Aberrant hypermethylation in primary tumours and sentinel lymph node metastases in paediatric patients with cutaneous melanoma.

BACKGROUND: Debate on how to manage paediatric patients with cutaneous melanoma continues, particularly in those with sentinel lymph node (SLN) metastases who are at higher risk of poor outcomes. Management is often based on adult algorithms, although differences in clinical outcomes between paediatric and adult patients suggest that melanoma in paediatric patients differs biologically. Yet, there are no molecular prognostic studies identifying these differences. OBJECTIVES: We investigated the epigenetic (methylation) regulation of several tumour-related genes (TRGs) known to be significant in adult melanoma progression in histopathology(+) SLN metastases (n = 17) and primary tumours (n = 20) of paediatric patients with melanoma to determine their clinical relevance. METHODS: Paediatric patients (n = 37; ≤ 21 years at diagnosis) with American Joint Committee on Cancer stage I-III cutaneous melanoma were analysed. Gene promoter methylation of the TRGs RASSF1A, RARß2, WIF1 and APC was evaluated. RESULTS: Hypermethylation of RASSF1A, RARß2, WIF1 and APC was found in 29% (5/17), 25% (4/16), 25% (4/16) and 19% (3/16) of histopathology(+) SLNs, respectively. When matched to adult cutaneous melanomas by Breslow thickness and ulceration, hypermethylation of all four TRGs in SLN(+) paediatric patients with melanoma was equivalent to or less than in adults. With a median follow-up of 55 months, SLN(+) paediatric patients with melanoma with hypermethylation of > 1 TRG vs. ≤ 1 TRG had worse disease-free (P = 0·02) and overall survival (P = 0·02). CONCLUSIONS: Differences in the methylation status of these TRGs in SLN(+) paediatric and adult patients with melanoma may account for why SLN(+) paediatric patients have different clinical outcomes. SLN biopsy should continue to be performed; within SLN(+) paediatric patients with melanoma, hypermethylation of TRGs can be used to identify a subpopulation at highest risk for poor outcomes who warrant vigilant clinical follow-up.

Genetic analysis of the human tumor necrosis factor alpha/cachectin promoter region in a macrophage cell line.

The 615-bp 5' flanking region of the human TNF-alpha/cachectin gene was isolated and ligated to the luciferase reporter gene. In addition, a series of truncated promoter constructs was generated by exonuclease III digestion. The promoter activity of these constructs was studied in a transient transfection system using the TNF-alpha-producing U937 cell line. Full-length and truncated TNF promoter constructions extending from -615 to -95 bp relative to the transcription start site (TSS) could be induced by phorbol esters. A construct truncated to within 36 bp of the TSS (and within 11 bp of the TATAA box) was inactive. Therefore, the phorbol ester responsive is localized in the TNF/cachectin promoter to a relatively short region proximal to the TATAA box.

Human osteosarcomas: immunologic evidence suggesting an associated infectious agent.

Immunofluorescent studies have revealed a high incidence of antibodies to osteosarcomas in the serums of patients with this disease and their close associates which react with a common antigen (or antigens) in osteosarcomas. The distribution of these antibodies suggests the association of an infectious agent with this neoplasm which is capable of producing unrecognized infections in healthy contacts of these patients.

Evidence for in vivo reaction of antibody and complement to surface antigens of human cancer cells.

The immune adherence test was used to determine whether antibody and complement in cancer patients are fixed in vivo to tumor cells. Human erythrocytes adhered in vitro to the surface of human cancer cells obtained from autopsy and biopsy. Adherence was enhanced by further addition of the C2 and C3 components of complement, and was diminished by preliminary treatment with antibody to C3 (that is, to beta1C-globulin). The results suggest that tumor associated membrane antigens form complexes in vivo with antibodies and complement.

Microcytotoxicity test: detection in sarcoma patients of antibody cytotoxic to human sarcoma cells.

A microcytotoxicity test has been used to detect a factor cytotoxic for human sarcoma cells; the factor was found in serums from 70 percent of sarcoma patients, 58 percent of their family members, and 8 percent of serums from normal blood donors. This cytotoxin is an antibody against a common cell surface sarcoma antigen since it is specific for sarcoma cells, is complement dependent, and is extractable with the globulin fraction of serum.

Human liposarcomas: tissue cultures containing foci of transformed cells with viral particles.

Viral particles occurring in foci of human liposarcoma cells in tissue culture are morphologically similar to the sarcoma viruses of the avian and murine species. Antibodies in the serum of the patient from whom the culture was originated reacted with cytoplasmic antigens in the original liposarcoma and in the cultured liposarcoma cells.

Studies of a melanoma tumor-associated antigen detected in the spent culture medium of a human melanoma cell line by allogeneic antibody. I. Purification and development of a radioimmunoassay.

A tumor-associated antigen (TAA) was isolated from spent culture medium (chemically defined serum-free medium) of human melanoma cell line UCLA-SO-M14 (M14). The isolation procedures included concentration, ultrafiltration, gel filtration chromatography, and chloroform:methanol (C:M) extraction. The melanoma TAA activity recovered from the organic phase of the C:M extract was subsequently fractionated by gel filtration and radiolabeled with Na125I. The radioiodinated antigen was further purified by Sephacryl S-200 gel filtration and allogeneic antibody affinity chromatography. With the use of previously characterized anti-TAA allogeneic sera from melanoma patients and 125I-labeled TAA, a radioimmunoassay (RIA) was developed. Protein A-bearing Staphylococcus aureus was used to separate bound and unbound 125I-labeled TAA. The coefficient of variation between experiments and within experiments with unlabeled melanoma TAA as the competitor in the competitive RIA ranged from 8.9 to 20.4%. These variations were consistently lower (8.9-13.6%) at high levels (6 micrograms melanoma TAA/ml) of the competitor than they were (17.3-20.5%) at low levels (0.5 microgram melanoma TAA/ml), suggesting reasonable reproducibility of the assay. A logit versus log plot of the competitive RIA data and analysis by linear regression yielded a straight line. This line represented a 5- to 1,000-ng detection range for melanoma TAA. Analysis of C:M-extracted and Sephacryl S-200-purified melanoma TAA by the competitive RIA revealed a 695-fold purification of the antigen that represented a 37.5% recovery from the spent culture medium. The greatest enrichment of the melanoma TAA was achieved by the C:M extraction step. This step separated the melanoma TAA from other antigens, e.g., fetal antigen and human leukocyte antigens.

Studies of a melanoma tumor-associated antigen detected in the spent culture medium of a human melanoma cell line by allogeneic antibody. II. Immunobiologic characterization.

A tumor-associated antigen (TAA) was isolated from spent culture medium of human melanoma cell line UCLA-SO-M14 (M14). After radioiodination and further purification, it was used in a radioimmunoassay (RIA). When tested in RIA for anti-TAA activity, 56% (111/200) of sera from melanoma patients, 21% (21/100) of sera from sarcoma patients, and 19% (19/100) of sera from carcinoma patients, as well as 12% (6/50) sera from pregnant women and 10% (10/100) apparently normal sera, were positive. The variations (0.25-3.8 micrograms/mg total protein) in specific TAA activity in 20 different batches of spent media collected for 18 months exhibited patterns of gradual increases and decreases, suggesting a cell growth cycle-related phenomenon. Binding between the allogeneic antibody and 125I-labeled TAA was inhibited by 70% (32/47) of melanoma spent media and cells (from culture or biopsy). Conversely, only 7% (3/33) spent medium or membrane extracts of other tumor cells and 0% (1/69) of human normal, fetal, or placental cells were positive in competitive RIA. The TAA was immunologically different from normal tissue antigens, blood-group antigens, human leukocyte antigens, and microbial antigens. These immunobiologic properties of the TAA isolated from M14 spent culture medium and used in the RIA suggested that such TAA was melanoma associated.

Studies of a melanoma tumor-associated antigen detected in the spent culture medium of a human melanoma cell line by allogeneic antibody. III. Physicochemical properties.

A melanoma tumor-associated antigen (TAA), isolated from spent culture medium of human melanoma cell line UCLA-SO-M14, was purified to mean homogeneity to determine its physical and biochemical nature. Gel filtration and native polyacrylamide gel electrophoretic analyses of the 125I-labeled melanoma TAA revealed that the antigen had a molecular weight in the range of 180,000-190,000. However, ultracentrifugation of melanoma 125I-labeled TAA in a 5-20% sucrose density gradient revealed a sedimentation coefficient to be 4.96 +/- 0.24. Melanoma 125I-labeled TAA focused at a pH of 6.5 by isoelectric focusing. No carbohydrates were detectable by various colorometric methods in the highly purified melanoma TAA fraction, and melanoma TAA failed to bind with several lectins. Its antigenic activity was destroyed by the proteolytic enzymes but was not affected by the glycosidic enzymes or phospholipase A2. The results suggested that the melanoma TAA was most likely a lipoprotein that had a molecular weight of 180,000-190,000, a sedimentation coefficient of approximately 5, and a pl value of 6.5. The protein portion apparently formed the antibody binding site(s).

Growth and metastasis of human melanoma xenografts in the hamster host.

Cells from three human melanoma cell culture lines (UCLASO-M-12, UCLASO-M-7, and WEG-1) were transplanted into the cheek pouches of immunosuppressed hamsters. Tumor nodules were found in the cheek pouches of hamsters receiving any one of these lines, and by 90--100 days after transplantation, nearly all hamsters had grossly visible lung metastases. Metastases were noted only in hamsters receiving continuous immunosuppression during the transplant period.

Immunochemical characterization of fetal antigen isolated from spent medium of a human melanoma cell line.

A fetal antigen (FA) was isolated from spent culture medium of a melanoma (M14) cell line. Allogeneic serum samples from melanoma patients, previously characterized with respect to anti-FA activity, were used as the source of anti-FA antibody. The FA activity was partially purified by membrane ultrafiltration, gel filtration, and chloroform:methanol extraction. The partially purified FA was then used to develop an enzyme-linked immunosorbent assay (ELISA). By indirect ELISA both the IgG and IgM classes of anti-FA antibodies were detected in the sera of cancer patients and normal volunteers. The incidences of anti-FA antibodies in the sera of cancer patients and normal volunteers were not significantly different. As detected by competitive inhibition in ELISA, FA activity was widely distributed among melanoma, sarcoma, and carcinoma tumor tissues and cultured tumor cells, as well as among fetal brain, skin, and muscle tissues. FA activity was destroyed by treatment with beta-galactosidase and hyaluronidase, but it was not destroyed by proteolytic and lipolytic enzymes. The antigen bound to immobilized ricin, peanut, and soybean lectins. FA activity in material purified by ricin-affinity chromatography was associated with molecules in the 60,000- to 70,000-dalton region as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. These results suggest a glycoprotein nature for the FA isolated from the spent culture medium of melanoma (M14) cells; this FA apparently elicits formation of natural antibodies in the cancer patients and normal donors.

Oncofetal antigen: a tumor-associated fetal antigen immunogenic in man.

Oncofetal antigen (OFA) has been defined with the use of human natural antibodies as a membrane antigen of human cancer cells that cross-reacts with human fetal brain tissues. The immunogen that elicits the antibody is unknown. The present study was undertaken to examine the immunogenicity of the OFA found on tumor cells. Postoperative melanoma patients were immunized with OFA-positive melanoma cells. Anti-OFA reactivities in the immunized sera were titrated by the immune adherence assay with the use of a known OFA-positive cultured melanoma cell line, M14, as target cell. Alloantibodies were excluded by absorption with lymphoblastoid cells autologous to M14. Anti-OFA antibody then was identified by absorption with fetal brain. In 6 months of immunization, 19 of 23 patients produced increased anti-OFA antibodies. The peak titers ranged from 1:16 to 1:2,048. Sera from 18 patients who were not immunized also were tested for 6 months postoperatively, and none had significant increases in antibody titers. The increase of anti-OFA antibody titer in response to the immunization with OFA-positive tumor cells suggests the immunogenic capability of tumor-related OFA in man.

Oncofetal antigen-I: distribution in human tumors.

Oncofetal antigen-I (OFA-I) is a human tumor-associated fetal antigen that cross-reacts with fetal brain. The biological distribution of this antigen was studied in a variety of histologic types of human tumors and normal tissues. A total of 170 human tumors and 89 normal tissues obtained from biopsied and autopsied specimens were tested by immune adherence absorption assay. The antigen was most prevalent in malignant melanomas (82%); next in order of incidence were lung carcinomas (71%), sarcomas (61%), brain tumors (58%), and breast carcinomas (53%). However, OFA-I was found in only 15% of gastrointestinal tumors. The four hepatoma specimens tested were negative for OFA-I. OFA-I was also present in metastases and normal-appearing tissues of patients with OFA-I positive tumors but was not found in normal tissues from noncancer patients. These studies demonstrated that OFA-1 is widely distributed among human tumors. It appears to be more prevalent in tumors of ectodermal and mesodermal origin.

Gangliosides of human melanoma.

The ganglioside composition of human malignant melanoma was studied with the use of 80 melanoma specimens, including 52 surgical specimens and 28 cultured cell lines. A ganglioside fraction was isolated and purified from each of these tissues, and the amount of each component ganglioside was assessed by thin-layer chromatography (TLC) and a TLC scanner. Five gangliosides (GM3, GD3, GM2, GD2, and alkali-labile ganglioside) were most commonly expressed by these melanomas. However, the total ganglioside amount (ranging from 33 to 302 micrograms/g wet wt of tissue) as well as the distribution of each ganglioside were widely heterogeneous in both biopsied and cultured melanomas. When the ganglioside expressions of cultured and biopsied melanomas were compared, GM2 and GD2 were minor components of biopsied melanomas but often became major components of cultured melanoma cells. Conversely, alkali-labile ganglioside was expressed more strongly on biopsied melanomas. This heterogeneity suggests that it will be necessary to analyze the ganglioside composition of biopsied melanomas before using monoclonal antibodies to melanoma-associated gangliosides for melanoma diagnosis or therapy.

Prolonged survival for melanoma patients with elevated IgM antibody to oncofetal antigen.

Antibody directed against a cultured melanoma cell line known to express an oncofetal antigen was measured in sera obtained from patients with stage II melanomas. This study was undertaken to determine if an inhibiting or enhancing effect on tumor growth would be suggested by a positive or negative correlation of antibody levels with tumor recurrence or patient survival. A positive correlation with disease-free interval and survival was detected among patients who had high levels of the IgM class of antibody before and shortly after surgery for state II disease. The IgG class of antibody over the period measured did not correlate consistently with tumor recurrence. Absorption of the IgM antibody with fetal brain confirmed that the dominant detectable reactivity was directed to the oncofetal antigen. The relevance of these findings is related to that of other evidence indicating that immunity to fetal antigens expressed on tumor cells is a participant in host-tumor interaction.

Systemic administration of human leukocyte interferon to melanoma patients. I. Effects on natural killer function and cell population.

Natural killer (NK) cytotoxicity was assessed against K562 targets in 14 melanoma patients who received daily im doses of human leukocyte interferon (IFN) for 42 consecutive days. The most common pattern of NK activity was a decline from pretreatment levels 1 day after initiation of treatment, followed by increasing cytotoxicity with peak activity at day 7 and a subsequent gradual decline to pretreatment levels during the remaining weeks of treatment. This pattern was particularly apparent in patients who received 3 x 10(6) or 9 x 10(6) U IFN/day, while patients who received 1 X 10(6) U IFN/day tended to lack the decline at day 1 and maintained the elevated NK activity past day 7. Changes in NK activity could not be related to changes in absolute lymphocyte counts; to proportions of cells bearing membrane receptors for erythrocyte-antibody-complement, of cells bearing FC gamma receptor; or to clinical response to IFN.

Comparison of histocompatibility antigens on cultured human tumor cells and fibroblasts by quantitative antibody absorption and sensitivity to cell-mediated cytotoxicity.

Human tumor and febroblast tissue culture cells were compared to determine the suitability of fibroblasts as control cells in experiments on human tumor serology and cellular immunology. Fibroblasts expressed the same HL-A antigen profile as did melanoma cells. Furthermore, the quantitative expression of the determinants was similar on both cell types. In four of five pairs tested, the fibroblasts displayed similar sensitivity to effector cells generated by mixed lymphocyte culture as did the tumor cells from the same donor, but there were some differences in the effects of specific alloimmune effector cells at high and low effector-to-target ratios on the two types of target cells. Results indicated that fibroblasts are legitimate control target cells for studies in human tumor immunology, if screening assays are done to verify their antigenicity and sensitivity to cell-mediated cytolysis.

Suggestive evidence for in vivo binding of specific antitumor antibodies of human melanomas.

Antibodies eluted from homogenates of human melanoma cells reacted against melanoma cells reacted against melanoma antigens in a complement fixation test. Before elution, sonically treated homogenate did not react significantly against autologous serum but, following elution, antigenic activity increased markedly (up to 32-fold). Eluate of one melanoma reacted with the sonically treated residue of other melanomas but not with similarly prepared residues of sarcoma, carcinomas, or normal tissues. Melanoma eluates comtained more IgG than IgA. Traces of IgM were found in two melanoma eluates. Eluates of normal tissues (lung, kidney, and muscle) were devoid of serum proteins and did not react with the soncially treated melanoma residues. These results support the hypothesis that antitumor antibodies are bound to melanoma cells in vivo and that these antigens are cross-reactive.

An epitope common to gangliosides O-acetyl-GD3 and GD3 recognized by antibodies in melanoma patients after active specific immunotherapy.

GD3 is a major ganglioside of human melanoma and was shown to be an effective target for passive immunotherapy with murine monoclonal antibodies. It was noted earlier that GD3 neither purified nor in melanoma cell vaccine (MCV), could elicit an antibody response in melanoma patients. In this study, we demonstrate that melanoma patients who received MCV had autoantibodies against a derivative of GD3, O-acetylated GD3 (O-AcGD3), a minor ganglioside expressed on human melanoma cells, and that the antibodies cross-reacted with GD3. Thin layer chromatographic immunostaining revealed that all of the sera containing antibodies against O-AcGD3 also reacted to GD3. None of the other sera responded only to GD3, although the MCV contained 7- to 12-fold higher GD3 than O-AcGD3. Furthermore, the antibody activity was completely abolished by absorption with animal erythrocytes expressing either O-acetyl disialogangliosides or GD3, indicating that the antibodies recognize an epitope commonly shared by GD3 and O-AcGD3. The antibodies bound only to the sialyloligosaccharide moiety but not to the ceramide portion of GD3 after endoglycosylceramidase treatment. The antibodies failed to bind to GD3 after neuraminidase treatment. These results indicate that the sialyloligosaccharides of the gangliosides are important components of the epitope. Periodate oxidation abolished reactivity of the antibodies to GD3 but not that to O-AcGD3, revealing that the glycerol side chain of the sialic acids in both GD3s was an important structure of the epitope. The binding of the antibodies to melanoma cell surface gangliosides was confirmed by an absorption with a GD3- and O-AcGD3-positive melanoma cell line. These results in the light of previous reports on the inability of GD3 to elicit immune response in humans suggest that anti-GD3 antibodies found in the melanoma patients were induced by immunization with O-AcGD3 and O-AcGD3 present in the MCV would serve as an antigen source for GD3-targeted active specific immunotherapy of melanoma.

A membrane antigen common to human cancer and fetal brain tissues.

Membrane antigens of a cultured human melanoma line, UCLASO-M14, were studied using immune adherence techniques. Allogeneic sera from melanoma patients that were reactive with the M14 but nonreactive with lymphoid cells of the M14 donor were used as antibodies. The antigen responsible for the reaction between M14 and the antibodies was searched for in other cancer, normal, and fetal tissues using antibody absorption techniques. The antigen was found in a variety of different histological types of biopsied and cultured cancer cells as well as in melanomas. The antigen did not exist in biopsied normal tissues, but it appeared in cultured normal skin and muscle. Neither normal lymphocytes nor cultured lymphoid cells showed any antigenicity. The antigen was present in human fetal tissues and was the strongest in fetal brain tissues at 22 weeks of development. Liver, spleen, thymus, and small intestine from the same fetus were negative for antigen.