1192 head and neck malignancies (oral and nasal cavities, paranasal sinuses, oropharynx, nasopharynx, hypopharynx, larynx, trachea, ear and salivary glands) were retrieved from the archives of the Pathology and Oral Pathology departments of the University College Hospital, Ibadan, Nigeria between 1990 and 2008. 142 poorly-differentiated and undifferentiated neoplasms including anaplastic (undifferentiated) or poorly-differentiated carcinomas, anaplastic large cell lymphomas, pleomorphic sarcomas, malignant fibrous histiocytoma, esthesioneuroblastoma and spindle cell sarcomas were selected. Cases where the original paraffin block could not be obtained were excluded from analysis. Only 86 of the 142 undifferentiated and poorly-differentiated head and neck malignancies diagnosed during the study period satisfied the inclusion criteria.
Freshly prepared sections from each case were stained with hematoxylin-eosin (H&E) and a panel of antibodies to leukocyte common antigen (CD45), cytokeratin AE1/AE3, vimentin, desmin, myogenin and neuron-specific enolase (NSE) using the specifications of the manufacturer (Dako Cytomation, USA).
The sections for immunohistochemistry were de-paraffinized, hydrated and then rinsed in Phosphate Buffered Solution (PBS). They were immersed in heat induced epitope retrieval citrate buffer diluted to 1:10 with distilled water and incubated at 90°C for 1 hour. They were then placed in fresh citrate, cooled in water for 20 minutes and then rinsed in PBS. Positive controls (skin for cytokeratin AE1 or AE3, tonsils for CD45; neural tissue for Neuron-specific enolase; skeletal muscle for Myogenin and Vimentin; and smooth muscle for desmin) and negative controls were employed for each antibody. 3% hydrogen peroxide was added to each section for 10 minutes and the sections were rinsed in 0.1% PBS. The specimens were incubated for an hour with 40-130 μl of appropriately diluted Dako mouse primary antibody, followed by incubation with undiluted labeled polymer Horse Radish Peroxidase conjugated antimouse secondary antibody for 30 minutes. One ml of Diaminobenzidene solution was added to cover the specimen, followed by incubation in a humidity chamber for 15 minutes. The sections were then immersed in aqueous hematoxylin and rinsed in distilled water. The tissue was then dehydrated and subsequently rinsed with xylene. DPX (Distyrene, Plasticizer and Xylene) mounting fluid was then applied and a cover slip placed.
All the seven antibodies used in the panel for one specimen were reviewed sequentially and the pattern and intensity of staining was observed and scored as: negative (0), weakly positive (+1), moderately positive (+2) and strongly positive (+3) [7]. The slides were reviewed without reference to initial histology diagnosis to eliminate bias. The final immunohistochemical findings were then correlated with the H&E stained slides in order to arrive at a final diagnosis.
The data was analyzed using version 16 of the Statistical Package for Social Sciences (SPSS16). Qualitative data were compared using chi-square statistics. Quantitative data were summarized using mean, standard deviation and confidence interval and compared using student t- and/or one-way analysis of variance test. The level of significance was set at p < 0.05. Sensitivity and specificity were calculated using immunohistochemistry as the gold standard to which the original H&E diagnosis was compared. The positive predictive value, negative predictive value, accuracy and degree of agreement were also determined. For degree of agreement, Kappa value >0.75 = excellent agreement, 0.4-0.75 = fair to good agreement and <0.4 = moderate to poor agreement [8].