Initial clinical evaluation of radiolabeled MX-DTPA humanized BrE-3 antibody in patients with advanced breast cancer.

To evaluate radiometal-labeled humanized BrE-3 (huBrE-3) monoclonal antibody as a radioimmunolocalization and therapeutic agent in breast cancer patients, tumor localization, pharmacokinetics, radiation dosimetry, and immunogenicity of (111)In-labeled combined 1-p-isothiocyanatobenzyl 3-methyl- and 1-p-isothiocyanatobenzyl 4-methyldiethylenetriamine pentaacetic acid (MX-DTPA) huBrE-3 were studied. Seven women with BrE-3 antigen-positive, metastatic breast carcinoma underwent (111)In huBrE-3 infusion (5 mCi; 50 mg), followed by serial gamma camera imaging and plasma sampling. Region of interest analysis of images was used to make radiation absorbed dose estimates for (111)In huBrE-3. Data were extrapolated to 90Y huBrE-3. Human anti-human antibody (HAHA) response was measured in serum samples obtained up to 3 months after infusion. Patients tolerated infusions well. Seventy-six percent of 105 known sites of disease were identified on planar and single-photon emission computed tomography scans. For six of seven patients, a biexponential model fit the plasma time-activity curve best with an average T1/2alpha=10.6+/-8.5 (SD) h and average T1/2beta=114.2+/-39.2 h. Radiation absorbed dose estimates for (111)In huBrE-3 for whole body averaged 0.53+/-.08 rads/mCi. Dose estimates for 90Y huBrE-3 for marrow averaged 8.4+/-11.9 rads/mCi, and for tumors, 70+/-31.5 rads/mCi. Liver radioactivity uptake averaged 19.7+/-8.8% injected dose at 24 h after infusion, translating into an average radiation absorbed dose 21.1+/-12 rads/90Y mCi administered. Only one of seven patients demonstrated a low level of HAHA response. Although the plasma half-lives are longer and marrow dose higher for radiolabeled huBrE-3 compared with the murine construct, the excellent tumor localization, good tumor dosimetry, and low immunogenicity support the use of 90Y-huBrE-3 antibody for radioimmunotherapy of breast cancer.

Conjugation of interferon alpha to a humanized monoclonal antibody (HuBrE-3vl) enhances the selective localization and antitumor effects of interferon in breast cancer xenografts.

Human mammary carcinoma xenografts (MCF-7) growing in nude mice were treated with natural interferon alpha (n-IFN-alpha) alone or conjugated to a humanized monoclonal antibody (MoAb) anti-breast mucin (HuBrE-3vl) or to irrelevant human IgG1kappa. The IFN and the conjugates were administered as 20 intra-lesional (i.l.) injections to 1 of 2 xenografts in each mouse, or i.p. The growth inhibitory effects of HuBrE-3vl/nIFN-alpha were significantly greater than those of nIFN-alpha used as a single agent or conjugated to HuIgG1kappa. These effects occurred locally in the tumors receiving i.l. injections and systemically, although to a slightly lesser extent, in the noninjected tumors of mice treated i.l. and in the xenografts of mice treated i.p. Biodistribution studies showed that the uptake of 125I-HuBrE-3vl/nIFN-alpha by the tumors 24 hours after i.l. or s.c. injection was greater than that of 125I-HuIgG1kappa/nIFN-alpha, 125I-nIFN-alpha alone, or by normal tissues, documenting a tumor targeting effect and favorable tumor:normal tissues (T:NT) ratios. The targeting effects and the resulting tumor growth inhibition were favored by the IFN-mediated up-regulation of the HuBrE-3vl reactive antigen, which was more prominent after 3 weeks of treatment with HuBrE-3vl/nIFN-alpha. These results were superior to those we obtained previously with nIFN-alpha conjugated to another MoAb of the same group (Mc5). These studies point out the potential usefulness of HuBrE-3vl/nIFN-alpha for the local and systemic treatment of breast cancer lesions by providing a means of delivering high doses of IFN to the tumors while minimizing the amount of IFN binding to normal tissues.

Biodistribution of anti-breast mucin HuBrE3-derived inverted Fabs (IFabs) in nude mice carrying MX-1 xenografts.

In radioimmunotherapy, the long circulation times of antibody radioconjugates correlate with high relative radiation doses to nontumor tissues. Tumor/normal tissue ratios can be significantly improved by using targeting molecules with shorter circulation times. IFabs are multimers of VH-CH1-linker-VK-CK monomers. The lack of the Fc region in IFabs should lead to circulation times that are shorter than those of IgG molecules. The monomers assemble into disulfide-bond-stabilized multimers, 90% of which are 100 kDa dimers (IFab2). IFab2s should not be rapidly eliminated through kidney filtration because their molecular weight is above the threshold for renal passage. We report the first experimental in vivo tests for 125I-IFab radioconjugates derived from a humanized version of the anti-breast mucin monoclonal antibody BrE-3. Biodistributions are reported for athymic nude mice carrying human mammary tumor MX-1 xenografts. The T1/2 beta's for the different tissues ranged from 13.3 h for blood to 19.9 h for tumor. Therefore, IFab radioconjugates cleared the body with a rate comparable to that of F(ab')2 fragments. Except for stomach, tumor/nontumor dose ratios were significantly better for IFabs than for the parent antibody (BrE-3)4 days after injection.

Effect of multiple, repeated doses of radioimmunotherapy on target antigen expression (breast MUC-1 mucin) in breast carcinomas.

The effect of radioimmunotherapy (RIT) on target antigen expression was studied in breast carcinomas transplanted in immunodeficient mice. In nine separate experiments, a single dose of 1500 microCi of 131I-labeled monoclonal antibody (MAb) Mc5 was given to groups of mice carrying well-established, vascularized, transplantable breast tumors (MX-1). Mc5 recognizes an epitope on the tandem repeat of the breast epithelial MUC-1 mucin. This dose suppressed tumor growth for at least 20 days, after which the tumors began to regrow. At various times thereafter, tumors were removed and analyzed for target antigen expression by flow cytometry and immunohistochemistry. In no case was there any significant decrease in antigen content/cell in the tumors of treated mice compared to tumors in control untreated mice. Similar results were obtained with four other breast carcinomas (MCF-7, MDA-MB-331, MDA-MB-435, and MX-2A). To assess the effect of repeated RIT doses on target antigen expression, groups of mice with MX-1 tumors were given 2, 3, and 4 consecutive doses of 1200 microCi of 131I-labeled Mc5. One mouse each at 2, 3, and 4 doses (3 of 18) was cured of its tumor. Control mice were sacrificed after 50 days due to the excessive size of their tumors. Tumors from four mice from each group (2, 3, and 4 doses), after they began to regrow, were excised and analyzed for mucin content and compared to tumors from untreated mice with similar-size tumors transplanted at later dates. In none of the treated groups was there any decrease in mucin content. These results demonstrate that RIT with an anti-breast mucin MAb does not result in the appearance of antigen-negative tumor cells, thus indicating that repeated fractionated doses, which will most likely be necessary for an eventual cure of breast cancer with MAb therapy, are possible.

Biological activity of two humanized antibodies against two different breast cancer antigens and comparison to their original murine forms.

We have humanized two monoclonal antibodies (MoAbs), hu-BrE-3 and hu-Mc3, that are bound to two different antigens of the breast epithelial cell. They bind to the breast epithelial mucin (M(r) 400,000) and the BA46 antigen (M(r) 46,000). They could participate in a joint radioimmunotherapy strategy administering repeated or fractionated dosages, where increased irradiation could be delivered by their simultaneous administration. Both antibodies, hu-BrE-3 and hu-Mc3, had similar reactivity to their antigens and similar binding affinity as those of their original murine forms. However, because humanized MoAbs could have different pharmacokinetic and radioimmunotherapeutic characteristics than their original murine forms, the experimental biodistribution in vivo of both of these two humanized anti-breast tumor MoAbs was compared to their original murine forms. Biodistributions in immunodeficient mice grafted with transplantable human breast tumors, both after radioiodination and 111In labeling via 1-p-isothiocyanatobenzyl-methyl-diethylene-triaminepenta-ace tic acid (MXDTPA), demonstrated comparable tumor:normal tissue ratios for the humanized and murine forms. In radioimmunotherapy, the humanized forms for both MoAbs showed also similar tumoricidal activity as that of the original murine MoAbs. These results show that the new humanized forms are amenable to conjugation and radioisotope labeling without loss of biological activity. Furthermore, they demonstrate that these engineered molecules kept intact, both qualitatively and quantitatively, their binding ability, pharmacokinetics, and radioimmunotherapeutic characteristics after the humanization process.

Anti-BA46 monoclonal antibody Mc3: humanization using a novel positional consensus and in vivo and in vitro characterization.

Mc3 is a murine mAb that is highly effective in treating breast tumors in experimental radioimmunotherapy. Mc3 binds to BA46, a 46-kDa glycoprotein of the human milk fat globule membrane that is also expressed in breast carcinoma cells. We cloned and sequenced cDNAs encoding the variable regions of Mc3 and constructed an IgG1, kappa human/mouse chimeric antibody. We then humanized the variable regions of Mc3 using a positional consensus method and retaining residues that might either contact the complementarity-determining regions or the opposite chain. This positional consensus is novel in that it does not include residues with buried side chains. Humanized Mc3 retained full binding affinity, and fully competes with murine Mc3 for antigen binding. Humanized and murine 131I-labeled Mc3 behaved identically in athymic nu/nu mice biodistribution studies. The tumor uptake levels for both antibodies increased over a period of 4 days within a range of 13-20% of the injected dose per g with extremely favorable tumor:normal ratios. Also, a single therapeutic dose of 131I-labeled humanized Mc3 in the same animal model reduced the average tumor size and produced one of five cures while in the uninjected control tumor growth continued unabated. We believe that these results justify the implementation of Phase I human clinical trials for imaging and radioimmunotherapy of breast cancer.

Humanization of KC4G3, an anti-human carcinoma antibody.

We have previously constructed a chimeric version of KC4G3, a murine antibody that reacts with several human epithelial cancers and binds to the human breast epithelial mucin. We have now successfully humanized KC4G3 using positional consensus data, previously compiled after examining several other antibody structures, listing residues in the VH and V kappa frameworks that could influence antigen binding. We have previously showed that a fraction of the kappa chains of murine and chimeric KC4G3 migrates abnormally on SDS-PAGE most likely due to N-linked glycosylation in V kappa. The glycosylation signal has now been removed from V kappa, as a consequence of humanization. As expected, the humanized kappa chain migrates normally on SDS-PAGE. We detected no significant differences either in the affinities (1.6 x 10(9) M-1 vs. 1.4 x 10(9) M-1, respectively) or in the ability to compete for antigen binding, between the murine and the humanized antibodies. The humanized version is an IgG1, kappa immunoglobulin produced by mouse myeloma SP2/0-Ag14 cells and is designated HuKC4v2. The HuKC4v2 frameworks conform to the V kappa II and VHIII human consensus in all but six positions in V kappa and three positions in VH.

Cloning of cDNAs encoding the variable domains of antibody KC4G3 and construction of a chimeric antibody.

We have cloned and sequenced cDNAs encoding the variable regions of the light (VL) and heavy (VH) chains of monoclonal antibody KC4G3. VL belongs to group II and resulted from a V kappa-J kappa 2 recombination. VH belongs to group IIId and arose from a V-D9-JH3 recombination. The VL and VH frameworks are respectively 84% and 83% identical to the corresponding VL and VH human consensus frameworks. The deduced amino acid sequence of VL contains an asparagine-linked glycosylation site in framework 3 (N74 I75 S76). We have determined that a large fraction of the light chains are indeed glycolysated. We constructed an IgG1, kappa human/mouse chimeric antibody (by inserting the murine KC4G3 Fv-encoding cDNAs into plasmids encoding a human IgG1, kappa Fc domain) and expressed it in SP2/0-Ag14 mouse myeloma cells. This chimeric monoclonal antibody is designated ChiKC4. We have determined that the murine monoclonal antibody KC4G3 binds the human breast mucin with an affinity constant of 1.1 x 10(9) M-1. ChiKC4 binds the same antigen with an affinity constant of 1.2 x 10(9) M-1. ChiKC4 binds the carcinoma tissue sections by the ABC immunoperoxidase method in an identical manner as does KC4G3.

Cloning of cDNAs encoding the variable domains of antibody BrE-3 and construction of a chimeric antibody.

BrE-3 is an IgG1,kappa murine monoclonal antibody that binds to human breast epithelial mucin and that has been shown to be promising for imaging and treatment of breast cancer. We have cloned and sequenced cDNAs encoding the variable regions of the light (VL) and heavy (VH) chains of BrE-3. VL belongs to group II and resulted from a V kappa-J kappa 1 fusion. VH belongs to group IIIc and arose from a V-D-JH3 non-conservative fusion which left little or nothing of the original D minigene. Thus, the third VH CDR contains only 4 amino-acids. We constructed an IgG1,kappa human/mouse chimeric antibody (by joining the murine variable domains to human constant domains) and expressed it in SP2/0 myeloma cells. This chimeric monoclonal antibody stains breast carcinoma tissue sections by the ABC immunoperoxidase method. Its affinity for the BrE-3 antigen is 2.68 x 10(8) M-1, which, considering the experimental error, is indistinguishable from the affinity of the original murine antibody (3.75 x 10(8) M-1). The VL and VH domains alone are respectively 73%, and 63% identical to the human V kappa II and VHIII consensus sequences. If the CDRs are excluded, these numbers become respectively 82% and 80%. Therefore, we expect the reported chimeric BrE-3 to be considerably less immunogenic to humans than the original murine antibody, while retaining the original binding properties.

The use of natural interferon alpha conjugated to a monoclonal antibody anti mammary epithelial mucin (Mc5) for the treatment of human breast cancer xenografts.

An immunoconjugate composed of natural interferon alpha (nIFN alpha) bound in a noncleavable fashion to a monoclonal antibody (MoAb) recognizing a breast epithelial membrane mucin (Mc5) was used to to treat xenografts of a human mammary carcinoma cell line (MCF-7) growing in nude mice. The immunoconjugate (nIFN alpha/Mc5) was administered as 20 intralesional (i.l.) injections to 1 of 2 xenografts in each animal. It was found that nIFN alpha/Mc5 produced a significant enhancement of the growth inhibitory actions of nIFN alpha on the injected tumors. Further enhancement was obtained when nIFN gamma or nIFN gamma together with Mc5 (at a dose 10 times larger than that present in nIFN alpha/Mc5) were added to the immunoconjugate. Biodistribution experiments showed that the uptake of 125I-nIFN alpha/Mc5 by the tumors was greater and its elimination slower than for 125I-nIFN alpha alone or conjugated to irrelevant mouse IgG1. In addition, the immunoconjugate up-regulated the antigenic expression of a breast epithelial membrane mucin by the carcinoma cells, an up-regulation which was not significantly different from that produced by nIFN alpha alone. The contralateral noninjected tumors exposed to systemic levels of the immunoconjugate showed an enhancement of antitumor effects, but to a lesser extent than the injected tumors. These findings suggest that the enhancement of the growth inhibitory action of the immunoconjugate was related to the specific binding of Mc5 which targeted the IFN to the carcinoma cells and impeded its elimination. It is likely that the targeting was favored by the IFN-mediated up-regulation of antigenic expression by the carcinoma cells, thereby producing a cascade of interrelated effects. The results of this study point out the feasibility and potential usefulness of IFN treatment by means of immunoconjugates as well as the worth of pursuing and improving this form of therapy.

5-Iododeoxyuridine increases the efficacy of the radioimmunotherapy of human tumors growing in nude mice.

Recently, there has been much interest in the use of radionuclide conjugated monoclonal antibodies for the treatment of human malignancies. One way to potentially maximize the therapeutic effectiveness of radioimmunotherapy would be to sensitize tumor cells to the radiation dose delivered by the antibody. Since radioimmunotherapy can potentially treat disseminated disease, including micrometastasis, we chose to study a halogenated pyrimidine radiosensitizer, a class of compounds that affect nonhypoxic cells. 5-Iododeoxyuridine, administered with pyrimidine metabolism modulators, increased the therapeutic effectiveness of radioimmunotherapy, resulting in individual cures of human tumors growing in BALB/c nu/nu (nude) mice. 5-Iododeoxyuridine was administered with N-(phosphonacetyl)-L-aspartic acid and 5-fluoro-deoxycytidine plus tetrahydrouridine. This drug treatment was combined with radioimmunotherapy using 131I conjugated to a monoclonal antibody, Mc5. Mc5 binds to a mucin component of the human milk fat globule. This antigen is expressed on the surface of MX-1 cells, the transplantable human tumor used in this study. Tumor-bearing mice treated with both the drug protocol and 131I-Mc5 (540 microCi, 10 microCi/micrograms) showed a regression in average tumor volume. The average tumor volume was reduced below the initial size at treatment for 50 days; two of five cures were obtained. Neither cures nor regressions were observed with either the drug or antibody treatments alone. Our results indicate the potential for increasing the therapeutic effectiveness of radioimmunotherapy of human solid tumors with halogenated pyrimidines.

Development and characterization of breast carcinoma cell lines as in vitro and in vivo models for breast cancer diagnosis and therapy.

To establish a model system for preclinical radioimmunotherapy studies, attempts were made to graft 16 different human breast carcinoma cell lines into BALB/c nu/nu(nude) mice. Nine produced serially transplantable tumors growing at a variable rate, whereas seven failed to do so. Conversely, three new cell lines were established in monolayer culture from transplantable human breast tumors in nude mice. Twelve selected tumors and their corresponding cell lines were characterized for DNA ploidy, % S-phase, and breast epithelial mucin expression by immunohistochemistry and flow cytometry. A wide diversity of these cellular characteristics were found in that each tumor was unique and distinct from the others. DNA ploidy differed among the tumors but was not affected by switching between in vitro to in vivo growth. Some tumors expressed similar levels of the breast mucin both in vitro and in vivo, whereas most expressed lower levels as transplantable tumors. There was a good correlation between immunohistochemical and flow cytometric determination of surface and cytoplasmic mucin expression, and with both techniques estrogen and progesterone receptor positive tumors had significantly higher levels of mucin expression than receptor negative tumors. These 12 transplantable breast tumors, with their corresponding cell lines, provide an excellent model system for testing radioimmunotherapy and other therapeutic reagents because they exhibit diverse phenotypic characteristics that represented a mini-population of breast cancer patients' tumors, allowing assessment of the effect of therapy when confronted with different breast tumors' genotype and phenotype.

A novel serum assay for breast epithelial antigen using a fusion protein.

Serum levels of breast epithelial antigens (BrE-Ags) are presently used in the follow-up of breast cancer patients. Available assays do not have optimal sensitivity and rely on reagents that could vary in their source and purity. A novel competitive solid-phase radioimmunoassay was developed for BrE-Ags that consists of the NP5 fusion protein, produced in Escherichia coli, that is composed of beta-galactosidase and polypeptide sequence obtained from a breast carcinoma cell line cDNA library, and anti-human milk fat globule monoclonal antibody Mc5. The fusion protein carries an altered epitope sequence (mimotope) that is similar, but not identical, to that found in the native antigen. This new competitive assay configuration has two essential features, a solid-phase affinity step that purifies the fusion protein carrying the mimotope for Mc5 and a competitive step that provides quantitation. Serum values for this assay show high specificity and sensitivity for breast cancer patients when compared to normal subjects and post-surgical breast cancer patients during their disease-free period.

Epitope expression on the breast epithelial mucin.

Multiple epitope expression on the breast epithelial mucin was explored using a panel of monoclonal antibodies (MoAbs) created against milk and breast tissue preparations, against blood group determinants, and against other non-breast epithelial mucins. Since the breast epithelial mucin is now used in both diagnostic and therapeutic modalities for breast cancer, and also because altered or incomplete glycosylation in varying degrees is expected in breast carcinoma tissue, the antigenic target used here was the native mucin and sequential stages of deglycosylation introduced to it by HF treatment. Partial deglycosylation increased exposure of core peptide amino acid sequences increasing MoAb binding generally, while it either decreased or occasionally increased binding of blood group oligosaccharides. Cross reactivity of MoAbs to other mucins was low with the breast epithelial mucin (BEM). The study of the affinity binding constants of some of the anti-BEM peptide MoAbs predicted carbohydrate participation in their epitope structure. The identification of different epitopes on the BEM, investigations on their possible epitopic structure, and the study of MoAb binding during different stages of glycosylation of the molecule leads to knowledge on the contribution of carbohydrates to their epitopes and strengthens the ability to understand their performance in their diverse possible applications in breast cancer diagnosis, prognosis, and therapy.

Evidence of extensive phospholipid fatty acid methylation during the assumed selective methylation of plasma free fatty acids by diazomethane.

We compared a conventional method (Method I) for measuring plasma free fatty acid (FFA) concentrations with two more rapid procedures (Method II and Method III). Method I required total lipid extraction, separation of FFA by thin-layer chromatography, methylation, and gas-liquid chromatographic analysis of the fatty acid (FA) methyl esters. Method II was a colorimetric procedure. Method III relied upon diazomethane's presumed ability to selectively methylate FFA even in the presence of FA esters. The three methods were compared using plasma from fasted and from fed nude mice, tumor-bearing mice (MX-1 and ZR-75-1 human mammary carcinomas), and controls. Method II, was less reliable than Method I, but both gave similar mean values for plasma FFA levels in fasted mice. Both Methods I and II also showed similar lowering of plasma FFA levels after feeding previously fasted mice. Method III consistently gave values that were far greater than those obtained using Methods I and II. Moreover, highly significant differences between fasted and fed mice were obscured by direct methylation of plasma FFA with diazomethane (Method III). The excess FA methyl esters formed in Method III were derived from plasma phospholipids, but not from plasma triacylglycerols. After feeding fasted mice, plasma free palmitic acid and oleic acid levels fell (Method I); by contrast, the excess "FFA" formed by methylation of plasma phospholipid FA increased two-fold and fourteen-fold, respectively. Caution is therefore advised in the use of direct methylating agents when measuring total and individual plasma FFA levels.

Breast epithelial mucin serum clearance.

The breast epithelial mucin is one of the currently used serum breast cancer markers. Immunoassays detect elevations of this marker in breast cancer relapse; however, values obtained do not usually correlate with breast tumor load and false negatives are frequent. The causes for such results were investigated in an immunodeficient mouse grafted with transplantable human breast tumors. Increases in serum breast epithelial mucin values are only found after a significant threshold of tumor load is reached and only when a highly increased release of the marker is created by radioimmunotherapy. This indicates that substantial amounts of the marker must be released to overcome clearance. Further, a hepatic clearance mechanism is demonstrated since carbon tetrachloride toxicity increases breast epithelial mucin levels in tumor bearing mice.

Effect of dietary fat and monoclonal antibody therapy on the growth of human mammary adenocarcinoma MX-1 grafted in athymic mice.

A diet containing 11% per weight fish oils (Max-EPA) reduced the rate of growth of a transplantable human breast tumor (MX-1) grafted into immunodeficient mice (BALB/c, nu/nu) when compared to MX-1 tumors in mice fed polyunsaturated (corn oil) and saturated (lard) fatty acids; however, there was no difference between the corn oil and lard animal groups. Treatment with a cocktail of monoclonal antibodies (MoAbs) decreased the rat of growth of MX-1 in the corn oil fed animals but not in those fed lard or Max-EPA, but the rate of tumor growth of the Max-EPA treated group, either MoAb treated or not, was slower than that of the corn oil and lard groups.

Biochemical and histological characterization of antigens preferentially expressed on the surface and cytoplasm of breast carcinoma cells identified by monoclonal antibodies against the human milk fat globule.

The preparation of monoclonal antibodies (MAbs) against the human milk fat globule membrane with preferential binding to breast carcinoma cells is described. Using BALB/c mouse myeloma cells; inter-specific, intra-strain, and inter-strain hybridomas were isolated that identified three different components of the human milk fat globule of approximately 46,000, and 70,000 daltons and a mucin-like glycoprotein complex (NPGP) ranging from 400,000 to over a million daltons, respectively. Three MAbs (BrE1, BrE2, BrE3) identified the latter component which consists of at least three different size molecules for which the aforementioned MAb's have different binding specificities. MAbs, BrE2 and BrE3, bound to normal breast epithelial cells but to a lesser extent than to tumors and only at the apical surface facing the lumen, while they bound breast carcinomas strongly, and often in the cytoplasm as well as on the surface. Higher concentrations of BrE3 were required to stain normal breast compared to breast tumors. BrE1 also stained breast carcinomas both on the surface and cytoplasmically but did not stain normal breast tissue. The MAb, Mc13, as well as the previously reported MAb McR2, both against the 70,000 dalton component, did not significantly stain either normal or cancerous breast tissue in histological sections but did bind significantly to cultured breast epithelial cells and to the milk fat globule membrane. The MAbs, Mc8 and Mc3, reported previously to be against the 46,000 dalton component, stained histologically only malignant breast tissue but only weakly; however, they bound strongly to intact breast carcinoma cells and breast cell membrane preparations with a radioimmunobinding assay. These MAbs should be useful in characterizing the surface of breast epithelial cells, studying surface alterations in malignancy, and possibly in breast cancer diagnosis and therapy.