INTRODUCTION

The histologic subtyping of neuroendocrine lung tumors (NLT) is of critical clinical importance. The morphologic spectrum of typical carcinoid tumor (TC), atypical carcinoid tumor (AC), large cell neuroendocrine carcinoma (LCNEC), and small cell carcinoma (SCC) is associated with increased loss of differentiation and aggressive clinical behavior (1). Often, the histologic distinction between TC and AC can be subtle, and numerous classification schemes have been proposed (2, 3). Using strict and clarified criteria set forth by Travis et al. (4), the distinction between TC and AC has become increasingly important because of vastly different patient survival data. The 10-year survival of patients with TC is 87%, but this drops dramatically to 35% for patients with AC. At the opposite end of the NLT spectrum, high-grade neuroendocrine carcinomas (LCNEC and SCC) exhibit morphological diversity, but similar genetic abnormalities (1, 5, 6).

Differences between low (TC), intermediate (AC), and high-grade (LCNEC and SCC) NLT can be assessed by immunophenotyping for neuroendocrine markers, cell cycle abnormalities, and angiogenesis. In NLT, several studies have shown that increased Ki67 or MIB-1 expression, as a marker of increased proliferation, correlates with histological subtype and may be of prognostic value (7, 8, 9, 10, 11). In contradistinction, there have been few studies on angiogenesis in NLT. Slodkowka et al. (12) found no correlation between angiogenesis and the presence of regional lymph node metastases in TC or AC, but the full spectrum of NLT was not assessed. In this study, we analyzed angiogenesis and proliferation rates in the entire spectrum of NLT to shed light on potential differences in tumor biology.

MATERIALS AND METHODS

Twenty formalin-fixed, paraffin-embedded NLT (5 TC, 5 AC, 5 LCNEC, 5 SCC) were retrieved from the files in the Department of Pathology of Emory University Hospital, Atlanta, Georgia. The tumors were classified by one of the authors (AAG) according to the World Health Organization classification of lung tumors, based on the criteria proposed by Travis et al. for NLT (4, 13). The 4-μm sections were subjected to steam heat-induced epitope retrieval and then immunostained using an avidin-biotin complex technique (Ventana, Tucson, AZ). The cases were stained for CD 31 (DAKO, Carpinteria, CA, 1:50) and MIB-1 (1:50; Immunotech, Westbrook ME). 3,3′-diamino-benzidine was used as chromogen, and hematoxylin as counterstain. Positive controls included myometrium (CD 31) and tonsil (MIB-1). Negative controls had specific antibody replaced by buffer.

Angiogenesis was assessed by two authors (ZKA and CC) by quantitation of mean vessel density (MVD) in 2 high-power fields (200 ×) in one “hot spot” according to the method described by Weidner et al. (14). Cellular proliferation was assessed by manually counting the number of MIB-1 nuclear-positive cells in a 1000 cell count, converted into a percentage of total cells. All of the cases were scored by a single observer (ZKA). Analysis of variance was performed using ANOVA.

RESULTS

Representative photomicrographs of low- (TC) and high-grade NLT (LCNEC) stained with hematoxylin and eosin are depicted in Figure 1, A–B, respectively. The mean MVD was similar in all four tumor types (P =.395) with no statistically significant difference (Fig. 1C–D; Fig. 2).

FIGURE 1
figure 1

A, typical carcinoid tumor, hematoxylin and eosin stain (100 ×). B, large cell neuroendocrine carcinoma, hematoxylin and eosin stain (100 ×). C, typical carcinoid tumor, mean vessel density, CD-31 immunostain (100 ×). D, large cell neuroendocrine carcinoma, mean vessel density, CD-31 immunostain (100 ×). E, typical carcinoid tumor, cellular proliferation, MIB-1 immunostain (100 ×). F, large-cell neuroendocrine carcinoma, cellular proliferation, MIB-1 immunostain (100 ×).

Full size image
FIGURE 2
figure 2

Mean vessel density in neuroendocrine lung tumors. TC, typical carcinoid tumor; AC, atypical carcinoid tumor; LCNEC, large-cell neuroendocrine carcinoma; and SCC, small-cell carcinoma.

Full size image

Proliferation rates are significantly different between SCC and the three types of carcinoid tumor (P <.05; Fig. 1, E–F; Fig. 3). Analysis among the subtypes of carcinoid tumor revealed proliferation rates that were not significantly different. Hence, proliferation rates did not distinguish between TC and AC, nor between SCC and LCNEC. These latter two NLT had much higher proliferation rates than carcinoid tumors, but their rates were not significantly different.

FIGURE 3
figure 3

Proliferation rates of neuroendocrine lung tumors. TC, typical carcinoid tumor; AC, atypical carcinoid tumor; LCNEC, large-cell neuroendocrine carcinoma; and SCC, small-cell carcinoma.

Full size image

DISCUSSION

The pathogenesis of NLT is not well understood, despite increasing awareness of their morphological diversity. The separation of high- from low-grade NLT is prognostically relevant and generally reproducible among experienced pathologists (3). However, differentiation of intermediate-grade NLT may be diagnostically difficult because AC may demonstrate a propensity to invade and metastasize (2, 4, 15). To explain differences in clinical behavior, various mechanisms in tumor biology are under investigation.

Angiogenesis is a necessary process in tumor progression (16). Increased angiogenesis is required for tumor growth in animal models, whereas inhibition of angiogenesis results in increased tumor apoptosis and decreased metastases (17). Weidner et al. (14) have shown that microvessel count, a marker of angiogenesis, correlates with clinical outcome in carcinoma of the breast. Since that initial study, numerous studies in other solid tumors have confirmed the role of angiogenesis with tumor progression in vivo and have also confirmed that elevated microvessel counts are associated with poor prognosis (18, 19).

The addition of dominant oncogenes in vitro and in vivo to cells with defined mutations in tumor suppressor genes confers an angiogenic phenotype (20, 21, 22). These studies parallel the discovery of dominant oncogenes, whose transforming ability was discovered by their ability to convert preneoplastic cells with deletions in tumor suppressor genes into fully malignant and angiogenic neoplasms (23, 24, 25, 26). The loss of a nuclear tumor suppressor gene is an early event in the elaborate sequence that leads to tumorigenesis. There are, however, several examples of poorly angiogenic precursor lesions, including those of early squamous cell carcinomas of the skin, cutaneous melanomas, and cervical intraepithelial neoplasia/carcinoma (27, 28).

We have shown that NLT with good prognosis, such as TC, are angiogenic, yet they may reach appreciable sizes without clinical evidence of metastatic growth. Other examples of angiogenic neoplasms include angiomyolipomas of tuberous sclerosis (Arbiser et al., submitted), hemangioblastomas of von Hippel-Lindau disease, and Spitz nevi (29, 30, 31).

Our study showing that proliferation rate, but not MVD, correlates with NLT subtype and suggests that carcinoid tumors are highly vascular neoplasms with indolent behavior. Proliferative ability in vivo correlates with a progressive degree of cellular atypia and loss of tumor differentiation. Thus, the acquisition of the angiogenic phenotype is an early event, whereas tumor proliferative capabilities appear to be a later event. Recent studies of NLT have shown activation of ras (32, 33) or loss of function of menin in a subset of TC to be a common early event in low-grade NLT, whereas loss of p53 is seen in high-grade NLT (6, 32, 33, 34, 35, 36, 37).

CONCLUSION

Two major conclusions can be inferred from our study. First, increased proliferation rates are associated with loss of tumor differentiation, particularly in high-grade NLT, such as in LCNEC and SCC. Second, MVD is not appreciably different among the various NLT tumor types. These observations may be useful prognostically and in the development of novel preventive and therapeutic modalities for certain NLT.