Laryngeal biopsies, contrary to biopsies from many other sites of the body, very often contain minute amounts of tumour tissue that may consist of morphologically undifferentiated tumour only. In haematoxylin- and eosin-stained sections, there may be no indicative features of what specific tumour entity that is present. In the larynx, particularly small round cell neoplasms, primary or metastatic, often cause a diagnostic dilemma and where an incorrect diagnosis can induce substantial clinical consequences for the patient (e.g., primary neuroendocrine carcinomas vs metastatic variants, certain sarcomas). If sufficient/representative material has been obtained, the application of immunohistochemistry and/or molecular techniques should in virtually every case reveal the true nature of the malignancy. In cases with sparse amount of material, and therefore a limited number of sections to be cut, a careful and thoughtful stepwise approach is necessary to ascertain a reliable diagnosis, or at least guide the clinician to the most likely diagnoses. With today's advanced and widely available technology with an abundance of markers to discriminate different tumours, the use of the term "undifferentiated" should be largely unnecessary. In the exceptional, and indeed exceedingly rare cases, when a classification is not possible, even after repeat biopsy, we suggest that the laryngeal neoplasm is better termed "unclassified malignant neoplasm" rather than "undifferentiated malignant neoplasm".
Keywords: FISH; Immunohistochemistry; Larynx; RT-PCR; Unclassified malignancies.