Relationships between membrane antigens of human leukemic cells and oncogenic RNA virus structural components.

Leukemic cells from all human chronic granulocytic leukemia (CGL) and some acute myelomonocytic leukemia (AMML) donors are lysed by rabbit antisera to a purified glycoprotein of Friend murine leukemia virus (FLV gp71) in a microcytotoxicity assay. These antisera are not cytotoxic to cells from patients with acute myelocytic leukemia (AML), acute lymphocytic leukemia (ALL), chronic lymphocytic leukemia (CLL), or to peripheral blood lymphocytes from normal donors. A goat antiserum to gradient purified FLV in addition to reacting with cells from CGL and AMML donors also reacted with cells from AML patients and some ALL donors. However, this antiserum failed to react with cells from CLL patients. Peripheral blood and bone marrow leukocytes prepared from leukemic patients in clinical remission failed to react with antisera to FLV and FLV gp71. Absorption experiments demonstrated that the antigen on CGL cells which is reacting with the antiserum to FLV gp71 is also present on normal human platelets and neutrophils. Similar absorption studies showed that the antigen on AML cells detected by the FLV antiserum is not present on normal leukocytes and platelets and appears to be related to the major internal p30 antigens of mammalian RNA tumor viruses. Another antigenic relationship between oncornaviruses and membrane antigens of human leukemia cells was shown by the ability of FLV antigens to absorb the cytotoxic reactivity of nonhuman primate antisera detecting human leukemia-associated antigens. FLV and FLV gp71 antigens were able to absorb all cytotoxic activity of monkey and chimpanzee antisera to human myeloid leukemia antigens when these antisera were tested with CGL cells. These two approaches to an analysis of cross-reactivity indicate that the antigenic determinant(s) detected by the cytotoxic reactions of the FLV gp71 antiserum with human CGL cells is different from the determinant on FLV gp71 which is responsible for the inhibition of the reactivity of simian antisera with CGL cells. Since the goat and rabbit antisera to FLV and FLV gp71 are able to distinguish AML from CGL cells by direct cytotoxicity testing and absorption, they may be valuable reagents for the serological diagnosis of myeloid leukemia. In addition, since peripheral blood cells from AML and CGL patients in clinical remission were seronegative, the antisera may be valuable as management aids. The data in this report indicates that whatever the mechanism of leukemogenesis is in man, cells from CGL and AML patients possess certain membrane antigens which cross-react with FLV structural components such as p30 and gp71.

Human Ia beta chains and the invariant chain share a common antigenic determinant.

We have presented evidence that a mouse monoclonal antibody 1,227, which was previously shown to recognize a determinant on the light chains beta 1 and beta 2 of human Ia antigens, also recognizes a determinant on the invariant chain (1) associated with these molecules. Ia-specific proteins were resolved by two-dimensional (2D) PAGE and electrophoretically transferred onto nitrocellulose filters. The presence of a determinant shared between beta 1, beta 2 and I was established using "Western blotting" technique. In addition, we demonstrated that 1,227 can immunoprecipitate isolated beta 1, beta 2, and the invariant chain proteins following their separate elution from SDS-PAGE.

Distribution of common acute lymphoblastic leukemia antigen in nonhematopoietic tissues.

The common acute lymphoblastic leukemia antigen (CALLA), as defined by J-5 murine monoclonal antibodies, was detected on renal tubular and glomerular cells from fetal and adult donors by an indirect immunoperoxidase technique. CALLA could also be detected on epithelial cells of the fetal small intestine and on myoepithelial cells of adult breast but not on myoepithelial cells of the salivary gland. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis of immunoprecipitated 125I-labeled membrane antigens from dissociated renal cells demonstrated that the antigen migrated as a 90,000 mol wt antigen rather than the 98,000-100,000 mol wt antigen noted on CALLA-positive tissue culture cell lines. The data suggest that the determinant defined by the J-5 monoclonal antibody is neither a lymphoid cell-specific differentiation antigen nor a leukemia-specific antigen.

Monoclonal antibody against human T cell leukemia virus p19 defines a human thymic epithelial antigen acquired during ontogeny.

Using monoclonal antibody 12/1-2 against a 19,000-dalton human T cell leukemia virus (HTLV) protein (anti-p19), previously demonstrated to be reactive with HTLV-infected human cells, but not in numerous other uninfected cells, we found a reactive antigen to be expressed on the neuroendocrine component of human thymic epithelial cells but not on any other normal epithelial or neuroendocrine human tissues. Moreover, this reactive antigen is acquired on neuroendocrine thymic epithelium during thymic ontogeny--first appearing on fetal thymic epithelial cells between 8 and 15 wk gestation. While only a portion of thymic epithelial cells in the subcapsular cortical region of 15- and 24-wk fetal thymuses contained anti-p19+ epithelial cells, the entire subcapsular cortical region of newborn thymus epithelium was anti-p19+. By age 3 yr, normal subjects' entire subcapsular cortical and medullary thymic epithelium was anti-p19+. Using antibody against HTLV core protein, p24, and c-DNA probes for HTLV DNA, neither HTLV-specific p24 protein nor proviral DNA could be demonstrated in anti-p19+ thymic epithelial tissue. However, thymic epithelial extracts, disrupted HTLV extracts, as well as purified HTLV p19 antigen all inhibited the binding of anti-p19 antibody to thymic epithelium. Thus, anti-p19 may recognize a determinant on an HTLV-encoded 19,000-dalton structural protein that is shared by human thymic epithelium. Alternatively, anti-p19 defines a host encoded protein that is selectively expressed by normal thymic epithelium, and is induced to be expressed in HTLV-infected malignant T cells.

Antigens specific for human lymphocytic and myeloid leukemia cells: detection by nonhuman primate antiserums.

Primate antiserums to human leukemia cells can detect antigens specific for lymphocytic leukemia cells or antigens present on certain myeloid leukemia cells. The antigen specific for lymphocytic leukemia cells is destroyed by treatment with neuraminidase or trypsin. Tryptic digests of lymphocytic leukemia cells contain the antigen, which has a high molecular weight.

Membrane-bound immunoglobulins on human leukemic cells. Evidence for humoral immune responses of patients to leukemia-associated antigens.

Immunoglobulins were detected on the membranes of human leukemic cells by a microcytotoxicity technique. A significant percentage of lymphocytes from normal donors failed to react with goat antisera to human heavy chain determinants or to lambda-light chains. Lymphocytes from some normal donors, however, did react with antisera to k-light chains. A high percentage (50-90) of cells from some leukemia patients were killed by antisera to light chains and by one or more antisera to heavy chain determinants. Trypsin treatment of leukemic cells resulted in a loss of cytotoxic activity with all immunoglobulin antisera. Reactivity with the k-light chain antiserum was detectable 2 h after trypsinization of chronic myeloid leukemic (CML) cells and 8 h after treatment of acute lymphocytic leukemic (ALL) cells. Reactivity with the antisera to heavy chain determinants and lambda-light chains could not be detected 8 and 48 h after trypsinization of CML and ALL cells, respectively. The cytotoxic activity of the immunoglobulin antisera to heavy chains was abolished by absorption with the specific immunoglobulin used to define the antisera by precipitation. Eluates (pH 3.2) prepared from leukemic cells which reacted by cytotoxicity with the immunoglobulin antisera were shown to contain immunoglobulins of different heavy chain classes. In addition, some of the eluates had cytotoxic antibody activity to human leukemia cells. The specificity of the eluted antibodies is similar to the specificity previously described for cytophilic antibodies from leukemic patients and nonhuman primate antisera to human leukemia cells. The possible in vitro detection and in vivo significance of the eluted non-complement-fixing antibodies is considered.

Detection and partial characterization of human thymus-leukemia antigens.

Two monkey antisera against human thymocytes after absorption with human erythrocytes and peripheral blood leukocytes were shown to detect human thymus-leukemia (HTL)-like antigens. These sera were cytotoxic for thymocytes (> 90% lysis at a 1:10 dilution) but were nonreactive with enriched peripheral blood T- and B-lymphocytes or with cells from myeloid or B-cell lymphoid leukemias. Most (16/17) sheep erythrocyte rosette-forming acute lymphoblastic leukemia (ALL) cells reacted with these sera. Cells from patients with T-cell chronic lymphocytic leukemia, lymphoblastic lymphoma (LBL), and thymoma were also positive. Three of 4 T-cell lymphoblastoid lines derived from ALL patients reacted with these sera. Absorption of the sera with MOLT-4F cells, thymocytes, or LBL cells removed the reactivity against all types of cells tested. However, sera absorbed with the T-cell line HSB remained cytotoxic for thymocytes, MOLT-4F, and most (6/9) T-cell cancers tested. The peripheral blood cell-absorbed sera precipitated a molecule with an apparent molecular weight of 48,000 from lactoperoxidase-labeled thymocytes but not from similarly labeled peripheral blood lymphocytes. The ability of the sera to precipitate this antigen was decreased by absorption with thymocytes, MOLT-4, or LBL cells but not by absorption with HSB, SB, or non-T, non-B ALL cells. Sequential precipitation studies suggested that the HTL antigen was not associated with beta 2 microglobulin.

Monoclonal antibodies against human pancreatic adenocarcinoma: distribution of DU-PAN-2 antigen on glandular epithelia and adenocarcinomas.

This report describes the distribution of antigen DU-PAN-2, defined by a monoclonal antibody raised against human pancreatic carcinoma cells, on a variety of tumors and nonneoplastic tissues. With the use of an immunoperoxidase technique, the antigen was detected on 16 of 16 pancreatic carcinomas, on 5 of 5 gallbladder or bile duct carcinomas, on the great majority of stomach adenocarcinomas, and, less commonly, on adenocarcinomas of other primary sites. Substantial intratumor heterogeneity of antigen expression was noted. DU-PAN-2 antigen was present on many types of normal glandular epithelial cells but often was more weakly expressed than on the corresponding tumors. The immunomorphology of staining, coupled with biochemical information known about the antigen, supports the notion that the DU-PAN-2 antigen is a mucin-like substance. Its relative restriction of expression on different types of glandular epithelium suggests that DU-PAN-2 antibody might be a useful reagent for helping to determine the site of origin of adenocarcinomas.

Childhood lymphoblastic cancer: subgroups defined by nonhuman primate antisera to human lymphatic leukemia cell antigens.

Lymphoblasts from 23 children with acute lymphocytic leukemia (ALL) and 10 with lymphoblastic lymphoma (LBL) were studied by complement-dependent microcytoxicity tests with two nonhuman primate antisera defining leukemia-associated and lymphoma-associated antigens. Cells form 15 patients with ALL and 1 with LBL reacted only with antiserum to chronic lymphatic leukemia (CLL). These group-I patients were predominantly female. Most were pancytopenic and lacked mediastinal widening and T-cell markers; lymphoblasts from 15 were periodic acid-Schiff-positive. Cells from 8 male patients reacted only with antiserum to converted lymphosarcoma (LS). All these group-II patients expressed T-cell markers; 5 had mediastinal enlargement and 2, an abdominal mass. Six of the 8 were PAS-negative. Cells from 9 patients reacted with both antisera. The group-III patients demonstrated some characteristics of each of the above groups. Patients whose lymphoblasts reacted with CLL antiserum presented with clinical and laboratory features indicative of a good prognosis, i.e., ALL with PAS positivity and no T-cell markers or localized mass. Patients whose cells reacted with LS antiserum often had bad prognostic features: mediastinal or abdominal mass, expression of T-cell markers, and PAS negativity. These antisera appear able to differentiate childhood ALL from LBL. The distinction is important prognostically and perhaps therapeutically.

Human acute myelogenous leukemia antigens defined by simian antisera: evidence for leukemia-associated antigens distinct from immune response-associated alloantigens.

Leukemic blasts from a patient with acute myelogenous leukemia (AML) and peripheral blood T- and B-lymphocyte subpopulations from his genetically identical normal twin were analyzed with the use of the simian antiserum-defining AML antigens and a rabbit antiserum to immune response-associated (la)-like antigens. Blast cells from the patient consistently reacted with both reagents, whereas the B-lymphocyte populations from the patient's normal identical twin reacted only with the rabbit anti-la serum and in no instances reacted with the antiserum to AML cell antigens. Blast cells from the AML patient significantly stimulated the lymphocytes of his normal twin and his own remission leukocytes, whereas the cells from the normal twin failed to stimulate the cells of the patient. These results suggested the existence on AML cells of tumor-associated antigens that are distinct from various other well-characterized normal human alloantigens and differentiation antigens including B-cell antigens. Changes were reported in the expression of leukemia-associated antigens and Ia-like antigens on the cells of an AML patient undergoing chemotherapy as well as in the ability of the simian antisera to distinguish antigens specific for myeloid leukemias from lymphocytic types of leukemias.

Carcinoembryonic antigen in nonhuman primates.

Blood levels of carcinoembryonic antigen (CEA) have been measured in several nonhuman primate species. Only gorillas and chimpanzees were found to have significant elevations of CEA-like activity in their blood compared to the normal values of less than 2.5 ng/ml in humans. The average CEA level in 134 chimpanzees was 25.2 ng/ml (range, 4.2--95 ng/ml) and in 13 gorillas it was 32 ng/ml (range, 12.4--61.9 ng/ml). These levels were not related to sex. Blood levels repeatedly taken over a 1 1/2-year period remained relatively stable in both species. Analysis of parallelism of immunologic reactivity showed chimpanzee CEA to be similar to but not identical with human CEA. The molecular size of chimpanzee CEA was also similar to that of human CEA.

Detection of both T-cell and Ia-like antigens on cells from patients with acute myelomonocytic leukemia and chronic myelogenous leukemia in blast crisis.

Appropriately absorbed antisera to the lymphoblastoid cell lines HSB and SB detect a human T-lymphocyte-associated antigen (TLAA) and the human Ia-like antigens, respectively. Cells from some patients with acute myelomonocytic leukemia (AMML) and chronic myelogenous leukemia in blast crisis expressed both TLAA and Ia antigens when tested in a complement-dependent microcytotoxicity assay (greater than 90% lysis with both antisera). When patients were in remission, expression of TLAA and Ia antigens returned to normal values. Quantitative absorption of anti-TLAA serum with increasing numbers of AMML cells showed that these cells could remove reactivity of the serum for both HSB and human thymocytes. Similarly, absorption of anti-Ia serum with AMML cells removed all serological reactivity when this serum was tested on chronic lymphocytic leukemia cells or normal B-cells. These serological findings were confirmed by immunoprecipitation and sodium dodecyl sulfate-polyacrylamide gel electrophoresis studies using radiolabeled antigens. Cells from an AMML patient were labeled with 125I using lactoperoxidase; both the TLAA and Ia antigens were precipitated from the resulting solubilized membrane preparation. Leukemic cells from one AMML patient and one patient with chronic myelogenous leukemia in blast crisis were studied for Ia and TLAA antigens with a double fluorescence technique. Over 80% of the cells showed dual fluorescence.

Radiolocalization of human pancreatic tumors in athymic mice by monoclonal antibody DU-PAN 1.

Monoclonal antibodies that selectively bind to pancreatic tumors may be useful in the therapy and diagnosis of pancreatic carcinoma. In this study we have examined the tumor localization of radioiodinated DU-PAN 1, a mouse monoclonal antibody that is selective for a human pancreatic cancer-associated antigen. After radiolabeling, both DU-PAN 1 intact monoclonal antibody and F(ab')2 fragments retained immunoreactivity and showed high affinity for the pancreatic tumor cell line CA13 in vitro. Paired-label biodistribution studies in nude mice bearing CA13 s.c. xenografts were performed. Mice received both 131I-labeled DU-PAN 1 immunoglobulin G2a or F(ab')2 fragment and 125I-labeled mouse myeloma immunoglobulin G2a or F(ab')2 fragment. Tumor uptake for 5-micrograms doses of DU-PAN 1 immunoglobulin ranged from 4.8 to 11.83% injected dose/g. Tumor uptake values for mice given 5-micrograms doses of DU-PAN 1 F(ab')2 ranged from 3.9 to 6.9% injected dose/g. Tumor uptakes of the respective myeloma controls were lower in all cases when compared with the DU-PAN 1 preparations. Tumor localization indices for 5-micrograms doses of DU-PAN 1 immunoglobulin were 3.0 and 24 h and 2.9 at 48 h. For 5-micrograms doses of DU-PAN 1 F(ab')2, tumor localization indices were 29.9 at 24 h and 90.0 at 48 h. In most cases, tumor:normal tissue ratios were greater than 3 at all time points, indicative of tumor selectivity for both DU-PAN 1 preparations, but the ratios were considerably higher using the DU-PAN 1 F(ab')2. The F(ab')2 fragment thus displays better tumor localization characteristics when compared with the intact immunoglobulin. Protein doses of DU-PAN 1 F(ab')2 of between 5 and 10 micrograms gave the best localization, although protein doses of up to 100 micrograms could be administered before apparent tumor saturation was seen.

Polypeptide core of a human pancreatic tumor mucin antigen.

This work describes the molecular properties of the polypeptide core of a human pancreatic mucin antigen isolated from a human pancreatic adenocarcinoma cell line, HPAF. Pancreatic tumor mucin was isolated by a combination of molecular sieve chromatography and CsCl/4 M guanidine-HCl density gradient ultracentrifugation. Trifluoromethane sulfonic acid was used to remove carbohydrate units from purified mucin molecules. A rabbit monospecific polyclonal antibody was generated against pancreatic apomucin and reacted with a Mr greater than 200,000 species. The antibody binding data indicated that the rabbit antiserum raised against pancreatic apomucin cross-reacted with a breast mucin synthetic peptide. Northern blot and immunodot blot analyses of various cell line extracts revealed that a tandem repeat sequence and a similar mRNA were detected in both pancreatic and breast mucin-producing cell lines. These results suggest that pancreatic apomucin and breast apomucin share some similarity in tandem repeated nucleic acid and protein sequences.

Relationship between adenosine deaminase and a human thymus leukemia antigen isolated from MOLT 4 cells.

The relationship between adenosine deaminase (ADA) and a human membrane thymus leukemia (HTL) antigen-detected by an antithymocyte serum was explored. A freeze-thaw extract of the T-cell-derived cell line, MOLT 4, was applied to an immunoabsorbant column of a rabbit antiserum to calf ADA. The bound MOLT 4 ADA was eluted with 6 M urea. The recovered ADA had a specific activity of 490 mumol of adenosine deaminated per min per mg protein, and the yield was 32%. No HTL antigenic activity was detected in the purified ADA. In addition, no ADA activity was detected in the unbound fraction containing the HTL antigenic activity, supporting the conclusion that ADA and HTL antigen are independent molecules. Affinity-purified anti-calf ADA was not cytotoxic for several HTL antigen-positive cells, including thymocytes, MOLT 4, and thymus-derived acute leukemia lymphoblasts.

Isolation and properties of a human pancreatic adenocarcinoma-associated antigen, DU-PAN-2.

This work describes the molecular properties of a human pancreatic adenocarcinoma-associated mucin-like antigen defined by a murine monoclonal antibody (DU-PAN-2). DU-PAN-2 antigen is a large molecule, readily detected in the body fluids of patients with pancreatic adenocarcinoma and sensitive to neuraminidase treatment and alkaline reduction. The antigen binding activity of the DU-PAN-2 immunoglobulin M antibody is markedly reduced when coupled to an insoluble matrix. Therefore, the antigen was partially purified from the serum and ascites of patients with pancreatic adenocarcinoma by ammonium sulfate fractionation and Affi-Gel-Blue chromatography to remove most of the serum proteins. Noncovalently associated proteins were further separated on CsCl/CsBr density gradients and noncovalently associated lipids removed by organic solvent extraction. DU-PAN-2 antigen was monitored by a solid-phase competition radioimmunoassay. We have been able to detect antigen reactivity and analyze its electrophoretic pattern following transfer from 1% agarose gel onto nitrocellulose paper and immunoblotting with DU-PAN-2 antibody. Antigen was also labeled metabolically with various radioactive monosaccharides and sulfates. Radioimmunoprecipitation of labeled antigen with DU-PAN-2 antibody showed two distinct broad antigen bands consistent with the immunoblotting signals. The heavily glycosylated and polydisperse nature of this antigen and the results of various enzyme treatments suggest that the DU-PAN-2 epitope is expressed on a mucin-like molecule.

A novel ribonucleoprotein complex defined by monoclonal antibodies to NIH 3T3 cells transfected with human pancreatic adenocarcinoma DNA.

Three monoclonal antibodies elicited to NIH 3T3 cells transfected with DNA from a human pancreatic adenocarcinoma cell line recognized a novel ribonucleoprotein complex. Minimally, this ribonucleoprotein complex contained a Mr 240,000 protein (by sodium dodecyl sulfate-polyacrylamide gel electrophoresis) and two RNA species with apparent sizes of 1.5 and 3.0 kilobases (by formaldehyde agarose gel electrophoresis). In addition to a cytoplasmic and nuclear subcellular localization, the RNA antigen was secreted from human tumor cell lines and NIH 3T3 cells transfected with pancreatic tumor DNA (inhibitable by monensin) and was apparently not a viral or Mycoplasma contaminant. The ribonucleoprotein antigen was detected in some normal tissues by immunoperoxidase but was not found in or secreted from in vitro cultured normal human fibroblasts, nontransfected or spontaneously transformed NIH 3T3 cells, or normal peripheral blood leukocytes.

Antigens expressed on NIH 3T3 cells following transformation with DNA from human pancreatic tumor.

This report documents that the pancreatic adenocarcinoma cell line, HPAF, contains oncogene activity detected by transformation of NIH 3T3 cells through transfection with HPAF DNA. The HPAF transfected NIH 3T3 cells do not contain oncogenes homologous with c-H-ras, c-K-ras, c-N-ras, v-fms, c-myb, c-sis, v-fgr, c-mos, c-myc, c-fos, v-fes, v-src, v-erb A, v-erb B, c-N-myc, v-raf, or v-abl, other than the endogenous mouse genes. The transfectants do express proteins detected by two-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis which were not found in nontransfected NIH 3T3 cells. Monoclonal antibodies raised against the transfectants recognize proteins not found in untransfected NIH 3T3 cells that are antigenically identical to proteins found in the HPAF cells. These antigens are also detected on six other human pancreatic adenocarcinoma cell lines but show a much more restricted distribution on lymphoblastoid, melanoma, prostatic carcinoma, and normal skin fibroblast cell lines.

Humoral immunity against a tandem repeat epitope of human mucin MUC-1 in sera from breast, pancreatic, and colon cancer patients.

Using synthetic peptides 60,80, and 105 residues long, corresponding to 3, 4, and 5.25 tandem repeats of human mucin MUC-1 protein core, as antigens in a solid-phase enzyme-linked immunosorbent assay, we screened sera from 24 breast cancer patients, 10 colon cancer patients, and 12 pancreatic cancer patients, at various stages of disease, for the presence of mucin-specific antibodies. The 105-residue peptide was superior in allowing detection of high levels of anti-mucin antibodies in 10.9% of sera in each cancer group. Another 4.3% showed intermediate reactivity. Lower levels of detection were achieved with the 80-residue peptide, and no specific reactivity was detectable with the 60-residue peptide. Anti-mucin antibodies were previously undetectable when this assay was performed with purified whole mucin or short synthetic peptides. The presence or absence of antibody did not correlate with the levels of circulating mucin or stage of disease. One highly reactive serum sample was used to identify more precisely the epitope on the long synthetic peptide to which the reactivity was directed. The reactivity of this serum specific for the 105-residue peptide was blocked by a 9-residue peptide from the NH2-terminal region of the 20-residue tandem repeat containing the previously identified immunogenic epitope APDTRP. Another 9-residue mucin peptide, from the COOH-terminal region of the tandem repeat which does not contain the APDTRP epitope, had no effect. All the mucin-specific reactivity was found to be of the IgM isotype, indicating a helper T-cell-independent response, unusual for an antibody against a peptide epitope, but not unexpected for tandemly repeated epitopes.

Classification of human leukemia by membrane antigen analysis with xenoantisera.

Rabbit and monkey antisera after appropriate absorption were rendered specific for normal or leukemic lymphoid- and myeloid-associated antigens. Antisera defining a common peripheral blood T-cell antigen, a thymus leukemia antigen, HLA-DR or Ia-like antigen, common acute lymphoblastic leukemia antigen (CALLA), and a myeloid-monocyte (M) antigen were used in a microcytotoxicity assay to classify leukemic cells from 30 patients in a double blind study. The antisera to the M antigen reacted with adherent peripheral blood cells and polymorphonuclear leukocytes and failed to react with nonadherent mononuclear cells and enriched T-cells and chronic lymphocytic leukemia cells. The M antisera also reacted with U937, a monocytic-type cell line, and with HL60, a promyelocytic-type cell line, but failed to react with T and B lymphoblastoid cell lines. The specificities of the other antisera have been described in previous reports. Cells from three of the patients could not be phenotyped by microcytotoxicity testing. Cells from 25 patients had a consensus morphological or histochemical diagnosis of either acute lymphoblastic leukemia or acute nonlymphocytic leukemia. The serological classification of these patients using the five types of antisera listed above were consistent with the consensus diagnosis. In addition, the lymphoid cancers were further subclassified as to T-, B-, or thymus antigen types. There was no consensus lymphoid versus myeloid diagnosis on cells from two patient. The serological classification in both cases favored a diagnosis of myeloid rather than lymphoid leukemia.