- Email: [email protected]
Correlation between grade and prognosis in metastatic gastroenteropancreatic neuroendocrine tumors
Human Pathology (2009) 40, 1262–1268
www.elsevier.com/locate/humpath
Original contribution
Correlation between grade and prognosis in metastatic gastroenteropancreatic neuroendocrine tumors Jonathan Strosberg MD a,⁎, Aejaz Nasir MD a , Domenico Coppola MD a , Mark Wick MD b , Larry Kvols MD a a
H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL 33612, USA University of Virginia Medical Center, Charlottesville, VA 22908, USA
b
Received 6 November 2008; revised 14 January 2009; accepted 22 January 2009
Keywords: Neuroendocrine tumors; Carcinoid tumors; Pancreatic endocrine tumors; Islet cell tumors; Grade; Differentiation; Prognosis
Summary Three-tiered grading systems (low, intermediate, and high grade) have been proposed for neuroendocrine tumors. These classifications have not been rigorously evaluated in neuroendocrine malignancies of the digestive tract. We performed a retrospective chart analysis of 83 patients with metastatic gastroenteropancreatic neuroendocrine tumors, correlating tumor grade with overall survival. We also analyzed available biopsy specimens (on 40 patients), examining hematoxylin and eosin stains for mitotic rate and immunostaining for measurement of the Ki-67 index. Tumor grades were assigned based on the mitotic rate and the Ki-67 index, and the prognostic validity of each grading method was assessed. A highly significant correlation existed between the reported tumor grade and overall survival. Five-year survival rates for patients with low-, intermediate-, and high-grade tumors were 87%, 38%, and 0%, respectively. On biopsy specimen analysis, both mitotic rates and Ki-67 indexes correlated strongly with overall survival. We conclude that a 3-tiered grading classification for gastroenteropancreatic neuroendocrine tumors correlates with survival in the metastatic setting. Both mitotic rates and Ki-67 indexes are inversely associated with survival and can be analyzed independently for assignment of grade. © 2009 Elsevier Inc. All rights reserved.
1. Introduction Recent grading classifications for neuroendocrine tumors (NETs) are based on growing evidence that histologic markers of proliferation and differentiation correlate strongly with prognosis. The initial distinction between “typical” and “atypical” neuroendocrine histology originated with an analysis by Arrigoni et al [1] of 201 bronchial ⁎ Corresponding author. E-mail address: [email protected] (J. Strosberg). 0046-8177/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.humpath.2009.01.010
carcinoid tumors in the early 1970s. Various morphological factors including mitotic rate, cell pleomorphism, and necrosis were found to differentiate between indolent and aggressive tumor categories. The grading of pulmonary NETs was further refined by Travis et al [2] who distinguished between low-grade (typical carcinoid), intermediate-grade (atypical carcinoid), and high-grade (small cell and large cell neuroendocrine carcinoma) categories, based primarily on mitotic rates (b2, 2-10, and N10 mitoses per 10 high-power field [HPF], respectively). This 3-tiered classification was shown to correlate strongly with survival in an analysis of 200 tumor specimens.
Correlation between grade and prognosis in metastatic GEP-NETs Similar grading systems have been proposed for NETs of the thymus. Rosai et al [3] proposed classifying these tumors as carcinoid grade 1 (low grade), grade 2 (intermediate grade or atypical), and grade 3 (equivalent to small cell carcinoma). Moran and Suster [4] recommended a similar classification that abandoned the term carcinoid and replaced it with the more contemporary neuroendocrine carcinoma. This grading system defined well-differentiated neuroendocrine carcinomas as having low mitotic activity (3 or fewer mitoses per 10 HPF), minimal cytologic atypia, and small foci of necrosis; moderately differentiated carcinomas as having increased mitotic activity (4-10 mitoses per 10 HPF), areas of necrosis, and moderate cytologic atypia; and poorly differentiated carcinomas as having high mitotic activity (N10 mitoses per 10 HPF), extensive necrosis, and marked nuclear atypia. Analysis of the Ki-67 labeling index, typically identified by the MIB-1 monoclonal antibody, has offered an alternative immunohistochemical technique for assessing prognosis in NETs. Several studies have validated the role of Ki-67 staining in grading NETs of the lung [5,6] and gastrointestinal tract [7-10]; however, some investigators have questioned the use of Ki-67 as an independent prognostic variable [11-13]. Gastroenteropancreatic NETs (GEP-NETs) are predominantly indolent low-grade neoplasms; however, some are histologically and clinically aggressive. In the past, malignant GEP-NETs have been separated into 2 grades consisting of well-differentiated (low-grade) and poorly differentiated (high-grade) carcinomas [14-16]. Distinguishing characteristics of low-grade versus high-grade tumors have not been uniformly defined in the literature. Some sources have emphasized standard histologic criteria (atypia, necrosis, increased mitotic rate) [14-18], whereas others have focused on elevations in the Ki-67 index (N30%) as a defining feature of high-grade tumors [19]. In recent years, several authors have proposed a 3-tiered grading system for GEP-NETs modeled after the established categories of pulmonary and thymic neuroendocrine carcinomas [20,21]. Data supporting an intermediate-grade or “moderately differentiated” prognostic category of GEPNETs include the analysis of Hochwald et al [13] of 136 differentiated pancreatic endocrine tumors, identifying mitotic rate and necrosis as features distinguishing low versus intermediate-grade tumors. Similarly, an analysis by Van Eeden et al [22] of metastatic differentiated midgut carcinoid tumors identified a mitotic rate greater than 2 per 10 HPF as an independent negative prognostic factor. Other recent studies have focused on the Ki-67 index as a method of distinguishing low-grade from intermediate-grade tumors. For example, Tomassetti et al [23] identified a Ki-67 index of greater than 2.6% as a negative prognostic factor in patients with differentiated endocrine tumors of the ileum. This cutoff of approximately 2% was in agreement with results of other contemporary studies examining the prognostic role of Ki-67 in differentiated GEP-NETs [24-26]. The
1263
above data sets have motivated the European Neuroendocrine Tumor Society to propose a 3-tiered grading system for foregut [27], midgut [28], and hindgut [28] NETs in which low-grade (G1) tumors are defined by a mitotic rate of less than 2 per 10 HPF and/or Ki-67 index of 2% or less, intermediate-grade (G2) tumors defined by 2 to 20 mitoses per 10 HPF and/or Ki-67 index between 3% and 20%, and high-grade (G3) carcinomas defined by mitotic count greater than 20 per 10 HPF and/or Ki-67 index greater than 20%. It is thus evident, that 3-tiered grading systems for GEPNETs have been adopted and translated into routine practice, although the criteria proposed by different groups do not yet reflect a clear consensus on how to define individual grades. At the H. Lee Moffitt Cancer Center, Tampa, FL, we have roughly followed parameters suggested by Travis et al [2], Moran and Suster [4], and Wick [21], evaluating mitotic rates and histologic criteria (atypia, pleomorphism, necrosis) for grading GEP-NETs. Ki-67 immunostaining has not been routinely performed for grading purposes. Our hypothesis is that this 3-tiered grading system significantly correlates with overall survival in patients with metastatic GEP-NETs and can serve as a valid prognostic factor. We have therefore investigated the impact of tumor grade (as reported in the medical record) on overall survival in a large data set of patients with metastatic GEP-NETs. We have also investigated the correlation of tumor grade with other end points such as somatostatin receptor expression and presence of pathologic hormonal secretion. Finally, we have performed histologic and immunohistochemical analyses on a subset of available biopsy specimens to more precisely define grading parameters and compare the validity of grading based on the mitotic rate versus the Ki-67 index.
2. Materials and methods 2.1. Step 1: Chart review We conducted a medical chart analysis of patients with metastatic GEP-NETs seen at the H. Lee Moffitt Cancer Center between 1999 and 2005, selecting all patients characterized on pathologic reports as having high-grade carcinomas (also termed poorly differentiated or grade 3) and intermediate-grade carcinomas (also termed moderately differentiated or grade 2) and randomly selecting an equivalent number of patients with the more common lowgrade carcinomas (also termed well-differentiated or grade 1]). Neuroendocrine tumors of unknown primary were included if metastatic spread predominantly involved the liver, mesentery, and/or other infradiaphragmatic organs. Kaplan-Meier methodology was used to evaluate overall survival, stratified by tumor grade (as reported in the medical record). Other clinical features examined included presence of pathologically elevated hormone levels and evidence of
1264
J. Strosberg et al.
somatostatin receptor expression (as determined by detection of radiotracer uptake on 111In pentetreotide scintigraphy).
tissue included nonneoplastic lymph node, following the manufacturer recommendations. The negative control included tissue sections incubated with normal rabbit serum.
2.2. Step 2: Histologic analysis of NECA specimens 2.4. Step 4: Statistical analysis A team of 2 pathologists (A. Nasir and D. Coppola) with expertise in neuroendocrine malignancies evaluated hematoxylin and eosin–stained sections on all excisional and core biopsy specimens available among the list of the above patients. A mitotic rate was assessed in each case (number/ 10 HPF), and the presence or absence of tumor necrosis was documented.
2.3. Step 3: Ki-67 immunostaining and determination of proliferation index Sections from tumors were then submitted for immunohistochemical examination to evaluate the Ki-67 index. Each sample was fixed in 10% neutral buffered formalin for 9 hours. After fixation, the tissue samples were processed into paraffin blocks. Immunohistochemical staining was performed on Dako autostainer (Dako, Carpinteria, CA) using 3-μm-thick paraffin sections from each of the representative tumor block selected. The sections were deparaffinized, rehydrated, and incubated with 3% H2O2 to block endogenous peroxidase. After high-temperature steam-induced epitope retrieval, the sections were incubated with a rabbit polyclonal antibody to Ki-67 (K2, Ventana Medical Systems Product no. 790-2910, Cell Conditioning 1 Solution, Standard, proprietary dilution) for 60 minutes according to the manufacturer's instructions. The sections were then washed in TBS and incubated with a biotinylated secondary antibody (Dako), followed by strep-avidin-biotin complex (Dako) for 30 minutes. A solution of 3,3′diaminobenzidene tetrahydrochloride (Sigma, St Lois, MO) was used as a chromogen followed by sodium azide and 20 μL of H2O2 in 100 mL of Tris-HCl (50 μmol/L, pH 7.6). After light counterstain with Harris hematoxylin, the sections were examined under light microscope. The sections stained for Ki-67 were independently evaluated by 2 expert pathologists (A. Nasir and D. Coppola) without prior knowledge of the clinicopathologic data. The Ki-67 index was determined by assessing the percentage of positively staining tumor cell nuclei. These evaluations were made in the most viable areas of the stained tumor sections that also showed the highest degree of nuclear labeling. Any difference of interpretation was resolved by joint immunohistochemical review by the 2 pathologists. Positive control Table 1
The mitotic rates and Ki-67 indexes were entered on a scatterplot graph, and linear regression analysis was performed to assess the correlation between the 2 prognostic variables. Tumors were then assigned a grade based on mitotic rate (mitotic grade) and a separate grade based on the Ki-67 proliferation index (Ki-67 grade). Survival was estimated using Kaplan-Meier methodology stratified by both grading systems to assess the prognostic power of one model (mitotic grade) versus the alternative model (Ki-67 grade).
3. Results A total of 83 cases of metastatic GEP-NETs were analyzed. All 27 cases of high-grade (poorly differentiated) tumors and all 28 cases of intermediate-grade (moderately differentiated) tumors found on chart review were selected for analysis. Among several hundred patients with lowgrade (well-differentiated) tumors, 28 were randomly selected for analysis.
3.1. Patient and tumor characteristics 3.1.1. Age The median age at diagnosis was 57 years (range, 23-83 years). There was no significant correlation between tumor grade and age. 3.1.2. Sex The entire cohort is composed of 50 males and 33 females. Males represented 39% of the low-grade, 64% of the intermediate-grade, and 78% of the high-grade cohort (P b .025). 3.1.3. Primary tumor site The most common sites of primary tumor were small bowel (28 patients), pancreas (24 patients), colon (10 patients), and rectum (5 patients). Twelve patients had metastatic tumors with unknown primaries that were suspected to be gastrointestinal based on their infradiaphragmatic pattern of metastatic disease. There were no significant
Sites of metastatic dissemination
Low grade Intermediate grade High grade
Liver
Mesentery
Lungs
Bone
Brain
Ovary
Other LN
25 26 23
7 8 1
0 3 2
0 5 6
0 1 5
1 2 0
2 0 5
Correlation between grade and prognosis in metastatic GEP-NETs
1265
differences in the distribution of primary tumors among the low-grade and intermediate-grade cohort (preponderance of pancreatic and ileal primary tumors). In the high-grade cohort, a substantially higher fraction of primary tumors were colorectal or occult. 3.1.4. Sites of metastatic spread The liver was the most common site of metastases (89%) followed by the mesentery (19%) and bone (11%). Skeletal metastases were observed only in intermediate- and highgrade tumors. Brain metastases occurred in 5 high-grade tumors and one intermediate-grade tumor but were not observed in low-grade tumors (Table 1). 3.1.5. Somatostatin receptor status Among 27 patients with low-grade tumors who underwent somatostatin receptor scintigraphy (OctreoScans), all (100%) had evidence of radiotracer uptake, indicating presence of somatostatin receptors. Thirteen (56%) of 23 scanned patients with intermediate-grade tumors and only 1 (14%) of 7 scanned patients with high-grade tumors had positive OctreoScans (Fig. 1). 3.1.6. Secretory activity Eighteen (64%) of 28 patients with low-grade tumors, 14 (50%) of 28 patients with intermediate-grade tumors, and 2 (7%) of 27 patients with high-grade tumors had clinical and laboratory evidence of hormonal syndrome (Fig. 2). The carcinoid syndrome was most commonly observed. Other hormonal syndromes included VIPoma, glucagonoma, insulinoma, gastrinoma, and Cushing's syndrome.
3.2. Survival analysis 3.2.1. Median survival From diagnosis of metastases until death from any cause, median survival was 13 months in the high-grade cohort, was
Fig. 1 Percentage of tumors with radiotracer uptake on somatostatin receptor scintigraphy.
Fig. 2 Percentage of tumors associated with a hormonal syndrome.
39 months in the intermediate-grade cohort, and was not reached in the low-grade cohort of patients (Fig. 3). 3.2.2. Two-year and 5-year survival rates The estimated survival rates at 2 years (by Kaplan-Meier analysis) were 100% in the low-grade, 57% in the intermediate-grade, and 22% in the high-grade cohort. At 5 years, the respective estimated survival rates were 87%, 38%, and 0% (Table 2).
3.3. Pathologic specimen analysis Among the 83 cases selected for chart review, pathologic biopsy specimens were available on 40 patients. The specimens included 29 surgical excisions, 9 core needle
Fig. 3 Overall survival of patients with metastatic GEP-NETs stratified by tumor grade.
1266
J. Strosberg et al.
Table 2 One-year, 2-year, and 5-year survival rates of patients with metastatic GEP-NETs, stratified by tumor grade
12 mo 24 mo 60 mo
Low grade (%)
Intermediate grade (%)
High grade (%)
P
100 100 87
78 57 37
51 22 0
b.001 b.001 b.001
biopsies, and 2 endoscopic biopsies. An analysis of mitotic rate (number/10 HPF) and Ki-67 proliferation index was performed in each case. Most of the specimens belonged to the low-grade and high-grade cohorts, with a relatively small number of intermediate-grade specimens available for histologic examination. 3.3.1. Correlation between mitotic rate and Ki-67 index On linear regression analysis, a highly significant correlation between the mitotic rate and the Ki-67 index was observed with a Pearson correlation coefficient of 0.90 (Fig. 4). 3.3.2. Assignment of grade based on mitotic rate and Ki-67 index Because of the relatively small number of biopsy specimens (particularly in the intermediate-grade cohort), a 2-tiered (rather than a 3-tiered) grading system was used. Tumors with mitotic rates of 0 to 10/10 HPF were classified as low/ intermediate grade, whereas those with mitotic rates greater than 10/10 HPF were classified as high grade. Likewise, tumors with a Ki-67 index of 0% to 20% were classified as low/intermediate grade, whereas those with a Ki-67 index greater than 20% were classified as high grade. Using these grading thresholds, there was a complete agreement between the mitotic and Ki-67 grades with the exception of a single specimen that was classified as low/intermediate grade by mitotic rate but as high-grade by Ki-67 index.
Fig. 4 Correlation between the mitotic rate and Ki-67 proliferative index in patients with metastatic GEP-NETs.
Fig. 5 Overall survival stratified by mitotic rate (0-10 mitoses/10 HPF versus N10 mitoses/10 HPF).
3.3.3. Survival analysis stratified by mitotic grade and Ki-67 grade The classification of tumors into low/intermediate-grade and high-grade cohorts by mitotic rate and by Ki-67 index correlated strongly with survival (Figs. 5 and 6). Whether grades were defined based on the mitotic rate or the Ki-67 index, the survival rates were identical at 2 and 5 years (100% and 85%, respectively, in the low/intermediate-grade cohort; 23% and 0%, respectively in the high-grade cohort). This suggests that both grading criteria (mitotic rate and Ki67 index), if assessed objectively, can provide clinically relevant information.
Fig. 6 Overall survival stratified by the Ki-67 proliferative index (0%-20% versus N20%).
Correlation between grade and prognosis in metastatic GEP-NETs
4. Discussion Gastroenteropancreatic NETs are heterogeneous neoplasms with markedly distinct prognoses varying by grade. In our chart review of 83 metastatic GEP-NETs, we have found that a 3-tiered histologic grading system correlated strongly with overall survival (5-year survival rates of 87%, 38%, and 0% in low-, intermediate-, and high-grade cohorts, respectively). We have also demonstrated an inverse relationship between tumor grade and somatostatin receptor expression as well as an inverse association between grade and presence of a functioning hormonal syndrome. Our findings add evidence to the argument that a 3-tiered grading system is prognostically valid in GEP-NETs, as has been previously demonstrated in pulmonary [2] and thymic [4] NETs. Our comparison of the mitotic rate versus the Ki-67 index in 40 biopsy specimens demonstrated that these 2 markers of proliferation are highly correlated with each other. Using a mitotic rate of 10 per 10 HPF and a Ki-67 index of 20% as cutoff points, we found that both mitotic grade and Ki-67 grade were identical in their ability to distinguish goodprognosis versus poor-prognosis patients. We conclude that although immunohistochemical analysis of the Ki-67 index is not always necessary for assignment of grade, it represents a clinically relevant criterion to predict survival in patients with metastatic GEP-NETs, particularly when the mitotic rate is not easily assessed or when the clinical presentation is discordant with the histologic appearance of the tumor. The number of biopsy specimens available for pathologic review in our study was insufficient to establish a precise threshold for separating low-grade from intermediate-grade tumors. Until larger studies define unique prognostic parameters for GEP-NETs, we support a uniform 3-tiered grading system for NETs, modeled after the proposals of Moran and Suster [4] and Wick [21], that is, using mitotic rate thresholds of 0 to 3, 4 to 10, and greater than 10 as the primary method of distinguishing low-, intermediate-, and high-grade tumors. A consistent and standardized system for grading NETs of the lungs, mediastinum, and gastroenteropancreatic tract would add much clarity to NET taxonomy and aid in stratification of future clinical trials.
4.1. Conclusions A 3-tiered grading classification for GEP-NETs (low grade, intermediate grade, and high grade) correlates strongly with survival in the metastatic setting. The mitotic rate (number of mitoses per 10 HPF) is strongly associated with the Ki-67 proliferation index. Both markers of proliferation are inversely correlated with survival and can be analyzed independently for assignment of grade.
References [1] Arrigoni MG, Woolner LB, Bernatz PE. Atypical carcinoid tumors of the lung. J Thorac Cardiovasc Surg 1972;64:413-21.
1267
[2] 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 1998;22:934-44. [3] Rosai J, Levine G, Weber WR, Higa E. Carcinoid tumors and oat cell carcinomas of the thymus. Pathol Annu 1976;11:201-26. [4] Moran CA, Suster S. Neuroendocrine carcinomas (carcinoid tumor) of the thymus. A clinicopathologic analysis of 80 cases. Am J Clin Pathol 2000;114:100-10. [5] Costes V, Marty-Ane C, Picot MC, et al. Typical and atypical bronchopulmonary carcinoid tumors: a clinicopathologic and KI-67– labeling study. HUM PATHOL 1995;26:740-5. [6] Bohm J, Koch S, Gais P, Jutting U, Prauer HW, Hofler H. Prognostic value of MIB-1 in neuroendocrine tumours of the lung. J Pathol 1996; 178:402-9. [7] von Herbay A, Sieg B, Schurmann G, Hofmann WJ, Betzler M, Otto HF. Proliferative activity of neuroendocrine tumours of the gastroenteropancreatic endocrine system: DNA flow cytometric and immunohistological investigations. Gut 1991;32:949-53. [8] Moyana TN, Xiang J, Senthilselvan A, Kulaga A. The spectrum of neuroendocrine differentiation among gastrointestinal carcinoids: importance of histologic grading, MIB-1, p53, and bcl-2 immunoreactivity. Arch Pathol Lab Med 2000;124:570-6. [9] Rindi G, Azzoni C, La Rosa S, et al. ECL cell tumor and poorly differentiated endocrine carcinoma of the stomach: prognostic evaluation by pathological analysis. Gastroenterology 1999;116:532-42. [10] La Rosa S, Sessa F, Capella C, et al. Prognostic criteria in nonfunctioning pancreatic endocrine tumours. Virchows Arch 1996; 429:323-33. [11] Zimmermann ME, Bosman FT. Proliferative activity of well differentiated neuroendocrine tumours of the gut. Histol Histopathol 2003;18:353-8. [12] Al-Khafaji B, Noffsinger AE, Miller MA, DeVoe G, Stemmermann GN, Fenoglio-Preiser C. Immunohistologic analysis of gastrointestinal and pulmonary carcinoid tumors. HUM PATHOL 1998;29:992-9. [13] Hochwald SN, Zee S, Conlon KC, et al. Prognostic factors in pancreatic endocrine neoplasms: an analysis of 136 cases with a proposal for low-grade and intermediate-grade groups. J Clin Oncol 2002;20:2633-42. [14] Capella C, Heitz PU, Hofler H, Solcia E, Kloppel G. Revised classification of neuroendocrine tumours of the lung, pancreas and gut. Virchows Arch 1995;425:547-60. [15] Kloppel G, Heitz PU, Capella C, Solcia E. Pathology and nomenclature of human gastrointestinal neuroendocrine (carcinoid) tumors and related lesions. World J Surg Feb 1996;20:132-41. [16] Kloppel G, Perren A, Heitz PU. The gastroenteropancreatic neuroendocrine cell system and its tumors: the WHO classification. Ann N Y Acad Sci 2004;1014:13-27. [17] DeLellis RA, Lloyd RV, Heitz PU, Eng C, editors. World Health Organization classification of tumours, pathology, and genetics of tumours of endocrine organs. Lyon: IARCPress; 2004. [18] Kloppel G, Solcia E, Capella C, Heitz PU. Classification of neuroendocrine tumours. Ital J Gastroenterol Hepatol 1999;31 (Suppl 2):S111-6. [19] Rindi G, Kloppel G. Endocrine tumors of the gut and pancreas tumor biology and classification. Neuroendocrinology 2004;80(Suppl 1): 12-5. [20] Wick MR, Graeme-Cook FM. Pancreatic neuroendocrine neoplasms: a current summary of diagnostic, prognostic, and differential diagnostic information. Am J Clin Pathol 2001;115(Suppl):S28-S45. [21] Wick MR. Neuroendocrine neoplasia. Current concepts. Am J Clin Pathol 2000;113:331-5. [22] Van Eeden S, Quaedvlieg PF, Taal BG, Offerhaus GJ, Lamers CB, Van Velthuysen ML. Classification of low-grade neuroendocrine tumors of midgut and unknown origin. HUM PATHOL 2002;33:1126-32. [23] Tomassetti P, Campana D, Piscitelli L, et al. Endocrine pancreatic tumors: factors correlated with survival. Ann Oncol 2005;16:1806-10.
1268 [24] Panzuto F, Nasoni S, Falconi M, et al. Prognostic factors and survival in endocrine tumor patients: comparison between gastrointestinal and pancreatic localization. Endocr Relat Cancer 2005;12: 1083-92. [25] Rorstad O. Prognostic indicators for carcinoid neuroendocrine tumors of the gastrointestinal tract. J Surg Oncol 2005;89:151-60. [26] Furlan D, Cerutti R, Uccella S, et al. Different molecular profiles characterize well-differentiated endocrine tumors and poorly differ-
J. Strosberg et al. entiated endocrine carcinomas of the gastroenteropancreatic tract. Clin Cancer Res 2004;10:947-57. [27] Rindi G, Kloppel G, Alhman H, et al. TNM staging of foregut (neuro) endocrine tumors: a consensus proposal including a grading system. Virchows Arch 2006;449:395-401. [28] Rindi G, Kloppel G, Couvelard A, et al. TNM staging of midgut and hindgut (neuro) endocrine tumors: a consensus proposal including a grading system. Virchows Arch 2007;451:757-62.
www.elsevier.com/locate/humpath
Original contribution
Correlation between grade and prognosis in metastatic gastroenteropancreatic neuroendocrine tumors Jonathan Strosberg MD a,⁎, Aejaz Nasir MD a , Domenico Coppola MD a , Mark Wick MD b , Larry Kvols MD a a
H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL 33612, USA University of Virginia Medical Center, Charlottesville, VA 22908, USA
b
Received 6 November 2008; revised 14 January 2009; accepted 22 January 2009
Keywords: Neuroendocrine tumors; Carcinoid tumors; Pancreatic endocrine tumors; Islet cell tumors; Grade; Differentiation; Prognosis
Summary Three-tiered grading systems (low, intermediate, and high grade) have been proposed for neuroendocrine tumors. These classifications have not been rigorously evaluated in neuroendocrine malignancies of the digestive tract. We performed a retrospective chart analysis of 83 patients with metastatic gastroenteropancreatic neuroendocrine tumors, correlating tumor grade with overall survival. We also analyzed available biopsy specimens (on 40 patients), examining hematoxylin and eosin stains for mitotic rate and immunostaining for measurement of the Ki-67 index. Tumor grades were assigned based on the mitotic rate and the Ki-67 index, and the prognostic validity of each grading method was assessed. A highly significant correlation existed between the reported tumor grade and overall survival. Five-year survival rates for patients with low-, intermediate-, and high-grade tumors were 87%, 38%, and 0%, respectively. On biopsy specimen analysis, both mitotic rates and Ki-67 indexes correlated strongly with overall survival. We conclude that a 3-tiered grading classification for gastroenteropancreatic neuroendocrine tumors correlates with survival in the metastatic setting. Both mitotic rates and Ki-67 indexes are inversely associated with survival and can be analyzed independently for assignment of grade. © 2009 Elsevier Inc. All rights reserved.
1. Introduction Recent grading classifications for neuroendocrine tumors (NETs) are based on growing evidence that histologic markers of proliferation and differentiation correlate strongly with prognosis. The initial distinction between “typical” and “atypical” neuroendocrine histology originated with an analysis by Arrigoni et al [1] of 201 bronchial ⁎ Corresponding author. E-mail address: [email protected] (J. Strosberg). 0046-8177/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.humpath.2009.01.010
carcinoid tumors in the early 1970s. Various morphological factors including mitotic rate, cell pleomorphism, and necrosis were found to differentiate between indolent and aggressive tumor categories. The grading of pulmonary NETs was further refined by Travis et al [2] who distinguished between low-grade (typical carcinoid), intermediate-grade (atypical carcinoid), and high-grade (small cell and large cell neuroendocrine carcinoma) categories, based primarily on mitotic rates (b2, 2-10, and N10 mitoses per 10 high-power field [HPF], respectively). This 3-tiered classification was shown to correlate strongly with survival in an analysis of 200 tumor specimens.
Correlation between grade and prognosis in metastatic GEP-NETs Similar grading systems have been proposed for NETs of the thymus. Rosai et al [3] proposed classifying these tumors as carcinoid grade 1 (low grade), grade 2 (intermediate grade or atypical), and grade 3 (equivalent to small cell carcinoma). Moran and Suster [4] recommended a similar classification that abandoned the term carcinoid and replaced it with the more contemporary neuroendocrine carcinoma. This grading system defined well-differentiated neuroendocrine carcinomas as having low mitotic activity (3 or fewer mitoses per 10 HPF), minimal cytologic atypia, and small foci of necrosis; moderately differentiated carcinomas as having increased mitotic activity (4-10 mitoses per 10 HPF), areas of necrosis, and moderate cytologic atypia; and poorly differentiated carcinomas as having high mitotic activity (N10 mitoses per 10 HPF), extensive necrosis, and marked nuclear atypia. Analysis of the Ki-67 labeling index, typically identified by the MIB-1 monoclonal antibody, has offered an alternative immunohistochemical technique for assessing prognosis in NETs. Several studies have validated the role of Ki-67 staining in grading NETs of the lung [5,6] and gastrointestinal tract [7-10]; however, some investigators have questioned the use of Ki-67 as an independent prognostic variable [11-13]. Gastroenteropancreatic NETs (GEP-NETs) are predominantly indolent low-grade neoplasms; however, some are histologically and clinically aggressive. In the past, malignant GEP-NETs have been separated into 2 grades consisting of well-differentiated (low-grade) and poorly differentiated (high-grade) carcinomas [14-16]. Distinguishing characteristics of low-grade versus high-grade tumors have not been uniformly defined in the literature. Some sources have emphasized standard histologic criteria (atypia, necrosis, increased mitotic rate) [14-18], whereas others have focused on elevations in the Ki-67 index (N30%) as a defining feature of high-grade tumors [19]. In recent years, several authors have proposed a 3-tiered grading system for GEP-NETs modeled after the established categories of pulmonary and thymic neuroendocrine carcinomas [20,21]. Data supporting an intermediate-grade or “moderately differentiated” prognostic category of GEPNETs include the analysis of Hochwald et al [13] of 136 differentiated pancreatic endocrine tumors, identifying mitotic rate and necrosis as features distinguishing low versus intermediate-grade tumors. Similarly, an analysis by Van Eeden et al [22] of metastatic differentiated midgut carcinoid tumors identified a mitotic rate greater than 2 per 10 HPF as an independent negative prognostic factor. Other recent studies have focused on the Ki-67 index as a method of distinguishing low-grade from intermediate-grade tumors. For example, Tomassetti et al [23] identified a Ki-67 index of greater than 2.6% as a negative prognostic factor in patients with differentiated endocrine tumors of the ileum. This cutoff of approximately 2% was in agreement with results of other contemporary studies examining the prognostic role of Ki-67 in differentiated GEP-NETs [24-26]. The
1263
above data sets have motivated the European Neuroendocrine Tumor Society to propose a 3-tiered grading system for foregut [27], midgut [28], and hindgut [28] NETs in which low-grade (G1) tumors are defined by a mitotic rate of less than 2 per 10 HPF and/or Ki-67 index of 2% or less, intermediate-grade (G2) tumors defined by 2 to 20 mitoses per 10 HPF and/or Ki-67 index between 3% and 20%, and high-grade (G3) carcinomas defined by mitotic count greater than 20 per 10 HPF and/or Ki-67 index greater than 20%. It is thus evident, that 3-tiered grading systems for GEPNETs have been adopted and translated into routine practice, although the criteria proposed by different groups do not yet reflect a clear consensus on how to define individual grades. At the H. Lee Moffitt Cancer Center, Tampa, FL, we have roughly followed parameters suggested by Travis et al [2], Moran and Suster [4], and Wick [21], evaluating mitotic rates and histologic criteria (atypia, pleomorphism, necrosis) for grading GEP-NETs. Ki-67 immunostaining has not been routinely performed for grading purposes. Our hypothesis is that this 3-tiered grading system significantly correlates with overall survival in patients with metastatic GEP-NETs and can serve as a valid prognostic factor. We have therefore investigated the impact of tumor grade (as reported in the medical record) on overall survival in a large data set of patients with metastatic GEP-NETs. We have also investigated the correlation of tumor grade with other end points such as somatostatin receptor expression and presence of pathologic hormonal secretion. Finally, we have performed histologic and immunohistochemical analyses on a subset of available biopsy specimens to more precisely define grading parameters and compare the validity of grading based on the mitotic rate versus the Ki-67 index.
2. Materials and methods 2.1. Step 1: Chart review We conducted a medical chart analysis of patients with metastatic GEP-NETs seen at the H. Lee Moffitt Cancer Center between 1999 and 2005, selecting all patients characterized on pathologic reports as having high-grade carcinomas (also termed poorly differentiated or grade 3) and intermediate-grade carcinomas (also termed moderately differentiated or grade 2) and randomly selecting an equivalent number of patients with the more common lowgrade carcinomas (also termed well-differentiated or grade 1]). Neuroendocrine tumors of unknown primary were included if metastatic spread predominantly involved the liver, mesentery, and/or other infradiaphragmatic organs. Kaplan-Meier methodology was used to evaluate overall survival, stratified by tumor grade (as reported in the medical record). Other clinical features examined included presence of pathologically elevated hormone levels and evidence of
1264
J. Strosberg et al.
somatostatin receptor expression (as determined by detection of radiotracer uptake on 111In pentetreotide scintigraphy).
tissue included nonneoplastic lymph node, following the manufacturer recommendations. The negative control included tissue sections incubated with normal rabbit serum.
2.2. Step 2: Histologic analysis of NECA specimens 2.4. Step 4: Statistical analysis A team of 2 pathologists (A. Nasir and D. Coppola) with expertise in neuroendocrine malignancies evaluated hematoxylin and eosin–stained sections on all excisional and core biopsy specimens available among the list of the above patients. A mitotic rate was assessed in each case (number/ 10 HPF), and the presence or absence of tumor necrosis was documented.
2.3. Step 3: Ki-67 immunostaining and determination of proliferation index Sections from tumors were then submitted for immunohistochemical examination to evaluate the Ki-67 index. Each sample was fixed in 10% neutral buffered formalin for 9 hours. After fixation, the tissue samples were processed into paraffin blocks. Immunohistochemical staining was performed on Dako autostainer (Dako, Carpinteria, CA) using 3-μm-thick paraffin sections from each of the representative tumor block selected. The sections were deparaffinized, rehydrated, and incubated with 3% H2O2 to block endogenous peroxidase. After high-temperature steam-induced epitope retrieval, the sections were incubated with a rabbit polyclonal antibody to Ki-67 (K2, Ventana Medical Systems Product no. 790-2910, Cell Conditioning 1 Solution, Standard, proprietary dilution) for 60 minutes according to the manufacturer's instructions. The sections were then washed in TBS and incubated with a biotinylated secondary antibody (Dako), followed by strep-avidin-biotin complex (Dako) for 30 minutes. A solution of 3,3′diaminobenzidene tetrahydrochloride (Sigma, St Lois, MO) was used as a chromogen followed by sodium azide and 20 μL of H2O2 in 100 mL of Tris-HCl (50 μmol/L, pH 7.6). After light counterstain with Harris hematoxylin, the sections were examined under light microscope. The sections stained for Ki-67 were independently evaluated by 2 expert pathologists (A. Nasir and D. Coppola) without prior knowledge of the clinicopathologic data. The Ki-67 index was determined by assessing the percentage of positively staining tumor cell nuclei. These evaluations were made in the most viable areas of the stained tumor sections that also showed the highest degree of nuclear labeling. Any difference of interpretation was resolved by joint immunohistochemical review by the 2 pathologists. Positive control Table 1
The mitotic rates and Ki-67 indexes were entered on a scatterplot graph, and linear regression analysis was performed to assess the correlation between the 2 prognostic variables. Tumors were then assigned a grade based on mitotic rate (mitotic grade) and a separate grade based on the Ki-67 proliferation index (Ki-67 grade). Survival was estimated using Kaplan-Meier methodology stratified by both grading systems to assess the prognostic power of one model (mitotic grade) versus the alternative model (Ki-67 grade).
3. Results A total of 83 cases of metastatic GEP-NETs were analyzed. All 27 cases of high-grade (poorly differentiated) tumors and all 28 cases of intermediate-grade (moderately differentiated) tumors found on chart review were selected for analysis. Among several hundred patients with lowgrade (well-differentiated) tumors, 28 were randomly selected for analysis.
3.1. Patient and tumor characteristics 3.1.1. Age The median age at diagnosis was 57 years (range, 23-83 years). There was no significant correlation between tumor grade and age. 3.1.2. Sex The entire cohort is composed of 50 males and 33 females. Males represented 39% of the low-grade, 64% of the intermediate-grade, and 78% of the high-grade cohort (P b .025). 3.1.3. Primary tumor site The most common sites of primary tumor were small bowel (28 patients), pancreas (24 patients), colon (10 patients), and rectum (5 patients). Twelve patients had metastatic tumors with unknown primaries that were suspected to be gastrointestinal based on their infradiaphragmatic pattern of metastatic disease. There were no significant
Sites of metastatic dissemination
Low grade Intermediate grade High grade
Liver
Mesentery
Lungs
Bone
Brain
Ovary
Other LN
25 26 23
7 8 1
0 3 2
0 5 6
0 1 5
1 2 0
2 0 5
Correlation between grade and prognosis in metastatic GEP-NETs
1265
differences in the distribution of primary tumors among the low-grade and intermediate-grade cohort (preponderance of pancreatic and ileal primary tumors). In the high-grade cohort, a substantially higher fraction of primary tumors were colorectal or occult. 3.1.4. Sites of metastatic spread The liver was the most common site of metastases (89%) followed by the mesentery (19%) and bone (11%). Skeletal metastases were observed only in intermediate- and highgrade tumors. Brain metastases occurred in 5 high-grade tumors and one intermediate-grade tumor but were not observed in low-grade tumors (Table 1). 3.1.5. Somatostatin receptor status Among 27 patients with low-grade tumors who underwent somatostatin receptor scintigraphy (OctreoScans), all (100%) had evidence of radiotracer uptake, indicating presence of somatostatin receptors. Thirteen (56%) of 23 scanned patients with intermediate-grade tumors and only 1 (14%) of 7 scanned patients with high-grade tumors had positive OctreoScans (Fig. 1). 3.1.6. Secretory activity Eighteen (64%) of 28 patients with low-grade tumors, 14 (50%) of 28 patients with intermediate-grade tumors, and 2 (7%) of 27 patients with high-grade tumors had clinical and laboratory evidence of hormonal syndrome (Fig. 2). The carcinoid syndrome was most commonly observed. Other hormonal syndromes included VIPoma, glucagonoma, insulinoma, gastrinoma, and Cushing's syndrome.
3.2. Survival analysis 3.2.1. Median survival From diagnosis of metastases until death from any cause, median survival was 13 months in the high-grade cohort, was
Fig. 1 Percentage of tumors with radiotracer uptake on somatostatin receptor scintigraphy.
Fig. 2 Percentage of tumors associated with a hormonal syndrome.
39 months in the intermediate-grade cohort, and was not reached in the low-grade cohort of patients (Fig. 3). 3.2.2. Two-year and 5-year survival rates The estimated survival rates at 2 years (by Kaplan-Meier analysis) were 100% in the low-grade, 57% in the intermediate-grade, and 22% in the high-grade cohort. At 5 years, the respective estimated survival rates were 87%, 38%, and 0% (Table 2).
3.3. Pathologic specimen analysis Among the 83 cases selected for chart review, pathologic biopsy specimens were available on 40 patients. The specimens included 29 surgical excisions, 9 core needle
Fig. 3 Overall survival of patients with metastatic GEP-NETs stratified by tumor grade.
1266
J. Strosberg et al.
Table 2 One-year, 2-year, and 5-year survival rates of patients with metastatic GEP-NETs, stratified by tumor grade
12 mo 24 mo 60 mo
Low grade (%)
Intermediate grade (%)
High grade (%)
P
100 100 87
78 57 37
51 22 0
b.001 b.001 b.001
biopsies, and 2 endoscopic biopsies. An analysis of mitotic rate (number/10 HPF) and Ki-67 proliferation index was performed in each case. Most of the specimens belonged to the low-grade and high-grade cohorts, with a relatively small number of intermediate-grade specimens available for histologic examination. 3.3.1. Correlation between mitotic rate and Ki-67 index On linear regression analysis, a highly significant correlation between the mitotic rate and the Ki-67 index was observed with a Pearson correlation coefficient of 0.90 (Fig. 4). 3.3.2. Assignment of grade based on mitotic rate and Ki-67 index Because of the relatively small number of biopsy specimens (particularly in the intermediate-grade cohort), a 2-tiered (rather than a 3-tiered) grading system was used. Tumors with mitotic rates of 0 to 10/10 HPF were classified as low/ intermediate grade, whereas those with mitotic rates greater than 10/10 HPF were classified as high grade. Likewise, tumors with a Ki-67 index of 0% to 20% were classified as low/intermediate grade, whereas those with a Ki-67 index greater than 20% were classified as high grade. Using these grading thresholds, there was a complete agreement between the mitotic and Ki-67 grades with the exception of a single specimen that was classified as low/intermediate grade by mitotic rate but as high-grade by Ki-67 index.
Fig. 4 Correlation between the mitotic rate and Ki-67 proliferative index in patients with metastatic GEP-NETs.
Fig. 5 Overall survival stratified by mitotic rate (0-10 mitoses/10 HPF versus N10 mitoses/10 HPF).
3.3.3. Survival analysis stratified by mitotic grade and Ki-67 grade The classification of tumors into low/intermediate-grade and high-grade cohorts by mitotic rate and by Ki-67 index correlated strongly with survival (Figs. 5 and 6). Whether grades were defined based on the mitotic rate or the Ki-67 index, the survival rates were identical at 2 and 5 years (100% and 85%, respectively, in the low/intermediate-grade cohort; 23% and 0%, respectively in the high-grade cohort). This suggests that both grading criteria (mitotic rate and Ki67 index), if assessed objectively, can provide clinically relevant information.
Fig. 6 Overall survival stratified by the Ki-67 proliferative index (0%-20% versus N20%).
Correlation between grade and prognosis in metastatic GEP-NETs
4. Discussion Gastroenteropancreatic NETs are heterogeneous neoplasms with markedly distinct prognoses varying by grade. In our chart review of 83 metastatic GEP-NETs, we have found that a 3-tiered histologic grading system correlated strongly with overall survival (5-year survival rates of 87%, 38%, and 0% in low-, intermediate-, and high-grade cohorts, respectively). We have also demonstrated an inverse relationship between tumor grade and somatostatin receptor expression as well as an inverse association between grade and presence of a functioning hormonal syndrome. Our findings add evidence to the argument that a 3-tiered grading system is prognostically valid in GEP-NETs, as has been previously demonstrated in pulmonary [2] and thymic [4] NETs. Our comparison of the mitotic rate versus the Ki-67 index in 40 biopsy specimens demonstrated that these 2 markers of proliferation are highly correlated with each other. Using a mitotic rate of 10 per 10 HPF and a Ki-67 index of 20% as cutoff points, we found that both mitotic grade and Ki-67 grade were identical in their ability to distinguish goodprognosis versus poor-prognosis patients. We conclude that although immunohistochemical analysis of the Ki-67 index is not always necessary for assignment of grade, it represents a clinically relevant criterion to predict survival in patients with metastatic GEP-NETs, particularly when the mitotic rate is not easily assessed or when the clinical presentation is discordant with the histologic appearance of the tumor. The number of biopsy specimens available for pathologic review in our study was insufficient to establish a precise threshold for separating low-grade from intermediate-grade tumors. Until larger studies define unique prognostic parameters for GEP-NETs, we support a uniform 3-tiered grading system for NETs, modeled after the proposals of Moran and Suster [4] and Wick [21], that is, using mitotic rate thresholds of 0 to 3, 4 to 10, and greater than 10 as the primary method of distinguishing low-, intermediate-, and high-grade tumors. A consistent and standardized system for grading NETs of the lungs, mediastinum, and gastroenteropancreatic tract would add much clarity to NET taxonomy and aid in stratification of future clinical trials.
4.1. Conclusions A 3-tiered grading classification for GEP-NETs (low grade, intermediate grade, and high grade) correlates strongly with survival in the metastatic setting. The mitotic rate (number of mitoses per 10 HPF) is strongly associated with the Ki-67 proliferation index. Both markers of proliferation are inversely correlated with survival and can be analyzed independently for assignment of grade.
References [1] Arrigoni MG, Woolner LB, Bernatz PE. Atypical carcinoid tumors of the lung. J Thorac Cardiovasc Surg 1972;64:413-21.
1267
[2] 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 1998;22:934-44. [3] Rosai J, Levine G, Weber WR, Higa E. Carcinoid tumors and oat cell carcinomas of the thymus. Pathol Annu 1976;11:201-26. [4] Moran CA, Suster S. Neuroendocrine carcinomas (carcinoid tumor) of the thymus. A clinicopathologic analysis of 80 cases. Am J Clin Pathol 2000;114:100-10. [5] Costes V, Marty-Ane C, Picot MC, et al. Typical and atypical bronchopulmonary carcinoid tumors: a clinicopathologic and KI-67– labeling study. HUM PATHOL 1995;26:740-5. [6] Bohm J, Koch S, Gais P, Jutting U, Prauer HW, Hofler H. Prognostic value of MIB-1 in neuroendocrine tumours of the lung. J Pathol 1996; 178:402-9. [7] von Herbay A, Sieg B, Schurmann G, Hofmann WJ, Betzler M, Otto HF. Proliferative activity of neuroendocrine tumours of the gastroenteropancreatic endocrine system: DNA flow cytometric and immunohistological investigations. Gut 1991;32:949-53. [8] Moyana TN, Xiang J, Senthilselvan A, Kulaga A. The spectrum of neuroendocrine differentiation among gastrointestinal carcinoids: importance of histologic grading, MIB-1, p53, and bcl-2 immunoreactivity. Arch Pathol Lab Med 2000;124:570-6. [9] Rindi G, Azzoni C, La Rosa S, et al. ECL cell tumor and poorly differentiated endocrine carcinoma of the stomach: prognostic evaluation by pathological analysis. Gastroenterology 1999;116:532-42. [10] La Rosa S, Sessa F, Capella C, et al. Prognostic criteria in nonfunctioning pancreatic endocrine tumours. Virchows Arch 1996; 429:323-33. [11] Zimmermann ME, Bosman FT. Proliferative activity of well differentiated neuroendocrine tumours of the gut. Histol Histopathol 2003;18:353-8. [12] Al-Khafaji B, Noffsinger AE, Miller MA, DeVoe G, Stemmermann GN, Fenoglio-Preiser C. Immunohistologic analysis of gastrointestinal and pulmonary carcinoid tumors. HUM PATHOL 1998;29:992-9. [13] Hochwald SN, Zee S, Conlon KC, et al. Prognostic factors in pancreatic endocrine neoplasms: an analysis of 136 cases with a proposal for low-grade and intermediate-grade groups. J Clin Oncol 2002;20:2633-42. [14] Capella C, Heitz PU, Hofler H, Solcia E, Kloppel G. Revised classification of neuroendocrine tumours of the lung, pancreas and gut. Virchows Arch 1995;425:547-60. [15] Kloppel G, Heitz PU, Capella C, Solcia E. Pathology and nomenclature of human gastrointestinal neuroendocrine (carcinoid) tumors and related lesions. World J Surg Feb 1996;20:132-41. [16] Kloppel G, Perren A, Heitz PU. The gastroenteropancreatic neuroendocrine cell system and its tumors: the WHO classification. Ann N Y Acad Sci 2004;1014:13-27. [17] DeLellis RA, Lloyd RV, Heitz PU, Eng C, editors. World Health Organization classification of tumours, pathology, and genetics of tumours of endocrine organs. Lyon: IARCPress; 2004. [18] Kloppel G, Solcia E, Capella C, Heitz PU. Classification of neuroendocrine tumours. Ital J Gastroenterol Hepatol 1999;31 (Suppl 2):S111-6. [19] Rindi G, Kloppel G. Endocrine tumors of the gut and pancreas tumor biology and classification. Neuroendocrinology 2004;80(Suppl 1): 12-5. [20] Wick MR, Graeme-Cook FM. Pancreatic neuroendocrine neoplasms: a current summary of diagnostic, prognostic, and differential diagnostic information. Am J Clin Pathol 2001;115(Suppl):S28-S45. [21] Wick MR. Neuroendocrine neoplasia. Current concepts. Am J Clin Pathol 2000;113:331-5. [22] Van Eeden S, Quaedvlieg PF, Taal BG, Offerhaus GJ, Lamers CB, Van Velthuysen ML. Classification of low-grade neuroendocrine tumors of midgut and unknown origin. HUM PATHOL 2002;33:1126-32. [23] Tomassetti P, Campana D, Piscitelli L, et al. Endocrine pancreatic tumors: factors correlated with survival. Ann Oncol 2005;16:1806-10.
1268 [24] Panzuto F, Nasoni S, Falconi M, et al. Prognostic factors and survival in endocrine tumor patients: comparison between gastrointestinal and pancreatic localization. Endocr Relat Cancer 2005;12: 1083-92. [25] Rorstad O. Prognostic indicators for carcinoid neuroendocrine tumors of the gastrointestinal tract. J Surg Oncol 2005;89:151-60. [26] Furlan D, Cerutti R, Uccella S, et al. Different molecular profiles characterize well-differentiated endocrine tumors and poorly differ-
J. Strosberg et al. entiated endocrine carcinomas of the gastroenteropancreatic tract. Clin Cancer Res 2004;10:947-57. [27] Rindi G, Kloppel G, Alhman H, et al. TNM staging of foregut (neuro) endocrine tumors: a consensus proposal including a grading system. Virchows Arch 2006;449:395-401. [28] Rindi G, Kloppel G, Couvelard A, et al. TNM staging of midgut and hindgut (neuro) endocrine tumors: a consensus proposal including a grading system. Virchows Arch 2007;451:757-62.