- Email: [email protected]
Improving resectability and survival in patients with primary duodenal carcinoma
Improving Resectability and Survival in Patients With Primary Duodenal Carcinoma Romano
Delcore, MD, James H. Thomas, MD, Jameson Forster, MD, A d o S. Hermreck, MD, KansasCity,Kansas
Of 35 patients with primary duodenal carcinoma (PDC), 13 were treated between 1960 and 1974 (group I) and 22 between 1975 and 1990 (SToup I I ) . PDCs were found in the first 5 portions of the duodenum ( 14% ), second 18 ( 5 1 % ) , third 8 ( 2 3 % ) , and fourth 4 ( 1 2 % ) . Five patients (38%) in group I were deemed to have unresectable disease compared with only one patient (5%) in group II. Eight patients (62%) in group I underwent resection by either pancreatoduodenectomy ( 4 ) or segmental resection ( 4 ) , and 2 0 patients ( 9 5 % ) in group II had pancreatoduodenectomy ( 1 7 ) or segmental resection ( 3 ) . Operative mortality was 31% in group I and 0% in group II. Mean survival was 7 months (range: 0 to 22 months) in group I and 4 8 months (range: 6 to 2 1 8 months) in group II. None of the patients in group I survived for 2 years, whereas the 5-year survival for patients in group II was 62%. This experience suggests that resectability, operative mortality, and survival in patients with PDCs have improved markedly in recent years.
From the Departmentof Surgery,Universityof KansasMedicalCenter, KansasCity,Kansas. Requestsforreprintsshouldbc addressedto Romano Delcore,MD, Room 4945, Murphy Building,Universityof Kansas MedicalCenter, 3901 Rainbow Boulevard,Kansas City,Kansas 66160-7309. Presentedat the 45th Annual Meeting of the SouthwesternSurgicalCongress,Monterey,California,April 18-21, 1993.
626
duodenal carcinoma (PDC) is rare and is generally considered to have a low resectahility rate as Fwell}rimary as a dismal prognosis [1,2]. PDCs account for just 0.3% of all gastrointestinal malignancies but represent 30% to 45% of small intestinal adenocarcinomas [1-5]. Because fewer than 800 cases have been reported in the literature, awareness of the condition is limited, diagnosis is often delayed, and definitive surgical treatment for this lesion remains controversial [6]. Recently, improved results have been reported in the treatment of ampullary adenocarcinomas [7] and pancreatic adenocarcinomas [8], but such a trend has not been generally'recognized in the treatment of PDCs [1-5,9-11]. This study was undertaken to better characterize this unusual neoplasm and to determine if improvements in resectability and survival following surgical treatment have occurred in recent years comparable with those reported for other periampullary neoplasms. PATIENTS AND METHODS The records of all patients with PDCs treated at the University of Kansas Medical Center between 1960 and 1990 were reviewed. All periampullary lesions were excluded unless their origin from the duodenal mucosa could be unequivocally demonstrated. In particular, all patients with primary ampullary adenocarcinomas were excluded and have been previously reported [7]. Tumors were staged according to the staging system published by the American Joint Committee on Cancer [12]. Follow-up information was obtained through office visits and telephone contact with the patients and/or the referring physicians and was complete through 1993 or to the time of the patient's death in all cases. Operative mortality was defined as any death occurring prior to or within 30 days after hospital discharge. In an effort to detect recent trends, patients were divided into two groups on the basis of two 15-year time periods: those treated between January 1960 and December 1974 (group I), and those treated between January 1975 and December 1990 (group II). Patient groups were compared for mean age, duration of symptoms, and tumor size using the Student's unpaired t-test, x 2 analysis was used for all other comparisons between groups of patients. The estimated survival rate of patients discharged from the hospital was determined by the Kaplan-Meier product limit method of life-table analysis. A p value of less than 0.05 was used to indicate a significant difference. RESULTS Thirty-five patients with PDCs were treated between 1960 and 1990. The mean age of the patients was 62
THE AMERICAN JOURNAL OFSURGERY VOLUME166 DECEMBER1993
PRIMARY DUODENAL CARCINOMA
years (range: 37 to 84 years), and 22 (63%) patients were men. Patients were symptomatic for a mean of 7 months (range: 1 to 24 months) prior to diagnosis. Presenting symptoms were vague epigastric or abdominal pain (60%); obstructive symptoms including nausea, vomiting, and postprandial abdominal discomfort (60%); weight loss (more than 10% of body weight) (54%); occult and overt gastrointestinal bleexiing (43%); anemia (hemoglobin level of less than 10.0 g/dL) (40%); and jaundice (serum bilirubin level of greater than 2 mg/dL) (22%). Most patients underwent multiple preoperative diagnostic studies. Conventional upper gastrointestinal barium meal (UGI) studies were performed in 29 patients and were positive in 93%; in 2 patients with false-negative UGI studies, PDCs were located in the second (n = 1) and fourth (n -- 1) portions of the duodenum. Esophagogastroduodenoscopy (EGD) was performed in 20 patients. A lesion was identified in 19 (95%) patients, and biopsy specimens were obtained in 18 (95%) of these patients. One false-negative EGD occurred in a patient with PDC of the fourth portion of the duodenum that was correctly diagnosed by conventional UGI study. The re,suits of endoscopic biopsy revealed adenocarcinoma in 15 (83%) patients and villous or tubulovillous adenomatous tissue without evidence of neoplasia in 3 (17%) patients. Computed tomographic scans of the abdomen were obtained in 15 (43%) patients and were diagnostic or revealed abnormalities suggestive of an obstructive lesion in 67%. PDCs were located in the fh-st portion of the duodenum in 5 patients, the second in 18, the third in 8, and the fourth in 4. Of 18 tumors located in the second portion, 10 extended to involve the ampulla and 8 were infra-ampullary. All 35 patients had invasivc adenocarcinomas. Histologically, tumors were well differentiated (grade 1) in 9 patients, moderately differentiated (grade 2) in 19 patients, and poorly differentiated (grade 3) in 7 patients. The tumor was confmed to the duodenal wall (no extension beyond the serosa, TNM: T3) in four (11%) patients. Invasion through the serosa into the periduodenal fat (more than 2 cm) or invasion of other organs such as the pancreas, stomach, colon, jejunum, perirenal fat, or superior mesenteric vessels (TNM: T4) was present in 31 (89%) patients. At the time of surgical exploration, 22 (63%) patients had metastases involving the lymph nodes (n = 22), the liver (n -- 5), and the omentum, mesentcry, or peritoneum (n = 6). All patients with positive lymph nodes had PDCs that extended beyond the periduodenal fat and directly invaded other adjacent organs (TNM: T4). Three of four patients with tumors confined to the duodenal wall (TNM: T3) had negative lymph nodes. The fourth patient had gross distortion of the serosa without direct extension into the surrounding fat or pancreas but had lymph node and liver metastases. Mean tumor size was 5 cm (range: 2 to 10 cm). There was no correlation between tumor size and the presence or absence of either local invasion or metastases. PDCs were associated with villous adenomas in four (11%) patients, adenomatous polyps in three (9%) patients, and duodenal diverticula in two (6%) patients.
TABLE I Characteristics of 35 Patients With Primary Duodenal Carcinoma
Mean age (y) Mean duration of symptoms (mo) Pain Obstructive symptoms Gastrointestinal bleeding Upper gastrointestinal study Esophagogastroduodenoscopy
Groupl
Groupll
(n = 13) (%)
(n =22) (%)
63 9 8 (62) 9 (69) 9 (69) 13 (100) 3 (23)
62 6 13 (59) 12 (55) 6 (27)* 16 (73) 17 (77)*
*p <0.05 by x2analysis.
TABLE II Tumor Characteristics in 35 Patients With Primary Duodenal Carcinoma
Mean size (cm) Positive lymph nodes Metastases 1"4 Grade 1 2 3 Stage II III IV
Group I (n = 13)
Group II (n = 22)
(%)
(%)
6 8 (62) 4 (31) 12 (92)
5 14 (64) 6 (27) 19 (86)
3 (23) 6 (46) 4 (31)
6 (27) 13 (59) 3 (14)
5 (38) 4 (31) 4 (31)
8 (36) 8 (36) 6 (27)
Thrce other patients had previous colectomies for colon carcinoma (two with familial adcnomatosis coli/ Gardner's syndrome), and thrce patients had previous operations for peptic ulcer disease. Thirtcen patients were treated betwcen January 1960 and December 1974 (group I), and 22 patients were treated between January 1975 and December 1990 (group II). There was no significant difference between groups in terms of age, symptoms (with the exception of bleeding), duration of symptoms, operability, type of operation, metastases, tumor size, extent (TNM: T3 or T4), or grade (Tables I and H). There were no stage 0 or stage I patients in either group, and there was no significant difference in the pathologic staging of tumors (Table II) or in the use of adjuvant therapy between the groups. The two groups were significantly different with respect to preoperative gastrointestinal bleeding, number of patients evaluated by EGD, number of curative resections, and operative morbidity and mortality (Tables I and HI). Thirty-four (97%) patients underwent exploration (1 patient in group II died preoperatively of a cercbrovascular accident), and 28 (82%) underwent resection for cure. In group I, 8 (62%) patients underwent curative resection
THE AMERICAN JOURNAL OF SURGERY
VOLUME 166 DECEMBER 1993 627
DELCOREETAL
T A B L E IIl Results of S u r g i c a l T r e a t m e n t in 35 Patients With Primary Duodenal Carcinoma Group I (n = 13)
Group II (n = 22)
13 (100) 4 (31) 5 (38) 4 (31) 8 (62) 5 (38) 4 (31) 7 0 0
21 (96) 1 (5) 3 (14) 17 (81) 20 (95)* 2 (10)* 0* 48t 14 (67)* 62*
(%)
(%)
Laparotomy Bypass only Segmental resection Whipple's resection Curative resection Operative morbidity Operative mortality Overall survival (rno) Alive at 2 years 5-Year survival (%) * p < 0 . 0 5 by x 2 analysis. tp < 0 . 0 5 by Student's unpaired t-test.
100
80 i-
GROUP II (n=20)*
70 I= 9J
60 I-
~<
50 I=
I:o: 0
1
2
3
4
I 5
I 6
I 7
I 8
I 9
10
YEARS Figure 1. Estimated survival rates for 25 patients with primary duodenal carcinomas after curative resection. *p < 0 . 0 5 by X 2
analysis.
by pancreatoduodenectomy (n = 4) or segmental resection (n = 4), whereas 20 (95%) of 21 operated patients in group II underwent curative resections by pancreatoduodenectomy (n = 17) or segmental resection (n = 3) (p <0.05). Curative segmental resections in both groups were performed only for tumors located in the distal third (n ffi 4) or fourth portions (n = 3) of the duodenum. Five patients (38%) in group I and one patient (5%) in group II were not candidates for resection for cure due to extensive local invasion of vital structures or distant metastases. Of these six patients, one in group I underwent palliative segmental resection but had unresectable hepatic metastases, and five patients underwent palliative bypass operations only. Postoperative complications occurred in five (38%) patients in group I and two (10%) patients in group II. In group I, one patient developed an intra-abdominal abscess after segmental resection and required reoperation; he subsequently recx)vered and was discharged from the hospital. The other four patients in group I experienced lethal complications (operative mortality: 31%): one duodenojejunostomy leak after segmental resection, one thrombosis of the superior mesenteric vein after extended pancreatoduodenectomy that required portal vein resec628
THE AMERICAN JOURNAL OF SURGERY
tion with interposition grafting, one myocardial infarction after palliative bypass in an 84-year-old patient, and one gastric outlet obstruction after pancreatoduodenectomy in a patient with renal failure. In group II, one patient had a wound infection complicated by dehiscence, and one patient developed gastric outlet obstruction, both after pancreatoduodenectomies. These two patients required reoperation but recovered and were discharged from the hospital. There were no operative deaths in group II. A significant increase in resectability for cure and a significant decrease in overall operative morbidity and mortality were observed during the more recent time period (Table III). A significant difference in overall survival was also noted between the two groups (p <0.005) (Figere 1). All patients in group I had tumorrelated deaths, and none survived beyond 22 months, whereas 67% of patients in group II survived for at least 2 years after operation. The only patient who underwent palliative bypass in group II had both peritoneal and hepatic metastases at the time of operation and died of tumor-related complications (survival: 28 months), Three patients in group II died of unrelated causes and were free of tumor at 5, 12, and 36 months, respectively. Overall, the presence of gastrointestinal bleeding and preoperative symptoms of 4 months' duration or longer significantly affected survival. Patients with either occult or overt gastrointestinal bleeding had an overall 2-year survival rate of 17%, but none of the patients with overt bleeding survived for 2 years. Patients without bleeding survived for a mean of 60 months and had a 5-year survival rate of 60% (p <0.025). Patients who were symptomatic for 4 months or longer had a 2-year survival rate of 23%, whereas all patients who were symptomatic for less than 4 months are still alive after a mean follow-up of 7 years (p < 0.005). The presence of lymph node metastases also adversely affected survival after curative resection, whereas location, size, local invasion, and histologic grade of the tumor did not. Patients with lymph node metastases had a 2-year survival rate of 30%, whereas patients with negative lymph nodes had a 5-year survival rate of 72% (p <0.05). Of the 10 patients with stage IV disease, none survived for 2 years. Curative resections were attempted by pancreatoduodenectomy and liver resection in three of these patients. These three patients survived for 6, 10, and 22 months, respectively. The 5-year survival rate for patients with stage II disease was 72% compared with 18% for patients with stage III disease, but this difference did not reach statistical significance (p <0.1). Patients treated by curative pancreatoduodenectomy had a 5-year survival rate of 52%, whereas none of the patients treated by curative segmental resection survived for 2 years. Due to the small number of patients treated by segmental resection, however, this difference did not reach statistical significance (p <0.1). COMMENTS Although first described in 1746, PDC is a rarely encountered neoplasm found in only 0.035% of autopsies,
VOLUME 166 DECEMBER 1993
PRIMARY DUODENAL CARCINOMA
with just 694 cases recorded in the world literature between 1746 and 1974 [2]. The index of suspicion for this lesion remains very low, and the characteristic lack of specific symptoms or physical signs has consistently delayed diagnosis and treatment. In this study, patients treated after 1975 were diagnosed earlier but the delay for aggressive investigation and definitive diagnosis was still 6 months. The nonspecific symptoms of PDC are compatible with many benign conditions and are commonly attributed to the much more prevalent etiology of peptic ulcer disease. The most common presenting symptom was vague abdominal pain. This pain is usually epigastric and can be postprandial (obstructing lesions), intermittent (ulcerating lesions), or constant and gnawing (penetrating lesions) [4]. Obstructive symptoms such as early satiety, nausea, and vomiting occurred with the same frequency as pain in this study, and weight loss and anemia were present in approximately half of the patients. Occult gastrointestinal bleeding was present in 5 patients, but, more importantly, overt gastrointestinal bleeding with melena, hematochezia, or hematemesis was present in 10 patients and was an ominous prognostic factor [11]. None of the patients with overt gastrointestinal bleeding survived 2 years. Jaundice was present in eight patients with PDCs of the second portion but was usually preceded by weeks or months of other symptoms. The evaluation of patients with PDCs has changed significantly in recent years with the advent of EGD. All patients treated prior to 1979 were evaluated by UGI studies but only 4 of 17 patients underwent EGD. In contrast, 75% of patients treated after 1979 underwent UGI studies, whereas 90% were evaluated by EGD. Both UGI and EGD proved to be excellent studies for the evaluation of patients suspected of having PDCs. However, both tests were limited by their relative inability to routinely and fully evaluate the distal duodenum. Two of three false-negative results occurred in patients with lesions located in the distal duodenum. This observation emphasizes the need for meticulous evaluation of the entire length of the duodenum by either study. When in doubt, the examination should be repeated with a long scope or in conjunction with hypotonic duodenography [2,6,13]. Although computed tomography has been reported to reliably demonstrate sites of distant metastases, vascular encasement, and contiguous\ organ invasion [14], it was less accurate (67%) when compared with UGI and EGD in the primary diagnosis of PDCs in the present study. The rarity of PDC has precluded adequate study of possible etiologic factors, and no cause has yet been identified for this neoplasm [1,3]. PDCs arising from adenomatous polyps have been reported in patients with the familial adenomatosis coli/Gardner's syndrome, and two patients in this study had this syndrome [15,16]. The premalignant nature of duodenal villous adenomas is well known, and the frequency of adenocarcinoma in villous adenomas (30% to 45%) demands that this finding be regarded with great concern [17-19]. Three patients in this study had biopsy results that demonstrated only benign villous adenoma tissue but after resection were
found to have invasive adenocarcinomas. Only complete resection with adequate margins and careful histologic evaluation of the entire adenoma can reliably exclude the presence of invasive adenocarcinoma. Occasionally, it is technically possible to s u ~ s f u l l y remove small villous adenomas by local excision. However, careful endoscopic follow-up is essential, since one patient in this study developed PDC after complete local resection of a duodenal villous adenoma. In fact, the difficulty in making an accurate diagnosis and the high chance of recurrence after local excision have prompted several authors to recommend pancreatoduodenectomy as the initial form of treatment for duodenal villous adenomas [17,19]. Surprisingly, there was no significant difference in the staging of patients in groups I and II (Table II), suggesting that, although the mean duration of symptoms was shorter and EGD more frequently performed in patients in group II, no significant improvement has been made in the early detection of PDCs during the past 30 years. Duration of symptoms, bleeding, tumor size, grade, location, or local invasion, and lymph node metastases have all been suggested to influence survival of patients with PDC by some investigators, but others have not confirmed these findings [1-6,9-11,13]. In this study, duration of symptoms, overt gastrointestinal bleeding, and lymph node metastases were the only factors found to significantly affect long-term survival after curative resection. Despite positive lymph nodes, five patients in this series are alive and free of disease 22, 24, 28, 30, and 45 months, respectively, after curative pancreatoduodenectomy. The presence of regional lymph node metastases normally removed as part of the pancreatoduodenectomy specimen should not be considered a contraindicati0n to curative resection since it offers the only chance for cure and significant palliation. In this study, only distal lesions were treated by curative segmental duodenal resection. Of 12 patients with PDCs of the third or fourth portion of the duodenum, 7 were treated by segmental resection, 3 had palliative bypasses, and 2 underwent pancreatoduodenectomy. No patient treated by segmental resection survived for more than 2 years, and the only cure amongst these 12 patients occurred in a patient treated by pancreatoduodenectomy (survival: 18 years). Occasional survivors have been reported after segmental resection [11], and a recent study suggests that this should be the operative treatment of choice for patients with PDCs of the third and fourth portions of the duodenum [13]. Nevertheless, pancreatoduodenectomy, with en bloc resection of all adjacent tissues and removal of the regional lymph nodes, remains the only procedure that satisfies the technical principles of an adequate, curative cancer operation. In the past, pancreatoduodenectomywas associated with high operative morbidity and mortality rates, which led some surgeons to abandon curative resection in favor of lesser operations such as segmental resection or routine palliative bypass procedures for the treatment of duodenal and other periampullary carcinomas. Recently, however, a marked reduction in operative morbidity and mortality has been consistently reported from specialized centers,
THE AMERICAN JOURNAL OF SURGERY
VOLUME 166
DECEMBER 1993
629
DELCOREETAL
DISCUSSION Claude H. Organ, Jr. (Oakland, CA): The Kansas group has again delved into their clinical experience and presented a series of patients with primary duodenal carcinomas treated at their Medical Center during a 30-year period. They have provided evidence of improved results: (1) no mortality in the most recent 15-year period (30% versus 0%), (2) decreased morbidity (30% versus 10%), and (3) an overall improved survival rate (0 versus 62%). Their surgical management in this same time frame has utilized fewer bypass procedures, fewer segmental resections (38% versus 14%), and an increase in the number of pancreatoduodenal resections (26% versus 95%). These dramatic improvements clearly lead to the conclusion REFERENCES that they were indeed better surgeons than their known I. Alwmark A, AnderssonA, LassonA. Primary carcinomaof the predecessors, many of whom were their teachers. duodenum. Ann Surg 1980; 191: 13-8. In 17% of the patients, Whipple's procedure was per2. Spira IA, Ghazi A, Wolff WI. Primary adenecarcinomaof the formed without a positive histologic diagnosis. In 13 of 35 duodenum. Cancer 1977; 39: 1721-6. 3. Cortese AF, Cornell GN. Carcinoma of the duodenum. Cancer patients (37%), resectional procedures were undertaken in the presence of distant metastases. I am concerned that 1972; 29: 1010-5. 4. Lillemoe K, Imbembo AL. Malignant neoplasms of the duode- 51% of this patient population had disease in the second num. Surg Gynecol Obstet 1980; 150: 822-6. part of the duodenum. Is your definition of primary duo5. van Ooijen B, KalsbeekHL. Carcinoma of the duodenum. Surg denal carcinoma made solely on the basis of excluding Gynecol Obstet 1988; 166: 343-7. ampullary or periampullary carcinoma? There are im6. Lai ECS, Doty JE, Irving C, Tompkins RK. Primary adenocar- portant structures in close juxtoposition with the duoder of the duodenum: analysis of survival. World J Surg 1988; nal mucosa, and 86% of your study group had disease of 12: 695-9. 7. DelcoreR, Connor CS, Thomas JH, Friesen SR, Hermreck AS. the cmbryologic mid-gut. Does this make for an invalid Significanceof tumor spread in adenocareinomaof the ampulla of comparison with the 14% rising from the foregut since there is increasing evidence that foregut tumors are more eater. Am J Surg 1989; I58: 593-8. 8. Crist DW, Sitzmann JV, Cameron JL. Improvedhospital mor- much aggressive than those in the mid-gut? bidity, mortality, and survival after the Whipple procedure. Ann Your database in some parts of the manuscript is a bit Surg 1987; 206: 358-65. confusing. If all 35 patients had invasive carcinoma as 9. Nakase A, Matsumoto Y, Uchida K, Honjo I. Surgical treat- your text suggests, what is the justification for the 22% ment of cancer of the pancreas and the periampullaryregion:cumu- segmental rate? In your text, 13 or 37% had metastases to lative results in 57 institutionsin Japan. Ann Surg 1977; 185:52-7. the liver, the omentum, the mesentary, or the peritoneum. I0. Michelassi F, Erroi F, Dawson PJ, et al. Experiencewith 647 It is unclear from your data whether the denominator is consecutivetumors of the duodenum, ampulla head of the pancreas 34 or 35. Furthermore, in the introduction to the paper, and distal common bile duet. Ann Surg 1989; 210: 544-56. you state that the past literature demonstrates 800 cases 11. Jcesting DR, Beart RW, Van Heerden JA, Weiland LH. Improving survival in adenocareinoma of the duodenum. Am J previously reported and yet in the text of the manuscript, you say 690. Finally, two further questions: (1) Are your Surg 1981; 141: 228-31. 12. Small intestine. In: Beahrs OH, Henson DE, Hutter RVP, improved results due to your improved management of Kennedy BJ, editors. Manual for staging of cancer. Philadelphia: the pancreaticoenteric anastomosis? Is your group comJB Lippincott, 1992: 69-73. posed mainly of duct-to-mucosa surgeons, or are they 13. Lowell JA, Rossi RL, Munson L, Braaseh JW. Primary adeno- stuffers of the pancreas (i.e., simple end-to-end anastocarcinoma of third and fourth portions of duodenum; Favorable moses)? (2) Are there other reasons to justify the aggresprognosis after resection. Arch Surg 1992; 127: 557-60. sive role of pancreatoduodenal resection for palliative 14. Kazerooni EA, Quint LE, Francis IR. Duodenal neoplasms: predictive value of CT for determining malignancy and tumor therapy? Frank Johnson (St. Louis, MO): It is difficult to resectability. AIR Am J Roentgcnol 1992; 159: 303-9. 15. Schnur PL, David E, BrownPW, et al. Adenocarcinomaof the separate duodenal carcinoma, pancreas carcinoma, and duodenumand the Gardner syndrome.JAMA 1973;223:1229-32. ampullary carcinoma from one another. Could you spec16. Iida M, Yao T, Itoh H, et al. Natural history of duodenal ulate on whether the diagnostic criteria you used to inlesions in Japanese patients with familial adenomatosiscoli (Gard- clude patients in your study might somehow have affectner's syndrome). Gastroenterology 1989; 96: 1301-6. ed the conclusions that you reached? That is to say, might 17. Ryan DP, Sehapiro RH, Warshaw AL. Villous tumors of the a few patients with duodenal carcinoma have been classiduodenum. Ann Surg 1986; 203: 301-6. fied as having pancreas cancer or ampullary carcinoma 18. Galandink S, Hermann RE, Jagelman DG, Fazio VW, Sivak and might this have affected your results? Were there MV. Villoustumors of the duodenum.Ann Surg 1988;207: 234-9. 19. Chappuis CW, Divincenti FC, Cohn I. Villous tumors of the any patients with duodenal wall gastrinoma or other unusual histologies that might have accounted for the reladuodenum. Ann Surg 1989; 209: 593-9. 2 0 . Delcore R, Thomas JH, Pierce GE, Hermreek AS. Panereato- tively good results? Romano Delcore (closing): We believe that the reagastrostomy: a safe drainage procedure after panereatoduodenectomy. Surgery 1990; 108: 641-7. son for the much better operative morbidity and mortaliincluding our own, where a small number of surgeons perform a large number of pancreatoduodenectomies in less time and with less blood loss [8,20]. Perhaps the most significant finding in this study is the absence of operative mortality and the sharply reduced operative morbidity (10%) that has occurred since 1975. During the last 15 years, all but 4 of 22 patients with PDCs were aggressively treated by curative pancreatoduodenectomy, resulting in a markedly improved 5-year survival rate of 62%. This survival is comparable with that of patients with primary ampullary carcinoma, which has been considered the most curable of all upper gastrointestinal malignancies.
630
THE AMERICAN JOURNAL OF SURGERY VOLUME 166 DECEMBER 1993
PRIMARY DUODENALCARCINOMA
ty in the recent group of patients is our use of pancreatogastrostomy as opposed to pancreatojejunostomy. As previously reported, we have found pancreatogastrostomy to be much safer than conventional pancreatojejunostomy ([20] in the original paper). It is occasionally very difficult to differentiate bet w i n the four different histologic typos of l~riampullary carcinomas. In this study, it was fairly simple because 4 years ago we presented to this Congress our experience with primary ampuUary carcinomas ([7] in the original paper). Therefore, we had already evaluated the exclusion criteria and separated primary ampullary, duodenal, common bile duct, and pancreatic neoplasms. Specifically, whenever the pathologist was unable to determine the epithelium Of origin with certainty, the tumor histologic origin was classified as unknown and was excluded from
the study. In this study, only carcinomas that unequivocally originated from duodenal mucosal epithelium were considered primary duodenal carcinomas. In fact, it has probably been previous inclusion of duodenal primary tumors in scrips of pancreatic carcinomas that has fictitiously provided pancreatic carcinoma with a better survival than it actually has, rather than the opposite being true. To address the question about duodenal gastrinomas, we previously reported to this Congress our experience with 15 primary duodenal wall gastrinomas (Delcore R, Cheung LY, Friesen SR. Characteristics of duodenal wall gastrinomas. Am J Surg 1990; I60: 621-4). These tumors were also excluded from the present study. Finally, survival rates were calculated based on 34 patients, since we excluded 1 patient who died during the preoperative evaluation.
THE AMERICAN JOURNAL OF SURGERY vOLUME 166 DECEMBER 1993 631
Delcore, MD, James H. Thomas, MD, Jameson Forster, MD, A d o S. Hermreck, MD, KansasCity,Kansas
Of 35 patients with primary duodenal carcinoma (PDC), 13 were treated between 1960 and 1974 (group I) and 22 between 1975 and 1990 (SToup I I ) . PDCs were found in the first 5 portions of the duodenum ( 14% ), second 18 ( 5 1 % ) , third 8 ( 2 3 % ) , and fourth 4 ( 1 2 % ) . Five patients (38%) in group I were deemed to have unresectable disease compared with only one patient (5%) in group II. Eight patients (62%) in group I underwent resection by either pancreatoduodenectomy ( 4 ) or segmental resection ( 4 ) , and 2 0 patients ( 9 5 % ) in group II had pancreatoduodenectomy ( 1 7 ) or segmental resection ( 3 ) . Operative mortality was 31% in group I and 0% in group II. Mean survival was 7 months (range: 0 to 22 months) in group I and 4 8 months (range: 6 to 2 1 8 months) in group II. None of the patients in group I survived for 2 years, whereas the 5-year survival for patients in group II was 62%. This experience suggests that resectability, operative mortality, and survival in patients with PDCs have improved markedly in recent years.
From the Departmentof Surgery,Universityof KansasMedicalCenter, KansasCity,Kansas. Requestsforreprintsshouldbc addressedto Romano Delcore,MD, Room 4945, Murphy Building,Universityof Kansas MedicalCenter, 3901 Rainbow Boulevard,Kansas City,Kansas 66160-7309. Presentedat the 45th Annual Meeting of the SouthwesternSurgicalCongress,Monterey,California,April 18-21, 1993.
626
duodenal carcinoma (PDC) is rare and is generally considered to have a low resectahility rate as Fwell}rimary as a dismal prognosis [1,2]. PDCs account for just 0.3% of all gastrointestinal malignancies but represent 30% to 45% of small intestinal adenocarcinomas [1-5]. Because fewer than 800 cases have been reported in the literature, awareness of the condition is limited, diagnosis is often delayed, and definitive surgical treatment for this lesion remains controversial [6]. Recently, improved results have been reported in the treatment of ampullary adenocarcinomas [7] and pancreatic adenocarcinomas [8], but such a trend has not been generally'recognized in the treatment of PDCs [1-5,9-11]. This study was undertaken to better characterize this unusual neoplasm and to determine if improvements in resectability and survival following surgical treatment have occurred in recent years comparable with those reported for other periampullary neoplasms. PATIENTS AND METHODS The records of all patients with PDCs treated at the University of Kansas Medical Center between 1960 and 1990 were reviewed. All periampullary lesions were excluded unless their origin from the duodenal mucosa could be unequivocally demonstrated. In particular, all patients with primary ampullary adenocarcinomas were excluded and have been previously reported [7]. Tumors were staged according to the staging system published by the American Joint Committee on Cancer [12]. Follow-up information was obtained through office visits and telephone contact with the patients and/or the referring physicians and was complete through 1993 or to the time of the patient's death in all cases. Operative mortality was defined as any death occurring prior to or within 30 days after hospital discharge. In an effort to detect recent trends, patients were divided into two groups on the basis of two 15-year time periods: those treated between January 1960 and December 1974 (group I), and those treated between January 1975 and December 1990 (group II). Patient groups were compared for mean age, duration of symptoms, and tumor size using the Student's unpaired t-test, x 2 analysis was used for all other comparisons between groups of patients. The estimated survival rate of patients discharged from the hospital was determined by the Kaplan-Meier product limit method of life-table analysis. A p value of less than 0.05 was used to indicate a significant difference. RESULTS Thirty-five patients with PDCs were treated between 1960 and 1990. The mean age of the patients was 62
THE AMERICAN JOURNAL OFSURGERY VOLUME166 DECEMBER1993
PRIMARY DUODENAL CARCINOMA
years (range: 37 to 84 years), and 22 (63%) patients were men. Patients were symptomatic for a mean of 7 months (range: 1 to 24 months) prior to diagnosis. Presenting symptoms were vague epigastric or abdominal pain (60%); obstructive symptoms including nausea, vomiting, and postprandial abdominal discomfort (60%); weight loss (more than 10% of body weight) (54%); occult and overt gastrointestinal bleexiing (43%); anemia (hemoglobin level of less than 10.0 g/dL) (40%); and jaundice (serum bilirubin level of greater than 2 mg/dL) (22%). Most patients underwent multiple preoperative diagnostic studies. Conventional upper gastrointestinal barium meal (UGI) studies were performed in 29 patients and were positive in 93%; in 2 patients with false-negative UGI studies, PDCs were located in the second (n = 1) and fourth (n -- 1) portions of the duodenum. Esophagogastroduodenoscopy (EGD) was performed in 20 patients. A lesion was identified in 19 (95%) patients, and biopsy specimens were obtained in 18 (95%) of these patients. One false-negative EGD occurred in a patient with PDC of the fourth portion of the duodenum that was correctly diagnosed by conventional UGI study. The re,suits of endoscopic biopsy revealed adenocarcinoma in 15 (83%) patients and villous or tubulovillous adenomatous tissue without evidence of neoplasia in 3 (17%) patients. Computed tomographic scans of the abdomen were obtained in 15 (43%) patients and were diagnostic or revealed abnormalities suggestive of an obstructive lesion in 67%. PDCs were located in the fh-st portion of the duodenum in 5 patients, the second in 18, the third in 8, and the fourth in 4. Of 18 tumors located in the second portion, 10 extended to involve the ampulla and 8 were infra-ampullary. All 35 patients had invasivc adenocarcinomas. Histologically, tumors were well differentiated (grade 1) in 9 patients, moderately differentiated (grade 2) in 19 patients, and poorly differentiated (grade 3) in 7 patients. The tumor was confmed to the duodenal wall (no extension beyond the serosa, TNM: T3) in four (11%) patients. Invasion through the serosa into the periduodenal fat (more than 2 cm) or invasion of other organs such as the pancreas, stomach, colon, jejunum, perirenal fat, or superior mesenteric vessels (TNM: T4) was present in 31 (89%) patients. At the time of surgical exploration, 22 (63%) patients had metastases involving the lymph nodes (n = 22), the liver (n -- 5), and the omentum, mesentcry, or peritoneum (n = 6). All patients with positive lymph nodes had PDCs that extended beyond the periduodenal fat and directly invaded other adjacent organs (TNM: T4). Three of four patients with tumors confined to the duodenal wall (TNM: T3) had negative lymph nodes. The fourth patient had gross distortion of the serosa without direct extension into the surrounding fat or pancreas but had lymph node and liver metastases. Mean tumor size was 5 cm (range: 2 to 10 cm). There was no correlation between tumor size and the presence or absence of either local invasion or metastases. PDCs were associated with villous adenomas in four (11%) patients, adenomatous polyps in three (9%) patients, and duodenal diverticula in two (6%) patients.
TABLE I Characteristics of 35 Patients With Primary Duodenal Carcinoma
Mean age (y) Mean duration of symptoms (mo) Pain Obstructive symptoms Gastrointestinal bleeding Upper gastrointestinal study Esophagogastroduodenoscopy
Groupl
Groupll
(n = 13) (%)
(n =22) (%)
63 9 8 (62) 9 (69) 9 (69) 13 (100) 3 (23)
62 6 13 (59) 12 (55) 6 (27)* 16 (73) 17 (77)*
*p <0.05 by x2analysis.
TABLE II Tumor Characteristics in 35 Patients With Primary Duodenal Carcinoma
Mean size (cm) Positive lymph nodes Metastases 1"4 Grade 1 2 3 Stage II III IV
Group I (n = 13)
Group II (n = 22)
(%)
(%)
6 8 (62) 4 (31) 12 (92)
5 14 (64) 6 (27) 19 (86)
3 (23) 6 (46) 4 (31)
6 (27) 13 (59) 3 (14)
5 (38) 4 (31) 4 (31)
8 (36) 8 (36) 6 (27)
Thrce other patients had previous colectomies for colon carcinoma (two with familial adcnomatosis coli/ Gardner's syndrome), and thrce patients had previous operations for peptic ulcer disease. Thirtcen patients were treated betwcen January 1960 and December 1974 (group I), and 22 patients were treated between January 1975 and December 1990 (group II). There was no significant difference between groups in terms of age, symptoms (with the exception of bleeding), duration of symptoms, operability, type of operation, metastases, tumor size, extent (TNM: T3 or T4), or grade (Tables I and H). There were no stage 0 or stage I patients in either group, and there was no significant difference in the pathologic staging of tumors (Table II) or in the use of adjuvant therapy between the groups. The two groups were significantly different with respect to preoperative gastrointestinal bleeding, number of patients evaluated by EGD, number of curative resections, and operative morbidity and mortality (Tables I and HI). Thirty-four (97%) patients underwent exploration (1 patient in group II died preoperatively of a cercbrovascular accident), and 28 (82%) underwent resection for cure. In group I, 8 (62%) patients underwent curative resection
THE AMERICAN JOURNAL OF SURGERY
VOLUME 166 DECEMBER 1993 627
DELCOREETAL
T A B L E IIl Results of S u r g i c a l T r e a t m e n t in 35 Patients With Primary Duodenal Carcinoma Group I (n = 13)
Group II (n = 22)
13 (100) 4 (31) 5 (38) 4 (31) 8 (62) 5 (38) 4 (31) 7 0 0
21 (96) 1 (5) 3 (14) 17 (81) 20 (95)* 2 (10)* 0* 48t 14 (67)* 62*
(%)
(%)
Laparotomy Bypass only Segmental resection Whipple's resection Curative resection Operative morbidity Operative mortality Overall survival (rno) Alive at 2 years 5-Year survival (%) * p < 0 . 0 5 by x 2 analysis. tp < 0 . 0 5 by Student's unpaired t-test.
100
80 i-
GROUP II (n=20)*
70 I= 9J
60 I-
~<
50 I=
I:o: 0
1
2
3
4
I 5
I 6
I 7
I 8
I 9
10
YEARS Figure 1. Estimated survival rates for 25 patients with primary duodenal carcinomas after curative resection. *p < 0 . 0 5 by X 2
analysis.
by pancreatoduodenectomy (n = 4) or segmental resection (n = 4), whereas 20 (95%) of 21 operated patients in group II underwent curative resections by pancreatoduodenectomy (n = 17) or segmental resection (n = 3) (p <0.05). Curative segmental resections in both groups were performed only for tumors located in the distal third (n ffi 4) or fourth portions (n = 3) of the duodenum. Five patients (38%) in group I and one patient (5%) in group II were not candidates for resection for cure due to extensive local invasion of vital structures or distant metastases. Of these six patients, one in group I underwent palliative segmental resection but had unresectable hepatic metastases, and five patients underwent palliative bypass operations only. Postoperative complications occurred in five (38%) patients in group I and two (10%) patients in group II. In group I, one patient developed an intra-abdominal abscess after segmental resection and required reoperation; he subsequently recx)vered and was discharged from the hospital. The other four patients in group I experienced lethal complications (operative mortality: 31%): one duodenojejunostomy leak after segmental resection, one thrombosis of the superior mesenteric vein after extended pancreatoduodenectomy that required portal vein resec628
THE AMERICAN JOURNAL OF SURGERY
tion with interposition grafting, one myocardial infarction after palliative bypass in an 84-year-old patient, and one gastric outlet obstruction after pancreatoduodenectomy in a patient with renal failure. In group II, one patient had a wound infection complicated by dehiscence, and one patient developed gastric outlet obstruction, both after pancreatoduodenectomies. These two patients required reoperation but recovered and were discharged from the hospital. There were no operative deaths in group II. A significant increase in resectability for cure and a significant decrease in overall operative morbidity and mortality were observed during the more recent time period (Table III). A significant difference in overall survival was also noted between the two groups (p <0.005) (Figere 1). All patients in group I had tumorrelated deaths, and none survived beyond 22 months, whereas 67% of patients in group II survived for at least 2 years after operation. The only patient who underwent palliative bypass in group II had both peritoneal and hepatic metastases at the time of operation and died of tumor-related complications (survival: 28 months), Three patients in group II died of unrelated causes and were free of tumor at 5, 12, and 36 months, respectively. Overall, the presence of gastrointestinal bleeding and preoperative symptoms of 4 months' duration or longer significantly affected survival. Patients with either occult or overt gastrointestinal bleeding had an overall 2-year survival rate of 17%, but none of the patients with overt bleeding survived for 2 years. Patients without bleeding survived for a mean of 60 months and had a 5-year survival rate of 60% (p <0.025). Patients who were symptomatic for 4 months or longer had a 2-year survival rate of 23%, whereas all patients who were symptomatic for less than 4 months are still alive after a mean follow-up of 7 years (p < 0.005). The presence of lymph node metastases also adversely affected survival after curative resection, whereas location, size, local invasion, and histologic grade of the tumor did not. Patients with lymph node metastases had a 2-year survival rate of 30%, whereas patients with negative lymph nodes had a 5-year survival rate of 72% (p <0.05). Of the 10 patients with stage IV disease, none survived for 2 years. Curative resections were attempted by pancreatoduodenectomy and liver resection in three of these patients. These three patients survived for 6, 10, and 22 months, respectively. The 5-year survival rate for patients with stage II disease was 72% compared with 18% for patients with stage III disease, but this difference did not reach statistical significance (p <0.1). Patients treated by curative pancreatoduodenectomy had a 5-year survival rate of 52%, whereas none of the patients treated by curative segmental resection survived for 2 years. Due to the small number of patients treated by segmental resection, however, this difference did not reach statistical significance (p <0.1). COMMENTS Although first described in 1746, PDC is a rarely encountered neoplasm found in only 0.035% of autopsies,
VOLUME 166 DECEMBER 1993
PRIMARY DUODENAL CARCINOMA
with just 694 cases recorded in the world literature between 1746 and 1974 [2]. The index of suspicion for this lesion remains very low, and the characteristic lack of specific symptoms or physical signs has consistently delayed diagnosis and treatment. In this study, patients treated after 1975 were diagnosed earlier but the delay for aggressive investigation and definitive diagnosis was still 6 months. The nonspecific symptoms of PDC are compatible with many benign conditions and are commonly attributed to the much more prevalent etiology of peptic ulcer disease. The most common presenting symptom was vague abdominal pain. This pain is usually epigastric and can be postprandial (obstructing lesions), intermittent (ulcerating lesions), or constant and gnawing (penetrating lesions) [4]. Obstructive symptoms such as early satiety, nausea, and vomiting occurred with the same frequency as pain in this study, and weight loss and anemia were present in approximately half of the patients. Occult gastrointestinal bleeding was present in 5 patients, but, more importantly, overt gastrointestinal bleeding with melena, hematochezia, or hematemesis was present in 10 patients and was an ominous prognostic factor [11]. None of the patients with overt gastrointestinal bleeding survived 2 years. Jaundice was present in eight patients with PDCs of the second portion but was usually preceded by weeks or months of other symptoms. The evaluation of patients with PDCs has changed significantly in recent years with the advent of EGD. All patients treated prior to 1979 were evaluated by UGI studies but only 4 of 17 patients underwent EGD. In contrast, 75% of patients treated after 1979 underwent UGI studies, whereas 90% were evaluated by EGD. Both UGI and EGD proved to be excellent studies for the evaluation of patients suspected of having PDCs. However, both tests were limited by their relative inability to routinely and fully evaluate the distal duodenum. Two of three false-negative results occurred in patients with lesions located in the distal duodenum. This observation emphasizes the need for meticulous evaluation of the entire length of the duodenum by either study. When in doubt, the examination should be repeated with a long scope or in conjunction with hypotonic duodenography [2,6,13]. Although computed tomography has been reported to reliably demonstrate sites of distant metastases, vascular encasement, and contiguous\ organ invasion [14], it was less accurate (67%) when compared with UGI and EGD in the primary diagnosis of PDCs in the present study. The rarity of PDC has precluded adequate study of possible etiologic factors, and no cause has yet been identified for this neoplasm [1,3]. PDCs arising from adenomatous polyps have been reported in patients with the familial adenomatosis coli/Gardner's syndrome, and two patients in this study had this syndrome [15,16]. The premalignant nature of duodenal villous adenomas is well known, and the frequency of adenocarcinoma in villous adenomas (30% to 45%) demands that this finding be regarded with great concern [17-19]. Three patients in this study had biopsy results that demonstrated only benign villous adenoma tissue but after resection were
found to have invasive adenocarcinomas. Only complete resection with adequate margins and careful histologic evaluation of the entire adenoma can reliably exclude the presence of invasive adenocarcinoma. Occasionally, it is technically possible to s u ~ s f u l l y remove small villous adenomas by local excision. However, careful endoscopic follow-up is essential, since one patient in this study developed PDC after complete local resection of a duodenal villous adenoma. In fact, the difficulty in making an accurate diagnosis and the high chance of recurrence after local excision have prompted several authors to recommend pancreatoduodenectomy as the initial form of treatment for duodenal villous adenomas [17,19]. Surprisingly, there was no significant difference in the staging of patients in groups I and II (Table II), suggesting that, although the mean duration of symptoms was shorter and EGD more frequently performed in patients in group II, no significant improvement has been made in the early detection of PDCs during the past 30 years. Duration of symptoms, bleeding, tumor size, grade, location, or local invasion, and lymph node metastases have all been suggested to influence survival of patients with PDC by some investigators, but others have not confirmed these findings [1-6,9-11,13]. In this study, duration of symptoms, overt gastrointestinal bleeding, and lymph node metastases were the only factors found to significantly affect long-term survival after curative resection. Despite positive lymph nodes, five patients in this series are alive and free of disease 22, 24, 28, 30, and 45 months, respectively, after curative pancreatoduodenectomy. The presence of regional lymph node metastases normally removed as part of the pancreatoduodenectomy specimen should not be considered a contraindicati0n to curative resection since it offers the only chance for cure and significant palliation. In this study, only distal lesions were treated by curative segmental duodenal resection. Of 12 patients with PDCs of the third or fourth portion of the duodenum, 7 were treated by segmental resection, 3 had palliative bypasses, and 2 underwent pancreatoduodenectomy. No patient treated by segmental resection survived for more than 2 years, and the only cure amongst these 12 patients occurred in a patient treated by pancreatoduodenectomy (survival: 18 years). Occasional survivors have been reported after segmental resection [11], and a recent study suggests that this should be the operative treatment of choice for patients with PDCs of the third and fourth portions of the duodenum [13]. Nevertheless, pancreatoduodenectomy, with en bloc resection of all adjacent tissues and removal of the regional lymph nodes, remains the only procedure that satisfies the technical principles of an adequate, curative cancer operation. In the past, pancreatoduodenectomywas associated with high operative morbidity and mortality rates, which led some surgeons to abandon curative resection in favor of lesser operations such as segmental resection or routine palliative bypass procedures for the treatment of duodenal and other periampullary carcinomas. Recently, however, a marked reduction in operative morbidity and mortality has been consistently reported from specialized centers,
THE AMERICAN JOURNAL OF SURGERY
VOLUME 166
DECEMBER 1993
629
DELCOREETAL
DISCUSSION Claude H. Organ, Jr. (Oakland, CA): The Kansas group has again delved into their clinical experience and presented a series of patients with primary duodenal carcinomas treated at their Medical Center during a 30-year period. They have provided evidence of improved results: (1) no mortality in the most recent 15-year period (30% versus 0%), (2) decreased morbidity (30% versus 10%), and (3) an overall improved survival rate (0 versus 62%). Their surgical management in this same time frame has utilized fewer bypass procedures, fewer segmental resections (38% versus 14%), and an increase in the number of pancreatoduodenal resections (26% versus 95%). These dramatic improvements clearly lead to the conclusion REFERENCES that they were indeed better surgeons than their known I. Alwmark A, AnderssonA, LassonA. Primary carcinomaof the predecessors, many of whom were their teachers. duodenum. Ann Surg 1980; 191: 13-8. In 17% of the patients, Whipple's procedure was per2. Spira IA, Ghazi A, Wolff WI. Primary adenecarcinomaof the formed without a positive histologic diagnosis. In 13 of 35 duodenum. Cancer 1977; 39: 1721-6. 3. Cortese AF, Cornell GN. Carcinoma of the duodenum. Cancer patients (37%), resectional procedures were undertaken in the presence of distant metastases. I am concerned that 1972; 29: 1010-5. 4. Lillemoe K, Imbembo AL. Malignant neoplasms of the duode- 51% of this patient population had disease in the second num. Surg Gynecol Obstet 1980; 150: 822-6. part of the duodenum. Is your definition of primary duo5. van Ooijen B, KalsbeekHL. Carcinoma of the duodenum. Surg denal carcinoma made solely on the basis of excluding Gynecol Obstet 1988; 166: 343-7. ampullary or periampullary carcinoma? There are im6. Lai ECS, Doty JE, Irving C, Tompkins RK. Primary adenocar- portant structures in close juxtoposition with the duoder of the duodenum: analysis of survival. World J Surg 1988; nal mucosa, and 86% of your study group had disease of 12: 695-9. 7. DelcoreR, Connor CS, Thomas JH, Friesen SR, Hermreck AS. the cmbryologic mid-gut. Does this make for an invalid Significanceof tumor spread in adenocareinomaof the ampulla of comparison with the 14% rising from the foregut since there is increasing evidence that foregut tumors are more eater. Am J Surg 1989; I58: 593-8. 8. Crist DW, Sitzmann JV, Cameron JL. Improvedhospital mor- much aggressive than those in the mid-gut? bidity, mortality, and survival after the Whipple procedure. Ann Your database in some parts of the manuscript is a bit Surg 1987; 206: 358-65. confusing. If all 35 patients had invasive carcinoma as 9. Nakase A, Matsumoto Y, Uchida K, Honjo I. Surgical treat- your text suggests, what is the justification for the 22% ment of cancer of the pancreas and the periampullaryregion:cumu- segmental rate? In your text, 13 or 37% had metastases to lative results in 57 institutionsin Japan. Ann Surg 1977; 185:52-7. the liver, the omentum, the mesentary, or the peritoneum. I0. Michelassi F, Erroi F, Dawson PJ, et al. Experiencewith 647 It is unclear from your data whether the denominator is consecutivetumors of the duodenum, ampulla head of the pancreas 34 or 35. Furthermore, in the introduction to the paper, and distal common bile duet. Ann Surg 1989; 210: 544-56. you state that the past literature demonstrates 800 cases 11. Jcesting DR, Beart RW, Van Heerden JA, Weiland LH. Improving survival in adenocareinoma of the duodenum. Am J previously reported and yet in the text of the manuscript, you say 690. Finally, two further questions: (1) Are your Surg 1981; 141: 228-31. 12. Small intestine. In: Beahrs OH, Henson DE, Hutter RVP, improved results due to your improved management of Kennedy BJ, editors. Manual for staging of cancer. Philadelphia: the pancreaticoenteric anastomosis? Is your group comJB Lippincott, 1992: 69-73. posed mainly of duct-to-mucosa surgeons, or are they 13. Lowell JA, Rossi RL, Munson L, Braaseh JW. Primary adeno- stuffers of the pancreas (i.e., simple end-to-end anastocarcinoma of third and fourth portions of duodenum; Favorable moses)? (2) Are there other reasons to justify the aggresprognosis after resection. Arch Surg 1992; 127: 557-60. sive role of pancreatoduodenal resection for palliative 14. Kazerooni EA, Quint LE, Francis IR. Duodenal neoplasms: predictive value of CT for determining malignancy and tumor therapy? Frank Johnson (St. Louis, MO): It is difficult to resectability. AIR Am J Roentgcnol 1992; 159: 303-9. 15. Schnur PL, David E, BrownPW, et al. Adenocarcinomaof the separate duodenal carcinoma, pancreas carcinoma, and duodenumand the Gardner syndrome.JAMA 1973;223:1229-32. ampullary carcinoma from one another. Could you spec16. Iida M, Yao T, Itoh H, et al. Natural history of duodenal ulate on whether the diagnostic criteria you used to inlesions in Japanese patients with familial adenomatosiscoli (Gard- clude patients in your study might somehow have affectner's syndrome). Gastroenterology 1989; 96: 1301-6. ed the conclusions that you reached? That is to say, might 17. Ryan DP, Sehapiro RH, Warshaw AL. Villous tumors of the a few patients with duodenal carcinoma have been classiduodenum. Ann Surg 1986; 203: 301-6. fied as having pancreas cancer or ampullary carcinoma 18. Galandink S, Hermann RE, Jagelman DG, Fazio VW, Sivak and might this have affected your results? Were there MV. Villoustumors of the duodenum.Ann Surg 1988;207: 234-9. 19. Chappuis CW, Divincenti FC, Cohn I. Villous tumors of the any patients with duodenal wall gastrinoma or other unusual histologies that might have accounted for the reladuodenum. Ann Surg 1989; 209: 593-9. 2 0 . Delcore R, Thomas JH, Pierce GE, Hermreek AS. Panereato- tively good results? Romano Delcore (closing): We believe that the reagastrostomy: a safe drainage procedure after panereatoduodenectomy. Surgery 1990; 108: 641-7. son for the much better operative morbidity and mortaliincluding our own, where a small number of surgeons perform a large number of pancreatoduodenectomies in less time and with less blood loss [8,20]. Perhaps the most significant finding in this study is the absence of operative mortality and the sharply reduced operative morbidity (10%) that has occurred since 1975. During the last 15 years, all but 4 of 22 patients with PDCs were aggressively treated by curative pancreatoduodenectomy, resulting in a markedly improved 5-year survival rate of 62%. This survival is comparable with that of patients with primary ampullary carcinoma, which has been considered the most curable of all upper gastrointestinal malignancies.
630
THE AMERICAN JOURNAL OF SURGERY VOLUME 166 DECEMBER 1993
PRIMARY DUODENALCARCINOMA
ty in the recent group of patients is our use of pancreatogastrostomy as opposed to pancreatojejunostomy. As previously reported, we have found pancreatogastrostomy to be much safer than conventional pancreatojejunostomy ([20] in the original paper). It is occasionally very difficult to differentiate bet w i n the four different histologic typos of l~riampullary carcinomas. In this study, it was fairly simple because 4 years ago we presented to this Congress our experience with primary ampuUary carcinomas ([7] in the original paper). Therefore, we had already evaluated the exclusion criteria and separated primary ampullary, duodenal, common bile duct, and pancreatic neoplasms. Specifically, whenever the pathologist was unable to determine the epithelium Of origin with certainty, the tumor histologic origin was classified as unknown and was excluded from
the study. In this study, only carcinomas that unequivocally originated from duodenal mucosal epithelium were considered primary duodenal carcinomas. In fact, it has probably been previous inclusion of duodenal primary tumors in scrips of pancreatic carcinomas that has fictitiously provided pancreatic carcinoma with a better survival than it actually has, rather than the opposite being true. To address the question about duodenal gastrinomas, we previously reported to this Congress our experience with 15 primary duodenal wall gastrinomas (Delcore R, Cheung LY, Friesen SR. Characteristics of duodenal wall gastrinomas. Am J Surg 1990; I60: 621-4). These tumors were also excluded from the present study. Finally, survival rates were calculated based on 34 patients, since we excluded 1 patient who died during the preoperative evaluation.
THE AMERICAN JOURNAL OF SURGERY vOLUME 166 DECEMBER 1993 631