The Potential Role of Flow Cytometry in the Diagnosis of Small Cell Carcinoma

The diagnosis of small cell carcinoma (SCC) is typically based on a combination of characteristic morphology and neuroendocrine phenotype as demonstrated by immunohistochemical (IHC) staining. Among the neuroendocrine markers consistently expressed by SCC is the neural cell adhesion molecule, a member of the immunoglobulin superfamily of adhesion molecules, which play a role in cell-cell adhesion in normal and malignant tissues, especially in tumors with neuroendocrine differentiation.1,2 Neural cell adhesion molecule can exist in several isoforms, one of which, the 140-kd isoform, has been shown to be identical to the human leukocyte differentiation antigen CD56.3 

Small cell carcinoma is usually densely cellular and easily disaggregated by mechanical means. These features, together with expression of CD56 in virtually all cases, make SCC an excellent candidate for phenotypic analysis by flow cytometry (FCM).

Although the medical literature contains numerous articles on FCM analysis of the DNA content of SCC,4–6 there is virtually no information available on the contribution of FCM-determined immunophenotyping to the diagnosis of SCC.7 From specimens submitted to IMPATH, Inc (New York, NY) for FCM in the last 2 years, we identified 38 cases whose FCM characteristics and morphologic features suggested a diagnosis of SCC. Where sufficient tissue was available (27 cases), we applied a panel of conventional IHC markers used to identify neuroendocrine neoplasms. The results of FCM and IHC were compared, with attention to CD56 expression. We sought to determine the potential diagnostic usefulness of FCM in SCC, based on these 27 cases.

Twenty-seven cases were selected for IHC staining based on FCM demonstration of a clearly defined population of CD56⁺CD45⁻ cells along with cytologic/morphologic features of SCC, namely, cohesive, small to medium-sized cells with condensed stippled nuclear chromatin, inapparent nucleoli, nuclear molding, and often necrosis and streaming of nuclear chromatin. Tissue from the following sites was evaluated: intrathoracic, including mediastinum (n = 5) and lung (n = 1); peripheral lymph nodes (n = 15); bone marrow (n = 1); liver (n = 3); submandibular gland (n = 1); and nasal cavity (n = 1). Three cases represented tissue aspirates (1 bone marrow, 2 liver); the remaining 24 cases consisted of tissue biopsies.

Immunophenotyping by standard 3- or 4-color FCM was performed in accordance with guidelines outlined in the 1995 US-Canadian consensus conference on FCM.8 Briefly, cell suspensions were prepared from solid tissues using a manual dispersion method. Isolated cells were preincubated in RPMI 1640 medium supplemented with 10% fetal bovine serum to minimize nonspecific binding of antibodies. Erythrocytes in bone marrow specimens were lysed with a 0.008% solution of ammonium chloride. Cells were washed with phosphate-buffered saline and incubated with Becton Dickinson cocktails of antibodies conjugated to fluorescein isothiocyanate, phycoerythrin, peridinin chlorophyll protein, or allophycocyanin. The antibodies chosen comprised a panel used by IMPATH, Inc, to evaluate and characterize lymphoproliferative disorders. The last 12 cases also included a fluorescein isothiocyanate–conjugated antibody to cytoplasmic cytokeratin (Dako Corporation, Carpinteria, Calif). Data acquisition and analysis were performed on a FACScalibur FCM instrument. CellQuest software (Becton Dickinson Immunocytometry Systems, San Jose, Calif) was used for data analysis. Antigen expression as assessed by fluorescence intensity was depicted on multiple dual-parameter scattergrams. Nonviable cells were excluded by using 7-amino actinomycin D (7-AAD). Matched isotype controls were used in all FCM panels. Cells were considered positive for antigen expression if they were present in a well-defined population whose median fluorescence intensity was approximately 1 or more logs greater than that of its matched isotype control.

Tissue from 27 cases was available for immunostaining. After deparaffinization and standard antigen retrieval, immunostaining was performed on 4-μm tissue sections using the TechMate 500 automated immunostainer (Ventana Medical Systems, Tucson, Ariz) and the EnVision detection system (Dako). The following antibodies were used (name, dilution, and manufacturer): CAM5.2, 1:200, Becton Dickinson; AE1/AE3, 1:100, Dako; synaptophysin, 1:300, Dako; chromogranin, 1:16 000, Dako; neuron-specific enolase, 1:4000, Dako or Chemicon International, Inc, Temecula, Calif; CD56, 1:100, Neomarkers, Inc, Fremont, Calif; thyroid transcription factor-1 (TTF-1), 1:1000, Neomarkers.

The Table summarizes the pertinent flow cytometric and immunohistochemical findings. The SCC cell size, as determined by forward scatter, was always large, with only mild internal complexity (side scatter). The intensity of CD56 expression was moderate to bright (approximately 100–1000 times that of the matched isotype control) in 25 cases and dim to moderate (10–100 times that of the control) in 2 cases (Figure, A). CD38 expression was negative in all cases, arguing against a plasma cell neoplasm. Cytokeratin testing was performed in 12 cases; all exhibited positivity (1 dim, 10 moderate, 1 bright), confirming the epithelial nature of the nonhematopoietic neoplastic population (Figure, B).

Twenty-six of 27 cases in which tissue was available for IHC staining exhibited a profile consistent with a carcinoma of neuroendocrine differentiation, that is, positivity for cytokeratin and for 1 or more conventional neuroendocrine markers. (Although the AE1/AE3 stain yielded an indeterminate result in case 2, the case was felt to be an SCC based on cytomorphologic features and positivity for synaptophysin and neuron-specific enolase.) The results were as follows (No. positive/No. studied): AE1/AE3, 22/23; CAM5.2, 24/25; TTF-1, 16/24; CD56, 25/25; chromogranin, 20/27; synaptophysin, 23/25; and neuron-specific enolase, 24/25. Thyroid transcription factor-1 is a transcription factor expressed in epithelial cells of lung and thyroid origin and is usually negative in cells of gastrointestinal, genitourinary, breast, and skin origin. Positivity of TTF-1 in SCC is presumptive evidence of pulmonary origin, given the rarity of SCC of the thyroid.

In this group of patients, who were selected because of an FCM-determined CD56⁺CD45⁻ phenotype and cytomorphologic features of SCC, we found very high concordance with results obtained using a panel of conventional IHC markers for neuroendocrine neoplasms. These results suggest that FCM complements and may possibly be a satisfactory alternative to IHC when SCC is suspected.

Flow cytometry has a number of advantages over IHC. Intensity of antigen expression is quantifiable and can be expressed in objective numeric terms. In many cases, FCM can be performed on material obtained by fine-needle aspiration. This feature, together with the rapidity with which results can be available (often within hours), makes FCM a potentially valuable tool for situations in which more invasive procedures can be hazardous and a speedy diagnosis is desirable, for example, in patients symptomatic from superior vena cava syndrome. When aspirate samples are too scanty or are otherwise inadequate for FCM, IHC would continue to serve as the standard confirmatory modality for diagnosing SCC.

Positivity for the cytokeratin antibody, as demonstrated in all 12 cases in which it was tested by FCM, suggests that it may be a useful addition to a standard lymphoma antibody panel when a nonhematopoietic neoplasm such as SCC is in the differential diagnosis. Of possible interest, in the 11 cases not included in this series because of lack of additional tissue for IHC confirmation, 7 of 7 cases evaluated by FCM showed cytokeratin positivity.

A common problem for surgical pathologists is differentiation of SCC or other small blue cell tumors from malignant lymphoma because of overlapping cytomorphologic features. In our series, standard FCM graphic display of CD45 plotted against CD56 typically demonstrated the population of interest to be clearly separate from lymphoid and other hematopoietic cells (Figure, A), thereby providing a distinction from most lymphomas, as well as suggesting a neoplasm of neuroendocrine derivation. The only medical literature we could find that alluded to FCM phenotyping in the diagnosis of SCC was a recent abstract by investigators at the University of Washington.7 Eleven of 16 cases of small blue round cell tumors exhibiting strong expression of CD56 and negativity for CD45 by FCM were diagnosed as SCC, including 1 case with very low-level bone marrow involvement.

We emphasize that our series was selected based on the CD56⁺CD45⁻ phenotype, together with morphologic features consistent with SCC. We did not attempt to assess the sensitivity or specificity of this phenotype. CD56 is not unique to neural/neuroendocrine tumors and can be expressed, to some extent, by a wide variety of malignancies, including cancer of the gallbladder,9 malignant mesothelioma,10 and non–small cell cancer of the lung.11 In a study of 889 cases of non–small cell lung cancer, 86 (15%) of 575 cases of squamous cell carcinoma and 30 (11%) of 262 cases of adenocarcinoma demonstrated IHC positivity for neural cell adhesion molecule,11 findings which underscore the importance of careful correlation of cytomorphology with immunophenotypic findings. Plasma cell neoplasms are often CD56⁺ and CD45⁻, but they usually express CD38, unlike SCC, and their morphologic differentiation from SCC is generally straightforward. In rare cases of pleural effusion, a cell population with the immunophenotypic profile of SCC has been noted, but only benign mesothelial cells were seen on the cytospin preparation (J.W., unpublished data, April 2002).

The present study is retrospective and reflects an element of case bias. Material was submitted for FCM because lymphoma was suspected in many instances. Twenty of 27 specimens were lymph nodes, either mediastinal or extrathoracic. It is unclear whether the same results would have been obtained from an unselected series of patients with SCC undergoing fine-needle aspiration of lesions from a wider variety of sites. Despite these limitations, our results appear to indicate that the combination of the CD56⁺CD45⁻ phenotype, as determined by FCM, and typical small cell neuroendocrine-type morphology, in the proper clinical setting, is highly suggestive of SCC. Immunophenotyping by routine 3- or 4-color FCM is relatively simple, rapid, and efficient and provides information that can be useful clinically. Confirmatory studies on larger numbers of patients will be necessary to define the possible role of FCM in the approach to the diagnosis of SCC.