Fax: (011) 49-30-450553942
Neuroendocrine tumors of midgut and hindgut origin: Tumor-node-metastasis classification determines clinical outcome
Article first published online: 18 JAN 2011
Copyright © 2011 American Cancer Society
Volume 117, Issue 15, pages 3332–3341, 1 August 2011
How to Cite
Jann, H., Roll, S., Couvelard, A., Hentic, O., Pavel, M., Müller-Nordhorn, J., Koch, M., Röcken, C., Rindi, G., Ruszniewski, P., Wiedenmann, B. and Pape, U.-F. (2011), Neuroendocrine tumors of midgut and hindgut origin: Tumor-node-metastasis classification determines clinical outcome. Cancer, 117: 3332–3341. doi: 10.1002/cncr.25855
- Issue published online: 20 JUL 2011
- Article first published online: 18 JAN 2011
- Manuscript Accepted: 27 OCT 2010
- Manuscript Revised: 15 OCT 2010
- Manuscript Received: 3 MAY 2010
- carcinoid tumor;
- neuroendocrine tumor;
- TNM staging;
- survival analysis;
- prognostic stratification
Prognostic classification of neuroendocrine tumor (NET) patients is difficult due to the complexity of current classification systems. A recent proposal for a tumor-node-metastasis (TNM) classification and a grading system based on the proliferative fraction proved valid in NETs of foregut origin. The purpose of this study was to test the efficacy of a proposal for TNM staging and grading for midgut and hindgut NETs.
Two hundred seventy patients with histologically proven midgut and hindgut NETs were investigated. Epidemiological, clinicopathological, and tumor-specific data at initial diagnosis were recorded. Tumors were classified according to the World Health Organization (WHO) and the recent European Neuroendocrine Tumor Society-TNM staging and grading proposal. Survival analysis and statistical testing for independent prognostic factors were performed using log-rank tests and Cox regression.
Of 270 NETs originating in the midgut or hindgut, 7% (5-year survival rate [YSR], 100%) were stage 1, 8% (5-YSR, 100%) were stage 2, 19% (5-YSR, 89.5%) were stage 3, and 66% (5-YSR, 83.3%) were stage 4 NETs; 62% (5-YSR 95.2%) were grade 1, 32% (5-YSR 82.0%) were grade 2, and 6% (5-YSR, 51.4%) were grade 3 NETs. WHO classification significantly separated poorly from well-differentiated NET or carcinomas but did not further discriminate. TNM staging significantly separated stages 1, 2, and 3 from stage 4 NETs, as did grading according to proliferative capacity for all grades. Multivariate analysis confirmed these results, particularly for Ki67 grading.
The acquired data confirmed the prognostic relevance of the proposed TNM staging and grading system and demonstrated the applicability of these classification tools. The TNM system can therefore facilitate therapeutic stratification and comparison of data from different institutions. Cancer 2011. © 2011 American Cancer Society.
Neuroendocrine tumors (NETs) are rare epithelial neoplasms with specific tumor biological features and challenging diagnostic and therapeutic problems.1-4 NETs most frequently occur in the gastroenteropancreatic (GEP) system5, 6 and frequently manifest clinically via paraneoplastic hormone hypersecretion or functionality.3 The most frequent functional syndrome is the carcinoid syndrome,7-10 which is frequently observed in metastatic small intestinal NETs, which probably comprise the largest subgroup of GEP NETs.5, 6, 11-13 The term “neuroendocrine tumor” with reference to the respective organ of primary tumor localization has replaced the older term “carcinoid,” because the latter term was confusing and misleading.1, 2, 4, 14-16 GEP NETs are heterogeneous, mostly malignant neoplasms with a wide variety of clinical courses ranging from small, benign tumors curatively resected via endoscopy to highly malignant, diffusely metastasizing tumors with rapid progression and death shortly after diagnosis.1, 6, 7, 11, 13 Categorization and classification of GEP NETs has been difficult and to some extent confusing since 1907,1, 2 and only recent classification attempts have lead to some systematic classification with prognostic relevance.14, 17
Several prognostically relevant factors have been identified, including size of the primary tumor, presence of metastasis, clinical symptoms, proliferative index, chromogranin A expression, and successful surgical resection.6, 13, 18-21 In addition, histopathological characteristics such as local invasiveness of the primary tumor, histological differentiation grade, and angio-, lymphangio-, or neural invasion have been used for pathological classification purposes.14, 17 However, the classifications proposed in 1995 by Capella et al17 and by the World Health Organization (WHO) in 200014 have not achieved wide acceptance, likely due to lack of clinical information for the classifying pathologist, unfamiliarity in less experienced centers, and hesitancy to adopt a complex classification system.2, 4 In 2006, an international consensus conference organized by the European Neuroendocrine Tumor Society (ENETS) proposed a tumor-node-metastasis (TNM) classification system for foregut NETs based on features specific for NET entities4 that subsequently proved valid and easily applicable.22-25 In particular, the suggested grading system—which uses either Ki67-index or mitotic count—was shown to be of strong prognostic value in foregut NETs22 or pancreatic NETs from a surgical,23 oncological,24 or pathological25 cohort. Furthermore, staging according to the proposed TNM system was a valid predictor of long-term outcome in all of these studies.22-25 Locoregional disease, however, was not significant in all studies.23, 24 However, in these cohorts, patients with successful oncological resection had an improved outcome, thus surgical intervention was the more relevant prognostic factor rather than the preoperative tumor load. Taken together, in the foregut—or more specifically, in pancreatic NETs—the TNM staging und grading system as proposed by the ENETS demonstrates simple applicability and strong predictive value for the prognostic stratification of these patients, a combination that makes it attractive for a wider use in routine and scientific use.
In 2007, a comparable proposal for NET of midgut and hindgut origin again based on tumor-specific literature was published26 by ENETS but has not yet been validated. The proposal includes the simple grading system based on the proliferative capacity of NET determined either by mitotic index or Ki67-labeling index. The proliferative fraction proved to be one of the most potent independent prognostic factors in NET independently from primary tumor anatomical site.6, 18, 21, 27-30
To our knowledge, this is the first validation study of the new ENETS-TNM staging and grading system in an international bicentric cohort of well characterized midgut and hindgut NETs compared with current classification systems. Furthermore, NET-specific features such as functionality, primary tumor localization, and histological differentiation were considered. Survival analysis was performed, and the relevance of prognostic stratification according to the newly proposed TNM classification including the grading system defined.
MATERIALS AND METHODS
The medical records of 270 patients with histopathologically proven GEP NET from the midgut or hindgut treated between 1984 and 2008 at either at the Department of Hepatology and Gastroenterology at the Charité in Berlin, Germany (n = 207), or at the Department of Gastroenterology and Pancreatology at Hŏpital Beaujon, in Clichy, France (n = 63) were retrospectively analyzed. Histopathological confirmation of the diagnosis of a NET was required before inclusion in the study. Primary tumor localization was confirmed either via endoscopic or surgical proof of the primary tumor (n = 253) or by conclusive imaging studies such as somatostatin receptor imaging and computed tomography or magnetic resonance imaging (n = 17). The patient files were systematically reviewed for date of initial diagnosis, localization of the primary tumor, histopathological diagnosis, tissue site from which samples were taken for histopathological diagnosis, clinical staging at initial diagnosis (ie, results of imaging and surgical procedures), and presence and quality of a clinically manifest hormone hypersecretion, including transthoracic echocardiography for right-heart involvement in suspected carcinoid heart disease. All patient files were reviewed, and disease and tumor-specific information was recorded (H. J., O. H.). Data were re-evaluated for accuracy and consistency (U.-F. P.). Survival data were obtained via chart review.
Special attention was paid to histopathological data, as acquired from the 2 contributing institutions. The histopathological diagnosis was initially made in 1 of the 2 centers in 153 (58.6%) patients. In the remaining 113 (41.4%) patients, the histopathological diagnosis was reviewed, revised, and/or completed when incorrect or incomplete. This was performed by a staff pathologist from either of the 2 contributing institutions (A. C., M. K., C. R.). Specifically, histological differentiation grade, immunohistochemistry, mitotic index, and Ki67-labeling index were amended as suggested by Rindi et al4, 26 if necessary and when sufficient tumor tissue was available. Based on the existing information, the WHO classification14 for histological typing of endocrine tumors and the recently published TNM staging and grading system suggested by Rindi et al26 were applied whenever possible.
Statistical analysis was performed using SPSS software version 17.0.2 and SAS version 9.2 (SAS Institute, Cary, NC). Approximately normally distributed data are presented as the mean with standard deviation (SD), standard error of mean (SEM), and/or 95% confidence interval (CI). Otherwise, the median and range are given. Overall and disease-specific (ie, NET-related) survival were both calculated as primary endpoints of survival analysis using the Kaplan-Meier method and log-rank tests for univariate analysis of potentially prognostic factors. Analyzed prognostic factors were WHO classification subgroups, tumor stages, tumor grades, and presence or absence of a carcinoid syndrome. The hazard ratios of NET-related death for both TNM staging and tumor grading were calculated using Cox regression models in a multivariate analysis. Grading and staging subgroups as analyzed by univariate analysis were included in the multivariate model for Cox regression analysis. Cox regression analysis was performed unadjusted and adjusted for age and sex. To account for groups without events (no NET-related deaths), Firth's modification for maximum likelihood estimation was used. P<.05 was considered statistically significant; all tests were 2-sided. No adjustments for multiple comparisons were performed.
Two hundred seventy cases of midgut and hindgut NETs treated at both institutions were retrospectively analyzed. Sex distribution in the cohort was almost even (men/women, 134/136), mean age (± SD) at initial diagnosis was 54.6 ± 13.6 years (median, 55 years; range, 10-85 years). A carcinoid syndrome was present at initial diagnosis in 115 cases (42.6%); no information on presence or absence of a hormone hypersecretion syndrome was available in 7 cases. Primary tumor localizations, distribution of TNM stages and grades, WHO classification categories, and cases with carcinoid syndrome at initial diagnosis in each group are given in Table 1. The mean follow-up time (± SD) was 53.9 ± 48.3 months (median, 42.5 months; range, 1-285 months). A selection bias toward more advanced and oncologically complicated cases may have played a role in the composition of the studied cohort, because both centers are national tertiary referral centers for NET. This was particularly true for the relatively high number of advanced appendiceal NETs.
|Site of Primary Tumor||Total (N=270)||Stagea (n=262)||Grading (n=189)||World Health Organization (n=215)|
|Jejunum||26||0 (0)||2 (0.8)||6 (2.3)||18 (6.9)||10 (5.3)||10 (5.3)||2 (1.1)||0 (0)||21 (9.8)||1 (0.5)|
|Carcinoid syndrome||9||—||0 (0)||2 (1.8)||7 (6.2)||3 (3.6)||3 (3.6)||0 (0)||0 (0)||7 (7.1)||0 (0)|
|Ileum||188||2 (0.8)||5 (1.9)||36 (13.7)||138 (52. 7)||86 (45.5)||43 (22.8)||2 (1.1)||1 (0.5)||153 (71.2)||1 (0.5)|
|Carcinoid syndrome||99||0 (0)||0 (0)||10 (8.8)||87 (77.0)||46 (55.4)||25 (30.1)||0 (0)||0 (0)||83 (84.7)||1 (1.0)|
|Appendix||21||4 (1.5)||12 (4.6)||1 (0.4)||4 (1.5)||5 (2.6)||3 (1.6)||0 (0)||6 (2.8)||4 (1.9)||1 (0.5)|
|Carcinoid syndrome||2||0 (0)||0 (0)||0 (0)||2 (1.8)||1 (1.2)||1 (1.2)||—||0 (0)||2 (2.0)||0 (0)|
|Cecum||12||0 (0)||1 (0.4)||5 (1.9)||6 (2.3)||4 (2.1)||2 (1.1)||3 (1.6)||0 (0)||8 (3.7)||3 (1.4)|
|Carcinoid syndrome||2||—||0 (0)||0 (0)||2 (1.8)||1 (1.2)||1 (1.2)||0 (0)||0 (0)||2 (2.0)||0 (0)|
|Colon, ascending||2||0 (0)||1 (0.5)||1 (0.4)||1 (0.4)||1 (0.5)||0 (0)||1 (0.5)||0 (0)||1 (0.5)||1 (0.5)|
|Carcinoid syndrome||0||—||0 (0)||0 (0)||0 (0)||0 (0)||—||0 (0)||0 (0)||0 (0)||0 (0)|
|Colon, sigmoid||1||1 (0.4)||0 (0)||0 (0)||0 (0)||1 (0.5)||0 (0)||0 (0)||1 (0.5)||0 (0)||0 (0)|
|Carcinoid syndrome||0||0 (0)||—||—||—||0 (0)||—||—||0 (0)||0 (0)||0 (0)|
|Rectum||20||11 (4.2)||1 (0.4)||1 (0.4)||6 (2.3)||10 (5.3)||3 (1.6)||3 (1.6)||6 (2.8)||6 (2.8)||1 (0.5)|
|Carcinoid syndrome||2||1 (0.9)||1 (0.9)||0 (0)||0 (0)||1 (1.2)||0 (0)||1 (1.2)||1 (1.0)||1 (1.0)||0 (0)|
|Total||270||18 (6.9)||21 (8.0)||50 (19.1)||173 (66.0)||117 (61.9)||61 (32.3)||11 (5.8)||14 (6.5)||193 (89.8)||8 (3.7)|
|Carcinoid syndrome||114||1 (0.9)||1 (0.9)||12 (10.6)||98 (86.7)||52 (62.7)||30 (36.1)||1 (1.2)||1 (1.0)||95 (97.9)||1 (1.0)|
Sufficient information on categorization was available for the classifications according to Capella et al17 in 201 (74.4%) patients, the WHO of 200014 in 215 (79.6%) patients, and the TNM classification according to Rindi et al26 in 262 (97.0%) patients. Grading according to immunohistochemical Ki67-labelling was available in 189 (70.0%) patients. The distributions of documented grades in relation to tumor stages and primary tumor localizations are given in Table 2. TNM staging methods were pathological (pTNM) and clinical (cTNM). Adequate pTNM staging was possible in 136 (50.4%) patients; complete cTNM staging was possible in 222 (82.2%) patients.
|Stage 1||Stage 2||Stage 3||Stage 4|
Overall and NET-Related Survival
Fifty-three (19.6%) patients died during the follow-up period, 34 of them (64.1%) due to NET-related causes (eg, tumor cachexia, liver failure due to NET metastases, septic syndrome due to malignant ascites). The mean overall survival (± SEM) was 156.5 ± 12.9 months (95% CI, 131.1-181.8 months), the median overall survival was 161.0 months (range, 4-229 months) with 2-, 5-, and 10-year survival rates (YSRs) of 93.0%, 84.6%, and 58.4%, respectively. The mean NET-related survival (± SEM; calculated by censoring non-NET-related deaths) was 187.8 ± 15.7 months (95% CI, 157.1-218.6 months), the median NET-related survival was 229.0 (range, 4-229 months) with 2-, 5-, and 10-YSRs of 96.1%, 88.7%, and 70.1%, respectively. Because of the difference in survival times and rates, we chose to use NET-related survival as the primary endpoint of our analyses, assuming that it better reflects the natural course of the disease.
Survival Analysis of NET Classified According to WHO Classification
Of the 215 cases available for survival analysis after WHO classification,14 43 patients died, 28 of them due to NET-related causes. None of the patients classified as having well-differentiated endocrine tumor (WDET) (n = 14) died; none of them was followed up for 10 or more years. Table 3 shows data for 2-, 5- and 10-YSRs of patients with WDET or well-differentiated endocrine carcinoma (WDEC) (n = 193, 89.8% of all analyzed patients). Patients with poorly differentiated endocrine carcinoma (PDEC) (n = 8) demonstrated 2- and 5-YSR of 45.0%. PDEC showed a significantly shorter NET-related survival than WDEC (P = .003) and WDET (P = .012), whereas no statistically meaningful numbers were reached between WDET and WDEC (P = .358) (Fig. 1).
|Classification||No. of Deaths (%)||Mean Survival ± SEM (95% CI), mo||Survival Rate, %|
|2 Years||5 Years||10 Years|
|World Health Organization (2000)14||28/215 (13.0)|
|WDEC||24/193 (12.4)||160.9±8.4 (144.4-177.4)||97.2||89.5||69.9|
|PDEC||4/8 (50.0)||71.4±21.0 (30.2-112.6)||45.0||45.0||ND|
|Clinical TNM staging26||33/262 (12.6)|
|Stage 1||0/18 (0)||—||100.0||100.0||100.0|
|Stage 2||0/21 (0)||—||100.0||100.0||100.0|
|Stage 3||3/50 (6.0)||151.3±7.5 (136.6-166.0)||97.6||97.6||86.0|
|Stage 4||30/173 (17.3)||172.0±16.5 (139.7-204.3)||94.8||83.3||59.4|
|G1||6/117 (5.1)||243.9±17.2 (210.1-277.7)||99.0||95.2||80.3|
|G2||11/61 (18.0)||172.8±16.2 (141.1-204.6)||94.7||82.0||66.2|
|G3||5/11 (44.5)||89.6±32.8 (25.3-153.8)||51.4||51.4||34.3|
Survival Analysis of NET Classified According to TNM Staging
Two hundred sixty-two (97.0%) patients were available for analysis of cTNM staging according to Rindi et al.26 Fifty-one of these patients died, 33 of them due to NET-related causes. Table 3 provides data on death rates, mean survival, and 2-, 5-, and 10-YSRs; Table 4 provides data on survival rates in various parts of the midgut and hindgut. Survival analysis revealed a significantly poorer survival for stage 4 versus stage 3 tumors (P = .040) and for stage 4 tumors versus stage 1/2 tumors (P = .025) (Fig. 2A). Comparisons of stage 3 tumors with stages 1 and 2 (stage 1, P = .436; stage 2, P = .379; stage 1/2, P = .240) were not statistically significant. However, grouping stages 1, 2, and 3a together as “limited disease” (ie, neither locoregional lymph node nor distant metastasis) and comparing them against stages 3b and 4 as “extensive disease” (ie, locoregional lymph node and/or distant metastasis) showed again a statistically significant difference (P = .040) (Figure 2B). Similarly, analyzing stages 1 and 2 (ie, locally confined tumor growth) together against stages 3 and 4 (clearly invasive tumor growth) lead to statistically significant results (P = .045). Further analysis of TNM staging of 207 jejunoileal NETs, which correspond to the classical “carcinoid tumor,” revealed a clear tendency to improved survival with lower tumor stages but did not reach statistical significance due to the limited number of cases (stages 1/2 vs stage 3, P = .328; stages 1/2 vs stage 4, P = .155; stage 3 vs stage 4, P = .091).
|Tumor Group||Survival Rate, %|
|2 Years||5 Years||10 Years|
|Mid- and hindgut||96.1||88.7||70.1|
Survival Analysis of NET Classified According to Grade
Detailed survival analysis for NET-related survival according to the grading system proposed by Rindi et al26 was performed in 189 patients; 32 of them died, including 22 who died of NET-related causes. Details of death rates, median survival, and 2-, 5-, and 10-YSR are given in Tables 3 and 4. As shown in Tables 1 and 2, grade 3 NETs are very rare in the midgut and hindgut groups, particularly when compared with the overall number of cases in the individual localization. The majority were grade 1 (G1) (117/189, 61.9%) and grade 2 (G2) NETs (61/189; 33.3%). In both of these groups, a carcinoid syndrome was present in almost half of all cases at initial diagnosis (G1, 52/117 [44.4%]; G2, 30/61 [49.2%]) (Table 1). Only 4 cases of grade 3 (G3) NET were found in the jejunoileum (4/153 [2.6%]), whereas 3 of 9 cases (33.3%) of cecal NET and 3 of 16 cases (18.8%) of rectal NET were G3 NET. Only one carcinoid syndrome was seen in 11 G3 NETs (9.1%), reflecting the extreme rarity of such a functional syndrome in this subgroup of NET.
NET-related patient survival in the whole cohort was significantly worse for G3 tumors when compared with G1 and G2 tumors (G1, P<.001; G2, P = .008) and for G2 compared with G1 tumors (P = .027) (Fig. 3). Additional analysis of jejunoileal NET (“carcinoid tumors”) showed similar results that did not reach statistical significance between G1 and G2 NET (G1 vs G2, P = .213; G1 vs G3, P = .001; G2 vs G3, P = .017) but a trend comparable to the results from the whole study cohort could be observed. It may be speculated that this was due to the smaller number of cases in the purely jejunoileal cohort because of the otherwise concordant results to the analysis of the whole cohort. However, tumor biology or factors caused by disease management, which were not an integral part of this study, may also have contributed to this result.
Influence of Carcinoid Syndrome
In 115 cases of carcinoid syndrome, no relevant sex preference (62 men, 53 women) was observed. A single case of a metastasized upper jejunal VIPoma was excluded from further analysis of influence of hypersecretion syndromes. Symptoms of carcinoid syndrome included facial flushing in 88.0%, diarrhea in 80.6%, and otherwise unexplained bronchospasm in 3.8% of all cases. Carcinoid heart disease was diagnosed in 19.6% of patients. Carcinoid syndrome was more frequently present in ileal (54.1%) than in jejunal (37.5%) NET. Although similarly frequent in G1 and G2 NET (see above), carcinoid syndrome was found in early stage NET (stages 1 or 2) only in 1 cecal and 2 rectal NET. In stage 3 NET, carcinoid syndrome was observed in 2 of 6 jejunal (33.3%) and 10 of 36 ileal (27.8%) NETs. In stage 4 NET, carcinoid syndrome was present in 37.8% of jejunal and 63.0% of ileal NETs (Table 1), 80.8% (84/104) of which showed liver metastases. Furthermore, 2 cases each of stage 4 appendiceal and cecal NET were noted.
Although there was a faint tendency toward a better 5- and 10-YSR in nonfunctional NET (5-YSR, 93.7%; 10-YSR, 79.7%) compared with patients with carcinoid syndrome (5-YSR, 85.6%; 10-YSR, 64.2%), this was not the case for 2-YSR (nonfunctional, 97.6%; carcinoid syndrome, 96.9%) or on log-rank testing regardless of the presence of a carcinoid syndrome at initial diagnosis (nonfunctional vs carcinoid syndrome, P = .146) or at any time later (nonfunctional vs carcinoid syndrome, P = .236) (Figure 4).
Cox Regression Analysis for TNM Staging and Tumor Grading
Because all 3 grading subgroups were significantly different from each other on univariate analysis, they were included in a multivariate Cox regression analysis with the lowest subgroup (G1) as the reference group (Table 5). Using this multivariate analysis, relative risk of death increased to approximately 3- and 11-fold (adjusted 3- and 18-fold) for G2 and G3 NET, respectively, compared with G1 NET. Although statistical significance was only reached for combined stages 1 and 2 as well as stage 3 each versus stage 4 on univariate analysis, we performed a Cox regression analysis that showed an increase of the relative risk for death in stage 3 and stage 4 NET by 4- and 10-fold (adjusted 3- and 8-fold, respectively) compared with combined stage 1 and 2 NETs. These results, however, were not statistically significant on multivariate analysis due to the relatively small numbers of endpoint events (ie, NET-related deaths).
|Variable||Hazard Ratio (95% CI), Unadjusted||P||Hazard Ratio (95% CI), Adjusted for Age and Sex||P|
|G2||2.9 (1.07-7.98)||.036||3.18 (1.15-8.80)||.026|
|G3||11.38 (3.46-37.43)||<.001||18.18 (5.04-65.58)||<.001|
|Stage 1/2 (reference)||1||1|
|Stage 3||3.6 (0.17-74.40)||.411||2.76 (0.13-58.23)||.514|
|Stage 4||10.2 (0.58-179.16)||.111||7.99 (0.45-141.37)||.156|
To our knowledge, this is the first study to demonstrate that the novel TNM classification system for midgut and hindgut neuroendocrine tumors (NET) as proposed by Rindi et al26 is applicable and valid for prognostic stratification of midgut and hindgut NET of the gastrointestinal tract. We demonstrate in a large patient cohort collected from 2 national referral centers in 2 different European countries that the proposed TNM staging system is statistically significantly able by univariate analysis to discriminate in a prognostically relevant manner between NET with only local (ie, stages 1 and 2) or locoregional (ie, stages 1, 2, and 3a; Figure 2B) spread and more extensive tumor disease (stages 3 and 4 or 3b and 4, respectively). Furthermore, the proposed grading system based on the proliferative capacity of the individual tumors as judged by Ki67 labeling index, a well-known prognostic factor in NET,6, 13, 14, 18, 27-30 proved to indicate statistically significant different prognosis between all 3 grades on univariate and multivariate analysis.
Classification of NET has been difficult since its first recognition1, 2 because of its heterogeneous tumor biology and clinical manifestations. The first classification by Williams and Sandler31 categorized NET according to the embryonic origin of the organ from which the NET derived, thereby separating foregut, midgut, and hindgut NET. Although this classification provides anatomical information, it is not prognostically meaningful per se, and other classification systems have been published.2, 14, 17 The most recent WHO classification system published in 200014 is increasingly being shown to be of prognostic relevance, mostly in foregut NET.22-25, 32, 33 However, the WHO classification is a highly complex classification system requiring clinical and pathological information that is often unavailable when a pathological diagnosis is reached, resulting in incomplete classifications and improper prognostic patient stratification.1, 4, 22, 26 Furthermore, it does not discriminate between different tumor stages of locally advanced or metastatic NET, which is of high clinical relevance. Finally, because TNM staging is common stratification practice for oncological patients, organ-specific TNM stages developed for adenocarcinomas have been improperly used for routine NET patient management, leading to incorrect prognostic estimation and inappropriate or insufficient treatment. This issue was also raised by a study reporting poor response to biotherapy in NET classified as WDEC by the WHO.33
The TNM staging and grading system proposed by ENETS4, 26 was found to be of prognostic value in foregut22 and more specifically in pancreatic NET.23-25 Therefore, improvement of prognostic classification can also be expected for midgut and hindgut NET, particularly jejunoileal NET. In addition, a recent consensus was reached between European and North American pathologists who have defined an essential data set for pathological reporting of neuroendocrine tumors.34 This group also stressed the importance of a TNM-based staging approach that takes clinical and imaging information into consideration, thereby advocating cTNM staging for optimal therapeutic decision making.34, 35 In addition, the recently published guideline for diagnosis and management of NET by the North American Neuroendocrine Tumor Society36 emphasizes the importance of both TNM staging and grading for adequate classification of NET. Finally, validation studies such as this one are demanded by all authors.26, 34-26 Until evidence from validation studies is available, the authors of these publications34-36 recommend detailed pathological findings in the reports, thereby facilitating future comparison between these recent histopathological classification systems.
In our study, the proposed TNM staging system was able to discriminate between a more benign prognosis in midgut and hindgut NET with either localized disease (stages 1 and 2) or with locoregional spread only (stages 1 to 3a, “limited disease”). This is in accordance with data from The Netherlands,37 where the 5-YSRs for local disease, regional disease, and distant metastasis were 93%, 74%, and 19%, respectively. In this population-based cohort, no significant differences depending on primary tumor localization were observed. However, these data stem from patients treated in the 1990s, the poorer outcome thus reflecting the few therapeutic options available at that time. More recently, data from the United States12 support a similar observation with decreasing median survival time and increasing tumour load. Single-institution data also report the dependence of survival on tumor spread.6, 13, 38, 39 Although the most reliable conclusions from our study can be drawn for jejunoileal NET, the largest subgroup of the present series, our TNM staging observation reflects the strict relationship existing between tumor spread and survival for mid- and hindgut NET.
No statistically significant difference was observed in stage 1 and 2 NET, the excellent outcome of these patients being most likely due to complete curative resection. The lack of statistically significant differences between stages 1/2 versus stage 3 is likely due to the fact that, in spite of few NET-related deaths in stage 3 patients, the overall analyzed case number was too small to detect statistically significant differences. However, it cannot be excluded that considerably larger patient numbers may reveal a statistically significant effect, as indicated by the trend observed in the current study. Similarly, the conclusions drawn from our data are mostly relevant to midgut NETs, because we included only a relatively small number of hindgut NETs. The reason for this is a selection bias: colonic NETs are very rare, whereas the less infrequent rectal NETs are mostly detected at early stages and therefore will not be presented to tertiary referral centers. Future studies possibly performed by some of the large, less selection-biased nationwide registries30 may contribute important data on this topic.
The proliferative capacity of NET is a well-characterized prognostic factor6, 13, 14, 18, 27-30; however, until the proposal by Rindi et al,26 no standard cutoff values were proposed for classification of gastrointestinal NET to allow treatment comparison in routine patient management or in clinical trials. Our results indicate that the proposed grading system, which relies on widely available histopathological techniques (ie, mitotic count or Ki67 labeling), in fact allows a statistically significant prognostic stratification as confirmed via multivariate analysis. The present data in mid- and hindgut NETs overlap with what was observed in foregut NETs,22-25 where a similar significant prognostic relevance of this grading system was demonstrated. A recent analysis from the Spanish Cancer Registry also supports the relevance of Ki67 index as an independent prognostic factor.30 Ki67 index was suggested as a tool for therapeutic stratification in NET,40 and the present observations strongly support this approach as a general strategy in NET. The advantage of the grading system will be the standardized classification allowing a more standardized stratification and treatment, leading to effective comparison in different series and hopefully improvement in patient care, as also suggested by the recent consensus statements.34-36
The presence or absence of a carcinoid syndrome in our analysis had no impact on the clinical outcome, although carcinoid syndrome was more frequently present in advanced tumor stages, particularly stage 4 NETs. This suggests that the presence of carcinoid syndrome is not an independent prognostic factor, but rather an epiphenomenon of tumor load with highly relevant clinical and therapeutic implications.
In conclusion, the proposed TNM staging and grading system has been shown to be valid in midgut and hindgut NET and particularly in jejunoileal NET. It has been shown to significantly stratify NET patients into prognostically relevant subgroups, in turn potentially allowing effective therapeutic stratification. This should support simpler and better routine management of NET patients and strengthen the conclusions drawn from clinical trials to provide better basis for treatment.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 8Carcinoid heart disease: the role of urinary 5-hydroxyindoleacetic acid excretion and plasma levels of atrial natriuretic peptide, transforming growth factor-beta and fibroblast growth factor. Cancer. 2003; 97: 1609-1615., , , , , .
- 14Histological typing of endocrine tumours. In: WHO: International Histological Classification of Tumours. Berlin, Germany: Springer; 2000., , .
- 15Prognostic factors in neuroendocrine lung tumors: a Spanish Multicenter Study. Spanish Multicenter Study of Neuroendocrine Tumors of the Lung of the Spanish Society of Pneumonology and Thoracic Surgery (EMETNE-SEPAR). Ann Thorac Surg. 2000; 70: 258-263., , , et al.
- 17Revised classification of neuroendocrine tumors of the lung, pancreas and gut. Vichows Arch. 1995; 452: 547-560., , , , .
- 28Endocrine 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. 1996; 27: 1124-1134., , , et al.