Population-Based Study of Islet Cell Carcinoma

In order for us to make advances in the diagnosis and management of patients with islet cell carcinoma, we must improve our understanding of the epidemiology, natural history, and prognostic factors for this relatively rare disease. Because of the rarity of neuroendocrine carcinoma and the lack of a staging system, much of the information previously published has been based on case series and anecdotal experiences. In this study, we take advantage of the vast amount of data collected by the SEER Program to examine the largest series of islet cell carcinomas reported to date. To our knowledge, this study represents the only population-based study of islet cell carcinoma in published literature.

Prevalence is defined as the number people alive on a certain date in a population that ever had a diagnosis of the disease. In this study, the counting method15 was used to estimate prevalence from incidence and follow-up data obtained from the SEER 9 registries. Complete prevalence can be established by this method by using registries of very long duration. The SEER 9 registries have the longest follow-up duration and contain data suitable for prevalence analyses for the past 28 years. Given the longer survival duration often experienced by patients with neuroendocrine carcinoma, we report only limited duration prevalence data which may somewhat underestimate the complete prevalence. By incidence, islet cell carcinomas account for only 1.3% of all pancreatic cancers. However, because of the better outcome generally experienced by patients with islet cell carcinoma, they represent almost 10% of pancreatic cancers in prevalence analyses.

We acknowledge that analyses from the SEER registries underestimate the total number of patients with islet cell tumors. All cases from the SEER database were denoted to be malignant. Thus, it is likely that small, benign-appearing tumors were not included in the SEER registries (for example, insulinomas and small nonfunctioning tumors). Although histologic evidence of invasion of basement membrane defines malignant behavior for most epithelial malignancies, the definition of malignant behavior for pancreatic neuroendocrine neoplasms is more complex. In the absence of malignant behavior such as direct invasion of adjacent organs, metastases to regional lymph nodes or distant sites, it may be difficult to classify an islet cell tumor as benign or malignant. Pancreatic endocrine tumors are classified as benign, uncertain malignant potential, and malignant on the basis of size, the presence or absence of lymphovascular invasion, and the number of mitoses and Ki-67–positive cells by immunohistochemistry. However, there is a considerable overlap among these histopathologic features in benign and malignant neuroendocrine tumors and there is even heterogeneity within different areas of the same tumor. Many small islet cell tumors may have been considered benign or of unclear malignant potential and were excluded from the SEER registries. However, size is likely a function of when a tumor is diagnosed; left untreated, it is likely that most islet cell tumors will eventually grow locally into adjacent structures or soft tissues, and/or spread to distant organs. Therefore, outside of small insulinomas, all islet cell neoplasms should be considered potentially malignant. Thus, although the SEER registry data provide important information about malignant islet cell tumors, the extent to which it underestimates the frequency of smaller islet cell neoplasms is unknown. In our experience, even small islet cell tumors without clear evidence of invasion or metastasis at the time of initial surgical resection may recur and spread years later.

We found a wide distribution of age at diagnosis, with a median of 59 years. Separated by race, white patients were older at the time of diagnosis. When we compared stage by histologic type, we found that VIPomas were less likely to be metastatic at the time of diagnosis. Several findings may have contributed to this observation. One possible explanation is that VIPomas are generally associated with profound watery diarrhea, which may cause patients to seek medical attention earlier than if they had nonfunctional tumors. Second, as previously mentioned, small insulinomas and gastrinomas may have been considered benign (in the absence of invasion of adjacent organs, lymph node metastases, or distant metastases) and therefore not captured for analysis thereby enriching the population of study patients with metastatic functioning tumor.

In an earlier publication that was based on data from the SEER Program, Modlin et al.9 described the five-year overall survival of patients with carcinoid tumors to be 59.5% to 67.2%. The survival of patients with islet cell carcinoma seems less favorable. In the present study, we observed a median overall survival of 38 months. This is identical to the median survival observed in a large retrospective series of 163 cases from the University of Texas M. D. Anderson Cancer Center.16 The survival duration of patients with distant metastases was also similar (23 months in current study vs. 25 months in the M. D. Anderson series). We did, however, observe improvements in the outcome of patients with islet cell carcinoma over time. These improvements were observed among all SEER stage groups (data not shown), and are likely due in part to improvements in supportive care.

By multivariate survival analysis, we found that stage of disease, primary tumor location, and age at diagnosis were important predictors of outcome. The difference in survival by primary tumor location can be attributed to several possible factors. In this study, patients with tumors classified as having overlapping location had the worst outcome. This is likely because of the larger tumor size in this group. Tumor located at the head of pancreas may be diagnosed earlier because of hyperbilirubinemia resulting from biliary obstruction. This is suggested by our finding of a trend (nonsignificant) for a lower rate of metastastic disease at the time of diagnosis on the basis of tumor location in the pancreatic head. However, adjusted for stage, patients with a tumor located in the pancreatic head were more likely to have a worse outcome than patients with tumors located in the body or the tail of the pancreas. One possible explanation is that tumors arising in the pancreatic head are of greater malignant potential. For example, most insulinomas (57%), glucagonomas (53%), and VIPomas (64%) were located in the tail of the pancreas. Such functioning tumors were also associated with improved survival compared with nonfunctioning tumors. In addition, the fact that tumors in the pancreatic head are more likely to cause biliary obstruction (with a risk for cholangitis), invade the duodenum (resulting in hemorrhage or obstruction), or involve the peripancreatic mesenteric vasculature (resulting in pain, mesenteric foreshortening, and malabsorption) may contribute to local tumor morbidity which may influence survival duration. There is also likely to be a general tendency on the part of physicians to avoid surgical resection of the pancreatic head (pancreaticoduodenectomy or Whipple procedure) when faced with a large tumor or low volume metastatic disease. This is not the case with tumors in the body or tail of the pancreas that can be surgically excised with a distal pancreatectomy. However, it is the pancreatic head tumor that carries the greatest risk for tumor associated morbidity such as hemorrhage and biliary or gastric outlet obstruction.16 To what degree patients with pancreatic head tumors suffer morbidity and mortality from local disease progression rather than distant metastases cannot be determined from this data set.

Finally, increasing age at diagnosis was a predictor of poor outcome in our study. The differences in survival were seen across all SEER stage groupings. Certainly, pancreatic resection carries considerable operative risks and the medical comorbidities that are associated with advancing age may have precluded some patients from resection while increasing the risk to those taken to surgery. For patients with advanced disease, systemic chemotherapy often includes drugs that are considered toxic.17,18 For example, doxorubicin is recognized to be cardiotoxic, and streptozocin may cause worsening of diabetes. Thus, heart disease and diabetes which are more common among older patients may have limited the use of these drugs.

At present, surgery is the only curative treatment for islet cell carcinoma. Surgery should be recommended for most patients in whom cross-sectional imaging suggests that complete resection is possible.19 Although islet cell carcinoma has a better prognosis than adenocarcinoma of the pancreas, the disease remains incurable once multifocal unresectable metastatic disease exists. Although survival beyond 10 years has been described in the literature for some patients with metastatic disease, the survival duration for most with advanced disease is far shorter. Although streptozocin-based chemotherapy can induce objective response in 32% to 39% of patients,17,20 second-line treatment options are limited. Newer approaches such as peptide receptor radiotherapy, systemic agents targeting vascular endothelial growth factor, and mTOR are under development.

Optimal management of patients with islet cell carcinoma requires an understanding of the disease process and a multimodal approach. A better understanding of the molecular biology of this disease may lead to improved clinical models for predicting outcome and developing novel treatment strategies for this relatively rare but complex disease. Until then, an understanding of the natural history of the disease as provided herein is necessary to allow physicians and patients to accurately assess the risks and potential benefits of treatment alternatives that are based on the extent of disease and the age and performance status of the patient.