SEARCH

SEARCH BY CITATION

Keywords:

  • neuroendocrine tumors;
  • neuroendocrine neoplasm;
  • gastroenteric tract;
  • prognosis;
  • classification;
  • TNM;
  • World Health Organization

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

BACKGROUND:

Gastroenteric neuroendocrine neoplasms (GE-NENs) display highly variable clinical behavior. In an attempt to assess a better prognostic description, in 2010, the World Health Organization (WHO) updated its previous classification, and the European Neuroendocrine Tumors Society (ENETS) proposed a new grading and TNM-based staging system. In the current study, the authors evaluated the prognostic significance of these models and compared their efficacy in describing patients' long-term survival to assess the best prognostic model currently available for clinicians.

METHODS:

The study cohort was composed of 145 patients with extrapancreatic GE-NEN who were observed from 1986 to 2008 at a single center and were classified according to the WHO and ENETS classifications. Survival evaluations were performed using Kaplan-Meyer analyses on 131 patients. Only deaths from neoplasia were considered. A P value < .05 was considered significant. Prognostic efficacy was assessed by determining the Harrell concordance index (c-index).

RESULTS:

Both the 2010 WHO and the ENETS classification were able to efficiently divide patients into classes with different prognoses. According to the model comparison, the ENETS TNM-based staging system appeared to be the strongest. All combined models were effective prognostic predictors, but the model that included the 2010 WHO classification plus ENETS staging had a higher c-index.

CONCLUSIONS:

Both the 2010 WHO classification and the ENETS staging system are valid instruments for GE-NENs prognostic assessment, with TNM-based stage appearing to be the best available choice for clinicians, both alone and in association with other classifications. Cancer 2013. © 2012 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

The heterogeneity of neuroendocrine neoplasms (NENs) both in biology and in clinical behavior has led to the formulation of several guidelines for their diagnosis and treatment1, 2; however, until 2010, there was no unique classification or prognostic evaluation for these neoplasms.1, 3-7 A classification of endocrine tumors based on their site of origin8 comprises the foregut (respiratory system, stomach, duodenum, proximal jejunum, and pancreas), the midgut (distal jejunum, ileum, appendix, and right-sided colon), and the hindgut (transverse and left-sided colon and rectum) and has prognostic relevance.6, 9

A classification of gastroenteric NENs was proposed by the World Health Organization (WHO) in 2000 (WHO-2000)10 and was further radically updated in 2010 (WHO-2010).11, 12 WHO-2000 identified well differentiated NENs that had benign (WDET-B) or uncertain malignant (WDET-U) behavior, well differentiated neuroendocrine carcinomas (WDECs), and poorly differentiated neuroendocrine carcinomas (PDECs). This classification had prognostic relevance13-19 but did not discriminate benign behavior from low-grade malignant behavior.20

In an attempt to integrate WHO-2000, the European Neuroendocrine Tumor Society (ENETS) proposed a new set of criteria2, 4, 5, 21 for grading these tumors based on the Ki-67 index (grade 1, 0%-3%; grade 2, 3%-20%; and grade 3, 20%-100%) and for staging disease based on the TNM classification of gastroenteric NENs (GE-NENs) by dividing tumors into 5 stages: stage 0 (in situ invasion [Tis], no lymph node metastases [N0], and no distant metastases [M0]), stage I (T1/T1a/T1b,N0,M0: greatest tumor dimension ≤1 cm in the foregut and midgut and ≤2 cm in the hindgut [T1/T1a/T1b]), stage II (T2N0M0, T3N0M0: greatest tumor dimension >1 cm in the foregut and midgut and >2 cm in the hindgut [T2]; invasion of serosa into the stomach, lower jejunum, and ileum; invasion of the pancreas or retroperitoneum of the duodenum and proximal jejunum; and invasion of subserosa or pericolic or perirectal fat of the hindgut [T3]), stage III (T4,N0,M0 or anyT,N1,M0: invasion of other structures [T4] and the presence of lymph node metastases [N1]), and stage IV (anyT,anyN,M1: the presence of distant metastases [M1]). This classification had prognostic efficacy,13, 14, 7, 20, 22-26 but only a few studies have focused on extrapancreatic GE-NENs.23, 25, 26

In the more recent WHO-2010,11 purely endocrine neoplasms are divided into grade 1 neuroendocrine tumors (NETs), grade 2 NETs, and grade 3 neuroendocrine carcinomas (NECs), according not only to their histopathologic characterization but also to their proliferation index.12 The ENETS staging system is strongly recommended but is not considered necessary. Currently, this classification still lacks clinical validation.12

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

The Patient Sample

In total, 145 patients who underwent radical surgery or excisional biopsy of an extrapancreatic GE-NEN at the San Raffaele Scientific Institute during the period from 1986 to 2008 were included in the current study cohort. Histologic and immunohistochemical analyses were performed on 136 patients by the Pathology Unit of the same institute, and 9 patients were excluded for lack of adequate histologic material. Patients were followed by the Endocrine Tumors: Multidisciplinary Approach (TEAM) Group. Informed consent was obtained from all patients.

Diagnosis was based on conventional histology and immunohistochemical studies (chromogranin A, synaptophysin, neuron-specific enolase). Tumor staging was based on conventional imaging studies, including ultrasonography, computed tomography, and magnetic resonance, and on the assessment of local or metastatic spread during surgery. Tumor grading was performed through pathologic evaluation of the Ki-67 index and the mitotic count.

Patients who had NENs were classified according to: 1) the embryologic area of origin (foregut, midgut, or hindgut),8, 27 2) WHO-2000,10 3) ENETS classification4, 5 (information for TNM staging was available for 134 patients [96%], and information on grading was available for 105 patients [77%]), and 4) WHO-201011, 12 (information on this classification was available for 103 patients [76%]). Survival information was obtained for 131 of the 136 patients who were included in this study (2.9% were lost to follow-up). The cause of death was verified by reviewing death certificates or hospital records or by contacting the patients' families.

Statistical Analyses

Statistical analyses were performed with the Stata software package (version 11.1; StataCorp, College Station, Tex). Continuous data were expressed as the mean ± standard deviation if normally distributed. Normality was assessed with both the Skewness-Kurtosis test and the Shapiro-Wilk W test. Categoric variables were expressed as numbers and percentages. A P value < .05 was considered statistically significant. Continuous variables were compared with analyses of variance. Post-hoc multiple pairwise comparisons were performed using a 2-sided t test with Bonferroni correction. Categoric variables were compared using the Fisher exact test. Post-hoc multiple pairwise comparisons were performed using the Fisher exact test with Bonferroni correction.

Survival analysis was performed according to the Kaplan-Meier method. Survival was calculated from the date of diagnosis to the date of disease-specific death or the end of observation. Censoring was applied to the last follow-up date for patients who remained alive. Patients without disease-specific death were included in the analyses only if they had at least 30 months of follow-up. During the observation period, 14 patients died from causes unrelated to GE-NENs, and their survival was censored on the date of death.

Univariate analysis was used to identify potential predictors of survival by means of the log-rank test with P < .20 considered statistically significant. In this setting, age was categorized according to quartiles into 4 categories: first quartile, ages 0 to 45 years; second quartile, ages 46 to 60 years; third quartile, ages 61 to 70 years; and fourth quartile, aged ≥71 years. WHO-2000, WHO-2010, TNM stage, and grade were always considered categoric variables; and comparisons between scores were conducted using the lowest risk category of each scale as the reference category. In the survival analysis that included the TNM stage variable, 19 patients (13.9%) were excluded because they had disease classified as stage 0 or I, and no events were observed during the study in those categories.

Cox proportional hazards modeling was used to study the effect on survival of variables that were identified as significant in univariate analysis. Four models were built with WHO-2000, WHO-2010, TNM stage, or grade as the main predictor. WHO-2010 and grade models were adjusted for age and site, whereas the TNM stage model was adjusted only for age, because tumor site is part of TNM staging. Four additional models were built in which WHO-2000+grade (adjusted for age and site), WHO-2000+TNM stage, WHO-2010+TNM stage, and grade+TNM stage (adjusted for age) were the main predictors. Proportionality of hazards was verified for each variable by visual inspection of “log-log” plots and with the proportional-hazards assumption test based on Schoenfeld residuals.

To compare the discrimination of survival models, the Harrell concordance index (c-index) was calculated for each model.28 This index is defined as the proportion of usable patient pairs in which the outcome is concordant with the prediction produced by the model. A c-index value of 0.5 indicates no predictive discrimination, and perfect prediction of patient outcomes yields a c-index value of 1.0. Concordance indexes were then compared according to the method described by Newson.29 This method provides confidence limits and P values for differences between c-indexes of different models, thus allowing statistical comparison of model accuracy. Estimating a model's c-index in the same data set in which the model has been produced leads to overoptimistic estimates of predictive power, because the model is optimized for the data set in which the predictive power is measured. To avoid testing accuracy on the same data set that yielded the model (the so-called “training set”), the Jacknife data resampling technique was applied in calculating concordance indexes and their 95% confidence intervals. With this technique, the c-index is obtained by repeatedly computing it once for each observation in the data set, each time omitting the associated observation, and calculating the mean of the resulting values.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Sample Characteristics

Of the 136 patients with GE-NEN in this study, there were 70 men (51.5%) and 66 women (48.5%). All patients were Caucasians, and only 1 had a hereditary NENs syndrome (MEN1). Primary tumor sites were: esophagus (1 patient, 0.7%), stomach (22 patients, 16.2%), duodenum (20 patients, 14.7%), jejunum (2 patients, 1.5%), ileum (40 patients, 29.4%), gallbladder (2 patients, 1.5%), caecal appendix (32 patients, 23.5%), colon (10 patients, 7.3%), and rectum (7 patients, 5.2%). Of the 136 tumors, 43 tumors (31.1%) originated in the foregut, 76 (55.5%) originated in the midgut, and 17 (12.4%) originated in the hindgut. The mean patient age at diagnosis was 57 ± 18.4 years.

The characteristics of the sample and classification according to WHO-2000, WHO-2010, ENETS stage, and grade are summarized in Table 1. According to the WHO-2000 classification, the distribution of classes differed by sex (P = .031), because women more often developed a benign phenotype. The mean age at diagnosis differed significantly (P < .0001), and older patients more often developed poorly differentiated neoplasms. Finally, a significant difference was observed in the embryologic area of origin (P < .001): The midgut displayed the most benign phenotype, and the hindgut displayed the worst.

Table 1. General Characteristics of Patients and Classification According to the 2000 World Health Organization Classification, the 2010 World Health Organization Classification, and European Neuroendocrine Tumor Society Stage and Grade
    Site of Origin: No. (%)
ModelNo. of Patients (%)Women: No. (%)Age: Mean±SD, yForegutMidgutHindgut
  • Abbreviations: PDEC, poorly differentiated neuroendocrine carcinomas; SD, standard deviation; WDEC, well differentiated neuroendocrine carcinomas; WDET-B, well differentiated neuroendocrine neoplasms with benign behavior; WDET-U, well differentiated neuroendocrine neoplasms with uncertain malignant behavior; WHO-2000, 2000 World Health Organization classification; WHO-2010, 2010 World Health Organization classification.

  • a

    According to WHO-2010, purely endocrine neoplasms are divided into grade 1 neuroendocrine tumors, grade 2 neuroendocrine tumors, and grade 3 neuroendocrine carcinomas.

WHO-2000      
WDET-B33 (24.3)23 (34.9)48.9±22.910 (23.3)20 (26.3)3 (17.7)
WDET-U13 (9.6)7 (10.6)43.8±18.36 (14)6 (7.9)1 (5.9)
WDEC68 (50)28 (42.4)61.1±14.316 (37.2)47 (61.8)5 (29.4)
PEDC22 (16.2)8 (12.1)66.2±13.211 (25.6)3 (4)8 (47.1)
WHO-2010a      
Grade 171 (68.9)35 (68.6)57.8±18.218 (56.3)49 (87.5)4 (26.7)
Grade 211 (10.7)7 (13.7)64.4±9.74 (12.5)5 (8.9)2 (13.3)
Grade 221 (20.4)9 (17.7)68.9±11.110 (31.3)2 (3.6)9 (60)
TNM stage      
03 (2.2)1 (1.5)66.3±9.03 (7.1)0 (0)0 (0)
I16 (11.9)9 (13.4)54.1±22.76 (14.3)8 (11)2 (10.5)
II41 (30.6)23 (34.3)49.1±21.39 (21.4)30 (41.1)2 (10.5)
III33 (24.6)17 (25.4)64.9±13.311 (26.2)15 (20.6)7 (36.8)
IV41 (30.6)16 (23.9)63.2±10.813 (31)20 (27.4)8 (42.1)
Grade      
173 (69.5)35 (68.6)57.7±18.019 (57.6)50 (87.7)4 (26.7)
211 (10.5)7 (13.7)64.4±9.74 (12.1)5 (8.8)2 (13.3)
321 (20)9 (17.7)68.9±11.110 (30.3)2 (3.5)9 (60)

According to the WHO-2010 classification, classes did not differ between sexes (P = .535), but the mean age at diagnosis was significantly older in patients who had grade 2 tumors (P < .001) and grade 3 tumors (P < .001) compared with patients who had grade 1 tumors. The embryologic area of origin also was different (P < .001): The midgut, once again, displayed the most benign class, and hindgut displayed the worst.

No significant difference was observed in TNM stage according to patient sex (P = .54) or embryologic area of origin (P = .05). In contrast, the mean age at diagnosis varied significantly (P < .0001), in particular between patients with localized (stage II) and metastatic (stage III and IV) disease.

ENETS grade did not vary significantly between different sexes (P = .052), but a higher grade was observed in older patients (P = .01), in particular between grades 1 and 3 (P = .019). Similarly, tumors that originated in the midgut had a significantly lower grade than tumors that originated in the hindgut (P < .001).

Survival Study

Of 131 patients who had survival information available, 52 patients (39.6%) died from causes related to the GE-NEN. Survival rates after 12 months, 36 months, 60 months, and 120 months were 86%, 70%, 64%, and 59%, respectively. Six variables were predictive of survival in univariate analysis: age at diagnosis (P = .003), site of origin (P < .0001), WHO-2000 classification (P < .0001), WHO-2010 classification (P < .0001), TNM stage (P < .0001), and grade (P < .0001). When TNM variables were considered individually, the tumor (T) variable was not predictive of survival in univariate analysis (P = .02), whereas both the lymph node (N) variable (P < .0001) and the metastasis (M) variable (P < .0001) were predictive of survival. Sex was not predictive of survival (P = .2). Figure 1 illustrates the Kaplan-Meier curves with survival stratified according to each of the 6 significant predictors. Table 2 reports individual Cox models for WHO-2000, WHO-2010, TNM stage, and grade as predictors. In the WHO-2000 model (P < .0001), only patients with PDEC had significantly poorer survival than patients with WDET-B. In the WHO-2010 model (P < .0001), patients who had grade 3 tumors had poorer survival than patients who had grade 1 tumors (P < .0001) and grade 2 tumors (P = .024), and those who had grade 2 tumors had poorer survival than those who had grade 1 tumors (P = .030). In the TNM stage model (P < .0001), patients who had stage IV disease had poorer survival than patients who had stage III disease (P = .002), and those who had stage II disease (P < .0001) and stage III disease had poorer survival than patients who had stage II disease (P = .003). Regarding the “T” variable, patients who had locally advanced disease (T4) exhibited poorer survival compared with patients who had other T classifications (P < .05), although no statistical significance was observed between patients who had T1, T2, and T3 tumors. Regarding both the “N” and “M” variables, significantly poorer survival was observed among patients who had involved lymph nodes or distant metastases (P < .0001). In the grade model (P < .0001), patients who had grade 3 tumors exhibited poorer survival than patients who had grade 1 tumors (P < .0001) and grade 2 tumors (P = .03), and patients who had grade 2 tumors exhibited poorer survival than patients who had grade 1 tumors (P = .04).

thumbnail image

Figure 1. Estimated disease-specific survival was stratified according to (A) age at diagnosis, (B) embryologic area of origin, (C) 2000 World Health Organization (WHO) classification, (D) 2010 WHO classification, (E) European Neuroendocrine Tumor Society (ENETS) grade, and (F) ENETS stage. WDET-B indicates well differentiated neuroendocrine neoplasms with benign behavior; WDET-U, well differentiated neuroendocrine neoplasms with uncertain malignant behavior; WDEC, well differentiated neuroendocrine carcinomas; PDEC, poorly differentiated neuroendocrine carcinomas.

Download figure to PowerPoint

Table 2. Single Classification Models
ModelHR (95% CI)aP
  • Abbreviations: CI, confidence interval; HR, hazard ratio; PDEC, poorly differentiated neuroendocrine carcinomas; WDEC, well differentiated neuroendocrine carcinomas; WDET-B, well differentiated neuroendocrine neoplasms with benign behavior; WDET-U, well differentiated neuroendocrine neoplasms with uncertain malignant behavior; WHO-2000, 2000 World Health Organization classification; WHO-2010, 2010 World Health Organization classification.

  • a

    WHO-2000, WHO-2010, and HRs for grade were adjusted for age and origin area; HRs for TNM stage were adjusted for age.

  • b

    According to the WHO-2010, purely endocrine neoplasms are divided into grade 1 neuroendocrine tumors, grade 2 neuroendocrine tumors, and grade 3 neuroendocrine carcinomas.

WHO-2000  
WDET-B1.0
WDET-U0.8 (0.2-4.2).823
WDEC1.6 (0.6-3.4).325
PDEC4.8 (1.7-13.3).003
WDEC vs WDET-U1.9 (0.4-8.5).399
PDEC vs WDET-U5.7 (1.3-26.3).024
PDEC vs WDEC3.0 (1.4-6.6).005
WHO-2010b  
Grade 11.0
Grade 22.8 (1.2-7.3).030
Grade 39.6 (3.7-25.2)< .0001
Grade 2 vs grade 33.4 (1.2-9.8).024
TNM stage  
II1.0
III6.8 (1.9-24.1).003
IV18.4 (5.6-60.8)< .0001
IV vs III2.7 (1.5-5.0).002
Grade  
11.0
22.7 (1.04-6.77).04
38.5 (3.3-21.6)< .0001
3 vs 20.3 (0.1-0.9).03

The predictive discrimination of c-indexes of these models were compared. No model provided significantly more discrimination than another, except the TNM stage model (c-index = 0.78), which exhibited significantly better discrimination compared with the WHO-2000 model (c-index = 0.69; P = .036), but not compared with the grade model (c-index = 0.74; P = .268) or the WHO-2010 model (c-index = 0.73; P = .243).

Table 4. Comparison of the Prognostic Efficacy of Single and Combined Models
  P
Classification ModelHarrell C-Index (99% CI)Models 1-2Models 1-3Models 1-4Models 2-3Models 2-4Models 3-4
  1. Abbreviations: CI, confidence interval; WHO-2000, 2000 World Health Organization classification; WHO-2010, 2010 World Health Organization classification.

Single models       
1. WHO-20000.69 (0.62-0.77).179.125.036   
2. WHO-20100.73 (0.66-0.81).350.243 
3. Grade0.74 (0.66-0.81) .268
4. TNM stage0.78 (0.72-0.85)   
Combined models       
1. WHO-2000+grade0.73 (0.66-0.81).020.002.002   
2. WHO-2000+TNM stage0.82 (0.77-0.87).844.860 
3. WHO-2010+TNM stage0.83 (0.77-0.88) .484
4. TNM stage+grade0.83 (0.77-0.88)   

Four additional survival models were generated in which the association of WHO-2000+grade, WHO-2000+stage, WHO-2010+TNM stage, and TNM stage+grade were studied. All of these combined models were significantly predictive of survival (P < .0001) (Table 3). C-indexes of these models were compared (Table 4) to determine which model was best in stratifying patients' risk. The WHO-2000+grade model (index = 0.73) provided significantly less discrimination than the WHO-2000+TNM stage model (c-index = 0.82), the WHO-2010+TNM stage model (c-index = 0.83), and the TNM stage+grade model (c-index = 0.83; P = .020, P = .002, and P = .002, respectively).

Table 3. Combined Classification Models
Combined ModelHR (95% CI)P
  • Abbreviations: PDEC, poorly differentiated neuroendocrine carcinomas; SD, standard deviation; WDEC, well differentiated neuroendocrine carcinomas; WDET-B, well differentiated neuroendocrine neoplasms with benign behavior; WDET-U, well differentiated neuroendocrine neoplasms with uncertain malignant behavior; WHO-2000, 2000 World Health Organization classification; WHO-2010, 2010 World Health Organization classification.

  • a

    WHO-2000 HRs were calculated with constant grade classification, HRs for grade were calculated with constant WHO-2000 classification, and the model was adjusted for age and area of origin.

  • b

    WHO-2000 HRs were calculated with constant stage classification, HRs for TNM stage were calculated with constant WHO-2000 classification, and the model was adjusted for age.

  • c

    HRs for TNM stage were calculated with constant WHO-2010 classifications, HRs for WHO-2010 were calculated with constant TNM stage classification, and the model was adjusted for age.

  • d

    According to WHO-2010, purely endocrine neoplasms are divided into grade 1 neuroendocrine tumors, grade 2 neuroendocrine tumors, and grade 3 neuroendocrine carcinomas.

  • e

    HRs for TNM stage were calculated with constant grade classification, HRs for grade classification were calculated with constant TNM stage classification, and the model was adjusted for age.

WHO-2000+gradea  
WHO-2000  
WDET-B1.0
WDET-U0.9 (0.1-6.9).959
WDEC1.2 (0.3-5.5).769
PDEC1.7 (0.3-10.2).546
WDEC vs WDET-U1.3 (0.3-6.3).730
PDEC vs WDET-U1.8 (0.3-11.4).521
PDEC vs WDEC1.4 (0.5-3.8).526
Grade  
 11.0
 22.4 (0.9-6.4).092
 36.2 (1.8-21.1).003
 3 vs 22.6 (0.8-8.4).099
WHO-2000+TNM stageb  
WHO-2000  
 WDET-B1.0
 WDET-U0.2 (0.4-1.3).100
 WDEC0.2 (0.1-0.5).001
 PDEC0.6 (0.2-2.1).449
 WDEC vs WDET-U0.7 (1.1-3.1).611
 PDEC vs WDET-U2.7 (0.6-13.5).213
 PDEC vs WDEC4.1 (2.0-8.2)< .0001
TNM stage  
 II1.0
 III10.1 (2.6-38.5).001
 IV33.4 (8.8-127.8)< .0001
 IV vs III3.3 (1.7-6.7).001
WHO-2010+TNM stagec  
WHO-2010d  
Grade  
 11.0
 21.9 (0.8-4.9).167
 35.6 (2.6-11.9)< .0001
 3 vs 22.9 (1.1-8.1).039
Stage  
 II1.0
 III4.3 (1.2-16.1).028
 IV13.5 (3.9-46.8)<.0001
 IV vs III3.1 (1.6-6.2).001
TNM stage+gradee  
Stage  
 II1.0
 III4.4 (1.2-16.5).026
 IV14.1 (4.1-48.9)<.0001
 IV vs III3.2 (1.6-6.3).001
Grade  
 11.0
 21.9 (0.7-4.7).183
 35.6 (2.7-11.9)<.0001
 3 vs 23.0 (1.1-8.3).034

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

The objectives of this study were to compare the prognostic efficacy of all GE-NEN classifications, considered alone or combined, and to determine the best prognostic tool currently available. In our series, less differentiated neoplasms preferentially affected man and were localized in the hindgut or the foregut. In contrast, well differentiated neoplasms were likely to originate from midgut in women and were associated with a good prognosis and a low proliferation index. Older age at diagnosis was correlated with less differentiated tumors, which had higher TNM stage and grade. These data are similar to previously presented findings in the literature.16

Because both age and site of origin of the tumor significantly affect survival, we corrected prognostic models for these 2 variables (with the exception of TNM stage models, which were corrected only for age, because TNM stage already considers the area of origin). However, although the area of origin classification in the foregut, midgut, and hindgut is well accepted in the scientific literature,6-9 each area includes different areas of origin with different effects on survival based on the behavior of various tumors. This is particularly evident in the midgut, in which appendiceal carcinoids have a low impact on survival, because they are well differentiated and are usually diagnosed at an earlier stage (the mortality rate for appendiceal tumors was only 10% over a 20-year period, but these tumors are classified 75% as WDETs, 88% as WHO-2010 grade 1, 90% as ENETS-stage I or II, and 88% as Grade 1. The issue of homogeneity of the areas of origin with respect to differentiation and survival is beyond the scope of the current work.

In our single-model analysis, we compared the prognostic efficacy of WHO-2000, WHO-2010, ENETS stage, and grade. The analysis confirmed that the WHO-2000 classification discriminates poorly between mild and aggressive tumors when both tumors present a well differentiated phenotype. This limitation seems to have been overcome in the WHO-2010 classification, which stratifies patients' prognoses effectively into 3 main classes. According to ENETS staging, the survival analysis indicated a higher mortality rate for patients who had stage III and IV (metastasized) disease compared with patients who had stage II (local) disease and a worse prognosis for patients who had stage IV (distantly metastasized) disease compared with patients who had stage III (locally metastasized) disease, thus confirming the prognostic value of the ENETS TNM-based staging system. However, a mortality estimate for patients who had stage 0 and I disease was not possible, because no deaths were observed among those patients during the observation period. This suggests that these 2 classes may be incapable of differentiating patients' prognoses. An evaluation TNM variables individually indicated that tumor size was not correlated with patients ‘survival, whereas both local and distant metastases had a significant effect on long-term survival, as previously reported.30-33 The American Joint Committee on Cancer/International Union Against Cancer TNM staging system34 also was considered for appendiceal tumors, but no substantial difference was demonstrated in patients’ survival stratification compared with ENETS staging.

Considering ENETS grade, patients who had grade 2 and 3 tumors had a significantly worse prognosis than patients who had grade 1 tumors, and survival was significantly worse for patients who had grade 3 tumors compared with patients who had grade 2 tumors. This analysis suggested that ENETS grade also is capable of accurately differentiating patients' prognoses within its classes. In an initial analysis, we also evaluated an alternative Ki-67 cutoff level (ie, 5% instead of 3%), as suggested by Scarpa et al20 for pancreatic NENs, but we did not observe any substantial difference in ENETS grade (ie, only 3 patients changed grade).

The current comparison of prognostic models demonstrated that ENETS stage had a higher c-index, and its prognostic value appeared to be superior to that of the WHO-2000 classification, whereas its prognostic efficacy was comparable to that of the WHO-2010 classification and grade. Although it had a higher c-index, ENETS grade and the WHO-2010 classification had similar prognostic efficacy and were not statistically superior to the WHO-2000 classification. This latter result was hardly surprising, because both of these systems rely on identical Ki-67 cutoff values; however, in our opinion, a formal comparison was necessary to demonstrate the equivalence of these 2 classification systems in the assessment of GE-NETs. These data also suggest that the WHO-2000 classification alone is not sufficient for proper patient stratification and, thus, should always be associated in clinical practice with either the ENETS grade (ie, the WHO-2010 classification) or the TNM stage (ie, the ENETS classification).

Our combined model analyses suggested that the strongest models were TNM stage+grade and TNM stage+WHO-2010, but the efficacy of those models was comparable to that of the TNM stage+WHO-2000 model, and all of these combined models had a c-index that was significantly higher that of the WHO-2000+grade model, suggesting that TNM stage can greatly improve stratification and risk assessment in patients who are diagnosed with GE-NENs.

We tried to minimize the impact of the retrospective nature of this study and the 20-year diagnostic process using accurate data collection based on computerized records that significantly reduced data loss. However, methodological concerns may arise, because we studied 2 survival models in which the WHO-2000 and the WHO-2010 were combined with TNM stage. Two variables—tumor size and the presence of metastases—were computed in the TNM stage and in the WHO-2000 and WHO-2010 classifications. Therefore, in studying models in which TNM stage is combined with WHO-2000 or WHO-2010, these variables actually were “counted twice,” thus inflating the correspondent c-index. Conversely, our analysis was not meant to formally build new scales but, rather, to assess the predictive usefulness of widely used scales if they are applied together by clinicians.

In conclusion, with a minimum observation time of 3 years and with histologic data provided by a single pathologist, this large, single-center study demonstrated that, for a good prognostic evaluation of patients with GE-NENs, a wide range of information about the tumor is needed. Our results demonstrate that the ENETS TNM-based staging system is the best prognostic tool currently available for patients who are diagnosed with GE-NENs. This system improved the prognostic efficacy of other models and provided a risk assessment that also may guide clinicians in the therapeutic management of these patients. However, because GE-NENs are rare tumors, further multicenter studies are needed to validate our results in larger series of patients and in different clinical settings.

FUNDING SOURCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

No specific funding was disclosed.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES
  • 1
    Kloppel G, Couvelard A, Perren A, et al. ENETS Consensus Guidelines for the Standards of Care in Neuroendocrine Tumors: towards a standardized approach to the diagnosis of gastroenteropancreatic neuroendocrine tumors and their prognostic stratification. Neuroendocrinology. 2009; 90: 162-166.
  • 2
    Plockinger U, Rindi G, Arnold R, et al. Guidelines for the diagnosis and treatment of neuroendocrine gastrointestinal tumours. A consensus statement on behalf of the European Neuroendocrine Tumour Society (ENETS). Neuroendocrinology. 2004; 80: 394-424.
  • 3
    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.
  • 4
    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.
  • 5
    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: 757-762.
  • 6
    Kloppel G. Classification and pathology of gastroenteropancreatic neuroendocrine neoplasms. Endocr Relat Cancer. 2011; 18(suppl 1): S1-S16.
  • 7
    Ramage JK, Ahmed A, Ardill J, et al. Guidelines for the management of gastroenteropancreatic neuroendocrine (including carcinoid) tumours (NETs). Gut. 2012; 61: 6-32.
  • 8
    Williams RA, Whitehead R. Non-carcinoid epithelial tumours of the appendix—a proposed classification. Pathology. 1986; 18: 50-53.
  • 9
    Kloppel G. Tumour biology and histopathology of neuroendocrine tumours. Best Pract Res Clin Endocrinol Metab. 2007; 21: 15-31.
  • 10
    Hamilton SR, Aaltonen LA, eds. World Health Organization Classification of Tumours. Pathology and Genetics of Tumours of the Digestive System. Lyon, France: IARC Press; 2000.
  • 11
    Bosman TF, Carniero F, Hruban RH, Theise ND, eds. WHO Classification of Tumours of the Digestive System. Lyon, France: IARC Press; 2010.
  • 12
    Rindi G, Wiedenmann B. Neuroendocrine neoplasms of the gut and pancreas: new insights. Nat Rev Endocrinol. 2011; 8: 54-64.
  • 13
    Pape UF, Jann H, Muller-Nordhorn J, et al. Prognostic relevance of a novel TNM classification system for upper gastroenteropancreatic neuroendocrine tumors. Cancer. 2008; 113: 256-265.
  • 14
    Ekeblad S, Skogseid B, Dunder K, Oberg K, Eriksson B. Prognostic factors and survival in 324 patients with pancreatic endocrine tumor treated at a single institution. Clin Cancer Res. 2008; 14: 7798-7803.
  • 15
    Oberg K, Castellano D. Current knowledge on diagnosis and staging of neuroendocrine tumors. Cancer Metastasis Rev. 2011; 30(suppl 1): 3-7.
  • 16
    Yao JC, Hassan M, Phan A, et al. One hundred years after “carcinoid”: epidemiology of and prognostic factors for neuroendocrine tumors in 35,825 cases in the United States. J Clin Oncol. 2008; 26: 3063-3072.
  • 17
    Fischer L, Kleeff J, Esposito I, et al. Clinical outcome and long-term survival in 118 consecutive patients with neuroendocrine tumours of the pancreas. Br J Surg. 2008; 95: 627-635.
  • 18
    Lepage C, Rachet B, Coleman MP. Survival from malignant digestive endocrine tumors in England and Wales: a population-based study. Gastroenterology. 2007; 132: 899-904.
  • 19
    Kloppel G, Rindi G, Anlauf M, Perren A, Komminoth P. Site-specific biology and pathology of gastroenteropancreatic neuroendocrine tumors. Virchows Arch. 2007; 451(suppl 1): S9-S27.
  • 20
    Scarpa A, Mantovani W, Capelli P, et al. Pancreatic endocrine tumors: improved TNM staging and histopathological grading permit a clinically efficient prognostic stratification of patients. Mod Pathol. 2010; 23: 824-833.
  • 21
    Rindi G, de Herder WW, O'Toole D, Wiedenmann B. Consensus guidelines for the management of patients with digestive neuroendocrine tumors: why such guidelines and how we went about It. Neuroendocrinology. 2006; 84: 155-157.
  • 22
    Arnold R, Chen YJ, Costa F, et al. ENETS Consensus Guidelines for the Standards of Care in Neuroendocrine Tumors: follow-up and documentation. Neuroendocrinology. 2009; 90: 227-233.
  • 23
    Falconi M, Bettini R, Scarpa A, Capelli P, Pederzoli P. Surgical strategy in the treatment of gastrointestinal neuroendocrine tumours. Ann Oncol. 2001; 12(suppl 2): S101-S103.
  • 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: 1083-1092.
  • 25
    Jann H, Roll S, Couvelard A, et al. Neuroendocrine tumors of midgut and hindgut origin: tumor-node-metastasis classification determines clinical outcome. Cancer. 2011; 117: 3332-3341.
  • 26
    La Rosa S, Inzani F, Vanoli A, et al. Histologic characterization and improved prognostic evaluation of 209 gastric neuroendocrine neoplasms. Hum Pathol. 2011; 42: 1373-1384.
  • 27
    Anlauf M, Garbrecht N, Bauersfeld J, et al. Hereditary neuroendocrine tumors of the gastroenteropancreatic system. Virchows Arch. 2007; 451: 29-38.
  • 28
    Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996; 15: 361-387.
  • 29
    Newson R. Comparing the predictive power of survival models using Harrell's c or Somers' D. Stata J. 2010; 10: 339-358.
  • 30
    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: 532-542.
  • 31
    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: 1126-1132.
  • 32
    Mullen JT, Wang H, Yao JC, et al. Carcinoid tumors of the duodenum. Surgery. 2005; 138: 971-977; discussion 977-978.
  • 33
    Weber HC, Venzon DJ, Lin JT, et al. Determinants of metastatic rate and survival in patients with Zollinger-Ellison syndrome: a prospective long-term study. Gastroenterology. 1995; 108: 1637-1649.
  • 34
    Kloppel G, Rindi G, Perren A, Komminoth P, Klimstra DS. The ENETS and AJCC/UICC TNM classifications of the neuroendocrine tumors of the gastrointestinal tract and the pancreas: a statement. Virchows Arch. 2010; 456: 595-597.