Penetration and binding of antibodies in experimental human solid tumors grown in mice.

Monoclonal antibody localization to human carcinoma xenografts in nude mice was investigated by injecting biotinylated antibody. The antibody used in most experiments, MA103, reacts with a widely distributed human antigen present at a high density on the cell surface and has a relatively high effective affinity, 5.8 x 10(9) M-1. Various doses of antibody were injected, and tumors were collected at various times after injection. Frozen sections were stained by the immunoperoxidase method to detect bound antibody, and adjacent sections were stained with the same antibody in vitro to demonstrate total antigen distribution. In complementary experiments, unconjugated MA103 was injected in high doses to determine whether subsequent in vitro staining of frozen sections by biotinylated MA103 would be inhibited. The results demonstrate that: (a) approximately 0.5 mg antibody was sufficient to saturate antigenic sites on viable tumors cells; (b) saturation was achieved on viable tumor cells throughout the tumor 3 days after injection; (c) in necrotic areas, some antigen was accessible to antibody in vivo, but a considerable fraction of the antigen was inaccessible; (d) at 1 day after injection, stained cells were confined to areas near the blood vessels in the stroma; (e) for i.p. tumors, there was no difference between i.p. and i.v. injection of antibody and no indication of solid tumor penetration directly from the peritoneal cavity; and (f) nonspecific localization of a nonreactive biotinylated monoclonal antibody was weakly detectable and when present was seen as diffuse staining in the connective tissue only. A more limited range of experiments was performed with a second biotinylated antibody, MH99, which reacts with an epithelial cell surface antigen also recognized by many other antibodies, including 17-1A, AUA1, and KS1/4. Results were generally similar, except that a higher dose, 1.5 mg, was required to obtain homogeneous dark staining of tumor cells, and there appeared to be a considerable amount of antigen in viable areas of the tumor that was not accessible in vivo, possibly because it was intracellular. This approach appears to demonstrate tumor localization of antibody more impressively than other methods that have been used. This is partly because of the excellent resolution attained, since the reactive antibody produces a sharp membrane-staining pattern that is totally absent using a nonreactive control antibody. In addition, use of a target antigen that is predominantly on the cell surface, as opposed to cytoplasmic or secreted, is probably important. These data will be of value in attempting to use antibodies for immunotherapy.