Expression of developing neural transcription factors in lung carcinoid tumors


  • Disclosure/conflict of interest: The authors declare no conflict of interest.


In lung tumors, the association between carcinoids and high-grade neuroendocrine tumors (HGNETs) is controversial. To understand the phenotypic similarities/differences between lung carcinoids and HGNETs, we comparatively investigated the expression of three kinds of developing neural transcription factors (DNTFs: BRN2, TTF1 and ASCL1) and multiple endocrine neoplasia type 1 (MEN1) as well as RB1 and P53 using 18 carcinoids and 16 HGNETs. The DNTFs were expressed in 10 of the 18 carcinoids and in all the HGNETs, while normal neuroendocrine cells, which are considered the major cell origin of lung carcinoids and small cell carcinomas, did not express DNTFs. Both the DNTF- and DNTF+ carcinoids contained typical and atypical carcinoids. All the DNTF- carcinoids examined were formed in the bronchial wall. All the MEN1- carcinoids examined were classified into the DNTF- carcinoids, while all the HGNETs expressed MEN1. This finding suggests that DNTF- MEN1- carcinoids are unlikely to be precursors of HGNETs. Although the status of RB1 and P53 between carcinoids and HGNETs were apparently different, the DNTF+ carcinoids of two male patients and one female patient revealed morphologies resembling HGNET cells and relatively high Ki67 indices. Further investigation of DNTF expression in carcinoids might provide important clues to understand the association between carcinoids and HGNETs.

The World Health Organization (WHO) classifies morphologically identifiable lung neuroendocrine (NE) tumors (NETs) into four categories: typical carcinoid (TC), atypical carcinoid (AC), small cell carcinoma (small cell lung cancer, SCLC) and large cell NE carcinoma (LCNEC).[1-3] Carcinoid tumors are histologically and clinicopathologically subclassified as TC (low-grade NET) or AC (intermediate-grade NET) according to mitotic count and the presence/absence of necrotic foci,[3] whereas SCLC and LCNEC are considered high-grade NETs (HGNETs).

The NET cells belonging to these distinct subsets are thought to originate from NE cells that exist solely or that form neuroepithelial bodies in normal airways; these NET cells share the morphologic, immunohistochemical and ultrastructural characteristics of NE cells. Accordingly, a spectrum of NETs ranging from low- to high-grade NETs has been proposed.[3, 4] However, because of differences in their incidence and clinical, epidemiologic, genetic, histologic and survival characteristics, these NETs are categorized separately from each other in the WHO classification.[1-3] Clinically, approximately 20–40% of patients with TCs and ACs are nonsmokers, whereas almost all patients with SCLCs or LCNECs are cigarette smokers. In terms of gene abnormalities, the mutation or inactivation of two major tumor suppressor genes, RB1 and P53, occurs frequently in SCLCs and LCNECs but is rarely observed in TCs. Furthermore, mutations in the multiple endocrine neoplasia type 1 (MEN1) gene and loss of heterozygosity at the MEN1 gene locus, 11q13, have been demonstrated in 30–40% of sporadic lung carcinoid tumors but not in HGNETs.[1, 5-11] These reports raise the question as to whether HGNETs and carcinoid tumors should be dealt as a spectrum of NETs. If not, then how do HGNETs arise? The histogenesis of SCLCs and LCNECs remains unclear, mainly because extremely early (preinvasive) SCLC/LCNEC lesions have not been isolated.

Since thyroid transcription factor 1 (TTF1) is stably expressed in peripheral lung epithelial cells consisting of terminal respiratory units (TRUs), such as nonciliated cuboidal bronchiolar epithelial cells (Clara cells), type II pneumocytes, and bronchiolar basal cells, and since lung adenocarcinomas that arise in TRUs frequently express TTF1, TTF1 is used as a diagnostic marker for lung adenocarcinomas exhibiting TRU differentiation.[12] However, more than 90% of SCLCs express high levels of TTF1.[13] We recently reported that BRN2, a neural lineage-specific homeoprotein also known as POU3F2, OCT7 or N-Oct3, is expressed and directly involved in TTF1 overexpression in SCLCs.[14] The crucial role played by BRN2 in the development of the endocrine hypothalamus, which is part of the diencephalon, suggests that TTF1 is an important transcription factor for fetal diencephalon morphogenesis. Accordingly, the expression of BRN2 in SCLCs suggests that SCLCs may possess a phenotype similar to that of developing neural cells.[15-17] The presence of TTF1 expression in lung carcinoid tumors is controversial because positivity for TTF1 in carcinoid tumors ranges from 0% to 69%, suggesting the existence of cellular phenotypic variations among lung carcinoid tumors.[18-20] Achaete-scute homolog-like 1 (ASCL1) is another developing neural cell-specific transcription factor, which contains a basic helix-loop-helix structure and is overexpressed in HGNETs of the lung.[21, 22] We recently reported that ASCL1 contributes to the expression of an NE marker gene, synaptophysin, and that ASCL1 expression is upregulated by BRN2.[23, 24] These facts suggest that an examination of the levels of BRN2, TTF1 and ASCL1 in carcinoid tumors might provide clues to understand the developmental neural character of carcinoid tumors and the relationship between carcinoid tumors and HGNETs.

In this study, we aimed to define the phenotypic similarities and differences among lung carcinoid tumors and HGNETs by examining the expression patterns of three types of developing neural cell-specific transcription factors (DNTFs: BRN2, TTF1 and ASCL1) and MEN1 in 18 cases of carcinoid tumors and 16 cases of HGNETs.

Materials and Methods


This study was approved by the institutional ethics board of Kyorin University School of Medicine. Eighteen surgically resected cases of lung carcinoid tumors (10 TCs and eight ACs) were collected from the pathological files of lung cancer patients who had been treated at Kyorin University Hospital between 1987 and 2014. Eight surgically resected cases of SCLCs and eight cases of LCNECs were also selected for comparative examinations. Histological diagnosis and pathological staging were performed according to the criteria of the WHO classification and the seventh lung cancer TNM classification.[1-3] The NE differentiation of the tumors was confirmed by positive immunohistochemical staining for the following NE markers: chromogranin A, synaptophysin and neural cell adhesion molecule 1 (NCAM1, also known as CD56). All the NE tumors assessed in this study expressed at least two of these three NE markers. The clinicopathological features of the patients are shown in Table 1. Several female patients had carcinoid tumors, whereas all eight SCLCs and seven of eight LCNECs examined were from males. Fourteen carcinoid tumors were diagnosed as pathological stage (p-stage) IA, three were as p-stage IB, and one was as p-stage IIA. None of these tumors have recurred. The SCLCs comprised four cases of p-stage IA, two cases of p-stage IB and two cases of p-stage IIIA. The LCNECs consisted of four cases of p-stage IA, one case of p-stage IB, one case of p-stage IIA, one case of p-stage IIB and one case of p-stage IIIA. The tumors have recurred in one SCLC patient (p-stage IIIA) and in two LCNEC patients (p-stages IIA and IIIA).

Table 1. Clinicopathological features of surgically resected lung neuroendocrine tumors examined in this study
 TC (n = 10)AC (n = 8)SCLC (n = 8)LCNEC (n = 8)
  1. TC, typical carcinoid; AC, atypical carcinoid; SCLC, small cell lung cancer; LCNEC, large cell neuroendocrine cancer; M, male; F, female.
Mean age (range) (years)57.9(40–83)61.5(42–73)66.8(56–73)70.3(52–82)
M : F2:85:38:07:1
Mean tumor size(range)(mm)20.8(5–35)18.0(7–48)21.8(8–36)24.3(14–28)

Histology and immunohistochemistry

Formalin-fixed, paraffin-embedded tissue sections containing NET lesions were examined in this study. Tissue sections 4 μm in thickness were cut from paraffin blocks and mounted onto silane-coated glass slides. The sections were deparaffinized in xylene, rehydrated in a graded series of ethanol solutions, and then either stained with hematoxylin and eosin (HE) or immunohistochemically analyzed using primary antibodies against Ki67, P53, RB1, BRN2, TTF1, ASCL1 and MEN1. Endogenous peroxidase activity was blocked with a 0.3% H2O2-methanol solution for 30 min at room temperature. Details of the primary antibodies and antigen retrieval treatment used in this study are presented in Supporting Table S1. After antigen retrieval, immunohistochemistry was performed using an EnVision detection system (DakoCytomation, Glostrup, Denmark) according to the manufacturer's instructions. Primary antibodies were applied to the slides for 1 h at room temperature or overnight at 4°C. After being thoroughly washed with Tris-buffered saline (pH 7.6) containing 0.05% Tween 20 (TBS-T) (DakoCytomation), the slides were incubated with a peroxidase-conjugated EnVision polymer and anti-rabbit and anti-mouse secondary antibodies (DakoCytomation) for 30 min at room temperature. After thorough washes, the peroxidase activity was visualized with 3-3′-diaminobenzidine tetrahydrochloride (DakoCytomation). The immunoreactivity, except for Ki67 immunostaining, was evaluated using the following scoring system (scores of 0 to 2): 0 indicated no staining; 1 indicated 1–50% positive cells; and 2 indicated > 50% positive cells. The Ki67 labeling index was calculated as the percentage of Ki67-positive cells in a sample of 5000 tumor cells.

Reverse transcription-polymerase chain reaction (RT-PCR)

RT-PCR analyses were conducted to confirm the expression of the DNTFs, BRN2, TTF1 and ASCL1. Tumor lesions were dissected from the formalin-fixed, paraffin-embedded (FFPE) tissue sections, and total RNA was extracted using an RNase-free deparaffinization solution (Qiagen, Hilden, Germany) and an RNeasy FFPE kit (Qiagen) according to the manufacturer's instructions. RT-PCR was performed using 500 ng of total RNA and an RNA-LAPCR kit (Takara, Shiga, Japan), according to the manufacturer's instructions. The nucleotide sequences of the forward and reverse primer sets are listed in Supporting Table S2. β-actin served as an internal control and 35 PCR cycles were performed.

Fluorescent double immunostaining

To examine whether normal NE cells expressed BRN2, TTF1 and ASCL1, we performed fluorescent double immunostaining with fluorescently labeled secondary antibodies on formalin-fixed, paraffin-embedded normal lung tissues that had been surgically resected due to lung adenocarcinomas. Synaptophysin was used as a marker for NE cells (Supporting Table S1). After antigen retrieval treatment (Supporting Table S1), a mixture of a rabbit monoclonal anti-synaptophysin antibody and the desired mouse monoclonal antibody was applied to the tissue section-mounted slides, which were then incubated overnight at 4°C. After being thoroughly washed with 15 mmol/L phosphate-buffered saline (PBS) (pH 7.4), the slides were incubated with a mixture of an Alexa Fluor 488-labeled anti-rabbit secondary antibody (1:500, 4412S, Cell Signaling Technology, Beverly, MA, USA) and an Alexa Fluor 594-labeled anti-mouse secondary antibody (1:500, 8890S, Cell Signaling Technology) for 1 h at room temperature. After thorough washes with PBS, the nuclei were counterstained with 4′6-diamidino-2-phenylindole (DAPI). The analyses were performed using a fluorescence microscope (Olympus, Tokyo, Japan).


Histological and immunohistochemical verification of lung carcinoid tumors

The patients' gender, smoking history and tumor-originating site, as well as the immunohistochemistry results for Ki67, RB1, and P53, are summarized in Table 2.

Table 2. Smoking history of patients, tumor-originating site, Ki67 labeling indices, and expressions of RB1, P53, BRN2, TTF1, ASCL1 and MEN1 in surgically resected lung neuroendocrine tumors examined in this study
CaseGenderBIOriginating siteKi67 (%)RB1P53BRN2TTF1ASCL1MEN1
  1. AC, atypical carcinoid; BI, Brinkman index; F, female; LCNEC, large cell neuroendocrine cancer; M, male; SCLC, small cell lung cancer; TC, typical carcinoid.

Eight of the 10 patients with TCs were female (80%), whereas five of the eight patients with ACs (63%) and 17 of 18 the patients with HGNETs (94%) were male. Five of the eight female patients with TCs (63%) and two of the three female patients with AC (67%) were never-smokers, whereas all the patients with HGNETs were smokers.

The tumor-originating site was defined as either the bronchus or the bronchiole. If the tumor tissue included bronchial cartilage and/or bronchial glands, the tumor was considered to originate from the bronchus. Otherwise, the tumor was considered to originate from the bronchiole. Nine of the 10 TCs (90%) originated from the bronchial wall, whereas two of the eight ACs (25%) originated from the bronchiolar wall. Although the data concerning tumor-originating site of the SCLC might be biased due to usage of surgically resected SCLCs in this study, five of the eight SCLCs (63%) and seven of eight LCNECs (88%) originated from the bronchiolar wall.

The Ki67 labeling indices of the TCs were less than 5%, whereas those of the ACs ranged from 5.2% to 19.2% (Fig. 1). These results were consistent with a previous report in which low-grade NE tumors (TCs) exhibited low proliferation rates (Ki67 indices ≤ 5%) and the Ki67 indices of intermediate-grade NE tumors (ACs) were between 5% and 20%.[25] All the SCLCs and LCNECs examined in the present work had much higher Ki67 labeling indices (ranging from 58.8% to 93.2%) than the TCs and ACs (Fig. 1).

Figure 1.

Histology and immunohistochemical staining for RB1, P53 and Ki67 in representative cases of typical carcinoid (TC), atypical carcinoid (AC), small cell lung cancer (SCLC) and large cell neuroendocrine cancer (LCNEC) of the lung. The TC (TC-2) and AC (AC-3, AC-4 and AC-8) cells shown here diffusely express RB1 and exhibit low or no P53 expression. The proportions of Ki67-positive cells were higher in the ACs, especially AC-3, AC-4 and AC-8, compared with TC (TC-2). In contrast, the SCLC (SCLC-1) and LCNEC (LCNEC-2) cells did not express RB1, which was expressed in inflammatory cells and stromal cells (arrows). P53 and Ki67 were diffusely expressed in the SCLC and LCNEC cells. Bar = 50 μm.

RB1 expression was diffusely present in all the carcinoid tumors but was absent from the SCLCs and LCNECs (Fig. 1). The P53 signals in the carcinoid tumors were focal and weak; however, a small number of tumor cells displayed intense P53 expression in two AC cases (AC-3 and AC-8) (Fig. 1). In contrast, all the SCLCs and LCNECs showed diffuse and intense P53 signals, suggesting the abnormal accumulation of altered P53 molecules (Fig. 1). These findings are consistent with previous reports.[1, 7-11, 26]

The expression of DNTFs and MEN1 in lung carcinoid tumors and HGNETs

The results of immunohistochemical staining for DNTFs and MEN1 are summarized in Table 2. BRN2 expression was observed in five of the 10 TCs (50%), four of the eight ACs (50%) and all the HGNETs (100%). TTF1 expression was detected in five of the 10 TCs (50%) and four of the eight ACs (50%), whereas most of the HGNETs, especially SCLCs, exhibited diffuse and strong TTF1 expression (Fig. 2). ASCL1 expression was observed in five of the 10 TCs (50%), five of the eight ACs (63%), and all the HGNETs (100%); the intensity in the SCLCs was conspicuously high (Fig. 2). The expression of BRN2, TTF1 and ASCL1 was confirmed by RT-PCR analyses using mRNA extracted from the NETs and specific primer sets for each molecule. The tumors in which BRN2, TTF1 or ASCL1 were detected immunohistochemically also expressed the respective mRNA (Fig. 3). In accordance with a previous report that MEN1 is expressed ubiquitously in any nucleated cell,[27] the non-neoplastic bronchial/bronchiolar epithelial cells, alveolar cells, interstitial cells, endothelial cells and infiltrating lymphocytes examined in the present work displayed positive signals for MEN1. However, the MEN1 expression was absent in two of the 10 TCs (20%) and two of the eight ACs (25%), whereas all the HGNETs expressed MEN1 (Fig. 2).

Figure 2.

Histology and immunohistochemical staining for BRN2, TTF1, ASCL1 and MEN1 in representative cases of typical carcinoid (TC), atypical carcinoid (AC), small cell lung cancer (SCLC) and large cell neuroendocrine cancer (LCNEC) of the lung. TC-2 and AC-1 did not express BRN2, TTF1, ASCL1 or MEN1, whereas endothelial cells in the tumors expressed MEN1 (arrows). TC-5, AC-3, AC-4 and AC-8 were positive for BRN2, TTF1, ASCL1 and MEN1 as well as SCLC-3 and LCNEC-2. Bar = 25 μm.

Figure 3.

Results of RT-PCR analysis of BRN2, TTF1 and ASCL1. Specific BRN2, TTF1 and ASCL1 signals were amplified from mRNA extracted from typical carcinoid (TC)-6 and atypical carcinoid (AC)-3. However, no specific signals were detected in TC-3 or AC-1, which also exhibited no immunohistochemical staining for BRN2, TTF1 or ASCL1. Small cell lung cancer (SCLC)-2 and SCLC-4 were used as positive controls. β-actin (Act-B) served as an internal control. bp, base pair.

Re-evaluation of lung carcinoid tumors in terms of their expression of DNTFs

Having observed that the expression of BRN2, TTF1 and ASCL1 was closely coordinated in the carcinoid tumors and HGNETs, we re-evaluated the carcinoid tumors with regard to their expression of these DNTFs. The clinicopathological findings are summarized in Table 3. The patients with carcinoids were relatively younger than those with HGNETs, and both the DNTF- and DNTF+ groups consisted of middle-aged and older patients. Half of the patients (50%) in the DNTF- group were male, whereas seven of the 10 patients (70%) in the DNTF+ group were female. Notably, all the DNTF- carcinoids were formed in the bronchial wall, whereas three of the ten DNTF+ carcinoids (30%) were located in the bronchioles. Both the DNTF- and DNTF+ groups contained TCs and ACs. MEN1-negative tumors were observed only in the DNTF- group; specifically, four of the eight DNTF- carcinoids (50%) were MEN1-negative. The Ki67 labeling indices of the DNTF+ group exhibited a wider range than those of the DNTF- group; this discrepancy was mainly caused by two male cases (AC-3 and AC-4) and one female case (AC-8) with relatively high Ki67 labeling indices (15.0%, 19.2% and 14.6%, respectively) (Table 2). The tumor cells of AC-3 and AC-4 had enlarged cytoplasm and nuclei with prominent nucleoli compared with the other carcinoid tumors, whereas the tumor cells of AC-8 had relatively scant cytoplasm and small nuclei with vesicular chromatins (Fig. 2).

Table 3. Clinicopathological features of surgically resected DNTF and DNTF+ carcinoids and their comparison with those of surgically resected SCLCs and LCNECs
 DNTF carcinoid (n = 8)DNTF+ carcinoid (n = 10)SCLC (n = 8)LCNEC (n = 8)
  1. AC, atypical carcinoid; DNTF, developing neural transcription factor; F, female; LCNEC, large cell neuroendocrine cancer; M, male; SCLC, small cell lung cancer; TC, typical carcinoid.
Mean age (range) (years)59.4(40–72)59.6(42–83)66.8(56–73)70.3(52–82)
M : F4:43:76:07:1
Originating site(bronchus : bronchiole)8:07:33:51:7
Histologic type(TC : AC)5:35:5--
MEN1 expression(negative : positive)4:40:100:80:8
Ki67 labeling index(range)(%)0.9-8.50.7–19.282.5–93.258.8–90.9

Expression patterns of DNTFs in normal NE cells

To compare the expression patterns of DNTFs in normal NE cells, which are considered the major cell type from which SCLCs originate, we performed immunofluorescent staining for BRN2, TTF1 and ASCL1. The immunofluorescent double-staining images demonstrated that the synaptophysin-expressing normal airway NE cells located in the bronchi, bronchioles and terminal to respiratory bronchioles did not express BRN2, TTF1 or ASCL1 (Fig. 4).

Figure 4.

Immunofluorescence microphotographs of normal bronchi, bronchioles and terminal to respiratory bronchioles. Double immunostaining was performed using anti-synaptophysin and anti-BRN2, -TTF1 or -ASCL1 antibodies. Neuroendocrine cells that expressed synaptophysin and that localized solely or formed clusters in the airway mucosa did not express BRN2, TTF1 or ASCL1. A developing neural transcription factor-positive (DNTF+) carcinoid tumor was used as a positive control. The nuclei are stained with 4′6-diamidino-2-phenylindole (DAPI). Bar = 25 μm.


The current WHO classification system categorizes lung carcinoid tumors as TCs or ACs based on the proliferative activities of NE tumor cells.[3] Because the metastatic potential and prognosis of ACs surpass those of TCs,[3, 28] this clinicopathological classification system is quite useful for diagnostic pathology. The purpose of this study was to compare the cellular characteristics of carcinoids and HGNETs using the immunohistochemical findings of BRN2, TTF1, ASCL1, MEN1, RB1, P53 and Ki67 in a series of lung NETs. We estimated carcinoid tumors from the standpoint of DNTF expression manner and found that half of the lung carcinoid tumors (9/18) expressed DNTFs. Therefore, we classified the carcinoid tumors into two phenotypic subtypes, DNTF- and DNTF+ and re-evaluated their clinicopathological features.

In this study, we assessed the expression of MEN1 using immunohistochemistry, since somatic mutations in the MEN1 gene have been reported in sporadic lung carcinoids and these mutations result in the truncation or (potentially) the complete loss of MEN1.[5] Consequently, we found that all four MEN1- carcinoids were classified into the DNTF- carcinoid group and that all the HGNETs examined were MEN1+. In a review article, Swarts et al. recently proposed that lung carcinoids and HGNETs are separate biological entities, rather than belonging to a single spectrum of lung NETs.[29] We partly agree with their opinion because our results reveal possibilities that MEN1- DNTF- carcinoids are phenotypically different from MEN1+ carcinoids and are unlikely to be precursors of HGNETs.

In accordance with previous reports,[1-3, 7-11, 25-27] the patients with carcinoid tumors in this study were not overwhelmingly male, whereas most of the patients with HGNETs were male. The remarkable gender disproportion and the differences in genetic abnormalities between carcinoids and HGNETs are the two main reasons why researchers cannot easily accept dealing with carcinoids and HGNETs as a spectrum of lung NE tumors.[30, 31] However, the possibility that some DNTF+ carcinoids are precursor lesions of HGNETs is suggested by the following findings: (i) the DNTF+ carcinoids and the HGNETs had similar tumor-originating loci and expression patterns of BRN2, TTF1, ASCL1 and MEN1, although their RB1 and P53 statuses were discrepant; and (ii) the DNTF+ carcinoid tumor cells exhibiting increased growth activity showed LCNEC cell-like enlarged cytoplasm and nuclei with prominent nucleoli (AC-3 and AC-4) or SCLC cell-like small-sized hyperchromatic nuclei and scant cytoplasm (AC-8). However, since none of the lung carcinoid tumors of our panel have recurred, we cannot obtain evidence for prognostic difference between DNTF- and DNTF+ carcinoids, nor for progression of DNTF+ carcinoids to HGNETs. Further investigations, including clinicopathological studies evaluating a greater number of carcinoid tumors, are necessary to define the association between lung carcinoids and HGNETs.

As mentioned above, it is of interest that AC-3 and AC-4, both which originated from the bronchiole, revealed LCNEC-like morphology and that AC-8, which originated from the bronchus, showed SCLC-like morphology. There are morphological differences between SCLCs and LCNECs,[1, 2] and the expression levels of RB1 have been reported to be higher in LCNECs than in SCLCs.[32, 33] Furthermore, as most LCNECs are generated in the peripheral areas of lung,[34] the tumor-originating loci of SCLCs and LCNECs examined in this study were rather different; seven of the eight LCNECs originated from the bronchiole, while SCLCs originated from either bronchus or bronchiole. However, it has been reported that SCLCs and LCNECs could not be clearly distinguished based on gene expression profiles.[35] D'Adda et al. demonstrated a close genetic relationship between SCLC and LCNEC elements in combined HGNETs, suggesting that both elements have a monoclonal origin from a common ancestor.[36] These reports suggest that SCLCs and LCNECs, both of which are categorized as HGNETs, have similar gene expression profiles that are associated with a malignant neuroendocrine phenotype, regardless of the morphology and originating locus/cell. Further investigations are also necessary to elucidate what causes morphological differences between SCLC and LCNEC and to resolve the long-standing issue concerning therapeutic strategy for LCNEC.


This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (grant #24590428 to Takuya Yazawa). We thank Ms. Michiru Umino and Ms. Ayumi Sumiishi for their technical support.