Radical Pancreatectomy for Ductal Cell Carcinoma of the Head of the Pancreas

TADAO MANABE, MD, GAKUJI OHSHIO, MD, NOBUO BABA, MD, TADASHI MIYASHITA, MD, NOBORU ASANO, MD, KOlCHlRO TAMURA, MD, KENlCHlRO YAMAKI, MD,

ATSUSHI NONAKA, MD, AND TAKAYOSHI TOBE, MD

Seventy-four patients were treated with a radical or a nonradical pancreatectomy for ductal cell carcinoma of the head of the pancreas. Their survival rates and the selection of the operative procedure were evaluated. In 32 patients, a radical pancreatectomy was attempted where there was sufficient clearance of regional or juxta-regional lymph nodes beyond the group of suspected metastatic nodes, as well as a resection of a greater margin of soft tissue around the pancreas. These patients’ cumulative 5-year survival rate was 33.4%. In 14 Stage I or Stage 11 patients, the cumulative 5-year survival rate was 46.4%. In 18 Stage 111 or Stage IV patients, the cumulative 5-year survival rate was 20.7%. For 42 patients treated with a nonradical pancreatectomy with the dissection of lymph nodes adjacent to the pancreas or of regional lymph nodes but with insufficient clearance of the soft tissue around the pancreas, the cumulative 2-year and 3-year survival rates were 5.4% and 0%, respectively. In seven patients with Stage I1 carcinoma, the survival rate was 16.7% after 2 years and 0% after three years. In 35 Stage I11 or Stage IV patients, the survival rate was 3.2% after 2 years and 0% after 3 years. Thus, the survival rates were significantly higher in patients treated with radical operation than in patients who had nonradical operation. These results indicate that a radical pancreatectomy with sufficient lymph node clearance with the surrounding connective tissue around the pancreas is indispensable to cure patients with ductal cell carcinoma of the pancreas.

Cancer 64:1132-1137. 1989.

ESPITE the development of new diagnostic aids such D as ultrasonography, computerized tomography, en- doscopic retrograde pancreatocholangiography, and an- giography, the prognosis for patients with pancreatic car- cinoma, particularly pancreatic ductal cell carcinoma, is poor because of the tumor’s low resectability and a limited postoperative survival time when compared with other gastrointestinal malignant neoplasms. In most cases, tu- mors extend to the outer margin of the pancreas and in- filtrate the pancreatic capsule and adjacent vessels. ’,* Even if these lesions appear to be resected in the course of pan- createctomy, they are often overlooked since invisible mi- croscopic lesions may be left behind.

Therefore, the primary lesion should be removed with as much surrounding, apparently noncancerous tissues as is feasible, including an en bloc removal of the regional

From the First Department of Surgery, Faculty of Medicine, Kyoto University, Kyoto, Japan.

Supported in part by a grant of Scientific Research (B-62480282) from the Ministry of Education, Science and Culture of Japan.

Address for reprints: Tadao Manabe, MD, First Department of Surgery, Faculty of Medicine, Kyoto University, 54 Shogoin-Kawaracho, Sakyoku, Kyoto 606, Japan.

Accepted for publication March 7, 1989.

lymphatic drainage basin.3 If the involvement of the portal system and the hepatic and superior mesenteric arteries is suspected, these structures should be resected along with adjacent soft tissue^.^

This report analyzes the results of the pancreatectomy for carcinoma of the head of the pancreas from the point of view of radicality and shows that a radical pancreatec- tomy with extensive lymph node resection and the clear- ance of a greater margin of soft tissue around the pancreas is a potential cure for carcinoma of the head of the pan- creas.

Patients and Methods

From 1966 to 1987, 340 patients with carcinoma of the pancreas were treated at Kyoto University Hospital. In 224 of these patients the tumor was in the head of the pancreas, in 106 it was in the body and tail of the pancreas, and in 10 it was in the entire pancreas. A resection was performed in 83 of 224 patients (37%) with carcinoma of the head of the pancreas, in 41 of 106 patients (39%) with carcinoma of the body and tail of the pancreas, and in one of ten patients (10%) with carcinoma of the entire pancreas. In the remaining 2 15 patients, bypass operations

1132

No. 5 RADICAL PANCREATECTOMY FOR PANCREATIC CA + Manabe et a/. 1133

or exploratory laparotomies were performed. This study evaluates the outcomes of 74 patients (52 men and 22 women) 3 1 to 85 years of age (mean age, 59.8 years) with ductal cell carcinoma of the head of the pancreas treated by pancreatectomy. Patients with cystadenocarcinoma, islet cell carcinoma, acinar cell carcinoma, or ductal cell carcinoma with distant metastasis were excluded from this study.

The carcinomas were classified into four stages as pro- posed by the Japanese Pancreatic Society.’ Stage I was a tumor size of 2 cm or less in diameter (Tl), no regional lymph node metastasis (NO), no capsular invasion (SO), no retroperitoneal invasion (RpO), or no involvement of the portal system (PVO). Stage I1 was a tumor size of 2.1 to 4.0 cm in diameter (T2), lymph node metastasis in the primary group adjacent to the tumor (Nl), suspected cap- sular invasion (S l), suspected retroperitoneal invasion (Rpl), or suspected involvement of the portal system (PV1). Stage I11 was a tumor size of 4.1 to 6.0 cm in diameter (T3), lymph node metastasis in the secondary group regarded as regional lymph nodes between the pri- mary and tertiary group (N2), marked capsular invasion (S2), marked retroperitoneal invasion (Rp2), or marked involvement of the portal system (PV2). Stage IV was a tumor size of more than 6.1 cm in diameter (T4), lymph node metastasis in the tertiary group regarded as juxta- regional lymph nodes (N3), direct invasion of the adjacent viscera (S3), extensive retroperitoneal invasion (Rp3), marked involvement ofthe portal system (PV3), or distant metastasis.

The 74 patients were divided into a radical pancre- atectomy group (32 patients) and a nonradical pancre- atectomy group (42 patients), according to the extent of the clearance of the lymph node and resection of con- nective tissue around the pancreas. In radical pancre- atectomy, patients had either a subtotal pancreatectomy that removed more than 60% of the proximal portion of the pancreas (2 1 patients) after obtaining microscopic findings of negative cancer cells in the cut margin of the pancreas during the operation, or a total pancreatectomy (1 1 patients), removal of the bile duct below the bifur- cation with cholecystectomy, a subtotal gastrectomy, and a total duodenectomy. Clearance of the regional or the juxta-regional lymph nodes beyond the group of suspected metastatic lymph nodes and the resection of the connec- tive tissue was accomplished by skeletonization of the portal system, the hepatic and superior mesenteric arteries, and the celiac axis. When involvement of the portal sys- tem, the superior mesenteric vein, or the common hepatic artery was suspected, a combined resection of the pan- creatic segment of the arterial and/or portal system was performed together with the pancreatectomy. The portal and superior mesenteric veins or the common hepatic artery were reconstructed by an end-to-end anastomosis

without a graft. In patients treated with a nonradical pan- createctomy, a pancreaticoduodenectomy was performed in 29 patients and a total pancreatectomy in 13 patients. Resection of the bile duct with a cholecystectomy, a sub- total gastrectomy, and a total duodenectomy was per- formed. If the portal system was involved, a resection of the portal vein also was combined. Lymph nodes adjacent to the head of the pancreas or regional lymph nodes were resected with a pancreatectomy, however, there was in- sufficient clearance of the soft tissue in the postpancreatic space including the perivascular region.

Four Stage I and ten Stage I1 patients were treated with radical operations (a pancreaticoduodenectomy in eight, including two with a combined resection of the portal system, and a total pancreatectomy in six, including two with a combined resection of the portal vein and one with combined resection of the portal vein and the common hepatic artery). Seven Stage I1 patients had nonradical operations (a pancreaticoduodenectomy in five and a total pancreatectomy in two). Sixteen Stage 111 and two Stage IV patients had radical operations (a pancreaticoduode- nectomy in 14, including 11 with a combined resection of the portal system, and a total pancreatectomy in four, including three with a combined resection of the portal system). Thirty-five Stage 111 and Stage IV patients had a nonradical pancreatectomy. A pancreaticoduodenectomy was performed in 24 patients, including two with a com- bined resection of the portal vein, and a total pancre- atectomy was performed in 11 patients, including seven with a combined resection of the portal system.

Results

The size of the tumor, the presence of lymph node me- tastases, capsular and retroperitoneal invasions, and portal system involvement were recorded in 74 patients who were treated by the resection of the pancreas (Table 1). No correlation was found between tumor size, capsular invasion, or portal system involvement and radicality. Capsular invasion of the tumor was observed in 36% of the radical and 43% of the nonradical cases of Stage I and Stage I1 carcinoma and in 72% of the radical and 83% of the nonradical cases of Stage I11 and Stage IV carcinoma. Portal system involvement was found in 2 1% of the radical cases of Stage I and Stage I1 carcinoma and in 83% of the radical and 80% of the nonradical cases of Stage 111 and Stage IV carcinoma. Lymph node metastasis or retroper- itoneal invasion showed a negative correlation with rad- icality. Lymph node metastasis was found in 21% of the radical and 86% of the nonradical cases of Stage I and Stage I1 carcinoma and in 50% of the radical and 77% of the nonradical cases of Stage 111 and Stage IV carcinoma. Retroperitoneal invasion was demonstrated in 14% of the radical and 43% of the nonradical cases of Stage I and

1134 CANCER September 1 1989 Vol. 64

TABLE 1 . Comuarison of Clinical Characteristics of Patients Havine. Radical or Nonradical Pancreatectomies

Tumor size Lymph node metastasis Capsular invasion Retroperitoneal invasion Portal system involvement

Operation TI T2 T3 T4 NO N1 N2 N3 SO SI S2 S3 RpO Rpl Rp2 Rp3 PVO PVI PV2 PV3

Stage 1 and 11

Values in parentheses are percentages.

Stage I1 carcinoma and in 55% of the radical and 57% of the nonradical cases (including extensive invasion in 20% of the cases) of Stage I11 and Stage IV carcinoma. Lymph node metastasis and extensive retroperitoneal invasion were decisive factors affecting the radicality.

The 30-day postoperative mortality rate was 6.2% (two of 32) in the radical operation group, 9.5% (four of 42) in the nonradical operation group, and 8.3% (six of 74) in total. In patients with Stage I and Stage I1 carcinoma, the mortality rate was zero. In patients with Stage 111 and Stage IV carcinoma, the mortality rate was 1 1.3% (six of 53). Among the six patients who died soon after surgery, four with pancreaticoduodenectomy died of intraperito- neal bleeding (two patients), leakage from the pancreati- cojejunal anastomosis (one patient), or cardiac failure (one patient). Two patients who underwent a pancreaticodu- odenectomy with a combined resection of the portal vein died of bleeding or hepatic failure. The cumulative post- operative survival rates for the entire 74 patients were 43.3% after 1 year, 19.3% after 2 years, 15.4% after 3 years, 13.5% after 4 years, and 13.5% after 5 years.

Thirty of 32 patients who had a radical pancreatectomy survived the operation. One among the Stage I patients treated with a pancreaticoduodenectomy and three among the Stage I1 patients similarly treated, one with a total pancreatectomy, and two with a total pancreatectomy with the resection of the portal vein died of recurrence within 2 years (5.6 to 19 months). However, one Stage I patient treated with a pancreaticoduodenectomy lived for 54 months and died of a hepatic abscess of unknown causes, and two Stage I patients (one treated with a pan-

creaticoduodenectomy and the other with a pancreati- coduodenectomy with the resection of the portal vein) have been alive 70 and 40 months, respectively, after sur- gery. One Stage I1 patient treated with a total pancreatec- tomy and the removal of the hepatic vein lived for 24.9 months and died of a stroke. Two Stage I1 patients (one treated with a pancreaticoduodenectomy and the other with a total pancreatectomy) lived for 155.5 and 72 months, respectively, and one patient treated with a pan- creaticoduodenectomy and the resection of the portal vein has been alive for 5.8 months after surgery. One Stage 111 patient treated with a total pancreatectomy and the re- section of the portal vein died of hepatic failure 1.5 months after surgery. One Stage 111 patient treated with a pancre- aticoduodenectomy, seven treated with a pancreatico- duodenectomy and the resection of the portal vein, and two treated with a total pancreatectomy and the resection of the portal vein died of recurrences within 3.0 to 17.7 months. One Stage 111 patient treated with a pancreati- coduodenectomy died of a recurrence after 36 months, one Stage 111 patient treated with a pancreaticoduode- nectomy died as a result of a traffic accident 98 months after surgery, and one Stage I11 patient treated with a pan- creaticoduodenectomy and the resection of the portal vein has been alive for 13.5 months after surgery. One Stage IV patient treated with a pancreaticoduodenectomy and portal vein resection and one with a total pancreatectomy have been alive 20.5 and 11 1 months, respectively, after surgery. The cumulative postoperative survival rates for all 32 patients treated with a radical pancreatectomy were 64.7% after 1 year, 38.2% after 2 years, 38.2% after 3 years,

No. 5 RADICAL PANCREATECTOMY FOR PANCREATIC CA . Munube et al. I135

" 1 2 3 4 5

YEARS POSTOPERATIVE PERIOD

FIG. 1. Cumulative survival rate for 32 patients treated with a radical pancreatectomy and 42 patients treated with a nonradical pancreatectomy (Kaplan-Meyer method) (P < 0.001).

33.4% after 4 years, and 33.4% after 5 years (Fig. 1). There have been six 4-year survivors and five 5-year survivors; the mean survival time was 29.8 months. For 14 Stage I and Stage I1 patients, the cumulative I-year, 2-year, 3- year, 4-year, and 5-year survival rates were 85. I%, 46.4%, 46.4%, 46.4%, and 46.4%, respectively (Fig. 2). There have been four 4-year survivors and three 5-year survivors; the mean survival time was 36.0 months. For 18 Stage I11 and Stage IV patients, the survival rates were 46.7%, 31.1%, 20.7%, 20.7%, and 20.7%, respectively (Fig. 3); there have been two 4-year survivors and two 5-year sur- vivors (mean survival time, 24.0 months).

Thirty-eight of 42 patients who had a nonradical pan- createctomy survived the operation. Among the seven Stage I1 patients treated with a nonradical pancreatectomy, one with a pancreaticoduodenectomy lived for 32.5 months, one with a total pancreatectomy died of septic shock of an unknown cause 2 months after surgery, and four with a pancreaticoduodenectomy and one with a total pancreatectomy died of a recurrence 8.6 to 17 months after surgery. One Stage I11 patient treated with a total pancreatectomy died of a recurrence at 32.5 months, and one Stage IV patient with a total pancreatectomy and the resection of the portal vein died of diabetes 2 months after surgery. Eight other Stage 111 patients treated with a pancreaticoduodenectomy (one with a total pancreatec- tomy and two with a total pancreatectomy and the resec- tion of the portal vein) and ten Stage IV patients treated with a pancreaticoduodenectomy (two with a pancreati- coduodenectomy and the resection of the portal vein, two with a total pancreatectomy, and two with a total pan- createctomy and the resection of the portal vein) died of a recurrence 1.7 to 19.2 months after surgery. The cu- mulative postoperative survival rates for all 42 patients treated with a nonradical pancreatectomy were 27.1 % after

%

RADICAL PANCREATECTOMY 4

3 v)

NON-RADICAL PANCREATECTOMY 20 -

1 2 3 4 5 YEARS

POSTOPERATIVE PERIOD

FIG. 2. Cumulative survival rate for Stage I and Stage 11 patients who had a radical pancreatectomy (n = 14) or a nonradical pancreatectomy (n = 7) (Kaplan-Meyer method) (0.1 < P < 0.05).

1 year, 5.4% after 2 years, and 0% after 3 years (Fig. 1); the mean survival time was 7.8 months. Patients treated with a radical pancreatectomy survived longer than those treated with a nonradical pancreatectomy. The difference in the survival times between the radical and the nonrad- ical pancreatectomy groups was Significant ( P < 0.001 as determined by the generalized Wilcoxon test). The cu- mulative 1 -year, 2-year, and 3-year survival rates of seven Stage I1 patients treated with nonradical surgery were 66.7%, 16.7%, and 0%, respectively (Fig. 2), with a mean survival time of 16.4 months. For Stage I and Stage I1 patients, radical surgery resulted in higher survival rates than those for nonradical surgery (0.05 < P < 0.1). The cumulative I-year, 2-year, and 3-year survival rates of 35 Stage I11 and Stage IV patients treated with nonradical

% 100

80

W + d 60

Q 1 2 40

-I

3 v)

20

NON-RADICAL PANCREATECTOMY

0 I . I

1 2 3 4 5 YEARS

POSTOPERATIVE PERIOD

FIG. 3. Cumulative survival rate for Stage 111 and Stage IV patients who had a radical pancreatectomy (n = 18) or a nonradical pancreatec- tomy (n = 35) (Kaplan-Meyer method) ( P < 0.05).

1136 CANCER September 1 1989 Vol. 64

operations were 19.4%, 3.2%, and 0%, respectively (Fig. 3); the mean survival time was 8.8 months. The survival rates of Stage 111 and Stage IV patients after radical op- erations were significantly better than those after nonrad- ical operations (P < 0.05).

Discussion

An analysis of the results in the published data of sur- gery for carcinoma of the pancreas is difficult because there is no classification of the carcinoma or evaluation of its operative radicality. In this study, we report on the results of pancreatectomies for ductal cell carcinoma of the head of the bancreas based on the stage and the rad- icality of the operation.

Our resectability rate of 37% in the treatment of car- cinoma of the head of the pancreas is better than that reported in the literature. This high rate of resectability is due mainly to the extensive procedure of en bloc radical pancreatectomy for advanced pancreatic carcinomas, particularly those involving the adjacent portai system and/or a r t e r i e~ .~ .~ Our postoperative mortality rates after the pancreatectomy were 0% for Stage I and Stage I1 car- cinomas, 11.3% for Stage 111 and Stage IV carcinomas, and 8.3% for the entire series. These results are better than the cumulative figures reported by Connolly ef aL6 in a review of published reports of more than 50 patients in each series, although there were great differences among the series. They reported that the mortality rates were 2 1.8% ( I 58 of 723 patients) after a pancreaticoduodenec- tomy and 23.9% (55 of 230 patients) after a total pancre- atectomy. An extended radical pancreaticoduodenectomy or a total pancreatectomy did not raise the mortality rate. Among our 28 patients treated with an extended pancre- atectomy combined with the resection of the adjacent vessels, only two patients died within 30 days after the operation.

The 5-year survival rates after a radical operation were 33.4% for the entire series, 46.4% for Stage I and Stage 11, and 20,770 for Stage 111 and Stage IV patients. These survival rates for ductal cell adenocarcinoma compare favorably with those in recent reports, which range from 0% to 26.7%.6 Kellum et al.' and Connolly ef aL6 pointed out that improved long-term survival rates are not due to advances in surgical techniques but to the fact that more tumors of less than 3 cm in diameter are being found and that many tumors are not histologically proven adeno- carcinomas. In our series, however, all of the tumors were ductal cell adenocarcinomas; cystoadenocarcinomas, islet cell carcinomas, and acinar cell carcinomas were excluded. Various sizes of resectable ductal cell carcinomas were included. We classified the stage of each resectable ductal cell carcinoma and divided the patients into radical and nonradical groups according to the extent of clearance of

the lymph node with the surrounding soft tissue and the state of the lymph node metastases in the line of resection. Resection was considered to be radical when the lymph nodes, beyond the group of lymph nodes with macro- scopic metastases, were negative by performing a regional or a juxta-regional lymph node dissection and by the clearance of the soft tissue from around the pancreas. This was performed by a skeletonization of major vessels to attain a sufficient free margin away from the carcinoma.

Resection was considered to be noncurative when the cleaning of the lymph nodes and the clearance of soft tissue around the pancreas were insufficient and when macroscopic examinations showed that cancer-containing tissues were suspected to be left behind. This strict intra- operative differentiation between radical and nonradical operations seems to be one of the reasons why our results are better than those of other reported series. However, even without this type of differentiation, the postoperative survival rates have been fairly good in the reported se- ries,63839 presumably because extensive radical pancrea- tectomies, which have the possibility of being curative, contribute to a longer survival time.

Among the factors influencing the prognosis of pan- creatic carcinoma, retroperitoneal invasion and portal system involvement are reduced radically by a standard pancreaticoduodenectomy. Fortner et al. lo suggested that even if the lesions involving adjacent vessels are resected during the course of a standard pancreatectomy, they are often incompletely removed. In fact, most patients in whom surgery was nonradical have been treated with a standard pancreaticoduodenectomy or a standard total pancreatectomy. In patients with large tumors and ret- roperitoneal or vessel involvement, lymph node metastasis and capsular invasion also were frequent.2 Lymph nodes adjacent to the portal system are frequent sites of metas- tasis because retroperitoneal lymph nodes are in the nor- mal direct pathway between the primary lesion in the head of the pancreas and celiac lymph nodes. Further- more, carcinoma of the pancreas often spreads beyond the usual pancreaticoduodenal resection line.'

Curative resection by an en bloc regional pancreatec- tomy with the removal of the portal vein and the adjacent soft tissues for advanced carcinoma of the pancreas, which was first advocated by Fortner4 in 1973 as a new approach, is an attractive procedure in view of the high probability of lymph node involvement and retroperitoneal invasion. There are many critical reports'* concerning the extended radical pancreatectomy, particularly about its unaccept- ably high operative mortality, considerable postoperative morbidity, and disappointingly low survival rate. Most reports of series in which extended radical pancreatecto- mies were performed are considered to include lesions that are too far advanced to make a sufficiently wide free margin around the tumor possible. We think that a radical

No. 5 RADICAL PANCREATECTOMY FOR PANCREATIC CA Manabe et al. 1137

pancreatectomy with a sufficiently wide free margin around the cancerous tissue by lymphatic and connective tissue clearance beyond the suspected lesions can be cu- rative. In patients with small localized carcinomas of less than 2 cm in diameter, fairly good results can be obtained by a pancreaticoduodenectomy if the radical operation is carried out with a sufficient margin of soft tissue and the clearance of regional lymph nodes around the pancreatic head.I3 However, only four of our 74 patients had Stage I carcinomas. The other patients with small carcinomas had lymph node metastases, capsular invasion, portal sys- tem involvement, or advanced stage tumors that needed extensive operations to be considered radical.

Our results substantiate the concept that a radical op- eration with the clearance of regional or juxta-regional lymph nodes beyond the suspected metastatic nodes and a greater margin of soft tissue resection is indispensable as a possible cure for ductal carcinoma of the pancreas.

REFERENCES

I . Monge JJ, Judd ES, Gaze RP. Radical pancreatectomy: A 22-year experience with complications, mortality rate, and survival rate. Ann Surg 1964; 160:711-722.

2. Manabe T, Suzuki T, Tobe T. Evaluation of en bloc radical pan-

createctomy for carcinoma of the head of the pancreas involving the adjacent vessels. Dig Surg 1985; 2:27-30.

3. Fortner JG. Surgical principles for pancreatic cancer: Regional total and subtotal pancreatectomy. Cancer 1981; 47:1712-1718.

4. Fortner JG. Regional resection of cancer of the pancreas: A new surgical approach. Surgery 1973; 73:303-320.

5. Japanese Pancreatic Society. General Rules for Surgical and Pathological Studies on Cancer of the Pancreas, ed. 3. Tokyo: Kanehara, 1986; 9-15 (in Japanese).

6. Connolly MM, Dawson PJ, Michelassi F ef al. Survival in 1001 patients with carcinoma of the pancreas. Ann Surg 1987; 206:366-373.

7. Kellum JM, Clark J, Miller HH. Pancreatoduodenectomy for re- sectable malignant periampullary tumor. Surg Cynecol Obsfet 1983; 157:

8. Longmire WP Jr. Cancer of the pancreas: Palliative operation, Whipple procedure, or total pancreatectomy. WorldJSurg 1984; 8:872- 879.

9. Nagakawa J, Kurachi M, Konishi K ef al. Translateral retroperi- toneal approach in radical surgery for pancreatic carcinoma. Jpn J Surg 1982; 12:229-233.

10. Fortner JG, Kim DK, Cubilla A ef al. Radical pancreatectomy: En bloc pancreatic portal vein and lymph node resection. Ann Surg

I I . Tryka AF, Brooks JR. Histopathology in the evaluation of total pancreatectomy for ductal carcinoma. Ann Surg 1979; 190:373-38 1.

12. Moosa AR, Scott MH, Lavelle-Jones M. The place of total and extended total pancreatectomy in pancreatic cancer. World JSurg 1984;

13. Manabe T, Miyashita T, Ohshio G ef al. Small carcinomas of the pancreas: Clinical and pathological evaluations of 17 patients. Cancer

362-366.

1977; 186:42-50.

8:895-899.

1988; 62: 135- 14 1.