SUMMARY A series of 41 fresh and 36 routinely processed malignant melanomas were immunostained with a panel of 12 monoclonal antibodies reactive against a range of epithelial, lymphoid, and melanoma associated antigens. The aim of the study was to determine whether this panel of antibodies would be useful in diagnostically difficult cases for differentiating melanomas from other tumours, particularly carcinomas and lymphomas. The results confirmed that most unequivocal malignant melanomas can be identified by positivity for S100 protein and for the antigen recognised by antibody NK1/C3, and by negativity for epithelial and lymphoid antigens. The incidence of melanomas expressing cytokeratin antigens was higher, however, particularly in cryostat sections than has previously been reported. It is therefore suggested that a panel of antibodies with more than one marker in each category should be used for identifying melanomas in clinical practice.When the origin of a malignant tumour is unknown the two most likely alternative diagnoses are lymphoma and carcinoma. Recently, studies performed in several laboratories have established the patterns of antigenic expression characteristic of these two cate-

Summary Despite the presence of a lymphocytic infiltrate in solid cancers, the failure for tumour growth to be contained suggests an inadequate immune response to the tumour. Poor cytotoxicity exerted by tumour-infiltrating lymphocytes (TILs) against tumour cells in vitro, combined with continued tumour growth in vivo, suggests deficiencies in TIL function or numbers. Various theories have been postulated to explain how tumour cells may escape immunosurveillance and control. One of the many hypotheses is the failure of production of cytokines, which are necessary for T cells to mediate their function. Thus, the expression of cytokine mRNA in human breast tumour sections was investigated by reverse transcriptase polymerase chain reaction (RT-PCR) with cytokine-specific primers. A relatively consistent finding was detection of interleukin (IL) 10 mRNA among the tumours. No IL-2 and little IL-4 mRNA was detected in the tumours. IL-6 and IL-10 mRNA was detected in only one and two of the normal breast tissues respectively. IL-2, IL-4 and tumour necrosis factor (TNF)-a mRNA was not detected in any of the normal breast tissues. The reduced function of TILs may be related to IL-10, which has known inhibitory effects on T-cell activation.

Seven large bowel carcinomas were examined by light and electron microscopy for the presence of five oncofoetal antigens. Ultrastructural investigations involved a novel method whereby thick sections of gluteraldehyde-fixed material were cut on a vibratome and then labelled using slight modifications of a standard unlabelled antibody-enzyme (PAP) technique, before further processing. Ultrastructural preservation, staining properties and the retention of antigen activity was seemingly better than that achieved by other investigators. Specific, positive labelling for carcinoembryonic antigen (CEA), colon specific antigen (CSA) and pregnancy-specific beta-1-glycoprotein (SP1) was seen in every case. Clear positive labelling for placental alkaline phosphatase (PLAP) and human chorionic gonadotropin (HCG) was seen in two cases. Extracellular labelling was found in areas of cell debris, free lying or in phagocytic cells and on tumour cell brush borders. The pattern of intracellular labelling, however, was different for each antigen and reflected the probable sites of synthesis and release from the cells. Thus CEA, a complex glycoprotein, was localised within the golgi apparatus, small apical cytoplasmic vesicles and mucous droplets in relatively well differentiated tumour cells. CSA, a chemically related glycoprotein, had a similar, but less dense distribution. SP1, by contrast, was localised within basally-located vesicles associated with the ribosomal endoplasmic reticulum and appeared to be released and persist as debris or taken up by phagocytic cells below the basal lamina. PLAP and HCG, both proteins, were found within simple single membrane-bound vesicles within relatively undifferentiated cells.

SUMMARY Using a computed video image analysis system, the staining intensity for both neurone specific enolase (NSE) and S100 protein was measured in sections from 19 malignant melanomas and 16 benign melanocytic lesions. The results of this study confirm previous reports that NSE and S100 protein are useful markers for malignant melanoma. NSE staining intensity in the cases of malignant melanoma was significantly higher than that in benign naevi (p = 0 01 1). Intensity of staining for S100 protein was not significantly higher in the malignant melanomas. There was, however, a significant S100 gradient when comparing superficial and deep intradermal portions of these tumours (p = 0-003). This feature was not seen in benign naevi. The greatest intensity of S100 protein staining was found in the deeper portions of the malignant melanomas. This gradient difference was not seen with staining for NSE. Although it seems that the overall intensity of staining for NSE is more effective in differentiating between benign and malignant lesions, the difference in staining intensity between the superficial and deep portions of the tumour may be the better indicator of adverse behaviour in lesions in which the diagnosis of malignancy is uncertain. Behaviour in individual cases of malignant melanoma is exceedingly difficult to predict, although the depth of invasion seems to be one of the most reliable prognostic indicators at present.' 4 In a previous study we reported that staining intensity and pattern for neurone specific enolase (NSE) were of diagnostic value in differentiating benign from malignant melanocytic lesions and that differences in staining intensity were of potential prognostic value.5 There are problems with this approach. Not least is the fact that the studies, which have so far been undertaken, entail subjective assessment of the level of staining intensity and may therefore not be reproducible between laboratories or even between observers in the same laboratory.This can be overcome by having an objective measure of staining intensity rather than the usual subjective plus scale scoring. In this study the optical density of staining in tumour tissue was measured using a microcomputer video image analysis system.6 Nineteen cases of malignant melanoma and 16 benign melanocytic lesions were stained for both NSE and S100 protein and the average staining intensity between the superficial, deep, and total tumour regions was examined.Accepted for publication 30 April 1986 Material and methods Archived material from the files of the Flinders Medical Centre was used for this study. The tumours selected came from a batch reviewed by a panel of four consultant pathologists. Only tumours in which there was unanimous independent agreement as to the histological diagnosis were measured.Sections of 19 malignant melanomas and 16 benign compound naevi that had been formalin fixed and paraffin embedded were cut at 5 im and stained to show the presence of NSE and S1OO protein. STAININGThe details of the method have been d...

No abstract