected. Limitations of this technique include an inability to assess direct tumor invasion into the mesocolon and an inability to evaluate the effect of peritumoral pancreatitis. Likewise, similar to

PANCREAS

other imaging modalities, although endoscopic US detects lymph node enlargement, metastases cannot be differentiated from benign lymphadenopathy.

Is color Doppler US useful for detecting vascular involvement? Changes in portal vein blood flow velocity are seen with tumor compression or invasion. Resectability is suggested by lack of tumor contact with major vessels as detected by Doppler US. Likewise, vessel encasement signifies unresectability. Problems are encountered in those patients with contact but no encasement between tumor and vessel, and the Doppler US results are mixed in suggesting resectable.

The value of laparoscopic US in staging pancreatic cancers is still uncertain. At times laparoscopic US detects liver metastases not seen by other imaging modalities.

Intraportal endovascular US is feasible intraoperatively via the superior mesenteric vein or preoperatively via a percutaneous transhepatic approach. Sensitivities and specificities >90% can be achieved in detecting portal vein invasion, and in select individuals such an approach is potentially useful. Portal vein invasion is implied by obliteration of the portal vein echogenic outline. Such an endovascular approach has also been suggested for detecting adjacent extrapancreatic nerve plexus invasion (92).

Magnetic Resonance Imaging

Magnetic resonance studies available to evaluate and stage pancreatic cancer include breathhold T1and T2-weighted sequences, MRCP, and breath-hold MRI using gadolinium and mangafodipir contrast, at times in a 3D presentation. Such an approach provides multiple specific MR examinations, all obtained as part of one study. Similar to other applications, MRI preoperative staging depends on the magnet field strength and scanning parameters used. Evolving study parameters make predictions for optimal tumor staging techniques unpredictable. Different techniques appear advantageous for gauging vascular invasion and detecting adenopathy. One MR indication appears to be in patients with inconclusive CT, especially those with a CT finding of an indeterminate pancreatic head enlargement.

Both preand postcontrast MRI is useful in staging. Precontrast T1-weighted images are useful in evaluating peripancreatic tumor extension, but tissue planes between tumor and peripancreatic fat are not as well seen postcontrast; on the other hand, tumor margins and surrounding vasculature are better identified on postcontrast images (Fig. 9.18). Due to fibrosis, MRI tends to overestimate vascular invasion.

A combination of conventional MR, 3D MRCP,and venous and arterial MRA both detect a cancer and can suggest unresectability due to vascular invasion and identify any bile duct or pancreatic duct obstruction.

Angiography

Currently arteriography is little used for staging pancreatic cancer. Angiography tends to overestimate tumor invasion.

Theoretically,percutaneous transhepatic portography and intravascular US hold promise in evaluating portal and superior mesenteric vein involvement by tumor, but preliminary results reveal high false positive and negative rates (93); also, these procedures are not without significant complications.

Therapy

Therapy for pancreatic ductal adenocarcinoma can be divided into attempted cure and palliation.

Resection

Some of the reported survival rates after resection of pancreatic carcinoma are rather optimistic and need to be placed in a proper perspective. Little consistency in reporting survivors is evident in the literature, reported patient selection bias exists and some studies do not exclude cystic and neuroendocrine tumors with their more favorable survival. In particular, 5-year survival rates of about 30% after resection should be viewed critically. The overall survival rate is probably <1% and, in spite of some reports, resection does not have any discernible impact on patient survival.

The pros and cons of preoperative biliary drainage in a setting of obstructive jaundice are discussed in Chapter 8.