Myc represses miR-15a/miR-16-1 expression through recruitment of HDAC3 in mantle cell and other non-Hodgkin B-cell lymphomas.

Our recent study demonstrated miR-15a/16-1 downregulation in mantle cell lymphoma (MCL). Here, we investigated mechanisms of miR-15a/16-1 transcriptional repression and its epigenetic regulation by c-Myc and histone deacetylase (HDAC) in MCL. c-Myc expression was detected in MCL cell lines and in the primary MCL samples, and pri-miR-15a/16-1 mRNAs were significantly upregulated in Mino and Jeko-1 cells with c-Myc knockdown by small interfering RNAs (siRNAs). Our co-immunoprecipitation analysis showed that c-Myc interacted with HDAC3. Moreover, using chromatin immunoprecipitation, we demonstrated that both c-Myc and HDAC3 co-localized to the two promoters of the miR-15a/16-1 cluster gene, DLEU2, and inhibition of HDAC3 increased histone acetylation of the DLEU2 promoters. Luciferase reporter assay confirmed the dependence of Myc-mediated DLEU2 transcriptional repression on HDAC3. Treatment with the pan-HDAC inhibitor, suberoylanilide hydroxamic acid and HDAC3 siRNA resulted in increased miR-15a/16-1 expression. The regulatory mechanism of miR-15a/16-1 was further demonstrated in Burkitt lymphoma and Myc overexpressing cell lines. These findings highlight the role of HDAC3 in Myc-induced miR-15a/16-1 changes and reveal novel mechanisms for c-Myc-driven microRNA suppression and malignant transformation in aggressive B-cell malignancies.

Cdc2: a monopotent or pluripotent CDK?

Cell cycle progression is controlled by both extracellular and intracellular signalling molecules. It has been generally believed that cdc2/CDK1 only control G(2)-M transition in mammalian and many other higher eukaryotic cells. Accumulating evidence shows that cdc2 not only promotes G(2)-M transition but is also capable of regulating G(1) progress and G(1)-S transition via association with multiple interphase cyclins; cdc2 activity can be inhibited by p21 and p27, two traditional G(1) CDK inhibitors. In addition, cdc2-cyclin B controls pronuclear union in interphase fertilized eggs. These data suggest that cdc2 may be a pluripotent CDK. Although mechanisms responsible for the multiple functions of cdc2 remain to be further investigated, interactions of cdc2 with pRb and with several important transcription factors may provide a clue to the pluripotent role of cdc2.

Follicular dendritic cell-induced microRNA-mediated upregulation of PRDM1 and downregulation of BCL-6 in non-Hodgkin's B-cell lymphomas.

B-cell lymphoma 6 (BCL6) and PR domain containing 1 (PRDM1) are considered as master regulators for germinal center (GC) formation and terminal B-cell differentiation. Dysregulation of BCL6 and PRDM1 has been associated with lymphomagenesis. Here, we show for the first time that direct cell-cell contact between follicular dendritic cells (FDC) and B-lymphocytes, by influencing the expression of a set of microRNAs (miRNAs), regulates the expression of BCL6 and PRDM1. We identify that, on cell adhesion to FDC, FDC induces upregulation of PRDM1 expression through downregulation of miR-9 and let-7 families and induces downregulation of BCL-6 through upregulation of miR-30 family in B-lymphocytes and lymphoma cells. We further demonstrate that the miR-30 family directly controls BCL-6 expression and miR-9-1 and let-7a directly control PRDM-1 expression through targeting their 3'UTR, mediating the FDC effect. Our studies define a novel regulatory mechanism in which the FDC, through induction of miRNAs in B-lymphocytes, orchestrates the regulation of transcription factors, promotes germinal center B-cell survival and differentiation. Dysregulation of miRNAs may interfere with B-cell survival and maturation, thus representing a novel molecular mechanism, as well as a potential therapeutic target in B-cell lymphomas.

TGFbeta inhibits GM-CSF-induced phosphorylation of ERK and MEK in human myeloid leukaemia cell lines via inhibition of phosphatidylinositol 3-kinase (PI3-k).

OBJECTIVES: Activation of SMAD-independent p44/42 MAPK (ERK1/2) signalling by TGFbeta has been recently reported in various cell types. However, the mechanisms for the linkage between the SMAD-dependent and -independent pathways are poorly understood. In this study, we investigated whether TGF-beta activates the ERK pathway and how TGFbeta communicates with the MAP kinase signals induced by a mitogen, in human myeloid leukaemia cells. MATERIALS AND METHODS AND RESULTS: TGFbeta dramatically suppressed proliferation of MV4-11 and TF-1 cells without detectable phosphorylation of ERK1/2 and MEK1/2 for the duration of 48 h, as detected by MTT assay and Western blot analysis, respectively. In contrast, GM-CSF induced rapid and transient phosphorylation of MEK1/2 and ERK1/2 and up-regulated cell proliferation. Both GM-CSF-induced ERK1/2 activation and cell proliferation were significantly inhibited by TGFbeta. GM-CSF also induced transient phosphorylation of the p85 subunit of PI3-kinase. Corresponding to this change, phosphorylated p85 was found to bind to the GM-CSF receptor-alpha subunit, as detected by immunoprecipitation and Western blot analysis. PD98059, a selective inhibitor of MEK, blocked GM-CSF-induced phosphorylation of MEK and ERK but not p85. However, TGFbeta and LY294002, a potent inhibitor of PI3-kinase, significantly inhibited phosphorylation of both p85 and ERK1/2. CONCLUSIONS: These studies thus indicate that TGFbeta does not activate the ERK pathway but turns off the GM-CSF-induced ERK signal via inhibition of the PI3-kinase-Akt pathway, in these human leukaemia cells.

Lymphoma cell adhesion-induced expression of B cell-activating factor of the TNF family in bone marrow stromal cells protects non-Hodgkin's B lymphoma cells from apoptosis.

This study explores whether lymphoma cell adhesion-induced B cell-activating factor (BAFF) expression in bone marrow stromal cells (BMSCs) protects B lymphoma cells from apoptosis. We first showed protection of lymphoma cells from apoptosis by conditioned medium of a stromal cell-lymphoma cell coculture, either spontaneous or induced by mitoxantrone, implying a role for soluble factor(s) in lymphoma cell survival. Addition of BAFF counteracted mitoxantrone-induced apoptosis and elicited a reduction in spontaneous apoptosis in primary lymphomas, suggesting a role of BAFF in sustaining B-cell survival. Abundant BAFF was detected in the BMSC cell line (HS-5) and primary BMSCs by flow cytometry, RT-PCR and immunoblotting. BAFF levels were 20- to 200-fold higher in BMSCs than in lymphoma cells, and lymphoma cell adhesion to BMSCs augmented BAFF secretion twofold through upregulation of BAFF gene expression. Finally, neutralization of BAFF by TACI-Ig or depletion of BAFF by small hairpin RNA (shRNA) in BMSCs significantly enhanced lymphoma cell response to chemotherapy and overcame stroma-mediated drug resistance, suggesting that lymphoma cells use BMSC-derived BAFF as a survival factor. These findings support the hypothesis that lymphoma cells interact with BMSCs, resulting in stromal niches with high BAFF concentration, and identify BMSC-derived BAFF as a functional determinant for B lymphoma cell survival in the bone marrow environment.

Bone marrow stromal cells prevent apoptosis of lymphoma cells by upregulation of anti-apoptotic proteins associated with activation of NF-kappaB (RelB/p52) in non-Hodgkin's lymphoma cells.

Stromal cells are an essential component of the bone marrow microenvironment that regulate or supports tumor survival. In this study we therefore studied the role of stromal cells in lymphoma cell survival. We demonstrated that adhesion of the B-cell lymphoma cell lines SUDH-4 and 10 to bone marrow stroma inhibited mitoxantrone-induced apoptosis. This adhesion-dependent inhibition of mitoxantrone-induced apoptosis correlated with decreased activation of caspases-8 and 9, and cleavage of caspase 3 and PARP. Electrophoretic mobility shift assays (EMSA) analysis demonstrated significantly increased NF-kappaB binding activity in lymphoma cells adhered to stroma cells compared to lymphoma cells in suspension. This DNA binding activity could be attributed to cell adhesion-mediated proteolysis of the NF-kappaB precursor, p100 (NF-kappaB2). This resulted in the generation of active p52, which translocated to the nucleus in complex with p65 and RelB. Coculture with stromal cells also induced expression of the NF-kappaB-regulated anti-apoptotic molecules, XIAP, cIAP(1) and cIAP(2). Inhibition of NF-kappaB significantly suppressed HS-5-induced protection against apoptosis in lymphoma cell lines as well as in primary lymphoma cells. Thus, bone marrow stroma protects B-cell lymphoma cells against apoptosis, at least in part through activation of NF-kappaB dependent mechanism involving up-regulation of NF-kappaB regulated antiapoptotic proteins. Consequently, this study suggests a new approach to decrease the resistance of lymphoma to chemotherapy.

An IkappaBalpha inhibitor causes leukemia cell death through a p38 MAP kinase-dependent, NF-kappaB-independent mechanism.

Treatment of U937 cells with an IkappaBalpha phosphorylation inhibitor, Bay 11-7085, induced a rapid phosphorylation of p38 mitogen-activated protein (MAP) kinase, significant apoptosis, extensive necrosis, and a weak phosphorylation of MAP kinase kinase. Bay 11-7085 had no effect on the basal levels of phosphorylated IkappaBalpha but completely inhibited phorbol 12-myristate 13-acetate-induced phosphorylation of IkappaBalpha. Although Bay 11-7085 prevented phorbol 12-myristate 13-acetate-induced NF-kappaB nuclear translocation, SN50, a specific inhibitor of nuclear translocation and function of NF-kappaB, did not induce any significant nuclear/DNA fragmentation, caspase 3 activation, or cell death. The p38 MAP kinase-specific inhibitor, SB203580, completely inhibited the phosphorylation of p38 MAP kinase and significantly decreased Bay 11-7085-induced apoptosis. In contrast, the MAP kinase kinase-specific inhibitor PD98059 had no effect on Bay 11-7085-induced apoptosis. Caspase-specific inhibitor, z-Val-Ala-Asp-fluoromethyl ketone prevented Bay 11-7085-induced activation of caspase 3 but was not able to block Bay 11-7085-induced phosphorylation of p38 MAP kinase. These data suggest that Bay 11-7085 induces apoptosis through a p38 MAP kinase-dependent, NF-kappaB-independent mechanism.

IgM Myeloma: An IgM Gammopathy Distinct from Waldenstrom's Macroglobulinemia.

A 65-year-old patient with an IgM gammopathy, plasma cell infiltration of the bone marrow, lytic lesions of the skeleton, and symptoms of hyperviscosity was evaluated. A diagnosis of IgM myeloma was made. Treatment for multiple myeloma was initiated and resulted in a significant clinical response. There is ongoing debate whether IgM myeloma exists as a unique entity, oras a variant of Waldenstrom's macroglobulinemia. We reviewed the English literature and discovered only six cases documented as IgM multiple myeloma. Further investigation revealed 30 reported cases, which had characteristics typical of an IgM myeloma, but were documented as variations of B-cell neoplastic disease. We acknowledge IgM myeloma to be an entity distinct from Waldenstrom's macroglobulinemia. It is characterized by an IgM monoclonal gammopathy, a predominance of plasma cells infiltrating the bone marrow, and typically associated with osteolytic lesions and/or osteoporosis. Recognizing IgM myeloma is necessary for appropriate disease management.

Prolonged activation of the mitogen-activated protein kinase pathway is required for macrophage-like differentiation of a human myeloid leukemic cell line.

The role of the mitogen-activated protein kinase (MAPK) signal transduction pathway in the proliferation of mammalian cells has been well established. However, there are relatively few reports concerning cell differentiation being mediated by MAPK. The effect of phorbol 12-myristate 13-acetate (PMA) on cell differentiation and signal transduction in a human myeloid leukemia cell line, TF-1a, was investigated. When TF-1a cells were treated with 10(-6), 10(-7), 10(-8), and 10(-9) M PMA for 24 h, they underwent 98, 93, 91, and 51% macrophage-like differentiation, respectively. PMA treatment rapidly (10 min) induced phosphorylation of MAPK kinase (MEK and p44/42 MAPK), which persisted for at least 24 h. p44/42 MAPK immunoprecipitates from lysates of PMA-treated cells had increased ability to phosphorylate the transcription factor Elk-1. This is important because phosphorylated Elk-1 can be considered an "end-product" of the MAPK pathway. In contrast, treatment of TF-1a cells with granulocyte/macrophage-colony stimulating factor induced only transient activation of MEK and p44/42 MAPK (10-20 min) and an increase (approximately 50%) in cell proliferation, without any change in cellular differentiation. These results suggest that macrophage-like differentiation may be dependent on prolonged activation of the MAPK pathway. Additional support for this conclusion was obtained from experiments showing that treatment of TF-1a cells with antisense oligonucleotides for MEK1 coding sequences prior to adding PMA inhibited macrophage-like differentiation. Furthermore, transient transfection with an inactive, dominant-negative MEK mutant also inhibited PMA-induced differentiation, whereas transient transfection with a plasmid coding for constitutively activated MEK led to macrophage-like differentiation in the absence of PMA.

CD44 expression and MMP-2 secretion by mouse glioma cells: effect of interferon and anti-CD44 antibody.

We have previously reported that invasiveness of mouse glioma G-26, which expresses CD44 adhesion molecule, was inhibited in vitro following treatment with anti-CD44 antibody or mouse interferon alpha/beta (MuIFN alpha/beta). Here, we evaluated whether the expression of transmembrane CD44 adhesion molecule and/or secretion of extracellular matrix metalloproteinases (MMPs) were affected when glioma cell invasion was inhibited. Flow cytometric evaluation of CD44 adhesion molecule expression in G-26 glioma using anti-CD44 antibody, confirmed that G-26 cells were CD44+. Following 3-day treatment with MuIFN alpha/beta at 8 x 10(2) or 8 x 10(3) IU/ml of glioma cells, the expression of CD44 was not significantly affected as reflected by CD44+ cell number and fluorescence intensity. The pretreatment of glioma cells for 1 day with anti-CD44 antibody resulted in a 30-60% decrease of CD44 expression. This coincided with significantly (p < 0.05) lower cell activity as judged by MTT assay for mitochondrial activity. The zymographic evaluation of MMP activity in the G-26 glioma cell culture showed a high level of the active form of MMP-2. This level of MMP-2 was decreased following 3 day treatment of G-26 glioma cells with either 8 x 10(2) or 8 x 10(3) IU/ml of MuIFN alpha/beta but only the latter concentration produced statistically significant 55% decrease. However, following a 1 day treatment of G-26 glioma cells with anti-CD44 antibody, the level of active MMP-2 form was not significantly affected. These findings indicate that while the inhibitory effect of IFN on glioma invasion was accompanied by a decreased level of the active form of MMP-2 released extracellularly, the expression of the transmembrane CD44 adhesion molecule was not affected. Conversely, anti-CD44 antibody pretreatment of G-26 glioma, which led to the inhibition of glioma invasion, resulted in decreased CD44 expression and lower cell activity but had no effect on the MMP-2.

Transforming growth factor beta inhibits growth of more differentiated myeloid leukemia cells and retinoblastoma protein phosphorylation at serine 795.

Transforming growth factor beta (TGF-beta) has been shown to be a specific inhibitor of early human myeloid progenitors. We show here that TGF-beta1 potentially inhibited not only the growth of primitive but also more mature myeloid leukemic cells. Surprisingly, those apparently more mature progenitor cells, such as MV4-11 and Mo7e cells, are very sensitive to the action of TGF-beta. The addition of TGF-beta1 to liquid cultures of these cells significantly inhibited their proliferation, with as much as 72% inhibition of growth of MV4-11 cells. The suppressive effect by TGF-beta1 was not reversed or prevented by granulocyte-macrophage colony-stimulating factor or interleukin 3 used to promote cell growth in TF-1a and MV4-11 cells. TGF-beta1 completely abolished the clonal growth of MV4-11 cells in soft agar and inhibited Mo7e, KG-1, K562, TF-1, and TF-1a colony growth by 99%, 90%, 63%, 53%, and 43%, respectively. The cells treated with TGF-beta1 showed progressive accumulation in the G1 phase of cell cycle. Maximal G1 arrest (93%) was observed in MV4-11 cells. Using anti-retinoblastoma protein (pRb) and anti-specific phosphorylated-pRb antibodies, we demonstrated that TGF-beta1 greatly inhibited pRb phosphorylation at serine 795 in MV4-11 and Mo7e cells. Taken together, our data suggest that the sensitivity of myeloid leukemic progenitor cells to growth inhibition by TGF-beta may not be inversely correlated with their maturation stage, and the inhibition of the cells appeared to be linked to the suppression of pRb phosphorylation at serine 795.

Characterization of a unique factor-independent variant derived from human factor-dependent TF-1 cells: a transformed event.

A factor-independent variant (TF-1a) has been isolated from the factor-dependent TF-1 cell line. The subline has been grown continuously in culture for > 1.5 years without added cytokines. The cells retain the ability to respond to multicytokines, with a different response pattern from its parental cell line. The TF-1 cells appeared singly in liquid culture. In contrast. TF-1a cells formed aggregates which increased markedly in size and in number upon TGFbeta1 treatment and showed a diminished TGFbeta-mediated growth inhibition. TF-1a, but not TF-1 cells, formed colonies in soft agar culture in the absence of any added growth factors, and developed the capacity to generate an invasive tumor(s) in nude mice. There was a constitutive activation of MAPK and MEK in TF-1a but not in TF-1 cells, which may be one of the mechanisms leading to factor-independent growth of TF-1a cells. Phenotypically, TF-1 cells were CD34+ /CD38+, whereas TF-1a cells were CD34+ /CD38-. This suggests that TF-1a may represent a less mature hematopoietic cell than TF-1. In conclusion, TF-1a is different from TF-1 in many important aspects which are associated with neoplastic transformation. The variant appears to be an excellent model for studying the process of progressive malignant transformation of myeloid cells and for studying signal pathways involved in the spontaneous and factor-induced growth of the cells.

Assessment of intracellular TAP-1 and TAP-2 in conjunction with surface MHC class I in plasma cells from patients with multiple myeloma.

Immunotherapy involving cytotoxic T lymphocytes (CTLs) is an attractive alternative for treatment of various malignancies, including multiple myeloma. For tumour cells to be recognized and killed by CTLs they must express cell surface major histocompatibility complex (MHC) class I molecules and the transporter associated with antigen processing (TAP). However, loss of MHC class I and the TAP protein are common among several types of solid tumours. This study assessed the expression of TAP protein (by intracellular flow cytometry) and cell surface MHC class I molecules in three human myeloma cell lines as well as the plasma cell population (CD38+ bright) in bone marrow specimens from 13 multiple myeloma patients. In all of the patients, 100% of the plasma cell population expressed both the TAP subunits and cell surface MHC class I molecules, but at varying intensities. Both TAP and MHC class I were also expressed in the three myeloma lines. Additionally, the function of the antigen transport machinery was evaluated by a peptide transporter assay in the three myeloma lines. TAP transporter activity was readily detectable in two out of three myeloma lines, whereas the diminished activity in the third cell line was completely restored by co-culturing with recombinant interferon-gamma (rIFN-gamma).

Dexamethasone, cyclophosphamide, idarubicin and etoposide (DC-IE): a novel, intensive induction chemotherapy regimen for patients with high-risk multiple myeloma.

We evaluated toxicities and responses to a novel, dose intensive and time sequenced, chemotherapy programme (DC-IE) in 45 patients with high-risk myeloma. DC-IE consisted of: dexamethasone (days 1-4); cyclophosphamide (day 5); idarubicin and etoposide (days 8-10). Complete response (CR) was achieved in four patients, six patients achieved near complete responses (nCR) and 21 patients achieved a partial remission (PR). Overall response rate was 76% (CI 56-94%) for newly diagnosed patients (n = 21) and 62% (CI 36-81%) for relapsed/refractory patients (n = 24). Toxicities were limited to myelosuppression; two patients died of sepsis during neutropenia (4%). DC-IE is active and tolerable for high-risk multiple myeloma, including patients with relapsed or refractory disease to anthracycline containing regimens.

High-sensitivity detection of minimal residual breast carcinoma using the polymerase chain reaction and primers for cytokeratin 19.

We have developed a reverse transcriptase-polymerase chain reaction (RT-PCR) assay to identify breast carcinoma cells in bone marrow aspirates with high sensitivity and specificity. This assay relies on the detection of cytokeratin 19 (K19) RNA by nested primer PCR followed by annealing to a (32P)-labeled internal sequence probe and autoradiography. In reconstitution experiments, this assay is capable of detecting 10 fg of admixed mammary tumor RNA in 1 microgram of normal marrow RNA (a dilution of 1:10(7)). Thirty of 30 primary breast tumor specimens, 19 of 19 cytologically positive bone marrow aspirate specimens, and three of 11 aspirate negative/biopsy positive specimens showed detectable K19 transcript. This assay shows high specificity, with 50 of 52 negative control aspirates showing no detectable amplification product. False-positive amplification was noted in two of 18 aspirates obtained from patients with active chronic myelogenous leukemia. Of stage II and III postsurgical breast carcinoma patients with histologically negative bone marrows and no radiographic bone disease, 14 of 30 were K19 positive by PCR. RT-PCR analysis of K19 transcript is a highly sensitive and specific method of detecting and monitoring low-level metastatic disease in patients with primary carcinoma of the breast. The presence of K19 RNA in histologically negative bone marrows suggests that this assay may prove a powerful monitor for patients undergoing curative therapy as well as a novel prognostic indicator.

Clinical significance of bone marrow metastases as detected using the polymerase chain reaction in patients with breast cancer undergoing high-dose chemotherapy and autologous bone marrow transplantation.

PURPOSE: The present study evaluates the clinical significance of detection of cytokeratin 19 (K19) in the bone marrow of patients with breast cancer undergoing high-dose chemotherapy (HDCT) and autologous bone marrow transplantation (ABMT). PATIENTS AND METHODS: We studied retrospectively cryopreserved bone marrow aspirates from 83 patients with high-risk stage II, III, and IV breast cancer obtained before bone marrow harvest but after induction chemotherapy. All samples were histologically negative for metastases. Polymerase chain reaction (PCR) for K19 was performed according to methods described previously and results were correlated with the probability of relapse following HDCT and ABMT. RESULTS: The incidence of occult metastases as defined by PCR for K19 message was 52% for 19 stage II, 57% for 14 stage III, and 82% for 50 stage IV patients (two-tailed P = .0075, chi 2 test). The probability of relapse at 3 years after ABMT was 32% and 94% for K19-positive stage II/III and stage IV patients, respectively, versus 10% and 14% for K19-negative stage II/III and stage IV patients, respectively. The difference was significant for stage IV patients (two-tailed P = .0002). CONCLUSION: It has been shown that PCR is a highly sensitive method to detect K19 message in the bone marrow. The incidence of K19 positivity in bone marrow increases significantly with advancing stage. In patients with breast cancer, especially metastatic breast cancer, undergoing HDCT and ABMT, the presence of K19 is associated with a poor prognosis.

Recommendations for the use of routine bone marrow aspiration and lumbar punctures in the follow-up of patients with retinoblastoma.

PURPOSE: Patients with metastatic retinoblastoma have a poor outcome. Hope that early detection of extraocular spread will improve survival has led to routine monitoring with bone marrow and cerebrospinal fluid (CSF) examinations. In light of cost and patient morbidity, the clinical utility of this practice is questioned. PATIENTS AND METHODS: We have performed 254 serial bone marrow aspirations and 164 lumbar punctures in 60 children with retinoblastoma. RESULTS: Two patients with extensive intraocular disease at diagnosis developed positive bone marrow aspirations, although no patient died of distant metastasis. Three patients developed positive CSF examinations. All had neurologic symptoms at the time of CSF positivity. CONCLUSIONS: We recommend performing staging bone marrow and CSF evaluations only in patients with clinical, histologic, or radiologic evidence of local or systemic extension (Pratt stage III-IV), or in patients presenting with one Reese-Ellsworth group V eye and retrolaminar or extrascleral extension of their tumor. We recommend limiting follow-up bone marrow and CSF evaluations to patients who develop objective signs and symptoms of metastatic or regionally recurrent disease.

Urokinase plasminogen activator in ovarian cancer.

Invasion and metastasis require the destruction of the extracellular matrix and basement membranes to facilitate growth or migration of tumor cells into vascular and lymphatic spaces. These processes are mediated by proteolytic enzymes. Malignant cells produce urokinase which is a protease known to enhance the invasiveness of many tumor cells. The relationship between urokinase and various prognostic factors was investigated in 16 patients with epithelial ovarian cancer. Tissue concentrations of urokinase were measured in tumor cytosols using enzyme-linked immunoassays. Urokinase levels were lower in ovarian tumors of low malignant potential (median 9.5, range 3.5-18.3 pg/mg protein, n = 4) than invasive cancers (median 44.6, range 16.1-210.6 pg/mg protein, n = 12), p < 0.01. In the invasive carcinomas urokinase levels did not vary significantly with tumor stage or cell type. Grade 3 tumors had higher levels of urokinase (median 120.4, range 21.4-397.1 pg/mg protein, n = 6) than grade 1 and 2 tumors (median 29.2, range 16.1-51.8 pg/mg protein, n = 6), p < 0.05. Urokinase levels were higher in recurrent (median 120.4, range 51.8-210.6 pg/mg protein, n = 4) than in primary (median 29.2, range 16.1-97.1 pg/mg protein, n = 8) tumors, p < 0.05. These results support the hypothesis that urokinase plays a role in invasion and metastasis of ovarian epithelial cancers and suggest that tissue levels of urokinase may have prognostic value.