Diagnosis of Gastroenteropancreatic Neuroendocrine Tumors
Results are reported following the REMARK guidelines (REporting recommendations for tumor MARKer prognostic studies).
An initial survival analysis of the entire series revealed a striking prognostic difference when the primary tumor anatomic site was considered. Neuroendocrine tumors arising in the appendix (n = 19) showed a cumulative 5-year survival rate of 94.4% Figure 1. They belonged to favorable categories: G1 grade; good differentiation; early stage; no vascular invasion, lymph node involvement, or necrosis; small size (<1 cm); and low proliferation rate (<2 mitosis 10× high-power field and <2% Ki-67 expression). These observations suggest that the biology underlying the development of NETs of the appendix may differ from NETs of other gastrointestinal anatomic origin. For this reason, we decided to exclude these cases from the series and study them separately.
(Enlarge Image)
Figure 1.
Primary tumor location and survival rates. Kaplan-Meier survival analysis of the 110 patient series is stratified according to the primary tumor anatomic site of origin.
No significant survival differences were observed among the other anatomic sites (Figure 1). A subset analysis based on site was performed, but strong statistical conclusions could not be drawn due to sample size. Thus, the results reported here refer to GEP NETs as a whole, with the appendix excluded.
In the healthy intestinal neuroendocrine cells, E-cadherin and β-catenin immunostaining was detected in the membrane with a linear pattern Image 1A and Image 1B. Both proteins form the E-cadherin/β-catenin complex, which is essential in maintaining tissue homeostasis. N-cadherin and vimentin expression was negative in healthy intestinal neuroendocrine cells. None of the transcriptional repressors showed a nuclear signal in healthy intestinal neuroendocrine cells Image 1C and Image 1D. In tumors, E-cadherin, β-catenin, and N-cadherin were highly expressed in 43 (47.3%), 46 (50.5%), and 61 (67.0%) of 91 cases, respectively. Vimentin expression was detected in 10 (11.0%) of 91 cases, 7 of which were G1 NETs. According to the E-cadherin/β-catenin complex integrity status, 37 (41.6%) of 89 tumors were classified as having lost complex integrity. In tumors, Snail1, Snail2, Twist, and Foxc2 were positive in 45 (50.0%) of 90 cases and in 40 (44.0%), 43 (47.3%), and 52 (57.1%) of 91 cases, respectively. Representative immunostainings of E-cadherin, β-catenin, Snail, and Foxc2 are shown in Image 2 (G1 NETs), Image 3 (G2 NETs), Image 4 (G3 NEC-LC), and Image 5 (G3 NEC-SC).
(Enlarge Image)
Image 1.
Representative double immunostaining of healthy intestinal mucosa: synaptophysin expression identifies the neuroendocrine cell (red) with normal E-cadherin (brown) (A) and β-catenin (brown) (B) expression (arrows). Nuclear expression of Snail1 (C) and Foxc2 (D) is negative (arrows) in neuroendocrine cells (×400).
(Enlarge Image)
Image 2.
Representative immunostaining of grade 1 neuroendocrine tumors: E-cadherin (A), β-catenin (B), Snail1 (C), and Foxc2 (D) (×200).
(Enlarge Image)
Image 3.
Representative immunostaining of grade 2 neuroendocrine tumors: E-cadherin (A), β-catenin (B), Snail1 (C), and Foxc2 (D) (×200).
(Enlarge Image)
Image 4.
Representative immunostaining of grade 3 large-cell neuroendocrine carcinomas: E-cadherin (A), β-catenin (B), Snail1 (C), and Foxc2 (D) (×200).
(Enlarge Image)
Image 5.
Representative immunostaining of grade 3 small-cell neuroendocrine carcinomas: E-cadherin (A), β-catenin (B), Snail1 (C), and Foxc2 (D) (×200).
Most tumors arising in the appendix showed no Snail1, Snail2, Twist, Foxc2, or vimentin expression and high E-cadherin expression (P < .05 for all).
Grade 3 NEC-LC and NEC-SC showed the following changes in EMT marker expression more often than G1 and G2 NETs: E-cadherin cytoplasmic pattern (P = .001), weak β-catenin expression (P = .04), high Foxc2 expression (P = .02), and a trend toward high Snail1 expression (P = .08). Tumor grade and the presence of necrosis were associated with a cytoplasmic pattern of E-cadherin (P = .001 and P = .047, respectively), weak N-cadherin expression (P = .003 and P = .004, respectively), high Snail1 expression (P = .04 and P = .04, respectively), and high Foxc2 expression (P = .005 and P = .004, respectively).
High N-cadherin expression was associated with the favorable findings of absence of necrosis (P = .004) and G1 status (P = .0001). Moreover, a cytoplasmic N-cadherin pattern was associated with an absence of nodal involvement (P = .04) and early stage tumors (I–II; P = .049). Tumors of patients with excessive alcohol consumption showed an altered E-cadherin/β-catenin complex (P = .01).
One of the aims of our study was to find molecular variables that might help in the differential diagnosis of NET subgroups. Thus, we established comparisons of each protein expression variable between the following subgroups: G1 NET vs G2 NET and G3 NEC-LC vs G3 NEC-SC. We found that weak β-catenin and N-cadherin expression discriminated between G3 NECs, being more common in the small cell subtype than in the large cell subtype (P = .009 and P = .002, respectively).
E-cadherin protein expression levels were independent of the Snail1 levels. However, E-cadherin cytoplasmic localization was correlated with high Snail1 expression (P = .03) and, as expected, with the loss of the integrity of the E-cadherin/β-catenin complex (P = .001). Also, elevated N-cadherin expression was associated with the conserved E-cadherin/β-catenin complex (P = .04). An association between high Twist expression and high N-cadherin level at the membrane (P = .007) was observed. The elevated Foxc2 expression levels correlated with high Snail1 (P = .0001), Snail2 (P = .002), and Twist expression (P = .001). No evidence of the described "cadherin switch" was found.
The mean follow-up was 58 months (range, 1–130 months), and the 5-year cumulative survival rate was 56.8%. A negative impact on patient survival was found for the following clinicopathologic variables: age, diagnosis, tumor grade, mitotic index, proliferation index (Ki-67), histologic differentiation, and necrosis. The molecular variables with a negative impact on survival were loss of integrity complex Figure 2A, reduced N-cadherin expression Figure 2B, high Snail1 level Figure 2C, and the cytoplasmic E-cadherin pattern Figure 2D. Mean survival times for each category are shown in Figure 3. In the multivariate survival analysis (Cox) in which significant variables were included according to Kaplan-Meier analysis, the variables with an independent prognostic value were the mitotic index (2.6-fold increased risk of death by disease, P = .0001) and Snail1 expression (2.17-fold increased risk of death by disease, P = .05).
(Enlarge Image)
Figure 2.
Molecular findings affecting survival rates. Cumulative Kaplan-Meier survival curves are stratified according to the integrity of the E-cadherin/β-catenin complex (P = .04) (A), N-cadherin expression level (P = .055) (B), Snail1 protein expression levels (P = .02) (C), and E-cadherin pattern (P = .005) (D).
(Enlarge Image)
Figure 3.
Mean survival times for each category of the molecular variables with an impact on survival.
Grade 1 NETs show the peculiarity that their histologic differentiation does not correlate with tumor invasion and metastasis. This type of tumor was the most prevalent in our series (n = 49), which allowed for a statistical analysis. In a search of variables with prognostic significance, we performed a survival analysis and found that age younger than 60 years and high β-catenin expression identified subgroups of patients with better survival rates (P = .04 and P = .04, respectively).
Results
Results are reported following the REMARK guidelines (REporting recommendations for tumor MARKer prognostic studies).
Previous Considerations
An initial survival analysis of the entire series revealed a striking prognostic difference when the primary tumor anatomic site was considered. Neuroendocrine tumors arising in the appendix (n = 19) showed a cumulative 5-year survival rate of 94.4% Figure 1. They belonged to favorable categories: G1 grade; good differentiation; early stage; no vascular invasion, lymph node involvement, or necrosis; small size (<1 cm); and low proliferation rate (<2 mitosis 10× high-power field and <2% Ki-67 expression). These observations suggest that the biology underlying the development of NETs of the appendix may differ from NETs of other gastrointestinal anatomic origin. For this reason, we decided to exclude these cases from the series and study them separately.
(Enlarge Image)
Figure 1.
Primary tumor location and survival rates. Kaplan-Meier survival analysis of the 110 patient series is stratified according to the primary tumor anatomic site of origin.
No significant survival differences were observed among the other anatomic sites (Figure 1). A subset analysis based on site was performed, but strong statistical conclusions could not be drawn due to sample size. Thus, the results reported here refer to GEP NETs as a whole, with the appendix excluded.
Protein Expression in Healthy vs Tumor Tissue
In the healthy intestinal neuroendocrine cells, E-cadherin and β-catenin immunostaining was detected in the membrane with a linear pattern Image 1A and Image 1B. Both proteins form the E-cadherin/β-catenin complex, which is essential in maintaining tissue homeostasis. N-cadherin and vimentin expression was negative in healthy intestinal neuroendocrine cells. None of the transcriptional repressors showed a nuclear signal in healthy intestinal neuroendocrine cells Image 1C and Image 1D. In tumors, E-cadherin, β-catenin, and N-cadherin were highly expressed in 43 (47.3%), 46 (50.5%), and 61 (67.0%) of 91 cases, respectively. Vimentin expression was detected in 10 (11.0%) of 91 cases, 7 of which were G1 NETs. According to the E-cadherin/β-catenin complex integrity status, 37 (41.6%) of 89 tumors were classified as having lost complex integrity. In tumors, Snail1, Snail2, Twist, and Foxc2 were positive in 45 (50.0%) of 90 cases and in 40 (44.0%), 43 (47.3%), and 52 (57.1%) of 91 cases, respectively. Representative immunostainings of E-cadherin, β-catenin, Snail, and Foxc2 are shown in Image 2 (G1 NETs), Image 3 (G2 NETs), Image 4 (G3 NEC-LC), and Image 5 (G3 NEC-SC).
(Enlarge Image)
Image 1.
Representative double immunostaining of healthy intestinal mucosa: synaptophysin expression identifies the neuroendocrine cell (red) with normal E-cadherin (brown) (A) and β-catenin (brown) (B) expression (arrows). Nuclear expression of Snail1 (C) and Foxc2 (D) is negative (arrows) in neuroendocrine cells (×400).
(Enlarge Image)
Image 2.
Representative immunostaining of grade 1 neuroendocrine tumors: E-cadherin (A), β-catenin (B), Snail1 (C), and Foxc2 (D) (×200).
(Enlarge Image)
Image 3.
Representative immunostaining of grade 2 neuroendocrine tumors: E-cadherin (A), β-catenin (B), Snail1 (C), and Foxc2 (D) (×200).
(Enlarge Image)
Image 4.
Representative immunostaining of grade 3 large-cell neuroendocrine carcinomas: E-cadherin (A), β-catenin (B), Snail1 (C), and Foxc2 (D) (×200).
(Enlarge Image)
Image 5.
Representative immunostaining of grade 3 small-cell neuroendocrine carcinomas: E-cadherin (A), β-catenin (B), Snail1 (C), and Foxc2 (D) (×200).
Most tumors arising in the appendix showed no Snail1, Snail2, Twist, Foxc2, or vimentin expression and high E-cadherin expression (P < .05 for all).
Associations Between Clinicopathologic and Protein Expression Variables in GEP NETs
Grade 3 NEC-LC and NEC-SC showed the following changes in EMT marker expression more often than G1 and G2 NETs: E-cadherin cytoplasmic pattern (P = .001), weak β-catenin expression (P = .04), high Foxc2 expression (P = .02), and a trend toward high Snail1 expression (P = .08). Tumor grade and the presence of necrosis were associated with a cytoplasmic pattern of E-cadherin (P = .001 and P = .047, respectively), weak N-cadherin expression (P = .003 and P = .004, respectively), high Snail1 expression (P = .04 and P = .04, respectively), and high Foxc2 expression (P = .005 and P = .004, respectively).
High N-cadherin expression was associated with the favorable findings of absence of necrosis (P = .004) and G1 status (P = .0001). Moreover, a cytoplasmic N-cadherin pattern was associated with an absence of nodal involvement (P = .04) and early stage tumors (I–II; P = .049). Tumors of patients with excessive alcohol consumption showed an altered E-cadherin/β-catenin complex (P = .01).
Differential Diagnosis of GEP NETs
One of the aims of our study was to find molecular variables that might help in the differential diagnosis of NET subgroups. Thus, we established comparisons of each protein expression variable between the following subgroups: G1 NET vs G2 NET and G3 NEC-LC vs G3 NEC-SC. We found that weak β-catenin and N-cadherin expression discriminated between G3 NECs, being more common in the small cell subtype than in the large cell subtype (P = .009 and P = .002, respectively).
Associations between Protein Expression Findings
E-cadherin protein expression levels were independent of the Snail1 levels. However, E-cadherin cytoplasmic localization was correlated with high Snail1 expression (P = .03) and, as expected, with the loss of the integrity of the E-cadherin/β-catenin complex (P = .001). Also, elevated N-cadherin expression was associated with the conserved E-cadherin/β-catenin complex (P = .04). An association between high Twist expression and high N-cadherin level at the membrane (P = .007) was observed. The elevated Foxc2 expression levels correlated with high Snail1 (P = .0001), Snail2 (P = .002), and Twist expression (P = .001). No evidence of the described "cadherin switch" was found.
Survival Analysis
The mean follow-up was 58 months (range, 1–130 months), and the 5-year cumulative survival rate was 56.8%. A negative impact on patient survival was found for the following clinicopathologic variables: age, diagnosis, tumor grade, mitotic index, proliferation index (Ki-67), histologic differentiation, and necrosis. The molecular variables with a negative impact on survival were loss of integrity complex Figure 2A, reduced N-cadherin expression Figure 2B, high Snail1 level Figure 2C, and the cytoplasmic E-cadherin pattern Figure 2D. Mean survival times for each category are shown in Figure 3. In the multivariate survival analysis (Cox) in which significant variables were included according to Kaplan-Meier analysis, the variables with an independent prognostic value were the mitotic index (2.6-fold increased risk of death by disease, P = .0001) and Snail1 expression (2.17-fold increased risk of death by disease, P = .05).
(Enlarge Image)
Figure 2.
Molecular findings affecting survival rates. Cumulative Kaplan-Meier survival curves are stratified according to the integrity of the E-cadherin/β-catenin complex (P = .04) (A), N-cadherin expression level (P = .055) (B), Snail1 protein expression levels (P = .02) (C), and E-cadherin pattern (P = .005) (D).
(Enlarge Image)
Figure 3.
Mean survival times for each category of the molecular variables with an impact on survival.
Differences in Survival Between G1 NETs
Grade 1 NETs show the peculiarity that their histologic differentiation does not correlate with tumor invasion and metastasis. This type of tumor was the most prevalent in our series (n = 49), which allowed for a statistical analysis. In a search of variables with prognostic significance, we performed a survival analysis and found that age younger than 60 years and high β-catenin expression identified subgroups of patients with better survival rates (P = .04 and P = .04, respectively).
SHARE
