Primary pancreatic lymphoma

Primary pancreatic lymphoma

Primary pancreatic lymphoma Michael Bouvet, MD, Gregg A. Staerkel, MD, Francis R. Spitz, MD, Steven A. Curley, MD, Chusilp Charnsangavej, MD, Fredrick...

117KB Sizes 0 Downloads 3 Views

Primary pancreatic lymphoma Michael Bouvet, MD, Gregg A. Staerkel, MD, Francis R. Spitz, MD, Steven A. Curley, MD, Chusilp Charnsangavej, MD, Fredrick B. Hagemeister, MD, Nora A. Janjan, MD, Peter W. T. Pisters, MD, and Douglas B. Evans, MD, Houston, Texas

Background. Primary pancreatic lymphoma is a rare neoplasm that may be confused with pancreatic adenocarcinoma. We reviewed retrospectively our contemporary experience with this disease to define more clearly the clinical presentation of this disease and the proper role for percutaneous fine-needle aspiration biopsy and surgery. Methods. From 1980 to 1995, 11 patients with primary pancreatic lymphoma were treated at The University of Texas M. D. Anderson Cancer Center. Patient demographics, radiographic studies, fineneedle aspiration biopsy findings, operative procedures, and other treatment data were reviewed. Results. The median age of the 11 patients was 64 years (range, 37 to 74 years). Abdominal pain was the most common symptom at presentation. Five patients had an elevated lactate dehydrogenase level, and only two patients had hyperbilirubinemia. Computed tomography scan demonstrated encasement of the superior mesenteric artery or superior mesenteric–portal vein confluence in six patients. Seven patients underwent computed tomography–guided fine-needle aspiration; five had findings of lymphoma. Two patients underwent distal pancreatectomy and splenectomy, and one underwent pancreaticoduodenectomy. All patients were treated with combination chemotherapy, and seven received radiotherapy. Only two patients have died of disease (12 and 16 months after diagnosis) at a median follow-up time of 67 months. Conclusions. In the majority of patients, pancreatic lymphoma can be distinguished from pancreatic adenocarcinoma on the basis of symptoms, laboratory and radiographic findings, and fine-needle aspiration biopsy results. Once the diagnosis is established, all patients should undergo systemic chemotherapy followed by involved-field radiotherapy if the tumor has not been resected. (Surgery 1998;123:382-90.) From the Departments of Surgical Oncology, Anatomic Pathology, Diagnostic Radiology, Hematology, and Radiation Oncology, University of Texas, Houston, Texas

IN CONTRAST TO HODGKIN’S DISEASE, in which spread outside the lymphatic system is unusual, nonHodgkin’s lymphoma frequently involves extranodal sites. The gastrointestinal tract, particularly the stomach and small bowel, is the most common extranodal site. The pancreas and adjacent organs may be involved, usually in the setting of massive retroperitoneal and periaortic lymphadenopathy.1 However, isolated primary pancreatic lymphoma is rare, with less than 1% of extranodal nonHodgkin’s lymphomas arising in the pancreas.2 In a review of 207 cases of malignant tumors of the pancreas, adenocarcinoma of the pancreas was by far the most common pathologic subtype, with only three (1.5%) cases of pancreatic lymphoma.3

Accepted for publication Sept. 3, 1997. Reprint requests: Douglas B. Evans, MD, Department of Surgical Oncology, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Box 106, Houston, TX 77030. Copyright © 1998 by Mosby, Inc. 0039-6060/98/$5.00 + 0 11/56/86538

382 SURGERY

Only 33 patients with this disease have been reported from three large referral centers.4-6 Accurate diagnostic imaging is essential in the evaluation of a patient with a presumed neoplasm of the pancreas. Contrast-enhanced helical computed tomography (CT) performed with thin-section technique can accurately assess the relationship of the low-density tumor to the celiac axis, superior mesenteric artery (SMA), and superior mesenteric–portal vein (SMPV) confluence.7 A patient is deemed to have CT evidence of locally advanced, unresectable disease when there is loss of the normal fat plane between the tumor and the SMA or celiac axis, or occlusion of the SMPV confluence. CT-guided percutaneous fine-needle aspiration (FNA) is the diagnostic procedure of choice for establishing a cytologic diagnosis in patients with locally advanced or metastatic pancreatic cancer and is also commonly used in the diagnosis and classification of lymphoma.8,9 By using such an approach for the evaluation of a pancreatic mass, it may be possible to make the diagnosis of primary pancreatic lymphoma, thereby eliminating

Bouvet et al. 383

Surgery Volume 123, Number 4 unnecessary laparotomy. We reviewed our experience with pancreatic lymphoma to define more clearly the clinical presentation of this disease and the proper role for percutaneous biopsy and surgery. MATERIAL AND METHODS A computerized search using the medical informatics and lymphoma databases at The University of Texas M. D. Anderson Cancer Center was performed. A total of 11 patients with the diagnosis of primary pancreatic lymphoma treated between 1980 and 1995 were identified. Patients with pancreatic lymphoma who also had superficial lymphadenopathy, mediastinal lymphadenopathy, or other extranodal disease were excluded. Patient demographic, diagnostic, and treatment information was recorded, including signs and symptoms at presentation, physical examination findings, laboratory abnormalities, radiographic findings, operative procedures, and outcome. During the study period most CT scans were performed with contrast enhancement and thinsection technique by using either the 9800 HighLight HTD scanner or the 9800 High-Light Advantage scanner (GE Medical Systems, Milwaukee, Wis.). However, during the last 4 years, scans of the pancreas were performed on a helical CT unit (9800 High Speed Advantage; GE Medical Systems). Patients received 240 to 280 ml of 2% barium sulfate suspension (readi-Cat; E-Z-EM Co. Inc., Westbury, N.Y.) or water to opacify the stomach and small bowel. Non-contrast-enhanced scans were obtained through the liver and pancreas at 10 mm slice thickness and 10 mm scan interval to localize the pancreas. Localization of the pancreas allowed the entire gland to be included in the thinsection portion of the evaluation. Intravenous contrast enhancement was achieved with nonionic contrast material administered by an automatic injector at a rate of 1.5 to 2 ml/sec for a total of 150 ml. Fifty seconds after intravenous iohexol administration was started, a dynamic series of scans through the pancreas was performed at 1.5 or 3 mm slice thickness and 5 mm scan interval. After completion of the bolus phase, the remainder of the abdomen was evaluated at 5 or 10 mm slice thickness and 8 or 10 mm scan interval. When helical CT was used, the scan of the pancreas was done at 3 mm slice thickness at a pitch factor of 1.5 or 2 with a 60-second scan delay after the start of intravenous contrast material. CT criteria for resectability included the following: (1) the absence of extrapancreatic disease, (2) a patent SMPV confluence, and (3) no direct tumor extension to the celiac axis

Table I. Symptoms at presentation in 11 patients with pancreatic lymphoma Symptom Abdominal pain Weight loss Back pain Jaundice Fever/chills Ascites

No. of patients (%) 9 (82) 4 (36) 3 (27) 2 (18) 2 (18) 1 (9)

or SMA. All available CT scans were reviewed by a single diagnostic radiologist. Percutaneous FNA of the pancreas was performed in an interventional radiology suite under CT guidance by using a 22-gauge needle placed through the anterior abdominal wall (the posterior approach is now used for most pancreatic head tumors). Cytologic smears were prepared on glass slides after each needle-aspiration procedure (average number of passes was three). Slides were then stained by using both the Diff-Quik and Papanicolaou methods. Residual tissue within the needle and syringe was rinsed in RPMI-1640 tissue culture medium. All stained slides were immediately assessed by a cytopathologist; if lymphoma was considered, a portion of the RPMI rinse solution was allocated for immunophenotyping (immunocytochemical staining) to confirm lymphoma. Immunophenotyping on FNA specimens was performed on air-dried cytospin preparations fixed in absolute acetone at 4° C for 10 minutes. Tissues for these cytospin preparations were obtained from the RPMI rinse solution after separation of the white cells by using density-gradient methodology. Immunostaining for immunoglobulin light chains was detected with a direct peroxidase procedure using goat antihuman peroxidase labeled κ and λ light chains at 1:25 dilution (Tago Corp., Burlingame, Calif.). For detection of other surface marker antigens, the avidin-biotin-peroxidase complex method was used with monoclonal antibodies to Leu-1 (CD5), Leu-4 (CD3), Leu-12 (CD19), and Ki-1 (CD30). As part of this study, all cytologic and histologic specimens were re-reviewed by a single pathologist. For complete staging, all patients underwent bone marrow aspiration, chest radiography, and/or chest CT. The tumors were classified by lymphoma type according to the Working Formulation, and patients were staged according to the Ann Arbor staging system for non-Hodgkin’s lymphomas10,11: stage IE disease was confined to the pancreas and stage IIE disease involved the pancreas and regional nodes. The International

384 Bouvet et al.

Surgery April 1998

Table II. CT findings in 11 patients with pancreatic lymphoma Patient 1 2 3 4 5 6 7 8 9 10 11

Location of tumor

Size (cm)

Tail 7 Head 4.5 Head 5 Body and tail 10 Body 5 Body and tail N/A Head and uncinate 8 Head 7 Head 5 Tail 3 Tail 13

Density Not evaluable Hypodense Hypodense Hypodense Hypodense N/A Hypodense Hypodense Hypodense Mixed Mixed

SMA encasement

SMPV confluence

Not evaluable Yes Yes Yes Yes N/A Yes Yes No No No

Not evaluable Not evaluable Not evaluable Patent Occluded N/A Occluded Occluded Patent Patent Patent

Resectable* Not evaluable No No No No N/A No No Yes Yes Yes

SMA, superior mesenteric artery; SMPV, superior mesenteric portal vein; not evaluable, CT scan performed early in series and not of sufficient quality to assess vascular involvement; N/A, CT scan not available for review. *No extrapancreatic disease, no encasement of the celiac axis or SMA, and a patent SMPV confluence.

Index score was used to assess the prognosis of the patient.12 Follow-up was obtained by review of the patient’s medical records or by correspondence with the patient, relatives, or personal physician when necessary. Complete remission was defined as complete disappearance of disease as determined by clinical, radiographic, and laboratory evaluation. Survival was measured from the date of histologic or cytologic diagnosis to the date of death and recorded in months. RESULTS Patient symptoms, laboratory results, and radiographic findings. The median age of the 11 patients was 64 years (range, 37 to 74 years). There were six men and five women. The most common symptom at presentation was abdominal pain, which, in contrast to pain from adenocarcinoma of the pancreas, uncommonly radiated to the back (Table I). Two patients presented with jaundice and one with ascites. Physical examination findings were nonspecific in most cases, although an abdominal mass or fullness was palpable in three patients. Abnormal laboratory values included an elevated alkaline phosphatase level (>126 IU/L) in seven patients, an elevated lactate dehydrogenase level (>618 IU/L) in five patients, and an elevated total bilirubin level (>1.0 mg/100 ml) in only two patients. Both patients with hyperbilirubinemia had tumors of the pancreatic head causing mechanical obstruction of the intrapancreatic portion of the common bile duct. Five of the 11 tumors were located in the head of the pancreas, three in the body, and three in the tail. Ten patients underwent pretreatment abdominal CT; nine studies were of adequate technical

quality to allow analysis of tumor characteristics (Table II). In the one remaining patient, the pancreatic mass was discovered at the time of cholecystectomy (Fig. 1). Seven (78%) of the nine evaluable tumors were hypodense. Six of the nine CT studies demonstrated unresectable tumors as evidenced by encasement of the SMA (six) and occlusion of the SMPV confluence (three). Three tumors were considered “radiographically resectable” as defined by the absence of arterial encasement and a patent SMPV confluence; two were in the pancreatic tail, and one was in the pancreatic head. Only one of 11 patients presented with a potentially resectable, hypodense mass in the pancreatic head. This patient also had extrahepatic biliary obstruction. The other two patients with radiographically resectable disease had lesions confined to the pancreatic tail without visible metastatic disease—a rare finding in patients with pancreatic adenocarcinoma. Cytologic and histologic findings and staging information. Seven of the 11 patients underwent FNA biopsy (Table III). Of these, five specimens were diagnostic of lymphoma, one was diagnostic of malignancy but inconclusive for the type of tumor (original report’s diagnosis was “rare malignant cells consistent with carcinoma”), and one was a nondiagnostic specimen, consisting of benign pancreatic tissue only (sampling error). The sensitivity of FNA biopsy in this setting was 71% (five of seven). Two of the five patients with diagnostic FNA specimens underwent surgical biopsy. These surgical biopsies were done because FNA biopsy was performed early in the series, before complete familiarity with this method of diagnosis (one patient), and because of an inability to morphologically sub-

Surgery Volume 123, Number 4

Bouvet et al. 385

Fig. 1. Abdominal CT scan of patient who was found to have pancreatic lymphoma at time of cholecystectomy. (A) Before treatment. (B) Ten years after diagnosis, patient has no evidence of disease.

classify the lymphoma as to small cleaved cell versus large cell type (one patient). Four of the 11 patients did not have FNA biopsy. One of these four had a core biopsy, one had an open biopsy at the time of cholecystectomy, one had an open biopsy at an outside institution before referral, and one refused FNA biopsy. Ten patients had large cell lymphoma and one had follicular mixed cell lymphoma. Four patients had Ann Arbor stage IE disease and seven had Ann Arbor stage IIE disease. The International Index score was favorable for the group of patients, ranging from 0 to 2. These results are summarized in Table III. Treatment and follow-up. Laparotomy was performed in 10 patients, either as part of the initial diagnostic evaluation or as part of restaging. Three patients underwent resection of all gross disease; two by distal pancreatectomy and one by pancreaticoduodenectomy with en bloc resection of the SMPV confluence and reconstruction of the con-

fluence with an internal jugular vein interposition graft.13,14 These three patients all had radiographically resectable disease and FNA was either not done or did not suggest lymphoma (patients 9 to 11, Tables III and IV). Surgery in the remaining seven patients consisted of incisional biopsy. All patients received four to six cycles of combination chemotherapy, usually consisting of cyclophosphamide (750 mg/m2 intravenously on day 1), doxorubicin (50 mg/m2 intravenously on day 1), vincristine (2 mg intravenously on day 1), prednisone (100 mg/day orally on days 1 through 5), and bleomycin (15 mg intravenously on day 1). Specific variations in chemotherapeutic regimens are noted in Table IV. Three patients received chemotherapy after surgical resection, and eight patients received chemotherapy without surgical resection. Five of the eight unresected patients had complete responses to chemotherapy; two of the eight had minimal response to chemotherapy and have died of disease at 12 and 16 months after diagnosis. One

386 Bouvet et al.

Surgery April 1998

Table III. Comparison of age, gender, FNA findings, cell type, and stage in 11 patients with pancreatic lymphoma Patient 1 2 3 4 5 6 7 8 9 10 11

Age (yr)

Gender

52 74 72 66 70 64 58 44 53 72 37

F M M F M M M F M F F

FNA cytology

Cell type

None‡ Diagnostic None Diagnostic Diagnostic None Diagnostic Diagnostic Nondiagnostic Suggestive of carcinoma None

Diffuse, large cell Diffuse, large cell Follicular, mixed cell Diffuse, large cell Diffuse, large cell Diffuse, large cell Diffuse, large cell Diffuse, large cell Diffuse, large cell Diffuse, large cell Diffuse, large cell

Stage*

International Index score†

IIEB IIEA IEA IIEB IIEA IIEA IIEA IEA IEA IEA IIEA

1 1 1 2 2 1 2 0 1 2 0

*Ann Arbor staging system abbreviations: E, extranodal; A, absence or B, presence of unexplained weight loss > 10% body weight, unexplained fever > 38° C, or night sweats. †International

Index score, 1 point for each of the following: age > 60 years, Ann Arbor stage III or higher, two or more extranodal sites, two or higher Zubrod performance status, elevated lactate dehydrogenase level; total possible score 0 to 5.

‡This

patient had a percutaneous core biopsy specimen diagnostic of lymphoma.

Table IV. Comparison of surgical treatment, chemotherapy, radiotherapy, disease status, and survival in 11 patients with pancreatic lymphoma Patient

Surgical procedure

1 2

None Biopsy (to confirm FNA)

3 4

Biopsy (at time of cholecystectomy) Restaging laparotomy, splenectomy 1 yr after diagnosis Gastrojejunostomy

5 6 7 8

9 10 11

Biopsy Biopsy (for lymphoma subclassification) Restaging laparotomy 2 yr after diagnosis Pancreaticoduodenectomy Distal pancreatectomy, splenectomy Distal pancreatectomy, splenectomy

Radiation dose (Gy)

Chemotherapy IM-etop, HOP CHOP, OAP-bleo, MIME CHOP-bleo CHOP-bleo CHOP-bleo, CMED, ESHAP CHOP ESHAP, BACOS, MINE CHOP-bleo, MINE, ESHAP, Autologous BMT CHOP, OPEN CHOP, OPEN CHOP-bleo, OPEN

Outcome

Survival (mo)

None 44

NED Dead

191 16

45 44

159 80

30

NED Died of other causes Dead

40 40

NED NED

67 55

40

NED

44

None None None

NED NED NED

23 13 11

12

BACOS, bleomycin, doxorubicin, cyclophosphamide, vincristine, methylprednisolone; BMT, bone marrow transplant; CHOP-bleo, cyclophosphamide, doxorubicin, vincristine, prednisone, bleomycin; CMED, cyclophosphamide, methotrexate, etoposide, dexamethasone; ESHAP, etoposide, methylprednisolone, high-dose cytarabine, and cisplatin; HOP, doxorubicin, vincristine, prednisone; IM-etop, ifosfamide, methotrexate, etoposide; MIME, mesna, ifosfamide, methotrexate, and etoposide; MINE, mesna, ifosfamide, mitoxantrone, etoposide; NED, no evidence of disease; OAP-bleo, vincristine, cytarabine, prednisone, bleomycin; OPEN, vincristine, prednisone, etoposide, mitoxantrone.

patient had a partial response and underwent FNA of a residual pancreatic mass that was nondiagnostic. Therefore the patient underwent laparotomy with biopsy, which confirmed the diagnosis of residual lymphoma. She then underwent high-dose chemotherapy followed by autologous bone marrow transplant (based on these positive findings of

residual disease at restaging laparotomy) and is currently without evidence of disease. External beam irradiation was administered to seven of the eight unresected patients after the delivery of combination chemotherapy. Four patients did not receive irradiation, including the three patients who underwent pancreatectomy and

Surgery Volume 123, Number 4

Bouvet et al. 387

Fig. 2. Outcome of patients treated for pancreatic lymphoma. DOD, died of disease (lymphoma); NED, no evidence of disease.

one patient treated early in our experience with chemotherapy alone. Radiation dosages ranged from 30 to 45 Gy. A variety of beam energies were used, but high energy photons (>18 Mv) were used most commonly. In all cases, anterior/posterior fields localized to the pancreatic region were used; extension of the field to administer 25 Gy to the spleen was performed in only one patient who had a 5-year disease-free interval before experiencing a distant relapse in the neck that was controlled by radiation alone. Radiation was well tolerated and no significant late sequelae have been observed. Nine of the 11 patients had a complete response to treatment; eight of them are alive without evidence of disease and one died of an unrelated cause 80 months after diagnosis (Fig. 2). The median duration of follow-up was 67 months (range, 11 to 191 months). The surviving patients include three patients who underwent pancreatic resection followed by chemotherapy, four patients treated with chemotherapy and radiation, and one treated with chemotherapy alone. DISCUSSION CT evidence of a large pancreatic neoplasm in the absence of weight loss, back pain, and extrahepatic biliary obstruction (often with an elevated lactate dehydrogenase [LDH] level) should cause the physician to include lymphoma in the differential diagnosis of a pancreatic mass. Only two patients in our series presented with small (≤5 cm), potentially resectable mass lesions in the pancreas. One patient underwent pancreaticoduodenectomy for presumed adenocarcinoma and was found to have lymphoma on pathologic evaluation of the resected specimen (Fig. 3). A second patient underwent a distal pancreatectomy for a small (3 cm) mixed-density lesion in the pancreatic tail

believed to represent an exocrine or endocrine carcinoma. The remaining nine patients had large tumors in the pancreatic head (four), body (three), or tail (two); only one of these patients underwent resection (distal pancreatectomy). These findings are in agreement with previous reports of CT findings suggestive of pancreatic lymphoma: large homogeneous masses with diameters greater than 7 cm with or without retroperitoneal lymphadenopathy.15-17 Percutaneous CT-guided FNA biopsy of the pancreas is the technique of choice for obtaining tissue diagnosis in patients with locally advanced pancreatic tumors as defined by our objective CT criteria (no evidence of tumor extension to the celiac axis or SMA and a patent SMPV confluence). However, accurate interpretation of FNA specimens requires an experienced cytopathologist and the ability to perform immunocytochemical staining to differentiate lymphoma from nonlymphomatous neoplasms. In addition, the rare T-cell lymphoma may require gene rearrangement studies, which can also be performed on FNA specimens.18-21 Common to the five aspirates diagnostic for lymphoma in our report were numerous large atypical lymphocytes distributed singly in a background of lymphoglandular bodies, which represent small fragments of cytoplasm that have detached from lymphoid cells (Fig. 4). Their presence is helpful in indicating a lesion’s lymphoid nature, and they are not recognizable on histologic sections. Additional findings that suggest a cytologic diagnosis of lymphoma include significant nuclear atypia, as indicated by membrane irregularity, fine chromatin, nucleolar prominence, and an overall monomorphic appearance to the lymphoid cells.22,23 Inadequate sampling may hamper the cytologic diagnosis of lymphoma.24 Our one nondiagnostic

388 Bouvet et al.

Fig. 3. Histology. Lymphoma cells (right) are seen infiltrating pancreatic acinar tissue (left). Acinar gland atrophy and destruction are evident. Hematoxylin-eosin stain; original magnification ×400.

Fig. 4. Cytology. Large lymphocytes with irregular nuclear outlines, prominent nucleoli, vesicular fine chromatin, and minimal cytoplasm predominate in this smear (diagnosed as large cell lymphoma). Inconspicuous lymphoglandular bodies (arrows) are seen in the background.

Fig. 5. Histologic section shows small cleaved cell lymphoma (right) with focus of large cell transformation (left). Inadequate sampling could result in incorrect subtyping of this lymphoma. Hematoxylin-eosin stain; original magnification ×600.

Surgery April 1998 case represented a sampling problem in that the malignancy was missed (patient 9, Table III). The case diagnosed as “malignant, consistent with carcinoma” was, on re-review, considered malignant but insufficient for cytologic subclassification (patient 10, Table IV). The lack of sufficient cellularity did not permit a recognizable pattern of lymphoma or the use of ancillary studies to confirm the presence of carcinoma versus lymphoma. Critical to the interpretation of all cytologic reports by the clinician is an accurate analysis of patient signs, symptoms, and radiographic findings. If the cytologic diagnosis is at variance with the clinical findings (as in this case), re-review or repeat biopsy is indicated. Inadequate sampling can also cause difficulties when the lymphomatous process is not uniform. One of our cases considered diagnostic of lymphoma illustrates this point. Although a correct diagnosis of lymphoma was made, the subtype of lymphoma (small cleaved vs large cell) could not be determined. Appropriately, the patient underwent laparotomy for surgical biopsy. This biopsy specimen revealed small cleaved cell lymphoma with adjacent areas of large cell transformation (Fig. 5). At our institution, lymph node biopsy is commonly used to classify most lymphomas. For those patients with intraabdominal or intrathoracic masses without peripheral adenopathy who would require laparotomy or thoracotomy for surgical biopsy, percutaneous CT-guided FNA biopsy is a potential alternative for diagnosis. Most intraabdominal lymphomas are of the diffuse, large B-cell type. Therefore, if the FNA sample is adequate and a definitive diagnosis of large-cell lymphoma is made and confirmed with immunophenotyping, patients can be spared the potential morbidity of laparotomy for open biopsy. The role of surgery for pancreatic lymphoma remains controversial. Webb et al.5 reported a series of nine patients with primary pancreatic lymphoma. With the use of combination chemotherapy consisting of cyclophosphamide, doxorubicin, and prednisone, a complete remission occurred in six of nine patients. Four patients with obstructive jaundice treated initially with nonhepatotoxic agents avoided surgical biliary bypass and had rapid resolution of jaundice. The authors concluded that the only role for surgery was to aid in establishing the diagnosis when percutaneous biopsy was nondiagnostic. In contrast were the findings of Behrns et al,4 who reported on the Mayo Clinic experience with 12 patients with primary pancreatic lymphoma treated between 1952 and 1991. A single pancreati-

Bouvet et al. 389

Surgery Volume 123, Number 4 coduodenectomy was performed 1 year after a full course of chemotherapy had failed. Mean survival was 13 months for patients who received chemotherapy alone, 22 months for those treated with radiotherapy alone, and 26 months for those who received combined radiotherapy and chemotherapy. In view of the poor survival rate with radiotherapy and chemotherapy, they advised a more aggressive surgical approach including pancreatic resection or tumor debulking. In our series, surgical resection was performed in the three patients who had radiographically resectable primary pancreatic neoplasms as defined by objective CT criteria. All three patients were presumed to have pancreatic exocrine or endocrine carcinoma. Preoperative FNA was performed in two of these three patients and was nondiagnostic in one patient and incorrectly diagnosed as consistent with carcinoma in the other. Although all three patients are alive and without evidence of disease, it is likely that a similar clinical result could have been achieved with chemotherapy and irradiation without surgical pancreatectomy. A number of prognostic factors have been identified for non-Hodgkin’s lymphoma including age, presence of B symptoms, performance status, serum LDH and β-2-microglobulin levels, and stage. These factors can be summarized by the International Index score or the M. D. Anderson Tumor Score.25 Because β-2-microglobulin levels were not available for the patients treated early in this series, the International Index score was used. Patients with low-risk stages IE and IIE disease have a 5-year disease-free survival rate of 84% after treatment with four cycles of cyclophosphamide, doxorubicin, vincristine, and prednisone plus regional irradiation.26,27 All 11 patients in this report were considered low to moderate risk as defined by the International Index score. Nine of 11 patients achieved a complete clinical and radiographic remission, with follow-up greater than 3 years in six of nine patients. Primary pancreatic lymphoma is rare, making it difficult to define the optimal approach to diagnosis and treatment. Our data suggest that in the majority of patients pancreatic lymphoma can be distinguished from pancreatic adenocarcinoma on the basis of patient symptoms, laboratory and radiographic findings, and FNA biopsy results. Contrast-enhanced helical CT performed with thin-section technique can accurately assess the relationship of the primary tumor to the celiac axis, SMA, and SMPV confluence. In patients with locally advanced disease (as defined by objective CT criteria), FNA, performed via CT or endoscopic ultra-

sonography, is the technique of choice for obtaining a tissue diagnosis. If FNA results are diagnostic of large cell lymphoma, laparotomy can be avoided. FNA biopsy specimens that lack adequate cellularity for diagnostic certainty should prompt repeat FNA and, if needed, open surgical biopsy. Surgical resection is reserved for the anecdotal case that mimics primary exocrine or endocrine carcinoma and appears resectable on high-quality CT images. The primary treatment for patients in whom the diagnosis of lymphoma can be established with minimally invasive techniques is combination chemotherapy with involved-field radiotherapy. REFERENCES 1. Boddie AW, Eisenberg BD, Mullins JO, Schlichtemeier AL. The diagnosis and treatment of obstructive jaundice secondary to malignant lymphoma: a problem of multidisciplinary management. J Surg Oncol 1980;14:111-23. 2. Freeman C, Berg JW, Cutler SJ. Occurrence and prognosis of extranodal lymphomas. Cancer 1972;29:252-60. 3. Reed K, Vose PC, Jarstfer BS. Pancreatic cancer: 30 year review (1947 to 1977). Am J Surg 1979;138:929-33. 4. Behrns KE, Sarr MG, Strickler JG. Pancreatic lymphoma: is it a surgical disease? Pancreas 1994;9:662-7. 5. Webb TH, Lillemoe KD, Pitt HA, Jones RJ, Cameron JL. Pancreatic lymphoma: is surgery mandatory for diagnosis and treatment? Ann Surg 1989;209:25-30. 6. Mansour GM, Cucchiaro GC, Niotis MT, et al. Surgical management of pancreatic lymphoma. Arch Surg 1989;124:1287-9. 7. Fuhrman GM, Charnsangavej C, Abbruzzese JL, et al. Thinsection contrast enhanced computed tomography accurately predicts resectability of malignant pancreatic neoplasms. Am J Surg 1994;167:104-13. 8. Robins DB, Katz RL, Evans DB, Atkinson EN, Green EL. Fine-needle aspiration of the pancreas: in quest of accuracy. Acta Cytol 1995;39:1-10. 9. Carter TR, Feldman PS, Innes DJ, Frierson HF, Frigy AF. The role of fine needle aspiration cytology in the diagnosis of lymphoma. Acta Cytol 1988;32:848-53. 10. The Non-Hodgkin’s Lymphoma Pathologic Classification Project. National Cancer Institute sponsored study of classifications of non-Hodgkin’s lymphomas: summary and description of a working formulation for clinical usage. Cancer 1982;49:2112-35. 11. Carbone PP, Kaplan HS, Musshoff K, et al. Report of the Committee on Non-Hodgkin’s Disease Staging Classification. Cancer Res 1971;31:1860-1. 12. The International Non-Hodgkins Lymphoma Prognostic Factors Project. A predictive model for aggressive NHL. N Engl J Med 1993;329:987-94. 13. Evans DB, Abbruzzese JL, Rich TA. Cancer of the pancreas. In: DeVita VT, Hellman S, Rosenberg SA, editors. Cancer, principles and practice of oncology. 5th ed. Philadelphia: JB Lippincott Co; 1997. p 1054-87. 14. Cusack JC, Fuhrman GM, Lee JE, Evans DB. Management of unsuspected tumor invasion of the superior mesentericportal venous confluence at the time of pancreaticoduodenectomy. Am J Surg 1994;168:352-4. 15. Teefey SA, Stephens DH, Sheedy PF. CT appearance of primary pancreatic lymphoma. Gastrointest Radiol 1986;11413.

390 Bouvet et al.

16. Cappell MS, Yao F, Cho KC, Axiotis CA. Lymphoma predominantly involving the pancreas. Dig Dis Sci 1989;34:942-7. 17. Prayer L, Schurawitzki H, Mallek R. CT in pancreatic involvement of non-Hodgkin lymphoma. Acta Radiol 1992;33:123-7. 18. Robins DB, Katz RL, Swan F, Atkinson EN, Ordonez NG, Huh YO. Immunotyping of lymphoma by fine-needle aspiration: a comparative study of cytospin preparations and flow cytometry. Am J Clin Pathol 1994;101:569-76. 19. Katz RL, Hirsch-Ginsberg C, Childs C, et al. The role of gene rearrangements for antigen receptors in the diagnosis of lymphoma obtained by fine-needle aspiration: a study of 63 cases with concomitant immunophenotyping. Am J Clin Pathol 1991;96:479-90. 20. Erwin BC, Brynes RK, Chan WC, et al. Percutaneous needle biopsy in the diagnosis and classification of lymphoma. Cancer 1986;57:1074-8. 21. Hu E, Horning S, Flynn S, Brown S, Warnke R, Sklar J. Diagnosis of B-cell lymphoma by analysis of immunoglobulin gene rearrangements in biopsy specimens obtained by fine needle aspiration. J Clin Oncol 1986;4:278-83.

Surgery April 1998 22. DeMay RM. Lymph nodes. In: The art and science of cytopathology. Chicago: ASCP Press; 1996. p 779-846. 23. Pontifex AH, Klimo P. Application of aspiration biopsy cytology to lymphomas. Cancer 1984;53:553-6. 24. Ortell SR, Skinner JM. The typing of non-Hodgkin’s lymphomas using fine needle aspiration cytology. Pathology 1982;14:389-94. 25. Rodriguez J, Cabanillas F, McLaughlin P, et al. A proposal for a simple staging system for intermediate grade lymphoma and immunoblastic lymphoma based on the ‘tumor score’. Ann Oncol 1992;3:711-7. 26. Tondini C, Zanini M, Lombardi F, et al. Combined modality treatment with primary CHOP chemotherapy followed by locoregional irradiation in stage I or II histologically aggressive non-Hodgkin’s lymphomas. J Clin Oncol 1993;11:720-5. 27. Connors JM, Klimo P, Fairey RN, Voss N. Brief chemotherapy and involved field radiation therapy for limited stage histologically aggressive lymphoma. Ann Intern Med 1987;107:25-30.