This study is based on our retrospective analysis of 19 patients with solid tumors whose diagnoses were made from bone marrow seen at the Department of Hematology, Uludag University Medical Faculty, over a period of 9 years. Patients with non-Hodgkin's lymphomas and Hodgkin's diseases were not included in this study and patients with known neoplastic disease were also excluded.
The standard technique was employed in obtaining the samples from posterior iliac crest using a Jamshidi needle (Regular/Adult, 11-gauge). All of the trephine biopsies were performed unilaterally, because clinically none of the diseases that focally involve the bone marrow were included in the differential diagnosis prior to the biopsy procedure. Length of the biopsy cores ranged between 1.2 cm and 2.2 cm (mean 1.7 cm). Trephine biopsies were fixed in 10% neutral buffered formaline for at least 24 hours, and then decalcified overnight in a decalcifying solution which is a mixture of 8% HCl and 10% formic acid at equal amounts of volume. Following the automated tissue processing, biopsies were embedded in parafin blocks, and 0.3 micrometer sections were cut. In some cases the sternum was sampled using a Rosenthal needle for aspiration. Touch preparations were done if the aspiration resulted in a "dry tap" or if aspiration material was considered to be technically inadequate for evaluation, or if it was hemodilute. Bone marrow aspiration, touch preparations and peripheral blood smears which were obtained at the same time by biopsy were stained by May Grunwald-Giemsa. Hematoxylene-eosin, Giemsa and reticulin stains were routinely performed in biopsy sections. If the nonhematologic properties of the tumor could be identified with routine stains, and if the primary of the metastatic tumor could easily be diagnosed morphologically with routine hematoxylene-eosin stains, as in the cases of signet ring cell carcinoma; no immunohistochemical study was held. The pathologist's approach to definitive diagnosis of the patient with metastasis of unknown primary effectively followed a few sequential steps: First of all we tried to determine the cell line of differentiation e.g. carcinoma, lymphoma, melanoma, sarcoma, or germ cell, with the help of morphological findings and if needed, immunohistochemical stains. Our panel of antibodies contained pancytokeratin, HMB45, Leukocyte Common Antigen (LCA), Vimentin and Placental Alkalen Phosphatase (PLAP). If it was an epithelial tumor we tried to determine the cytokeratin (CK) type or types of distribution in the tumor cells, since some subsets of cytokeratins are uniqe to certain tumor types. Our panel contained AE1/AE3, CAM5.2, CK7, CK20, and 34 beta E12. In our further studies we tried to determine if there was expression of supplemental antigens of epithelial or germ cell derivation, that was Carcinoembryonic Antigen (CEA), Epithelial membrane Antigen (EMA) or PLAP. Last step was to determine if there was expression of cell-specific structures or receptors that are unique identifiers of cell types, for example neuroendocrine granules, peptide hormones, thyroglobulin, Prostate Specific Antigen (PSA), Gross Cystic Disease Fluid Protein (GCDFP) or Thyroid Transcription Factor-1 (TTF-1). Our panel of antibodies contained Synaptophysin, Chromogranin, Neuron Specific Enolase (NSE), Thyroglobuline, PSA, GCDFP, and TTF-1. Cases with sarcomatous properties were immunostained with Desmin, Smooth Muscle Actin, S100, Vimentin and Myoglobulin. After the confirmation of original peripheral blood and bone marrow cytological findings and histopathological diagnoses by a senior hematologist and an expert pathologist the patient charts were reviewed. Cases with a history of malignancy at the time of presentation were excluded.
Patient characteristics were recorded in each case, including: presenting symptoms, onset of symptoms, physical examination findings, peripheral blood counts, peripheral blood morphology, diagnostic evaluation, management, and survival.