- Email: [email protected]
Diagnosis and staging of pancreatic adenocarcinoma with dynamic computed tomography
Diagnosis and Staging of Pancreatic Adenocarcinoma With Dynamic Computed Tomography Patrick C. Freeny, MD, L. William Traverso, MD, John A. Ryan, MD, Seattle,Washington
We assessed the accuracy of dynamic contrast-enhanced computed tomography (CT) in the diagnosis and staging of 213 patients with pancreatic carcinoma and compared it with the accuracy of angiography and surgery. A correct CT diagnosis of pancreatic carcinoma was made in 207 of 213 (97%) patients. Tumors were located in the pancreatic head in 64%, the body in 22%, and the tail in 10%, and enlarged the pancreas diffusely in 4%. CT staged 25 (12%) patients as having potentially resectable tumors and 188 (88%) as having unresectable tumors on the basis of local extension (72%), contiguous organ invasion (4~%), vascular invasion ( 8 2 % ) , and distant metastases ( 5 0 % ) . Compared with angiography in 6 0 patients, CT detected vascular invasion missed on angiograpby in 20%, and angiography detected invasion missed by CT in 5%. In these latter cases, other CT criteria of unresectability were present, and angiography provided no significant staging information. Compared with surgery in 71 patients, c r accurately predicted unresectable tumors in 100% of patients and resectable tmnors in 72% of patients. Eleven of the patients with CT-resectable tumors underwent resection. Median survival was 22.7 months, with four patients alive at a median of 15.5 months postoperatively. Palliative resections were performed in six patients, and median survival was 14.4 months.
omputed tomography (CT) has become the primary modality for the evaluation of patients with suspected pancreatic carcinoma, virtually eliminating the use of endoscopic retrograde cholangiopancreatography (ERCP) and angiography for initial diagnosis [1,2]. The use of dynamic scanning techniques that employ automatic table incrementation during rapid power injection of a large intravenous bolus of iodinated contrast medium has significantly increased the efficacy of CT for the detection, diagnosis, and staging of pancreatic carcinoma. We assessed the accuracy of high-resolution dynamic CT and compared it with angiography and surgery in the detection and staging of 213 patients with pancreatic carcinoma.
C
PATIENTS AND M E T H O D S A prospective series of 213 incremental dynamic bolus CT scans of the abdomen performed in patients with ductal adenocarcinoma of the pancreas between 1982 and 1992 were reviewed. The diagnosis of pancreatic adenocarcinoma was confirmed by percutaneous biopsy in 80 patients (38%), surgical exploration and biopsy in 71 patients (33%), or with a combination of clinical follow-up studies and combined results of CT, ERCP, and angiography in 62 patients (29%). CT techniques: Incremental dynamic abdominal CT was performed with a General Electric C T / T 8800 or 9800 scanner (General Electric Medical Systems, Milwaukee, WI). Scans were obtained using contiguous 5- to 10-mm collimated slices during rapid intravenous bolus power injection of 150 mL of a 60% iodinated contrast agent injected at a rate of 2 to 2.5 mL,/sec for 20 to 25 seconds and 1 mL/sec for the remainder of the volume. The scan sequence was performed using individual breath-held scans or clusters of three scans beginning 30 to 40 seconds after the initiation of the contrast bolus. The entire liver and pancreas were scanned within 180 seconds in all patients. CT scan interpretation: The original CT radiographic reports were used to record the CT diagnosis, regardless of the findings on retrospective review of the CT scans. A C T diagnosis of pancreatic carcinoma was made if, in the absence of changes of chronic or acute pancreatiFrom the Departments of Radiology (PCF) and General Surgery tis, a focal or diffuse pancreatic mass was present with or (LWT, JAR), Universityof WashingtonSchoolof Medicine,Virginia without evidence of local or distant tumor spread, or if a Mason MedicalCenter, Seattle,Washington. dilated main pancreatic and common bile duct were idenRequests for reprints should be addressed to Patrick C. Freeny, tified to the level of the pancreatic head or ampulla in the MD, Department of Radiology SB-05, University of Washington absence of a pancreatic mass. Schoolof Medicine,Seattle,Washington98195. The 213 CT scans were reviewed retrospectively at Presentedat the 79th Annual Meetingof the North PacificSurgical Association,Tacoma,Washington,November13-14, 1992. the conclusion of the study, and the following 6 parame600
THE AMERICAN JOURNAL OF SURGERY VOLUME165 MAY 1993
PANCREATIC ADENOCARCINOMA
ters were evaluated: (1) pancreatic mass (focal or diffuse; location; tumor size; attenuation); (2) main pancreatic duct and intrahepatic and extrahepatic bile duct appearance (normal; not seen; dilated: main pancreatic duct dilated if the diameter was greater than 5 mm in the head or greater than 3 mm in the tail; intrahepatic ducts dilated if seen greater than 2 cm from the hilus and measuring greater than 5 mm in diameter; extrahepatic bile ducts dilated if common hepatic duct greater than 6 mm in diameter and common bile duct greater than 9 mm in diameter); (3) local tumor extension (anterior or posterior to the tumor or into the porta hepatis or splenic hilum); (4) contiguous organ invasion (stomach, duodenum, root of the small bowel mesentery; liver, spleen, adrenal gland, kidney); (5) vascular invasion or tumor-vessel contiguity (tumor partially or completely surrounding a major extrapancreatic artery or vein; arterial or venous obstruction with arterial collateral vessels or mesenteric varices; tumor-vessel contiguity: tumor immediately contiguous with vessel with no intervening normal fat plane); and (6) metastases (liver; lymph nodes). Staging of pancreatic carcinoma: Pancreatic carcinoma was staged on the basis of CT (213 patients), angiography (60 patients), and surgical findings (71 patients). CT staging, Tumors were staged as unresectable if one or more of the following four findings were seen: local tumor extension, contiguous organ invasion, vascular invasion, and hepatic metastases (isolated lymph node metastases, i.e., nodes larger than 2 cm in short axis, were not considered to indicate tumor unresectability). Angiographic staging. Tumors were considered to be unresectable if one or more of the following features were present: tumor encasement or obstruction of major extrapancreatic arteries or veins and hepatic metastases. Surgical staging. Tumors were considered to be unresectable for cure if one or more of the following findings were observed during laparotomy: hepatic or peritoneal metastases; direct tumor extension into the peripancreatic soft tissues, major extrapancreatic vascular structures, contiguous organs, or mesenteries. In some patients, the primary tumors were not resected if frozen section biopsy showed the presence of regional lymph node metastases. RESULTS CT diagnosis: The prospective CT diagnoses in the 213 patients with pancreatic carcinoma were pancreatic carcinoma (207 of 213, 97%), normal CT (4 of 213, 2%), duodenal carcinoma (1 of 213, 0.5%), and panereatitis ( 1 of 213, 0.5%). Normal CTscan. Retrospective review of the four CT scans interpreted as normal showed no abnormality in two patients. The third patient had a small, hypodense mass that was not mentioned on the CT report, despite having been identified on a preoperative sonogram. At surgery, the patient was believed to have acute biliary pancreatitis, and a cholecystectomy was performed. Repeat CT 6 weeks later showed a 4-cm mass in the pancreatic head with local extension and vascular invasion. The fourth patient had pancreatic head enlargement. A re-
TABLE I C T Findings in 213 Patients with Pancreatic Carcinoma*
CT Findings
Total (%)
Tumor mass Focal Diffuse Decreased attenuation
205 197 8 167
MPD/CBD dilatation Biductal (no mass) MPD only Biductal (mass) CBD only
168 (79)
Contiguous organ invasion
(96) (92) (4) (6)
Head (%)
Location of Tumor Body Tail (%) (%)
129 (61) 47 (22)
21 (10)
102 (48) 43 (20)
16 (8)
6 (3)
6 (3)
25 {12) 117 (55) 28 {12)
20 (9) 109 (51) 22 (10)
5 (2) 8 (4) 3 (1)
Diffuse (%)
6 (3)
1 (0.5)
91 (43)
57 (27) 15 (7)
14 (7)
5 (2)
Local tumor extension
154 (72)
89 (42) 39 (I 8)
19 (9)
7 (3)
Vascular invasion Arterial only Venous only Arterial + venous
175 7 48 120
(82) (3) (23) (56)
6 (3) 1 (0.4) 32 (15) 6 (3) 70 (33) 33 (15)
0 9 (4) 10 (5)
0 1 (0.4) 7 (3)
Metastases Liver Lymph nodes
106 (50) 72 t34) 57 (27)
34 (16) 22 (10) 29 (14) 18 (8)
13 (6) 6 (3)
3 (1) 4 (2)
CT = computed tomography; MPD = main pancreatic duct: CBD = common bile duct. *Numbers in parenthesesreferto the percentageof the total (213) patients.
peat scan 8 weeks later showed a 5-cm mass in the pancreatic head and liver metastases. Incorrect CT scan interpretation. Two patients had pancreatic masses that were interpreted incorrectly. One patient was believed to have a primary duodenal carcinoma. CT showed no fat plane between the tumor and the caudal margin of the pancreatic head and no pancreatic or common bile duct dilatation. At surgery, a primary carcinoma of the pancreatic head was found growing inferiorly and invading the duodenum and the root of the small bowel mesentery. The second patient had enlargement of the pancreatic head and peripancreatic soft tissue changes believed to represent pancreatitis. ERCP showed segmental obstruction of the main pancreatic and common bile duct in the head of the pancreas and intraductal filling defects within a dilated upstream pancreatic duct. A pancreatic duct stent was placed, and percutaneous biopsy was positive for adenocarcinoma. Surgical exploration showed the tumor to be unresectable due to extension into the small bowel mesentery.
CT findings of pancreatic 'carcinoma (Table I): The pancreatic tumors were located in the head of the gland in 137 patients (64%), the body in 47 (22%), and the tail in 21 (10%), and diffusely involved the gland in 8 (4%). The CT findings in the 213 patients with pancreatic carcinoma are detailed in Table I.
THE AMERICAN JOURNAL OFSURGERY
VOLUME 165
MAY 1993
601
FREENY ET AL
Figure 1. Large unresectable pancreatic carcinoma. Top. Dynamic CT shows a large mass (M) in the head of the pancreas, dilatation of main pancreatic duct in the body and tail with atrophy of the surrounding parenchyma (arrows), and posterior tumor extension (curved arrows} to the aorta (a) and inferior vena cava (c). s -superior mesenteric vein. Middle. Scan at a lower level shows tumor (T) encasing the celiac and hepatic arteries (arrows) and contiguous with the portal vein (pv). Bottom. CT-guided biopsy shows needle (arrow) within the mass. 602
THE A M E R I C A N J O U R N A L OF SURGERY
VOLUME 165
MAY 1993
PANCREATIC ADENOCARCINOMA
Figure 2. Small unresectable pancreatic carcinoma. Dynamic CT at the level of the pancreatic head shows a 1.8-cm hypodense tumor (arrowheads) extending anteriorly into the peripancreatic soft tissues between (arrow) and encasing the gastroduodenal artery (A) and the anterior pancreaticoduodenalvein (V). Arterial enc a s e m e n t was c o n f i r m e d on angiography. S = superior masenteric vein; D = duodenum.
Figure 3. Resectable pancreatic carcinoma. Top. Dynamic CT shows normal pancreatic (P) body and tail with no duct dilatation. PV = portal vein. Bottom. Scan at the level of the pancreatic head shows a 10-mm hypodense tumor (arrowhead) confined to the uncinate, a --- superior mesenteric artery; v = superior mesenteric vein. THE AMERICAN JOURNAL OFSURGERY
VOLUME 165
MAY 1993
603
FREENY ET AL
The most common compilation of findings was a pancreatic mass (96%), pancreatic and/or common bile duct dilatation, and one or more findings of extrapancreatic tumor extension (88%) (Figures 1 and 2). Extrapancreatic tumor extension encompassed the following four findings: (1) tumor involvement of contiguous organs (91 of 213 patients, 43%), most commonly the duodenum, stomach, root of the small bowel mesentery, and spleen; (2) local extension of the tumor into the anterior or posterior peripancreatic fat and soft tissues, the porta hepatis, and the splenic hilum (154 of 213 patients, 72%); (3) arterial encasement or obstruction by tumor (127 of 213 patients, 60%), most often involving the celiac, splenic, superior mesenteric, hepatic, and gastroduodenal arteries; and (4) venous encasement or obstruction with associated abdominal varices (168 of 213 patients, 79%), most commonly involving the splenic and superior mesenteric veins. Isolated biductal dilatation (i.e., no associated tumor mass seen by CT) was seen in six patients. The remaining 162 (76%) patients with pancreatic and/or bile duct dilatation had associated pancreatic masses. Correlation of angiography, CT, and surgical findings: Angiography was performed in 60 patients, and the findings were correlated with the corresponding dynamic CT findings in each patient and with the findings at surgery in 41 patients. Angiography and CT showed equal and similar findings in 45 (75%) patients. CT showed vascular involvement (tumor-vessel contiguity or encasement of the superior mesenteric artery a n d / o r vein) not seen by angiography in 12 (20%) patients, and, in 3 patients (5%), vascular involvement (common hepatic artery, replaced right hepatic artery, gastroduodenal artery) was seen angiographically but missed by CT. Additional surgical correlation was made in 41 patients. Angiography and CT showed similar and equal findings, which were confirmed at surgery in 28 patients (12 with CT-resectable tumors and 16 with CT-unresectable tumors), whereas CT showed vascular involvement not demonstrated by angiography in 11 patients (tumorvessel contiguity in 8 patients, vascular encasement in 3 patients). In two patients with unresectable tumors, vascular involvement not seen by CT was demonstrated angiographically and confirmed at surgery. Overall, angiography showed vascular involvement not seen by CT in only 3 of 60 patients (5%), but other CT criteria of tumor unresectability were present in all 3 patients. Thus, angiography contributed no additional or significant information to the staging of pancreatic carcinoma. Staging: CT and surgical results: CT-resectable tumors. Twenty-five patients (25 of 213, 12%) had potentially resectable tumors on the basis of CT (i.e., no identifiable CT criteria of unresectability) (Figure 3). This group included two patients with normal CT scans (tumor subsequently identified by ERCP and confirmed by percutaneous biopsy in both patients), the patient with pancreatic head enlargement and the patient with a small tumor mass in the pancreatic head, both missed on the preoperative CT scans, and the patient believed to have pancreatitis. 604
Six (6 of 25, 24%) of the patients who had CT-resectable tumors were not deemed to be operative candidates due to severe coexisting medical diseases. One patient was believed to have biliary pancreatitis at the time of surgery and was discussed earlier. The remaining 18 (18 of 25, 72%) patients with CT-resectable tumors underwent laparotomy. Twelve (12 of 18, 67%) of these patients were found to have resectable tumors, and 11 underwent resection (8 pancreaticoduodenectomies for carcinoma of the head of the pancreas; 2 total pancreatectomies for carcinoma of the head and body of the pancreas; 1 distal pancreatectomy for carcinoma of the tail of the pancreas). The other patient had severe cardiovascular disease, was not believed to be able to tolerate a pancreatic resection at the time of laparotomy, and thus had only a palliative drainage procedure. Seven of the 11 patients who underwent resection have died (range: 7 to 42 months; mean: 22.7 months), and 4 are alive (range: 9 to 22 months; mean: 15.5 months). Ten of the 11 resected tumors were in the head of the pancreas, and 1 was in the tail. The tumors ranged in size on the resected specimens from 2 to 3 cm in maximum diameter. Three patients had no pathologic evidence of local or regional spread of tumor. The other eight patients showed positive lymph nodes (six), perineural invasion (three), duodenal wall invasion (two), and posterior extension of the tumor into the retropancreatic soft tissues (one). Five (5 of 25, 20%) of the patients with CT-resectable tumors were found to have unresectable tumors at the time of surgery and had only palliative procedures. The tumors in these patients were in the head of the pancreas and ranged in size on the CT scans from 2 to 3.5 cm in diameter. In these patients, CT missed tumor invasion into the root of the small bowel mesentery (one) and transverse colon (one), metastatic lymph nodes in the porta hepatis (two), and posterior tumor extension with a metastatic node in the porta (one). CT-unresectable tumors. One hundred eighty-eight patients (188 of 213, 88%) had CT criteria of tumor unresectability as detailed above. Most of these patients (69%) had three to five CT criteria of tumor unresectability, whereas only 31% had one or two criteria. The 20 patients with only 1 criteria had either hepatic metastases, contiguous organ invasion, local tumor extension, or vascular invasion. Fifty-three patients with CT-unresectable tumors underwent laparotomy. Forty-seven (89%) were found to have unresectable tumors and had palliative drainage procedures (biliary, gastric), whereas 6 patients (12%) had surgical confirmation of the CT criteria of unresectability but had palliative resections. Five of these patients have died (range: 6 to 27 months; mean, 14.4 months), and one is alive at 6 months but has widespread hepatic metastases. Two patients had body-tail tumors with splenic vein obstruction and perigastric varices. They developed recurrent tumor at 10 and 11 months postoperatively and died at 15 and 27 months, respectively. Three patients had superior mesenteric vein involvement, and one of these had duodenal invasion. One patient developed liver metastases 4 weeks after pancreaticoduodenec-
THE AMERICAN JOURNAL OF SURGERY VOLUME165 MAY 1993
PANCREATIC ADENOCARCINOMA
tomy and is alive at 6 months, one patient developed postoperative superior mesenteric vein thrombosis and died at 6 months, and the third patient developed postoperative superior mesenteric artery thrombosis and died at 6 months. The final patient had duodenal invasion and positive lymph nodes, developed recurrent tumor at 10 months, and died at 15 months. Thus, CT correctly predicted unresectable tumors in 100% of cases (53 of 53) but correctly predicted resectable tumors in only 72% of cases (13 of 18). COMMENTS CT diagnosis of pancreatic carcinoma: This study confirms the high degree of accuracy of high-resolution incremental dynamic contrast-enhanced CT in the detection and staging of pancreatic carcinoma [3,4]. Although dynamic CT routinely can detect tumors as small as 1 to 1.5 cm in diameter as hypodense masses, its high degree of accuracy in this study (207 of 213, 97%) reflects to a large extent the fact that most symptomatic patients with pancreatic carcinoma had unresectable tumors at the time of initial clinical presentation. Only 25 (12%) patients had no CT evidence of extrapancreatic spread of tumor, whereas 188 (89%) had unresectable tumors by CT criteria, and 69% of these patients had 3 to 5 CT criteria of unresectability, indicating widespread disease (Table I). It should be emphasized that the CT findings of a pancreatic mass are nonspecific, even when ancillary criteria of unresectability are present. This nonspecificity was not addressed specifically in this paper, but it has been discussed in a previous report [3]. Focal pancreatitis, other types of pancreatic neoplasms, particularly nonfunctioning islet cell carcinoma, metastases, and lymphoma, and variations in size and shape of the normal pancreas can mimic the CT appearance of pancreatic carcinoma. Thus, biopsy or ancillary procedures (ERCP or angiography) are recommended to substantiate the CT diagnosis (Figure 1). Percutaneous biopsy of potentially resectable pancreatic carcinoma remains controversial. Some current treatment protocols that utilize radiotherapy and chemotherapy prior to resection require biopsy proof of tumor prior to starting therapy. Warshaw [5] has shown an increased positivity rate of peritoneal cytology in patients who have had percutaneous biopsy, but this has not been substantiated by other authors. Thus, the decision to biopsy potentially resectable tumors requires consideration of the institutional treatment protocols. Staging pancreatic carcinoma: Recent reports indicating improved survival after surgical resection of pancreatic carcinoma have led to increasing interest in more precise preoperative staging [6-10]. CT and angiography are widely used and highly accurate [3,11]. More recently, laparoscopy combined with peritoneal cytology has been recommended as a final staging procedure in patients with tumors that show no evidence of unresectability by CT and angiography [5,12]. A series of 40 patients with potentially resectable tumors by CT and angiography reported by Warshaw et al [12] underwent laparoscopy, and an additional 14 patients (35%) were found to
have small hepatic or peritoneal metastases that excluded them from surgical resection. In a subsequent report combining CT, angiography, and laparoscopy, CT and angiography alone were 92% and 95% accurate in predicting tumor unresectability, respectively, but were only 45% and 54% accurate in predicting tumor resectability [10]. Performing laparoscopy in patients with no signs of unresectability by CT and angiography increased the accuracy for predicting tumor unresectability to 89% and for predicting tumor resectability to 90%. In the current series, although 25 patients (12%) had CT-resectable tumors, only 18 of these patients underwent surgical exploration. The accuracy of CT in predicting tumor resectability in these 18 patients was 72% (13 of 18), substantially greater than the figure of 45% reported by Warshaw and colleagues [10]. The findings missed by CT in the five patients found to have unresectable tumors at the time of surgery included tumor invasion into the root of the small bowel mesentery (one) and transverse colon (one), metastatic lymph nodes in the porta hepatis (two), and posterior tumor extension with a metastatic node in the porta (one). Although some of these findings may have been found with laparoscopy, none of the patients was excluded from surgery only on the basis of small liver or peritoneal metastases. Reports by Warshaw [I0] and Dooley [13] and coworkers have indicated that angiography is superior to CT in staging pancreatic carcinoma. In the series of Dooley et al [13], angiography detected major vascular involvement in 31% of 90 patients with negative CT scans. Although Warshaw et al [10] did not directly compare angiography and CT, they indicated that CT was neither significantly superior nor inferior to angiography. Angiography was performed in 60 patients in our series and did not contribute any additional or significant staging information compared with dynamic CT. Preoperative angiography was been found to be valuable, however, in displaying vascular anatomic variations prior to pancreatic resection. Survival: The 97% accuracy of high-resolution CT in detecting pancreatic carcinoma, even its ability to detect routinely tumors as small as 1 to 2 cm, has had little effect on patient survival. This lack of impact is due to the fact that only 12% of patients had potentially resectable tumors and that only three of the patients who underwent resection for cure had no pathologic evidence of tumor spread, i.e., positive lymph nodes, perineural invasion, or local extension. In our experience, surgery has had little effect on survival. In our original series published in 1988, survival after resection for cure was 19.5 months [3]. In the current series, 11 patients underwent resection for cure. Seven patients died with a median survival of 22.7 months, and four patients are alive at a median of 15.5 months (range: 9 to 22 months) postoperatively. Five patients in the original series who had CT criteria of tumor unresectability underwent palliative resection and had a median survival of 14.6 months. In the current series, despite aggressive surgery, six patients had palliative resections with a median survival of only 14.4 months (one patient is alive at 6 months but has widespread hepatic metastases).
THE AMERICAN JOURNAL OF SURGERY
VOLUME 165
MAY 1993
605
FREENY ET AL
It seems clear that future improvements in survival are likely to come only when pancreatic carcinoma can be detected by a screening test prior to the time that the patient becomes symptomatic.
CONCLUSIONS High-resolution dynamic contrast-enhanced CT is highly accurate in the detection (97%) and staging (93%) of pancreatic carcinoma. Angiography was not found to add any significant information regarding staging but was helpful in defining the vascular anatomy prior to the anticipated resection. REFERENCES 1. Freeny PC, Marks WM, Ball TJ. Impact of high-resolution computed tomography of the pancreas on utilization of ERCP and angiography. Radiology 1982; 142: 35-9. 2. Freeny PC. Radiology of the pancreas: two decades of progress in imaging and intervention. AJR Am J Roentgenol 1988; 150: 975-81. 3. Freeny PC, Marks WM, Ryan JA, et al. Pancreatic ductal adenocarcinoma: diagnosis and staging with dynamic CT. Radiology 1988; 166: 125-33. 4. Ward EM, Stephens DH, Sheedy PF II. CT characteristics of
606
pancreatic carcinoma: an analysis of 100 cases. RadioGraphics 1983; 3: 547-65. 5. Warshaw AL. Implications of peritoneal cytology for staging of early pancreatic cancer. Am J Surg 1991; 161: 26-9. 6. Cameron JL, Crist DW, Sitzman JV, et al. Factors influencing survival after pancreaticoduodenectomy for pancreatic cancer. Am J Surg 1991; 161: 120-5. 7. Trede M, Schwall G, Saeger HD. Survival after pancreatoduodenectomy. Ann Surg 1990; 211: 447-58. 8. Tsuchiya R, Noda T, Harada M, et al. Collective review of small carcinomas of the pancreas. Ann Surg 1986; 203: 77-81. 9. Warshaw AL, Fern'andez del Castillo C. Pancreatic carcinoma. N Engl J Med 1992; 326: 455-65. 10. Warshaw AL, Gu Z-Y, Wittenberg J, Waltman AC. Preoperative staging and assessment of resectability of pancreatic cancer. Arch Surg 1990; 125: 230-3. 11. Bradley EL IlL Pseudocysts in chronic pancreatitis. In: Beger H, Biichler M, Ditchuneit H, Malfertheiner P, editors. Chronic pancreatitis. Berlin: Springer-Verlag, 1990: 306-96. 12. Warshaw AL, Tepper JE, Shipley WU. Laparoscopy in the staging and planning of therapy for pancreatic cancer. Am J Surg 1986; 151: 76-80. 13. Dooley WC, Cameron JL, Pitt HA, Lillemoe KD, Yue NC, Venbrux AC. Is preoperative angiography useful in patients with periampullary tumors? Ann Surg 1990; 211: 649-55.
THE AMERICAN JOURNAL OF SURGERY VOLUME 165 MAY 1993
We assessed the accuracy of dynamic contrast-enhanced computed tomography (CT) in the diagnosis and staging of 213 patients with pancreatic carcinoma and compared it with the accuracy of angiography and surgery. A correct CT diagnosis of pancreatic carcinoma was made in 207 of 213 (97%) patients. Tumors were located in the pancreatic head in 64%, the body in 22%, and the tail in 10%, and enlarged the pancreas diffusely in 4%. CT staged 25 (12%) patients as having potentially resectable tumors and 188 (88%) as having unresectable tumors on the basis of local extension (72%), contiguous organ invasion (4~%), vascular invasion ( 8 2 % ) , and distant metastases ( 5 0 % ) . Compared with angiography in 6 0 patients, CT detected vascular invasion missed on angiograpby in 20%, and angiography detected invasion missed by CT in 5%. In these latter cases, other CT criteria of unresectability were present, and angiography provided no significant staging information. Compared with surgery in 71 patients, c r accurately predicted unresectable tumors in 100% of patients and resectable tmnors in 72% of patients. Eleven of the patients with CT-resectable tumors underwent resection. Median survival was 22.7 months, with four patients alive at a median of 15.5 months postoperatively. Palliative resections were performed in six patients, and median survival was 14.4 months.
omputed tomography (CT) has become the primary modality for the evaluation of patients with suspected pancreatic carcinoma, virtually eliminating the use of endoscopic retrograde cholangiopancreatography (ERCP) and angiography for initial diagnosis [1,2]. The use of dynamic scanning techniques that employ automatic table incrementation during rapid power injection of a large intravenous bolus of iodinated contrast medium has significantly increased the efficacy of CT for the detection, diagnosis, and staging of pancreatic carcinoma. We assessed the accuracy of high-resolution dynamic CT and compared it with angiography and surgery in the detection and staging of 213 patients with pancreatic carcinoma.
C
PATIENTS AND M E T H O D S A prospective series of 213 incremental dynamic bolus CT scans of the abdomen performed in patients with ductal adenocarcinoma of the pancreas between 1982 and 1992 were reviewed. The diagnosis of pancreatic adenocarcinoma was confirmed by percutaneous biopsy in 80 patients (38%), surgical exploration and biopsy in 71 patients (33%), or with a combination of clinical follow-up studies and combined results of CT, ERCP, and angiography in 62 patients (29%). CT techniques: Incremental dynamic abdominal CT was performed with a General Electric C T / T 8800 or 9800 scanner (General Electric Medical Systems, Milwaukee, WI). Scans were obtained using contiguous 5- to 10-mm collimated slices during rapid intravenous bolus power injection of 150 mL of a 60% iodinated contrast agent injected at a rate of 2 to 2.5 mL,/sec for 20 to 25 seconds and 1 mL/sec for the remainder of the volume. The scan sequence was performed using individual breath-held scans or clusters of three scans beginning 30 to 40 seconds after the initiation of the contrast bolus. The entire liver and pancreas were scanned within 180 seconds in all patients. CT scan interpretation: The original CT radiographic reports were used to record the CT diagnosis, regardless of the findings on retrospective review of the CT scans. A C T diagnosis of pancreatic carcinoma was made if, in the absence of changes of chronic or acute pancreatiFrom the Departments of Radiology (PCF) and General Surgery tis, a focal or diffuse pancreatic mass was present with or (LWT, JAR), Universityof WashingtonSchoolof Medicine,Virginia without evidence of local or distant tumor spread, or if a Mason MedicalCenter, Seattle,Washington. dilated main pancreatic and common bile duct were idenRequests for reprints should be addressed to Patrick C. Freeny, tified to the level of the pancreatic head or ampulla in the MD, Department of Radiology SB-05, University of Washington absence of a pancreatic mass. Schoolof Medicine,Seattle,Washington98195. The 213 CT scans were reviewed retrospectively at Presentedat the 79th Annual Meetingof the North PacificSurgical Association,Tacoma,Washington,November13-14, 1992. the conclusion of the study, and the following 6 parame600
THE AMERICAN JOURNAL OF SURGERY VOLUME165 MAY 1993
PANCREATIC ADENOCARCINOMA
ters were evaluated: (1) pancreatic mass (focal or diffuse; location; tumor size; attenuation); (2) main pancreatic duct and intrahepatic and extrahepatic bile duct appearance (normal; not seen; dilated: main pancreatic duct dilated if the diameter was greater than 5 mm in the head or greater than 3 mm in the tail; intrahepatic ducts dilated if seen greater than 2 cm from the hilus and measuring greater than 5 mm in diameter; extrahepatic bile ducts dilated if common hepatic duct greater than 6 mm in diameter and common bile duct greater than 9 mm in diameter); (3) local tumor extension (anterior or posterior to the tumor or into the porta hepatis or splenic hilum); (4) contiguous organ invasion (stomach, duodenum, root of the small bowel mesentery; liver, spleen, adrenal gland, kidney); (5) vascular invasion or tumor-vessel contiguity (tumor partially or completely surrounding a major extrapancreatic artery or vein; arterial or venous obstruction with arterial collateral vessels or mesenteric varices; tumor-vessel contiguity: tumor immediately contiguous with vessel with no intervening normal fat plane); and (6) metastases (liver; lymph nodes). Staging of pancreatic carcinoma: Pancreatic carcinoma was staged on the basis of CT (213 patients), angiography (60 patients), and surgical findings (71 patients). CT staging, Tumors were staged as unresectable if one or more of the following four findings were seen: local tumor extension, contiguous organ invasion, vascular invasion, and hepatic metastases (isolated lymph node metastases, i.e., nodes larger than 2 cm in short axis, were not considered to indicate tumor unresectability). Angiographic staging. Tumors were considered to be unresectable if one or more of the following features were present: tumor encasement or obstruction of major extrapancreatic arteries or veins and hepatic metastases. Surgical staging. Tumors were considered to be unresectable for cure if one or more of the following findings were observed during laparotomy: hepatic or peritoneal metastases; direct tumor extension into the peripancreatic soft tissues, major extrapancreatic vascular structures, contiguous organs, or mesenteries. In some patients, the primary tumors were not resected if frozen section biopsy showed the presence of regional lymph node metastases. RESULTS CT diagnosis: The prospective CT diagnoses in the 213 patients with pancreatic carcinoma were pancreatic carcinoma (207 of 213, 97%), normal CT (4 of 213, 2%), duodenal carcinoma (1 of 213, 0.5%), and panereatitis ( 1 of 213, 0.5%). Normal CTscan. Retrospective review of the four CT scans interpreted as normal showed no abnormality in two patients. The third patient had a small, hypodense mass that was not mentioned on the CT report, despite having been identified on a preoperative sonogram. At surgery, the patient was believed to have acute biliary pancreatitis, and a cholecystectomy was performed. Repeat CT 6 weeks later showed a 4-cm mass in the pancreatic head with local extension and vascular invasion. The fourth patient had pancreatic head enlargement. A re-
TABLE I C T Findings in 213 Patients with Pancreatic Carcinoma*
CT Findings
Total (%)
Tumor mass Focal Diffuse Decreased attenuation
205 197 8 167
MPD/CBD dilatation Biductal (no mass) MPD only Biductal (mass) CBD only
168 (79)
Contiguous organ invasion
(96) (92) (4) (6)
Head (%)
Location of Tumor Body Tail (%) (%)
129 (61) 47 (22)
21 (10)
102 (48) 43 (20)
16 (8)
6 (3)
6 (3)
25 {12) 117 (55) 28 {12)
20 (9) 109 (51) 22 (10)
5 (2) 8 (4) 3 (1)
Diffuse (%)
6 (3)
1 (0.5)
91 (43)
57 (27) 15 (7)
14 (7)
5 (2)
Local tumor extension
154 (72)
89 (42) 39 (I 8)
19 (9)
7 (3)
Vascular invasion Arterial only Venous only Arterial + venous
175 7 48 120
(82) (3) (23) (56)
6 (3) 1 (0.4) 32 (15) 6 (3) 70 (33) 33 (15)
0 9 (4) 10 (5)
0 1 (0.4) 7 (3)
Metastases Liver Lymph nodes
106 (50) 72 t34) 57 (27)
34 (16) 22 (10) 29 (14) 18 (8)
13 (6) 6 (3)
3 (1) 4 (2)
CT = computed tomography; MPD = main pancreatic duct: CBD = common bile duct. *Numbers in parenthesesreferto the percentageof the total (213) patients.
peat scan 8 weeks later showed a 5-cm mass in the pancreatic head and liver metastases. Incorrect CT scan interpretation. Two patients had pancreatic masses that were interpreted incorrectly. One patient was believed to have a primary duodenal carcinoma. CT showed no fat plane between the tumor and the caudal margin of the pancreatic head and no pancreatic or common bile duct dilatation. At surgery, a primary carcinoma of the pancreatic head was found growing inferiorly and invading the duodenum and the root of the small bowel mesentery. The second patient had enlargement of the pancreatic head and peripancreatic soft tissue changes believed to represent pancreatitis. ERCP showed segmental obstruction of the main pancreatic and common bile duct in the head of the pancreas and intraductal filling defects within a dilated upstream pancreatic duct. A pancreatic duct stent was placed, and percutaneous biopsy was positive for adenocarcinoma. Surgical exploration showed the tumor to be unresectable due to extension into the small bowel mesentery.
CT findings of pancreatic 'carcinoma (Table I): The pancreatic tumors were located in the head of the gland in 137 patients (64%), the body in 47 (22%), and the tail in 21 (10%), and diffusely involved the gland in 8 (4%). The CT findings in the 213 patients with pancreatic carcinoma are detailed in Table I.
THE AMERICAN JOURNAL OFSURGERY
VOLUME 165
MAY 1993
601
FREENY ET AL
Figure 1. Large unresectable pancreatic carcinoma. Top. Dynamic CT shows a large mass (M) in the head of the pancreas, dilatation of main pancreatic duct in the body and tail with atrophy of the surrounding parenchyma (arrows), and posterior tumor extension (curved arrows} to the aorta (a) and inferior vena cava (c). s -superior mesenteric vein. Middle. Scan at a lower level shows tumor (T) encasing the celiac and hepatic arteries (arrows) and contiguous with the portal vein (pv). Bottom. CT-guided biopsy shows needle (arrow) within the mass. 602
THE A M E R I C A N J O U R N A L OF SURGERY
VOLUME 165
MAY 1993
PANCREATIC ADENOCARCINOMA
Figure 2. Small unresectable pancreatic carcinoma. Dynamic CT at the level of the pancreatic head shows a 1.8-cm hypodense tumor (arrowheads) extending anteriorly into the peripancreatic soft tissues between (arrow) and encasing the gastroduodenal artery (A) and the anterior pancreaticoduodenalvein (V). Arterial enc a s e m e n t was c o n f i r m e d on angiography. S = superior masenteric vein; D = duodenum.
Figure 3. Resectable pancreatic carcinoma. Top. Dynamic CT shows normal pancreatic (P) body and tail with no duct dilatation. PV = portal vein. Bottom. Scan at the level of the pancreatic head shows a 10-mm hypodense tumor (arrowhead) confined to the uncinate, a --- superior mesenteric artery; v = superior mesenteric vein. THE AMERICAN JOURNAL OFSURGERY
VOLUME 165
MAY 1993
603
FREENY ET AL
The most common compilation of findings was a pancreatic mass (96%), pancreatic and/or common bile duct dilatation, and one or more findings of extrapancreatic tumor extension (88%) (Figures 1 and 2). Extrapancreatic tumor extension encompassed the following four findings: (1) tumor involvement of contiguous organs (91 of 213 patients, 43%), most commonly the duodenum, stomach, root of the small bowel mesentery, and spleen; (2) local extension of the tumor into the anterior or posterior peripancreatic fat and soft tissues, the porta hepatis, and the splenic hilum (154 of 213 patients, 72%); (3) arterial encasement or obstruction by tumor (127 of 213 patients, 60%), most often involving the celiac, splenic, superior mesenteric, hepatic, and gastroduodenal arteries; and (4) venous encasement or obstruction with associated abdominal varices (168 of 213 patients, 79%), most commonly involving the splenic and superior mesenteric veins. Isolated biductal dilatation (i.e., no associated tumor mass seen by CT) was seen in six patients. The remaining 162 (76%) patients with pancreatic and/or bile duct dilatation had associated pancreatic masses. Correlation of angiography, CT, and surgical findings: Angiography was performed in 60 patients, and the findings were correlated with the corresponding dynamic CT findings in each patient and with the findings at surgery in 41 patients. Angiography and CT showed equal and similar findings in 45 (75%) patients. CT showed vascular involvement (tumor-vessel contiguity or encasement of the superior mesenteric artery a n d / o r vein) not seen by angiography in 12 (20%) patients, and, in 3 patients (5%), vascular involvement (common hepatic artery, replaced right hepatic artery, gastroduodenal artery) was seen angiographically but missed by CT. Additional surgical correlation was made in 41 patients. Angiography and CT showed similar and equal findings, which were confirmed at surgery in 28 patients (12 with CT-resectable tumors and 16 with CT-unresectable tumors), whereas CT showed vascular involvement not demonstrated by angiography in 11 patients (tumorvessel contiguity in 8 patients, vascular encasement in 3 patients). In two patients with unresectable tumors, vascular involvement not seen by CT was demonstrated angiographically and confirmed at surgery. Overall, angiography showed vascular involvement not seen by CT in only 3 of 60 patients (5%), but other CT criteria of tumor unresectability were present in all 3 patients. Thus, angiography contributed no additional or significant information to the staging of pancreatic carcinoma. Staging: CT and surgical results: CT-resectable tumors. Twenty-five patients (25 of 213, 12%) had potentially resectable tumors on the basis of CT (i.e., no identifiable CT criteria of unresectability) (Figure 3). This group included two patients with normal CT scans (tumor subsequently identified by ERCP and confirmed by percutaneous biopsy in both patients), the patient with pancreatic head enlargement and the patient with a small tumor mass in the pancreatic head, both missed on the preoperative CT scans, and the patient believed to have pancreatitis. 604
Six (6 of 25, 24%) of the patients who had CT-resectable tumors were not deemed to be operative candidates due to severe coexisting medical diseases. One patient was believed to have biliary pancreatitis at the time of surgery and was discussed earlier. The remaining 18 (18 of 25, 72%) patients with CT-resectable tumors underwent laparotomy. Twelve (12 of 18, 67%) of these patients were found to have resectable tumors, and 11 underwent resection (8 pancreaticoduodenectomies for carcinoma of the head of the pancreas; 2 total pancreatectomies for carcinoma of the head and body of the pancreas; 1 distal pancreatectomy for carcinoma of the tail of the pancreas). The other patient had severe cardiovascular disease, was not believed to be able to tolerate a pancreatic resection at the time of laparotomy, and thus had only a palliative drainage procedure. Seven of the 11 patients who underwent resection have died (range: 7 to 42 months; mean: 22.7 months), and 4 are alive (range: 9 to 22 months; mean: 15.5 months). Ten of the 11 resected tumors were in the head of the pancreas, and 1 was in the tail. The tumors ranged in size on the resected specimens from 2 to 3 cm in maximum diameter. Three patients had no pathologic evidence of local or regional spread of tumor. The other eight patients showed positive lymph nodes (six), perineural invasion (three), duodenal wall invasion (two), and posterior extension of the tumor into the retropancreatic soft tissues (one). Five (5 of 25, 20%) of the patients with CT-resectable tumors were found to have unresectable tumors at the time of surgery and had only palliative procedures. The tumors in these patients were in the head of the pancreas and ranged in size on the CT scans from 2 to 3.5 cm in diameter. In these patients, CT missed tumor invasion into the root of the small bowel mesentery (one) and transverse colon (one), metastatic lymph nodes in the porta hepatis (two), and posterior tumor extension with a metastatic node in the porta (one). CT-unresectable tumors. One hundred eighty-eight patients (188 of 213, 88%) had CT criteria of tumor unresectability as detailed above. Most of these patients (69%) had three to five CT criteria of tumor unresectability, whereas only 31% had one or two criteria. The 20 patients with only 1 criteria had either hepatic metastases, contiguous organ invasion, local tumor extension, or vascular invasion. Fifty-three patients with CT-unresectable tumors underwent laparotomy. Forty-seven (89%) were found to have unresectable tumors and had palliative drainage procedures (biliary, gastric), whereas 6 patients (12%) had surgical confirmation of the CT criteria of unresectability but had palliative resections. Five of these patients have died (range: 6 to 27 months; mean, 14.4 months), and one is alive at 6 months but has widespread hepatic metastases. Two patients had body-tail tumors with splenic vein obstruction and perigastric varices. They developed recurrent tumor at 10 and 11 months postoperatively and died at 15 and 27 months, respectively. Three patients had superior mesenteric vein involvement, and one of these had duodenal invasion. One patient developed liver metastases 4 weeks after pancreaticoduodenec-
THE AMERICAN JOURNAL OF SURGERY VOLUME165 MAY 1993
PANCREATIC ADENOCARCINOMA
tomy and is alive at 6 months, one patient developed postoperative superior mesenteric vein thrombosis and died at 6 months, and the third patient developed postoperative superior mesenteric artery thrombosis and died at 6 months. The final patient had duodenal invasion and positive lymph nodes, developed recurrent tumor at 10 months, and died at 15 months. Thus, CT correctly predicted unresectable tumors in 100% of cases (53 of 53) but correctly predicted resectable tumors in only 72% of cases (13 of 18). COMMENTS CT diagnosis of pancreatic carcinoma: This study confirms the high degree of accuracy of high-resolution incremental dynamic contrast-enhanced CT in the detection and staging of pancreatic carcinoma [3,4]. Although dynamic CT routinely can detect tumors as small as 1 to 1.5 cm in diameter as hypodense masses, its high degree of accuracy in this study (207 of 213, 97%) reflects to a large extent the fact that most symptomatic patients with pancreatic carcinoma had unresectable tumors at the time of initial clinical presentation. Only 25 (12%) patients had no CT evidence of extrapancreatic spread of tumor, whereas 188 (89%) had unresectable tumors by CT criteria, and 69% of these patients had 3 to 5 CT criteria of unresectability, indicating widespread disease (Table I). It should be emphasized that the CT findings of a pancreatic mass are nonspecific, even when ancillary criteria of unresectability are present. This nonspecificity was not addressed specifically in this paper, but it has been discussed in a previous report [3]. Focal pancreatitis, other types of pancreatic neoplasms, particularly nonfunctioning islet cell carcinoma, metastases, and lymphoma, and variations in size and shape of the normal pancreas can mimic the CT appearance of pancreatic carcinoma. Thus, biopsy or ancillary procedures (ERCP or angiography) are recommended to substantiate the CT diagnosis (Figure 1). Percutaneous biopsy of potentially resectable pancreatic carcinoma remains controversial. Some current treatment protocols that utilize radiotherapy and chemotherapy prior to resection require biopsy proof of tumor prior to starting therapy. Warshaw [5] has shown an increased positivity rate of peritoneal cytology in patients who have had percutaneous biopsy, but this has not been substantiated by other authors. Thus, the decision to biopsy potentially resectable tumors requires consideration of the institutional treatment protocols. Staging pancreatic carcinoma: Recent reports indicating improved survival after surgical resection of pancreatic carcinoma have led to increasing interest in more precise preoperative staging [6-10]. CT and angiography are widely used and highly accurate [3,11]. More recently, laparoscopy combined with peritoneal cytology has been recommended as a final staging procedure in patients with tumors that show no evidence of unresectability by CT and angiography [5,12]. A series of 40 patients with potentially resectable tumors by CT and angiography reported by Warshaw et al [12] underwent laparoscopy, and an additional 14 patients (35%) were found to
have small hepatic or peritoneal metastases that excluded them from surgical resection. In a subsequent report combining CT, angiography, and laparoscopy, CT and angiography alone were 92% and 95% accurate in predicting tumor unresectability, respectively, but were only 45% and 54% accurate in predicting tumor resectability [10]. Performing laparoscopy in patients with no signs of unresectability by CT and angiography increased the accuracy for predicting tumor unresectability to 89% and for predicting tumor resectability to 90%. In the current series, although 25 patients (12%) had CT-resectable tumors, only 18 of these patients underwent surgical exploration. The accuracy of CT in predicting tumor resectability in these 18 patients was 72% (13 of 18), substantially greater than the figure of 45% reported by Warshaw and colleagues [10]. The findings missed by CT in the five patients found to have unresectable tumors at the time of surgery included tumor invasion into the root of the small bowel mesentery (one) and transverse colon (one), metastatic lymph nodes in the porta hepatis (two), and posterior tumor extension with a metastatic node in the porta (one). Although some of these findings may have been found with laparoscopy, none of the patients was excluded from surgery only on the basis of small liver or peritoneal metastases. Reports by Warshaw [I0] and Dooley [13] and coworkers have indicated that angiography is superior to CT in staging pancreatic carcinoma. In the series of Dooley et al [13], angiography detected major vascular involvement in 31% of 90 patients with negative CT scans. Although Warshaw et al [10] did not directly compare angiography and CT, they indicated that CT was neither significantly superior nor inferior to angiography. Angiography was performed in 60 patients in our series and did not contribute any additional or significant staging information compared with dynamic CT. Preoperative angiography was been found to be valuable, however, in displaying vascular anatomic variations prior to pancreatic resection. Survival: The 97% accuracy of high-resolution CT in detecting pancreatic carcinoma, even its ability to detect routinely tumors as small as 1 to 2 cm, has had little effect on patient survival. This lack of impact is due to the fact that only 12% of patients had potentially resectable tumors and that only three of the patients who underwent resection for cure had no pathologic evidence of tumor spread, i.e., positive lymph nodes, perineural invasion, or local extension. In our experience, surgery has had little effect on survival. In our original series published in 1988, survival after resection for cure was 19.5 months [3]. In the current series, 11 patients underwent resection for cure. Seven patients died with a median survival of 22.7 months, and four patients are alive at a median of 15.5 months (range: 9 to 22 months) postoperatively. Five patients in the original series who had CT criteria of tumor unresectability underwent palliative resection and had a median survival of 14.6 months. In the current series, despite aggressive surgery, six patients had palliative resections with a median survival of only 14.4 months (one patient is alive at 6 months but has widespread hepatic metastases).
THE AMERICAN JOURNAL OF SURGERY
VOLUME 165
MAY 1993
605
FREENY ET AL
It seems clear that future improvements in survival are likely to come only when pancreatic carcinoma can be detected by a screening test prior to the time that the patient becomes symptomatic.
CONCLUSIONS High-resolution dynamic contrast-enhanced CT is highly accurate in the detection (97%) and staging (93%) of pancreatic carcinoma. Angiography was not found to add any significant information regarding staging but was helpful in defining the vascular anatomy prior to the anticipated resection. REFERENCES 1. Freeny PC, Marks WM, Ball TJ. Impact of high-resolution computed tomography of the pancreas on utilization of ERCP and angiography. Radiology 1982; 142: 35-9. 2. Freeny PC. Radiology of the pancreas: two decades of progress in imaging and intervention. AJR Am J Roentgenol 1988; 150: 975-81. 3. Freeny PC, Marks WM, Ryan JA, et al. Pancreatic ductal adenocarcinoma: diagnosis and staging with dynamic CT. Radiology 1988; 166: 125-33. 4. Ward EM, Stephens DH, Sheedy PF II. CT characteristics of
606
pancreatic carcinoma: an analysis of 100 cases. RadioGraphics 1983; 3: 547-65. 5. Warshaw AL. Implications of peritoneal cytology for staging of early pancreatic cancer. Am J Surg 1991; 161: 26-9. 6. Cameron JL, Crist DW, Sitzman JV, et al. Factors influencing survival after pancreaticoduodenectomy for pancreatic cancer. Am J Surg 1991; 161: 120-5. 7. Trede M, Schwall G, Saeger HD. Survival after pancreatoduodenectomy. Ann Surg 1990; 211: 447-58. 8. Tsuchiya R, Noda T, Harada M, et al. Collective review of small carcinomas of the pancreas. Ann Surg 1986; 203: 77-81. 9. Warshaw AL, Fern'andez del Castillo C. Pancreatic carcinoma. N Engl J Med 1992; 326: 455-65. 10. Warshaw AL, Gu Z-Y, Wittenberg J, Waltman AC. Preoperative staging and assessment of resectability of pancreatic cancer. Arch Surg 1990; 125: 230-3. 11. Bradley EL IlL Pseudocysts in chronic pancreatitis. In: Beger H, Biichler M, Ditchuneit H, Malfertheiner P, editors. Chronic pancreatitis. Berlin: Springer-Verlag, 1990: 306-96. 12. Warshaw AL, Tepper JE, Shipley WU. Laparoscopy in the staging and planning of therapy for pancreatic cancer. Am J Surg 1986; 151: 76-80. 13. Dooley WC, Cameron JL, Pitt HA, Lillemoe KD, Yue NC, Venbrux AC. Is preoperative angiography useful in patients with periampullary tumors? Ann Surg 1990; 211: 649-55.
THE AMERICAN JOURNAL OF SURGERY VOLUME 165 MAY 1993