In vivo antitumor activity of a panel of four monoclonal antibody-vinca alkaloid immunoconjugates which bind to three distinct epitopes of carcinoembryonic antigen.

A panel of four murine monoclonal antibodies apparently directed against three distinct epitopes of carcinoembryonic antigen (CEA) was conjugated via oxidized carbohydrate groups to 4-desacetylvinblastine-3-carboxyhydrazide. The resulting antibody-vinca conjugates were evaluated for antitumor activity against 2-9-day-established LS174T human colorectal carcinoma xenografts. The antibodies (immunoglobulin G, IgG) employed in this study were 11.285.14 (IgG1), 14.95.55 (IgG2a), CEM231 (IgG1), ZCE025 (IgG1). Additive immunofluorescence studies indicated that CEM231 and ZCE025 recognized the same or a closely related epitope(s) on CEA which was distinct from the two epitopes bound by 11.285.14 and 14.95.55. The in vivo antitumor efficacy studies demonstrated that chemoimmunoconjugates prepared from 14.95.55 and ZCE025 were more active than the conjugates constructed from the 11.285.14 and CEM231 antibodies. The 14.95.55 and ZCE025 immunoconjugates were also more efficacious than free drug or drug conjugated to irrelevant murine IgG. The presence of increased carbohydrate content on the light chain of ZCE025 may have been responsible for the ability to construct ZCE025-vinca conjugates with about twice the drug content (approximately 10 mol of vinca/mol of IgG) than was achieved with the other antibodies. The highly conjugated form of ZCE025 demonstrated similar efficacy but was much less toxic than a ZCE025 conjugate containing 5 mol of vinca/mol of IgG. These data indicated that significant differences existed in the ability of monoclonal antibodies to target a cytotoxic agent for effective antitumor activity even when the immunoconjugates recognized the same antigen or even the same or closely related antigen epitope(s). Furthermore, these differences could not have been identified without extensive in vivo evaluation for antitumor efficacy.

In vivo antitumor activity of a monoclonal antibody-Vinca alkaloid immunoconjugate directed against a solid tumor membrane antigen characterized by heterogeneous expression and noninternalization of antibody-antigen complexes.

It is widely believed that antigen heterogeneity and noninternalization of antigen-antibody complexes will severely limit the antitumor activity of monoclonal antibody-drug conjugates. The B72.3 monoclonal antibody binds to a tumor-associated antigen which is heterogeneously expressed in human carcinomas (J. Schlom, Cancer Res., 46: 3225-3238, 1986). We therefore performed studies to assess the degree of internalization of B72.3 antibody-antigen complexes and the level of in vivo antitumor activity that could be achieved with B72.3 conjugated to 4-desacetyl vinblastine-3-carboxhydrazide. Internalization studies were performed on LS174T colorectal carcinoma and OVCAR-3 ovarian carcinoma cells using iodinated B72.3 as well as an iodinated antibody that binds to the human transferrin receptor, IIB21. These data indicated that, in contrast to HB-21, the B72.3 antigen-antibody complex was not internalized. The B72.3-Vinca alkaloid immunoconjugate demonstrated significant antitumor activity against LS174T xenografts, although complete regressions of established tumors were not achieved. Immunohistochemical analyses indicated that the B72.3 antigen was heterogeneously expressed in the LS174T xenografts and that tumor cells which were not killed by high doses of B72.3-Vinca also expressed the B72.3 antigen. These studies indicated that significant antitumor activity may be achieved by monoclonal antibody-drug conjugates even when antigen heterogeneity and noninternalization of antigen-antibody complexes are encountered. The data also suggested that the formulation of antibody-drug conjugate cocktails to counteract antigen heterogeneity may not be sufficient to eradicate all malignant cells within a solid tumor mass.

In vivo selection of human tumor cells resistant to monoclonal antibody-Vinca alkaloid immunoconjugates.

UCLA-P3 human lung adenocarcinoma cells were grown in nude mice and given repetitive treatments of a monoclonal antibody-Vinca alkaloid immunoconjugate. Although this therapy resulted in a greater than 4-fold reduction in mean tumor mass of the established tumors, some animals experienced a reinitiation of tumor growth after cessation of conjugate treatment. Two such animals were treated again with high doses of monoclonal antibody-Vinca but one of the tumors was no longer regressed by the drug conjugate. The tumor was excised, enzymatically dissociated, and grown in tissue culture. Cultured cells were reimplanted in nude mice and subjected to further therapy with a monoclonal antibody-Vinca conjugate. The resulting tumors were also refractory to the immunoconjugate therapy. This cycle was repeated for a total of three times and resulted in the serial in vivo selection of three conjugate resistant variants. The mechanism responsible for the in vivo resistance of human tumor cells to the monoclonal antibody-Vinca immunoconjugate is unknown but does not appear to involve antigen modulation, altered tumor cell growth rate, or an apparent decrease in tumor targeting in vivo. The resistance was also not accompanied by any detectable elevation in multidrug resistance 1 mRNA or P-glycoprotein expression. Significantly, the resistance pattern was observed only in vivo and was not maintained by cells grown in vitro.

In vivo efficacy of monoclonal antibody-drug conjugates of three different subisotypes which bind the human tumor-associated antigen defined by the KS1/4 monoclonal antibody.

A panel of three hybridomas has been isolated each of which secretes a single species of monoclonal antibody (MoAb) directed against the KS1/4 tumor-associated antigen originally described by Varki et al. (Cancer Res 44: 681, 1984). These MoAbs were designated L1-(IgG2b), L2-(IgG1), and L4-(IgG2a)KS. Binding specificity, immuno-precipitation, and competitive binding analyses indicated that these MoAbs each recognize the same epitope of the KS1/4 antigen. The immunoprecipitation studies indicated that the MoAbs recognized a major antigenic component of 42 kDa and a minor component of 35 kDa. The L-KS antibodies were evaluated as MoAb-drug conjugates against a variety of human tumor targets grown in vivo as nude mouse xenografts. The MoAb-drug conjugates were constructed using protein-A-purified MoAbs conjugated to 4-desacetyl-vinblastine-3-carbohydrazide. Efficacy was determined using various dosing protocols on 2-14 day established tumors of lung, pharynx, colon, and skin origin. Control experiments included the use of dual-flank antigen-positive and negative tumors, free MoAbs, free drug, and mixtures of MoAbs and drug. These studies indicated that significant tumor growth suppression and actual tumor regression could be achieved by the MoAb-vinca conjugates and that this activity was antigen-mediated. The drug conjugates were more efficacious than free drug or free MoAbs administered either singly or in combination with each other.

Development of a dual label fluorescence technique that can be utilized to elucidate the mechanism of action of monoclonal antibody-drug conjugates.

A method is described that allows the simultaneous visualization and relative assessment of both the antibody and drug components of monoclonal antibody-drug conjugates at the target cell membrane. The antibody is detected by a fluorescein-conjugated anti-mouse immunoglobulin serum while the drug is visualized by rhodamine avidin or phycoerythrin-streptavidin binding to a biotinylated anti-Vinca alkaloid monoclonal antibody. This technique was effective in demonstrating the cell surface localization of a monoclonal antibody-Vinca alkaloid conjugate to human lung adenocarcinoma cells grown in vitro and was also used to demonstrate targeting of the conjugate in vivo to the membranes of these same tumor cells grown as a nude mouse xenograft. This method was also utilized to help elucidate the mechanism of action of monoclonal antibody-drug conjugates.