Distribution of Transforming Growth Factor-& in Human Astrocytomas HEINZ-A. HORST, MD, BERND W. SCHEITHAUER, MD, PATRICK J. KELLY, MD, AND JOHN S. KOVACH, MD
We used immunohistochemical techniques to study the distribution of transforming growth factor-B, (TGF-B,) and infiltrating lymphocytes and macrophages in human astrocytomas. Thirteen of 15 grade 4 astrocytomas (glioblastomas) showed staining with anti-TGF-8, antibody, predominantly in proliferating endothelial complexes and surrounding small and medium-sized blood vessels. Brain tissue microscopically free of tumor cells (n = 8) and more differentiated astrocytomas of varying grade (1 to 3; n = 6) devoid of endothelial proliferation did not stain with anti-TGF-8,. Normal brain contained only rare lymphoreticular cells. The majority of astrocytomas studied, however, contained T lymphocytes and macrophages with smaller numbers of B lymphocytes. The lymphoreticular infiltrates were concentrated primarily in close proximity to blood vessels. Within an individual tumor perivascular regions staining for TGF8, never contained more than occasional T lymphocytes. Perivascular regions not staining for TGF-8, frequently contained low to high numbers of T lymphocytes. The inverse relationship in the distribution of TGF-& and lymphocyte infiltrates is compatible with a functional relationship between this cytokine and an immune effector cell response to glioblastomas. HUM PATHOL 23:12841288. Copyright 0 1992 by W.B. Saunders Company Transforming a family functions, ferentiation,
of cell growth and
These diverse activities of TGF-/3 are exerted by the binding of TGF-/3 to specific receptors present on almost every cell type.’ The ubiquitous nature and the high degree of evolutionary conservation of the amino acid sequence of TGF-P2 strongly suggest an important regulatory role for TGF-P in many physiologic and pathophysiologic processes. Recently, Bodmer et al” showed that mRNA for TGF-PI and TGF-& is expressed in two human glioblastoma cell lines and three primary human glioblastomas, whereas a human melanoma and a neuroblastoma cell line expressed only TGF-PI mRNA. The demonstration of mRNA for TGF-P, and TGF-& in huextracellular
From the Departments of Oncology, Laboratory Medicine and Pathology, and Neurosurgery, Mayo Clinic and Foundation. Rochester, MN. Accepted for publication February 17, 1992. Supported by the Mayo Comprehensive Cancer Center grant CA 15083-D2, National Cancer Institute, Department of Health and Human Services. Dr Horst was the recipient of a scholarship from the Dr Mildred Scheel Foundation for Cancer Research, Bonn. Germany. Key words: transforming growth factor-p. astrocytoma, neovascularization, tumor-infiltrating lymphocytes, tumor-infiltrating macrophages, immunohistochemistry. Address correspondence and reprint requests to John S. Kovach. MD, Department of Oncology, Mayo Clinic, Rochester, MN 55905. Copyright 0 1992 by W.B. Saunders Company 0046-8177/92/2311-0015$5.00/O
man glioblastomas but not in normal fetal and adult brain tissue,” and the report by Kuppner et al” that few tumor-infiltrating lymphocytes and macrophages are present in human astrocytomas prompted us to look in situ for the distribution of TGF-0, and to study the relationship between the distribution of TGF-P, and lymphoreticular cells in human astrocytomas. MATERIALS
Tissues We studied 16 human astrocytomas (eight fibrillary, six gemistocytic, one small cell, and one pilocytic) and five mixed astrocytoma-oligodendrogliomas. Eight of the surgically resected specimens consisted in part of brain tissue that microscopically was either free of tumor or contained only isolated tumor cells. Such tissues are referred to as “tumor free.” Tumor tissue was defined as heavily infiltrated brain parenchyma or solid neoplasm. All specimens were removed as part of a clinically indicated surgical procedure. The astrocytomas were graded according to the St Anne-Mayo system described by Daumas-Duport et al.” The tumor specimens were divided into two parts. One portion was fixed in 10% formalin for 24 hours and subsequently placed in Bouin’s fixative for 4 hours prior to paraffin embedding; the other portion was snap-frozen in liquid nitrogen and stored at -70°C.
Antibodies The polyclonal anti-TGF-8, (anti-CC) antibody was generously provided by K.C. Flanders and M.B. Sporn (National Institutes of Health, Bethesda, MD). The antibody was produced in rabbits after immunization with the N-terminal amino acids 1 to 30 of TGF-P, .6 Purification and characterization of the antibody have been described in detail.6,7 The antibody binds to secreted extracellular TGF-P, , the presumed active form of this regulatory protein, and does not bind to TGF&.* The monoclonal antibodies Leu-1 (CD5), anti-IL-PR (CD25), Leu-14 (CD22), and Leu-M3 (CD 14) were purchased from Becton Dickinson (Mountainview, CA).
Staining for TGF-/3, was done on paraffin-embedded sections for technical reasons. Reliable immunohistochemical differentiation of lymphocyte subsets (eg, activated IL-2 receptor-positive lymphocytes) was possible only in cryopreserved material. Transforming growth factor-/& was detected by a modified peroxidase anti-peroxidase method as described by Heine et al.’ T lymphocytes (Leu-1+), lymphocytes expressing the IL2 receptor (anti-IL-2R+), B lymphocytes (Leu-14+), and macrophages (Leu-MJS) in lymphoreticular infiltrates were
TGF-P, IN HUMAN GLIOBLASTOMAS (Horst et al) dctt~~rt-tl witll .I rlll~er-step inll~iutloperoxidase ~II1-lllic~k \crial set tiolls from fl-cbzen samples.
‘l’ullloI.-iIltlltr;ltillg lymphocytes and macrophages were in secl.ions stained for- [email protected]
, and counterstained with hrmatoxvlin as wrll as in serial sections stained with hematox\ lir~~eosik andI in t11e iiilnlullostained frozen sections. The frecluenc) distribution of lymphocytes and macrophages was assessed s~irliqll;Intit;ltively using a modification of a grading schc~~r based on the number ot’ c.ells per high-power field developed hv Brooks et al.“’ I.ymphoreticular infiltration in the tumor p&enchynla was analyzed separately from that in ~lre perivascular spaces. The degree of infiltration into the 1~IIIO~ parenchvma has graded as follows: no specifically stained cells, onI\, oc.casional stained cells. and low (one positive CellI. modelare (two IO six positive cells), and high (> 4s positivle culls) number of immunostained cells. The degrew of infiltration into thr perivascular spaces was graded as follows: no specifically stained cells, only occasional stained tc~~s, and lo\! (a c.omplete Iaver of cells), model-ate (two to rhr