The impact of somatostatin receptor type 2 (SSTR-2a) expression levels on outcomes in patients with pancreatic neuroendocrine tumors (PNETs) has not been evaluated.
The impact of somatostatin receptor type 2 (SSTR-2a) expression levels on outcomes in patients with pancreatic neuroendocrine tumors (PNETs) has not been evaluated.
Correlations between clinicopathologic characteristics, including SSTR-2a expression and outcomes, were retrospectively studied in 79 patients with pancreatic neuroendocrine tumors (PNETs).
The SSTR-2a score was 0 in 27% of patients, 1 in 24% of patients, 3 in 30% of patients, and 4 in 18% of patients. The overall survival rate was 87% at 1 year, 77% at 3 years, and 71% at 5 years. On univariate analysis, a pancreatic tumor that measured ≥20 mm in greatest dimension, stage IV disease, vascular invasion, neuroendocrine carcinoma (NEC), and an SSTR-2a score of 0 were associated significantly with poor outcomes. On multivariate analysis, NEC (P = .000; hazard ratio, 28.8; 95% confidence interval, 7.502-111.240) and an SSTR-2a score of 0 (P = .001; hazard ratio, 3.611; 95% confidence interval, 1.344-9.702) were related independently to poor outcomes.
The current analysis of prognostic factors in patients with PNETs demonstrated that NEC and an SSTR-2a score of 0 both were significant independent predictors of poor outcomes. The results suggest that the assessment of SSTR-2a may facilitate the selection of treatment regimens and the prediction of outcomes. Because a considerable proportion of patients with NEC have SSTR-2a-positive tumors, further analyses of the usefulness of somatostatin analogues are warranted in patients who have SSTR-2a-positive NEC. Cancer 2013. © 2013 American Cancer Society.
Pancreatic neuroendocrine tumors (PNETs) have recently received considerable attention. However, PNETs remain poorly understood.
In recent years, it has been demonstrated that somatostatin analogues are useful for the management of unresectable, well differentiated PNETs.[1-3] Octreotide has high affinity for somatostatin receptor type 2A (SSTR-2a). Therefore, the evaluation of SSTR-2a status in histopathologic tumor specimens may facilitate predicting the response of PNETs to somatostatin analogues. To our knowledge, the impact of the SSTR-2a expression level on outcomes has not been evaluated previously in patients with PNETs. To gain insight into these issues, we retrospectively studied the relations of outcomes to clinical characteristics and histopathologic characteristics, including SSTR-2a scores, in 79 patients who had PNETs diagnosed histopathologically from 1988 through 2012.
The study group comprised 79 patients with PNETs who were treated in Kitasato University East Hospital from 1988 through 2012. A diagnosis of PNET was established histopathologically on the basis of immunostaining of biopsy specimens obtained by endoscopic ultrasound-guided fine-needle aspiration or surgically resected specimens. We retrospectively reviewed the medical records of these patients. Written informed consent for the use of tissue specimens was obtained from all patients.
The clinical variables studied were age at disease onset, sex, main symptoms, the presence or absence of functional tumors, tumor location, greatest pancreatic tumor dimension, the presence or absence of calcification, the presence or absence of cystic changes in tumors, the presence or absence of metastasis, and tumor stage according to the tumor-lymph node-metastasis (TNM) classification, as recommended by the American Joint Committee on Cancer (AJCC)/International Union Against Cancer (UICC).
Histopathologic examinations were performed to evaluate tumors based on the 2010 World Heath Organization (WHO) classification, venous and lymphatic invasion, invasion of the main pancreatic duct, the Ki-67 labeling index, and the SSTR-2a score. Formalin-fixed, paraffin-embedded blocks of tumors were used for immunohistochemical studies.
At the time of diagnosis of PNETs, immunostaining to establish a definitive pathologic diagnosis was assigned the highest priority. Consequently, the 2010 WHO classification and the SSTR-2 score could not be evaluated in some patients who had only small tissue specimens available, because they did not undergo surgery. The 2010 WHO classification was evaluated in 76 of the 79 patients, and SSTR-2 scores were evaluated in 66 of the 79 patients. Vascular invasion was evaluated histopathologically in patients who underwent surgery and in 1 patient at autopsy. Involvement of the main pancreatic duct was histologically evaluated in surgically resected specimens. In patients in who had tissue samples obtained only by biopsy or endoscopic ultrasound-guided fine-needle aspiration, disruption of the main pancreatic duct on endoscopic retrograde pancreatography was considered to indicate tumor invasion.
For immunohistochemical studies, working dilutions of the following primary antibodies were used: monoclonal anti-Ki-67 antibody (clone MIB-1; 1:100 dilution; Dako, Glostrup, Denmark) and polyclonal anti-SSTR-2a antibody (1:1000 dilution; Gramsch Laboratories, Schwabhausen, Germany). For both types of staining, PT Link (Dako) was used for antigen retrieval. After incubation with primary antibodies, the tissue slides were processed with an autostainer (Ventana HX System; Roche, Basel, Switzerland) according to the manufacturer's manual. Counterstaining was performed with Mayer hematoxylin.
After staining for Ki-67, fields in which positively stained nuclei were most densely distributed (so-called “hot spots”) were identified at lower magnification. Then, with the objective lens set at ×40 magnification, >1000 cells were examined in at least 5 high-power fields of the hot spots, and cells that were unequivocally positive for nuclear staining were counted. The percentage of nuclei with positive staining was defined as the Ki-67 labeling index. For the SSTR-2a score, expression profiles were classified with a 4-tiered system (with scores from 0 to 3), according to the methods proposed by Volante et al (Fig. 1).
The Ki-67 labeling index and the SSTR-2a score were evaluated first independently by an expert histopathologist (T.M.) and by 1 well trained gastroenterologist (K.O.). Any discrepancies between the 2 observers were resolved by consensus, which was reached while they reviewed the specimens in question under a discussion microscope.
Treatment regimens (rates of surgery and regimens used to treat unresectable tumors) and treatment outcomes (postoperative recurrence rate, time to recurrence, and site of recurrence) were studied.
The follow-up period was defined as the interval from the date of diagnosis of PNET to the date of death, with data censored at the last known date that the patient was alive, or to December 31, 2012. For statistical analysis, the likelihood test of independence between 2 groups was performed using the Fisher exact test. Survival rates were calculated with the Kaplan-Meier method. Risk factors potentially related to outcomes were analyzed using log-rank tests. A Cox proportional-hazards regression model was used to analyze independent risk factors. Statistical analyses were performed with the statistical package SPSS Base 17.0 (SPSS Inc., Chicago, Ill). P values < .05 were considered statistically significant.
The clinical characteristics of the patients are provided in Table 1. The median age of the 79 patients (37 men and 42 women) was 58 years (age range, 16-86 years). The most common conditions leading to diagnosis were abdominal pain (29%), hypoglycemia (13%), and back pain (6%). In total, 33 patients were asymptomatic and had tumors either detected on ultrasonographic examinations during heath check-ups (17 patients) or identified incidentally on imaging studies during follow-up for other diseases (16 patients). Fifteen patients (19%) had functional tumors, including 11 insulinomas, 3 gastrinomas, and 1 glucagonoma. The other 64 patients (81%) had nonfunctional tumors. Three patients had multiple endocrine neoplasia type 1 (MEN-1). Four of the 79 patients had 2 ectopic lesions in the pancreas. The ectopic lesions included 2 insulinomas, 1 glucagonoma, and 1 nonfunctional tumor. The sites of the 83 lesions in the 79 patients were as follows: head of the pancreas, 27 lesions; body of the pancreas, 30 lesions; and tail of the pancreas, 26 lesions. The median greatest tumor dimension of the 83 pancreatic lesions was 24 mm (range, 2-220 mm). Of the 83 lesions, calcification was identified in 10 lesions (3 functional tumors and 7 nonfunctional tumors), and cystic changes were identified inside the tumor or at its margin in 16 lesions (1 functional tumor and 15 nonfunctional tumors). At the time of diagnosis, 20 patients (25%) had distant metastases. The site of metastasis was the liver in 19 patients and the lymph nodes in 1 patient. Disease stage according to the TNM classification system as proposed by the AJCC/UICC was stage I in 28 patients, stage II in 19 patients (stage IIA in 12 and stage IIB in 7), stage III in 15 patients (stage IIIA in 1 and stage IIIB in 14), and stage IV in 17 patients.
|Characteristic||No. of Patients (%)|
|Age: Median [range], y||58 [16–86]|
|Upper abdominal pain or/and back pain||28 (35)|
|Health examination||17 (22)|
|During follow-up for other diseases||16 (20)|
|Presence of MEN-1||3 (4)|
|Location of tumora|
|Head of pancreas||27 (33)|
|Body of pancreas||30 (36)|
|Tail of pancreas||26 (31)|
|Greatest tumor dimension: Median [range], mma||24 [2–220]|
|Cyst in the tumora||16 (19)|
|TNM stage: AJCC/UICC|
The 2010 WHO classification of the 76 tumors that could be evaluated histopathologically was grade 1 neuroendocrine tumor (NETG1) in 38 lesions (50%), grade 2 neuroendocrine tumor (NETG2) in 25 lesions (33%), neuroendocrine carcinoma (NEC) in 12 lesions (16%), and mixed adenoneuroendocrine carcinoma (MANEC) in 1 lesion (1%). According to the 2010 WHO classification, the TNM classification for NETG1 and NETG2 was stage I in 43% of tumors, stage II in 27%, stage III in 17%, and stage IV in 13%. The TNM classification for NEC was stage I in 0%, stage II in 8%, stage III in 25%, and stage IV in 67%. The TNM classification for the 1 patient with MANEC was stage II. Among the 59 primary tumors that could be evaluated in surgically resected specimens or pathologic autopsy specimens, 27 (46%) were associated with vascular invasion. The rate of vascular invasion according to the 2010 WHO classification was 17% for NETG1, 70% for NETG2, 100% for NEC 100%, and 100% for MANEC. One patient in whom vascular invasion was evaluated at autopsy died shortly after diagnosis, but there was no evidence of vascular invasion on histopathologic examination of autopsy specimens, suggesting that the findings were similar at the time of diagnosis. Among the 67 lesions that could be evaluated in surgical specimens and on endoscopic retrograde pancreatography, 14 (21%) were accompanied by invasion of the main pancreatic duct. The rate of invasion of the main pancreatic duct according to the 2010 WHO classification was 6% for NETG1, 25% for NETG2, 40% for NEC, and 100% for MANEC. The SSTR-2a score in the 66 tumors that could be evaluated histopathologically was 0 in 18 tumors (27%), 1 in 16 tumors (24%), 2 in 20 tumors (30%), and 3 in 12 tumors (18%). According to the SSTR-2a score, the TNM classification for an SSTR-2a score of 0 was stage I in 35% of tumors, stage II in 18%, stage III in 12%, and stage IV in 35%. The TNM classification for an SSTR-2a of score 1 to 3 was stage I in 39% of tumors, stage II in 27%, stage III in 18%, and stage IV in 16%. According to the 2010 WHO classification, the SSTR-2a score for NETG1 was 0 in 22% of tumors, 1 in 19%, 2 in 34%, and 3 in 25%. The SSTR-2a score for NETG2 was 0 in 18% of tumors, 1 in 27%, 2 in 36%, and 3 in 18%. The SSTR-2a score for NEC was 0 in 58% of tumors, 1 in 33%, 2 in 8%, and 3 in 0%. The SSTR-2a score for MANEC was 0 in 0% of tumors, 1 in 100%, 2 in 0%, and 3 in 0%.
In total, 59 patients (75%) underwent radical surgery as first-line treatment. The postoperative recurrence rate was 15% (9 patients), and the median follow-up until postoperative recurrence was 18.7 months (range, 3-128 months). The site of recurrence was the remnant pancreas in 1 patient, the liver in 7 patients, and the liver and lymph nodes in 1 patient. The preoperative disease stage according to the TNM classification in the 9 patients who had a postoperative recurrence was stage IIB in 2 patients and stage IIIB in 7 patients. The 2010 WHO classification was NETG2 in 7 patients and NEC in 2 patients. The patients who did no undergo surgery received 1 or more of the following treatments (excluding those who refused therapy): somatostatin analogues, chemotherapy, and local treatment for liver metastasis (radiofrequency ablation or transcatheter arterial chemoembolization). Three patients with NETG2 received a somatostatin analog (octreotide). One of those patients had unresectable disease with multiple liver metastases at initial diagnosis, and the other 2 had recurrences with multiple liver metastases after surgery.
The median follow-up in the study group as a whole was 37.5 months (range, 0.7-244.4 months). Twenty patients died during follow-up. The overall survival rate was 87% at 1 year, 77% at 3 years, and 71% at 5 years (Fig. 2). On univariate analysis, a pancreatic tumor measuring ≥20 mm in greatest dimension (P = .0007), stage IV disease (P = .0000), vascular invasion (P = .0231), NEC (P = .0000), and an SSTR-2a score of 0 (P = .0000) were associated significantly with poor outcomes (Table 2). On multivariate analysis, NEC (hazard ratio, 28.889; 95% confidence interval, 7.502-111.240) and an SSTR-2a score of 0 (hazard ratio, 3.611; 95% CI, 1.344-9.702) were each independently related to poor outcomes (Table 2).
|Univariate Analysis||Multivariate Analysis|
|Variable||No. of Patients||Pa||HR (95% CI)||Pb|
|Age at diagnosis, y||.3669|
|Greatest tumor dimension, mm||.0007|
|Cyst in the tumord||.0774|
|2010 WHO classificationg||.0000||28.889 (7.502–111.240)||.000|
|Grade 1 NET||64|
|Grade 2 NET/NEC||12|
|SSTR-2a scoreh||.0000||3.611 (1.344–9.702)||.001|
The survival rate according to the 2010 WHO classification was 97% at 1 year, 97% at 3 years, and 91% at 5 years in patients with NETG1; 96% at 1 year, 92% at 3 years, and 87% at 5 years in patients with NETG2; and 42% at 1 year, 8% at 3 years, and 0% at 5 years in patients with NEC. Survival was significantly poorer in patients with NEC than in those with NETG1 or NETG2 (P = .0000) (Fig. 3). The survival rate according to the SSTR-2a score was 58% at 1 year, 51% at 3 years, and 35% at 5 years in patients with an SSTR-2a score of 0; 88% at 1 year, 74% at 3 years, and 74% at 5 years in patients with an SSTR-2a score of 1; 94% at 1 year, 80% at 3 years, and 80% at 5 years in patients with an SSTR-2a score of 2; and 100% at 1 year, 100% at 3 years, and 100% at 5 years in patients with a SSTR-2a score of 3. Survival was significantly poorer in patients who had an SSTR-2a score of 0 than in those who had an SSTR-2a score of 1 to 3 (P = .0000) (Fig. 4). The proportion of patients with NEC was significantly higher among patients with an SSTR-2a score of 0 than among those with an SSTR-2a score of 1 to 3 (P = .0023) (Table 3). The survival rate among patients with NETG1 or NETG2 was 89% at 1 year, 89% at 3 years, and 71% at 5 years in patients with an SSTR-2a score of 0 (n = 10) and 98% at 1 year, 95% at 3 years, and 91% at 5 years in patients with an SSTR-2a score of 1 to 3 (n = 44). The survival rate among patients with NEC was 60% at 1 year and 20% at 3 years in patients with an SSTR-2a score of 0 (n = 5) and 14% at 1 year and 0% at 3 years in patients with an SSTR-2a score of 1 to 3 (n = 7). An SSTR-2a score of 0 was associated with significantly poorer outcomes than an SSTR-2a score of 1 to 3 among patients with NETG1 or NETG2 (P = .0309), but not among patients with NEC.
|Pattern of SSTR-2a Score: No. of Patients|
|Variable||SSTR-2a Score 0, n = 17||SSTR-2a Score 1–3, n = 49||Pa|
|Age: Median [range], y||60 [19–81]||58 [16–86]||.6973|
|Location of tumora||.5917|
|Greatest tumor dimension: Median [range], mm||40 [4–220]||20 [2–110]||.0958|
|Cyst in tumorc||.4009|
|2010 WHO classificationf||.0023|
|Grade 1 NET||10||44|
|Grade 2 NET/NEC||7||3|
In the current study, we retrospectively studied the relations of outcomes to clinical characteristics as well as histopathologic characteristics based on the 2010 WHO classification, including SSTR-2a scores, in 79 patients who had PNETs diagnosed histopathologically from 1988 through 2012. Univariate analysis demonstrated that a pancreatic tumor that measured ≥20 mm in greatest dimension, stage IV disease according to the TNM staging system as proposed by the AJCC/UICC, vascular invasion, NEC according to the 2010 WHO classification, and an SSTR-2a score of 0 were associated significantly with poor outcomes. On multivariate analysis, NEC and an SSTR-2a score of 0 were each independently related to poor outcomes.
Corleto et al reported the impact of the presence or absence of SSTR-2a gene expression on the outcomes of PNETs in 33 patients with well differentiated gastroenteropancreatic endocrine carcinomas. To our knowledge, however, no previous study has reported the clinicopathologic characteristics and outcomes from a large series of patients with confirmed diagnoses of primary PNETs. Somatostatin is a cyclic peptide that investigators incidentally discovered could potently inhibit growth hormone secretion from the pituitary gland. Somatostatin receptors are classified into 5 subtypes, and SSTR-2 is subclassified into 2 variants according to the presence or absence of splicing: SSTR-2a and SSTR-2b.[10, 11] SSTR-2a has been implicated in the inhibition of endocrine hormone secretion, the suppression of growth factor-induced cell cycle progression, and the induction of apoptosis.[11-13] Octreotide, a drug that binds with high affinity to SSTR-2a, was developed to take advantage of these effects. Recent studies have demonstrated the therapeutic usefulness of octreotide in patients with unresectable, well differentiated PNETs.[2, 3, 14]
In our study, SSTR-2a was positive in significantly higher proportions of NETG1 and NETG2 than in NEC. With regard to correlation between SSTR-2a expression and Ki-67 labeling index, SSTR-2a expression reportedly was significantly lower in tumors with a Ki-67 labeling index ≥2% or higher than in those with a Ki-67 labeling index <2%. Conversely, some investigators have observed no difference in the expression levels of somatostatin receptor subtypes between NET and NEC.[1, 15] Thus, these findings remain controversial. In the current study, an SSTR-2a score of 0 was a significant independent predictor of a poor outcome. This observation may be attributed to the finding that some patients with NEC, which carries an extremely poor prognosis, were positive for SSTR-2a. Moreover, SSTR-2a was negative in 10 of the 54 patients with NETG1 or NETG2, including 3 patients who died. Consequently, our study demonstrated that NEC and an SSTR-2a score of 0 were each independent prognostic factors. Corleto et al reported that tumors with a Ki-67 labeling index <2% that were positive for both SSTR-2 and SSTR-5 were associated with significantly better outcomes than tumors with a Ki-67 labeling index ≥2% that were negative for both SSTR-2 and SSTR-5. When groups of tumors were compared according to the presence or absence of SSTR expression, tumors that were positive for SSTR-2 and SSTR-5 were associated with slightly (but not significantly) better outcomes than those that were negative for SSTR-2 and SSTR-5. This may be attributed to the observation that approximately 50% of the patients in the study by Corleto et al had primary NETs in organs other than the pancreas and that well differentiated endocrine carcinomas were studied. Although an SSTR-2a score of 0 was an independent predictor of poor outcomes in our study, the hazard ratio was about 8 times higher in patients with NEC. To confirm our finding that negativity for SSTR-2 is strongly related to poor outcomes in patients with PNETs, further studies in larger numbers of patients are required.
Our results reconfirmed the usefulness of the TNM staging system and the 2010 WHO classification. Stage IV disease according to the TNM classification was not an independent poor prognostic factor on multivariate regression analysis, probably because stage IV disease was significantly more common among patients with NEC than among those with NETG1 or NETG2 (P < .01; Fisher exact test) and because considerable numbers of long-term survivors were included among the patients with stage IV disease who had NETG1 or NETG2. With regard to the 2010 WHO classification, however, NEC was associated significantly with poorer outcomes than other carcinomas, but outcomes did not differ significantly between NETG1 and NETG2. Recently, Hamilton et al reported that classifying NETG1, NETG2, and NEC on the basis of Ki-67 labeling indexes of <5%, 5% to 9%, or ≥9%, instead of the currently used WHO recommendation of <2%, 2% to 20%, or >20%, correlated more strongly with outcomes in patients with PNETs, suggesting that reassessment of staging criteria is warranted. Another problem is that the 2010 WHO classification does not clearly include vascular invasion, which is an extremely important biologic characteristic of tumor cells. In our study, vascular invasion was related significantly to poor outcomes. Further studies of larger numbers of patients should be performed to confirm this finding.
Ito et al recently reported the results from a large epidemiologic study of PNETs in Japan. Both the annual prevalence and the annual incidence of PNETs were higher in Japan than in Western countries. This finding may be attributed to excellent health check-up programs in Japan, particularly the routine use of abdominal ultrasonography on a trial basis and easy access to advanced imaging examinations (eg, endoscopic ultrasonography, computed tomography, and magnetic resonance imaging). This medical environment is enabled by the National Health Insurance system in Japan. In our study, 42% of the 79 patients were asymptomatic. One of the characteristics of our hospital is that few patients have functional tumors or tumors associated with MEN-1. This is attributed to the observation that patients with functional tumors and classic hormone-related symptoms do not present at our hospital, because there is no department of endocrinology. The resection rate in our study was 75%, which was lower than that in the study by Ito et al. This lower resection rate suggests that many of our patients had nonfunctional tumors with a risk of metastasis and that most functional tumors were insulinomas, which have a low risk of metastasis.
In patients with PNETs, the main treatment goal is total surgical resection and control of tumor growth. At the time of diagnosis, about 20% of patients already have metastases to other organs. In addition to surgical resection to reduce tumor volume and local treatment, such as radiofrequency ablation and transcatheter arterial chemoembolization, medical therapy also plays an important role in the management of PNETs. With recent advances in molecular biology, high expression levels of multiple growth factors and their receptors have been demonstrated in NETs. Clinical trials of molecular-targeted agents that inhibit the proliferative activities of these factors and receptors have been conducted.[18, 19] Representative molecular-targeted agents include sunitinib, a tyrosine kinase inhibitor, and everolimus, a mammalian target of rapamycin (mTOR) inhibitor. Both of these inhibitors have produced good outcomes, suggesting that many patients with well differentiated PNETs would benefit from treatment. However, a regimen for NEC remains to be established, and outcomes remain extremely poor. Currently, combination chemotherapy, including a platinum compound, is used to treat NEC, similar to small cell lung cancer, which has similar pathologic and clinical features. Unfortunately, satisfactory results have not been obtained.[20, 21] In our study, SSTR-2a was positive in an appreciable proportion of patients with NEC and was associated with extremely poor outcomes. Therefore, future studies should investigate whether patients who have SSTR-2a-positive NEC are sensitive to somatostatin analogues.
Our study had several important limitations. It was a retrospective clinical trial of a relatively small group of patients with PNET who were treated at a single hospital. There were extreme imbalances between the number of patients with NEC versus those with NETG1 or NETG2, as well as between the number of patients with an SSTR 2A of score 0 versus those with scores from 1 to 3. Because our study period was approximately 24 years, therapy evolved considerably, and patients who received treatment in the early part of the study did not receive somatostatin analogues or molecular-targeted agents, which are now widely used to treat well differentiated tumors, NETG1, and NETG2. Such differences in treatment most likely had an appreciable impact on patient outcomes. Ideally, a large, multicenter, prospective study should be performed within a well defined period. Another limitation was that we used immunostaining to evaluate SSTR-2a expression. Evaluation of SSTR-2a mRNA expression would have been more quantitative, objective, and ideal. However, the results of immunohistochemical analysis are considered adequate. Evaluation of SSTR expression by indium-111 pentetreotide scintigraphy (octeroscan) would have been desirable, but this technique is not widely used as an examination in Japan, because it is not covered by the Japanese National Health Insurance system. Therefore, we did not use octreoscan imaging.
In our study, not only NEC but also an SSTR-2a score of 0 was a significant independent predictor of poor outcomes, suggesting that the evaluation of SSTR-2a may be useful for selecting treatment regimens and predicting outcomes in the future. Because a considerable proportion of patients with NEC have SSTR-2a–positive tumors, further analyses of the usefulness of somatostatin analogues are warranted in patients with these tumors.
No specific funding was disclosed.
The authors made no disclosures.