Haematologica 2002; 87:(03)ECR09
[Medline] [prev] [index] [next]Extramedullary myeloid cell tumor/granulocytic sarcoma with predilection for serosal surfaces
Maurilio Ponzoni,* Carlo Patriarca,°# Gianluigi Arrigoni,* Umberto Gianelli,* Aurelio Vicari,@ Franco Folli,@
*Department of Pathology and @ Internal Medicine, Scientific Institute Ospedale S.Raffaele HSR; *II Department of Pathology, Ospedale S. Paolo, University of Milan, Milan, Italy. #Present address: Servizio di Anatomia Patologica, A.O. Melegnano, Milan, Italy
Correspondence: Maurilio Ponzoni, M.D., Department of Pathology, IRCCS Ospedale S.Raffaele, via Olgettina 60, 20132 Milano, Italy. Phone: international +39.02.26432544. Fax: international +39.02.26432409. E-mail: ponzoni.maurilio@hsr.it
A granulocytic sarcoma with tropism for serosal surfaces and minimal skin and bone marrow involvement is reported. Serosal disease consisted both of a 'superficial component' and a 'tumor forming' counterpart. Chromosome analysis showed hypotriploid karyotype. Bone marrow involvement pattern and adhesion molecule immunoreactivity (IR) of neoplastic cells are discussed.A 78-year old woman with a previous diagnosis of common variable hypogammaglobulinemia eight years early (two reviewed bone marrow biopsies taken at that time disclosed megakaryocytic hyperplasia and rare reactive lymphoid aggregates), followed by a diagnosis of idiopathic thrombocytopenia three months later, was hospitalized because of asthenia and malaise. Professional or enviromental hazards were absent. Physical evaluation revealed pale skin/mucous membranes, a few trunk red-purple nodules not aching or blanching on pressure, ascites, and bilateral lower limb edema. Abnormal laboratory findings included normocytic normochromic anemia (Hb 9.2 g/dL), thrombocytopenia (77,000x109/L), neutropenia (2200x109/L), high serum LDH (7210 mU/dL), high uric acid (14.2 mg/dL) and hypogammaglobulinemia (350 mg/dl); blasts were absent from a peripheral blood smear. Thoraco-abdominal CT scan showed no abnormalities. The ascitic fluid and one of the biopsied skin nodules showed hematopoietic-looking neoplastic elements, suggestive of myeloid origin. A bone marrow biopsy performed 10 days before death disclosed a trilineage dys-hematopoiesis (Figure 1B) admixed to two edematous loosely arranged aggregates, that were characterized by large cells with irregular nuclei and abundant cytoplasm, immunoreactive1 for myeloperoxidase (Figure 1A), CD68, CD68R and not for CD3, CD15, CD20, CD30, CD34, CD43, CD45, CD45RO, CD79a, anti-hemoglobin, factor VIII and cytokeratin. Abnormal localized immature precursors (ALIPs) or an increase of CD34+ blasts were not observed. A bone marrow aspiration showed trilineage dys-hematopoiesis. Cytogenetic analysis of bone marrow cells revealed a normal karyotype, while in ascitic fluid an hypotriploid karyotype was present in all metaphases analyzed.2 The patient died of acute peritonitis 40 days after admission. At autopsy, gross examination revealed multiple nodules, 0.5 cm in diameter, distributed along pleural, pericardial and peritoneal surfaces, with a less irregular nuclear profile (Figure 1C). The serosal surface involvement was characterized by a biphasic pattern: a superficial, linear component mixed with mesothelium and an adjacent, 'tumor forming' component microinfiltrating the parenchyma lying below, without differences in cellular composition between the two patterns. The results of immunoreactivity for the tested adhesion molecules (Figure 1D) are summarized in Table 1. All remaining organs were not involved, with the exception of rare, small collections of neoplastic cells in the retroperitoneum and the same bone marrow infiltration pattern. No HHV-8 sequence was found in any of the tissues examined, as previously described,3 by PCR. Serosal presentation in extramedullary myeloid tumor (EMT) accounts for less than 10 % of cases, always associated with pulmonary or cardiac involvement.4 The present EMT shows a striking predilection for serosal surfaces and the absence of unequivocal parenchymal disease, with the exception of rare, minute skin and bone marrow aggregates. The absence of a 'mass forming' disorder could be responsible for the delay in diagnosis and questions the suitability of using the term tumor in this EMT, while the virtually exclusive presence of neoplastic cells along serosal surfaces may suggest the hypothesis of their primary involvement of mesothelial milieu. The modality of bone marrow involvement deserves further discussion. The neoplastic cells (Figure 1A) are far from looking myeloid-like in composition; therefore immunohistochemistry appears critical for their correct interpretation. The focal involvement may be responsible for the lack of detection of neoplastic cells by a bone marrow smear. Interestingly, the lack of IR for CD43 contrasts with recent data.5 Although EMT may develop within the context of underlying myeloproliferative/myelodysplastic disorders, the failed IR of tumor cells for CD34 is similar to previously reported granulocytic sarcoma subsets without underlying myelodysplastic syndrome.6 The simultaneous discordance of cytogenetic results in marrow and ascitic fluid cells contrasts to what expected in EMT.7 All the above mentioned results question therefore whether this peculiar neoplasia could share some aspects with the dissemination properties of 'solid tumors' , thus rendering the bone marrow a target rather than the original source of disease. Our EMT reacted for CD49a and CD61 (Table 1), which is not the rule for acute myeloid leukemias (AML),4,8,9 and for CD51 and CD56, as expected for extramedullary formation disposition,8 CD51 and CD61 IR was restricted to the 'superficial' component of the mesothelium, leaving the 'tumor forming' counterpart not stained; the "de novo" CD51 and CD61 IR in the serous spreading component of this case, as opposed to the majority of AML,9 may be partly responsible for this unique diffusion pattern. Likewise, CD49a, CD49c and CD49d selective IR in the 'tumor forming' component hints at a specific function of these integrins in the various morphologic patterns of this case. More extensive studies are needed to confirm potential molecular adhesion IR differences between EMT and AML.
Acknowledgments
The authors thank Giuseppe Torelli, M.D and Mario Luppi, M.D., PhD from University of Modena, Italy for their support in investigating HHV-8 by PCR, Mrs Tara Zoll for the excellent editing work, Elisabetta Ferrero, M.D. for the gift of anti- alpha 4 antibody and Laura Seghezzi, Ph.D. for useful comments.References