You are on page 1of 10

ORIGINAL ARTICLES

En Bloc Vascular Resection for Locally Advanced Pancreatic


Malignancies Infiltrating Major Blood Vessels
Perioperative Outcome and Long-term Survival in 136 Patients
Emre F. Yekebas, MD,* Dean Bogoevski, MD,* Guellue Cataldegirmen, MD,* Christina Kunze, MD,*
Andreas Marx, MD, Yogesh K. Vashist, MD,* Paulus G. Schurr, MD,* Lena Liebl, MD,*
Sabrina Thieltges, MD,* Karim A. Gawad, MD,* Claus Schneider, MD,* and Jakob R. Izbicki, MD*

Background: To assess in-hospital complication rates and survival


duration after en bloc vascular resection (VR) for infiltration of
pancreatic malignancies in major vessels.
Methods: Between 1994 and 2005, 585 patients underwent potentially curative pancreatic resection without adjuvant chemotherapy.
Four hundred forty-nine patients (77%) underwent standard oncologic resection (VR), whereas 136 (23%) received VR (VR). For
calculation of in-hospital morbidity and mortality rates, all 136
patients who underwent VR were considered. In contrast, for survival analysis, only pancreatic adenocarcinoma patients (n 100)
were included.
Results: One hundred twenty-eight VR patients underwent portal
or superior mesenteric vein resection and 13 hepatic artery (HA) or
superior mesenteric artery (SMA) resection. In 5 patients, synchronous VR addressing both the mesenterico-portal axis and either the
HA or SMA was performed. In-hospital morbidity and mortality
rates of VR patients (39.7%/4.0%) nearly equaled that of VR
patients (40.3%/3.7%). From the 100 patients with pancreatic adenocarcinoma, histopathology confirmed true vascular invasion in
77 patients. Twenty-three patients had peritumoral inflammation,
mimicking tumor invasion. Median survival was 15 months (11.2
18.8) in patients with histopathologic proven vascular invasion and
16 months (14.0 17.9) in those without (P 0.86). Two-year
survival probabilities were 36% (without) versus 34% (with vascular
invasion; P 0.9). Among VR patients with histopathologically
evidenced vascular invasion, 19 survived longer than 30 months,
and 6 were still alive 5 years after surgery. Multivariate modeling
identified nodal involvement (N1) and poor grading (G3) as the only
predictors of decreased survival. Evidence of vascular invasion had
no adverse impact on survival.
Conclusion: Postoperative morbidity and mortality rates after en
bloc VR are comparable with standard pancreatectomy proce-

From the *Department of General, Visceral and Thoracic Surgery, and


Institute of Pathology, University Medical Centre Hamburg-Eppendorf,
University of Hamburg, Hamburg, Germany.
Reprints: E. F. Yekebas, MD, Department of Surgery, University Hospital Hamburg-Eppendorf, Martinistrasse 52, 20246 Hamburg, Germany. E-mail: yekebas@uke.uni-hamburg.de.
Copyright 2008 by Lippincott Williams & Wilkins
ISSN: 0003-4932/08/24702-0300
DOI: 10.1097/SLA.0b013e31815aab22

300

dures. Median survival of 15 months in patients with vascular invasion


is superior to that of patients who undergo palliative therapy and nearly
equals that of patients who are not in need for VR.
(Ann Surg 2008;247: 300 309)

lthough early detection of pancreatic malignancies is the


only theoretical chance for cure,1 the majority of pancreatic cancers is diagnosed at an advanced or even incurable
stage. Narrowing or encasement of the celiac axis, superior
mesenteric, or splenic vessels visualized either by state-ofthe-art computed tomography (CT) scanning2 or intraoperatively are considered as signs of locally advanced tumors,
albeit differential diagnosis of true vascular tumor infiltration
and peritumoral inflammation is often difficult.
Until the early 1990s, the presence of portomesenteric
invasion was generally accepted as a contraindication for
curative surgery. However, institutional experiences gained
in several high-volume pancreatic centers, which evidenced
that advanced pancreatic carcinoma infiltrating major
peripancreatic vessels can be safely treated by right-sided,
left-sided, subtotal or total pancreatectomy with en bloc
resection of infiltrated major vascular structures led to increasingly aggressive surgical approaches. This attitude was
supported by the reported prognosis of these patients that
seems to resemble that in patients without resection of major
vascular structures.37
Therefore, the apodictic assumption that involvement
of mesenteric, portal, and splenic veins indicates irresectability is with regard to technical aspects and prognosis at least
highly debatable, if not even no more sustainable. A more
controversial area concerns an infiltration of the hepatic
artery (HA), the superior mesenteric artery (SMA) and of the
celiac trunk. In patients undergoing distal pancreatectomy
with en bloc celiac axis resection for advanced cancer of the
pancreatic body, a Japanese group has most recently reported
a R0 resection rate of 91% (21/23), a mortality rate of 0%,
and an estimated 5-year survival probability of 42%.8 Institutional experiences with en bloc vascular resection (VR) of
Annals of Surgery Volume 247, Number 2, February 2008

Annals of Surgery Volume 247, Number 2, February 2008

even more than one vessel in selected patients, as reported in


the presented series, have not yet been published in the
surgical literature.
Matters of debate are as follows: (1) Is vascular infiltration
always a sign of advanced (and therefore unresectable) cancer or
is it a consequence of tumor location? (2) Are perioperative
morbidity and mortality rates comparable between patients with
and without VR? (3) Does the widespread assumption that
oncologic overall prognosis is worse in patients with vascular
involvement withstand an accurate analysis of patients?
Therefore, we conducted an analysis based on a cohort of
585 curatively operated patients with histopathologically proven
pancreatic, ampullary or distal bile duct malignancies who were
spared from adjuvant chemotherapy. Among these patients, 136
underwent extended surgical procedures with en bloc VR. For
comparison of perioperative morbidity and mortality rates, the
entire cohort of patients undergoing VR, irrespective of whether
histopathology confirmed true vascular invasion was included.
In contrast, the analysis of oncologic long-term outcome was
limited to patients with ductal adenocarcinoma.

PATIENTS AND METHODS


Patient Characteristics, Inclusion Criteria
A total of 585 consecutive patients with pancreatic,
ampullary and distal common bile duct (CBD) cancer who

En Bloc Vascular Resection in Pancreatic Surgery

underwent potentially curative resection at our institution


from April 1994 to July 2005 were analyzed. Patients that
received postoperative adjuvant chemotherapy under controlled study conditions (until 2004) or in the routine clinical
setting (after 2004) were excluded. The ethic committee of
the chamber of physicians in Hamburg approved this prospective study. Informed consent was obtained from all
patients before including them in a prospective database.
The age of patients ranged from 32 to 90 years, with a
median of 61 years. There was a slight predominance of male
patients (Table 1). Patients with macroscopically incomplete
resection (R2-status) were excluded from analysis. Depending on the location of tumors, patients were subjected to
classic pancreatico-duodenectomy (c-PD), pylorus-preserving pancreatico-duodenectomy (pp-PD), distal spleno-pancreatectomy (DSP), subtotal pancreatectomy (st-P), and total
pancreatico-duodenectomy (t-PD). Guide-mark stitches of
the resected vessels after the removal of the specimen enabled
to examine whether or not intraoperatively suspected vascular
tumor invasion could be confirmed by histopathology and, in
turn, to calculate the rate of cases in which peritumoral
inflammatory changes mimicked vascular involvement. Tumor stage and grade were classified according to the sixth
edition of the tumor-node-metastasis classification of the
International Union against Cancer.9

TABLE 1. Characteristics of the Study Population (n 585)

Gender (M/F)
Age
Type of operation
c-PD (Whipple)
pp-PD (Whipple)
DSP
St-P
t-PD
Resection margins
R0
R1
Operation time
Surgical morbidity
Hemorrhage
Pancreatic fistula
Bile leak
PV thrombosis
Arterial thrombosis
Sepsis
Medical complications
No complications
Periop. mortality

Total Patients
(n 585)

Without VR
(n 449)

With VR
(n 136)

349/236
61 (3290)

264/185
61 (3282)

85/51
62 (3390)

0.441
0.949

277 (47%)
174 (30%)
55 (9%)
41 (7%)
38 (6%)

215 (48%)
144 (32%)
45 (10%)
23 (5%)
22 (5%)

502 (85.8%)
83 (14.2%)
360 (220520)

383 (85.3%)
66 (14.7%)
350 (220520)

31 (5.3%)
45 (7.7%)
29 (5.0%)
5 (0.8%)
2 (0.3%)
26 (4.4%)
95 (16.2%)
352 (60.2%)
23 (3.9%)

25 (5.6%)
36 (8.0%)
20 (4.5%)
3 (0.7%)
1 (0.2%)
19 (4.2%)
74 (16.5%)
271 (60.4%)
18 (4.0%)

62 (46%)
30 (22%)
10 (7%)
18 (13%)
16 (12%)
0.963
119 (87.5%)
17 (12.5%)*
360 (250500)
6 (4.4%)
9 (6.6%)
9 (6.6%)
2 (1.5%)
1 (0.7%)
7 (5.1%)
21 (15.4%)
81 (59.6%)
5 (3.7%)

0.23
0.719

0.895

*In only one VR patient, histopathology showed microscopic tumor invasion extending to the resected vessel margins,
whereas in all other patients, R1 status involved the retroperitoneal, eg SMA resection margin.

Arterial thrombosis involved the main hepatic artery in one patient without VR. In a second patient, thrombosis of a
venous graft replacing the SMA occurred.
pp indicates pylorus-preserving; DSP, distal spleno-pancreatectomy; st-P, subtotal pancreatectomy; t-PD, total pancreatico-duodenectomy; PV, portal vein.

2008 Lippincott Williams & Wilkins

301

Yekebas et al

Preoperative evaluation included abdominal ultrasonography (US) and contrast computed tomography (CT).
Helical CT with arterial, pancreatic, and hepatic phases was
routinely done. Endoscopic retrograde cholangiopancreatography, and endoscopic ultrasound were in individual patients
performed. Distant metastases, celiac trunk infiltration, or
thrombosis of the mesenterico-portal axis with evident cavernomatous transformation evidenced either by preoperative
imaging or intraoperatively were considered contraindications
for curative surgery. In contrast, narrowing of the mesentericoportal venous axis without cavernomatous thrombosis of the
portal vein and encasement of the hepatic and mesenteric arteries were accepted an indication for explorative laparotomy.

Surgical Technique
Type of Pancreatic Resection and Resectability
Criteria
After abdominal exploration, patients with distant metastases, peritoneal dissemination, arterial infiltration of the
celiac trunk, mesenteric root encasement, in either case histopathologically proven by frozen sections, underwent palliative procedures. All other patients were considered potentially eligible for resection. c-PD and pp-PD represented the
standard procedures for masses located right to the mesenterico-portal axis, whereas in those left to the SMV, DSP was
the procedure of choice. Pancreatic body malignancies were
usually either treated by st-P (extended left pancreatectomy)
or by TP. Bile duct and pancreatic transsection margins were
routinely checked by frozen section. Standard lymphadenectomy as described elsewhere was performed.10,11 Irrespective
of the type of resection (c-PD, pp-PD, DSP, st-P, TP),
mobilization of the specimen was performed before VR,
hereby resulting in en bloc resection of the specimen including the involved vessel as the last step of the operation.
Technical and general eligibility criteria for en bloc resection
were as follows: (1) presumable achievement of R0-status;
(2) tumor infiltration proximal to the peripheral branching of
the SMV and SMA; (3) tumor respecting the celiac trunk; (4)
no evidence for hypercoagulopathy syndromes.

Technique of VR
When tumors infiltrated the mesenterico-portal axis, the
vessel was freed from its surrounding tissue proximally and
distally to ensure sufficient vascular control. For SMV lesions, the proximal vascular clamp was placed distal to the
venous confluence. In case of tumors invading the lateral
aspect of the lower PV, proximal clamps were usually placed
at an angle to the spleno-portal confluence without sacrificing
the splenic vein. For tumor infiltration involving the mid or
high PV, the splenic vein was consistently divided to achieve
mobility. Whether tangential resection of the lateral SMV/PV
or segmental sleeve resection was performed depended on
tumor location. Tangential resection in case of attachment of
the tumor to the right-sided SMV/PV was usually reconstructed by simple venous suture. Rarely, when more than
one-third of the lateral wall had to be resected, autologous
venous patches (internal jugular, saphenous, inferior mesenteric vein) were performed to avoid venous narrowing. Vas-

302

Annals of Surgery Volume 247, Number 2, February 2008

cular infiltration of more than half of the venous circumference was consistently treated by segmental sleeve resection.
The preferred reconstruction technique after segmental resection was primary end-to-end anastomosis. Tension-free anastomosis was facilitated by mobilization of the mesenteric
root. Occasionally, to achieve additional mobility of the
proximal stump, the hepatic ligaments were transsected. Only
when despite both maneuvers tension-free anastomosis were
technically not feasible, venous continuity was restored by
interposition of autologous veins (see above). The most
recently published technique of using the left renal vein as an
autologous conduit to restore the mesenterico-portal continuity12 that, in contrast to harvesting the internal jugular or
saphenous veins, protects patients from additional cervical or
inguinal surgery was not applied in any of the reported
patients. When the venous confluence had to be resected, the
decision on whether or not the splenic vein stump was
reinserted was based on the presence or absence of adequate
collateralization of the splenic hilum via the short gastric
veins (Fig. 1).
In 5 among 13 patients undergoing arterial sleeve
resection, combined VR procedures addressing both the mesenterico-portal axis and the hepatic or mesenteric artery were
performed (Table 2). Such extended procedures were restricted to patients in good preoperative, clinical condition
(ASA III). When tension-free reconstruction of the arterial
continuity by end-to-end anastomosis was not feasible, an
autologous vein interposition using the internal jugular, saphenous, or inferior mesenteric vein was performed. Perioperative, systemic anticoagulation intending partial thromboplastin time of 60 to 70 s was only performed in patients who
underwent arterial resection. In all other patients, conventional, prophylactic heparinization by low molecular weight
heparin was performed.

In-Hospital Parameters
The following parameters were routinely assessed, online included in a prospective data base, and analyzed in this
study: perioperative morbidity, especially surgical complications, eg, occurrence of postpancreatectomy hemorrhage,
pancreatic, and biliary fistula, thrombosis of the PV, SMV,
SMA, and HA in patients undergoing VR, and of sepsis/intraabdominal abscess formation. Perioperative mortality was
defined as in-hospital mortality and deaths within the first
month after discharge of patients.

Follow-Up
Follow-up (median: 14 months; range 3139 months)
was either performed by interviewing the patients general
practitioners, in our institution on an outpatient basis, or,
when no other information could be provided, from the
regional Cancer Registry. These evaluations included regularly scheduled physical examinations, imaging tests and
studies of tumor markers (carcinoembryonic antigen and CA
19-9). Events considered were death, local recurrence, and
distant metastasis. When no events were recorded, the patients were censored at the last contact with them.
2008 Lippincott Williams & Wilkins

Annals of Surgery Volume 247, Number 2, February 2008

En Bloc Vascular Resection in Pancreatic Surgery

FIGURE 1. Reconstruction after en


bloc resection of the portal vein or
superior mesenteric vein. Sixty-five
patients underwent tangential VR (A).
Whether portal blood flow was restored by simple suture (n 57) or
venous patch depended on the extent of vessel wall resection. Segmental vessel resection was performed in
63 patients with different sites of
anastomosis. In the majority of these
patients, the splenic vein could be
maintained (B,C), whereas in only 12,
it had to be killed to achieve sufficient
mobility (D,E). Five patients in whom
a tension-free anastomosis could not
be performed required venous graft
interposition (F).

Statistical Analysis
Associations between categoric and continuous variables at surgery were assessed by Fisher exact test and
Wilcoxon tests, where appropriate. The Kaplan-Meier
method was used to estimate survival probability 24 and 60
months after surgery. Differences between patient groups
with respect to their survival were assessed by log-rank tests,
considering differences to be statistically significant at a P
value of 0.05. Multivariate modeling including variables
with a P value 0.05 in univariate log-rank test was fit to the
survival data using Cox proportional hazards methods. In this
case, significance statements refer to P values of 2-tailed tests
with a P value 0.05. Those 23 pancreatic adenocarcinoma
patients undergoing VR in whom histopathology only
showed tumor-mimicking involvement of vessels were, regarding their perioperative course, pooled with patients who
had histologically proven true vascular tumor invasion (Table
2008 Lippincott Williams & Wilkins

1), whereas for long-term survival analysis, VR patients


without histopathologically evidenced vascular infiltration
were shifted in the subset of VR patients without vascular
involvement. To avoid potential bias because of different
histopathologic diagnosis (ductal adenocarcinoma, ampullary, ie, duodenal and papillary carcinoma, bile duct carcinoma), comparison of survival between VR and VR
patients only considered patients with true pancreatic ductal adenocarcinoma.

RESULTS
Procedures and Perioperative Patient
Characteristics
Table 1 shows patient characteristics, surgical procedures, and perioperative outcome in the entire study cohort
irrespective of the tumor entity (ductal, ampullary, distal

303

Annals of Surgery Volume 247, Number 2, February 2008

Yekebas et al

TABLE 2. En Bloc Arterial Resection With or Without Mesenterico-Portal Resection


Patient No.
Synchronous VR
(PV/SMVHA/SMA)
1
2
3
4
5
Arterial resection
(HA/SMA)
6
7
8
9
10
11
12
13

Resected
Vessels

Arterial
Reconstruction

Pancreatic
Resection

PV/SMVHistology

HA/SMAHistology

N
Status

Perioperative
Outcome

Survival
(months)

HASMV
HAPV
HAPV
SMASMV
HAPV

E/E
E/E
E/E
Venous graft
E/E

c-PD
t-PD
t-PD
c-PD
t-PD

Positive
Positive
Positive
Positive
Positive

Positive
Positive
Positive
Negative
Positive

pN1
pN0
pN1
pN0
pN1

Survived
Survived
Survived
Died
Survived

18
25
11

14

HA
HA
HA
HA
HA
HA
SMA
SMA

E/E
E/E
E/E
E/E
E/E
E/E
Venous graft
E/E

t-PD
t-PD
st-P
t-PD
t-PD
c-PD
t-PD
st-P

Positive
Negative
Positive
Positive
Positive
Positive
Positive
Positive

pN1
pN0
pN0
pN1
pN1
pN1
pN0
pN1

Survived
Survived
Survived
Survived
Survived
Survived
Survived
Survived

9
Alive (39 mo)
Alive (32 mo)
6
19
15
Alive (29 mo)
10

Actual survival of patients with pancreatic adenocarcinoma who had true vascular invasion in main arterial branches confirmed by histopathology. Five among 13 patients
undergoing resection of the HA or SMA had simultaneous VR of the SMV or PV (Patient No. 1-5). All patients who survived longer than 2 years (no. 2, no. 7, no. 8, no. 12) were
staged as pN0. Regarding the latter, histopathology confirmed in 3 patients tumor invasion in the resected artery.
PV indicates portal vein; SMV, superior mesenteric vein; HA, hepatic artery; SMA, superior mesenteric artery; E/E, end-to-end anastomosis; c-PD, classical pancreaticoduodenectomy; t-PD, total pancreatico-duodenectomy; st-P, subtotal pancreatectomy.

CBD carcinoma) and of whether or not true vascular invasion


was evidenced histopathologically. c-PD was performed in
277 patients, pp-PD in 174 patients, DSP in 55 patients, st-P
in 41 patients, and t-PD in 38 patients (Table 1). Overall,
these procedures were performed in 449 patients without
vascular resection (VR), whereas 136 patients underwent
en bloc VR. Resection margins were tumor-free (R0) in 85%
of VR patients and in 87.5% of VR patients (P 0.963),
resulting, in the entire study cohort, in R0-status in 86%, and
in R1-status in 14% of patients. With respect to operation
times and median volume of intraoperatively transfused
blood units that accounted for a median of 3 units in both
groups, no significant differences between VR and VR
patients were detected. Postoperative hemorrhage and pancreatic fistula represented the most frequent surgical complications, overall accounting for 5.3% and 7.7%, respectively.
Portal vein thrombosis occurring in 3 VR patients and in 2
VR patients (0.7% and 1.5%, respectively) was lethal in 2
patients (1 VR, 1 VR). In 2 patients, arterial thrombosis
complicated the postoperative course: one patient developing
HA thrombosis not related to VR after TP for intraductal
papillary mucinous neoplasia (IPMN) with invasive carcinoma had, as would be expected, long-lasting biliary complications, but survived. In the second patient who underwent
synchronous en bloc Whipple procedure with SMA and SMV
resection, thrombosis of the venous graft replacing the SMA
due to pancreatic fistula with subsequent intraabdominal infection was lethal (Table 2). In the whole cohort of 585 patients, 22
(4%) died perioperatively, ie, during their hospital stay or within
30 days after discharge. Overall, morbidity and mortality rates of
VR and VR patients were comparable.

304

Type of VR and Reconstruction


Among the 136 patients who underwent VR, 128 were
subjected to a resection of the mesenterico-portal axis. In 13
patients a resection of the HA or SMA was performed with an
overlap of 5 patients undergoing synchronous VR of the PV
or SMV. Distribution of portal venous reconstruction techniques, depending on the tumor location, is shown in Figure
1. Tangential resection of either the SMV or the PV was
performed in 65 patients among which in 57, simple venous
suture was sufficient to restore portal blood flow, whereas 8
underwent a venous patch (Fig. 1A). Sixty-three patients
were subjected to segmental mesenterico-portal sleeve resection. In the majority of the latter (n 58), the mesentericoportal axis was reconstructed by end-to-end anastomosis (Fig.
1BE). Only 5 patients required venous graft interposition
(Fig. 1F). Overall, in 10 from 17 patients that underwent a
transsection of the splenic vein, a reinsertion of the venous
stump was performed, hereby creating a neo-confluence
(Fig. 1D, E).
In 13 patients, tumor invasion involved the HA (n 10)
or the SMA (n 3), respectively (Table 2). Among the 5
patients undergoing simultaneous VR of the mesenterico-portal
axis, 3 were subjected to t-PD and 2 to c-PD. In 11 patients,
end-to-end anastomosis was technically feasible, whereas 2 with
VR of the SMA required autologous vein interposition. Further
analysis of this subset showed that all patients surviving longer
than 2 years had been staged as pN0.

Histopathology
In 482 of 585 patients (82%), histopathology confirmed
pancreatic ductal adenocarcinoma, whereas the remainder
2008 Lippincott Williams & Wilkins

Annals of Surgery Volume 247, Number 2, February 2008

En Bloc Vascular Resection in Pancreatic Surgery

FIGURE 2. Overall survival of patients with pancreatic ductal adenocarcinoma grouped according to the VR and histologically
proven infiltration in the blood vessels. No survival benefit was detected in VR patients compared with VR patients irrespective of whether histopathology showed true infiltration or tumor-mimicking, inflammatory pseudo-infiltration (P 0.948
and P 0.279). Of the patients with VR and histologically proven infiltration 33.7% survived more than 30 months compared
with 30.8% for patients without VR.

had ampullary (duodenal, papillary; other, eg, NET) (n 54)


or distal bile duct carcinoma (n 49). To avoid statistical
bias caused by the different biologic characteristics of ampullary and distal CBD carcinoma, as compared with pancreatic carcinoma, only patients with true pancreatic adenocarcinoma were subjected to survival analysis (Fig. 2).
Among the 482 patients with ductal adenocarcinoma,
100 (21%) underwent VR. The analysis of the VR subset
revealed histopathologic evidence of vascular invasion in 77
patients. In contrast, 23 patients had only tumor-mimicking
lesions without histopathologic proven vascular involvement.
For survival analysis, these 23 VR patients without vascular invasion in histopathology were pooled with VR patients. This resulted in 405 patients without and 77 patients
with vascular invasion representing the data base for further
survival analysis (Table 3).
Comparison of tumor characteristics showed slight differences of borderline significance regarding the primary
tumor stage (T-stage). The higher rate of negative resection
margins in patients with histopathologically evidenced vascular invasion (90%) as compared with those without (82%)
did not reach statistical significance (P 0.1). In the entire
subset of VR patients, only one (distal CBD carcinoma)
was identified in whom microscopic involvement (R1) concerned the resection margin of the resected vessel. In all other
VR patients with histopathologically evidenced R1 status,
tumors extended to the retroperitoneal resection margin,
whereas en bloc resected vessels were found to be tumor-free.
Only minimal differences regarding nodal status and tumor
2008 Lippincott Williams & Wilkins

grading between patients with and without histopathologic


vascular invasion were assessed (Table 3).

Survival
Median follow-up times were 13 months (range, 4 123)
in VR patients with histopathologically evidenced vascular
invasion, 15 months (range, 4 112) in VR patients without
vascular invasion, and 14 months (range, 4 139) in VR
patients, respectively. Overall, median survival of patients with
histopathologically confirmed vascular invasion was 15 months
(95% confidence interval CI; 11.218.8 months), whereas that
of patients without vascular involvement was 16 months (95%
CI; 14.0 17.9 months, P 0.856). In those 23 VR patients
without histologically proven infiltration, median survival was
23 months (95% CI, 14 32).
Overall 2- and 5-year survival probabilities accounted
for 35.9% and 17% for patients without, and for 33.7% and
14.6% for patients with vascular invasion (P 0.9). In the
subanalysis of the VR group, 2-, and 5-year survival probabilities of patients in whom histopathology did not confirm
true vascular invasion were 41.1% and 24.2%, respectively.
In the entire cohort of 77 patients with vascular invasion,
19 survived more than 30 months. Six patients were still alive 5
years after surgery and one patient survived more than 10 years.
In the arterial en bloc resection group (n 13) involving either
the HA or SMA, 8 patients survived more than 1 year and 4 even
more than 2 years. Five from these patients were subjected to
simultaneous VR of the SMV or PV (Table 2).

305

Annals of Surgery Volume 247, Number 2, February 2008

Yekebas et al

TABLE 3. Histopathology in Pancreatic Ductal Carcinoma (n 482)

Standard procedures

En bloc Vascular Resection

Total

With Vascular Infiltration


in Histopathology En bloc
Vascular Resection

382

23

405

77

53 (14)
122 (32)
199 (52)
8 (2)

2 (9)
7 (30)
12 (52)
2 (9)

55 (14)
129 (32)
211 (52)
10 (2)

5 (6)
23 (30)
44 (57)
5 (6)

60 (21)
322 (79)

1 (13)
22 (87)

61 (15)
344 (85)

8 (10)
69 (90)

53 (14)
210 (55)
119 (31)

3 (13)
12 (52)
8 (35)

56 (14)
222 (55)
127 (31)

10 (13)
37 (48)
30 (39)

313 (82)
69 (18)

21 (91)
2 (9)

334 (82)
71 (18)

69 (90)
8 (10)

Without Vascular Infiltration in Histopathology

No.
T stage
T1
T2
T3
T4
N stage
N0
N1
Grading
G1
G2
G3
R status
R0
R1

P (1)

P (2)

0.095

0.057

0.283

0.231

0.421

0.404

0.10

0.10

Values in the parentheses are given in percentage.


P (1): Patients without histopathologically evidenced vascular invasion (standard pancreatic resections en bloc VR with negative vascular histopathology, n 405) versus
en bloc VR with histopathologically confirmed vascular involvement (n 77).
P (2): Standard pancreatic resections (n 382) versus en bloc VR with histopathologically confirmed vascular involvement (n 77).

Further stratification of the entire cohort of 482 patients


with ductal adenocarcinoma according to nodal involvement
showed, irrespective from vascular invasion status, significantly better median survival of 24 months (95% CI, 18.4
29.6) in pN0 patients compared with a median survival of 11
months (95% CI, 9.6 12.4) in pN1 patients (P 0.0001).
Additional stratification of pN0 patients according to vascular
invasion revealed that the slightly better median overall
survival of 24 months (95% CI, 16.8 31.2) in patients without vascular involvement did not reach statistical significance
(P 0.12) when compared with those with histopathologically proven vascular infiltration (median survival 13 months;
95% CI, 5.220.8).
The overall survival did neither differ between different
VR techniques (primary suture vs. end-to-end vs. patch vs.
graft interposition, data not shown, P 0.45) nor between
mesenterico-portal versus arterial resection (data not shown,
P 0.41).

Multivariate Analysis
Multivariate modeling selecting 7 variables using a
stepwise regression model identified only lymph node involvement (pN vs. pN1) and histologic grading (G1/2 vs. G3)
to be independent predictive factors for survival. Histologically proven vascular infiltration, as would be expected from
univariate survival analysis, was not found to have an adverse
independent influence on long-term outcome of patients.
Also, the resection margin status did not independently affect
survival (Table 4).

DISCUSSION
The considerable improvement of postoperative morbidity and mortality rates after major pancreatic resection for

306

TABLE 4. Multivariate Analysis of Variables Potentially


Predictive of Survival After Major Pancreatic Resection
Sex
Age
Vascular invasion
pT 1 & 2
pN0
R0
G1/2

Relative Risk

95.0% CI

0.730
1.001
1.063
0.768
2.320
0.886
2.464

0.4301.237
0.9911.012
0.8601.315
0.4471.319
1.7943.000
0.6081.292
1.9393.132

0.242
0.798
0.569
0.338
0.0001
0.531
0.0001

95% CI indicates 95% confidence interval.

pancreatic carcinoma in the past 2 decades has decreased the


threshold in operating upon patients with surgically challenging, locally advanced tumors. The impetus of extended vascular en bloc resection in case of preoperatively assumed or
intraoperatively assessed tumor invasion in adjacent vessels,
as long as distant metastases are absent, is to achieve a
potentially curative resection. This concerns in particular the
SMV and PV, whereas the role of arterial en bloc resection of
the HA and the SMA or even of the celiac trunk itself is still
highly controversial.
The presented series is based on an analysis of patients
who were spared from adjuvant chemotherapy. Until 2004,
we restricted the use of adjuvant chemotherapy to patients
who were enrolled in controlled studies, eg, the ESPAC-II
and ESPAC-III trials, whereas most of curatively resected
patients did not receive additional therapy. The rationale for
our reluctance towards the routine use of adjuvant chemotherapy under clinical settings was that adequately powered,
randomized studies evidencing the beneficial impact of adju 2008 Lippincott Williams & Wilkins

Annals of Surgery Volume 247, Number 2, February 2008

vant chemotherapy were not only scarce but also inconsistent.


A Japanese randomized controlled trial reported even a tendency
to worse 5-year survival probability in patients who received
adjuvant chemotherapy compared with control patients that only
underwent curative surgery (11.5% vs. 18.0%, not significant).13
After 2004, when the ESPAC-I14 and CONKO-00115 trials
evidenced statistically robust survival benefits of adjuvant
chemotherapy, our institutional attitude substantially changed.
Although one may argue that these studies, especially the
ESPAC-I trial, were afflicted with some weaknesses regarding the study design, the use of adjuvant chemotherapy has
therefore become our institutional standard in the clinical
setting after primary curative resection of pancreatic cancer.
Opponents of extended en bloc VR raise 2 major
counter-arguments against its use. The first argument arguing
that morbidity is substantially elevated,7,16 19 has been rebutted by several surgical series that evidenced comparable
in-hospital morbidity and mortality rates after VR and oncologic standard procedures.37,10,17,20 Nonetheless, the belief
in the usefulness of VR is still controversial. This is reflected
by considerable differences between experienced US centers
with respect to the rates of VR that range from 3%20 to 38%.4
In the present series, even mean operative time and
intraoperative blood loss, generally reported to be elevated in
case of en bloc resection,4,20,21,22 did not substantially differ
in patients with VR compared with those without. Also,
vascular reconstruction was not associated with increased
prevalence of specific vascular complications, such as hemorrhage and thrombosis. Overall, the analysis of in-hospital
morbidity and mortality rates of VR patients based on a
total of 136 patients with pancreatic adenocarcinoma (n
100) and with malignancies of nonpancreatic origin (ampullary/distal CBD cancer, other; n 36) were nearly identical
compared with VR patients.
The second counter-argument directed against VR is
related to the putative limited survival benefit once the tumor
invades major vessels. This attitude ignores several aspects.
First, definitive assessment of tumor adherence is hardly
possible by preoperative imaging. Even intraoperatively, it is
frequently misjudged to what degree vascular involvement is
caused by peritumoral inflammatory changes or by true
invasion. Therefore, in the individual patient, decision-making on whether or not to perform VR is often based on
institutional or even individual experiences and discretion
rather than on histopathologic evidence. In the present study,
a considerable rate of almost one-fourth (23/100) of patients
who underwent VR for pancreatic cancer was found to have
tumor-mimicking lesions without proven vascular involvement in histopathology. Prima facie, this rate of intraoperative misjudgement seems to be high. On close inspection,
however, it is even rather lower than that reported in the most
previous series according to which the rate of histopathologic
confirmation of intraoperatively suspected vascular infiltration ranges from 26% to 85%.4,6,7,16,20,2330
Furthermore, the notion that histopathologically evidenced vascular tumor invasion represents per se an adverse
factor, as argued by some authors,16 has been challenged or
even disproved by several studies that showed comparable
2008 Lippincott Williams & Wilkins

En Bloc Vascular Resection in Pancreatic Surgery

long-term survival in these patients compared with those


without vascular invasion.6,7 The most conclusive explanation for this is that the crucial factor, which determines
long-term outcome of patients with pancreatic cancer is the
presence or absence of early tumor cell dissemination to
distant organs undetectable by routine imaging techniques at
the time of potentially curative surgery. In support of this, one
may argue that all attempts aiming at achieving better overall
outcome by increasing the extent of peripancreatic lymphadenectomy eventually failed. In this context, it was the impetus
of several randomized trials in the past decade to decrease the
risk of local recurrence by extended lymphadenectomy procedures supposed to be beneficial as a result of its undoubtedly superior lymphatic clearance.20,3134 Better local control, so was the assumption, would result in improved longterm survival. However, although these trials reported a
significantly increased lymph node yield by extended lymphadenectomy, overall survival of patients who underwent extended lymphadenectomy was nearly identical compared with
patients subjected to standard lymphadenectomy.
In the light of these considerations, it was a key finding
of the presented series that the median survival of 15 months
of patients with true pancreatic adenocarcinoma who had
histopathologically proven vascular invasion (n 77) nearly
equaled that of 382 patients without VR. Interestingly, although not reaching statistical significance, VR patients in
whom histopathology showed only tumor-mimicking vascular involvement (n 23) had the longest median survival of
23 months. In this context, it is important to stress that survival
analysis only addressed patients with pancreatic carcinoma.
Patients with malignancies of nonpancreatic origin and those
who underwent adjuvant chemotherapy were excluded.
It has been recently suggested that the likelihood of R1
resection margin status is increased in large tumors that
require VR35; a finding, which is inconsistent with the data
reported herein. Several studies report a poor survival associated with VR due to a high incidence of positive resection
margins.10,16,21,29,36 Although not significant, we observed an
even lower R1 rate of 10% (true infiltration) and 9%
(infiltration-mimicking vascular involvement) after VR than
that after standard resection (18%). One may hypothesize that
factors such as large tumor size, R1 resection margin status
and the need for VR, frequently lacking to have significant,
independent importance in multivariate modeling, are associated with a higher likelihood of N1 status that is currently
regarded the most important independent variable for prediction of overall prognosis. In the present study, the only factor,
which had, apart from nodal metastasis, an independent
influence on survival was tumor grading (G1/2 vs. G3). All
other variables, including tumor size, resection margin status
and, in particular, histopathologic vascular invasion, had no
statistical significance when introduced in multivariate regression analysis.
That R1 status had even in univariate analysis no
independent adverse influence on survival directs attention to
the issue of histopathologic staging accuracy. It is an issue of
increasing concern that the assessment of a R1 status is
closely dependant on the technique of histopathologic exam-

307

Annals of Surgery Volume 247, Number 2, February 2008

Yekebas et al

ination. The R1 rates of approximately 10% to 20% in the


presented series, depending on whether or not VR was performed, were similar to those reported in the majority of
studies published in the surgical literature. In contrast, some
recent studies report considerably higher R1 rates of 80% and
over by implementing more sophisticated procedures of axial
slicing techniques of Whipple specimens.37 This raises the
question whether disparities in margin status reporting derive
from different histopathologic examination techniques rather
than from true differences in tumor staging, hereby resulting
in understaging of manyif not even mostpatients with
pancreatic malignancies due to the lack of standardized
guidelines for the histopathologic and reporting of specimens.
In summary, to argue that locally advanced tumors with
vascular invasion may have to some extent a worse prognosis
than early tumor stages is, in the individual patient, not
helpful in decision-making. First, the final histopathologic
examination will confirm surgically suspected true vascular
invasion in the majority (77% in this series; partly far below
this rate in other studies), but not all patients with a considerable rate of tumor-mimicking lesions. Second, even in those
patients in whom clinical or intraoperative suspicion is confirmed by histopathology, overall outcome seems to be dependent from other variables, such as nodal involvement and
tumor grading rather than from vascular tumor invasion that,
in the present study, had no an independent adverse impact on
survival. This may even be the case in selected patients in
whom VR of the PV/SMV is combined with a resection of
either the HA or the SMA. In the presented series, such
combined procedures were performed in 5 patients. Apart
from one patient who died perioperatively, actual survival in
these patients rangedin the absence of any adjuvant treatmentfrom 11 to 25 months, which is far longer than that to
be expected when therapy in these patients would have been
restricted to palliative measures.
In conclusion, our data suggest that major pancreatic
surgery can be safely combined with en bloc VR in case of
suspected or evidenced vascular invasion. With adequate
institutional experience, in-hospital morbidity and mortality
rates are comparable with that of standard procedures. The
need for vascular reconstruction does not adversely impact
long-term survival. When potentially curative resection is
achieved, 2-, and 5-year survival probabilities of 35% and
15% of patients with histopathologically evidenced vascular
invasion nearly equal that of patients without vascular involvement. Median survival after en bloc VR is far longer
than the reported survival of patients in whom therapy is
restricted to palliative surgical procedures and chemotherapy. Further trials are needed addressing whether neoadjuvant treatment options, especially radiotherapy, may downsize locally advanced pancreatic tumors suspicious of
vascular invasion.

4.

5.

6.

7.

8.

9.
10.

11.

12.

13.

14.

15.

16.

17.

18.
19.
20.

21.

22.

23.

REFERENCES

24.

1. Ariyama J, Suyama M, Ogawa K, et al. The detection and prognosis of


small pancreatic carcinoma. Int J Pancreatol. 1990;7:37 47.
2. Bluemke DA, Fishman EK. CT and MR evaluation of pancreatic cancer.
Surg Oncol Clin N Am. 1998;7:103124.
3. Yoshimi F, Asato Y, Tanaka R, et al. Reconstruction of the portal vein

25.

308

and the splenic vein in pancreaticoduodenectomy for pancreatic cancer.


Hepatogastroenterology. 2003;50:856 860.
Tseng JF, Raut CP, Lee JE, et al. Pancreaticoduodenectomy with
vascular resection: margin status and survival duration. J Gastrointest
Surg. 2004;8:935949; discussion 949 950.
Koniaris LG, Staveley-OCarroll KF, Zeh HJ, et al. Pancreaticoduodenectomy
in the presence of superior mesenteric venous obstruction. J Gastrointest
Surg. 2005;9:915921.
Harrison LE, Klimstra DS, Brennan MF. Isolated portal vein involvement in pancreatic adenocarcinoma. A contraindication for resection?
Ann Surg. 1996;224:342347; discussion 347349.
Leach SD, Lee JE, Charnsangavej C, et al. Survival following pancreaticoduodenectomy with resection of the superior mesenteric-portal vein
confluence for adenocarcinoma of the pancreatic head. Br J Surg.
1998;85:611 617.
Hirano S, Kondo S, Hara T, et al. Distal pancreatectomy with en bloc
celiac axis resection (DP-CAR) for locally advanced pancreatic body
cancer: long-term results. Ann Surg. 2007;246:46 51.
Sobin LH, Wittekind Ch. TNM Classification of Malignant Tumours. 6th
ed. New York: Wiley; 2002.
van Geenen RC, ten Kate FJ, de Wit LT, et al. Segmental resection and
wedge excision of the portal or superior mesenteric vein during pancreatoduodenectomy. Surgery. 2001;129:158 163.
Yeo CJ, Cameron JL, Lillemoe KD, et al. Pancreaticoduodenectomy
with or without distal gastrectomy and extended retroperitoneal lymphadenectomy for periampullary adenocarcinoma, Part 2: Randomized
controlled trial evaluating survival, morbidity, and mortality. Ann Surg.
2002;236:355366; discussion 366 368.
Smoot RL, Christein JD, Farnell MB. An innovative option for venous
reconstruction after pancreaticoduodenectomy: the left renal vein.
J Gastrointest Surg. 2007;11:425 431.
Takada T, Amano H, Yasuda H, et al. Is postoperative adjuvant chemotherapy useful for gallbladder carcinoma? A phase III multicenter
prospective randomized controlled trial in patients with resected pancreaticobiliary carcinoma. Cancer. 2002;95:16851695.
Neoptolemos JP, Stocken DD, Friess H, et al. A randomized trial of
chemoradiotherapy and chemotherapy after resection of pancreatic cancer. N Engl J Med. 2004;350:1200 1210.
Oettle H, Post S, Neuhaus P, et al. Adjuvant chemotherapy with
gemcitabine vs. observation in patients undergoing curative-intent resection of pancreatic cancer: a randomized controlled trial. JAMA.
2007;297:267277.
Allema JH, Reinders ME, van Gulik TM, et al. Portal vein resection in
patients undergoing pancreatoduodenectomy for carcinoma of the pancreatic head. Br J Surg. 1994;81:16421646.
Fuhrman GM, Leach SD, Staley CA, et al. Rationale for en bloc vein
resection in the treatment of pancreatic adenocarcinoma adherent to the
superior mesenteric-portal vein confluence. Pancreatic Tumor Study
Group. Ann Surg. 1996;223:154 162.
Harrison LE, Brennan MF. Portal vein resection for pancreatic adenocarcinoma. Surg Oncol Clin N Am. 1998;7:165181.
Sindelar WF. Clinical experience with regional pancreatectomy for
adenocarcinoma of the pancreas. Arch Surg. 1989;124:127132.
Bachellier P, Nakano H, Oussoultzoglou PD, et al. Is pancreaticoduodenectomy with mesentericoportal venous resection safe and worthwhile? Am J Surg. 2001;182:120 129.
Capussotti L, Massucco P, Ribero D, et al. Extended lymphadenectomy
and vein resection for pancreatic head cancer: outcomes and implications for therapy. Arch Surg. 2003;138:1316 1322.
Howard TJ, Villanustre N, Moore SA, et al. Efficacy of venous reconstruction in patients with adenocarcinoma of the pancreatic head.
J Gastrointest Surg. 2003;7:1089 1095.
Baulieux J, Adham M, Oussoultzoglou E, et al. Is pancreatectomy with
resection of the retro-pancreatic vessels for cancer justified?. Chirurgie.
1998;123:438 444.
Tashiro S, Uchino R, Hiraoka T, et al. Surgical indication and significance of portal vein resection in biliary and pancreatic cancer. Surgery.
1991;109:481 487.
Ishikawa O, Ohigashi H, Imaoka S, et al. Preoperative indications for
extended pancreatectomy for locally advanced pancreas cancer involving the portal vein. Ann Surg. 1992;215:231236.

2008 Lippincott Williams & Wilkins

Annals of Surgery Volume 247, Number 2, February 2008

26. Launois B, Franci J, Bardaxoglou E, et al. Total pancreatectomy for


ductal adenocarcinoma of the pancreas with special reference to resection of the portal vein and multicentric cancer. World J Surg. 1993;17:
122126; discussion 126 127.
27. Takahashi S, Ogata Y, Tsuzuki T. Combined resection of the pancreas
and portal vein for pancreatic cancer. Br J Surg. 1994;81:1190 1193.
28. Nakao A, Harada A, Nonami T, et al. Clinical significance of portal
invasion by pancreatic head carcinoma. Surgery. 1995;117:50 55.
29. Roder JD, Stein HJ, Siewert JR. Carcinoma of the periampullary region:
who benefits from portal vein resection? Am J Surg. 1996;171:170 174;
discussion 174 175.
30. Poon RT, Fan ST, Lo CM, et al. Pancreaticoduodenectomy with en bloc
portal vein resection for pancreatic carcinoma with suspected portal vein
involvement. World J Surg. 2004;28:602 608.
31. Pedrazzoli S, DiCarlo V, Dionigi R, et al. Standard versus extended
lymphadenectomy associated with pancreatoduodenectomy in the surgical treatment of adenocarcinoma of the head of the pancreas: a
multicenter, prospective, randomized study. Lymphadenectomy Study
Group. Ann Surg. 1998;228:508 517.

2008 Lippincott Williams & Wilkins

En Bloc Vascular Resection in Pancreatic Surgery

32. Nguyen TC, Sohn TA, Cameron JL, et al. Standard vs. radical pancreaticoduodenectomy for periampullary adenocarcinoma: a prospective,
randomized trial evaluating quality of life in pancreaticoduodenectomy
survivors. J Gastrointest Surg. 2003;7:19; discussion 9 11.
33. Riall TS, Cameron JL, Lillemoe KD, et al. Pancreaticoduodenectomy with
or without distal gastrectomy and extended retroperitoneal lymphadenectomy for periampullary adenocarcinoma, Part 3: Update on 5-year survival.
J Gastrointest Surg. 2005;9:11911204; discussion 1204 1206.
34. Farnell MB, Pearson RK, Sarr MG, et al. A prospective randomized trial
comparing standard pancreatoduodenectomy with pancreatoduodenectomy with extended lymphadenectomy in resectable pancreatic head
adenocarcinoma. Surgery. 2005;138:618 628; discussion 628 630.
35. Raut CP, Tseng J, Sun C, et al. Impact of resection status on pattern of
failure and survival after pancreatoduodenectomy for pancreatic adenocarcinoma. Ann Surg. 2007;246:52 60.
36. Launois B, Stasik C, Bardaxoglou E, et al. Who benefits from portal vein
resection during pancreaticoduodenectomy for pancreatic cancer? World
J Surg. 1999;23:926 929.
37. Verbeke CS, Leitch D, Menon KV, et al. Redefining the R1 resection in
pancreatic cancer. Br J Surg. 2006;93:12321237.

309