First published online July 5, 2006
British Journal of Radiology (2006) 79, 880-887
© 2006 British Institute of Radiology
doi: 10.1259/bjr/19684199
Pancreatic adenocarcinoma: signs of vascular invasion determined by multi-detector row CT
H Li, MD
1
M S Zeng, MD, Pro
2
K R Zhou, MD, Pro
2
D Y Jin, MD, Pro
3 and
W H Lou, MD, Pro
3
1Department of Radiology, The Affiliated First People's Hospital, Shanghai Jiao Tong University, 85 Wujin Road, 200080, Shanghai, Departments of 2Radiology and 3General Surgery, The Affiliated Zhongshan Hospital, Fudan University, 180 Fenglin Road, 200032, Shanghai, P R China
 |
Abstract
|
|---|
The purpose of this study was to analyse multi-detector row CT (MDCT) signs of peripancreatic arterial and venous invasion in pancreatic carcinoma. Among 101 patients with pancreatic carcinoma examined by MDCT, 54 candidates for surgery were pre-operatively evaluated for vascular invasion based on MDCT signs. The peripancreatic major vessels (including superior mesenteric artery, coeliac artery, common hepatic artery, superior mesenteric vein and portal vein) were examined carefully by surgeons during the operation. At surgical exploration, 78 of 224 vessels were invaded by tumour. The invaded peripancreatic major arteries (n = 29) and veins (n = 49) presented different MDCT signs: 43% of invaded veins (18/42, except for 7 occluded veins) were surrounded by tumour less than 50% of the vessel circumference compared with 97% (28/29) of the invaded arteries, which were surrounded by tumour more than 50% of the vessel circumference or were embedded in tumour (p<0.001). 69% (34/49) of the invaded veins had vascular stenosis or obliteration, compared with 41% (12/29) of the invaded arteries (p<0.05). Irregularity of the vein wall, 74% (31/42, except for 7 occluded veins); occurred more often than that of the artery wall, 45% (13/29) (p<0.05). In conclusion, the MDCT signs of peripancreatic arterial and venous invasion have different characteristics, which should be considered in pre-operative evaluation.
|
Introduction
|
|---|
One decisive factor for the irresectability of pancreatic carcinoma is the major vascular invasion [1]. Multi-detector row CT (MDCT) scanning of the pancreas enables multiphasic thin collimation scanning, with excellent spatial resolution, especially in the z-plane [2]. These volume data sets can be easily manipulated with three-dimensional imaging, potentially providing additional information to conventional axial display [3–6].
The purpose of our study was to assess the MDCT signs of arterial and venous invasion in pancreatic carcinoma.
|
Materials and methods
|
|---|
Patients
Between December 2001 and February 2004, 255 consecutive patients with presumed pancreatic carcinoma underwent pancreas examinations by triphasic MDCT. 101 patients with pancreatic carcinoma were confirmed by surgical-pathology (n = 54) or clinical follow-up (n = 47). 54 patients were pre-operatively judged to be resectable or requiring surgical bypass due to jaundice. Pancreatic adenocarcinoma was pathologically confirmed by biopsy or fine-needle aspiration. The remaining 47 patients deemed irresectable received chemotherapy or radiotherapy, and all of them survived less than 2 years.
MDCT images of the 101 patients were all prospectively evaluated for resectability: including vascular invasion and the presence of metastatic disease. Of the 54 candidates for surgery (37 males and 17 females, 40–79 years old, average 61.2 years), 17 patients underwent pancreaticoduodenectomy, and the others were found to be irresectable during surgical exploration. The time interval between MDCT examination and surgery was 
2 weeks.
Examination techniques
All MDCT examinations of the pancreas were performed on Mx8000 (Philips, Best, The Netherlands; four-slice spiral CT scanners). 600–800 ml water was routinely administered 10–20 min before the examination to distend the stomach, duodenum and proximal jejunum. Each patient received 120 ml of non-ionic contrast material (iopromide 350 mg ml–1; Ultravist; Schering, Berlin, Germany) via intravenous injection at the rate of 5 ml s–1.
Unenhanced and triphasic (arterial phase, pancreatic phase and hepatic phase) enhanced scans were performed. Unenhanced scan images were obtained from the top of the diaphragm to the caudal level of the uncinate process. The scanning parameters were: 120 kVp, 250 mA, a 0.875 pitch and 5 mm collimation. After a 20 s delay from the start of the intravenous infusion, the arterial phase was obtained from the level of the hepatic hilum through the entire pancreas. The collimation was altered to 2.5 mm, and the other parameters were the same as those of the unenhanced scan. The delay time of the pancreatic phase and the hepatic phase was 45 s and 80 s, respectively. The scanning range and parameters were the same as those of the arterial phase and the unenhanced scans, respectively.
After scanning, the data were transferred to the image server linked to an Mxview workstation with software version 3.5. This workstation was used for three-dimensional analysis of the local anatomy utilizing volume rendering (VR) combined with maximum intensity projection (MIP) and multiplanar reconstruction (MPR).
Image analysis and surgical correlation
CT signs of peripancreatic major vessels (including coeliac artery (CA); common hepatic artery (CHA); superior mesenteric artery (SMA); portal vein (PV); superior mesenteric vein (SMV)) were determined pre-operatively at consensus reading of the axial and 3D images by two observers (MSZ and HL, 10 years and 5 years of experience reading pancreatic imaging, respectively):
- Contiguity of tumour with the adjacent vessel was graded A
D: (grade A, fat plane or normal pancreatic tissue visible between tumour and vessel; grade B and grade C, tumour surrounding of less than and more than 50% of the vessel circumference; grade D, arterial embedment in tumour or venous occlusion.)
- Detailed vascular anatomic deformation: vessel stenosis presented a semi-circular or concentric smaller contour of the vessel. But a straight contour on one side was regarded as flattened, not as stenosed. Vessel wall infiltration presented an irregular and indented shape at the vascular margin abutting tumour.
224 vessels were examined carefully at surgery by two experienced surgeons (DYJ, WHL, 12 years and 8 years of experience operating pancreatic carcinoma, respectively) together. The surgical criterion of vascular ingrowth was that the vessel could be observed, or found by palpation, to be infiltrated or occluded at surgery. It was not considered vascular invasion if tumour was adherent to but could be separated from the vessel due to inflammatory or fibrotic reaction.
Statistical analysis
Chi-square tests were performed on the CT signs: (a) tumour surrounding more than 50% of the vessel circumference; (b) vessel stenosis or occlusion; (c) vessel wall irregularity, to check for any significant difference between the invaded arteries and veins. p < 0.05 was required to show statistical significance.
|
Results
|
|---|
43 of the 54 tumours at surgery were located in the pancreatic head or uncinate process, 11 in the body or tail. 17 tumours were surgically resectable, of which 16 were pre-operatively correctly diagnosed by MDCT. One was incorrectly judged irresectable because MDCT had shown tumour ingrown in the SMA (Figure 1
). Of 37 patients who received by-pass palliative surgery or surgical exploration, 6 tumours were incorrectly diagnosed resectable due to underdiagnosed invaded vessels (n = 4), missed hepatic metastases (n = 1) and peritoneal metastases (n = 1).
View larger version (50K):
[in this window]
[in a new window]
|
Figure 1. Pancreatic body carcinoma.(a,b) Consecutive axial images showed that the superior mesenteric artery (SMA) was surrounded to more than 50% of the vessel circumference by tumour and the vessel wall appeared infiltrated. Pre-operatively, this case was judged irresectable. At surgical exploration, the SMA was found not to be invaded (only fibrotic infiltration) by tumour and was successfully resected.
|
|
Of the 224 observed vessels, 146 vessels were found not to be invaded at surgery and 78 invaded. Except for the above-mentioned SMA, which MDCT incorrectly judged to be locally infiltrated by tumour on MDCT, the calibre of the other 145 not invaded vessels was unchanged and the vessel wall was regular. On MDCT, the correlation between the surgically confirmed vessels that were not invaded and tumour was:
- Grade A of contiguity of tumour with the adjacent vessel (fat plane or normal pancreatic tissue preserved) (n = 129, 22 SMAs, 29 CAs, 30 CHAs, 18 SMVs and 30 PVs)
- Grade B of contiguity of tumour with the adjacent vessel (<180°) (n = 9, 3 SMAs, 1 CA, 3 SMVs, 2 PVs)
- Grade C of contiguity of tumour with the adjacent vessels (>180°) (n = 8, 5 SMAs, 2 CAs, 1 CHA) (Figure 2)
View larger version (61K):
[in this window]
[in a new window]
|
Figure 2. Tumour in the uncinate process of the head of the pancreas.(a,b) Consecutive axial images showed the superior mesenteric artery (SMA) to be surrounded more than 50% of the vessel circumference by tumour. It was found not to be invaded by tumour at exploration and tumour was successfully resected.
|
|
The MDCT appearances of surgically confirmed invaded arteries (n = 29) and veins (n = 49) are shown in Tables 1 and 2, respectively. Statistical analysis related to invaded arteries and veins is shown in Table 3.
Circumferential involvement
43% of invaded veins (18/42, except for 7 occluded veins) were surrounded by less than 50% of the vessel circumference by tumour. 97% (28/29) of invaded arteries were surrounded by more than 50% of the vessel circumference by tumour or were embedded in tumour. Statistical analysis showed a significant difference between invaded arteries and veins (except for seven occluded veins) utilizing circumferential involvement of more than 50% of the vessel circumference by tumour (p<0.001).
Stenosis or occlusion
69% (34/49) of invaded veins presented stenosed or occluded (Figure 3). Eight veins (5 SMVs, 3 PVs), which were surrounded less than 50% of the vessel circumference by tumour, also appeared stenosed (Figure 4). 41% (12/29) of invaded arteries did not appear stenosed, and the difference was statistically significant (p<0.05). Nine arteries (2 CAs, 3 CHAs, 4 SMAs), embedded in tumour, did not appear stenosed (Figure 5).
View larger version (108K):
[in this window]
[in a new window]
|
Figure 3. Tumour in the uncinate process of the head of the pancreas.(a,b) Axial images and (c,d) 3D images showed that the tumour in the uncinate process of the head of the pancreas eroded a side of the superior mesenteric vein (SMV) vessel wall and penetrated it to form tumour thrombus (arrow). Axial images also showed that the superior mesenteric artery (SMA) was embedded. Surgery confirmed tumour invasion of the SMV and the SMA.
|
|
View larger version (61K):
[in this window]
[in a new window]
|
Figure 4. Pancreatic head carcinoma.(a,b) Consecutive axial images showed the portal vein (PV) vessel calibre (arrow) was changed, although it was surrounded less than 50% of the vessel circumference by tumour. (c) Volume rendering (VR) 3D image showed that a segment of PV was stenosed. The axial images also showed that the common hepatic artery (CHA) was embedded in tumour and the coeliac artery (CA) was surrounded by more than 50% of the vessel circumference. PV, CHA and CA were proven to be infiltrated at surgical exploration.
|
|
View larger version (69K):
[in this window]
[in a new window]
|
Figure 5. Tumour in the uncinate process of the head of the pancreas.(a,b) Axial images showed coeliac artery (CA) (short arrow) and common hepatic artery (CHA) (long arrow) embedded in tumour with regular vessel walls, although (c) volume rendering (VR) 3D image showed that the vessel calibre of CA and CHA was unchanged. Surgery confirmed tumour invasion of CA and CHA.
|
|
Infiltration
74% of invaded veins (31/42, except for 7 occluded veins) appeared infiltrated with an irregular wall. Three veins (2 SMVs, 1 PV) were found to be invaded only from their irregular wall, were surrounded by less than 50% of the vessel circumference and were not stenosed (Figure 6). 45% (13/29) of invaded arteries appeared infiltrated less often than invaded veins (except for 7 occluded veins) (p<0.05).
View larger version (125K):
[in this window]
[in a new window]
|
Figure 6. Pancreatic head carcinoma. The confluence of the portal vein(PV) and the superior mesenteric vein (SMV) was shown to be surrounded by less than 50% of the vessel circumference by tumour, with unchanged vessel calibre. But a focal area of vessel wall (arrow) was irregular and infiltrated. The confluence of the PV and the SMV was proven to be infiltrated at surgical exploration.
|
|
Other MDCT appearances of arterial and venous invasion
24% (7/29) of invaded arteries appeared deformed and seemed to be stretched or stiffened in three-dimension reconstruction images, such as VR (Figure 7). 10% (3/30) of invaded SMVs were diagnosed correctly because they appeared as "teardrop" in axial images (Figure 8). When invaded veins were highly stenosed or occluded, 20% (10/49), there were multiple venous collaterals.
View larger version (71K):
[in this window]
[in a new window]
|
Figure 7. Pancreatic head carcinoma.(a,b) Consecutive axial images showed that the superior mesenteric artery (SMA) was embedded in tumour. (c) Volume rendering (VR) 3D image showed a segment of the SMA to be stiffened (arrow) where the vessel calibre was low-grade stenosed. Surgery confirmed tumour invasion.
|
|
View larger version (107K):
[in this window]
[in a new window]
|
Figure 8. Tumour(T) in the uncinate process of the head of the pancreas tethered the superior mesenteric vein (SMV) (arrow) into a teardrop shape, although the SMV was surrounded by less than 50% of the vessel circumference and vessel wall remained regular. The superior mesenteric artery (SMA) (arrowhead) was also shown to be nearly embedded in tumour. The SMV and the SMA were proven to be infiltrated at surgical exploration.
|
|
|
Discussion
|
|---|
According to previous reports, using single-detector row CT the diagnostic accuracy of pancreatic irresectability was 95% and of pancreatic resectability was 7080% [7–10]. The most important reason for inaccurate assessment of resectability is underestimation of the vascular invasion.
To improve the accuracy of estimating invaded vessels, it is necessary to evaluate the MDCT signs of arterial and venous invasion separately because we have found clinically that the features of peripancreatic arterial and venous invasion on CT are different.
In our opinion, the major reason for the different CT signs of arterial and venous invasion is that the vein wall is thinner and weaker than the artery wall. When veins are surrounded and infiltrated by tumour, the wall tends to be irregular and the calibre becomes narrowed. At the same time, the flow rate in veins is slow, and tumour often penetrates the vein wall to form tumour thrombus, causing vascular occlusion (Figure 3
). As the artery wall is thicker and more flexible than the vein wall, and the artery calibre is smaller, even when the arteries are embedded in tumour, the calibre is not easily changed and the wall remains regular. In our study, some arteries were found not to be invaded at surgical exploration, although they were surrounded by more than 50% of the vessel circumference, and their calibre remained unchanged and the wall was regular.
Hough et al [11] found that tumours in the head of the pancreas could cause a tethered, teardrop appearance of the SMV in axial images. They believed that the teardrop SMV sign was a reliable indicator of irresectability. In their retrospective study, teardrop SMV was the only sign of irresectability in 13 of 17 patients. This sign presumably results from either direct tumour infiltration or peritumoural fibrosis adherent to the vessel that retracts or tethers the vessel, changing its normal round shape. In our series, there were three invaded SMVs confirmed by surgical exploration exhibiting the teardrop sign on axial images in which the vessels were surrounded by less than 50% of the vessel circumference, and the wall of which was regular.
Since pancreatic carcinoma may be accompanied by focal tissue fibrosis, the invaded arteries may appear stretched. In our study, seven stretched arteries were appreciated on three-dimensional reconstruction MDCT angiography (MDCTA) images [3].
Lepanto et al [12] found that CTA significantly increased the ability to identify venous invasion but did not improve the detection of arterial invasion. There are few MDCTA cases reported. As our experience has shown, the image quality of MDCTA exceeds that of CTA. MDCTA in particular allows the three-dimensional display of the relationship between tumour and vessel, but conventional axial images should also be reviewed as axial images appear better able to demonstrate the contiguity of tumour to vessel and change in calibre of the vessel wall.
A major limitation of our study is our gold standard of surgical palpation, because irresectable tumours could not be histologically assessed, although the surgical margins of resected tumours were confirmed by pathology.
In conclusion, invaded peripancreatic arterial and venous MDCT signs are different. It is important to pay attention to these differences in order to improve the accuracy of diagnosing vascular invasion and pancreatic resectability.
Received for publication July 28, 2005.
Revision received November 17, 2005.
Accepted for publication May 2, 2006.
|
References
|
|---|
- Hommeyer SC, Freeny PC, Crabo LG. Carcinoma of the head of the pancreas: evaluation of the pancreaticoduodenal veins with dynamic CT-potential for improved accuracy in staging. Radiology 1995;196:233–8.[Abstract/Free Full Text]
- McNulty NJ, Francis IR, Platt JF, Cohan RH, Korobkin M, Gebremariam A. Multi-detector row helical CT of the pancreas: effect of contrast-enhanced multiphasic imaging on enhancement of the pancreas, peripancreatic vasculature, and pancreatic adenocarcinoma. Radiology 2001;220:97–102.[Abstract/Free Full Text]
- Horton KM, Fishman EK. Multidetector CT angiography of pancreatic carcinoma: part 1, evaluation of arterial involvement. AJR Am J Roentgenol 2002;178:827–31.[Free Full Text]
- Horton KM, Fishman EK. Multidetector CT angiography of pancreatic carcinoma: part 2, evaluation of venous involvement. AJR Am J Roentgenol 2002;178:833–6.[Free Full Text]
- Horton KM, Fishman EK. 3D CT angiography of the celiac and superior mesenteric arteries with multidetector CT data sets: preliminary observations. Abdom Imaging 2000;25:523–5.[CrossRef][Medline]
- Nino-Murcia M, Jeffrey RB, Beaulieu CF, Li KCP, Rubin GD. Multidetector CT of the pancreas and bile duct system: value of curved planar reformations. AJR Am J Roentgenol 2001;176:689–93.[Free Full Text]
- Hommeyer SC, Freeny PC, Crabo LG. Carcinoma of the head of the pancreas: evaluation of the pancreaticoduodenal veins with dynamic CT-potential for improved accuracy in staging. Radiology 1995;196:233–8.[Abstract/Free Full Text]
- Raptopoulos V, Steer ML, Sheiman RG, Vrachliotis TG, Gougoutas CA, Movson JS. The use of helical CT and CT angiography to predict vascular involvement from pancreatic cancer: correlation with findings at surgery. AJR Am J Roentgenol 1997;168:971–7.[Abstract/Free Full Text]
- Diehl SJ, Lehmann KJ, Sadick M, Lachmann R, Georgi M. Pancreatic cancer: value of dual-phase helical CT in assessing resectability. Radiology 1998;206:373–8.[Abstract/Free Full Text]
- Kaneko K, Honda H, Hayashi T, Fukuya T, Irie H, Masuda K. Helical CT evaluation of arterial invasion in pancreatic tumors: comparison with angiography. Abdom Imaging 1997;22:204–7.[CrossRef][Medline]
- Hough TJ, Raptopoulos V, Siewert B, Matthews JB. Teardrop superior mesenteric vein: CT sign for unresectable carcinoma of the pancreas. AJR Am J Roentgenol 1999;173:1509–12.[Abstract]
- Lepanto L, Arzoumanian Y, Gianfelice D, Perreault P, Dagenais M, Lapointe R, et al. Helical CT with CT angiography in assessing periampullary neoplasms: identification of vascular invasion. Radiology 2002;222:347–52.[Abstract/Free Full Text]
Top
Abstract
Introduction
