Gastroenteropancreatic (GEP) neuroendocrine neoplasms (NENs) are currently classified as well differentiated (low-grade to intermediate-grade) neuroendocrine tumors (NETs) and poorly differentiated (high-grade) neuroendocrine carcinoma (NEC) based on morphology and the proliferation rate. Recent molecular discoveries and therapeutic advances have been based on studies of well differentiated low-grade NETs. Much less is known about high-grade NEC, and no prospective studies have evaluated high-grade NEC originating outside of the lung. In the 2000 World Health Organization (WHO) classification, morphologically aggressive NENs were designated as “poorly differentiated endocrine carcinoma.” In the 2010 WHO classification, the nomenclature for grade 3 NETs was altered to “neuroendocrine carcinoma,” including small cell carcinoma and large cell NEC (Table 1). NECs are defined as tumors with poorly differentiated morphology and a higher proliferation rate than well differentiated NETs. NECs are characterized by a high proclivity for metastatic dissemination even among patients with clinically localized tumors. The gastroenteropancreatic tract is the most common site for extrapulmonary NEC, accounting for 35% to 55% of all NECs outside the lung.[3-7] Treatment strategies for extrapulmonary NEC are often extrapolated from the treatment paradigm for small cell lung cancer (SCLC).[4, 8] However, some have questioned the rationale for this, as there are many clinical differences between SCLC and extrapulmonary NEC: SCLC is more strongly associated with smoking, brain metastases are more common, and responses to platinum-based chemotherapy are generally better.[4, 8-14] In this context, several recent retrospective, multicenter studies have focused on novel treatment strategies for GEP-NEC.[15-17] In this review, we summarize the current knowledge on GEP-NEC, discuss potential pitfalls in the current classification of NEC, and review emerging data on treatment.
|Neuroendocrine Tumor (NET)||Neuroendocrine Carcinoma (NEC) Grade 3|
|NET grade 1||NEC grade 3|
|NET grade 2||• Large cell|
|• Small cell|
|Grade 1||Mitotic count <2 per 10 HPF and/or Ki-67 index ≤2%|
|Grade 2||Mitotic count 2–20 per 10 HPF and/or Ki-67 index 3%-20%|
|Grade 3||Mitotic count >20 per 10 HPF and/or Ki-67 index >20%|
Search Strategy and Selection Criteria
A review to identify studies on GEP-NEC was performed by searching electronic databases; scanning reference lists of relevant articles, conference abstracts, and trial databases; and consultation with experts in the field. No limits were applied for language. An electronic search was applied to PubMed (1966-present), Embase (1974-present), the ClinicalTrials.gov registry, and American Society of Clinical Oncology conference abstracts. The final search was performed on November 1, 2013. We used the following search terms: neuroendocrine or endocrine, tumor or cancer or carcinoma, poorly differentiated or high-grade or undifferentiated. Lung was used as an exclusion term. Studies that reported on primary tumors outside the gastroentreopancreatic tract and on unknown primary tumors were excluded. In total, 1277 studies from the Embase/PubMed search were selected. Finally, we identified 74 studies that were highly relevant and used them for the current review. Only studies that reported on GEP-NEC were included because of the significant differences between pulmonary and extrapulmonary small cell carcinomas and between NECs of the gastroenteropancreatic tract versus other primary sites (head and neck, genitourinary, etc).[8, 11-13]
Poorly differentiated NECs account for as much as 10% to 20% of malignant digestive NENs. They are mainly located in the esophagus, pancreas, ampulla of Vater, and large bowel but very rarely in the ileum.[9, 16, 21-23] The number of patients with GEP-NEC is increasing,[16, 24] probably because of more precise pathologic classification. The Netherlands Cancer Registry recently reported an increase in incidence of GEP large cell NEC during the last 2 decades from 0.15 to 0.36 per 100,000 inhabitants, whereas the incidence of GEP small cell carcinoma was unaltered (from 0.15 to 0.18 per 100,000 inhabitants). Metastatic disease at diagnosis was present in 50% to 70% of patients, and no improvement in 5-year survival was observed. On the basis of the Surveillance, Epidemiology, and End Results (SEER) database, colorectal NEC has an incidence of 0.2 per 100,000 inhabitants, and 62% of patients have distant disease at diagnosis.
Diagnosis and Classification
Diagnosis of NEC is difficult and often is based on small biopsy specimens. Misclassification is frequent. NECs are characterized by the expression of immunohistochemical markers of neuroendocrine differentiation. Synaptophysin staining is usually positive, whereas chromogranin A (CgA) staining may be negative. NEC encompasses 2 histologic entities: small cell carcinoma and large cell NEC. Awareness of the latter is essential, because large cell NECs can be difficult to distinguish from poorly differentiated adenocarcinomas, and the need for routine immunostaining for synaptophysin and CgA in such tumors has been suggested. Up to 40% of NECs contain elements of non-neuroendocrine histology, which varies depending on the primary location. By definition, the neuroendocrine component in a mixed tumor has to exceed 30% to be classified as an NEC; otherwise, it is classified as an adenocarcinoma with neuroendocrine differentiation. Tumors are classified as mixed adenoneuroendocrine carcinoma (MANEC) if the gland-forming elements exceed 30%. The prognostic or predictive significance of a non-NEC component remains uncertain,[27-29] and, currently, treatments for NEC and MANEC are similar. The molecular pathogenesis of GEP-NEC is uncertain. It has been suggested that NECs either originate from multipotent stem cells, which also may be the origin of non-neuroendocrine adenocarcinomas, or arise from the dedifferentiation of well differentiated NENs.[30-35] The latter possibility is less likely, because it is rare to identify tumors that have separate components of well differentiated NET and poorly differentiated NEC.
The optimal histologic classification of NETs remains controversial.[16, 36-39] Current prognostic stratifications are based on tumor differentiation and grade. Differentiation refers to the resemblance of tumor histology to the normal counterpart cells, whereas grade refers to the proliferative activity of the tumor, which is measured according to either the mitotic rate or the Ki-67 index. In the WHO 2010 classification, high-grade NECs are defined as tumors with a high proliferation rate (>20 mitotic figures per 10 high-power fields [HPF] or a Ki-67 index >20%). These thresholds were established by the European Neuroendocrine Tumor Society (ENETS) based on a consensus of experts and are somewhat arbitrary. The classification differs from that used to define high-grade NECs in the lung and thymus (>10 mitotic figures per 10 HPF and no use of the Ki-67 index). Mitotic rate and Ki-67 values can be discordant, and measurement of the Ki-67 index often leads to a higher grade assignment. For example, among 65 patients who had well differentiated pancreatic NETs with a mitotic rate of 2 to 20 per 10 HPF, 23 individuals had a Ki-67 index above 20% and, thus, had their tumors reclassified as high-grade. The ENETS classification system has been formally validated in prognostic studies; however, the validation was only for its prognostic relevance and not to optimally stratify therapy.
In recent years, several studies have challenged the assumption that poorly differentiated histology and high tumor grade are equivalent. Some NETs that are histologically well differentiated contain foci with an elevated proliferative rate (usually the Ki-67 index) that places them into the high-grade category.[37-39, 41] In a study of 104 patients with large cell GEP-NEC, 20% of tumors were characterized as well differentiated despite a Ki-67 index >20%. In another series, half of patients with high-grade non-small cell NEC had histologically well differentiated tumors. The survival of patients who have tumors that are well differentiated but high-grade appears to be significantly longer than the survival of patients who have poorly differentiated high-grade tumors (40 months vs 17 months). In another study, patients with well differentiated high-grade pancreatic NETs had shorter overall survival (OS) than patients with intermediate-grade tumors (32 months vs 63 months) but longer OS than patients with poorly differentiated NEC (15 months). These studies suggest that the current WHO high-grade category should distinguish between high-grade well differentiated NET and high-grade poorly differentiated NEC. Adequate diagnosis by a systematic review of the pathology by expert pathologists and a shared definition of NEC based on tumor differentiation and grade constitute critical prerequisites for future progress.
Diagnostic workup of NEC differs from diagnostic evaluation of well differentiated NET. The role of tumor markers in NEC is not well established. It was recently demonstrated that CgA was elevated in approximately 66% of 188 patients who had advanced GEP-NEC. Neuron-specific enolase has been tested as a marker primarily in patients with pulmonary NEC.[42, 43] Routine measurement of urine 5-hydroxyindoleacetic acid is not useful, because GEP-NECs are rarely associated with serotonin secretion. Usually, 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) is strongly positive in NEC and can be useful for staging, especially in patients with locoregional presentation. The relevance of somatostatin receptor imaging is uncertain.[15-17, 45, 46] A positive imaging may indicate a relatively well differentiated tumor despite high proliferative activity. In contrast to patients with pulmonary NEC (SCLC), imaging of the brain is only indicated in patients with GEP-NEC who are symptomatic.
Prognosis varies based on disease stage at diagnosis. SEER data on 4054 patients with poorly differentiated NEC indicated a median survival of 34 months for patients with localized disease, 14 months for patients with regional disease, and 5 months for patients with distant disease. However, those results included pulmonary primaries. Therefore, we performed a separate analysis of high-grade gastrointestinal NECs in the SEER data that were diagnosed from 1973 to 2012 and identified 2546 patients. That analysis indicated that distant disease was present at initial diagnosis in 57% of patients. The median survival was 38 months (95% confidence interval [CI], 31-45 months) for patients with localized disease, 16 months (95% CI, 15-17 months) for patients with regional disease, and 5 months (95% CI, 4.7-5.4 months) for patients with distant disease (Fig. 1). For the patients who had distant disease at diagnosis, the 2-year survival rate was 11%, and the 3-year survival rate was 8%. Long-term survival was poor even for patients who initially had localized disease (Fig. 1),[12, 29] illustrating that micrometastases often are present at diagnosis or that the staging of patients is suboptimal. Tumor histology (large cell vs small cell) appears to be a significant prognostic factor. In a study, the 5-year survival rate for patients with large cell GEP-NEC was 32%, whereas patients with small cell GEP-NEC had a 5-year survival rate of 6%. The striking difference in 5-year survival was present in all disease stages: localized disease (large cell vs small cell GEP-NEC, 61% vs 12%, respectively), regional disease (37% vs 17%, respectively), and metastatic disease (18% vs 2%, respectively). The 5-year, stage-specific survival of patients with large cell GEP-NEC was higher for those who had pancreatic primaries compared with gastric and colorectal primaries.
The survival of patients with advanced disease is short (Table 2). The 5-year survival rate for patients with poorly differentiated NEC varies between 6% and 11%, depending on European region. Poor performance status, high proliferation rate, elevated lactate dehydrogenase, and thrombocytosis at baseline are negative prognostic factors.[16, 22] Patients who have metastatic disease with a Ki-67 index <55% have longer survival than patients who have tumors with a higher Ki-67 index.[15, 16, 51, 52] The primary tumor location also appears to be prognostic. The median survival for patients with advanced pancreatic NEC was 15 to 21 months in 2 recent studies[16, 49] compared with 11 months for all patients with GEP-NEC. However in a Japanese study, survival was shorter for patients who had hepatobiliary and pancreatic primaries compared with those who had other primary gastrointestinal tumor locations (7.9 months vs 13 months). In contrast to the epidemiologic data reported by Korse et al, many studies have not reported any significant difference in survival between patients with small cell NEC and large cell NEC.[16, 26, 28, 47] Patients with tumors that stain immunohistochemically for both synaptophysin and CgA have a better prognosis[15, 46]. Immunoreactivity for CD117 (a cell stem marker), and vascular invasion, are histologically negative prognostic markers for survival.
|Reference||No. of Patients||Cohort||Median Survival (95% CI), mo||2-Year Survival, %||3-Year Survival, %|
|Yao 2008||2027||All NEC (including lung)||5 (4.5-5.5)|
|SEER Program 2013||1389||GEP-NEC||5 (4.7-5.4)||11||8|
|Sorbye 2013||252||GEP-NEC; chemotherapy treated||11 (9.4-12.6)||14||9.5|
|Sorbye 2013||53||GEP-NEC; no treatment||1 (0.3-1.8)|
|Machida 2012||258||GEP-NEC; chemotherapy treated||11.5|
|Bernick 2004||38||Colorectal small cell NEC||10.5 (6.7-19)||26||13|
|Smith 2013||126||Colorectal NEC||13||5|
|Fuji 2001||53||Gallbladder, small cell NEC; chemotherapy treated; review||8||0|
|Strosberg 2011||32||Pancreatic NEC||21|
Surgery, neoadjuvant (perioperative) treatment, and adjuvant treatment
Guidelines for the treatment of limited disease are inconsistent. Retrospective studies have demonstrated that surgery alone is rarely curative, even for apparently localized disease. In a study of 199 patients with small cell esophageal carcinoma, survival was 20 months for patients who received systemic chemotherapy in addition to local treatment versus only 5 months if systemic chemotherapy was not received. Induction chemotherapy followed by consolidative chemoradiation in 25 patients who had limited esophageal small cell NEC led to a median survival of 20 months. Surgery may not be a necessary component of initial therapy for esophageal NEC in patients who have a poor performance status or locally advanced disease. Several authors suggest a sequence of neoadjuvant chemotherapy followed by definitive surgery.[10, 55] Altogether, most data support the view that, for patients who have limited disease, a combination of systemic platinum-based chemotherapy with local treatment consisting of radiotherapy, surgery, or both offers the best chance for long-term survival. Recent North American NeuroEndocrine Tumor Society (NANETS) guidelines for locoregional, poorly differentiated, extrapulmonary NEC recommend surgical resection for T1/T2N0 disease along with postoperative chemotherapy. When local treatment is needed in excess clinical stages, surgery or radiation can be considered if morbidity is low. Surgical metastatectomy is not recommended in the management of NEC. In small series, the median survival after partial hepatectomy for metastatic GEP-NEC was 6 to 15 months.[57, 58] However, there have been case reports of long-term survival after surgery or locoregional treatment for metastatic disease.[59, 60]
Most recurrences are distant and not local. The aggressive behavior of NEC warrants the consideration of adjuvant chemotherapy, although no prospective studies support this practice. For resected patients, NANETS guidelines recommend adjuvant therapy with 4 to 6 cycles of cisplatin or carboplatin plus etoposide.[3, 56, 61, 62] Postoperative radiologic staging, including FDG-PET imaging, should be considered before the initiation of adjuvant chemotherapy, as rapid recurrence often occurs after surgery.
First-line palliative chemotherapy
Metastatic GEP-NEC is an aggressive disease in which rapid referral to an oncologist is necessary. After a diagnosis of advanced disease, the median survival is only 1 month without chemotherapy, suggesting that the benefit from palliative chemotherapy may be substantial. Guidelines for the treatment of advanced GEP-NEC advocate the use of platinum-based chemotherapy combined with etoposide,[3, 61-63] although, until recently, only a few small first-line studies have been reported (Table 3). Treatment with cisplatin/etoposide has resulted in high response rates (range, 42%-67%) and a median survival of 15 to 19 months.[64, 65] A 3-drug regimen (carboplatin, etoposide, and paclitaxel) produced a higher response rate, but median survival did not seem to improve, and grade 3 or 4 toxicity was frequent. Recent data suggest that response rates and survival in patients with GEP-NEC may be lower than previously reported (Table 3). In a recent retrospective study, 252 patients with advanced GEP-NEC received cisplatin/etoposide or carboplatin/etoposide. In that study, the response rate was 31%, the median progression-free survival (PFS) was 4 months, and the median survival was 11 months. No differences in outcome were observed when comparing cisplatin-based treatment versus carboplatin-based treatment (Fig. 2A). The variables performance status, Ki-67 index, platelet count, lactate dehydrogenase, and primary tumor location were prognostic for response and survival (Fig. 2B-2D). Multivariate analyses identified performance status as the strongest prognostic factor. A Ki-67 threshold of 55% was predictive for response to first-line platinum-based chemotherapy. Tumors with a Ki-67 index <55% were much less responsive to platinum-based chemotherapy (response rate, 15% vs 42%), but the patients with those tumors had a significantly longer survival (median OS, 14 months vs 10 months) compared with the patients who had higher Ki-67 levels (Fig. 3). Given the limited benefit of platinum therapy for patients with a Ki-67 index <55%, it seems appropriate to consider other regimens in the first-line setting for these patients, such as temozolomide-based therapy. In a retrospective study of 258 patients who had poorly differentiated GEP-NEC, most patients (62%) received irinotecan/etoposide as first-line treatment. A multivariate analysis identified a primary hepatobiliarypancreatic tumor and a performance status of 2 as unfavorable prognostic factors for survival. Although the response rate and survival were numerically better for irinotecan/etoposide treatment compared with cisplatin/etoposide treatment, the treatment schedule was not an independent factor for survival.
|Reference||Chemotherapy Schedule||No. of Patients||RR, %||SD, %||PD, %||PFS, mo||Median OS, mo|
|Moertel 1991||Cisplatin and etoposide||18||67a||33||0||8||19|
|Mitry 1999||Cisplatin and etoposide||41 (27 GEP-NEC)||42||34||24||9||15|
|Hainsworth 2006: Unknown primary||Carboplatin, etoposide, and paclitaxel||78||53||30||9||7.5||14.5|
|Sorbye 2013||Cisplatin/etoposide or Carboplatin/etoposide||252||31||33||36||4||11|
|Machida 2012||Cisplatin/etoposide or Irinotecan/etoposide||258||50/27||5.2||11.5|
Second-line palliative chemotherapy
After first-line treatment, no further standard therapy has been established for GEP-NEC. Retreatment with cisplatin/carboplatin and etoposide is a valid option after a treatment break of at least 3 months, as it has been demonstrated that retreatment results in a response rate of 15%, and another 27% of patients achieve stable disease. Recently, several small, retrospective, second-line studies have been published (Table 4). Temozolomide was given with or without capecitabine and bevacizumab in a cohort of 25 patients with NEC after progression on cisplatin-based chemotherapy. The response rate was 33%, and the median OS was 22 months. On the basis of that study, temozolomide-based chemotherapy is commonly used as second-line treatment for NEC. In the same study, a Ki-67 index <60% was predictive for response to treatment and survival. Less encouraging results from a Danish second-line study with temozolomide monotherapy may have been due to the large percentage of patients with poor performance status. Again, patients with a Ki-67 index in the lower range (<55%) had an improved median survival (10.9 months vs 2.7 months). Irinotecan-based and oxaliplatin-based chemotherapy may have a role as second-line treatments after etoposide/platinum chemotherapy.[17, 52] PFS and OS were superior for patients who had a Ki-67 index <55% (median PFS, 6.2 months vs 3.6 months; median OS, 19.5 months vs 8.5 months). When patients received third-line chemotherapy (various schedules), the response rate was 7%, and another 34% achieved stable disease. Topotecan is recommended in several guidelines based on data from studies of small cell lung cancer, but there is no independent confirmation of the activity of topotecan in GEP-NEC. In summary, these studies show that many patients with GEP-NEC will benefit from additional lines of chemotherapy after failure on platinum/etoposide treatment.
|Reference||Chemotherapy Schedule||No of Patients||RR, %||SD, %||PD, %||PFS, mo||Median OS, mo|
|Welin 2011||Temozolomide ± capecitabine ± bevacizumab||25||33||38||29||6||22a|
|Sorbye 2013||Various: Taxane-based, n = 22; temozolomide-based, n = 35||100||18||33||49||3c||19a|
Other treatment options
There are no data supporting the use of somatostatin analogs in NEC. Peptide receptor radionucleotide therapy (PRRT) with radiolabeled somatostatin analogs is used regularly to treat well differentiated NET, whereas the benefit of PRRT for patients with high-grade NET is unproven. However, case reports have demonstrated long-lasting partial responses after PRRT treatment in NEC patients with high somatostatin receptor expression.[69-71] The role of liver embolization in NEC is unknown.
The current WHO classification places all neuroendocrine GEP tumors with a Ki-67 index >20% in a “poorly differentiated/high-grade” category, and this has probably led to grade migration. Tumors that were once considered “moderately differentiated” or “intermediate grade” are now included in the same category as small cell carcinomas. This has likely led to an overuse of platinum-based chemotherapy in patients who might benefit from other systemic treatments, such as temozolomide-based or streptozocin-based chemotherapy, targeted agents, or PRRT. Therefore, there is need for an improvement of the WHO classification. One proposed modification would involve the separation of high-grade tumors according to morphology, creating a “well differentiated high-grade NET” category and a “poorly differentiated high-grade NEC” category.[37-39] Another proposal is to separate high-grade tumors into 2 proliferative categories: those with a Ki-67 index from 20% to 55% and those with a Ki-67 index >55%.[16, 51, 52] Until a better classification is achieved for patients with NEC who have a Ki-67 index in the relatively lower range (20%-55%), additional factors like morphology (poorly differentiated vs well differentiated), mitotic rate, and clinical course also should be considered before selection of treatment. Clinical trials with drugs traditionally used for well differentiated NETs are now starting in patients with high-grade NEC. An ongoing French study is evaluating sunitinib in GEP-NEC with the objective of identifying predictive molecular markers of response to sunitinib (National Clinical Trial identifier NCT01215578). The mammalian target of rapamycin (mTOR) pathway is up-regulated in 70% to 80% of NEC tumors compared with only 27% of well differentiated tumors,[72, 73] and the mTOR inhibitor everolimus demonstrated an antitumor effect in a preclinical GEP-NEC model. A Nordic phase 2 study using everolimus and temozolomide as first-line treatment in patients who have metastatic GEP-NEC with a Ki-67 index from 20% to 55% will start accrual in 2014. A US intergroup is proposing a randomized phase 2 study comparing platinum/etoposide versus temozolomide/capecitabine as first-line treatment for metastatic GEP-NEC. A systematic pathologic review by expert pathologists will be critical to avoid the misinterpretation of new study results and to better understand the place of new therapeutic options within the “new subgroups” of NEC. Ongoing prospective registries for GEP-NEC will provide valuable new information in the future.[23, 38]
After decades with scarce knowledge about high-grade GEP NENs, new data are beginning to emerge. It is increasingly clear that the current WHO high-grade NEC category needs to be refined, because it includes tumors that are highly aggressive and platinum-sensitive as well as tumors that are less aggressive and relatively resistant to platinum-based chemotherapy. Prospective clinical trials as well as tumor registries will allow for the development of more sophisticated grading classifications and will provide clinicians with better prognostic and predictive tools for the selection of treatment.