D Y Mason

Hospital Universiti Kebangsaan Malaysia (HUKM), Kuala Lumpur, Kuala Lumpur, Malaysia

Are you D Y Mason?

Claim your profile

Publications (367)2031.46 Total impact

  • Source
  • [Show abstract] [Hide abstract]
    ABSTRACT: To study the expression of CD2-associated protein (CD2AP), an adaptor protein involved in T-cell signalling and renal function, in normal, reactive and neoplastic human lymphoid tissues. We used immunohistochemical techniques to evaluate monoclonal antibodies against CD2AP on over 400 formalin fixed paraffin embedded tissue blocks retrieved from the host institutions of three authors. The samples tested included normal, reactive and neoplastic lymphoid tissue. In lymphoid tissues, strong CD2AP staining was observed in plasmacytoid dendritic cells (pDCs), weak and variable in mantle zone B cells and moderate in rare germinal center cells. CD2AP labeled cortical and rare medullary thymocytes and isolated mononuclear cells in bone marrow trephines. Furthermore, epithelial and endothelial cells expressed CD2AP. Among neoplasms, the greatest number of CD2AP-positive cases were found in diffuse large B cell (21/94), NK T-cell lymphomas (7/67), "blastic plasmacytoid dendritic cell neoplasms" (9/10) and some types of solid tumor. Our finding that mature peripheral T cells are CD2AP-negative but immature cortical thymocytes are positive may prove useful for diagnostic purposes. Moreover, our results demonstrate that CD2AP represents a useful marker of normal and neoplastic pDC and may be used in a diagnostic panel in reactive or neoplastic lymphoid proliferations.
    Pathologica 04/2012; 104(2):56-64.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The hallmark of follicular lymphoma is the t(14;18)(q32;q21) chromosomal translocations that lead to deregulation of BCL2 expression in tumour cells. However, not all cases of follicular lymphoma express BCL2, nor is the t(14;18) translocation always present. Follicular lymphomas lacking the BCL2 rearrangement are less well studied with regards to their immunohistochemical and molecular features. This study aims to investigate the BCL2 protein expression pattern in t(14;18) negative follicular lymphomas. BCL2 protein expression pattern was analysed in 26 cases of t(14;18) negative follicular lymphoma [determined by fluorescence in situ hybridisation (FISH)], using antibodies against two-different epitopes, i.e., the widely-used antibody BCL2/124 and an alternative antibody E17. Two of the t(14;18) negative cases showed evidence of BCL2 amplification and trisomy 18. A total of 13 cases (50%) lacked BCL2 expression. In 10 cases (38%) the expression was heterogeneous and in only three cases (12%) the BCL2 expression was strongly positive. These cases could thus be subdivided into three subgroups: Group I, normal BCL2 genes (i.e., no evidence of translocation or amplification), and BCL2 protein negative; Group II, normal BCL2 genes but BCL2 protein positive; and Group III, presence of other genetic alterations, i.e., BCL2 amplification and trisomy 18, and BCL2 protein positive. This study suggests that it may be possible on the basis of staining to predict that the t(14;18) translocation is absent if a case is either negative for BCL2 protein with different antibodies or has heterogeneous BCL2 expression, possibly acquired through a physiological process of differentiation.
    Pathology 03/2012; 44(3):228-33. · 2.66 Impact Factor
  • Histopathology 03/2011; 58(4):633-6. · 2.86 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The t(14;18)(q32;q21) chromosomal translocation induces BCL2 protein expression in most follicular lymphomas. However, a small number of cases lack BCL2 expression despite carrying the t(14;18)(q32;q21) translocation. This study aims to explore the mechanism accounting for the lack of BCL2 protein expression when the t(14;18) translocation is present. BCL2 expression in the t(14;18) positive cell lines FL18, Karpas-422, SU-DHL-4 and SU-DHL-6, was analysed by Western blotting and by immunohistochemistry using two different antibodies. FISH analysis was performed to confirm the cytogenetic changes in the cell lines and real time quantitative PCR was used to evaluate the BCL2 mRNA level. Sequence analysis of translocated BCL2 was performed on FL18, Karpas-422, SU-DHL-4 and SU-DHL-6 cell lines. In FL18, Karpas-422, and SU-DHL-4, the BCL2 mRNA level correlated with the BCL2 protein expression. In contrast, BCL2 protein was not detected in SU-DHL-6 line using standard anti-BCL2 antibody (BCL2/124), despite the presence of the t(14;18) translocation and high level of mRNA. cDNA sequencing of translocated BCL2 showed three mutations in the SU-DHL-6 cell line, one of which resulted in an amino acid substitution (I48F) in the region recognised by the standard BCL2 antibody, whereas the other two were silent mutations at aa71 and aa72. Interestingly, when BCL2 expression was tested with an alternative antibody, E17, the protein was detected in SU-DHL-6, suggesting that the 'negativity' of SU-DHL-6 line for BCL2 using the standard antibody is spurious. Amino acid changes were found in Karpas-422 (G47D, P59L) and SU-DHL-4 (P59T, S117R) but these did not affect BCL2 detection. This study suggests that some somatic mutations of the translocated BCL2 gene may prevent epitope recognition by BCL2 antibodies, and hence cause false negative expression using the standard antibody. It is recommended that in practice all BCL2 negative cases should routinely be stained with an alternative antibody to prevent false negativity.
    Pathology 04/2010; 42(3):212-6. · 2.66 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To investigate the relationship between Bcl-2 protein expression and cell proliferation at single-cell level in B-cell lymphomas using double-labelling techniques. The relationship between Bcl-2 protein expression and cell proliferation was explored in 124 cases of B-cell lymphoma using double immunofluorescence labelling for Bcl-2 and Ki67. In follicular lymphoma, marginal zone lymphoma and a subset of chronic lymphocytic leukaemia/small lymphocytic lymphoma (CLL/SLL), neoplastic cells tended to lose Bcl-2 when they are in cell cycle. This pattern is usually maintained in both follicular lymphoma and CLL/SLL when they undergo high-grade transformation. In mantle cell lymphoma, diffuse large B-cell lymphoma and a subset of CLL/SLL, the inverse relationship (between Bcl-2 and Ki67) was not observed, i.e. the proliferating cells tended to show co-expression of Bcl-2. In low-grade lymphomas, including those that are transformed, Bcl-2 expression is lost when cell proliferate. However, in more aggressive tumours (i.e. mantle cell and de novo diffuse large B-cell lymphomas) the inverse Bcl-2/Ki67 relationship was not observed. It would be of interest to explore the clinical implications in lymphoma of the presence and absence of the inverse Bcl-2/Ki67 pattern.
    Histopathology 04/2010; 56(5):617-26. · 2.86 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: T follicular helper (T(FH)) cells reside in the light zone of germinal centers and are considered the cell of origin of angioimmunoblastic T-cell lymphoma. Recently, CXCL13, PD-1 and SAP were described as useful markers for T(FH) cells and angioimmunoblastic T-cell lymphoma but also reported in some peripheral T-cell lymphomas, not otherwise specified. In the present study the expression pattern of ICOS protein was investigated by immunohistochemistry-based techniques in routine sections of normal lymphoid tissues and 633 human lymphomas. Cells strongly positive for ICOS were restricted to the light zone of germinal centers and co-expressed T(FH)-associated molecules. In addition, weak to moderate ICOS expression was observed in a small proportion of FOXP3-positive cells. In lymphomas, ICOS expression was confined to angioimmunoblastic T-cell lymphoma (85/86), peripheral T-cell lymphomas of follicular variant (18/18) and a proportion of peripheral T-cell lymphomas, not otherwise specified (24/56) that also expressed other T(FH)-associated molecules. ICOS is a useful molecule for identifying T(FH) cells and its restricted expression to angioimmunoblastic T-cell lymphoma and a proportion of peripheral T-cell lymphomas, not otherwise specified (showing a T(FH)-like profile) suggests its inclusion in the antibody panel for diagnosing T(FH)-derived lymphomas. Our findings provide further evidence that the histological spectrum of T(FH)-derived lymphomas is broader than previously assumed.
    Haematologica 03/2010; 95(3):432-9. · 5.94 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Mutations in the C-terminal region of nucleophosmin in acute myeloid leukemia (AML) result in aberrant cytoplasmic nucleophosmin (cNPM) in leukemic blast cells which is detectable by immunocytochemistry in bone marrow trephine (BMT) biopsy sections. We tested whether cNPM is detectable by immunocytochemistry in air-dried smears of AML with nucleophosmin1 (NPM1) mutations. An immunoalkaline phosphatase method was developed using the OCI-AML3 cell line, known to have mutated NPM1, and assessed on blood and marrow smears of 60 AML cases. NPM was detectable in all blast cell nucleoli and cNPM in 21 of 31 of NPM1 mutated and 15 of 29 wild-type cases. Paired air-dried smears and BMT biopsies from the same case (mutated and wild-type) gave discrepancies in cNPM expression and there was no correlation in 10 of 22 cases. Due to the high false positive and negative rates for cNPM in cell smears, this method should not be used as a surrogate for NPM1 mutations in AML.
    Haematologica 12/2009; 95(4):670-3. · 5.94 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Two distinct groups of chronic lymphocytic leukaemia (CLL) are distinguished by the presence or absence of somatic hypermutation of the immunoglobulin heavy-chain gene. CLL without somatic hypermutation has an adverse outcome, but the precise biological differences that underlie this more aggressive clinical-course are unclear. Using a proteomic approach, we found that the two prognostic forms of CLL were consistently distinguished according to their protein expression pattern. The most important difference observed related to the different expression of nucleophosmin 1 between the two forms of CLL. This different expression was not related to apoptosis, proliferation or gene mutation. However, co-immunoprecipitation experiments identified an association between nucleophosmin 1 and ribosomal proteins. Using immunocytofluorescence, nucleophosmin 1 expression was identified in the nucleoli and nucleoplasm of all cells, but in a proportion of cells, nucleophosmin had been transferred from the nucleoplasm to the cytoplasm. Both the fluorescent intensity, and the frequency of cytoplasmic nucleophosmin 1 expression, was higher in CLL without somatic hypermutation. We propose therefore, that nucleophosmin 1, in association with ribosomal proteins, undergoes nucleo-cytoplasmic shuttling in CLL. This process is most prominent in un-mutated CLL and may signify altered protein biosynthesis.
    British Journal of Haematology 11/2009; 148(4):534-43. · 4.94 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Many diagnostic antibodies are generated by immunization with whole cells or cell extracts and are shown by screening on tissue sections to label specific cell populations. However, their target molecule then needs to be identified, and this can be technically demanding. Here we describe the use of protein arrays to define the targets of new or uncharacterized monoclonal antibodies. The technique involves screening protein arrays containing thousands of recombinant human proteins. An initial experiment showed that a well-characterized monoclonal antibody against nucleophosmin identified 22 clones on the array encoding this protein. Next, the antibody JJ166, for which the antigen had not yet been identified, was screened. This antibody was generated by immunizing with a nuclear extract of Jurkat cells and was detected in immunohistochemistry due to its distinctive nuclear staining of lymphoid tissue cells. However, its molecular target had remained unidentified using traditional approaches. A protein array screen rapidly identified the mitotic spindle-associated molecule NUMA1 (nuclear mitotic apparatus protein 1). To confirm this putative specificity, JJ166 was shown to react with COS-1 cells transiently transfected with the complementary DNA for NUMA1. Furthermore, a commercially available antibody against NUMA1 showed nearly identical staining in immunohistochemistry on human tissue and cells. Overall, this method represents an effective and quick strategy for defining the protein targets of new or uncharacterized monoclonal antibodies identified as having diagnostic or other potential value on the basis of their immunostaining patterns.
    Analytical Biochemistry 10/2009; 395(2):119-24. · 2.58 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: c-Maf, a leucine zipper-containing transcription factor, is involved in the t(14;16)(q32;q23) translocation found in 5% of myelomas. A causal role for c-Maf in myeloma pathogenesis has been proposed, but data on c-Maf protein expression are lacking. We therefore studied the expression of c-Maf protein by immunohistochemical analysis in myelomas and in a wide variety of hematopoietic tissue. c-Maf protein was detected in a small minority (4.3%) of myelomas, including a t(14;16)(q32;q22-23)/IgH-Maf+ case, suggesting that c-Maf protein is not expressed in the absence of c-Maf rearrangement. In contrast, c-Maf was strongly expressed in hairy cell leukemia (4/4) and in a significant proportion of T-cell (24/42 [57%]) and NK/T-cell (49/97 [51%]) lymphomas, which is in keeping with prior gene expression profiling and transgenic mouse studies. Up-regulation of c-Maf protein occurs in a small subset of myelomas, in hairy cell leukemia, and in T- and NK-cell neoplasms. Its detection may be of particular value in the differential diagnosis of small cell lymphomas.
    American Journal of Clinical Pathology 10/2009; 132(3):361-71. · 2.88 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Focal adhesion kinase (FAK) is a protein tyrosine kinase essential for intracellular regulatory events, such as cell growth, differentiation, migration and tumor metastasis. The aim of this study was to analyze the expression of FAK protein in a series of normal and neoplastic lymphoid tissues. An anti-FAK antibody was used to study the protein expression in paraffin-embedded samples of normal and neoplastic, hematolymphoid and non-hematolymphoid tissues by immunohistochemistry. In normal hematolymphoid tissue, the strongest expression of FAK was detected in germinal center and marginal-zone B cells; positive staining was also found in mantle zone B cells. In human lymphomas, FAK was expressed mostly in B-cell lymphomas and was predominantly negative in T-cell lymphoma. In Hodgkin lymphomas, FAK was found only in the neoplastic cells of lymphocyte predominant type, whereas the tumor cells of the classical form were FAK-negative. We demonstrate for the first time the expression of FAK in paraffin-embedded hematolymphoid tissue samples. Its differential expression in lymphomas may be of relevance for some B-cell neoplasms by using it as an additional marker to distinguish B- from T-lymphoblastic leukemia/lymphoma to further differentiate lymphocyte predominant from classical Hodgkin lymphoma.
    Pathology - Research and Practice 08/2009; 205(11):781-8. · 1.21 Impact Factor
  • D. C. Brown, K. C. Gatter, D. Y. Mason
    [Show abstract] [Hide abstract]
    ABSTRACT: Sixty-four cases of mainly lymphoid lesions were investigated by fine needle aspiration. Both conventional smears and cytospin preparations were made from this material. Diagnoses based on cytological and immunocytological criteria were made in 57 of these cases and include reactive hyperplasia, non-Hodgkin's and Hodgkin's lymphoma and metastatic carcinoma. Five cases were inadequate for diagnostic purposes and in 2 cases no definite diagnosis could be made. These diagnoses were confirmed using tissue sections in 48/51 cases (with only 3 cytological diagnoses significantly altered by histology) and supported by supplementary clinical information in 8 cases. These results demonstrate the value of FNA biopsy in the investigation of lymphoid lesions using both conventional cytology and immunocytochemistry.
    06/2009; 2(5):323-333.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The t(14;18)(q32;q21) chromosomal translocation induces BCL2 protein overexpression in most follicular lymphomas. However the expression of BCL2 is not always homogeneous and may demonstrate a variable degree of heterogeneity. This study analysed BCL2 protein expression pattern in 33 cases of t(14;18)-positive follicular lymphomas using antibodies against two different epitopes (i.e. the widely used antibody BCL2/124 and an alternative antibody E17). 16/33 (49%) cases demonstrated strong BCL2 expression. In 10/33 (30%) cases, BCL2 expression was heterogeneous and in some of these, its loss appeared to be correlated with cell proliferation, as indicated by Ki67 expression. Double immunofluorescence labelling confirmed an inverse BCL2/Ki67 relationship, where in 24/28 (86%) cases cellular expression of BCL2 and Ki67 was mutually exclusive. In addition, seven BCL2 'pseudo-negative' cases were identified in which immunostaining was negative with antibody BCL2/124, but positive with antibody E17. Genomic DNA sequencing of these 'pseudo-negative' cases demonstrated eleven mutations in four cases and nine of these were missense mutations. It can be concluded that in follicular lymphomas, despite carrying the t(14;18) translocations, BCL2 protein expression may be heterogeneous and loss of BCL2 could be related to cell proliferation. Secondly, mutations in translocated BCL2 genes appear to be common and may cause BCL2 pseudo-negative immunostaining.
    British Journal of Haematology 01/2009; 144(5):716-25. · 4.94 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Lymphomas originating from the lymphatic system comprise about 30 entities classified according to the World Health Organization (WHO). The histopathological diagnosis is generally considered difficult and prone to mistakes. Since non-random chromosomal translocations are specifically involved in different lymphoma entities, their detection will be increasingly important. Hence, a split-signal fluorescence in situ hybridisation (FISH) procedure would be helpful in discriminating the most difficult classifications. The Euro-FISH programme, a concerted action of nine European laboratories, has validated a robust, standardised protocol to improve the diagnostic approach on lymphoma entities. Therefore, 16 fluorescent probes and 10 WHO entities, supplemented with reactive cases, were selected. The results of the Euro-FISH programme show that all probes were correctly cytogenetically located, that the standardised protocol is robust, resulting in reliable results in approximately 90% of cases, and that the procedure could be implemented in every laboratory, bringing the relatively easy interpretation of split-signal probes within the reach of many pathology laboratories.
    Journal of Hematopathology 10/2008; 1(2):119-26.
  • [Show abstract] [Hide abstract]
    ABSTRACT: Double immunoenzymatic labeling of 2 different molecules in tissue sections is a widely used technique. However, it is time consuming since the 2 immunoenzymatic procedures are carried out in sequence, and they must also be optimally performed to avoid unwanted background labeling. In this paper, we report that double immunoenzymatic staining performed using automated immunostaining apparatus considerably reduces the requirements in terms of time and is also highly reproducible and free of background. Three tissue markers can also be visualized by performing (after immunoperoxidase labeling) 2 sequential immuno-alkaline phosphatase procedures using different substrates. Furthermore, single or double detection of mRNA by in situ hybridization can be combined with immunoenzymatic labeling. Finally, automated labeling could also be performed on peripheral blood and bone marrow smears, opening the possibility of using this procedure in the analysis of hematologic/cytology samples.
    Applied immunohistochemistry & molecular morphology: AIMM / official publication of the Society for Applied Immunohistochemistry 08/2008; 16(4):371-81. · 1.63 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Plasmacytoid dendritic cells (pDCs) are involved in innate immunity (eg, by secreting interferons) and also give rise to CD4+CD56+ hematodermic neoplasms. We report extensive characterization of human pDCs in routine tissue samples, documenting the expression of 19 immunohistologic markers, including signaling molecules (eg, BLNK), transcription factors (eg, ICSBP/IRF8 and PU.1), and Toll-like receptors (TLR7, TLR9). Many of these molecules are expressed in other cell types (principally B cells), but the adaptor protein CD2AP was essentially restricted to pDCs, and is therefore a novel immunohistologic marker for use in tissue biopsies. We found little evidence for activation-associated morphologic or phenotypic changes in conditions where pDCs are greatly increased (eg, Kikuchi disease). Most of the molecules were retained in the majority of pDC neoplasms, and 3 (BCL11A, CD2AP, and ICSBP/IRF8) were also commonly negative in leukemia cutis (acute myeloid leukemia in the skin), a tumor that may mimic pDC neoplasia. In summary, we have documented a range of molecules (notably those associated with B cells) expressed by pDCs in tissues and peripheral blood (where pDCs were detectable in cytospins at a frequency of <1% of mononuclear cells) and also defined potential new markers (in particular CD2AP) for the diagnosis of pDC tumors.
    Blood 05/2008; 111(7):3778-92. · 9.78 Impact Factor
  • D Y Mason
    Annales de Pathologie 12/2007; 27 Spec No 1:1S33-6. · 0.24 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To clarify the origin of Hodgkin (H) and Sternberg-Reed (SR) cells, frozen sections of lymph nodes from 30 patients with Hodgkin's disease were immunostained with a large panel of monoclonal antibodies reactive with cells of lymphoid tissue and granulopoiesis. The results showed that: (a) H and SR cells are devoid of markers specific to, or characteristic of B cells, macrophages, dendritic reticulum cells, interdigitating cells, or cells of erythropoietic or thrombopoietic origin; (b) the vast majority of H and SR cells contain granulocyte-related antigens detectable with the monoclonal antibodies TÜ9 and 3C4, but constantly lack other granulocytic cell markers (such as peroxidase and chloroacetate esterase). Monoclonal antibodies raised against a Hodgkin's disease-derived cell line included one, Ki-1, that was found to be selectively reactive with H and SR cells and a minute, but distinct cell population in normal lymphoid tissue and bone marrow. The latter, as yet unidentified cell population appears to be the normal equivalent of H and SR cells.
    Hematological Oncology 11/2007; 1(1):21 - 29. · 2.04 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The detection of genetic abnormalities (eg, translocations, amplifications) in paraffin-embedded samples by the fluorescence in situ hybridization (FISH) technique is usually performed on tissue sections. FISH analysis of nuclei extracted from paraffin-embedded samples is also possible, but the technique is not widely used, principally because of the extra labor involved and the loss of information on tissue architecture. In this article, we report that nuclei extracted from paraffin-embedded tissue often retain at least part of the surrounding cytoplasm. Consequently, immunocytochemical labeling for a range of cellular markers (eg, of lineage or proliferation) can be performed in combination with FISH labeling, allowing specific cell populations to be analyzed for genetic abnormalities. These cell preparations are largely free of the problems associated with tissue sections (eg, truncation artifact, signals in different focal planes) so that interpretation is easy and numerical chromosomal abnormalities are readily assessed. Cells isolated from paraffin sections can be stored in suspension so that arrays can be created as and when needed from a range of neoplasms for investigation by the immunoFISH technique (for example, for studying a new genetic abnormality). This procedure represents a novel methodology, which in some settings offers clear advantages over analysis of tissue sections.
    Journal of Molecular Diagnostics 10/2007; 9(4):479-89. · 3.95 Impact Factor

Publication Stats

19k Citations
2,031.46 Total Impact Points

Institutions

  • 2012
    • Hospital Universiti Kebangsaan Malaysia (HUKM)
      • Department of Pathology
      Kuala Lumpur, Kuala Lumpur, Malaysia
  • 1982–2011
    • Oxford University Hospitals NHS Trust
      • • Nuffield Department of Clinical Laboratory Sciences
      • • Department of Haematology
      • • Department of Histopathology
      Oxford, England, United Kingdom
  • 2010
    • National University of Malaysia
      • Department of Pathology
      Kuala Lumpur, Kuala Lumpur, Malaysia
  • 2006–2009
    • Stanford University
      • Department of Pathology
      Stanford, CA, United States
    • University Medical Center Schleswig-Holstein
      Kiel, Schleswig-Holstein, Germany
  • 1981–2009
    • University of Oxford
      • • Nuffield Division of Clinical Laboratory Sciences
      • • MRC Human Immunology Unit
      • • Molecular Haematology Unit
      • • Nuffield Department of Obstetrics and Gynaecology
      Oxford, ENG, United Kingdom
  • 2007
    • Centro Nacional de Investigaciones Oncológicas
      Madrid, Madrid, Spain
  • 2005
    • University of Barcelona
      • Department of Medicine
      Barcino, Catalonia, Spain
  • 2004
    • Humboldt-Universität zu Berlin
      Berlín, Berlin, Germany
  • 2003
    • University of Adelaide
      • Women's & Children's Health Research Institute (WCHRI)
      Adelaide, South Australia, Australia
  • 1997–2003
    • University of Bologna
      • Institute of Haematology
      Bologna, Emilia-Romagna, Italy
    • St. Jude Children's Research Hospital
      Memphis, Tennessee, United States
    • Osaka University
      • Department of Molecular Immunology
      Ibaraki, Osaka-fu, Japan
    • University of KwaZulu-Natal
      • Department of Anatomical Pathology
      Durban, KwaZulu-Natal, South Africa
  • 1999–2002
    • Università degli Studi di Perugia
      • Department of Internal Medicine
      Perugia, Umbria, Italy
  • 2000
    • IEO - Istituto Europeo di Oncologia
      • Department of Experimental Oncology
      Milano, Lombardy, Italy
  • 1994
    • Erasmus Universiteit Rotterdam
      • Department of Immunology
      Rotterdam, South Holland, Netherlands
  • 1993
    • Netherlands Cancer Institute
      • Division of Immunology
      Amsterdam, North Holland, Netherlands
    • Hôpital Henri Mondor (Hôpitaux Universitaires Henri Mondor)
      • Département de Pathologie
      Créteil, Ile-de-France, France
  • 1990–1993
    • University of Limoges
      Limages, Limousin, France
  • 1992
    • Cross Cancer Institute
      Edmonton, Alberta, Canada
  • 1991
    • Freie Universität Berlin
      • Institute of Social and Cultural Anthropology
      Berlin, Land Berlin, Germany
  • 1989–1990
    • University of Copenhagen
      København, Capital Region, Denmark
  • 1987
    • University of Chicago
      • Department of Pathology
      Chicago, IL, United States