|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2002 American Society for Clinical Oncology Prognostic Factors in Pancreatic Endocrine Neoplasms: An Analysis of 136 Cases With a Proposal for Low-Grade and Intermediate-Grade GroupsByFrom the Department of Surgery, University of Florida College of Medicine, Gainesville, FL, and Departments of Pathology and Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY. Address reprint requests to David S. Klimstra, MD, Department of Pathology, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021; email: klimstrd{at}mskcc.org
PURPOSE: In some organs (eg, the lung), endocrine tumors are classified on the basis of mitotic rate and necrosis. The purpose of this study was to evaluate prognostic factors in pancreatic endocrine neoplasms recently treated at a single institution. PATIENTS AND METHODS: In 136 patients undergoing surgery from 1979 to 1998, the influence on disease-free survival (DFS) and disease-specific survival (DSS) of tumor size, mitotic rate, vascular invasion, necrosis, metastases, and nuclear grade was determined. Cases were further grouped according to an existing proposed classification system and then regrouped on the basis of mitotic rate (< 2 mitoses per 50 high-power fields v higher) and necrosis (present or absent) into low- and intermediate-grade groups.
RESULTS: Correlations with DFS and DSS in univariate analysis included CONCLUSION: Pancreatic endocrine neoplasms exhibit a spectrum of biologic behavior, and the proposed benign (macroadenoma) and borderline groups contain potentially aggressive tumors. An alternative system based on mitotic rate and necrosis correlates strongly with survival without specifically designating any group as benign.
DURING EMBRYOGENESIS, pancreatic islets are known to form mostly through cellular buds originating from intralobular ductules. Although this process normally ends before birth, it may persist or reappear in many proliferative diseases of the endocrine pancreas, including solitary endocrine tumors arising in the adult pancreas and pancreatic involvement in multiple endocrine neoplasia type 1.1 Because islet cells often have hormone coexpression during early fetal development, it is thought that the origin of pancreatic endocrine tumors is from multipotent cells that can differentiate toward the cell lines found in these tumors.2 Pancreatic endocrine neoplasms (PENs) are rare benign or malignant epithelial tumors and broadly can be classified as functional or nonfunctional. Functional tumors have a clinical syndrome because of excessive peptide secretion. PENs present in diverse manners with varied treatment dilemmas. Presentation of functional tumors is usually due to symptoms from the hypersecretion of a particular hormone, whereas in nonfunctional tumors it is usually due to an effect of the tumor mass.3 It is unclear whether patients with nonfunctional tumors have a worse prognosis than those with functional islet cell tumors. Several studies have shown that the survival of patients with nonfunctional tumors is poorer than that of patients with functional tumors, but others have found no difference in survival between these groups.4 No consistent differences in histologic pattern have been found in nonfunctioning as compared to functioning tumors.1 The prognosis of patients with PENs is difficult to predict, in part because the definition of malignancy in PENs has been ambiguous. By some, PENs have been defined as malignant only when lymph nodes are involved or liver metastases are documented.3 Other investigators have included vascular invasion or invasion of adjacent structures as evidence of malignancy. However, the concept that a PEN removed successfully without recurrence was therefore biologically benign could be challenged. In fact, strict separation of PENs into benign and malignant groups may be less clinically useful than the definition of prognostic factors. The ideal prognostic classification system for PENs is not yet known. In an effort to more accurately predict the outcome of patients with these tumors, a recent system has been proposed but has yet to be independently validated. This system uses size, mitotic activity, and vascular invasion to separate tumors into microadenoma, macroadenoma, borderline tumor, and low-grade carcinoma groups.5 On the basis of these groupings, some PENs are categorized as benign. In other organs (eg, the lung), differentiated endocrine tumors are classified on the basis of mitotic rate and necrosis and are divided into low- and intermediate-grade groups (typical and atypical carcinoid tumors), although none are considered benign (other than small, incidental "carcinoid tumorlets").6 The purpose of this study was to identify prognostic criteria in a large group of PENs treated at a single institution recently. The usefulness of the currently proposed prognostic classification system was compared with a new system on the basis of mitotic rate and necrosis, similar to that used for endocrine tumors in the lung.
A retrospective review of patients with peripancreatic malignancies undergoing surgical evaluation at Memorial Sloan-Kettering Cancer Center from 1979 to 1998 was performed. From October 1982, patients were identified from a prospective database of patients with pancreatic tumors maintained by the Department of Surgery. On pathology review, 136 patients with differentiated PENs were identified and form the basis for this study. Incidentally detected PENs measuring less than 0.5 cm (so-called microadenomas) were not included. Cases of high-grade neuroendocrine carcinoma (defined as neuroendocrine neoplasms with widespread necrosis and > 10 mitoses per 10 high-power fields [HPFs]) were excluded. Three of 136 patients had a PEN associated with multiple endocrine neoplasia type 1 syndrome. The impact on survival of standard clinicopathologic parameters and additional staining for immunohistochemical markers for patients with either functional or nonfunctional tumors, and for the entire group, was determined. Samples of primary tumor were fixed in formalin and routinely processed to paraffin. Sections were stained with hematoxylin and eosin. The pathologic factors examined included tumor size, presence of lymph node or distant metastases, presence of extrapancreatic soft-tissue invasion, presence of vascular invasion, tumor nuclear grade, tumor mitotic rate per 50 HPFs, and presence of tumor necrosis. Immunohistochemical staining was performed with commercially available antibodies at the dilutions listed in Table 1. Sections (4 to 5 µm) from one representative paraffin-embedded, formalin-fixed block were deparaffinized, rehydrated in graded alcohols, and subjected to antigen retrieval by microwaving the slides in citrate buffer 10 mmol/L at pH 6 or by enzyme digestion. After incubation with the primary antibody, localization was performed via the standard streptavidin-biotin immunoperoxidase method and 3',3'-diaminobenzidine was used as the chromogen. Sections were counterstained with Harris modified hematoxylin. Positive staining, either nuclear (p53, and progesterone receptor protein) or cytoplasmic (remaining antibodies), was scored in a semiquantitative manner and expressed as percentage of tumor cells positive: 2 (> 50% cells positive), 1 (1% to 50% cells positive), and 0 (no staining) for all antibodies except MIB-1 (proliferative index). For MIB-1, nuclear staining was expressed only as the number of cells positive per 10 HPFs.
The utility of a recently proposed classification system for PENs was determined. This system divides tumors into four groups (microadenoma, macroadenoma, borderline tumor, and low-grade carcinoma) predominantly on the basis of tumor size, vascular invasion, and mitotic rate per 10 HPFs.5 Microadenoma and macroadenomas are considered to be benign in this system. Any tumor with metastases, gross local invasion, or blood vessel invasion is automatically a low-grade carcinoma, irrespective of the other parameters. A new classification for PENs was evaluated in which all tumors are considered potentially malignant and are divided into low-grade or intermediate-grade groups on the basis of tumor necrosis and mitotic rate per 50 HPFs.
Follow-up was obtained by patient interview and hospital chart review. Comparisons between groups were performed using
From 1979 to 1998, 136 patients underwent surgery for differentiated PENs at Memorial Sloan-Kettering Cancer Center. The demographics of the entire group are listed in Table 2. There were fewer functional tumors (n = 47) than nonfunctional tumors (n = 89). More patients with functional tumors were female as compared with patients with nonfunctional tumors, but this difference was not significant (P = .15). The median age was significantly lower in patients with functional tumors (53 years) compared with nonfunctional tumors (56 years, P = .03). The median symptom duration before patient evaluation and treatment at our institution was significantly shorter in patients with nonfunctional tumors (90 days) compared with functional tumors (365 days, P = .001).
Twenty-one of 47 functional tumors were insulinomas, with the remainder consisting of gastrinomas (n = 19), glucagonomas (n = 2), or vasoactive intestinal peptide secreting tumors (n = 5). Whereas patients with functional tumors presented with symptoms related to the production of excess hormones, patients with nonfunctional tumors presented with diverse signs and symptoms. The most frequent complaints were abdominal pain (n = 64 [72%]), weight loss (n = 38 [31%]), palpable mass (n = 15 [17%]), and jaundice (n = 13 [15%]). Ten patients had tumors detected while undergoing evaluation for other illnesses. Tumor locations within the pancreas are listed in Table 2. Approximately 50% of the lesions in patients with either nonfunctional or functional tumors were located in the pancreatic head. Median tumor size in patients with functional tumors (1.6 cm; range, 0.5 to 12.5 cm; n = 35) was significantly smaller than in patients with nonfunctional tumors (4.7 cm; range, 0.9 to 15 cm; n = 46; P = .005). Of 136 patients initially undergoing surgical evaluation, 87 (64%) underwent a potentially curative resection (defined as gross removal of all disease). In addition, 25 patients underwent a palliative resection in which gross tumor was left behind, and 24 patients underwent exploration but were found to have unresectable disease. The types of resection performed in 87 patients undergoing a potentially curative resection are listed in Table 3. The majority of patients with functional tumors underwent enucleation (n = 15 [41%]) or distal pancreatectomy (n = 15 [41%]). However, patients with nonfunctional tumors were more likely to undergo aggressive surgical resection consisting of pancreaticoduodenectomy (n = 23 [46%]) or distal pancreatectomy (n = 22 [44%]).
The results of immunohistochemical staining for multiple hormones are listed in Table 4. In general, unpredictable and infrequent staining for various peptide hormones was found in the nonfunctional tumors. In the functional tumors, there was infrequent staining for hormones other than those responsible for the clinical syndrome that was evident at patient presentation. Of note, approximately one third of tumors in both groups stained positively for pancreatic polypeptide. Also, many tumors expressed more than one peptide, usually in minor percentages of cells.
Immunohistochemical findings for other markers are listed in Table 5. Positive staining was most frequent for 013 and progesterone receptor; however, each was expressed in less than 50% of cases. Of note, except for MIB-1, there was no difference in survival for either functional or nonfunctional tumors when stratified by staining for the markers listed in this table (data not shown).
Among the 136 patients undergoing initial evaluation, there was no significant difference in disease-specific survival between patients with nonfunctional (n = 89; median, 100 months) and functional tumors (n = 47; median not reached; P = .34). In addition, with a median follow-up time of 48 months in those undergoing a curative resection, there was no significant difference in disease-free survival (DFS) (nonfunctional, median 110 months; functional, median 152 months; P = .61) (Fig 1) or disease-specific survival (DSS) (nonfunctional, median 141 months; functional, median not reached; P = .43) (Fig 2). In examining the figures, it appears that the survival curves for patients with functional and nonfunctional tumors are starting to separate, but with this length of follow-up, it is not statistically significant.
For those patients with nonfunctioning and functioning PENs undergoing a potentially curative resection (n = 87), factors that had an impact on DFS are listed in Table 6. On univariate analysis, multiple factors were significantly associated with decreased survival in both groups. Many of these factors did not overlap between the two groups. Whereas the presence of metastases was strongly associated with decreased survival in nonfunctioning tumors (P = .0004), this did not reach significance among the functioning tumors (P = .34). In nonfunctioning tumors, an MIB-1 value of more than 50 per 10 HPFs was significantly associated with decreased DFS (P = .002), similar to that found for the mitotic rate (P = .01). Of note, only the presence of tumor necrosis was significantly associated with decreased survival in both groups. Multivariate analysis was not performed within each group because of inadequate numbers of patients.
Factors that had an impact on DSS for patients with nonfunctioning and functioning PENs undergoing a potentially curative resection are listed in Table 7. Tumor mitotic rate of more than 2 per 50 HPFs was the only factor that was significantly associated with decreased survival in both groups.
Because of similar survival of patients with either nonfunctioning or functioning PENs, we combined these two patient populations to determine which factors had an impact on DFS and DSS for all 87 patients undergoing curative resection (Table 8). Although multiple factors were associated with a significantly worse DFS on univariate analysis, only the presence of tumor necrosis and the presence of lymph node or liver metastases were independently associated with this outcome. The four factors that predicted a significantly worse DSS on univariate analysis were the presence of vascular invasion (P = .04), tumor mitotic rate of more than 2 per 50 HPFs (P = .002), tumor size greater than 2 cm (P = .05), and an MIB-1 value of more than 50 per 10 HPFs (P = .05). However, on multivariate analysis, only mitotic rate more than 2 per 50 HPFs was associated with a significantly worse DSS (P = .02).
Classification into groups on the basis of the criteria proposed in the recent Armed Forces Institute of Pathology fascicle on the pancreas5 was possible in 74 patients undergoing a potentially curative resection and is summarized in Table 9. None of our cases fit the microadenoma category (< 0.5 cm). The majority of patients were classified as low-grade malignant, and patients in this group had a significantly decreased DFS as compared with the macroadenoma and borderline groups (P = .02). However, there was no difference in DSS among the three groups (P = .27). Within the macroadenoma and borderline groups, three of 18 and four of eight tumors recurred, respectively.
The tumors were then divided into low-grade and intermediate-grade groups on the basis of necrosis and mitotic rate, similar to the classification groups for typical and atypical carcinoid tumors in the lung. As listed in Table 10, there was a significantly decreased DFS and DSS in tumors that had evidence of necrosis or had  3 mitoses per 50 HPFs. Of note, the greatest decrease in survival was observed for those tumors that had evidence of necrosis or 2 mitoses per 50 HPFs (Figs 3 and 4). Median DFS was 47 months for those tumors that had evidence of necrosis or 2 mitoses per 50 HPFs (intermediate grade), whereas median survival has not been reached for those tumors with no necrosis and less than 2 mitoses per 50 HPFs (low grade).
The prognostic value of classification into low- and intermediate-grade groups was maintained for our starting group of 136 patients undergoing surgical exploration and with documented metastatic disease. In the patients with metastatic disease (n = 53), median survival was 100 months in patients with tumors in the low-grade group (n = 16), whereas the median survival was 61 months in those patients with tumors in the intermediate-grade group (n = 37, P = .03). A further analysis was performed to identify the prevalence of adverse prognostic features in the functioning PENs (Table 11). The insulinomas had a significantly smaller tumor size, lower mitotic counts, and decreased incidence of tumor necrosis and metastases compared with other functional PENs. Therefore, on the basis of mitotic rate and necrosis, the majority of insulinomas were categorized as low grade (81%), whereas most of the other functioning tumors were categorized as intermediate grade (69%). After surgical resection, there were two recurrences among the insulinomas; one of these was classified as intermediate grade.
It is difficult to determine the malignant potential of differentiated PENs on the basis of the histologic appearance. The incidence of malignant behavior in PENs varies between studies because of differences in the pathologic and clinical criteria for malignancy. Often, PENs have been defined as malignant only when metastases to lymph nodes or distant sites occur, or when there is invasion of extrapancreatic structures or vessels. In fact, some PENs lacking all of these features may recur, and others with locally invasive growth may not. The premise that a resected PEN that does not recur or metastasize was therefore benign can be questioned, because the natural history of the tumor was interrupted by surgery. Recently, attempts have been made to define histologic, immunohistochemical, and molecular prognostic factors that may aid in predicting the behavior of PENs after resection.7-11 Although some factors such as DNA ploidy and proliferative rate have been found to correlate with prognosis, none of these studies has successfully stratified patients into groups with differing biologic behavior. The main purpose of this study was to identify prognostic factors in a large series of patients with PENs recently treated at a single institution and having long-term follow-up. Specifically, an attempt was made to determine whether different prognostic groups could be defined that might separate patients into those at higher and lower risk for recurrence and death from disease. It has been suggested that many small functional PENs are discovered earlier because they produce symptoms related to hormone excess in the absence of biliary or gastrointestinal obstruction by the primary tumor.12 We found that the median duration of symptoms in patients with functional tumors (1 year) was actually significantly longer than for patients with nonfunctional tumors (3 months). Other authors13 have made similar observations. It seems that most symptomatic patients with PENs undergoing surgical treatment have had symptoms for a long time. However, because patients with nonfunctional tumors may be asymptomatic early in the course of the disease, the delay in the onset of symptoms may be a reason for the larger size of these tumors. Indeed, we found that nonfunctioning tumors were significantly larger than functioning tumors. In a series of patients with nonfunctioning PENs from the Mayo Clinic, the size of the primary tumor was more than 5 cm in 72% of cases.14 In an effort to better characterize the tumors, immunohistochemical staining for multiple markers was performed. In the nonfunctioning PENs, staining for insulin, gastrin, somatostatin, and vasoactive intestinal polypeptide was infrequent and generally focal (only 3% to 5% of tumors had widespread positivity), but more reactivity for glucagon and pancreatic polypeptide was found (10% to 11% widespread positivity and 21% to 22% focal positivity). These findings may partially explain the failure of such tumors to produce a hyperfunctional syndrome.2 In the functional tumors, the hormone causing the symptoms generally could be identified. However, in addition to the peptides expected to be found in functioning tumors, cells immunoreactive for other hormones also often were seen. Of note is the finding that 39% of functional tumors stained for pancreatic polypeptide. This finding confirms the data of other workers.2 Immunohistochemical staining for multiple other markers failed to enhance our ability to predict outcome. The most frequent positive stains were for progesterone receptor and 013; however, there was no correlation with survival. This is in contrast to the univariate analysis previously reported that demonstrated a significant correlation of progesterone receptor expression with survival.7 Stains for mucopeptides indicative of ductal differentiation (B72.3, CA19.9, carcinoembryonic antigen, and DUPAN-2) were positive at least focally in up to 18% of cases but were rarely diffusely expressed. Focal acinar differentiation in the form of positive staining for the enzymes trypsin or lipase was found in 5% of tumors, much less than previously reported.15 Although both ductal adenocarcinomas and acinar cell carcinomas are more aggressive tumors than PENs, the presence of focal ductal or acinar differentiation as detected immunohistochemically did not impact on prognosis. In our starting group of 136 patients and in those undergoing a potentially curative resection (n = 87), we found no statistically significant difference in DFS or DSS between patients with functional and nonfunctional tumors (Figs 1 and 2). Therefore, we felt it appropriate to evaluate factors predicting survival by grouping these tumor types together as well as analyzing them separately. It is still unclear whether patients with nonfunctional tumors have an inherently worse prognosis than those with functional islet cell tumors. Several studies have shown that the survival of patients with functional tumors is improved compared with that of patients with nonfunctional tumors. In a report from the Cleveland Clinic, the 10-year survival was 92% for patients with insulinomas, 68% for those with gastrinomas, and 55% for patients with nonfunctioning tumors.13 Researchers at the Mayo Clinic also found that survival was significantly worse in patients with nonfunctioning tumors than in those with gastrinomas: 91% v 58% at 3 years.16 However, in a more recent report from the same institution, no survival difference was noted between patients with nonfunctional and functional neoplasms,17 a finding confirmed in other studies.18 Our data reveal a trend toward longer survival in the functional group that did not reach statistical significance. Perhaps with a larger group of patients and longer follow-up, this trend may prove significant. Some authors indicate that the main determinant of prognosis may be the size of the primary tumor rather than functional status. In the study by Phan et al,19 the median size of functional PENs was 1.9 cm, whereas it was 4.0 cm in the nonfunctional tumors. Furthermore, malignancy rates were correspondingly lower in the functional tumors (47%) than in the nonfunctional tumors (60%). In the current study, tumor size on univariate analysis was found to correlate significantly with outcome, but did not remain an independent predictor of outcome on multivariate analysis (Table 8). Certainly, as shown in this and other studies, nonfunctional tumors often are not detected until the tumor becomes larger because of the absence of an endocrine syndrome. This may account for the poorer survival seen in some studies of patients with nonfunctioning tumors. Clearly, most insulinomas are symptomatic even at a small size, and because the majority of functioning PENs are insulinomas, the detection and removal of these tumors at a relatively early stage would contribute to a lower incidence of malignant behavior in the functional group. However, our data suggest that size alone is an insufficient criterion for predicting outcome of differentiated endocrine tumors. In our study, some of the prognostic factors had an impact on disease recurrence (DFS) but not on death from tumor (DSS). Certainly, it is important to evaluate DFS in PENs. Because of the relatively indolent nature of these tumors, survival can be prolonged even in the presence of recurrent tumor. We found in our statistical analysis of factors influencing DFS and DSS that mitotic rate and/or necrosis were repeatedly demonstrated to be of prognostic significance (Tables 6 through 8). However, some differences were identified in important prognostic factors between nonfunctioning and functioning tumors (and within these tumor groups), depending on the outcome being censored. This may certainly be related to biologic differences between nonfunctioning and functioning tumors but can also be explained by small numbers in some of the groups being analyzed and by the length of follow-up. Of note, whereas the presence of metastases is thought to be a powerful factor associated with decreased survival in PENs, we could not statistically demonstrate this in all of our analyses of patients undergoing a curative resection. Certainly, on univariate analysis, the presence of metastases in nonfunctioning tumors was significantly associated with tumor recurrence (P = .0004) but not death from disease (P = .50). However, this was not the case for functioning tumors where no impact of metastases on DFS could be determined (Table 6). This discrepancy may be explained in part because only nine of 37 patients with functioning tumors undergoing curative resection had metastatic disease. Prolonged follow-up in patients with PENs is essential to make meaningful determinations of which parameters have an impact on survival. Recently, a multiparametric approach including size, angioinvasion, perineural invasion, mitotic rate, and type of hormone secreted has been advocated as effective for the identification of PENs at higher risk of recurrence or metastasis.10 In this classification system, some tumors are considered benign, whereas others are considered borderline or low-grade malignant.10 When tumors in the current study were classified according to this system, we found that three of 15 patients in the benign group and four of eight patients in the borderline group had recurrence of tumor or died of disease. Therefore, small PENs with a low mitotic rate may not necessarily behave in a benign fashion. It seems that all PENs could be considered potentially malignant, with the exception of small endocrine microadenomas (< 0.5 cm). In addition, although this system relies heavily on tumor size and mitotic rate, any tumor with metastasis, gross local invasion, or blood vessel invasion is automatically considered low-grade malignant, irrespective of the other parameters. This results in a mixture of parameters that incorporates histologic factors as well as stage of disease and may not be applicable to every case.
In the current study, the two most important prognostic parameters to emerge in the multivariate analyses were mitotic rate and the presence of necrosis. When these two features were combined and used to separate the tumors into low-grade (no necrosis and < 2 mitoses per 50 HPFs) and intermediate-grade (presence of necrosis or Similar to the present study with PENs, in a recently reported multivariate analysis of survival in 200 pulmonary neuroendocrine tumors, mitotic counts were the only independent predictor of prognosis. On the basis of this finding, investigators divided pulmonary carcinoids into typical and atypical on the basis of mitoses and necrosis. Although there is theoretically a continuum among them, the typical carcinoids had a 90% 5-year survival and the atypical carcinoids had a 50% 5-year survival, two sharply different groups.6 These results match closely with the survival in our low- and intermediate-grade groups (Figs 3 and 4). Evaluation of mitoses and necrosis are relatively objective criteria and are more likely to be related to the fundamental biology of the tumor rather than parameters such as size and metastases, features that may simply reflect the stage at which the tumor was detected. Nevertheless, there are problems associated with the counting of mitoses. The issues of methodology and reproducibility related to mitosis counting are well known to pathologists. However, mitotic counts are used in the grading schemes for many tumors and are considered reproducible and of prognostic value if performed carefully. Other investigators have used immunohistochemical methods for assessing proliferative rates, especially using the MIB-1 antibody. Several studies have found the MIB-1 labeling index to have prognostic value in PENs.7,8 Although it is true that MIB-1 labels a greater proportion of tumor cells than those in active mitosis (and is therefore presumably more sensitive for detecting proliferating cells), we did not find an independent correlation between MIB-1 staining and prognosis in this study. Furthermore, the reliance on an immunohistochemical stain for classification is problematic, because immunohistochemical techniques may not be universally available and are subject to differences in techniques, variability with differences in fixation, and other problems. Outside of the context of a research study, counting mitoses is a more practical method of assessing proliferative rate, even if less technically sophisticated. The level of mitotic activity selected to separate the low-grade and intermediate-grade groups was chosen because it split the entire group roughly equally and it generated the most highly significant statistical difference. Nonetheless, other cutoff values were found to have statistically significant prognostic value (Table 10). Whether the value proposed in this study is optimal will require additional study. It is interesting to note that the mitotic rate used to separate typical from atypical pulmonary carcinoids has been revised several times since the original description of atypical carcinoid tumors in 1972, on the basis of the accumulation of additional data.20-22 This study affirms the clinical usefulness of a two-tiered classification of differentiated pancreatic endocrine neoplasms into low- and intermediate-grade groups. Without attempting rigidly to separately PENs into benign and malignant groups, the use of mitotic rate and necrosis defines subsets with distinctly different survival. Furthermore, this classification system can be applied independent of hormonal functional status.
1. Solcia E, Sessa F, Rindi G, et al: Pancreatic endocrine tumors: General concepts, nonfunctioning tumors and tumors with uncommon function, in Dayal Y (ed): Endocrine Pathology of the Gut and Pancreas. Boca Raton, FL, CRC Press, 1991, pp 105-131 2. Heitz PU, Kasper M, Polak JM, Kloppel G: Pancreatic endocrine tumors: Immunocytochemical analysis of 125 tumors. Hum Pathol 13: 263-271, 1982[Medline] 3. Hochwald SN, Conlon KC, Brennan MF: Nonfunctioning pancreatic islet cell tumors, in Doherty GM (ed): Surgical Endocrinology. Philadelphia, PA, Lippincott Williams & Wilkins, 2001, pp 361-373 4. White TJ, Edney JA, Thompson JS, et al: Is there a prognostic difference between functional and nonfunctional islet cell tumors? Am J Surg 168: 627-630, 1994[Medline] 5. Solcia E, Capella C, Kloppel G: Tumors of the endocrine pancreas, in Solcia E (ed): Atlas of Tumor Pathology: Tumors of the Pancreas. Washington, DC, Armed Forces Institute of Pathology, 1997, pp 145-209 6. Travis WD, Rush W, Flieder DB, et al: Survival analysis of 200 pulmonary neuroendocrine tumors with clarification of criteria for atypical carcinoid and its separation from typical carcinoid. Am J Surg Pathol 22: 934-944, 1998[CrossRef][Medline] 7. La Rosa S, Sessa F, Capella C, et al: Prognostic criteria in nonfunctioning pancreatic endocrine tumours. Virchows Arch 429: 323-333, 1996[Medline] 8. Pelosi G, Bresaola E, Bogina G, et al: Endocrine tumors of the pancreas: Ki-67 immunoreactivity on paraffin sections is an independent predictor for malignancy—A comparative study with proliferating-cell nuclear antigen and progesterone receptor protein immunostaining, mitotic index, and other clinicopathologic variables. Hum Pathol 27: 1124-1134, 1996[CrossRef][Medline] 9. Bottger T, Seidl C, Seifert JK, et al: Value of quantitative DNA analysis in endocrine tumors of the pancreas. Oncology 54: 318-323, 1997[Medline] 10. Solcia E, Rindi G, Paolotti D, et al: Clinicopathological profile as a basis for classification of the endocrine tumours of the gastroenteropancreatic tract. Ann Oncol 10: S9-S15, 1999 (suppl 2)
11.
Rigaud G, Missiaglia E, Moore PS, et al: High resolution allelotype of nonfunctional pancreatic endocrine tumors: Identification of two molecular subgroups with clinical implications. Cancer Res 61: 285-292, 2001 12. Bieligk S, Jaffe BM: Islet cell tumors of the pancreas. Surg Clin North Am 75: 1025-1040, 1995[Medline] 13. Broughan TA, Leslie JD, Soto JM, et al: Pancreatic islet cell tumors. Surgery 99: 671-678, 1986[Medline] 14. Kent RB, Van Heerden JA, Weiland LH: Nonfunctioning islet cell tumors. Ann Surg 193: 185-190, 1981[Medline] 15. Yantiss RK, Farraye FA, Chang HK, et al: Prognostic significance of acinar cell differentiation in pancreatic endocrine tumors (PET). Mod Pathol 14: 99A, 2001 (abstr) 16. Thompson GB, Van Heerden JA, Grant CS, et al: Islet cell carcinomas of the pancreas: A twenty-year experience. Surgery 104: 1011-1017, 1988[Medline] 17. Lo CY, Van Heerden JA, Thompson GB, et al: Islet cell carcinoma of the pancreas. World J Surg 20: 878-884, 1996[CrossRef][Medline] 18. White TJ, Edney JA, Thompson JS, et al: Is there a prognostic difference between functional and nonfunctional islet cell tumors. Am J Surg 168: 627-630, 1994[Medline] 19. Phan GQ, Yeo CJ, Hruban RH, et al: Surgical experience with pancreatic and peripancreatic neuroendocrine tumors: Review of 125 patients. J Gastrointest Surg 2: 473-482, 1998 20. Arrigoni MG, Woolner LB, Bernatz PE: Atypical carcinoid tumors of the lung. J Thorac Cardiovasc Surg 64: 413-421, 1972[Medline] 21. Travis WD, Linnoila RI, Tsokos MG, et al: Neuroendocrine tumors of the lung with proposed criteria for large-cell neuroendocrine carcinoma: An ultrastructural, immunohistochemical, and flow cytometric study of 35 cases. Am J Surg Pathol 15: 529-553, 1991[Medline] 22. Beasley MB, Thunnissen FB, Brambilla E, et al: Pulmonary atypical carcinoid: Predictors of survival in 106 cases. Hum Pathol 31: 1255-1265, 2000[CrossRef][Medline] Submitted October 11, 2001; accepted March 5, 2002.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2002 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 mitoses per 50 high-power fields (P = .001, P = .002), vascular invasion (P = .02, P = .04), size 
2 analysis. Survival within groups was analyzed with the Kaplan-Meier method with differences compared using the log-rank test. Multivariate analysis of survival was performed with the Cox proportional hazards model. Significance was defined as P < .05. 
