Adenosquamous Carcinoma of the Pancreas

Hypothesis  Adenosquamous carcinoma of the pancreas is a rare but particularly virulent variant of invasive ductal carcinoma. This review will demonstrate the aggressive biologic activity, histopathologic features, and DNA flow cytometric characteristics of this aggressive lesion. In addition, the outcome is less favorable than in other pancreatic neoplasms, in spite of aggressive surgical and postoperative adjuvant therapy.

Design  A retrospective review of 6 patients treated during an 8-year period.

Setting  A major urban university tertiary referral hospital.

Patients  There were 6 patients with this unusual tumor seen between 1990 and 1998. There were 4 men and 2 women, all white, with a mean ± SD age of 63.5 ± 14.7 years. Symptoms were similar to those in patients with more common pancreatic malignant neoplasms.

Results  Four patients with tumors in the head of the pancreas had pancreatoduodenectomy, and 2 with body and or tail lesions had distal pancreatectomy and splenectomy. Pathologically, all the tumors were poorly differentiated and aneuploid, and 5 of the 6 were locally metastatic. All but 1 patient had postoperative complications, but there were no operative deaths. One half of the patients received postoperative adjuvant chemotherapy and radiation therapy. Only 1 patient is still alive at 9 months after surgery, but has known residual cancer around his portal vein noted during palliative distal pancreatectomy.

Conclusions  Adenosquamous carcinoma of the pancreas is an uncommon variant of exocrine pancreatic neoplasm. It is characterized by an admixture of adenomatous and squamous cell elements and demonstrates aggressive biologic behavior. This series of 6 patients is similar to the 134 cases reported since 1907, in that survival is short despite aggressive surgical therapy. Few patients with this disease live more than 1 year. Aggressive therapy should be tempered by the realization of the uniform poor prognosis associated with this malignant neoplasm.

INVASIVE DUCTAL carcinoma accounts for the majority of pancreatic malignant neoplasms and has a poor prognosis with the exception of a series of highly selected patients undergoing radical surgical therapy for cure. Adenosquamous carcinoma of the pancreas is an unusual variant of pancreatic neoplasm. It has been variously referred to as adenoacanthoma, mixed squamous and adenocarcinoma, and mucoepidermoid carcinoma. These tumors are histologically characterized by adenomatous cell populations mixed with varying amounts of keratinized squamous cell elements. Major institutional reviews of autopsy and/or surgical specimens suggest an incidence of approximately 4% of all pancreatic neoplasms; however, only 134 cases have been reported in the accessible world's literature. Most of the reports have been small series or single case reports, and, of the large autopsy or surgical tissue reviews, only a few reports detail clinical, histopathologic, and patient outcome data. Survival times have been short in those patients found to have unresectable tumors, as well as in those who have undergone aggressive resection for attempted cure. The prognosis of this uncommon lesion appears to be even less favorable than the invasive ductal tumor, with only a few patients surviving more than 1 year. Because of its rarity, only anecdotal studies of adjunctive radiation or chemotherapy are available.

This series of 6 cases over an 8-year period demonstrates the aggressiveness of this tumor, and its rapid course from discovery to death, despite aggressive surgical treatment.

Between 1990 and 1998, six patients were diagnosed as having adenosquamous carcinoma of the pancreas and treated at the Indiana University Medical Center Hospitals, Indianapolis. There were 4 men and 2 women, all white, whose mean ± SD age was 63.5 ± 14.7 years. The duration of symptoms prior to presentation was 4.3 ± 3.9 months, and symptoms included abdominal pain and weight loss in 5 of the 6 patients, nausea and vomiting in 3 patients, and anorexia and jaundice in 2 patients. A single patient with a large tumor invading the duodenum presented with upper gastrointestinal tract bleeding and anemia. Physical findings included upper abdominal tenderness in 4 patients, jaundice in 2 patients with tumors in the head of the pancreas, and a palpable abdominal mass in 1 patient. Laboratory studies were unremarkable with the exception of elevated bilirubin and alkaline phosphatase levels in the 2 patients with jaundice and moderately elevated serum carbohydrate antigen 19-9 (Ca 19-9) of 200 U/mL (reference value, ≤70 U/mL) in 1 patient. Computed tomography and endoscopic ultrasound both accurately demonstrated and localized a pancreatic mass in all 6 patients (Figure 1). Endoscopic retrograde cholangiopancreatography was performed in 5 patients and demonstrated pancreatic ductal obstruction by tumor in the head or body, which corresponded with the tumor location on the imaging studies (Figure 2). Three patients had preoperative fine-needle aspiration biopsy of the tumor mass but none of the specimens were diagnostic of adenosquamous carcinoma. All 3 of the patients with tumors in the head of the pancreas underwent pancreatoduodenectomy, and those with tumors in the body and/or tail of the pancreas were treated by distal pancreatectomy and splenectomy.

There were no postoperative deaths, but 5 of the 6 patients had postoperative complications, including 3 with ventilator dependence and 2 with pancreatic fistulas that responded to nonoperative management. Postoperatively, survival was short, with a mean ± SD survival of 5.04 ± 3.58 months. Three of the patients received adjuvant chemoradiation therapy, but this did not result in significant prolongation of survival. One of the patients who had previously undergone pulmonary resection for lung carcinoma committed suicide at 4 months postoperatively. Four patients died of their malignancy at 1, 3, 5, and 12 months postoperatively. One patient with known residual tumor surrounding the portal vein is alive at 8 months and has undergone radiation therapy along with gemcitabine therapy.

Four tumors were located in the head of the pancreas and 2 were in the body and tail. Tumor size ranged from 1.2 to 6.5 cm, with a mean size of 4.2 cm. Grossly, the lesions were firm with a light tan to yellowish color and merged imperceptibly with the surrounding pancreatic parenchyma. Histologically, the tumors all displayed a mixture of ductal adenocarcinoma and squamous cell carcinoma, with the latter comprising more than 30% of the lesion. In a single case, the squamous cell component exceeded the glandular component by a 9:1 ratio. The adenocarcinoma consisted of ductlike structures lined by columnar cells having large vesicular nuclei and prominent nucleoli. Most of these cells had pale to clear cytoplasm with occasional mucin vacuoles. The malignant squamous cells were large and polygonal, having large hyperchromatic nuclei and eosinophilic cytoplasm with intercellular bridges and occasional squamous pearls. They were arranged in diffuse sheets and lobules. These squamous and glandular components were for the most part intimately admixed (Figure 3). The tumors elicited a desmoplastic response, which accounted for the firmness appreciated on gross examination.

Peripancreatic fat and neural invasion was present in all 6 cases. Lymphatic invasion was noted in 4 cases and the duodenal wall was invaded by tumor in 3 of the 4 tumors located in the head of the pancreas. Lymph node metastases were found in 5 of the 6 patients. DNA flow cytometry demonstrated aneuploidy in 5 patients with S-phase determination between 5.2% and 30.0% (Table 1).

Exocrine tumors of the pancreas have been classified histologically in several large institutional reviews.^1-4 The majority of cases are recognized as invasive ductal carcinomas. Other recognized variants include the mucinous tumors; pleomorphic, anaplastic, and large cell types; acinar cell carcinoma; spindle cell tumors; microadenocarcinomas; and oncocytic cancers.⁵ Adenosquamous and squamous cell carcinomas are recognized much less frequently and probably account for 1% to 4% of all reported tumors.

The first known report of adenosquamous carcinoma in the literature is credited to Herxheimer⁶ in 1907 in which he referred to this lesion as "cancroide." Subsequently, other authors have referred to this tumor of mixed columnar adenocarcinoma and keratin-containing squamous cell elements as mixed squamous and adenocarcinoma, mucoepidermoid carcinoma, and adenoacanthoma. This admixed tumor has been seen more commonly in other organ systems, such as the lungs, esophagus, colon, stomach, salivary glands, and the female reproductive organs. Several reports describe pancreatic tumors with a more unicellular squamous appearance without the glandular component.^7-9 Most pure squamous lesions of the pancreas are thought to represent metastatic disease, and it is suggested that in the absence of another primary lesion the squamous pancreatic lesion will usually be found to have a glandular component with more careful histological inspection.¹⁰ Since the pancreas does not normally contain any squamous cell elements, it has been proposed that squamous metaplasia occurs as a result of ductal inflammation due to chronic pancreatitis or obstruction by an adenomatous tumor. Currently, there are several theories, none well proven, of the origin of the adenosquamous tumor. The most favored theory is that of squamous metaplasia occurring as a result of obstruction and inflammation and differentiating into a malignant form.¹¹ The "collision theory" suggests that 2 histologically different tumors arise independently and join or coalesce. No reports have proven this transition phenomenon, but rather have demonstrated squamous cell components intermingled in the body of the predominantly adenocarcinoma. Electron microscopy has demonstrated 2 clearly different cell types. The adenocarcinoma cell demonstrates abundant endoplasmic reticulum and well-developed Golgi apparatus and secretory vesicles. The other cell has scant amounts of endoplasmic reticulum but prominent bundles of tonofilaments, similar to cells seen in other squamous cell tumors.¹⁰ A final theory is that of malignant differentiation by a pluripotential duct cell into the 2 histological types.¹² This theory has received support by an immunocytochemical study demonstrating positive stains for Ca 19-9, ST 439, and keratin in varying degrees in both the squamous and adenomatous cells.¹³ There is a report demonstrating carcinoma in situ in squamous metaplasia.¹⁴ In addition, there is a single report of an adenosquamous tumor focus in an equally unusual mucinous cystic neoplasm.¹⁵

The incidence of this unusual lesion is really not known, but estimates from review and classification of pancreatic tumors from autopsy and/or surgically resected specimens have been demonstrated to be approximately 4% (Table 2). Of the reported series, the incidence ranges from 0.9% to 11.1%, but this latter estimate was from a series of 27 surgical specimens.¹⁷ A number of reports suggest that many of these lesions are large and inoperable; therefore, analysis limited to surgical cases may not be totally indicative of the true incidence since many of these patients have not undergone subsequent surgery or autopsy.

A comprehensive review of 31 available reports between 1907 and 1997 identified 134 patients with adenosquamous tumors^1-36 (Table 3). From those reports with detailed clinical data, there were 69 men and 45 women, a ratio of 1.53:1. The mean age of these patients was 62.4 ± 11.7 years, with a range of 33 to 86 years. The major symptoms were no different than those of the usual invasive ductal lesions of the pancreas, and included abdominal and back pain, weight loss, anorexia, and jaundice in patients with lesions in the head of the pancreas. Tumor location was in the head of the pancreas in 31 patients, head and body in 4, body in 7, body and tail in 6, and tail in 8. It was diffuse in 4.

Whether or not surgical resection was attempted, survival was short, averaging 5.7 ± 4.1 months in 72 patients for whom survival data were available. Few studies report tumor marker data or flow cytometry results, but the present series provides evidence that these lesions are poorly differentiated, metastatic, and have aneuploid DNA characteristics. Because of the small numbers of reported cases, only anecdotal reports of adjunctive therapy are available and none of the therapies seem to prolong survival in this aggressive lesion.²⁰

Accurate preoperative diagnosis of this lesion is difficult. There are no characteristics on imaging studies to differentiate it from the more common exocrine neoplasms of the pancreas. A recent report by Kuji et al²¹ suggests that these lesions may selectively take up gallium 67 and be visualized by nuclear scanning. In several recent case reports, preoperative percutaneous and intraoperative cytologic studies have been diagnostically correct, but this did not alter treatment decisions or survival.^22-25

Once these lesions are identified, with or without accurate preoperative or intraoperative cytologic diagnosis, the question arises as to the extent of treatment. Since survival appears to be uniformly poor, should aggressive resection be undertaken, even if it appears potentially curative at the time of operation? From the review of the literature, there are only 5 known patients who have survived 1 year or longer, in contrast to reported series of highly selected patients with invasive ductal carcinomas in whom 5-year survival rates of 20% to 30% have been reported. In addition, mean survival with or without resection is less than 6 months in the present series, as well as those accumulated cases in the literature. One has to consider the risk and complications of major resective surgery in light of this rather short postoperative survival. If these lesions are small and there is no intraoperative evidence of nodal or distant metastasis, should these patients be denied an attempt at curative resection? Since pancreatoduodenectomy and distal pancreatectomy can currently be done with acceptably low morbidity and mortality, these procedures are probably the best palliative procedures available. In the near future, more effective chemotherapy combined with radiation may allow better survival with an acceptable quality of life.

Presented at the 106th Scientific Session of the Western Surgical Association, Indianapolis, Ind, November 18, 1998.

Reprints: James A. Madura, MD, Department of Surgery, Indiana University School of Medicine, 545 Barnhill Dr, EM 244, Indianapolis, IN 46202-5125.

Gerard V. Aranha, MD, Maywood, Ill: Dr Madura has presented us with 6 cases of adenosquamous carcinoma of the pancreas, a very rare variant of the usual adenocarcinoma. There were 4 patients with head lesions and 2 with lesions in the body; all were resected, but none survived more than 12 months, suggesting also a poorer prognosis than patients with adenocarcinoma. The pathology evaluation revealed that 5 of the 6 patients had peripancreatic or perineural invasion as well as lymph node metastasis and that 5 of the 6 had aneuploid tumors, all harbingers of poor prognosis. As far as theories go, there are many, but the one that makes the most sense is the one that you first alluded to—the one that says that there may be squamous metaplasia in an adenocarcinoma that leads to 2 pathologic entities in one tumor.

In your lymph node metastasis, which of the elements metastasized, was it the glandular type or the squamous type? Can you conjecture why the prognosis was so bad in these patients? Is it possible that squamous metaplasia occurs more often in adenocarcinoma and that we do not find it because we do not search enough for it? The incidence of this tumor you said was 4%, but in a Japanese study, it was 11%. If this variant occurs more, should we be reconsidering our adjunctive therapy? Our therapy now is geared toward adenocarcinoma. Should we be changing the chemotherapy or combining it with something geared toward squamous, let's say, platinum-based therapy?

You talked about ERCP (endoscopic retrograde cholangiopancreatography). If you had these patients first and you saw a CT scan clearly showing you a mass, would you put them through an ERCP?

Finally, one other question about biopsy. Because you saw 6 cases of this variant with bad prognosis, are you going to change your philosophy? I suspect you are like me and do not do a biopsy preoperatively or intraoperatively. Would you change your philosophy based on these 6 patients?

Dr Madura: Dr Aranha, your first question was about the cellular pattern of the lymph node metastases. We reviewed all of the lymph nodes in the 5 patients with metastases. Three of these patients had mixed adenocarcinoma and squamous cell carcinoma in the nodes, while 2 patients, interestingly, had only squamous cell elements. We also evaluated the pancreatic lesions for total content of each of the 2 specific cell types and found that 5 of the patients had 30% squamous cell cancer in their tumor, while 1 patient had more than 90% squamous cells in his tumor. You asked whether these patients should be treated with a chemotherapeutic regimen more suited for the squamous cell element. Unfortunately, these lesions are very unusual, and there are no series of patients treated with such an approach. Our sole survivor at this point has known residual disease but has survived nearly 9 months and he has been treated with gemcitabine and irradiation. I am not sure of this agent's efficacy with squamous cell tumors, but it does have some advantage in adenocarcinomas. I think your concept may be correct and in the future, we might change our adjuvant therapeutic approach.

You asked if we had diagnosed any of these lesions ahead of time, and, with the exception of those patients arriving with previous biopsy, we do not attempt a preoperative aspiration nor do we do any tissue sampling in the operating room. If the lesion looks amenable to removal on preoperative assessment and in the operating room, we go ahead with the resection. One of these patients gave us no choice since her tumor had eroded into the duodenum and was bleeding. I do not think that a diagnosis of adenosquamous carcinoma preoperatively would change our approach. It would give us an opportunity to inform the patient and their family about the prognosis, and perhaps allow us to stimulate our oncologic colleagues to consider a different postoperative regimen.