Please wait...



Goto Application


Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters

[ EN ]


This invention was partially made with funds provided by the National Cancer Institute under grants
CA-26184 and CA-08748. Accordingly, the United States
Government has certain rights in this invention.

This invention relates to a method for the
production of monoclonal antibodies (monoclonal antibodies) to restrictive antigenic human cell components especially in human ovarian tissues. Such monoclonal antibodies have use in cancer diagnosis and therapy, as well as other cell disorders.


Conventional antisera, produced by immunizing animals with tumor cells cr other antigens, contain a myriad of different antibodies differing in their specificity and properties. In 1975 Kβhler and Milstein (Nature, 256:495) introduced a procedure which leads to the production of quantities of antibodies of precise and reproducible specificity. The Kβhler-Milstein procedure involves the fusion of spleen cells (from an immunized animal) with an immortal myeloma cell line. By antibody testing of the fused cells (hybridomas), clones of the hybridomas are selected that produce antibody of the desired specificity. Each clone continues to produce only that one antibody, monoclonal antibody (monoclonal antibody). As hybridoma cells can be cultured indefinitely (or stored frozen in liquid nitrogen), a constant, adequate supply of antibody with uniform characteristics is assured.

Antibodies are proteins that have the ability to combined with and recognize other molecules, known as antigens. Monoclonal antibodies are no different from other antibodies except that they are very uniform in their properties and recognize only one antigen or a portion of an antigen known as a determinant.

In the case of cells, the determinant recognized is an antigen on or in the cell which reacts with the antibody. It is through these cell antigens that a particular antibody recognizes, i.e. reacts with, a particular kind of cell. Thus the cell antigens are markers by which the cell is identified.

These antigenic markers may be used to observe the normal process of cell differentiation and to locate abnormalities within a given cell system. The process of differentiation is accompanied by changes in the cell surface antigenic phenotype, and antigens that distinguish cells belonging to distinct differentiation lineages or distinguish cells at different phases in the same
differentiation lineage may be observed if the correct antibody is available.

The preparation of hybridoma cell liner can be
successful or not depending on such experimental factors as nature of the innoculant, cell growth conditions,
hybridization conditions etc. Thus it is not always possible to predict successful hybridoma preparation of one cell line although success may have been achieved with another cell line. But it is often true that selected monoclonal antibody may be representative of a class of monoclonal antibody raised by a particular immunogen.
Members of that class share similar characteristics, reacting with the same cell antigen. Thus the invention includes hybridoma cell lines and monoclonal antibody with like or similar characteristics.

Progress in defining cell surface antigens is of great importance in differentiation and disease as markers for normal and diseased cells, thereby furthering diagnosis and treatment. Thus work on melanocytes was made possible by the recently discovered technique of culturing melanocytes from normal skin (Eisinger, et al., Proc. Nat'l. Acad. Sci. USA, 79 2018 (March 1982). This method provides a renewable source of proliferating cells for the analysis of melanocyte differentiation antigens. Likewise, a large number of cell lines derived form melanomas have now been established and these have facilitated the analysis of melanoma surface antigens. The advent of monoclonal antibodies has greatly accelerated knowledge about the surface antigens of
malignant melanoma, cell markers on both melanomas and melanocytes have been identified. A panel of typing
monoclonal antibodies has been selected which recognizes differentiation antigen characteristics at each stage of development in both melanocytes and melanomas. These differentiation antigens may be used to classify melanocytes and melanomas and to group them into characteristic
sub-sets. [Dippold et al. Proc. Nat'l.Acad. Sci. U.S.A. 77, 6114 (1980) and Koughton, et al. J. Exp. Med. 156, 1755 (1982)]. Immunoassay of melanocytes and melanoma cells within sub-sets is thus made possible.

Initial recognition of differentiation antigens came about through analysis of surface antigens of T-cell leukemias of the mouse and the description of the TL, Thy-1, and Lyt series of antigens. (Old, Lloyd J., Cancer
Research, 41, 361-375, February 1981) The analysis of these T-cell differentiation antigens was greatly simplified by the availability of normal T cells and B cells of mouse and man. (See Patents #4,361,549-559; #4,364,932-37 and
#4, 363, 799 concerning monoclonal antibody to Human T-cell antigens).

The existence of human leukemia specific antigens has been suggested by studies using heterologous antibodies developed by immunization with human leukemic cells
[Greaves, M.F. et al. Clin. Immunol, and Immunopathol 4:67, (1975); Minowada, J., et al. J. Nat'l. Cancer Insti.
60:1269, (1978); Tanigaki, N., et al. J. Immunol. 123; 2906, (1979)] or by using autologous antisera obtained from patients with leukemia [Garret, T.J., et al., Proc. Nat'l. Acad. Sci. USA 74:4587, (1977); Naito, K., et al., Proc. Nat'l. Acad. Sci. USA, 80: 2341, (1983)]. The common acute lymphoblastic leukemia antigen (CALLA) which is present on leukemia cells from many patients with non-T, non-B, acute lymphoblastic leukemia (N-ALL), some chronic myelocytic leukemias (CML) in blast crisis and a few acute T-lymphoblastic leukemias (T-ALL) was originally described using conventional rabbit heteroantisera [Greaves, M.F. et al. Supra ].

By the autologous typing technique [Garret, T.J., et al. Supra; Naito, K., et al. Supra 1983; Old, L.J. Cancer Res. 41:361, (1981)], antibodies uniquely reacting with ALL cells were found in sera obtained from patients with ALL, and seemed to recognize very similar antigens to CALLA
(Garret, T.J., et al. Supra; Naito, K., et al. Supra).
Another leukemia associated antigen detected by heterologous antisera is the human thymus leukemia (TL)-like antigen, which is present on thymocytes as well as leukemia cells (Tanigaki, N. et al. Supra). This antigen, is therefore, a normal differentiation antigen which is composed of a heavy chain (MW 44,000-49,000) and light chain (MW 12,000-14,000) similar to the class I HLA antigens (Tanigaki, N., et al. Supra). These investigations have, however, been hampered by the need for vigorous absorptions with normal tissues as well as the relatively small quantity and low titer of the antisera.

In vitro production of monoclonal antibodies by the technique of Kδhler and Milstein, Supra has provided a better system for the identification and detection of leukemia specific antigens. A panel of monoclonal
antibodies detecting cell surface antigens of human
peripheral blood lymphocytes and their precursor cells have been investigated in detail [Reinherz, E.L., et al. Proc. Nat'l. Acad. Sci. USA 77:1588, (1980)]. While monoclonal antibodies detecting antigens characteristic for different lymphocyte lineages can be used for classification of human lymphocytic leukemia [Schroff, R.W., et al. Blood 59: 207, (1982)], such antibodies have only limited therapeutic applications. Monoclonal antibodies detecting human
leukemia associated antigens have also been produced. These include several antibodies detecting the human equivalents of the murine TL antigens. One TL-like antigen is
recognized by NA1/34 [McMichael, A.J., et al. Eur. J.
Immunol. 9:205, (1979)], OKT6 (Reinherz, E.L., et al. Supra) and Leu 6 (R. Evans, personal communication). A second TL-like antigen is recognized by M241 (Knowles, R.W., et al. Eur. J. Immunol. 12: 676,1982). Monoclonal antibodies with specificities for common acute lymphoblastic leukemia antigens J-5 (Ritz, J., et al. Nature 283:583, 1980), NL-1 and NL-22 (Ueda, R., et al. Proc. Nat'l. Acad. Sci. USA
79:4386, 1982) have also been produced. Recently, Deng, C-T, et al. Lancet. i:10, 1982) reported a complement fixing monoclonal antibody (CALLA-2) which reacts with most
cultured human T-ALL cell lines and also reacts with most fresh T-ALL cells.

Mouse monoclonal antibodies to human tumor cell surface antigens have been produced in many laboratories (Lloyd, K.O. (1983) In: Basic and Clinical Tumor Immunology, Vol. 1 (R.B. Herberman, Ed.), Nijhoff, The Hague (in
press)). The intention of these studies often has been to identify tumor-associated antigens that could be useful in tumor therapy or diagnosis. An inherent difficulty in this approach is the diversity of antigens on the cell surface. Although it has been possible to identify some antigens with a very restricted distribution, antibodies to antigens that elicit very weak immune responses may be missed due to their scarcity. These restricted antigens may be quite difficult to identify. Also, immunization with a complex mixture of antigens such as tumor cells may suppress the antibody response to relatively less imm.unogenic molecules, in a manner resembling antigenic competition (Taussig, M.J.
(1973). Curr. Top. Micro. Immuno. 600125). Thus production of monoclonal antibody to restricted cell sites is an especially difficult task. The present invention provide cancer diagnosis and therapy and overcome problems heretofor encountered in the prior art with respect to ovarian and endometrial human cell antigens.

The production and characterization of mouse monoclonal antibodies (mAbs) to human tumor cells has recently been an active area of research in recent years. This effort was stimulated by many earlier immunological studies which suggested that existence of human
tumor-specific antigens, but such evidence has not been conclusive, and has not been consistently reproduced in ether laboratories, making the status of such antigens uncertain (reviewed in Old, L.J. Cancer Res. 41:361, 1981; Herberman, R.B. Cancer Res. 19:207, 1974; Kedar, E., et al. Adv. Cancer Res. 38:171 1983; North, R.J. Adv. Immunol.
35:89, 1984 and Weiss, P.W., Curr. Topics Micro. Immunol. 89:1,1980). Mouse monoclonal antibodies also have not yet provided conclusive identification of a human tumor-specific antigen, although a few possibilities have been reported (Schlom, J., et al. (1985) Adv. Cancer Res., 43:143-174; Tsuji, Y., et al. (1985) Cancer Res., 45: 2358-2362; Tong, A.W., et al. (1984) Cancer Res., 44:4987-4992; Chin, J., et al. (1985) Cancer Res. 45:1723-1929). However, monoclonal antibodies to many new differentiation antigens have been obtained, and some of these have been recognized as having potential value in tumor diagnosis and therapy, particularly if they are expressed at higher levels in tumors than in normal cells. It should be considered that even monoclonal antibodies reacting with numerous normal cell types are more specific than current therapeutic agents (drugs and
radiation). In addition several antigens identified by monoclonal antibodies (all of which are mucin-like) appear to be valuable serum markers for particular cancer types (Herlyn, M., et al. (1982) J. Clin. Immunol., 2:135-140; Klug, T.L., et al. (1984) Cancer Res., 44:1048-1053; Lan, M.W., et al. (1985) Cancer Res., 45: 305-310; Papsidero, L.D., et al. (1964) Cancer Res., 44:4653-4647, Hirota, M., et al. (1985) Cancer Res., 45 : 1901-1905).

Ovarian carcinoma is a promising target for monoclonal antibody therapy (as all cancers of other
non-essential organs) in that tissue-specific differentation antigens are as useful as tumor-specific antigens. However, monoclonal antibodies specific for a particular epithelial cell type have been difficult to obtain. Of the large number of monoclonal antibodies obtained reacting with human carcinomas, only a few appear to be specific for a
particular histological type, namely the prostate (Raynor, R.H., et al. (1984) J. Nat. Cancer Inst., 73: 617-625;
Frankel, A.E., et al. (1982) Proc. Natl. Acad. Sci. USA, 79:903-907), the lung (Stahel, R.A., et al. (1985) Int. J. Cancer 35:11-17) and the breast (Menard, S., et al. (1983) Cancer Res., 43:1295-1300) and the breast, and these specificities have not yet been independently confirmed.

A number of ovarian tumor antigens have been detected using xenogeneic polyclcnal sera (reviewed in Lloyd, K.O. (1982) Serono Symposium No. 46 (M.I. Colnagki, G.L. Buraggi and M. Ghrone, Eds.) Academic press. N.Y. pp. 205-211) but none are related to the antigens of the invention. Other laboratories have also described
monoclonal antibodies to human ovarian carcinoma different from those of the invention. Bhattacharya et al.
(Bhattacharya, M., et al. (1982) Cancer Res., 42:1650-1654) produced an antibody to a saline-extracted antigen detected only in mucinous cyst adenocarcinomas of the ovary and in fetal intestine. Serous cyst adenocarcinomas, the most common ovarian carcinoma, did not contain this antigen.
Bast et al. produced an antibody (OC 125) reactive with an antigen present on 6/6 ovarian carcinoma cell lines and one melanoma of 14 non-ovarian cell lines tested. This antibody reacted with sections of 12/20 ovarian carcinomas and was nonreactive with 12 non-ovarian carcinomas and with most normal tissues, including normal adult and fetal ovary.
Weak reactivity was observed with adult fallopian tube, endometrium and endocervix (Bast, R.C., et al. (1981) J. Clin. Invest., 68: 1331-1336; Kabawat, S.E., et al. (1983) Amer. J. Clin. Pathol., 79:98-104).

We herewith incorporate by reference our previous work in the field namely U.S. patent application S.N.
562,465, and Intern J. of Gynocol. Pathol. 4: 121 (1985) concerning mAbs MH94, MF116, MD144, MH55, MF61, MH94 and MK99; also U.S. patent application 764,862 and J. Histochem. and Cvtochem (1985) 33:1095-1102 concerning mAbs MU78, MT334 and MQ49 and U.S. patent application S.N. 556,579 and
Hybridoma 2:pp 253-264 (1983) concerning mAb MH99.

Monoclonal antibody for ovarian cancers described. The antigenic profile of each of these monoclonal antibodies is presented with both serclogical and tissue reactivity studies in cancer and normal cell lines and tissue sections. These monoclonal antibodies form a panel useful for the diagnosis and therapy of ovarian cancers.

Ovarian epithelial cells form a simple cuboidal epithelium, and have no known function that distinguishes them from other epithelial cells; it is possible that there is no marker unique to these cells.

Also provided by this invention is an F(ab')2 fragment of the monoclonal antibody MX35.

This invention also provides a method of detecting human ovarian cancer in a subject which comprises obtaining a suitable sample from the subject, contacting the suitable sample with an amount of the aforementioned F(ab')2 MX35 fragment, labeled with a detectable marker, effective to and under conditions permitting the "fragment to form a complex with an antigen present on human ovarian cancer cells if present in the sample, and detecting any complexes so formed, thereby detecting human ovarian cancer in the subject.

This invention further provides a method of treating human ovarian cancer in a subject which comprises administering to the subject an amount of the aforementioned MX35 fragment, conjugated to a therapeutic agent, such as a radioactive therapeutic agent, effective to treat human ovarian cancer.

This invention further provides a method of detecting human ovarian cancer in a subject which comprises administering to the subject an amount of the aforementioned MX35 fragment, labeled with a detectable marker, effective to and under conditions permitting the fragment to specifically form a complex with an antigen present on human ovarian cancer cells if present within the subject, and detecting the detectable marker labelling the antibody so complexed.

Detailed Description of the Invention
This invention provides an F(ab')2 fragment of the monoclonal antibody MX35.

Also provided is the F(ab')2 MX35 fragment labeled with a detectable marker. Detectable markers useful in the subject invention can readily be ascertained by those of ordinary skill in the art. Useful detectable marker include, but are not limited to, enzymes, for example alkaline phosphatase or horseradish peroxidase;

substrates for enzymes; compounds capable of fluorescing, for example fluorescein; and radioactive marker. Examples of radioactive markers useful for the subject invention include, but are not limited to, radioactive isotopes, for example radioactive iodine such as 131I or 125I, radioactive technetium, and radioactive indium, and compounds containing such radioactive isotopes. Other detectable markers may be found by those of ordinary skill, and such markers are useful for purposes of the subject invention.

This invention also provides the aforementioned F(ab')2 MX35 fragment conjugated to a therapeutic agent. Therapeutic agents useful for the subject invention are those therapeutic agents which kill cancer cells, that is cells having a malignant phenotype. Therapeutic agents capable of killing cells having a malignant phenotype are well known to those of ordinary skill in the art, and any such therapeutic agent may be used in the subject invention. Useful therapeutic agents include drugs capable of killing malignant cells, for example doxorubicin; toxins; bacterial toxins, for example cholera toxin; and radioactive therapeutic agents, such as radioactive isotopes, for example 131I, radioactive technetium, and radioactive indium, or compounds containing such radioactive isotopes.

This invention also provides a method of detecting human ovarian cancer in a subject which comprises obtaining a suitable sample from the subject, contacting the suitable sample with an amount of the aforementioned F(ab')2 MX35 fragment, labeled with a detectable marker, effective to and under conditions permitting the fragment to form a complex with an antigen present on human ovarian cancer cells if present in the sample, and detecting any complexes so formed, thereby detecting human ovarian cancer in the subject.

Suitable samples for purposes of the subject invention include tissue samples, such as tissue biopsies. The tissue sample may be obtained from normal-appearing ovarian tissue. In one embodiment, the normal-appearing ovarian tissue sample is obtained from a subject who may be predisposed to developing ovarian cancer, for example a subject whose family has a history of ovarian cancer. The tissue sample may, however, be obtained from abnormal-appearing tissue, for example from a growth or tumor. Such abnormal-appearing tissue may be abnormal-appearing ovarian tissue, but may also be abnormally-appearing tissue of any organ in the peritoneal cavity of a subject.

Another suitable sample is a fluid sample, such as, but not limited to blood, plasma, serum, mucus (for example cervical mucus), fluid obtained from the peritoneal cavity, including ascites fluid (i.e. fluid which has collected in the peritoneal cavity of a subject as a result of infection occurring, for example, from a growth or tumor).

Other suitable samples useful for the subject invention may be ascertained by those of ordinary skill in the art.

Complexes are detected by detecting the detectable marker labelling the antibody fragment, and the means of detection will depend on the particular detectable marker chosen for use in the subject invention. For example, if the detectable marker is an enzyme, antibody fragment-antigen complexes may be detected by contacting the sample with a substrate for the enzyme and monitoring production of the enzyme product. As another example, if the detectable marker is a radioactive marker, the antibody fragment-antigen complexes may be detected by X-ray, by counting radioactive emissions, or by scintillation counting. An appropriate detection means may be ascertained by those of ordinary skill in the art.

In one embodiment, detecting human ovarian cancer in the subject comprises detecting a micrometastatic tumor or micrometastatic tumors in the suitable sample obtained from the subject. A micrometastatic tumor is a small tumor which has metastasized from an ovarian cancer tumor. In one embodiment, a micrometastatic tumor is less than about 1.5 cm in diameter. In another embodiment a micrometastatic tumor is less than about 1 cm in diameter. In a further, a micrometastatic tumor is less than about 5mm in diameter, and in another embodiment a micrometastatic tumor is less than about 1mm in diameter.

In another embodiment, detecting human ovarian cancer in the subject comprises detecting an epithelial ovarian carcinoma or epithelial ovarian carcinomas in the subject.

This invention further provides a method of treating human ovarian cancer in a subject which comprises administering to the subject an amount of the aforementioned MX35 fragment, conjugated to a therapeutic agent, such as a radioactive therapeutic agent, effective to treat human ovarian cancer.

For purposes of the subject invention, "treating human ovarian cancer" means killing ovarian cancer cells. Note that the ovarian cancer cells are not limited in location in the ovaries, but may comprise metastases or may be shed ovarian cancer cells located in other parts of the subject's body, for example in the peritoneal cavity of the subject.

Suitable modes of administration of antibody fragments are known in the art, and such method may be used in the invented treatment method. Examples of suitable forms of administration include intravenous injection and intraperitoneal injection. Administration may comprise administering the subject antibody fragment into the subject's peritoneal cavity.

The effective amount may be determined by methods known in the art. Typically, the subject is given a small dose of the antibody fragment conjugated to the therapeutic agent, and the dosage is increased until the subject cannot tolerate adverse side effects caused by the antibody fragment-conjugated therapeutic agent.

In one embodiment of the aforementioned method of treatment, the human ovarian cancer cells comprise a micrometastatic tumor or micrometastatic tumors. Micrometastatic tumors are described above.

This invention further provides a method of detecting human ovarian cancer in a subject which comprises administering to the subject an amount of the aforementioned MX35 fragment, labeled with a detectable marker, effective to and under conditions permitting the fragment to specifically form a complex with an antigen present on human ovarian cancer cells if present within the subject, and detecting the detectable marker labelling the antibody so complexed.

Appropriate methods of detection, as described above, depend on the chosen detectable marker. As described above, an appropriate method of detection can be determined by one of ordinary skill in the art.

In one embodiment of the above-described invented method of detecting human ovarian cancer in a subject, detecting human ovarian cancer in the subject comprises detecting a micrometastatic tumor or micrometastatic tumors in the subject. Micrometastatic tumors are as described above. In a further embodiment, the micrometastatic tumor or micrometastatic tumors are located in the subject's peritoneal cavity.

Suitable modes of administration include intravenous injection and intraperitoneal injection. Administration of the antibody fragment labelled with the detectable marker may comprise administering the fragment into the subject's peritoneal cavity.

This invention will be better understood from the Examples in the experiments which follow. However, one skilled in the art will readily appreciate that the specific methods and results discussed are merely illustrative of, and are not intended to, nor should they be construed to, limit the invention as described more fully in the claims which follow thereafter.

We have investigated monoclonal antibodies to ovarian carcinomas for 5 years in an attempt to identify tumor-specific or tissue-specific markers (Mattes, M.J., et al. (1984) Proc. Nat'l. Acad. Sci. USA, 811568-572;
Cordon-Cardo, C, et al. (1985) Int. J. of Gynocol. Pathol., 4:121-130; Mattes, M.J., et al. (1985) J. Histochem.
Cytochem., 33: 1095-1102). Initially cell lines were used for immunization and screening. Since ovarian carcinoma cell lines are very rare, and are likely to be
nonrepresentative of tumors occurring in vivo, we began another study in which we immunized mice with fresh tumor specimens, and used frozen tissue specimens as targets for screening. Here we describe 3 mAbs produced in this way, and a fourth, previously undescribed, produced earlier by immunization with an ovarian carcinoma cell line. These 4 monoclonal antibodies react with most of all fresh ovarian carcinomas and with a distinct range of normal epithelial cells. We describe their reactivity with fresh ascites carcinoma cells, and their lack of reactivity with normal mesothelium; these properties show a potential use in the effective intraperitoneal immunotherapy. We here describe the reactivity of these monoclonal antibodies on a large panel of normal and malignant cell lines and on frozen sections of normal human, tissues, as well as some
biochemical characteristics of the antigens recognized.

Target cells. The origin and culture of cell lines derived form human tumors, normal human fibroblasts and normal kidney epithelial cells have been described
(Mattes, M.J., et al. Proc. Natl. Acad. Sci. USA (1984)
Supra). A summary of the cell lines used is given in Table I. Normal tissues were obtained at surgery or autopsy. The ovarian cyst used for screening hybridoma supernants was a serious cystadenoma. Ovarian adenocarcinomas tested
included serous (7 specimens), endometrioid (2) and mucinous (1), tumor cell liver was obtained from human cancer
serology laboratory and further tissue specimens from Dr. C. Cordon-Cardo and Dr. Virginia K. Pierce at SKI. Tissues were covered with O.C.T. Compound (Miles laboratories, Elkhart, IN) and frozen in a slurry of 2-methyl-butane cooled in liquid nitrogen.

Production of mouse monoclonal antibodies. The monoclonal antibodies described were obtained from 3
fusions. For the generation of MT179, (Balb/c X C57BL/6)F1 mice were immunized the the ovarian carcinoma cell line SK-OV-4. Intraperitoneal injections of approximately 0.1 ml of packed cells were given twice at an interval of two weeks. The other monoclonal antibodies were obtained after immunizing mice with a mixture of 4 fresh ovarian carcinoma specimens, including 2 samples of ascites cells and 2 solid tumors, using approximately 107 x cells of each specimen. The mixture was suspended in 1.0 ml and injected i.p. three times at 3 week intervals. To immunize with solid tumor specimens, fragments of frozen tumor were thawed, placed in approximately 2 volumes of Dulbecco's phosphate-buffered saline (DPBS3, Gibco, Grand Island, NY), teased with scalpels and pressed through a fine still screen. This preparation was stored frozen for subsequent injections. Ascites cells were prepared as described in the accompanying paper; after thawing, they were washed once with DPBS.
Three days after the last injection, the fusion of immune spleen cells with mouse myeloma MOPC-21 NS/1 cells was performed as described (Mattes, M.J., et al. (1983)
Hybridoma 2:253-264). Initially cells were plated in 480 wells (Costar #3524, 24-well plates). Hybridoma cultures were subcloned at least twice by limiting dilution in
96-well plates on a feeder layer of normal mouse spleen cells. For mAb at MT179, culture supernatants were tested for antibody activity on a panel of cultured cells
consisting of immunizing cell line and other types of human tumor cells. For the other monoclonal antibodies,
supernatants were tested for reactivity on cryostate sections of a benign ovarian cyst. The specificity of epithelial-specific supernatants was tested further on various frozen sections, cell lines, and ABO blood group-related antigen preparations, as described below.
Since several months were required for this specificity analysis, cells from all original wells were frozen in 10% dimethyl sulfoxide, using 2 vials for each well, just after the original supernatants were collected. For cloning, cells were thawed, grown briefly in a well of a 24-well plate, re-tested for antibody activity and cloned using our usual procedure above. Cloned hybridoma cells were injected subcutaneously into nu/nu mice. Sera from mice with
progressively growing tumors were collected and used for serological and biochemical characterization. Antibody subclass was determined by immunodiffusion in agar with anti-Ig heavy chain-specific reagents (Bionetics,
Kensington, MD).

Serological procedures. Red cell rosetting methods for adherent cultured for 1-4 days and nonadherent target cells were carried out as described previously (Farr, A.G., et al. (1981) J. Immunol. Methods 47:129-144; Graham, R.C., et al. (1965) J. Histochem. Cytochem. 13: 150-152). The immune Rosetting assay was done as described (Carey, T.E., et al. (1976) Proc. Natl. Acad. Sci. USA 7333278. For adherent target cells, 200-500 trypsinized cells were plated in 0.01 ml in wells of Terasaki plates (Falcon Microtest plates 3034) and cultured for 1-4 days. Nonadherent target cells were attached to the wells by pretreating the wells for 45 min. at room temperature with Conconavalin A (Con A, grade IV, Sigma Chemicals, St. Louis Mo) at 1.0 mg/ml in DPBS. After washing the plates twice and blotting, target cells in DPBS were added and incubated for 45 min at room temperature Mattes, M.J. et al. J. Immunol. Meth 61: 145 (1983). To test for neuraminidase sensitivity, target cells were treated for 1 hr at 37 with Vibrio cholerae
neuraminidase (Calbiochem-Behring, La Jolla, CA) diluted 1:10 in 0.05 M citrate buffer pH 5.5, 0.1 M NaCl, 0.01 M CaCl2.

Cytoplasmic antigens were detected using an immunoperoxidase method as described (Farr, A.G., et al.
(1981) J. Immunol. Meth. 47:129; Graham, R.C. Jr., et al. (1965) J. Histochem. Cytochem. 13:150-152) washed twice with PBS, then fixed with 2.0% buffered formaldehyde (Farr, A.G., et al. (1981) J. Immunol. Methods 47: 129-144) for 30 min. All incubations were at room temperature. After 2 washes with PBS, they were incubated with 0.05% NP40 in PBS for 15 min. After 2 washes with PBS, 5% fetal bovine serum (FBS), monoclonal antibody was added, starting with a 1:500 dilution. After a 45 min incubation, plates were washed twice and peroxidase-conjugated rabbit anti-mouse Ig (DAKO P161, Accurate Chemicals, Westbury NY) was added (prepared immediately before use by mixing 1.0 ml 0.05 M acetate buffer pH 5.0, 0.05 ml 3-amino-9-ethyl carbazole at 4.0 mg/ml in N,N-dimethylformamide, and 0.005 ml 3.0% hydrogen peroxide (Graham, R.C. Jr., et al. (1965) J. Histochem.
Cytochem. 13: 150-152). After 15 min, the plates were washed twice with PBS, once with water, and examined.

Immunopercxidase staining of tissue sections using the ACC method was also carried out as described previously (Mattes, M.J., et al. (1985) J. Histochem. Cytochem.
33: 1095-1102).

Frocedures for absorption of Ab activity, using cells scraped from culture flasks, have been described
[Hirota, M., et al. (1985) Cancer Res. 45:1901-1905]. To test heat stability of antigens, cells were heated to 100°C for 5 min, then washed once before use in absorptions.
Blood leukocytes and erythrocytes were tested by
immunofluorescence as described (Mattes, M.J., et al. (1984) Proc. Natl. Acad. Sci. USA 81:568-572), using the monoclonal antibody at 1/50 and fluorescein-conjugated goat anti-mouse IgG (Cappel Laboratories, Cochranville, PA) at 1/40.
Reactivity with blood group A, B, H, Lewisa, Lewisb, X and Y determinants (Lloyd, K.O., et al. (1968) Proc. Natl. Acad. Sci. USA 61: 1470-1477) was determined by a solid phase enzyme-linked immunoassay as described (Lloyd, K.O., et al. (1983) Immunogenetics 177537-541), except that the antigen preparations were dissolved in water.

Biochemical analysis. Each antibody was tested for its ability to precipitate an antigen from the spent medium and from detergent-solubilized cell extracts after labeling by 3 methods: metabolic incorporation of
[3H]glucosamine (Mattes, M.J., et al. (1984) Proc. Natl. Acad. Sci. USA, 81:568-572), metabolic incorporation of
[35S]methionine (Mattes, M.J., et al. (1984) Supra) or chloramine T 125I labeling of solubilized cell membranes

(Mattes, M.J., et al. (1983) Hybridoma 2: 253-264). NP40 sclubilization of labeled cells (Mattes, M.J., et al. (1983)

Supra) and immunoprecipitation procedures (Mattes, M.J., et al. (1984) Supra) have been described previously. To test heat stability, radio labeled extracts were heated at 100 for 5 min; precipitated proteins were removed by
centrifugation (7,000 rpm, 15 min), then standard
immunoprecipitations were performed. Preparation of chloroform; methanol, 2:1 cell extracts and their use in inhibitions assays has been described (Mattes, M. J., et al.

(1984) Supra).

The examples below are for illustrative of the invention without limiting it.

Example l

Of the 4 monoclonal antibodies described here, one, MT179, was produced by immunization and screening with an ovarian carcinoma cell line, SK-OV-4. The assay employed was immunoperoxidase staining of fixed and permeabilized cells, which was intended to detect primarily cytoplasmic antigens. Initial selection of hybridomas to be cloned was based on nonreactivity with a panel of melanomas and
astrocytomas, so we expected that monoclonal antibodies restricted to epithelial differentiation antigens would be selected. Other monoclonal antibodies produced from the same fusion were previously described (Mattes, M.J., et al. (1985) J. Histochem. Cytochem., 33: 1095-1102). Three other monoclonal antibodies were obtained by immunization and screening with fresh tissue specimens. Immunization was with a mixture of 4 fresh ovarian carcinomas, as described. The mAbs were screened on cryostat sections of a benign serous ovarian cyst. This cyst specimen was chosen because the lining epithelial cells appeared morphologically similar to the simple cuboidal epithelial cells, of the normal ovary, but were preserved better than normal ovarian epithelial cells in frozen sections.. Therefore, this screening was primarily designed to detect tissue-specific epithelial differentiation antigens. Initial selection of hybridomas was on the basis of reactivity with epithelial cells and non-reactivity with connective tissue and blood vessels in the sections. Of 393 supernatants tested, 16 were
epithelial-specific. A much larger number, roughly 1/3 of the total supernatants, showed reactivity with all cells in the section. Before subcloning, we performed additional tests of specificity, in an effort to select
ovarian-specific monoclonal antibodies. The 16 supernatants were tested on sections of normal colon and the 5
supernatants, which were negative were retained, these supernatants were also negative on normal skin. Since many epithelial differentiation antigens have been identified as ABO blood group-related antigens, we tested these five supernatants on a panel of mucins containing the antigens A, B, O, Lea, Leb, X and Y. Two of the supernatants reacted with the A antigen. The remaining 3 supernatants,
designated MW162, MW207 and MX35, were then tested against cryostat sections of 5 solid ovarian tumors; all 3 reacted with 5/5 specimens. They were then tested on a panel of 30 cell lines, including ovarian carcinomas, other carcinomas and other tumor types. This was done partially to confirm the specificity of the antibodies, but also to select positive cell lines which could then be used as targets for screening subclones. Target cells were tested by 2 assays: a rosetting assay to detect cell surface antigens and an immunoperoxidase assay on fixed, permeabilized cells to detect primarily cytoplasmic antigens. All three
supernatant antibodies reacted with some ovarian carcinoma cell lines and were negative on melanoma and astrocytoma cell lines; they also reacted with some non-ovarian
carcinoma cell lines. Frozen cells from these three original wells were thawed, established in culture, tested for retention of reactivity, then cloned twice by limiting dilution before expansion.

The four monoclonal antibodies described were all non-reactive with blood leukocytes and erythrocytes by immuncflucrescence. Also, they did not react with any of the ABO blood group-related antigen preparations tested.

MT179. Ab MT179 is an IgGl. Although it reacts strongly in immunoperoxidase assays, it has not precipitated a detectable component from ovarian carcinoma SK-OV-4 or colon carcinoma SW 480 cells labeled with [35S]methionine,

[3H]glucosamine of 125I. In absorption experiments, the antigen was destroyed by heating at 100° for 5 min,
suggesting that it is a protein. MT179 was detected initially by immunoperoxidase staining of fixed,
permeabilized tissue culture cells, which produced
cytoplasmic staining, but MT179 was also detected on the cell surface of SK-OV-4 by a rosetting assay.

In frozen sections, MT179 was detected in a number of normal epithelial cells, namely in the colon, lung, skin, pancreas and breast (Table 1). Other epithelial cells and all non-epithelial cells were negative. MT179 was detected also in sections of 8/10 ovarian carcinomas. Expression in 147 tissue culture cells lines is summarized in Table 2.
MT179 was detected by immunoperoxidase staining in 3/8 ovarian carcinomas (SK-OV-4, SW626, A7), 2/2 uterine
carcinomas (SK-UT-2, ME-180), 9/11 colon carcinomas (SW480, SW620, SW1116, SW1222, SW1417, SK-CO-1, -13, CaCO-2, 4J), 5/9 bladder carcinomas (Scaber, RT4, 5637, JON, SW780), 3/4 pancreatic carcinomas (CAPAN-1, -2, A3), 8/11 lung
carcinomas (SK-LC-2, -3, -4, -5, -7, -9, -11, -12) 2/5 breast carcinomas (MCF-7, SK-BR-7), 2/17 renal carcinomas (SK-RC-17, -35), 1/2 prostate carcinomas (DU145), 1/1 bile duct carcinomas (Charles) and 1/1 choriocarcinoma (SVCC), with a reciprocal titer ranging from 500-8,000. It was negative on 21 melanomas, 16/17 astrocytomas, 5
neuroblastomas and 26 hematopoietic tumors of various types. Regarding cultured normal cells, 3 fibroblast cultures were negative and 2/3 kidney epithelial cultures were positive.

MW162. MW162 is an IgM. It precipitated an antigen from [ H]glucosamine labeled SK-OV-6 that migranted at the top of a 9% acrylamide gel (greater than 300,000 daltons). The antigen could still be precipitated after incubating the labeled extract at 100 for 5 min. These characteristics suggest that the antigen is a carbohydrate determinant on a mucin or proteoglycan. The antigen was not detectably precipitated from cell extracts labeled with
[35S]methionine or 125I. Reactivity of MW162 with SK-OV-4 was not affected by pretreatment of the cells, after
permeabilization, with neuraminidase. The antigen was not detected, by inhibition, in a glycolipid fractions of
SK-OV-6 prepared by chloroform: methanol extraction.

In frozen sections, MW162 was detected in many epithelial cells, namely in the esophagus, stomach,
bronchus, lung, kidney distal tubules, pancreas, thyroid, uterus and breast (Table 2. Other epithelial cells, such as in the colon and skin, were negative, as were all
non-epithelial cells examined. The antigen was detected in frozen. sections of 10/10 fresh ovarian carcinomas. Often staining in sections was concentrated at the luminal edge of cells.

Its distribution in 105 tissue culture cells is shown in Table 2. It was detected most readily in the cytoplasm by immunoperoxidase staining, so this assay was used for screening tissue culture cells; the antigen was detected weakly and inconsistently on the cell surface by rosetting. It was detected in 5/8 ovarian carcinomas
(SK-OV-3, -4, -6, Colo 316, A10), 2/9 colon carcinomas
(SW1116, SW1222), 3/6 bladder carcinomas (JON, VM-CUB-1, -2), 2/3 pancreatic carcinomas (CAPAN-1, -2), 6/8 lung carcinomas (SK-LC-1, -3, -7, -8, -17, -21), 4/4 breast carcinomas (MCF-7, SK-BR-5, -7, CAMA), 2/11 renal carcinomas (SK-RC-7, -18) and 1/1 choriocarcinoma (SVCC), with a reciprocal titer ranging from 500-2,000. It was negative on 10 melanomas, 10 astrocytomas, 4 sarcomas and 20
hematopoietic tumors of various types. Regarding normal cultured cells, it was negative on 4 fibroblast cultures and positive on 3/3 kidney epithelial cultures.

MW207. MW207 is an IgGl. Of the antigens
described here, it is the only one that was recognized initially as a cell surface antigen, so the rosetting assay, rather than the immunoperoxidase assay, was used to screen cell lines. It has not precipitated a detectable component from ovarian carcinoma SK-OV-6 or renal carcinoma SK-RC-18 labeled by any of the 3 isotopes described in Materials and Methods. In absorption experiments, it was destroyed by heating to 100 for 5 min, suggesting that the determinant recognized is a protein. [Three ml of packed cells absorbed all detectable Ab activity, while 10 times more heat-treated cells had no absorption activity.] In frozen sections, MW207 was detected on certain epithelial cells only, namely in the bronchus, lung, kidney proximal tubules, pancreas thyroid, uterus and breast (Table 1. In the pancreas, only the cells lining ducts were stained (Fig. 2). MW 207 was also detected in sections of 10/10 ovarian carcinomas. On 103 cell lines (Table 2) MW207 was present on 5/8 ovarian carcinomas (SW626, A7, A10, SK-OV-3, -6), 5/9 colon carcinomas (SW620, SW837, SW116, SW1222, SK-CO-10), 1/6 bladder carcinomas (VM-CUB-1), 2/3 pancreatic carcinomas (CAPAN-2, ASPC-1), 6/8 lung carcinomas (SK-LC-1, -7,-8,-14,-21,LcLL), 3/4 breast carcinomas (MCF-7, SK-BR-5, -7), 10/11 renal carcinomas (SK-RC-1, -7, -10, -15, -18, -29, -33, -42, -45, Caki 1), 1/1 teratocarcinomas
(Tera-1) and 1/1 choriocarcinoma (SVCC) with reciprocal titers ranging from 10 to 10 . The most strongly reactive target cells included carcinomas of the ovary, colon, bladder, lung, breast and kidney. MW207 was negative on 10 melanomas and 18 hematopoietic tumors, but reacted weakly with 2/10 astrocytomas and 1/4 sarcomas. Hence this Ab appears to be not strictly restricted to epithelial cells, although in frozen sections only epithelial cells were detectably stained. Regarding normal cells, MW207 was negative on 4 fibroblast cultures and positive on 2/3 kidney epithelial cell cultures.

MX35. MX35 is an IgGl. Although it reacted strongly in immunoperoxidase assays, it has not precipitated a detectable component from ovarian carcinoma SK-OV-6 or renal carcinoma SK-RC-18 cells, labeled by any of the 3 isoptopes described above. Heat stability could not be determined by absorption experiments, since the AB activity, diluted to near the endpoint, was not absorbed by an equal volume of unheated packed A10 cells, due perhaps to a low level of antigen exposure in scraped cells. MX35 was initially recognized as a cytoplasmic antigen, but was also detected by rosetting on the cell surface of the ovarian carcinoma line A10.

In frozen sections, MX35 was detected in
epithelial cells of the normal bronchus, lung, kidney collecting ducts, thyroid and uterus (Table 1). All other tissues examined were negative. MX35 was also detected in sections of 10/10 fresh ovarian carcinomas. Staining was often concentrated at the luminal edge of cells. In tissue culture cell lines, MX35 expression was rare (Table 2), being detected on only 3/8 ovarian carcinomas (A7, A10, SK-OV-6), 1/8 lung carcinomas (SK-LC-1) and 3/11 renal carcinomas (SK-RC-18), -33, -53), with reciprocal titer of 500-32,000. The most strongly positive cell lines were SK-LC-1, SK-RC-18 and SK-RC-53. Since, as noted above.

10/10 fresh ovarian carcinomas were positive, the data suggests that this antigen may be lost during adaptation of tumor cells to tissue culture. Regarding normal cells, MX35 was negative on 4 fibroblast cultures and positive on 1/3 kidney epithelial cell cultures. Considering the results with both normal tissues and cell lines, MX35 is the most restricted of the antigens described here.

We have described above 4 distinct epithelial differentiation, antigens identified by monoclonal
antibodies. From their distribution on normal tissues and cell lines, it is clear that the 4 antigens are different from each other. We are not aware of antigens described by other investigators that are likely to be identical to these. Three of the monoclonal antibodies did not
precipitate a detectable radio-labeled component, so little is known about the biochemical nature of the antigens recognized. The reason for the lack of immmunoprecipitation is not known, but possibilities include the following: 1) The antigen recognized is a minor cell constituent with a slow turn-over, so is not labeled adequately. 2) The antigen lacks glucosamine, methionine and tyrosine. 3) The antigen is either not extracted or denatured by the
detergents used to solubilize the cells. MT179 and MW207 were heat labile, suggesting that they are proteins.

MAbs to differentiation antigens have a number of possible applications in cancer diagnosis and therapy as well as in more basic studies for cell biology and
differentiation. In the example, we describe results of immunofluorescent staining of fresh ovarian carcinoma ascites cells, using these 4 monoclonal antibodies, which suggest that they are potentially useful in intraperitoneal therapy of such tumors. In addition, these Abs would probably be helpful in identifying rare carcinoma cells in peritoneal washings or lymph nodes, as has been described with other mABs (Ghosh, A.K., et al. (1983) J. Clin. Pathol. 36: 1150-1153; and Johnston, W.W., et al. (1985) Cancer Res., 45:1894-1900). Also, a significant number of
intraperitoneal carcinomas derive from an unknown or
uncertain primary. These monoclonal antibodies can be helpful in determining the origin of the primary, but such an application would require prior extensive tests with frozen sections of tumors of known histological types. MX35 seems most useful in this regard, since it is the most restricted of the antigens described here, being negative on carcinomas of the colon, bladder, pancreas and breast, but positive on a proportion of carcinomas of the ovary, lung and kidney. Sections of 10/10 ovarian carcinomas (the only type of carcinoma tested) were positive. Although many normal epithelial cells express these antigens, there may be certain tissue types in which the antigens are markers of malignancy. For example, MW162 and MW207 were negative on the normal colon, but were positive on some colon carcinoma cell lines. Further studies on cryostat sections of carcinomas of various types are required to investigate this possibility. Also, considering that most cancer serum markers defined by monoclonal antibodies have been
characterized as mucins (Herlyn, M., et al. (1982) J. Clin. Immunol. 2:135-140; Klung, T.L., et al. (1984) Cancer Res., 44:1048-1053; Len, M.S., et al. (1985) Cancer Res.,
45:305-310; Papsidero, L.D., et al. (1984) Cancer Res., 44: 4653-4647; Hirota, M., et al. (1985) Cancer Res.,
45: 1901-1905), the mucin-like antigen identified by MW162 should be a potential serum marker.

We have not yet obtained monoclonal antibodies to tumor-specific or tissue-specific antigens, and this
negative result warrants some discussion. Other
laboratories have had similar results, although one report of ovarian carcinomas tumor-specific monoclonal antibodies has recently appeared (Tsuji, Y., et al. (1985) Cancer Res., 45: 2358-2362). We have performed 22 fusions after
immunizing mice with various ovarian carcinoma cell lines. From these fusions, monoclonal antibodies that are tumor-or tissue-specific would have been detected, if they were present on the immunizing cell line. We conclude that more restricted antigens, possibly: 1) Do not exist; 2) Are not readily detected by current methods; or, 3) Are not present on cell lines. The few ovarian carcinoma cell lines are not likely to be representative of in vivo tumors; the
difficulty in establishing new lines of ovarian and other carcinomas (except renal) is well known. In this paper we present our results of 2 initial fusions using fresh tissue as immunogen and screening target, and we believe further similar studies will be productive. The strategy of freezing uncloned hybridomas, to allow time for specificity testing on frozen sections, made this approach possible, and should be widely applicable.

Example II
Example for Therapy
We have attempted to select monoclonal antibodies that might also be effective agents for diagnosis and intraperitoneal therapy or radioimmunodetection of human ovarian carcinoma. Antibodies were tested for reactivity with the surface of fresh Ovarian carcinoma ascites cells, and for non-reactivity with normal m «esothelial cells. The antibodies tested included 33 that had been identified previously as reacting with epithelial differentiation antigens. Five antibodies were selected with the desired specificity, MH99, MT179, MW162, MW207 and MX35, and these antibodies also reacted with cryostat sections of most of all ovarian carcinomas and benign ovarian cysts. All reacted also with certain normal epithelial cells. We also observed that the degree of heterogeneity of antigen expression on ascites carcinoma cells was dependent on the particular antigen being examined, and related to the biochemical nature of the antigen. In particular, most ABO blood group-related antigens showed a striking degree of heterogeneity. The rationale fcr intraperitoneal
immunotherapy and the criteria for selecting appropriate antibodies are discussed.

The effectiveness of monoclonal antibodies in cancer immunotherapy and immunolocalization depends on their specificity. The optimal target antigen would be
tumor-specific, and much effort has been directed to obtain such monoclonal antibodies. To date, no monoclonal antibody has been conclusively shown to identify a tumor-specific antigen; though there are some possible candidates for breast carcinoma (Schlom, J., et al. (1985) Adv. Cancer Res., 43:143-174), ovarian carcinoma (Tsuji, Y., et al.
(1985) Cancer Res., 45:2358-2362), pancreatic carcinoma (Chin, J., et al. (1985) Cancer Res., 45:1723-1729) and lung small cell carcinoma (Tong, A.W., et al. (1984) Cancer Res., 44:4987-4992), but characterization of these antigens is still preliminary. The search for tumor-specific antigens is based cn numerous prior immunological studies suggesting the presence of such antigens on human tumors (Old, L.J., et al. (1981) Cancer Res., 41:361-375; Herberman, R.B., et al. (1974) Adv. Cancer Res., 19:207-263; Kedar, E., et al.
(1983) Adv. Cancer Res., 38:171-288; Szigeti, R., et al.
(1985) Adv. Cancer Res., 43:241-306; Shuster, J.,e t al.
(1980) Proc. Exp. Tumor Res., 25:89-139; and Thomson,
D.M.P., et al. (1985) Int. J. Cancer 35:707-14). The published evidence primarily consists of data indicating an immune response to syngeneic tumors, as detected by assays for antibodies (Old, L.J., et al. (1981) Supra),
lymphocyte-mediated growth inhibition or cytotoxicity
(Herberman, R.B., et al. (1974) Supra), T
lyymphocyte-mediated cytotosicity (Kedar, E., et al. (1983) Supra), delayed hypersensitivity (Herberman, R.B., et al. (1974) Supra), macrophage migration inhibition (Szigeti, R., et al. (1985) Supra) or leukocyte adherence inhibition
(Shuster, J., et al. (1980) Supra and Thomson, D.M.P., et al. (1985) Supra). However, such approaches have not yet allowed definite characterization of any tumor antigen, and the results have generally not been consistently reproduced in different laboratories, so the presence of human tumor-specific antigens muct still be considered speculative (discussed in references Schlom, J., et al. (1985) Supra; Weiss, D.W., et al. (1980) Curr. Topics Micro. Immunol.
89:1-83; North, R.J., et al. (1984) Adv. Immunol.,

Although tumor-specific monoclonal antibodies are not available, monoclonal antibodies to differentiation antigens may be of value. Such Abs react with certain normal adult cells as well as turner cells of particular types, so toxicity arising from reactivity with normal cells is probable with some of all of such monoclonal antibodies. However, by adjusting the dose and by modifying the
monoclonal antibody (such as by preparation of antibody fragments, or conjugates with radioisotopes or toxins) it may be possible to obtain a therapeutic effect without major side effects. It should be considered that such monoclonal antibodies are more specific than current therapeutic agents. Chemotherapeutic agents were selected initially not for their specificity, but for their toxicity, and effective treatment requires the maximum tolerated dose; the same approach seems valid for mAB therapy. The major difference in this regard between monoclonal antibodies and new
chemotherapeutic drugs is that monoclonal antibodies cannot be pretested in animals; this makes determination of the optimal treatment regimen much more difficult.

We have focused on selecting mABs for therapy and radioimmunodiagnosis of ovarian carcinoma. This tumor type is the leading cause of death among patients in the United States with gynecologic malignancy, and is not treated effectively by current methods (Bender, H.G., and Beck, L. (Eds) (1983) Carcinoma of the Ovary. New York:Gustav Fischer Verlag, 1983 and Haije, W.G., et al. (1982) Ann. Clin.
Biochem., 19:258-262). Radioimmunodetection might decrease the need for "second-look" surgery, which is currently performed routinely to diagnose tumor recurrence. In regard to monoclonal antibody therapy, ovarian carcinoma
has two major advantages. First, as with other
non-essential organs, a tissue-specific antigen would be as useful as a tumor-specific antigen. This advantage however has not yet materialized, in that no differentiation antigen specific for ovarian epithelial cells has yet been
described. Second, the tumor grow primarily in the
peritoneal cavity; blood-borne metastases can occur, but most patients succumb prior to this (Bergman, F., et al. (1966) Acta. Obstet. Gynecol. Scand., 45:211-225).
Therefore, intraperitoneal therapy wold be expected to enhance interaction of the monoclonal antibody with tumor cells and to reduce interaction with normal, antigen- positive cells outside the peritoneal cavity. The
importance of this factor is difficult to evaluate at present. Serum proteins pass rapidly from the peritoneal cavity to the blood in normal animals (French, J.E., et al. (1960) Quart. J. Exper. Physiol., 45:88-103).

Antibodies injected i.p. into patients with ovarian carcinoma are initially exposed to only one type of normal cell, mesothelial cells, which line all surfaces of the peritoneal cavity. In this paper we describe the selection of monoclonal antibodies that react with the surface of fresh tumor cells but not with normal
mesothelium. Fresh ascites carinoma cells were used as targets in immunofluorescence. Ovarian carcinomas grow as both ascites and solid modules of tumor cells attached to the lining of the peritoneal cavity. To predict reactivity of monoclonal antibodies with tumor cells in vivo, use of ascites cells as targets seems more reliable than examining either tissue culture lines of ovarian carcinoma (which are rare and probably not representative) or frozen sections of fresh tumors (which would not indicate which antigens are accessible on the cell surface in vivo). Mesothelial cells were examined in frozen sections. Also examined were frozen sections of solid ovarian .carcinomas and benign ovarian cysts. The monoclonal antibodies tested were 33 Abs produced by our laboratories which, from previous studies, appeared to demonstrate specificity for epithelial cells, and included monoclonal antibodies reacting with ABO blood group-related antigens. We have identified 5 distinct monoclonal antibodies that reacted with the surface of most or all fresh tumor specimens and that were negative on mesothelium; they also reacted with various normal
epithelial cells. These antibodies appear suitable for further evaluation as potential therapeutic agents.

Antibodies. The 33 monoclonal antibodies tested in this study were generated and characterized previously. They were obtained from mice immunized with various human and were included in these experiments on the basis of preferential reactivity with epithelial cells in frozen sections of normal human tissues and in cultured cell lines. All have been tested on a wide range of cell lines and normal tissues, and also have been partially characterized biochemically by immunoprecipitation (which in some cases did not precipitate a detectable component). Based on these data, the monoclonal antibodies appeared different from each other. In the following list, they are grouped according to the type of carcinoma used for immunization and to the publication in which they are described: ovarian, MF61, MF116, MH55, MH94 (Mattes, M.J., et al. (1984) Proc. Natl. Acad. Sci. USA 81:568-572; Cordon-Cardo, C, et al. (1985) Int. J. Gynecol. Pathol., 4:111-130), MH99 (Mattes, M.J., et al. (1983) Hybridoma 2:253-264), MQ49, MT334 (Mattes, M.J., et al. (1985) J. Histochem. Cytochem., 33 : 1095-1102), MT179, MW162, MW207, MX35 (Mattes, M.J., et al. Four mouse
monoclonal antibodies to human epithelial differentiation antigens above), MR54, MT78, MV9 (M.J. Mattes, unpublished data); bladder, T16, T87 (Fradet, Y., et al. (1984) Proc. Nat. Acad. Sci. (Wash.), 81: 224-228); renal, S6 (Ueda, R., et al. (1981) Proc. Nat. Acad. Sci. (USA), 78:5122-5126); teratocarcinoma, K4 (Rettig, W.J., et al. (1985) in press); choriocarcinoma, LK26, SV19, SV63 (placental alkaline phosphatase) (Rettig, W. , et al. (1985) Int. J. Cancer
35:469-475); lung, F-15, F-16 (J. Feikert, unpublished data and U.S. patent application S.N. 474,225); and colon,
HT29-15, V-215, CLK314 (Sakamoto, J., et al . (1985) Fed. Proc. 42:792). We also tested 7 monoclonal antibodies to ABO blood group-related antigens, since these are epithelial differentiation antigens. These monoclonal antibodies also were obtained after immunizing mice with various human carcinomas, and include: anti-A, HT29-36 (Furukawa, K., et al. J. Immunol., in press); anti-B (Ueda, R., et al. (1981) Supra) anti-Lewisa, T174; anti-Lewisb, T218; anti-H type 2, H11; anti-X, P12; and anti-Y, F3 (Sakamoto, J., et al.
(1984) Molecular Immunol., 21:1093-1098).

Ascites cells. 0.5-2.0 1 ascites fluid from patients with serous adenocarcinoma of the ovary were filtered through 4-ply gauze and spun 5 min at 600g.
Pelleted cells were resuspended in 5-10 volumes of
supernatant, and 40 ml aliquots were underlaid with 10 ml Ficoll-Paque (Pharmacia, Piscataway, NJ). After spinning 15 min at 3,000g, the cells at the interface were collected, washed once with Dulbecco's phosphate-buffered saline (DPBS, Gibco, Grand Island, N.Y.), 7.5% fetal calf serum, 10% dimethylsulfoxide at various concentrations (1-10% packed cell volume/volume) and frozen in liquid nitrogen.
Immunofiucrescent staining was performed by standard
procedures (Mattes, M.J., et al. (1984) Proc. Natl. Acad. Sci. USA 81:568-572), using mAb sera at 1/50,
fluorescein-ccnjugated goat anti-mouse IgG (Cappel
Laboratories, Cochranville, PA) at 1/40 in medium containing 20% normal human serum, and 3-6 ul packed ascites cells per sample. Examination was by epi-illumination using a 75W xenon lamp, Leitz filter cube H, and a 40x objective. Some samples contained normal cells such as lymphocytes,
macrophages and mesothelial cells as well as tumor cells; therefore, evaluation was based on observation of clustered cells only, since carcinoma cells in ascites usually are in clusters, while the normal cells present are rarely in clusters. Samples of cells were processed and examined by the Pathology Laboratory at SKI (Sloan-Kettering Institute, N.Y., N.Y.) (Dr. Patricia Saigo), to confirm that virtually all cells in clusters were malignant. Photographs of immunofluorescence were prepared using Kodak Tri-X film and 90 second exposures. Bone marrow cells from normal donors were provided by the Bone Marrow Transplantation unit at SKI and examined similarly by immunofluorescence.

Crycstat sections. General methods for staining 0.007 mm sections by the avidin-biotin complex method have been described (Mattes, M.J., et al. (1985) J. Histochem. Cytochem. 33 : 1095-1102). Ovarian cysts included 1 serous cystadenoma, 3 mucinous cystadenomas, 1 simple cyst, 1 serous cystadenocarcinoma of low malignant potential, and 1 mucinous cystadenocarcinoma of low malignant potential.
Strips of the cyst wall were folded in pleats before
freezing. Tissues containing normal mesothelium included the diaphragm, body wall and pericardium. Ovarian carcinomas included serous (7 specimens), mucinous (1 specimen) and endometroid (2 specimens).

The 33 monoclonal antibodies evaluated were generated in our laboratories by immunization with ovarian, bladder, renal, lung, and colon carcinomas and with
choriocarcinomas. We anticipated that the expression of cell surface antigens on ovarian carcinoma ascites cells might be quite different from antigen expression on cultured cell lines, and therefore we tested 26 monoclonal antibodies reacting with a variety of epithelial differentiation antigens. ABO blood group-related antigens are epithelial differentiation antigens; therefore, a panel of 7 monoclonal antibodies to blood group-related structures, which were also obtained after immunizing mice with human carcinomas, was included.

As the initial screening for a most of the
monoclonal antibodies, we examined frozen sections of a benign ovarian cyst, to confirm the specificity of the monoclonal antibodies for epithelial cells. A benign cyst was used rather than a normal ovary because, in our
experience, epithelial cells are better preserved in frozen sections. Monoclonal antibodies reacting with connective tissue or blood vessels were eliminated at this stage, which included 8 monoclonal antibodies (MR54, T16, T87, V-215, CLK314, F-15, F-16 and LK29). None monoclonal antibodies produced the expected staining of epithelial cells only. Eight monoclonal antibodies were negative on the benign cyst, but were tested further for reactivity with fresh ascites cells. We observed that some monoclonal antibodies stained the outer surface of the cyst as well as the inner epithelial cells. The cells lining the outer surface are presumably drived from the normal ovarian epithelium, but due to cyst formation they appear flattened, like
mesothelial cells, in morphology. The monoclonal antibodies that stained the outer as well as the inner surface of the cyst, (MH94, MQ49, MT334, HT29-15 and F3), all react with heat stable (100°) antigens, which are probably
carbohydrate. One, F3, reacts with the Y blood
group-related antigen. Another, MQ49, reacts with both σlycolipids and mucin-like molecules (Mattes, M.J., et al. (1985) Supra). MH94, MT334 and HT29-15 also react with mucin-like molecules. The MK94 antigen was detected by immunoprecipitation after labeling with H235SO4, but not with [3H]glucosamine (Mattes, M.J., et al. Proc. Nat'l.

Acad. Sci USA 81:568 (1984) and unpublished data). Other monoclonal antibodies (MH99, MT179, K4 and SV19) reacted with cells lining the inner surface of the cyst only.

The 25 remaining monoclonal antibodies were tested against at least 2 specimens of fresh ovarian carcinoma ascites cells, by immunofluorescence. We had initially attempted, to facilitate assays, to attach these target cells to wells of Terasaki plates using concanavalin A, which has been effective with a wide range of nonadherent cell types (Mattes, M.J. et al., J. Immunol. Meth. 61:145 (1983). However, the ascites cells did not attach stably under these conditions, suggesting that they have unusual surface properties. We also tested 2 broadly reactive monoclonal antibodies, MA103 and AJ2 (Mattes, M.J., et al. (1983) Hybridoma 2:253), which are present on all human tissue culture cells: these monoclonal antibodies reacted with all ascites cells tested. Monoclonal antibodies negative with the 2 ascites specimens were not tested further (11 monoclonal antibodies: MF61, MFI 16, MH55, MV9 , K4, SV19, SV63, S6, HT29-15, HT29-36 and S8), but positive monoclonal antibodies were tested on additional ascites specimens. Results are summarized in Table 3. Six
monoclonal antibodies reacted with 5/5 or 4/5 specimens, and produced fairly homogeneous, ringed staining of tumor cells 2E). In these studies, we examined primarily clumped tumor cells since these were identified as malignant. The nature of the single cells was more variable, but many specimens clearly contained many malignant single cells, as indicated both by morphology and immunofluorescent staining.

Several monoclonal antibodies produced staining with certain unusual characteristics: 1) Tumor cells had a striking degree of heterogeneity. The fraction of positive tumor cells (in clumps) ranged from a few per cent to 50%. Positive cells were often extremely bright, although brightness was variable, so the positive and negative cell populations appeared to be distinct. A single clump of cells usually contained a mixture of strongly positive and negative cells. 2) In some specimens, many positive cells were not ringed, but were stained over a continuous portion, generally 1/3 to 2/3, of their surface. This occurred in both clumped cells and single tumor cells. In clumped cells, the stained portion of the membrane was usually the area not in contact with another cell. This antigen
distribution did not appear to be antibody-mediated, since it occurred in cells stained at 4° in the presence of 10 mM NaN3 (and which had been rretreated for 30 min with NaN3 before the first antibody incubation).

Of the monoclonal antibodies to ABO blood
group-related antigens tested, most that were positive on asciates cells produced this heterogeneous pattern, and this occurred with most specimens examined. The only other monoclonal antibodies producing this heterogeneous pattern were MQ49 and HT29-15, which also recognize carbohydrate determinants (Mattes, M.J., et al. (1985) J. Histochem.
Cytochem. 33:1095 Supra and Sakamoto, J., et al. (1985) Fed. Proc, 42:792). The exception among blood group-related monoclonal antibodies was F3 (anti-Y) which produced bright, ringed staining of most or all tumor cells in all specimens examined. This antigen, however, is present on erythrocytes (strongly on type 0, weakly on types A & B) (Sakamoto, J., et al. (1984) Molecular Immunol., 21: 1093-1098). Therefore, it does not appear to be a suitable target for tumor localization or therapy and has not been included in our subsequent studies of monoclonal antibody specificity.

The 5 monoclonal antibodies consistently reactive with ascites cells were tested further for specificity and results are included in Table 3. They reacted with frozen sections of most of all ovarian carcinomas tested, and with the epithelial ceils of many benign ovarian cysts. The most consistently positive monoclonal antibody was MH99, which reacted without exception with all cells derived form the ovarian epithelium. Table 4 summarizes the normal tissue reactivity of these monoclonals, which was described above. MH99 is most consistently positive and reacts with nearly all normal epithelial cells. THe other mAbs react with a subset of normal epithelial cells. Normal mesothelial cells in frozen sections of the lower surface of the diaphragm, body wall and pericardium were negative with all 5 mAbs. As described (Mattes, M.J., et al. (1983) Hybridoma 2:253 and "four mouse monoclonal antibodies to human epithelial differentiation antigens" Proc. Natl. Acad. Sci. USA) blood leukocytes and erythrocytes were negative by
immunofluorescence. Normal bone marrow cells, examined by immunofluorescence, were also negative.

The pattern of staining in frozen sections varied depending on the particular antibody. Antibodies MH99, MT179, and Mw207 stained all sides of the cells equally, in a pattern suggestive of membrane staining. In contrast, antibodies MW162 and MX35 often stained only the luminal edge of the cells and sometimes also produced a granular staining pattern.

To determine whether antibody-induced capping or other type of modulation cf surface antigens occurs, immunofluorescent staining of 2 ascites specimens was performed in the absence cf NaN3. Following staining, cells were incubated for 45 min at 37. This treatment had no effect on antigen distribution, relative to a control stained in the presence of NaN3, with any of the 5
monoclonal antibodies tested. We conclude that capping does not readily occur.

We also investigated the possibility that large amounts of soluble antigen in ascites fluid might inhibit monoclonal antibody binding in vivo. Fifty ul of autologous ascites fluid, collected at the same time as the cell specimen, was included during the first antibody incubation. This produced no detectable inhibition of immunofluorescent staining with any of the 5 monoclonal antibodies tested, suggesting that such inhibition would not occur in vivo.

We have identified 5 monoclonal antibodies that have potential value in intraperitoneal immunotherapy or immunodiαgnosis of ovarian carcinoma, since they react with the surface of fresh tumor cells but are negative on normal mesothelial cells. We attempted to examine cells as similar as possible to the cells that would be encountered by a monoclonal antibody injected i.p. We plan to examine cells frcrr biopsies obtained following injection cf radio-labeled antibody, to confirm the reactivity of the monoclonal antibodies in vivo. It is evident that examination of ovarian carcinoma cell lines, or cryostat sections of fresh tumors, does not provide reliable information regarding antigen expression on the surface of cells in vivo. For example, MT334 reacted with frozen sections of 10/10 tumors, but reacted weakly with only 1/5 ascites specimens.

The monoclonal antibodies described all react with some normal adult cells, but they may be effective in tumor localization or therapy without producing unacceptable side-effects. It must be considered that all current forms of cancer therapy are toxic, and the antibodies described are more specific than effects of chemotherapy or radiation. Moreover, after i.p. injection it is uncertain how efficiently the antibodies will reach the positive normal cells. For example, antigens located at the luminal edge of epithelial cells, such as MW162 and MX35, may not be exposed to the circulating antibodies. Also IgM's such as MW162 may not penetrate tissues sufficiently to reach the positive normal cells. If the antibody is fragmented or conjugated to other molecules, this may strongly affect the
localization in normal tissues and the side effects. In addition, monoclonal antibody binding to normal tissues may or may not produce toxicity. Radioisotope-conjugated monoclonal antibodies may damage tumor cells, which are relatively radio-sensitive, much more readily than normal cells to which they bind. Experimentation in humans is required to resolve these questions. Initially the
localization of radiolabeled antibodies will be investigated to determine whether the monoclonal antibodies bind
effectively to tumor and/or normal tissues. These
experiments will also indicate whether the monoclonal antibodies might be useful for radioimmunodetection of small tumor masses, which might eliminate in some cases the need for "second-look" surgery to detect tumor recurrence.

The answer to this question is obcured by a large number of reports describing putative tumor-specific antigens (Old, L.J., (1981) Supra; Herberman, R.B. (1974) Supra; Kedar, E., et al. (19S3) Supra; Szigeti, R. (1985) Supra; Shuster, J., et al. (1980) Supra; Thomson, D.M.P., et al. (1985) Supra; Weiss, D.W., (1980) Supra and North, R.J. (1984) Supra).
Until such antigens have been well-characterized, which has not been achieved in a single case, they must be considered as speculative. A more tissue-specific monoclonal antibody, reactive with fewer normal cells, is clearly preferable, but it should be considered that ovarian epithelial cells are relatively undifferentiated, and produce no known specific differentiaticn marker. Moreover, the truly tissue-specific markers, such as proteins secreted by the prostate, pancreas or breast, are often lost as a tumor progresses to a less differentiated state. We emphasize that the choice of the optimal target antigen for therapy depends not only on its restricted distribution. Equally important is its
consistent expression on nearly all tumor cells, at least from a particular patient, as well as on a low frequency of antigen-negative variants.

The monoclonal-antibodies used in these studies were generated in our laboratories, and may or may not be the optimal monoclonal antibodies for i.p. immunotherapy.

Other laboratories also have attempted to generate
monoclonal antibodies specific for ovarian carcinoma. Bast et. al. (Kabawat, S.E., et al. (1983) Int. J. Gynecol.
Pathol., 2:275-285) described monoclonal antibody OC125, which recognizes a high molecular weight antigen that appears to be a valuable serum marker for ovarian carcinoma (Klung, T.L., et al. (1984) Cancer Res., 44:1048-1053). In cryostat sections of normal tissues, only the epithelia of the uterus, fallopian tube and endocervix, and mesothelial cells of the peritoneum, pleura and pericardium were positive. Approximately 85% of ovarian serous carcinomas were positive. The reactivity with normal mesothelial cells causes this monoclonal antibody to be apparently,
inappropriate for i.p. immunotherapy. Also, expression of OC125 appeared to be heterogeneous in cryostat sections, with considerable negative cells being detected in positive tumors (Kabawat, S.E., et al. (1983) Amer. J. Clin. Pathol. 79:98-104). Gangopadhyay et al. (Gangopadhyay, A., et al. (1985) Cancer Res., 45:1744-1752) described the 1D3
monoclonal antibody, which reacted with essentially all ovarian mucinous carcinomas, and with only the normal colon among normal tissues examined; however, the more common serous carcinomas were negative. Tagliabue et al.
(Tagliabue, E., et al. (1985) Cancer Res., 45:379-385) obtained 2 monoclonal antibodies to ovarian carcinoma, MOv1 and MOv2. Mov2 reacted with most but not all ovarian carcinomas of all types, and with normal colon, stomach and breast. It reacted by immunofluorescence with most but not all fresh ovarian carcinoma ascites cells. MOvl reacted with mucinous but not serous ovarian carcinomas, and also reacted with the normal colon. Tsuji et al. (Tsuji, Y., et al. (1985) Supra) recently described 2 monoclonal
antibodies: 4C7 reacted with most ovarian mucinous,
endometroid and mesonephroid, but not serous carcinomas, while 3C2 reacted with most serous and endometroid, but not mucinous or mesonephroid carcinomas. These 2 antibodies did not react with any normal tissues, benign ovarian tumors or carcinomas of other organs.

Three monoclonal antibodies obtained to
breast-carcinoma-associated antigens have been found to react with ovarian carcinoma ascites cells. F36/22 reacted with 47/47 ascites specimens, and was negative with normal mesothelial cells in the same specimens (Croghan, G.A., et al. (1984) Cancer Res., 44:1954-1962). This monoclonal antibody also reacted with sections of ovarian carcinomas, but not with the normal ovarian epithelium, while benign ovarian tumors had weak and variable staining. However, many normal cells were positive with F36/22, including the breast, lung, sebaceous gland, sweat gland, uterus and kidney. Monoclonal antibody 3.14. A3, later called HMFG2, reacted with ovarian carcinomas and with normal bile duct, pancreas, sebaceous bland, salivary gland, kidney, lung, sweat gland and uterus (Arklie, J., et al. (1981) Int. J. Cancer, 28:23-29). This antibody, 131-I-labeled, was used in preliminary studies of i.p. injection for therapy of ovarian carcinomas (Epenetos, A. A., (1984) The Lancet,

B72.3 has also been detected in fresh ovarian carcinoma ascites cells (Johnston, W.U.V., et al. Cancer Res. 45:1894 (1985). It is positive on breast and other carcinomas, but negative cn all normal tissues examined. In sections of breast carcinomas, there was marked
heterogeneity in the expression of B72.3 (Schlom, J. et al. (1985) Supra). The presence of other well-characterized tumor markers, including carcinoembrycnic antigen,
alpha-fetoprotein and human chorionic gonadotropin, was also investigated in ovarian carcinomas; all were negative on more than 75% of serous carcinoma (Casper, S., et al. Am. J. Obstet. Gynecol. 149:154 (1984).

An interesting observation was that many
monoclonal antibodies to carbohydrate antigens produced striking heterogeneity in their staining of ascites cells. Tumor heterogeneity is of course a major obstacle in
effective treatment, and our observations suggest the important point that the level of heterogeneity can vary widely depending on the particular antigen, and that this level may be related consistently with the biochemical nature of the antigen. The mechanism for variation in blood group antigen expression is unknown, but may be related to the altered and variable state of differentiation of tumor cells. We suggest that this heterogeneity is directly related to the fact that blood group carbohydrate antigens, including ABO, Ii and T, account for many of the most consistent differences between malignant and normal cells of the same histological type (Feizi, T., (1985) Nature
314 : 53-57). That is, expression of carbohydrate antigens may be a very sensitive indicator of the state of
differentiation of a cell, allowing such antigens to be useful in diagnosis of malignancy. But, for the same reason, malignant cells may generally vary in expression of these antigens, making them unsuitable as targets for immunotherapy.

The rationale for i.p. immunotherapy is complex. Intraperitoneal therapy of ovarian carcinoma has been used extensively with a colloidal suspension of Cr32P04
(Rosenshein, N.B., et al. (1979) Obstet. Gyneco. Survey, 34:708-720) or with other chemotherapeutic drugs (Markman, M. in S.B. Howell (ed.) Intra-Arterial and Intracavitary Cancer Chemotherapy, PP. 61-69 Boston: Martiners Nijhoff, 1984), though evidently often without an understanding of the physiology of efflux from the peritoneal cavity, as noted by Leichner et al. (Leichner, P.K., et al. (1990) Radiology 134:729-734). Efflux of substances from the peritoneal cavity, including proteins, colloids, and
erythrocytes, is very rapid via lymphatics of the lower surface of the diaphragm. The lack of a substantial barrier to efflux is due to the presence of specilized mesothelial cells and lymphatics at this location (French, J.E., et al. (1960) Quart. J. Exper. Physiol., 45:88-103). This factor reduces the advantage of intraperitoneal therapy. However, there are several factors that can enhance the effectiveness of this approach. First, as noted for drug therapy
(Markman, M., (1984) In: S.B. Howell (ed.), Intra-Arterial and Intracavitary Cancer Chemotherapy, pp. 61-69. Boston: Martinus Nijhoff), the relative concentration in peritoneal fluid and in blood depends on both the rate of efflux from the peritoneal cavity and the rate of clearance from the blood. Substances cleared rapidly from the blood are therefore preferable. The clearance rate of monoclonal antibodies from the blood might be increased in various ways, such as removing sialic acid (Ashwell, G., et al.
(1974) Adv. Enzymol., 41:99-128). Second, patients with ovarian cancer have impaired efflux from the peritoneal cavity, due probably to blockage of lymphatics by tumor cells, which is presumably the reason for development of ascites (Feldman, G.B., et al. (1972) Cancer Res.,
32:1663-1666 and Coates, G., et al. (1972) Radiology,
107: 577-583). Third, it might be possible to decrease the rate of efflux in various ways which have been effective in experimental animals, such as by anaesthesia, blocking major lymphatic vessels (Courtice, F.C., et al. (1951) Austral. J. Exper. Biol. Med. Sci. 29:451-458), or disrupting the lymphatic capillaries on the lower diaphragm (Raybuck, H.E., et al. (1950) Am. J. Physiol. 199:1021-1024). Fourth, since the toxicity of monoclonal antibodies will result from binding to normal epithelial cells, if the antibody can be confined to the vascular system after it enters the blood, toxicity may be eliminated. This might be achieved by conjugating IgG to particles of diameter greater than 0.1u, which are too large to pass through the capillary
endothelium (Renkin, E.M., et al. (1977) Circ. Res.,
41:735-743 and Simionescu, N., et al. (1972) J. Cell Biol.

53:365-392). Regardless of the method of administration, some of the monoclonal antibody will enter the blood; this may be useful in that potentially metastatic tumor cells in the blood will be eliminated. The method developed for i.p. chemotherapy is to inject a large volume, and to withdraw it after several hours (Markman, M., (1984) Supra); this seems equally appropriate for monoclonal antibody therapy.
Intraperitoneal immunotherapy for ovarian cancer has
previously been attempted with polyclonal (Order, S.E., et. al. (1981) Cancer 48:590-596) and monoclonal (Epenetos, A.A., (1984) The Lancet, 1441-1443) antibodies in
preliminary experiments. Clearly many variables must be investigated in order to devise the optimal approach.

These hybridoma cell lines are on deposit at
Sloan-Kettering Institute 1275 York Avenue, New York, NY and at the American Type Culture Collection, 12301 Parklawn Drive, Rockville, Maryland 20852, a recognized international depository, under the following designations:

Radiolabeled monoclonal antibodies (Mabs) or their fragments are receiving considerable attention as tumor targeting agents for a wide variety of solid tumors

(Epenetos et al., 1982; Larson et al., 1983; Mach et al., 1983; Baum et al., 1986;

Delaloye et al., 1986; Chatal et al., 1987; Sharkey et al., 1990; Lloyd et al., 1993).

Although labeled Mabs are beginning to have an impact in the clinical domain, their specificity and sensitivity for tumor detection still remain to be improved.
Radioimmunotargeting with F(ab')2 and Fab fragments of antibodies has generally been favored over the use of the intact Ig due to their shorter biological half-life in the blood and rapid tissue distribution and clearance of the fragments resulting in higher tumor to normal tissue and tumor to blood ratios (Field et al., 1992).
The primary objective of our work was to study the biodistribution and pharmacokinetics of radiolabeled Mab MX-35 in a xenograft model of human ovarian cancer as a preliminary step to clinical studies using this antibody. Mab MX-35

(lgG1) was developed by immunization of mice with ascites and solid tumor ovarian cancer cells (Mattes et al., 1987; Mattes et al., 1989). It reacts strongly and relatively uniformly with 75-80% of ovarian carcinoma samples and with a few normal tissues

(Rubin et al., 1989, Rubin et al., 1991). The antigen detected is a cell surface 90,000 dalton, non-secreted glycoprotein (M. Welshinger, B.Y.T. Yin and K.O. Lloyd,
unpublished data).
In this study we place special emphasis on radiolabeled F(ab')2 fragments of Mab

MX35 as an alternative to the intact antibody. By comparing the tissue distribution of radiolabeled intact IgG with F(ab')2 fragments we demonstrated much better targeting using the fragments. More interestingly, compared with intact antibody, the F(ab')2 fragments gave markedly increased values of absolute tumor uptake of the antibody.

This result contrasts with previous results (Sands, 1990; Gerretsen et al., 1991;
Molthoff et al., 1992) obtained from comparative experiments.

Materials and Methods
Monoclonal Antibody Production
Monoclonal antibody MX-35, a murine lgG1, was produced from hybridoma ascites grown in BALB/c mice. High titer ascites batches were pooled for purification. MAb

MX-35 was purified from the ascites through several steps including removal of lipoprotein by ultracentrifugation at 100,000 g and ammonium sulfate precipitation

(50% saturation). Final purification was by protein A-agarose chromatography (Ey et
al., 1978).
Preparation of Fragments
Following overnight dialysis of purified MX-35 immunoglobuiin in 25mM sodium acetate pH 4.5.25 μl of pepsin (1.5 mg ml-1; Sigma Chemical Co., St. Louis, MO) was added to the antibody (2 mg in 400 μl) and incubated overnight at 37 C. F(ab')2 fragments were isolated using a kit obtained from Bio Chrom International, Tustin,

CA. High yield binding buffer (250 μl) containing 30 μl of anti-pepsin was added to the antibody-pepsin combination. The entire sample was diluted with 220 μl of high yield binding buffer and centrifuged to 2000 g for 15 min and the supernatant was placed on a protein-A-Avidchrome column. The unadsorbed fraction was concentrated using a Centricon 30 unit (Amicon, Beverly, MA) at 1075 g at 4 C. The final antibody concentration obtained was 1.95-2.34 mg ml-1 and the overall yield ranged from 63-65%. The identity of the fragments was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis under reducing conditions. Staining with

Coomassie blue, revealed bands of 23 kD and 25 kD corresponding to the light and
cleaved heavy chains, respectively (Figure 1).
Radiolabeling of Monoclonal Antibodies
Intact specific and control Mab or their F(ab')2 fragments were labeled using chloramine-T method as follows. MAb MX-35 (1 mg) were added to 0.5 ml of 0.15M

Na Cl in 0.05M phosphate buffer at pH 7.5. One mCi of 125l and 40μl of freshly dissolved chloramine-T (1 mg ml-1) were added on ice. After 10 min the reaction was terminated by addition of 45 μl sodium metabisulfate (1 mg ml-1). The protein was passed through a Sephadex G-25 column (10 ml) and fractions with the highest radioactivity were pooled. The immunoreactivity of the labeled product was determined by sequential absorptions with an antigen-expressing cell line (OVCAR-3). Between 60 and 75% of the radioactivity was adsorbed for both F(ab')2 fragments and intact antibody MX-35. Percent labeled protein was determined by the TCA precipitation method. For both F(ab')2 fragments and intact antibody incorporation of
iodine into protein was 90-95%.
Binding Assays
The method was described elsewhere (Mattes et al., 1989). Briefly, serial 2:3-fold dilutions of radiolabeled intact or F(ab')2 fragments of MAb MX-35 were incubated with 50 μl cell suspension (5 μl packed OVCAR-3 cells) mixed with 100 μl of labeled

MAb dilutions ranging from 0.05 μg ml-1 to 20 μg ml-1, for 5 hrs at 4 C. All assays were performed both with (2 mg ml-1) and without a high concentration of unlabeled

MAb MX-35. Free MAb was calculated by subtracting bound MAb from total bindable

MAb, and the equilibrium association constant Ka was determined by graphical analysis using the method of Scatchard. The off-rates for radiolabeled MAb bound to

OVCAR-3 cells were also determined as described by Mattes et al. (1989).
Biodistribution Studies
Athymic nude mice (BALB/c background) were implanted, intraperitoneally (IP), with fragments of the human ovarian cancer cell line, OVCAR-3, obtained by mincing tumor grown in another nude mouse. Visible tumors appeared approximately 3 weeks after injection. Each mouse had tumors of at least 300-500 mg. The expression of MX35 antigen in the tumor was confirmed by immunohistochemistry

(Rubin et al.1989). Mice bearing IP xenografts were randomized to be injected with either 20 μg of 125l-labeled MAb MX-35 [intact or F(ab')2] or the control antibody L6 anti-id 13B (intact or F(ab')2) [kindly provided by Dr. I. Hellstrom, Bristol-Myers Squibb

Pharmaceutical Research Institute, Seattle, WA] either intraperitoneally or intravenously. The footnote to Table 1 summarizes the number of mice sacrificed at various time points for the eight possible combinations of treatment groups. Tumor tissues and normal tissues from each time point were obtained at the time of sacrifice by dissection. All tumor was recovered from sites within the peritoneal cavity. From each animal 5 to 9 tumor samples were counted for 125l. These animal experiments were performed in compliance with the relevant national laws relating to the conduct
of animal experimentation.
Statistical Analysis
The uptake of antibodies was measured by calculating the % injected dose per gram of tissue. For tumor uptake and serum levels, mono-exponential models for the decay over time were fitted to the data using the method of analysis of covariance.

The half-life of MAb serum levels and tumor were estimated using the regression
estimate B as T ½=log2 / (-B) (Gibaldi & Perrier, 1975).
Tumor specific localization was measured by the data obtained from tumor to blood, tumor to liver and tumor to muscle ratios of % injected dose. The effects of the type and the form of antibody, administration route and time were examined using the method of analysis of variance. Log transformations on the ratios were performed to ensure normality. The localization index (II) (Beaumier et al., 1985) is defined as the

A localization index of 1 means that the antibodies have similar standardized uptakes

(Moshakis et al., 1981). An index greater than 1 indicates superior uptake of MAb
MX-35 over control antibody.
In this study, the control and specific antibodies were not co-administered to the same animals but to two independent experimental groups. The mean tumor to blood ratio (in log scale) for each treatment combination was used to calculate LI. The standard errors of LI were calculated using assumption of normality on the logarithm
of tumor to blood ratios.

Characteristics of Radiolabeled Antibody and Fragments
The radiolabeling procedure did not significantly change the immunologic properties of both the whole IgG and F(ab')2 fragments of MAb MX-35. They both retained 60- 70% immunological reactivity after radiolabeling. The binding affinities of the intact and F(ab')2 fragments of the antibody for OVCAR-3 cell line were determined by Scatchard analysis. Interestingly, the binding affinity of the intact MAb MX-35 (2.2 x 108 L M-1) was slightly lower than that obtained with F(ab')2 fragments (3.3 x 108 L M- 1). The number of binding sites/cell for the two forms was comparable (1.20 x 105 and 1.19 x 105 sites/cell, respectively). The off-rates for bound radiolabeled antibody were 23 hrs and 22 hrs for the intact antibody and fragment, respectively.
Phatmacokinetic Data
Serum clearance rates in 50 animals were determined for the 125l-labeled intact MAb

MX-35 and control antibody and in 58 animals for the 12Sl-labeled F(ab')2 fragments of

MAb MX-35 and the control antibody for both IV and IP routes of administration

(Table 1). The serum half-life was estimated as 31 hrs for IV and 39 hrs for IP injection for intact MAb MX-35 and 9 hrs for IV, 9.5 hrs for IP injection for the F(ab')2 fragments of MAb MX-35. We could not demonstrate any statistically significant difference in the magnitude of antibody accumulation in tumors and serum between

IV and IP routes of administrations for either intact or F(ab')2 fragments of-MAb MX- 35 or control antibody (all p> 0.07) (Table 1). However, the difference between the specific MAb MX-35 and the control antibody L6 anti-id 13B and the difference between intact and F(ab')2 fragments were significant for serum clearance (Figure 2A) and tumor clearance (Figure 2B). A separate mono-exponential model was thus fitted to each of the four combinations. The models explained the data well as indicated by the correlation estimates. It is clear from Table 1 and Figure 2 that the clearance in serum is consistently faster than that from tumor. For both tumor and serum, the half-life of F(ab')2 fragments are significantly shorter than that of intact

MAb (all p<0.01). Also the half-life of intact MAb MX-35 was shorter than for control antibody in both tumor and serum (both p<0.01). However, the half-life of F(ab')2 fragments of MAb MX-35 in tumor were longer than for the control antibody fragments

(P<0.01) and in serum there were no differences between the half-lives of F(ab')2 fragments of MAb MX-35 and the control antibody (P<0.1). As a result, the clearance of F(ab')2 fragments of MAb MX-35 is relatively faster in serum than in tumor.
Biodistribution Data
Following injection of radiolabeled antibody, peak percent injected dose (mean value) per gram of tumor tissue was 1.6 at 24 hrs for IV and 2.4 at 12 hrs for IP routes of injection for the intact MAb MX-35 (Figure 2b). The corresponding values at 12 hrs were 10.4 for IV and 8.2 for IP administration for F(ab')2 fragments of MAb MX-35.

There was no specific localization of the control MAbs in the tumor with peak % injected dose per gram tissue of 0.4 and 2.0 for intact antibody and F(ab')2 fragments,
respectively both IV and IP routes of injection.
Accumulation of radioactivity in nomal tissues was consistently lower, for both intact mAb and fragments, than in tumor tissue. Table 2 gives some representative data for one time point (24 hrs) and data for other time points are presented as tumor/normal tissue ratios in Table 3 and Figure 3. Ratios of tumor to blood, tumor to liver and tumor to muscle had similar patterns. The ratios of MAb MX-35 F(ab')2 fragments increased with time and, in most cases reached a maximum at 61 hrs for both IV and

IP routes of injection (Figures 3B, 3D, 3F). The ratios of intact MAb MX-35 also had an increasing trend when administered IV (Figures 3A, 3C, 3E). However, when given IP, neither intact MAb or F(ab')2 fragments ofthe control antibody had apparent changes over time. The mean ratios at 61 hrs for F(ab')2 fragments and the mean ratios at 72 hrs for intact antibody are tabulated in Table 3. This subset of data was chosen to be close to the peak area within the nature of the study design. Table 3 demonstrated that tumor to normal tissue ratios were much higher for MX-35 (Fab')2 fragments than for any other combination and that the ratios for the IP and IV
experimental groups were very similar.
Localization Data
Data for the localization of radiolabeled MAb in tumor relative to blood are shown in

Figure 4. The mean LI for F(ab')2 reached a value of 20.6 and 15.2 at 61.hrs for IV and IP injections, respectively, while mean LI for intact MAb was 4.8 and 4.3 for IV and IP injections, respectively with the former value being reached at 72 hrs and the latter at 24 hrs. The differences between IV and IP administrations were not significant except for that of intact antibody at 12 hrs (p<0.05) and 24 hrs (p<0.02). Generally, F(ab')2 fragments had higher LI than that of intact antibody (p<0.01 at 48

Immunologic binding characteristics of the antibody and the accessibility of the antigenic sites in the tumor are the fundamental basis for the selection of MAbs for

MAb-directed radio-immunodiagnosis or radioimmunotherapy of solid tumors.
Monoclonal antibodies with desirable characteristics should produce high tumor uptake and accompanying low background activity, i.e., a high target to nontarget ratios, and uniform MAb accumulation within the tumor (Sands et al., 1990; Sharkey et al., 1990; Waldman et al., 1991). The delivery of MAbs into the tumor is influenced by two major determinants; 1) MAb properties such as binding affinity, size of the antibody molecule, dose, immunoreactivity and internalization, and 2) intrinsic tumor properties such as histology, antigen density and homogeneity, vascularity, blood flow, permeability, and size of the tumor (Jain, 1987). In our study, intrinsic tumor properties and MAb properties remained constant between the two experiments except for changes in molecular size of the MAb MX-35.
As the transport of solute molecules into the tumor interstitium is governed by the biological and physicochemical properties of the diffusing molecule, the molecular weight of the administered antibody will strongly influence the rate of passive diffusion across both interstitial and extravascular subcompartments and cell membranes (Jain,

1987). Inaccessibility of antigen-bearing tumor cells to MAbs appear to be a major cause of inhomogeneous distribution of MAb in the tumor. In order to circumvent the problems associated with poor penetration of MAb into tumors, molecules of lower molecular weight are preferable for in vivo immunotargeting. Also, an additional advantage for low molecular-weight MAbs are their rapid clearance by urinary excretion (Bergardat et al., 1970; Buchegger et al., 1986; Endo et al., 1988).

Moreover, the fragments lack the Fc region responsible for nonspecific tissue uptake.

Radiolabeled fragments are, therefore, usually superior to whole IgG when used for in vivo immunotargeting (Buraggi et al., 1985; Andrew et al., 1986; Baum et al., 1986;

Andrew et al., 1988). Although these advantages hold true for both Fab and F(ab')2, reports dealing with studies on absolute tumor uptake of whole IgG, Fab and F(ab')2 often show a decrease in tumor uptake of fragments when compared to whole IgG, most likely the result of increased clearance from the blood and a decreased affinity inherent in the generation of fragments (Buchegger et al., 1986; Colapinto et al.,

1988; Endo et al., 1988); Wahl et al., 1983). Intact IgG usually gives the highest tumor uptake but this virtue may be overshadowed by high background levels of radioactivity. The renal clearance of whole IgG is relatively slow due to its high molecular weight, whereas fragments are rapidly cleared, thereby improving the
tumor/background ratio.
In our investigation, we compared the characteristics of whole antibody MX-35 and its

F(ab')2 fragments with regard to biodistribution in nude mice bearing OVCAR-3 xenografts. We noted significant differences in biodistribution between labeled F(ab')2 fragments and whole antibody MX-35, even though the immunoreactive fraction of the labeled antibody was very similar for both. Tumor to normal organ ratios were much higher for F(ab')2 fragments as compared to IgG. Additionally, tumor to blood ratios for the F(ab')2 fragments were approximately 7 times as high at 61 hrs as for IgG at 72 hrs or later (152 hrs) which is a similar finding to previous results obtained with intact MAbs and their fragments (Molthoff et al., 1992; Gerretsen et al., 1991).

Interestingly, there was an approximately 5-fold increase in % injected dose per gram with F(ab')2 fragments as compared to whole antibody resulting in a higher localization index (i.e., a higher specific/nonspecific tumor ratio). Our findings contrast with previous reports that have suggested that intact IgG's produce the highest levels of tumor uptake, while fragments produced significantly lower values.

Molthoff et al. (1992) reported maximum absolute tumor uptake for intact IgG ranging from 8.5 to 17.7% injected dose per gram for antibody whereas for the respective

F(ab')2 fragments the maximum values were 5.2% to 10% injected dose per gram. In the same context, Gerretsen et al. (1992) reported a mean tumor uptake as 14% for whole IgG and 7.2% for F(ab')2 fragments of the same antibody E48. However, in this particular study the investigators did not determine the binding affinities for the fragments. In a clinical study of mAb MOV18 in ovarian cancer patients, Buist et al.

(1993) reported %ID/kg of 6.2 and 0.9 for intact IgG and F(ab')2, respectively.

The basis for the superior tumor uptake by fragments is in our study not clear. The effect, could be due to the slightly higher affinity shown by the fragments. It is also possible that the number of effective binding sites in the tumor is higher with F(ab')2 fragments than with intact IgG due to better accessibility to the cell surface antigen as a consequence of reduced molecular size. Less surface area is occupied by the smaller F(ab')2 molecules (Molthoff et al.1992) which might escape a transportation barrier to which the intact antibody is vulnerable. Nevertheless, the number of binding sites/cell measured in vitro with OVCAR-3 cells was very similar for both forms of the MAb MX-35. Also, the off-rates for the cell-bound whole IgG and the fragment are very similar. Another possibility is that the intact MAb may be reacting with Fc receptors in tissues or in blood cells, although this is unlikely as mouse IgG, antibodies such as MAb MX-35, do not show this property. The explanation for the superior properties of the fragment will be the subject of further studies.
In summary, the use of F(ab')2 fragments of MAb MX-35 strongly improved absolute tumor uptake of the MAb when compared directly with intact MAb MX-35. An ongoing clinical study in patients with epithelial ovarian cancer on the localization of 125l-labeled whole MX-35 antibody has demonstrated modest accumulation (mean %ID g = 1.08 x 10-3 at 7 days) of the antibody in tumors (Rubin et al., 1993). These data are comparable to the values noted in the present animal study for whole antibody. This study therefore provides a rationale for a clinical study in patients with epithelial ovarian cancer patients using radiolabeled F(ab')2 fragments.

Andrew S. M., Pimm M. V., Perkins A. C. and Baldwin R. W. (1986) Comparative imaging and biodistribution studies with an anti-CEA monoclonal antibody and its F(ab')2 and Fab fragments in mice with colon carcinoma xenografts. Eur. J. Nucl.
Med.12, 168-175.
Andrew S. M., Perkins A. C, Pimm M. V. and Baldwin R. W. (1988) A comparison of iodine and indium labelled anti CEA intact antibody, F(ab')2 and Fab fragments by imaging tumour xenografts. Eur. J. Nucl. Med.13, 598-604.
Baum, R. P., Maul F. D., Senekowsch R., Lorenz M., Hottenrott C, Hopp J.. Standke

R., Chatal J. F., Saccavini J. C, Kriegel H. and Hor G. (1986) Radioimmuno-scintigraphy of CA 19-9/CEA producing tumors using 1-131 labelled F(ab')2 fragments of monoclonal antibodies (19-9/anti CEA radioinπmunococktail). In Nuclear Medicine in Clinical Oncology (Edited by Wmkler C), pp.197. Springer, Berlin.
Beaumier P. L., Krohn K. A., Carrasquillo J. A., Eary J., Hellstrom I., Hellstrom K. E.,

Nelp W. B. and Larson S. M. (1985) Melanoma localization in nude mice with
monoclonal Fab against p97. J. Nucl. Med.26, 1172-1179.
Bergard I. (1970) Plasma proteins in normal urine. In Proteins in Normal and

Pathological Urine (Edited by Manuel Y., Revillard J. P. and Betuel H.), pp.7-19.
Karger, Basel, NY.
Buchegger F., Mach J.-P., Leonnard P. and Carrel S. (1986) Selective tumor localization of radiolabeled anti-human melanoma monoclonal antibody fragment demonstrated in the nude mouse mode. Cancer 58, 655-662.

Buist M. R., Kenemans P., den Hollander W., Vermorken J. B., Molthoff C. J. M.,

Burger C. W., Helmerhorst T. J. M., Baak J. P. A., and Roos J. C. (1993) Kinetics and tissue distribution ofthe radiolabeled chimeric monoclonal antibody MOv18 IgG and F(ab')2 fragments in ovarian carcinoma patients. Cancer Res.53, 5413-5418.

Buraggi G. L., Callegaro L., Mariani G., Turrin A., Cascinelli N., Attili A., Bomberdieri

E., Terno G., Plassio G., Dovis M., Mazzuca N., Natali P. G., Scassellati G. A., Rosa

U. and Ferrone S. (1985) Imaging with l-131-labeled monoclonal antibodies to a
high-molecular-weight melanoma-associated antigen in patients with melanoma:efficacy of whole immunoglobulin and its F(ab')2 fragments. Cancer Res.
45, 3378-3387.
Chatal J.-F., Fumoleau P., Saccavini J.-C, Thedrez P., Curtet C, Bianco-Arco A.,

Chetanneau A., Peltier P., Kremer M. and Guillard Y. (1987) Immunoscintigraphy of recurrences of gynecologic carcinomas. J. Nucl. Med.28, 1807-1819.
Colapinto E. V., Humphrey P. A., Zalutsky M. R., Groothuis D. R., Friedman H. S., deTribolet N., Carrel S. and Bigner D. D. (1988) Comparative localization of murine monoclonal antibody Mel-14 F(ab')2 fragment and whole lgG2a in human glioma
xenografts. Cancer Res.48, 5701-5707.

Delaloye B., Bischof-Delaloye A., Buchegger F., Von Fliedner V., Grab J.-P., Volant

J.-C. Pettavel J. and Mach J.-P. (1986) Detection of colorectal carcinoma by emission-computerized tomography after injection of 123l-labeled Fab or F(ab')2 fragments from monoclonal anti-carcinoembryonic antigen antibodies. J. Clin. Invest.

77, 301-311.
Endo K., Kamma H. and Ogata T. (1988) Radiolabeled monoclonal antibody 15 and its fragments for localization and imaging of xenografts of human lung cancer. J.
Natl. Cancer Inst.80, 835-842.
Epenetos A. A., Britton K. E., Mather S., Shephard J., Granowska M., Taylor- Papadimitriou J., Nimmon C. C, Durban H., Hawkins L. R., Malpas J. S. and Bodmer

W. F. (1982) Targeting of iodine 123 labelled tumour associated monoclonal antibodies to ovarian breast and gastrointestinal tumors. Lancet 2, 994-1004.

Ey P.L, Prowse S. J. and Jenkin C. R. (1978) Isolation of pure IgG1, lgG2a, and lgG2D immunoglobulins from mouse serum using protein A-Sepharose. Biochemistry
15, 429-436.
Fjeld J. G., Michaelsen T. E. and Nustad K. (1992) The binding parameters of radiolabelled monoclonal F(ab')2 and Fab' fragments relative to immunoglobulin G in reactions with surface-bound antigens. Eur. J. Nucl. Med.19, 402-408.
Gerretsen M., Quak J. J., Suh J. S., Van Walsum M., Meijer C. J. L. M., Snow G. B. and Van Dongen G. A. M. S. (1991) Superior localization and imaging of radiolabeled monoclonal antibody E48 F(ab')2 fragment in xenografts of human squamous cell carcinoma of the head and neck of the vulva as compared to
monoclonal antibody E48 IgG. Br. J. Cancer 63, 37-44.
Gibaldi M. and Perrier D. (1975) Pharmacokinetics. Marcel Dekker, Inc, New York.

Jain R. K. (1987) Transport of molecules in the tumor interstitium: a review.
Cancer Res.47, 3039-3051.

Larson S. M., Carrasquillo J. A., Krohn K. A., Brown J. P., McGuffin R. W, Ferens J.

M., Graham M. M., Hill L. D., Beaumier P. L, Hellstrom K. E. and Hellstrom I. (1983)

Localization of 131l-labeled p97 specific Fab fragments in human melanoma as a basis
for radiotherapy. J. Clin. Invest.72, 2101-2114.
Lloyd K. O. (1993) Immunobiology of ovarian epithelial cancer. In Cancer of the

Ovary (Edited by Markman M. and Hoskins W. J.), pp.47-59. Raven Press, Ltd.,
New York.
Mach J.-P., Chatal J.-F., Lumbroso J.-D., Buchegger F., Forni M., Ritschard J.,

Berche C, Douillard J.-Y., Carrel S., Herlyn M., Steplewski Z. and Koprowski H.

(1983) Tumor localization in patients by radiolabeled monoclonal antibodies against
colon carcinoma. Cancer Res.43, 5593-5600.
Mattes J. M., Look K., Furukawa K., Pierce V. K., Old L. J., Lewis Jr. J. L, Lloyd K.

O. (1987) Mouse monoclonal antibodies to human epithelial differentiation antigen expressed on the surface of ovarian carcinoma ascites cells. Cancer Res.47, 6741- 6750.
Mattes J. M., Lloyd K. O. and Lewis Jr. J. L. (1989) Binding parameters of monoclonal antibodies reacting with ovarian carcinoma ascites cells. Cancer
Immunol. Immunother.28, 199-207.
Molthoff C. F. M., Pinedo H. M., Schluper H. M. M., Nijman H. W. and Boven E.
(1992). Comparison of the pharmacokinetics, biodistribution and dosimetry of monoclonal antibodies OC-125, OV-TL3 and 139H2 as IgG and F(ab')2 fragments in experimental ovarian cancer. Br. J. Cancer 65, 677-683.

Moshakis V., Mcllhinney R. A. J., Raghavan D. and Neville A. M. (1981)

Localization of human tumour xenografts after IV administration of radiolabelled
monoclonal antibodies. Br. J. Cancer 44, 91-99.
Rubin S. C, Finstad C. L, Hoskins W. J., Federici M. G., Lloyd, K. O., and Lewis Jr.

J. L. (1989) A longitudinal study of antigen expression in epithelial ovarian cancer.
Gynecol. Oncol.34, 389-394.
Rubin S. C, Finstad C. L, Hoskins W. J., Provencher D., Federici M. G., Lloyd K. O. and Lewis Jr. J. L. (1991). Analysis of antigen expression at multiple tumor sites n epithelial ovarian cancer. Am. J. Obstet. Gynecol.164, 558-563.
Rubin S. C, Kostakoglu L., Federici M. G.. Finstad C. L., Lloyd K. O., Larson S. M.,

Hoskins W. J. (1993). Clinical trial of pharmacokinetics, biodistribution and intraoperative radioimmunodetection of radiolabeled monoclonal antibody (MAb) MX- 35 in epithelial ovarian cancer. Gynecol. Oncol.49, 116-117.
Sands H. (1990) Experimental studies of radioimmunodetection of cancer; An
overview. Cancer Res. (SuppI.) 50, 809s-813s.
Sharkey R. M., Blumenthal R. D., Hansen H. J. and Goldenberg D. M. (1990)

Biological considerations for radioimmunotherapy. Cancer Res. (SuppI.) 50, 964s- 969s.
Wahl R. L., Parker C. W. and Philpott G. W. (1983) Improved radioimaging and
tumor localization with monoclonal F(ab')2. J. Nucl. Med.24, 316-325.
Waldman T. A. (1991). Monoclonal antibodies in diagnosis and therapy. Science
(Washington. D.C.) 252, 1657-1662.


b c

Brief Description of the Figures
Figure 1 : Sodium dodecyl sulfate-polyacrylamide gel electrophoretic analysis
under reducing conditions of MAb MX-35 (lanes a, b, c) and control MAb L6 anti-id 13B (lanes d, e, f) before enzyme treatment (lanes a, d); after pepsin digestion overnight at 37 C (lanes b, e); and following passage
through a protein A column to remove undigested intact IgG (lanes c, f).
The gel bands are visualized by Coomassie blue staining. Bands of Mr 50,000, Mr 25,000 and Mr 23,000 correspond to the heavy chains (hc), heavy chain fragments (hc-f), and light chains (Ic), respectively.

Figure 2: Mean percentages of injected radioactivity in blood (Panel A) and in
tumor (Panel B) showing clearance of intact IgG and F(ab')2 fragments of MAb MX-35 and control MAb L6 anti-id 13B. For intact IgG N=6 for all time points. For F(ab')2 fragments of MAb MX-35 N=7 at 12, 24, and 48 hours, N=8 at 38 hours and N=5 at 61 and 68 hours. The
experimental groups (IP and IV combined) are distinguished as follows:
intact MAb MX-35 (black dashed line); MAb MX-35 F(ab')2 fragments
(black solid line); control intact MAb L6 (stippled dashed line); and
control MAb L6 F(ab')2 fragments (stippled solid line). Vertical bars
indicate range or +/-2 standard errors.

Figure 3: Biodistribution data indicating tumor to blood ratios (A, B), tumor to
muscle ratios (C, D); and tumor to liver ratios (E, F) and comparing
intact IgG (A, C, E) and F(ab')2 fragments (B, D, F). All groups consisted of 3 or 4 mice per time point. The ratios for F(ab')2 fragments continued to increase up to 68 hours whereas intact antibody did not
show an apparent change over time. The experimental groups are
distinguished as follows: intact or fragmented MAb MX-35 by IP
injection (black solid line) or by IV injection (black dashed line); intact or fragmented control MAb L6 by IP injection (stippled solid line) or by IV
injection (stippled dashed line). Vertical bars indicate range or +/- 2
standard errors.

Figure 4: Comparison of tumor localization for intact IgG and F(ab')2 fragments.
The localization index reflects the ratio of activities in tumor of "specific" intact MAb MX-35 and F(ab')2 fragments over "nonspecific" control intact MAb L6 and F(ab')2 fragments and corrected for blood activity at the
same time. There was much more rapid development of specific uptake with F(ab')2 fragments of MAb MX-35 as compared to the intact MAb
MX-35. The experimental groups are distinguished as follows: intact
IgG by IP injection (stippled dashed line) or by IV injection (stippled solid line); F(ab')2 fragments by IP injection (black dashed line) or by IV
injection (black solid line). Vertical bars indicate range or +/-2 standard errors.

This ie a Phase I dose escalation study using 131I-MX35 F(ab')3 (2 mg/dose) for treatment of patients with lowvolume (≤ 0.5 cm diameter, documented at poet-chemotherapy Iaparotomy) MX35 antigen-expressing (≥ 25% of cells in an average high powerfield) ovarian carcinoma. Three patients per dose level will be followed forup to 8 weeks with biochemical and hematologictests fortoxicity, and CA-125 estimations for response. The firstcohort of patientswill receive 30mCI 131I laIbeled to 2 mg MX35 F(ab')2 antibody. Subsequentdose escalation will be In 30 mCI131I(labeled to 2 mg MX35 F(ab')2) Patients will betreated as outpatients in the Adult Day Hospital. Doses will be administered at intervals of4-5 days. Patients atthe first dose level will receive one outpatient infusion. Patients atsubsequent dose levelswill receive repeat outpatient Infusions at least4 days afterthe priorinfusion, with total patient radiation levels to be ≤ 5 mR/h at 1 meterat completion ofinfusion. All Infusionswill be completed within 14 days, to obviate/minimize anti-murine antibody responses during infusion. In the absence ofdisease progression and afterrecoveryfrom toxicity patients may be re-treated beginning 8weeks afterthe priorinfusion, fora total ofnot morethan three treatments.
A MaximumTolerated Dose (MTD)will bedetermined, defined asthe highest dose atwhich not morethan athird ofthe patients haveGrade IV hematopoietic or Grade III orgreaternon-hematopoletictoxicity. Serologicallyevaluable patients will be studied forpossible responses.

To determine the safety, toxicity and maximumtolerated dose (MTD) of131l-MX36 F(ab')2 administered byintraperitoneal (ip) injection to patients with residual ovarian cancer.
2.1 Curranttherapy ofovarian carcinoma.
Epithelial ovarian cancer(EOC) isthe fifth most common malignancy and the leading cause ofdeath fromgynecologicmalignancies among women in the United Stales, with an annual Incidence of26.000 cases. The disease
predominantly affects postmenopausal women in their sixth decade and
accountsforapproximately 15,000 deaths each year (1). Seventy percent of women presentwith advanced disease inwhich thetumorhas spread to the peritoneal surface ofthe upperabdomen. Extensive intraabdominal disease is difficultto eradicate completelyby surgery, and many patients have only a partial response to postoperative chemotherapy.
Atumor-associated antigen, CA-125, found in serum, is used as a
disease marker. Serum CA-125 level is elevated In approximately 80% of
patients with EOC and is a useful indicator ofdisease status during and after therapy In those patients.Adequate and complete surgical intervention is
mandatory primarytherapy forovariancarcinoma, permitting precise staging, accurate diagnosis and optimal cytoreduction.This is usually followed by
combination chemotherapythatinciudes aplatinum analogue - eithercisplatin or carboplatin. Aplatinum-containing regimen forms the cornerstone offirst-line .
treatment ofadvanced EOC. Chemotherapy is most effective in patients who have undergonemaximal eytoredudive surgeryorwho presentwith lowvolume disease. Preliminaryresults strongly suggestthat paelitaxel (Taxol) is very active in EOC. Six cycles ofchemotherapy have become the standard and yield clinical response rates ofapproximately60% to70% and 5-yearsurvivals of 10% to
20% (2).
Acombination ofCA-125testing and general physical and pelvic
examinations have been shown to detectprogression ofdisease in 90% of
patientswith recurrentEOC. However, benigndisease can give false positive predictions in up to 10% ofsuch patients. Seriel rise in CA-125 levels > 25%
have been shown in several studiesto predictprogressionwith almost 100% specificity (3,4,5). Routine radiological examinations (CT, MRI) have not been shown to improvedetection ofrecurrence, andthese studies are not routinely used In evaluation.
Currently available salvagetherapyforthe majority ofpatients with
relapsed EOC Is notcurative. Paelitaxel is the mostactivesingle agent for
treatment of relapse, even in patients refractory to platinum. It has a general response rate ofapproximately 35% (2). There is no evidence et presentthat earty reintroduetion ofchemotherapyis ofanysurvival benefit, orthat searching forthe site ofrelapsewill resultIn surgerythat can improve survival. Repeat surgical debutking maybenefit a subsetofhighlyselected patients with relapse, and serves to palliate complications such as bowel obstruction, to Improve the quality oflife. When a patient relapsesforthe second time, even with low-volume disease, there Is almost no possibility ofcure.
There Is thus a need ofdevelopmentfor alternativetreatment strategies in patients who demonstrate residual disease atthe time ofrepeat Iaparotomy

(carried out after Initial debutking and chemotherapy).
2.1.1 intraperitoneal radiolabeled antibody therapy In EOC.
Ovarian cancer remains largely confined to the peritoneal cavity during much ofits natural history. This has oncoureged intracavitary use ofa variety oftherapeutic radiopharmaceuticals, such as colloidal 32P
chromicphosphate, for advanced disease. Radiolabeled antibodies were first introduced by intracavitary injection by Epanetos etal(6). They
claimed that a higherconcentration ofradioactivity could be achieved in tumorafter local Instillstion than by systemic administration. Preliminary studies In ovarian cancer, using first 131l- and then 10Y- labeled HMFG
antibodies showed promise (7, 8). Patients with second relapse showed longerdisease-free survivalthan historical controls (15/16 free ofdisease @ 5 years) aftertreatmentwith 30Y-HMFG. The group also described the use ofa ehelete, administered to clearthe blood offree 30Y, as a way to improve the MTD by afactorof 3 (8). Toxicity was almost exclusively
Avarietyofotherinvestigators have subsequently shown the
benefits ofIP Injection ofradiolabeled antibody in ovanan cancer, with response rates of>50% even in advanced disease (9.10.11). However, moat series, with the exception ofthe studies by Epenetos etal. were in patients with bulky diseasewho had failed priorchemotherapy, and
responseswere ofshortduration.
Radiolabeled antibodies are most effective in small volume disease and the ideal trial designwould involve relapsed patients with small
volume disease. Complete responses should be possible in this setting.
In preliminary IPtrialswith 131l-B72.3 (12), radioactivity uptake in peritoneal implantswasshown to beconsiderably betterbythe IP route than bythe IV route, as opposed to morevaseubrized tumors.
Concentration was excellent and tumoricidal doses of50-100 Gy were predicted from the biodistribution (13).
In a Phese I dose escalation IPtrial with 131l-B72.3, doses of up to 125 mClwere welltolerated. One patient, with a gastrointestinal primary and peritoneal implents, Is now free ofdisease 8 years aftertreatment with 100 mCl 131l-872.3 IP (A. Raubttsehek, personal communication,
2.2 Antibody Background
24.1 Developmentand specificlty
Monoclonal antibody MX35 is a murine IgG, thatwas developed in the teboretory ofDr. Kenneth O.Lioyd atthe Sloan-Kettering Institute (14).

Fresh ovarian carcinoms specimens, including two solid tumors and two samples of ascites cells, were used as the immunogens. The antibody has beentestedforreactivity on a wide varietyofcryostatsections of normal and malignant human tissues, tissue culture cell lines, and ABO blood group-related antigen preparations, and biochemical
characterization has been performed. In fresh frozen sections, MX35 was detected in epithelial cells ofthe normal bronchus, lung, sweat glands, kidney collecting ducts, thyroid, fallopian tube, cervix and uterus. All other tissues examined, including normel mesothelium (peritoneum) were negative. MX35was detected in sections of420 of480 fresh frozen ovarian carcinomas. In tissue culture, MX35 wss detected on 3 of8 ovarian cancercell lines.
2.2.2 MX35 IgG clinical trial: In a dinical trial ofintact MX35
administered Intravenously or intraperitoneallyto ovarian cancer patients conducted at Memorial Sloan-Kettering (IRB 89-116), twenty-two patients with advanced ovarian cancer received 120l or 131l-labeled MX 35 in doses of2, 10, or 20 mg administered by intravenous (iV) or intraperitoneal (IP) injection. Ail patients
underwentlaperotomy et7 to 20 days following MAb injection to assess tumordistribution, obtain biopsies oftumor and normal tissue, and evaluate the use ofan intraoperative hand held
gamma-detecting device. Following IV injection, serum MAb half lifewas 38 hours. Tumor biopsies obtained at surgery showed
MAb accumulation offrom 3.8x10-3to 4.0x10-4 % injected
dose/gram oftissue. There was no correlation between absolute MAb accumulation in tumorend MAb dose administered. There was a general, though notstatistically significant relationship
between MAb uptake andthe level ofimmunohiatochemical
expression ofthe MX 35 antigen in a given patient'stumor.
Regression analysis showed a correlation between MAb
accumulation andthe interval between MAb Injection end surgery (P = 0.008). Specific localization of MAb in tumorwas
demonstrated by tumor:normel tissue ratios ranging from 2.3:1 to
34:1 (mean 8.0:1). Theturner-normaltissue ratios were not
significantly related to MAb dose, the level ofimmunohistochemical antigen expression, orthe interval between MAb injection and surgery. Duetotherelatively long serum halflife, mean
tumonserum ratioswere only 1.88 following IV injection. This ratio did notcorrelatewith MAb dose, daysfrom injection, or antigen expression, There was an excellent correlation (P.- <0.0001) between MAb uptake, as measured bythe intraoperative hand-held gamma counter, and direct gemma counting ofexcised tissues. No toxicitywasseen inthistrial (15). It is hoped that the use ofMX35 F(Ab')2fragmentswill speed clearance ofcirculating and non- specificallybound antibody and thus result in improved tumorto blood and tumorto normal tissue ratios.
2.2.3 Pre-Clinical MX35 P(Ab')2 Fragment Data in a nude mouse model of human ovarian cancerthe biodistribution of radiolabeled intact MX35 was compared to that of radiolabeled MX35 F(Ab')2 fragments (Kostakoglu, L., end Rubin, S., unpublished data, 1995). Immunodeficient micewere inoculated iP with the human ovarian cancercell line OVCAR-3. Afterestablishment oftumorgrowth the animalswere Injected either IP or IVwith 131-I labeled intact MX36 or F(Ab')2fragments. The nonreactive antibody anti-id L6 was used asa negative controlforthe intact antibody and fragments.
Animalsweresacrificed atvarious time points between 1 and 8 days fordirectgamma counting to determine blood, tumor, and normaltissue levels ofspecific and control antibody. Whole body gamma counting waa performed to assess antibody clearance.
Serum T½ was approximately30 hrs forthe intact antibody and 12 hours forthe F(ab')2 fragments, with no significantdifference by route ofinjection . Specific localization ofintact MX35 to tumorwas demonstrated forboth routes ofinjection, with peaktumorto
normaltissue ratios of 12:1 and 14:1 following IP and IV injection respectively . Peak ratiosforthe F(ab')2fragments were 85 and 33 bythe IP and IV routes respectively. There was no evidence for specific localization ofeithercontrol MAb. Absolute accumulation ofintactantibody Intumorwas in the range of0.5 to 3.0 % of
injected dose/gm, compared to7 to 10% forthe F(Ab'), fragments . Tumorto blood ratios, which never rose above one using intact antibody, reached a range of12to 17 at61 hoursfollowing IV or IP injection ofF(Ab')2fragments. These data suggest that the use of fragmentswillaccelerate clearance ofantibodyfromthe circulation, resulting in increased tumortonormal tissue and tumorto blood ratios.
2.2.4 Clinical trial With MX 35 F(ab')2.
A pre-surgical trial with redioiodineted MX 35 F(ab')2was recently carried out atthis Center(IRB #94-13). Five patients received 2mg and one patient received 10 mg MX 35 F(ab')2 bythe IVroute. Two patients received 2 mg MX 36 F(ab')2 IP. Thestudy revealed thattherewas specific targeting ofradioantibpdyto antigen-positive tumor cells, as determined bycontiguous slice comparison ofhistopethology.
Immunochemlstry, end autoradiography. Most "tumornodules" consisted largely offibrous tissuewith clusters oftumorcellswithin; conventional parameters ofantibodytargeting such as tumorserum ratios were not therefore useful. Besed on this smell sample size, it appears that MX 35 F(ab')2tumoruptake is notdose-dependent, andthattumor uptake is greaterwhenthe antibody is administered IP.
A majorconstraint precluding completion ofthis trial was accrual. Most patientswere unwilling to participate because theywould be unable to receive anytherapeutic antibody,
2.2.5 lodlne-131.
lodine-131 has been used in the therapy ofdifferentiated thyroid
cancerforseveraldecades. The biodistribution and kinetics offree iodide arewell known; uptake of radioactive iodine by thyroid and stomach can be effectively blocked by oral administration ofstable iodide; radioioidine can be stably conjugated to antibody; and experience with systemic
administration ofradioiodinated antibodies in solid tumors, lymphomas
and leukemias is considerable end growing.
Its relatively low (E800.6 MeV) energy beta-minus (β) emissions
may prove optimalforsmall volume disease; its gamma emission (364
KeV)will permitexternalimaging.
2.3 The Role ofintraperitoneal therapy in ovarian cancer
in spite ofthe high responses achieved with pletinum-based therapy in ovarian cancer, the numberofpatients achieving a pathologically documented negative second-look operation remains in the20-25% range. Forthose patients achieving a complete responseto front-line therapy, the use of intraperitoneal chemotherapy is potentially beneficial, particularly in the situation wheretumor is confined to the peritoneal cavity, an extravascularspace, thus allowing the
achievementofhigherdrugconcentrations to reachthe tumorthan achievable by intravenous therapy alone.
3 Summary:
The rationale forthe use of131l-MX35 F(ab'), in ovarian cancer Is based on the localization ofthe antibodystudied byautoradiography in a currentclinical trial (IRB #94-13) and on uniform distribution ofantigen in expressing ovarian tumors. The rationaleforintraperitoneal administration is based ontheobservationthatthe disease is lergely confined tothe peritoneal cavityand istherefore ideally suited for intracavitary therapy. Delivery ofthe radiolabeled antibody intothe peritoneal cavity allows greater concentration In tumors while reducing radiation exposureto normal tissue, perticuiariy bone marrow, lodine-131 isthe preferred radionuelide as it can be attached easilyto the antibody; there la little non-specificbinding; its beta-minus emission characteristics are optimal forradloimmunotherapy; because prior clinicalexperience has shown that the radiolabeled antibodytargetstotumor in vivo; and, given the proposed dosing schedule, outpatienttherapywill be feasible.
Tumorwill be obtained forimmunohistochemicaldetermination ofantigen expression. Only those patientsthat are antigen-positive (> 25% oftumorcells positive in an average high powerfield), willbe eligibleforthe trial, so thattherapeutic benefit may be expected. Patientswillbe treated between 4 and 6 weeks after repeat-look surgerywith the appropriate dose of131l-MX35 F(ab')2 ip. Theywillthen be observed for toxicity, especially bone marrow (videinfra).

4.1 Supply
Monoclonal antibody MX35 in a form satisfactory for human administration will be produced atthe Sloan-Kettering Laboratories underthe supervision of Dr. Lloyd J. Old.
4.2 Production
Hybridoma ascitesis produced in pristane treated, irradiated (BALB/c x
C57BL/6) FI mice. High titerascites batches are pooled for purification. Ascites is tested formurinevirus (MAP, M36V, EdlM. thymicvirus, LDH. MuLV[complete] LCM). Only escites batches negative inthese tests are pooled forpurification.
4.3 Purification
MX36 will be purified from ascites through several steps including removal of lipoprotein by urtracentrifugation at 100,000 g, ammonium sulfate precipitation, and chromatographicfractionation. Purified MX35will be tested forabsence of DNAand characterized by electrophoresis.
4.4 Production ofMX 35F(Ab')2fragments
MX 35 F(Ab')2 fragmentswill be produced In the laboratory ofDr. Kenneth O. Lioyd by pepsin digestion ofintact antibody, which cleaves the IgG molecule et the Fc region, generating an P(Ab')2 fragment ofapproximately 100,000
daltons, and various smallerFcfragments. The protease is subsequently
inactivated and complexed with an anti-pepsin antibody. The mixture is then passed through a proteinAcartridge, which removesthe complexed pepsin end undigested IgG. The F(Ab')2fragments are collected in the flowthrough
fractions, and thesmallerFcfragments (lass then 60,000 daltons) that may co- elute are removed bydialysis using e 80,000 MWcutoffmembrane.
The productischaracterized by SDS-polyacrylamide gel electrophoresis for its biochemical purity. It is tested for residual pepsin byWestern blotting with an anti-pepsin antibody, itisalsotested forthe presence ofblood group A
entigen (a possible contaminantofpepsin)with an anti-aantibody.
4.5 Preciinical SafetyTesting
Purified MX 35 F(AB')2 fragments will undergo safety tasting including the pyrogenicity assayin rabbits, sterllitytesting, end generalsafetytests in animals. Only batches ofMX 35 F(AB')2fragments that satisfy the purity and safety
criteria outlined above (inaccordance with current FDA guidelines) will be used in the proposed study.
4.6 Labeling ofMX35 F(ab')2with 131l
2 mg MX 35 F(ab')2will be labeled with 30 mCi 131l. Sterile technique and pyrogen-firee glasswareare used in all labeling steps. To the requisite amountof MAb F(AB')2, ata pH of7.4, will be added the requisite quantity of131l.50 μl of freshly prepared chlorernine-T (2 μg/μl) will be added and the solution incubated atroom temperature for2minutesafterwhich the reaction will be stopped by addition of50 μl ofsodium metabisulfite (10 μg/μl) and transferred to a 10 ml
Blogal P5 column (50-100 mesh). Fractionswith the highest amount of
radioactivity are pooled and filtered through a .22 μfilter. Immunoreactivity of labeled antibodywill be tested bysequential absorptive assaywith an antigen- expressing cell line.
The exactamountofradioiodinated MAb F(ab')2 to be injected will be 2.0 mg. The amount of131l in the final productwill be measured in a dose calibrator priorto injection. This will enablefuture pharmacokinetic and biodistribution quantitation.
4.6.1 Route ofadministration; Doses will be administered
intraporitoneally, in 100 mL5% human serum albumin, over 30
minutes. Up to 2 liters ofnormal saline will subsequently be
administered intraperitoneallyto aid distribution.
5.1 A Human AntiMouseAntibody response is invariablefollowing
administration ofmouseantibodytopatients with solidtumors. This may
interferewith seralogictests including CA-125. Potential risks to
intraperitoneal radioimmunotherapy (RIT) include myelosuppression. No
dose-limiting ematologlctoxicity orside-effects have been noted in therapytrials with up to75 mCl/m2 IV13'I.
5.2 Potential benefits include increased time to progression.
5.3 Thrombocytopenis has been the dose-limiting toxicitywith other RIT. The patient's plateletcountwillbe monitoredfrequently (videinfra) and platelet transfusions given per hospital guidelines when necessery. Neutropenis
will be managedwith G-CSF administrations pre guidelines. All patient
material and datathatis sentoutsidethe institutionwill be identified bythe hospital MRN number, the patient's identity will otherwise not be
5.4 Sincethe petienthes foiledstandard surgery/chemotherapy, the RIT may be ofbenefit, with otherwise minimal side-effects. Estimation of
therapeuticdoseshould moreoverpermit Phaae II trials in patients, with
potential efficacy.
5.5 There is no standard therapy availablefor patientswho will be eligible to
enterthis study. Otherexperimentaltherapies induct other IP
chemotherapy regimens, aswed as experimental systemic chemotherapy.

5.6 The patientswill be responsibleforstandard tests including CBC,
screening profiles, and CA-126 estimation. The patientwill also be
responsible forphysician charges. The interferon-gamma, radiolabeled
antibodyand allin vitrostudiesconsequentto the therapywill be offered
without charge, unlessthe rediopharmacautical and/orthe IFN-γ is
approved for use forthis indication bythe FDA.
8.1 Patients must heve histologicallyconfirmed ovarian carcinoma.
0.2 All patients with ovarian carcinoma scheduled to undergo surgical re- assessmentafterchemotherapywill beeligible.
8.3 Patients with minimal residual disease after standard surgical debulking and chemotherapy, who are eligible forintraperitoneal chemotherapy, will be eligibleforentry Into the protocol.
6.4 Patients must bewilling to receive further intraperitoneal chemotherapy if eligible.
6.5 Only those patientswhofulfil thefollowing criteria will be eligible forthe
6.5.1 MX35 antigen tumor expression ≥ 25% (ie. atleast 26% ofcells in an average high-powerfield should be reective with MX35 on
6.5.2 Residualdisease aftersurgical assessment ≤ 0.5 cm diameterin a petient deemed to be a good candidate for intraperitoneal therapy.
6.5.3 Uneventful placementofan indwelling intraperitoneal catheter.
8.6 Kamofsky performancestatus of ≥ 60%.
6.7 Patiente must have adequate organ function as defined by:
6.7.1 WBC≥3000Λmm3, platelet count≥100,000/mm3.
6.7.2 Bllirubin≤ 2.0 mg/100 ml .
8.7.3 Creatinine≤2.0 mg/100 ml or creatinine clearance≥ 40 ml/min.
(These tests may be carried outwithin 2 weeks ofstarting therapy.)
6.8 Noevidence ofactive infection which requires antibiotictherapy.
6.9 Signed informed consent
6.10 Patiente must be et least 18 years ofage.
6.1f Patients musthave recovered from the toxicity ofany priortherapy, and
notreceived chemotherapyor radiation therapy foratleast4weeks prior to entry into thetrial,
7.1 Patientswith no evidence ofresidual disease etsurgical assessment are ineligible.
7.2 Patients in whom surgical assessmentwas carried out > 45 days prior to entry into protocol are ineligible.
7.3 Patientswithdense intraperitonealadhesions preventing adequate
intraperitoneal distribution are ineligible.
7.4 MX35 tumorexpression < 25% (ie. <25% ofcells in an average high- powerfield reactivewith MX35 on immunohistoehemistry).
7.5 Concurrent radiotherapy otherthen 131l-MX35 F(ab')2is not permitted.
7.6 Patientswith anysignificantintercurrentmedical problems which will limit the amount of 131I-MX35 F(ab')2 they can tolerate are ineligible forthis
Phase l study.
7.7 Clinicallysignificantcardiacdisease (NewYork HeartAssociation Class
Ill/IV), orsevere debilitating pulmonarydisease.
7.8 Survival expectancyless than 12weeks.
7.0 Priortherapywith amurine monoclonal antibody.
7.10 Patients with a history of autoimmune hepatitie or history ofautoimmune disease.

8.1 Accrual Rate: The expected annual accrual rate is 20 patients.
8.2 131I-MX35F(ab')2 Administration:
8.2.1 2 mg monoclonal antibody MX35 P(ab')2will be radiolabeled with
131l. The mass amount of MX35 F(ab')2will be fixed at 2 mg. The
initial dose of 131l will be 30 mCl. lodine-131 dose will be escalated b repeat infusions. The schemawill be as follows: In orderto estimate kinetics of mAb distribution as
well asfor doslmetricestimates, 10 mCi Tc-99m HSA
will be infused along with the 131l-MX35 F(ab')2. A 2 ml
sample ofperitonealfluid will be obtained 15 minutes.
1 hour and 2 hours after completion ofthe infusion,
to allowdetermination ofmAband ip fluid kinetics. A
2 ml sample of blood wili be collected atthe same
8.2.2 Patients will receive 131l-MX35 F(ab')2intraperitoneally. Infusion will consist of 131l-MX35 F(ab')2dose diluted in 100 ml of5% human
serum albumin delivered as a 30 minute continuous intraperitoneal infusion,followed by900ml ofnormal saline. Up to an additional
one literofnormal saline will subsequently be infused. Thetotal
volumeofintraperitoneal fluid administered will not exceed two (2) liters, and will be recorded.
8.2.3 All administrations of 131l-MX35 F(ab')2will be performed under
clinical observation in theAdultDay Hospital. Acrash cartwith
medicationsforthetreatmentofanaphylaxis will be immediately
8.2.4 Radiation safety precautions will be observed by all personnel as
outlined in the appended guidelines. Patientswillbe kept in the
Adult Day Hospital until cleared by a memberofthe Radiation
Safety Service. We do not expectany patientto require radiation

8.3 Duration ofTreatment
8.3.1 Toxicity: Patients will be observed fortoxicity forat least eight weeks beforedoseescalation is carried out ifno Grade 3 or4 toxicity is observed among the
initialthree patients placed on a dose level, the dose
will beescalated fortne successive group ofthree
patients. Ifone instance ofGrade 3 or Grade 4 toxicity is
observed among the initial three patients placed on a
dose level, three additional patients will be treated at
that level, ifnofurther instances ofGrade 3 or4
toxicity are observed, the dosewill be escalated for
the successive group ofthree patients, If two instances ofGrade 3 or Grade 4toxicity are
observed fora given dose level, three additional
patients wll betreated etthat level. If no further
instances ofGrade 3 or4 toxicity are observed, the
dosewill bedetermined to bethe MTD, and no
furtherdose escalation will be carried out
8.3.2 Retreatment: None. Patients win receive only one course of therapy.
8.3.3 Furthertherapy: Afterrecovery tram toxicity if eny, and not
earlierthan 6 weeks or laterthan 12 weeks afterthe last dose of
131l-MX35 F(ab')2, theywill receive intraperitoneal chemotherapy if eligible.
8.4 CA-125 and KAMA
HAMAmayinterferewith CA-125 measurements. The serum for CA-125 will be batched end atthetime ofmeasurement 10μL of 10μg/ml ofMX35 will be edded totheassayto offset anyfalse elevations oftumormarkerevaluation.

8.5 Imaging
Gamma camera imaging ofall patients will be carried out in the Nuclear Medicineservice. Anteriorend posteriorimages will be obtained the day of the firstinfusion; beforeend aftersubsequentinfusions; daily until all infusions are complete; end three daysafterthe lastinfusion. Images will be designed to estimate quantitativedistribution ofradioactivity in the peritoneal cavity, and to estimate kinetics ofradioactivitytransferinto the systemiccirculation. SPECT imaging of releventareas willbecarried outas deemed necessary. Daily imaging time is not expectedto begreaterthan 2 hours.
8.6 Whole body counting
Whole body countswill be obtained wheneverthe patientis in the Nudear Medicine service. These countswill be obtained using the standard whole body counting crystal (with, ifnecessary, an appropriate lead shield) aswell aswith a radiation survey meter. Theformerwill provide percent retention estimates while the latterwill measure mR/h at 1 meter.

8.1 Progressive diseaseafter a minimum of6 weeks as defined in 14.1.5.
9.2 Intercurrentillness which prevents furtheredministration of 131l-MX35
9.3 Decision ofthe patientto withdraw from the study.
9.4 General orspecificchanges in the pattern's condition which render the
petient unacceptable forfurthertreatment in the judgment ofthe
10.1 Hazards and Protection
10.1.1 Radiation 131l-MX35 F(ab')2:The long-term toxicities of
intraperitoneal radiolabeled MoAbtherapy are not known.
Potantiallyany ofthe chronictoxicities associated with whole
abdominal external beamtnerapy such as bowel or bladder
fibrosis, liverdysfunction or peritonitis with adhesions could
10.1.2 Radiation Precautions; SeeAttached Radiation Safety
Procedure Guidelines. Laboratory specimens will be labeled
with radioactivitywarning labels, andwill be transported to
the Nuclear Medicine Lab by authorized personnel.
10.1.3 MoAb MX35 F(ab')2:There hes been minimal toxicitywith
the clinlcal administrationofmurine moncclonal antibodies
in over400 patients studied atthis institution. Over20
patients have received either intactorfragmented MX35
withoutside-effects. One ofthese patients (who received 10
mg 131l-MX35 F(ab')2 IV) had an idiosyncratic transient
hyperthyroidism,felt possiblyto be due to stable Iodine
toxicity (Jöd-BasedowSyndrome). No such toxicity has been
seen in over500 patientstreated atthis institution with
comparable doaes ofstable iodide.
10.1.4 Managementoftoxicityfrom MX38: Bronchoepasm and
anephytexis are rare complications ofmurine MoAb infusion.
Should such occur, the MoAb infusionwill be immediately
stoppedend epinephrine SC, steroids, respiratory
assistance end otherresuscitative measures undertaken.
Nofurther MoAb will be givento such a patient.
11.1 Known toxioltiee from 131l-MX35 F(ab')2
Forthe purposes ofreporting ADRs, the following are considered to be
known toxicities ofthe agentbeing studied: fever, diarrhea, pancytopenia and peritonitis. Repeated administration of murine MoAbs can result in
allergic reactions inducing serum sickness and anaphylaxis.

11.2 Adverse drug reactions (ADRa) to 131l-MX35 F(ab')2are to be reported
promptly to the institutional Review Board. The following schema is to be followed: Previously unknown Grade2 and Grade 3
reactions are to be reported within 10working days. Grade 4 reactions
and patient deathswhile on treatment are to be reported within 24 hours.
Awritten report is to followwithin 10working days.
12.1 Patientswill receive intraperitoneal chemotherapy after recoveryfrom
toxicity if any ≥ 6 weeks afterthe last dose of131l-MX35 F(ab')2.
12.2 Petients will notbe retreated with 131l-MX35 F(ab')2.
Therapeutic responses are notthe end point in this Phase I study. However, we will evaluate for response during the dose escalation portion ofthe study.
Further, as all patientswill have ≤ 0.5 cm diameterdisease, conventional
imaging studies will have limitedvalue inthe assessment ofextentofdisease and/or response. Serum CA 125 measurements may be used to monitordisease status in patientswith an initial elevated velue.
13.1 Patients WITHOUT bidimensionally measurable disease are evaluable by CA-125. Criteriafor evaluating disease status with CA-125 levels are
drawn from 'Tumourmarkers", by G.J.S Rustin, M.E.L van der Burg, end J.S. Berek.
13.1.1 Complete Response (CR): Normalization ofthe CA-125 for
3 successive evaluations, twoweeks apart.
13.1.2 Partial Response (PR): A50% foil in CA-125 efter 2 samples
confirmed by a fourth sample, twoweeks apart; or a serial
fallof50% overthree samples, two weeks apart; or a serial
fall overthree samplesto lessthan 25% ofthe first sample.
13.1.3 Stabilization (STAB): Patientswho do notmeetthe criteria
forPRorPROG for at least 90 days will be listed in a stable
disease category.
13.1.4 Progression (PROG): A25% risein CA-125 after2 samples
confirmed by afourth sample, twoweeks apart: or a serial
rise of50% overthree samples, two weeks apart; or
elevation above 100 μ/ml forover2 months.
13.1.5 Performance Status: A Kamofsky performance status score
win beassignedwhen evaluating patient. Thiswill allowfor
changes in CA.125 to be correlated with changes in
performance status.
13.2 Duration ofresponse:
Non-measurable patiente: Time fromwhich the CA-125 Is first noted to
have decreased by ≥ 60% forthree successive evaluations until a greater than 60% increase from the nadirvalue is documented on three successive determinations.
13.3 Patients withoutevaluabie disease will be assessed fortoxicity
All patientswill be required to sign a statement ofinformed consent.
This is a Phase I trial design, aimed to find the MTD of131l-MX36 F(ab')2
monocional antibody. The dose escalation scheme is standard. At any doss level an initial 3 patients will be treated. Ifno Grade 3 or4 toxicity is observed, dose escalation willoccur. If 1 or2 toxicities≥ Grade 3 are observed, 3 more patients will be added at the same level. Dosewillbe escalated ifonly 1 ofthe 6 treated has ≥ Grade 3 toxicity. The MTD is the maximum dose atwhich at most 1 out of three, if onlythree were treated atthat level, or2 out of6 show toxicity. The
following givesthe probability ofescalation fora given true probabilityoftoxicity.

Probability ofToxicity 10% 20% 30% 50% 75%
Probability of Escalation 91% 71% 48% 17% 1.8%
Since each dose level takes a maximum of 6 patients, assuming 4 dose levels, this trial would need a minimum of3 patients and a maximum of24 patients.

1. ParkerSL. Tong T. Bolden S. Wlngo PA. Cancer statistics, 1906. CA CancerJ Clin 46:5-27.1995.
2. NIH Consensus Development panel on Ovarian Cancer. Ovarian Cancer,
Screening, Treatment and Follow-up. JAMA 273:6:491-97, 1995
3. Vergote IB, BormerOP, AbelerVM. Evaluation ofCA-125 levels In the
monitoring ofovarian cancer. AmJObstetGynecal157:88-92.1967.
4. Alvarez RD, To A, Boots LR, et al. CA-125 as a serum merkerfor poor prognosis in ovsrian mallgnancies. GynecolOncol26:284-89, 1987.
5. Rustin GJS, Nelstrope A, Stilweil MA, et al. Savings obtained by CA-125
measurements during therapy forovarian caroinoma. EurJ Cancer28:79-92, 1992
6. Epenetos AA, Nimmon CC, Arklle J. Antibody-guided irradiation ofmalignant ascites in ovarian cancer: a newtherapeutic method possessing specificity
against cancer cells. J ObstetGyrr.68:71-74, 1986.
7. Epenetos AA. MunroAJ. StewartSetal. Antibody-guided irradiation of
advanced ovarian cancerwith intraperitoneally administered radiolabeled
monoclonal entibodies. J Clin Oncol; 5:1890-1899, 1987.
8. StewartJSW, HirdV, Snook D, et al. Intraperitoneal Yttrium-90-labeled
monoclonel antibodyin ovarian cancer. JClin Oncot, 8:1941-1950, 1890.
9. Muto MG, Finkler NJ, KassisAl, et al. Intraperitoneal radioimmunotherapy of refractory overian carcinome utilizing iodine-131 labeled monodonal antibody
OC125. Gyn Oncot, 45:265-272, 1992.
10. Crippa F. Radioimmunotherapyofovarian cancer. IntJBiolMarkers; 8:187- 191, 1993.
11. Crippa F. Radioimmunotherapy ofovarian cancer, intJBiolMarkers, 8:192- 197, 1993.
12. Coleher D, Eateban J. Carrasquillo. JA etal. Complementation ofintracavitary and intravenousadministrationofa monodonal antibody (572.3) in patients with careinoma. CancerRes4216-224, 1987.
13. Larson SM, Cerrasquillo JA, Colcher DC etal. Estimates ofradiation absorbed dose for intraperitoncally administered lodine-131 radiolabeled B72.3
monoclonal antibody in patientswith peritoneal cercinometosis. J Nucl Mod; 32; 1881-1667, 1991.
14. Mattes MJ, Look K, Furukawa K, etal. Mouse monodonal antibodiesto human epithelial differentiation antigens expressed on the surface ofovarian carcinoma ascites cells. CancerRes 1987; 8741-60.
15. Rubin S. Koetekogki L, FederidM, etal. Clinicaltrial of pharmacokinetics, biodistribution, and intraoperative radioimmunodetection ofradiolabeled
monoclonal antibody (MAb) MX35 in epithelial ovarian cancer. Gynecol Oncol (abstr) in press, 1995.

In the treatment of advanced epithelial ovarian cancer, the combined effect of surgery and chemotherapy has resulted in a complete response rate of 45% as confirmed by reassessment surgery [1]. However, the risk of recurrence remains high in patients with advanced disease (stages HI and IV) with 50% recurring within a median of 14 months after negative second-look Iaparotomy [2]. Patients with residual disease detected at second-look surgery or recurrent disease after completion of initial chemotherapy have a poor prognosis and few, if any, are cured by currently available salvage therapy. The potential of radiolabeled monoclonal antibodies (mAb) for the detection and quantitation of metastatic spread offers significant benefits for the subsequent management of these patients, as well as the possibility to actually treat micrometastatic disease with antibody carrying the appropriate therapeutic radionuclide, toxin or drug.
The application of radiolabeled antibodies for both radioimmunodiagnosis and treatment of ovarian carcinoma has been ongoing for more than 10 years [3-10]. Epenetos et al. [4] and Pateisky et al. [5] used 131I and 123l labeled antibodies (HMFG1 and HMFG2) against peptide epitopes of human milk fat globulin. Using gamma camera scintigraphy, they successfully demonstrated that >75% of patients having metastatic spread into the peritoneum imaged positively. Negative scans were attributed to the absence of disease or the presence of unresolvable microscopic foci only. The lack of solid tumor nodules >1.5 cm in diameter would render insufficient image contrast to enable specific antibody binding to be detected against a non-specific background [11]. Neither gamma camera imaging nor hand held surgical
radioactivity probes [8, 12] exhibited the sensitivity required to detect micrometastatic disease (<1 cm in diameter), due to insufficient contrast (rarely >10:1) of radionuclide activity accumulation within the tumor relative to the peri-tumor region [11].
Micrometastatic disease may, therefore, remain undetected by conventional nuclear medicine procedures. Moreover, in biodistribution studies using biopsied specimens, the presentation of radiolabeled antibody uptake and dosimetry as an activity per unit gram of tissue can be in significant error. This is because of the small size of the biopsy and the presence of only clusters of tumor cells within a large region of stromal tissue, endothelium and hematopoietic cells. Thus, including non-tumor cells in the activity per unit gram calculations can greatly dilute the tumor specific activity.

In order to explore ways around this problem, we have examined the use of storage phosphor screen technology to determine the distribution of radioactivity in surgical specimens obtained from an ongoing antibody imaging trial on the use of the radiolabeled murine monoclonal antibody (mAb) MX35 F(ab')2 fragment in patients with ovarian carcinoma having minimal residual disease. Digital images from scanned storage phosphor screens were compared with autoradiographic images obtained using film techniques and MX35 antigen localization determined by indirect immunohistochemistry in order to confirm the specific uptake of radiolabeled-mAb MX35 F(ab')2 to tumor cell foci. The data from phosphor digital images were used to evaluate the radionuclide distribution and to estimate accumulation in micrometastatic tumors, adjacent non-tumor tissue and other normal tissue samples. These estimates of tumor specific activity and radiation dose measurements were compared with traditional estimates of the percentage injected dose per gram of tissue (%ID/g) by well scintillation counting.

Patient Selection
Patients in this study had undergone prior surgery for epithelial ovarian cancer and had completed a prescribed course of platinum-based chemotherapy. Eligibility criteria included: known or suspected carcinoma of the ovary, a Karnofsky performance status greater than 60, no prior administration of murine mAb or fragment, and/or a negative human anti-murine antibody (HAMA) titer. Informed consent was obtained from all patients before participation in the study; the study and consent forms were approved by the Institutional Review Board of Memorial Hospital (IRB 94-13). Prior to participating in this trial, either paraffin-embedded tumor specimens or fresh, frozen tumor specimens from an earlier surgery were examined by immunohistochemistry for expression of the MX35 antigen on at least 75% of the carcinoma cells. The tissue specimens from six patients, who have participated in the ongoing monoclonal antibody imaging trial prior to a second-look surgery, are summarized in Table 1.
Preparation of Radiolabeled Monoclonal Antibody MX35 F(ab')2
Monoclonal antibody MX35, a murine IgGl, was generated from the
hybridoma fusion of NS-1 murine myeloma cells with splenocytes from a mouse immunized with 4 fresh ovarian carcinoma specimens [13] and purified as described previously [14]. For fragmentation, purified mAb MX35 IgG (2 mg in 400 μl) was dialysed overnight in 0.1 M citric acid buffer, pH 4.5. Pepsin (25 μl of 1.5 mg/ml;
Sigma Chemical Co., St Louis, MO) was then added to the antibody and digested for 3 hours at 37°C with agitation. F(ab')2 fragments were isolated using an Avid Chrom F(ab')2 Kit (UniSyn Technologies Inc., Tustin, CA). High yield binding buffer (250 μl) containing 30 μl of anti-pepsin was added to the antibody-pepsin combination. The entire sample was diluted with 220 μl of high yield binding buffer and centrifuged to 2000 rpm for 15 minutes and the supernatant was placed on a protein-A-Avid
Chrom cartridge. The unadsorbed fraction was concentrated using a Centricon 30 unit (Amicon, Beverly, MA) at 1075 g at 4°C. The final antibody concentration obtained was 9.3 to 13.6 mg/ml. The identity of the fragments was confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis under reducing
conditions; staining of the gel with Coomassie blue, revealed bands of 23 kD (light chains) and 25 kD (cleaved heavy chains).
The mAb MX35 F(ab')2 fragments were radiolabeled with iodine radionuclides using the chloramine-T method as follows: Two milligrams of antibody fragments were added to 0.5 ml of 0.10 M phosphate buffer, pH 7.4. To the radionuclides, 131l and 125I was added 100 μl of phosphate buffer and this solution was added to the antibody fragment solution. Chloramine-T (2 μg/ml) in phosphate buffer was added and after 2 minutes the reaction was quenched by addition of 50 μl sodium
metabisulfate (10 μg/ml). The protein was separated by passage through a Biogel P6 column (10 ml; BioRad, XXX) using 1% human serum albumin in 0.15 M NaCl as eluant. Terminal sterization was achieved by filtering through a 0.22 micron filter. Immunoreactivity of the labeled product was determined by sequential absorptions with an antigen-expressing cell line (OVCAR-3). Between 50 to 65 percent of the radioactivity was adsorbed using the method described by Mattes et al. [15]. Percent labeled protein was determined by thin layer chromatography and incorporation of radiolabeled iodine into protein were >95%. All procedures were performed aseptically with pyrogen-free material.
Administration of Radiolabeled Monoclonal Antibody MX35 F(ab')2
Beginning at least 24 hours prior to antibody administration and continuing to the time of surgery, patients were treated orally with 10 drops of a saturated solution of potassium iodide three times daily.131I/125I-labeled mAb MX35 F(ab')2 was administered by intravenous (IV) route in a 0.9% sodium chloride solution containing 5% human serum albumin (total volume 100 ml) through a 0.2 micron Millex G-V filter (Millipore, Bedford, MA) over a period of 1 hour. Radiolabeled mAb MX35 F(ab')2 was administrated by intraperitoneal (IP) route as follows: 500 ml of 0.9% sodium chloride was delivered using a catheter or a pre-existing IP port into the peritoneal cavity to facilitate antibody distribution; 100 ml of radiolabeled antibody was added in the same solution as the IV route; and an additional 500 ml of 0.9% sodium chloride was delivered. Five patients were entered at the 2 mg antibody dose labeled with both 131I and 125I. Three patients were injected by IV route and 2 patients by IP route. One patient was entered at the 10 mg antibody dose [2 mg of radiolabeled antibody plus 8 mg of unlabeled mAb MX35 F(ab')2] and injected by IV route.
Blood Samples and Tissue Biopsies
Blood samples were drawn pre-infusion of radiolabeled mAb, post-infusion (1 to 4 hours), immediately before surgery in order to compare radiolabeled antibody levels in the blood with those in the biopsied material, and 4 to 7 days post-surgery. Whole blood was centrifuged at < 2000 rpm for 10 minutes, serum was aspirated, and one ml of pre-surgery serum was weighed and counted in a Packard Cobra well scintillation counter (Packard Instrument Company, Douners Grove, IL).
Multiple biopsied specimens including adjacent normal tissue (fat, muscle and/or peritoneal wall) were retrieved from 6 patients during second-look surgery as summarized in Table 1. Fresh surgical biopsies were divided as follows: One half of each specimen was paraffin-embedded and used for routine histologic evaluation in our Department of Pathology. The other half of each biopsy was weighed and counted in a Packard Cobra well scintillation counter and then snap frozen in liquid nitrogen, embedded in Optimal Cutting Temperature (OCT) compound (Miles Laboratory Inc., Elkhart, IN) and stored at -80°C. A proportion of the frozen surgical biopsies from each case were cut using a motorized cryostat (Bright Instrument Co., Huntingdon, England, UK) and air dried onto microscope slides. Adjacent tissue sections (6 micron thickness) from each biopsy were then analyzed for MX35 antigen localization using an indirect immunoperoxidase procedure and for radionuclide distribution by autoradiography using film and storage phosphor screens.
The number of counts per minute (cpm) were obtained in two windows centered at 25 keV for 125I and 364 keV for 131I. The cpm was converted into activity by measuring 125I and 131I standards alongside the tissue specimens. The %ID/g for each radionuclide was determined for the blood and each tissue biopsy by dividing the specific activity (i.e., μCi/g) by the activity administered to the patient and multiplying by 100.
Immunohistochemical Analysis and MX35 Antigen Localization
Frozen tissue sections were fixed with cold acetone and analyzed for reactivity with mAb MX35 as described previously [16]. Antibody staining patterns were scored in a semiquantitative fashion. Specimens were classified as showing strong (+++) antigen expression when 75% or more of the tumor cells were stained,
heterogeneous (+ to ++) expression when 10 - 75% of the tumor cells stained, and no expression when negative or less than 10% of the tumor stained.
Autoradiography using Film for Distribution of Radionuclide Activity
Autoradiography was performed using Kodak X-OMAT AR film (Eastman Kodak, Rochester, NY). Tissue sections were covered with Saran wrap and overlaid with film alone or with an enhancer screen for exposure durations of between 1 day to 14 days. Films were developed in a Kodak RP X-OMAT processor (Eastman
Kodak). Film images were digitized using the Nikon CoolScan (Nikon, COMPANY) and compared to the digitized 35 mm Ektachrome film images of adjacent tissue sections immunostained with mAb MX35. Co-registration of the serial sections permitted visualization of the radiolabeled antibody activity over regions of tumor cells and non-tumor tissue.

Autoradiography using Storage Phosphor Screens for Distribution of Radionuclide Activity
For the analysis of storage phosphor screen images either a Model GS-250 or Model GS-350 Molecular Imager™ system (Bio-Rad Laboratories Inc., Genetics Systems Division, Hercules, CA) connected to a Macintosh XXX computer system (Apple Computers Inc., Cupertino, CA) was used. The storage phosphor screen (type BI) is fabricated from strontium sulfide commingled with elemental cerium and samarium (SrS; Ce, Sm) [17]. The interaction of ionizing radiation with the storage phosphor screen excites electrons into the conduction band from which they fall into electron traps. Quantitation of the number of filled traps is proportional to the amount of energy deposited in the screen. The detector signal is read out using an externally applied scanning laser (pulsed infrared diode at 910 nm) which scans the screen releasing the electrons from their traps, pixel by pixel. The process of electron de-excitation results in the emission of fluorescent photons, which are collected in a fiber optic pipe and counted using a photomultiplier tube. The resultant signal is processed using an analog to digital converter which provides the final 10 to 16 Mbyte image at 16 bit per pixel. The phosphor image data were analyzed on a Power Macintosh 7100/66 (Apple Computers Inc.) using Molecular Analyst™/Macintosh (Version 2.0) data analysis software (Bio-Rad Laboratories Inc.) and the public domain NIH Image (Version 1.55) program (written by Wayne Rasband, U.S.
National Institutes of Health) which is available from the Internet.
Tissue sections were covered with Saran wrap and then clamped against a storage phosphor screen in a light-tight cassette for a preliminary exposure duration of 24 hours and a second exposure duration of 4 to 16 days. The storage phosphor screen was erased after image read-out by exposure to infrared light for 15 minutes, and erased a second time immediately before re-exposure to the samples.
Calibration of the Storage Phosphor Screen using Radionuclide Standards
A storage phosphor screen image consists of a 2-dimensional array of intensities. The conversion of these intensities into specific activity units requires calibration of the storage phosphor screen response. The screen response was analyzed with three sets of radiolabeled standard sources. Strips containing graded standards of 14C (RPA.504 and RPA.511) and 125I (RPA.523) for autoradiographic calibration were purchased (DuPont, New England Nuclear COMPANY). The commerical 125I standards embedded in polymer were quoted at 20 μm tissue equivalent thickness. A set of 131I standards were made by dilution of a stock solution containing a known activity of 131I as follows: Eosin Y (1% alcoholic solution; Polysciences, Inc. Warrington, PA) was added to the radioactive solution, which was then mixed with the OCT compound until a uniform coloration was achieved. The samples were weighed and counted and the relative specific activities of the two blocks determined to be 0.536 μCi/g and 4.680 μCi/g respectively. The two 131I standards were sectioned at 6 mm thickness and dried onto microscope slides in the same way as the tissue sections prior to exposure to the screen for 24 hours. The response of the phosphor screen was determined to be 1814, 4740, and 830 counts per day (cpd) per pixel for 131I, 125I and 14C, respectively for a source of 1 μCi/g specific activity. The higher sensitivity of the storage phosphor screen means that exposure times are typically between 5 to 10 times shorter than film for the same image quality [18]. The 14C standards were placed alongside the iodine standards to evaluate the constancy of the storage phosphor screen over a long period.
The storage phosphor screen exhibits a slow signal fade during the signal acquisition time. Thus, for each sample exposure time, it was required to convolute the signal accumulation with the signal fade. The fade characteristics were
determined by repeatedly exposing the screen to the standards for the same one hour duration, and varying the interval before reading the screen, from immediate up to 14 days. The correction factor to account for signal fade (F) is given by the

where λF is the rate of signal loss

attributed to fade obtained by fitting the signal counts per day versus the time interval from exposure to read-out and λP is the physical decay constant for the respective radionuclide. Solving the integral, one obtains
F= e-λPto[1 - e-(λFP)to] / (λFP).
The rate constant λF for phosphor screen fade was found to be 0.0967 per day.
Inserting the decay constant λP for 131I, the above equation becomes
F = e-0.0862 to [ 1 - e-0.01051to / 0.0105,
where to is the time in days the phosphor screen is exposed to an 131l-labeled specimen.
Estimation of Dosimetry using Storage Phosphor Screens
The radiation dose is directly proportional to the cumulative specific activity of the radiolabeled antibody in the tumor. The dosimetry estimates presented in this paper were derived from the storage phosphor screen images. The film
autoradiographs were used for high resolution visualization only. We assumed a biological half-life (Tb) in the tumor of 15.5 hours for both IV and IP routes of injection [19]. This estimate was based upon measurements with the same radiolabeled 131I-labeled mAb MX35 F(ab')2 in a murine xenograft tumor model with OVCAR-3 human ovarian cancer cell line. This half-life is reasonable for F(ab')2 antibodies used in the treatment of ovarian carcinoma as evidenced by biological half-lives reported in patient trials after intraperitoneal injection with other radiolabeled mAbs, e.g., Tb = 21 hours for 111In-mAb OC125 [20] and Tb =14 hours for 186Re-mAb NR-CO-02 [21]. The physical half-life (Tp) is 8.04 days for 131I and 60 days for 125l. The biological half-life results in an effective half-life (Te) where Te = Tb· Tp/(Tb + Tp) = 14.3 hours for 131I and 15.3 hours for 125I. In this clinical trial the principal radionuclide dose contribution came from either 125I (Cases 1 and 2) or 131l (Cases 3-6) and the calculations which follow focused on the dosimetry for the appropriate radiolabeled mAb MX35 F(ab')2.
The "peak" specific activity in a biopsy specimen is back extrapolated from measurements of the whole biopsy by well scintillation counting, at 1, 4, or 5 days post infusion of the radiolabeled antibody, using the assumption of a
monoexponential clearance rate which equals Co e-λTe, where λ = 0.693/14.3 hours for 131l and λ = 0.693/15.3 hours for 125l and Co is the specific activity in μCi/g extrapolated back to time t=0.
The radiation absorbed dose resulting from the specific activity is given by the

integral of the specific activity Co ∫ e-λTe over time multiplied by 2.13 ΣniEi · Φi·

This equation consists of the product between the total energy (ΣniEi) released by the radionuclide emission, the fraction of the emission energy absorbed within the tumor (Φ), and 2.13 which is a units conversion factor. For 1311, the sum of the beta ray energy emitted per decay (ΣniEi = 0.187 MeV) multiplied by 2.13 equals 0.398 g-cGy/μCi-hr. For non-penetrating radiations, such as beta particles, it is
recommended by the MLRD committee to use unity for the absorbed fraction (Φ) [22]. However, for a micrometastatic deposit, containing less than a gram of tumor cells, the value of Φ is less than one. The absorbed fractions for several radionuclides including 131l are published by Humm [23] and Goddu, Rao and Howells [24]. For a 100μm micrometastatic lesion, Φ is equal to 0.17, i.e., 17% of the energy emitted within the lesion is deposited locally.

From an ongoing study on the localization of radiolabeled mAb MX35 F(ab')2 in patients with ovarian carcinoma, biopsied specimens were used to compare the distribution of radionuclide as determined by well scintillation counting of whole tissue specimens with storage phosphor screen autoradiography of tissue sections. These specimens were taken during second-look surgery after antibody
administration, 1 day to 5 days earlier. Biopsied specimens from 6 patients were analyzed (Table 1). All tumors were shown to express MX35 antigen as determined by immunohistochemical analysis. In total, 19 normal tissue biopsies and 16 biopsies containing tumor cell foci were studied in the laboratory. Tissue biopsies were counted for 131l and 125i activity in a well scintillation counter and cryostat sections, from these biopsies, were exposed to film and storage phosphor screens as outlined in the Materials and Methods section.
Determination of Specific Activity of Radiolabeled Antibody in Whole Tissue Biopsies using Well Scintillation Counting
The %ID/g was calculated for the whole biopsy specimens and blood sample for each case (Table 2). The %ID/g for biopsies with tumor ranged from 0.5 to 8.7 x 10- 3 (for 131I calculations) and from 0.3 to 6.4 x 10-3 (for 125I calculations) for samples analyzed between 4 to 5 days post surgery. The %ID/g for a single tumor sample studied one day post surgery was 22 x 10"3. The tumor:blood ratios ranged from 0.2:1 to 2.8:1 in the patients receiving antibody by the IV route and 3.7:1 to 5.6:1 by the IP route. The tumor:normal tissue (fat) ratios ranged from 0.9:1 to 39:1. The tumor:normal tissue ratios were greater in the two patients (cases 4 and 5) receiving antibody via the IP route. The percentage of tumor cells within the biopsy specimens was variable, ranging from <10 % to >75% of a sample (Table 1). In 2 of 6 cases (cases 2 and 4), greater than 50% of the biopsy consisted of tumor foci and in these cases the tumor:normal tissue ratios were significant (18:1). In 3 of 6 cases, less than 20% of the biopsy consisted of tumor foci and the tumor:normal tissue ratios were in the range between 0.9:1 to 8.9:1. One specimen (case 5) with <10% tumor cells in the biopsy had the highest tumor:normal tissue ratio of 36:1.
Immunohistochemical Delineation of Areas of Tumor Cells in Tissue Sections and Comparison with Film and Storage Phosphor Screen Autoradiography.
The autoradiographs from film and storage phosphor screens provide an image of the distribution of radionuclide devoid of its relation to the histology of the tissue section. Co-registration of phosphor images and/or digitized film images with the digitized Ektachrome images of tissue sections, stained by immunoperoxidase using mAb MX35 for antigen localization, allowed visual assessment of the efficacy of radiolabeled mAb targeting. Film images of tumor and normal tissue sections were available for 6 cases. Clusters of tumor cells were clearly detected by film visualization in 5 of 6 cases; in the other case single tumor cells and small clusters of <10 tumor cells were weakly detected by film (case 3). Phosphor screen images were analyzed for tumor and normal tissue sections in 5 cases; single tumor cells and small tumor cell islets did not image clearly in case 3.
Analysis of adjacent tissue sections confirmed the coincidence of the radiolabeled mAb MX35 F(ab')2 to regions of MX35-positive tumor cells. Figures 1-3 illustrate the results from three cases comparing film with storage phosphor screen images. Figure 1 illustrates a small laparoscopy specimen (case 2) with >80% of the specimen having strong MX35 antigen expression on the serous type ovarian carcinoma cells at high power (Figure 1a). Co-registration of the tissue section immunostained for antigen localization (Figure 1b) with both the film
autoradiographic image (Figures 1c) and the digital phosphor screen image (Figure 1d) shows variable intensity in the areas of radiolabeled mAb accumulation. Figure 2 illustrates a para aortic lymph node specimen (case 5) with a micrometastatic tumor cell cluster from a poorly differentiated ovarian carcinoma detected by
immunostaining for antigen localization at high power (Figure 2a) and low power (Figure 2b). The film image has saturated at the 7 day exposure (Figure 2c). The phosphor screen image has a greater dynamic range and therefore does not saturate (Figure 2d). The one day film and phosphor screen images showed specific
radionuclide localization to the same clusters of tumor cells as detected by
immunostaining (data not shown). Figure 3 illustrates an endometrioid carcinoma of the peritoneum (case 6). A micrometastatic tumor cell cluster was found near the surface of the ovary (upper panel) and multiple clusters of tumor cells were detected adjacent to the Fallopian tube epithelium (lower panel). The small lesion in the ovary (Figures 3a and 3b) is specifically targeted by radiolabelled mAb MX35 F(ab')2 as seen in the autoradiographic film and phosphor screen images (Figures 3c and 3d). Note that the Fallopian tube epithehum expresses MX35 antigen along the apical surface (Figures 3e and 3f) but the radiolabeled antibody preferentially localizes to the tumor cell clusters (Figures 3g and 3h).
Determination of Specific Activity of Radiolabeled Antibody Within Tissue
Specimens using Storage Phosphor Screens
Line profiles were drawn so as to traverse regions of micrometastatic tumor cell foci which impregnate the regions of normal stromal tissue. Ratios of the "peak" response (converted to cpm or cpd) overlying the MX35-positive tumor deposits relative to the normal stromal tissue background can differ significantly, as
illustrated in Figure 4 from case 4. By correcting the signal accumulation for physical decay and signal fade the "relative" specific activity for regions of interest within the phosphor screen image are obtained. These estimates render an assessment of the local relative specific activity of the radiolabeled antibody in the tumor and adjacent normal tissue within a tissue section. Estimates of the tumor activity was compared with measurements of the average activity per gram for the whole biopsied
specimen from a well scintillation counter (Table 3).
The phosphor screen images were analyzed retrospectively for all patients, after ascertaining the phosphor screen response, i.e., the counts per day per pixel per unit specific activity, as a function of exposure duration from standards. 131I, 125I, and 14C standards were exposed alongside sections of both normal and tumor tissues for patients 5 and 6. This allowed the unknown activity distribution in the biopsied specimen to be determined by direct scaling from the known activity of the standards and the ratio of the radionuclide response. The validity of our approach to patients 1 to 4, who were studied prior to the simultaneous exposure to standards, was verified by estimating the specific activity for patients 5 and 6 using both methods. The ratio between the values obtained by the new method involving simultaneous exposure to standards, to the previous method, by applying the known response and fade characteristics of the phosphor screen, was 0.86 and 0.95 for cases 5 and 6, respectively. Estimation of Dosimetry Within Tissue Specimens using Storage Phosphor Screens

Tables 2 and 3 summarize the dosimetry results for mAb MX35 F(ab')2 in tumor biopsy samples and normal tissues. The %ID/g and range of tumor:normal tissue ratios determined by a calibrated well scintillation counter on the day of biopsy, are presented in Table 2. In Table 3, the %ID/g from Table 2 are compared with estimates of local activity data derived from storage phosphor screens.
Assuming an effective half life of 14.3 hours for 131l and 15.3 hours for 125I, the %ID/g of tissue was back extrapolated to zero time and the cumulative specific activity calculated, i.e., the area under the curve using a monoexponential clearance.

The radiation dose estimates derived from the well scintillation counter measurements for the 4 patients showing good radionuclide localization to MX35-positive tumor cells were 70.1, 10.9, 106.3 and 7.7 cGy/mCi for patients 2, 4, 5 and 6, respectively. These estimates assume a uniform distribution of radionuclide activity within the tumor and do not take into account the complex microscopic distribution of the activity at the cellular level. The dose deposited in small microscopic foci of disease would be substantially reduced below estimates based upon the assumption of charged particle equilibrium [23] as a consequence of the small fraction of energy emanating from the high energy β-particles of 131I, which is deposited locally. The fraction of 131I β-ray energy locally absorbed within a 100μm diameter lesion is 0.17 [24]. This would reduce the dose estimates to only 8.9, 1.85, 18.07 and 1.31 cGy/mCi of 131I injected for patients 2, 4, 5 and 6, respectively. However, the storage phosphor screen autoradiographs show that the activity is not uniformly distributed through the biopsied specimens, but that "hot spots" of activity accumulate at the microscopic tumor lesions. From analysis of the counts per day overlying the tumor regions from the phosphor image data, we observed that specific activity was greater in tumor by a factor of between 4 to 12 times that predicted/estimated from well counter
measurement. This is because the well counter averages the activity per gram over both the tumor and stromal cells of the biopsied specimen. The %ID/g determined by the well counter and storage phosphor screen techniques are summarized in Table 3 alongside the dose estimates to the micrometastatic deposits by both methods. The absorbed doses determined by the storage phosphor screen are mostly greater than those predicted from the well counter method, even after consideration of the reduced absorbed fraction in a microscopic lesion. The phosphor screen method provides information about the activity distribution at a single time point from individual tissue sections and allows estimates to be made of the doses, that can be achieved within microscopic disease per mCi of 131I-mAb MX35 F(ab')2

In this study, we evaluated the radiolabeled antibody uptake of the murine mAb MX35 F(ab')2 in biopsied samples from patients with epithelial ovarian cancer by well scintillation counting and by autoradiography using storage phosphor screens. Specific localization of mAb in tumor was demonstrated by co-registration of the immunohistochemical staining in areas of tumor cell clusters with autoradiographic film and phosphor screen images. In all specimens with micrometastatic spread, radiolabeled mAb uptake showed specific localization to the carcinoma cells (Figures 1-3). Factors, other than antigen distribution, are involved in the localization of radiolabeled antibody in tissues. In this study we noted that antibody localized to tumor cell foci but did not accumulate in the adjacent normal antigen-positive Fallopian tube epithelium (Figure 3) in the tissue sample. In this case accessibility of the antibody to the luminal side of the ducts may be limited.
The radiolabeled antibody uptake (%ID/g) determined by well scintillation counting (1 to 5 days post mAb infusion) ranged between 223.5 to 5.2 x 10-4 %ID/g of tissues for 131I and 210.9 to 2.9 x 10-4 %ID/g for 125I. There was a general relationship between the radiolabeled mAb uptake in the tumor biopsy and both the level and the intensity of the immunohistochemical expression of the MX35 antigen in the corresponding tumor tissue section. Specific localization of mAb in tumor was demonstrated by tumor:normal tissue (fat) ratios ranging from 0.9:1 to 35.9:1 for 131l and from 0.9:1 to 39.0:1 for 125I. Significantly higher tumor:normal tissue ratios were calculated for the two patients given radiolabeled antibody by the IP route (e.g., 17.7:1 and 35.9:1 for 131I). These results can be compared to an earlier clinical trial in which mAb MX35 whole IgG was used [8]. In that study, tumor samples obtained at surgery (7 to 20 days post mAb infusion) showed a mAb accumulation of between 67.0 to 0.3 x 10-4 %ID/g of tissue and the tumor:normal tissue (fat) ratios ranged from 2.3:1 to 34.4:1. The tumor:normal tissue ratios were not significantly related to mAb dose, the level of immunohistochemical antigen expression or the interval between mAb infusion and surgery [8]. Also, in contrast to the present study, tumor:serum ratios rarely exceeded 1.0.
Our analysis provided a measure of the specific activity of biopsied specimens at one single time point; namely, at the time the biopsied specimen was removed and frozen. Because there was no means to determine the specific activity and microdistribution of the radiolabeled antibody before this time, we assumed an effective half-life for the 131I-labeled mAb MX35 F(ab')2 clearance of 14.3 hours based on a parallel murine study performed in our laboratory [19]. Quantitative autoradiography using film is complex with significant limitations, when the activity distribution is non-uniform, due to the limited linear response of film [28]. Therefore, we used storage phosphor screens which have a broader (>4 logs) linear response in this study. Measurements of the specific activity in the tumor, determined by the storage phosphor screen technique, are 4 to 12 times greater than those obtained by well scintillation counter estimates from the biopsied specimens (Table 1). These ratios do not directly translate to the absorbed dose estimates, which must further consider the size of the tumor cell clusters, and correct for the fraction of local energy absorption. Such corrections to absorbed dose are not possible when well counter methods are used since it is assumed that the radiolabel is uniformly distributed throughout the specimen. The storage phosphor screen approach can account for local energy depositions and thereby render accurate dose estimates with which to predict radiotoxicity. If dose limiting toxicity to the bone marrow is reached following a single administration of 100 mCi then the expected doses presented in Table 3 may be multiplied by 100. Thus, doses to the micrometastatic deposits within patients 2, 4, 5, and 6 could be as large as 5200, 2660, 8140 and 1470 cGy, respectively. In vitro studies on the radiosensitivity of ovarian carcinoma cell lines, OVCAR-3 and OVCA-433, have demonstrated that a radiation dose of approximately 1100 cGy is required to reduce the fraction of cell survivors to 0.001, ie., 3 logs of cell kill [28]. Estimates of the localization of radiolabeled mAb MX35 F(ab')2 within
micrometastatic lesions by phosphor screen techniques represent one of the most accurate assessments to date of the specific activity which can be targeted to
microscopic disease.
To improve the radiation dose estimates presented here, it would be necessary to determine the specific activity of the radiolabeled antibody in the tumor at multiple time points. This information is not available for microscopic disease.
Griffith et al. [25] proposed a method to implant thermoluminescent dosimeters, mounted at the tip of a catheter into tissue, in order to directly measure the radiation dose in situ. Yet, this method would not be readily applicable to microscopic disease, due to the uncertainty of the location of the tumor cells. A method using Positron Emission Tomography (PET) to assess quantitatively mAb localization in situ has been developed and tested with 124I-labeled mAb 3F8 in a patient with glioma [26]. High-resolution PET has been used to localize human ovarian cancer in nude rats using 124I-labeled mAb MX35 [27]. The high sensitivity of PET may allow the detection of microscopic disease. However, the resolution of current PET scanners in the abdomen is not better than 4 mm (General Electric Advance PET Scanner, Milwaukee, WI).
When an antibody can be shown convincingly to localize to micrometastatic tumors, epithelial ovarian cancer affords an ideal opportunity to use the antibody or antibody conjugates for therapy. The specific targeting of mAb MX35 F(ab')2 to micrometastatic disease as shown in this study demonstrates the potential of this radiolabeled antibody conjugate for such a therapeutic trial. The ability to target minimum residual disease may be a significant rationale for treating patients with refractory or recurrent ovarian cancer with radioimmunotherapy. Small tumors may be more uniformly accessible to monoclonal antibody and require lower doses of radiation than bulky disease. Ovarian cancer often spreads superfically on the surface of the peritoneum where it forms small tumor foci within the peritoneal cavity. Extraperitoneal metastases, other than spread to local lymph nodes, are rare. Administration of antibody through an IP route, therefore, provides an optimal mode for the treatment of this disease. An additional advantage of IP administration is that, although small tumors may be under vascularized, they will still be accessible to radiolabeled antibody by diffusion from the peritoneal fluid.
This study has demonstrated that mAb MX35, in its F(ab')2 form, avidly localizes to micrometastatic ovarian carcinoma deposits within the peritoneal cavity. Dose estimates to microscopic lesions ranged from 14.7 to 81.4 cGy/mCi injected. Based on conventional radiobiology, these doses per mCi injected would be therapeutic in patients with minimal residual disease.


1. XXX, XXX (Chapter 13). In: Cancer of the Ovary. Editors: Markman M. and Hoskins, W.J. Raven Press, New York, pp. xxx-xxx, 1993.

2. Rubin, S.C, Hoskins, W.J., Hakes, T.B., Markman, M., Cain, J.M. and Lewis, J.L.Jr. Recurrence after negative second-look Iaparotomy for ovarian cancer: Analysis of risk factors. Amer. J. Obstet. Gyn.159:1094-1098, 1988.

3. Granowska, M., Shepard, J., Britton, K.E., Ward, B., Mather, S., Taylor-Papadimitriou, J., Epenetos, A.A., Carroll, M.J., Nimmon, CC, Hawkins, L.A., et al.. Ovarian cancer: Diagnosis using 123I monoclonal antibody in comparison with surgical findings. Nuclear Med. Commun.5:485-499, 1984.

4. Epenetos, A., Shephard, J.H., Britton, K.E.„ Mather, S., Taylor-Papadimitriou, J., Granowska, M., Durbin, H., Nimmon, CC, Hawkins, L.R., Malpas, J.S. and
Bodmer, W.F. 123I radioiodinated antibody imaging of occult ovarian cancer.
Cancer 55:984-987, 1985.

5. Pateisky, N., Philipp, K., Skodler, W.D., Czerwenka, K., Hamilton, G. and Burchell, J. Radioimmunodetection in patients with suspected ovarian cancer. J. Nucl. Med.26:1369-1376, 1985.

6. Chatal, J.F., Fumoleau, P., Saccavini, J.C, Thedrez, P., Curtet, C, Biano-Arco, A., Chetanneau, A., Peltier, P., Kremer, M. and Guillard, Y. Immunoscintigraphy of recurrences of gynecologic carcinomas. J. Nucl. Med.28:1807-1819, 1987.

7. Stewart, J.S.W., Hird, V., Snook, D., Sullivan, M., Hooker, G., Coutenay-Luck, N., Sivolapenko, G., Griffiths., M., Myers, M.J., Lambert, H.E., Munro, A.J. and
Epenetos, A.A. Intraperitoneal radioimmunotherapy for ovarian cancer:
Pharmacokinetics, toxicity, and efficacy of I-131-labeled monoclonal antibodies. Int. J. Radiation Oncology Biol. Phys.16:405-413, 1989.

8. Rubin, S.C, Kostakoglu, L., Divgi, C, Federici, M.G., Finstad, C.L., Lloyd, K.O., Larson, S.M. and Hoskins, W.J. Biodistribution and intraoperative evaluation of radiolabeled monoclonal antibody MX35 in patients with epithelial ovarian cancer.

Gynecol. Oncol.51:61-66, 1993

9. Buist, M.R., Kenemans, P., Molthoff, C.J.M., Roos, J.C, Den Hollander, W., Brinkhuis, M. and Baak, J.P.A. Tumor uptake of intravenously administered radiolabeled antibodies in ovarian carcinoma patients in relation to antigen expression and other tumor characteristics. Int. J. Cancer 64:92-98, 1995.

10. Muto, M.G. and Kassis, A.I. Monoclonal antibodies used in the detection and treatment of epithelial ovarian cancer. Cancer 76:2016-2027, 1995.

11. Bradwell, A.R., Fairweather, D.S., Dykes, P.W., Keeling, A., Vaughan, A. and Taylor. J. Limiting factors in the localization of tumours with radiolabelled
antibodies. Immunology Today 6:163-170, 1985.

12. Daghighian, F., Mazziotta, J.C, Hoffman, E.J., Shenderov, P., Eshaghian, B., Siegel, S. and Phelps, M.E. Intraoperative beta probe: a device for detecting tissue labeled with positron or electron emitting isotopes during surgery. Med. Phys.21:153-157, 1994.

13. Mattes, M.J., Look, K., Furukawa, K., Pierce, V.K., Old L.J., Lewis, J.LJr. and Lloyd, K.O. Mouse monoclonal antibodies to human epithelial differentiation antigen expressed on the surface of ovarian carcinoma ascites cells. Cancer Res.
47:6741-6750, 1987.

14-15. Mattes, M.J., Lloyd, K.O. and Lewis, J.L.Jr. Binding parameters of monoclonal antibodies reacting with ovarian carcinoma ascites cells. Cancer Immunol.
Immunother.28:199-207, 1989.

16. Rubin, S.C, Finstad, C.L., Hoskins, W.J., Federici, M.G., Lloyd, K.O. and Lewis, J.L.Jr. A longitudinal study of antigen expression in epithelial ovarian cancer.
Gynecol. Oncol.34:389-394, 1989.

17. Nguyen, Q. and Heffelfinger, D.M. Imaging and quantitation of chemiluminescence using photoexcitable storage phosphor screen. Anal. Biochem.226:59-67, 1995.

18. Humm, J.L., Chin, L.M., Lanza, R.C., Schaeffer, CM., Speidel, M. and Greene, R.E. Digital Imaging for Autoradiography. Chapter 14. In: Computer Digital
Radiography in Clinical Practice. Editors: Greene, R.E. and Oestmann, J.W. Thieme Medical Publishers, Inc., New York, pp.167-172, 1992.

19. Kostakoglu, L., Rubin, S.C, Federici, M., Finstad, C.L., Larson, S.M., Hoskins, W.J. and Lloyd, K.O. A comparative study of intact and F(ab')2 fragments of monoclonal antibody MX35 in a xenograft model. J. Nucl. Med.34:216P, 1993.

20. Hnatowich, D.J., Gionet, M., Rusckowski, M., Siebecker, D.A., Roche, J., Shealy, D., Mattis, J.A., Wilson, J., McGann, J., Hunter, R.E., Griffin, T. and Doherty, P.W. Pharmacokinetics of 111In-labeled OC-125 antibody in cancer patients compared with the 19-9 antibody. Cancer Res.47:6111-6117, 1987.

21. Breitz, H.B., Weiden, P.L., Vanderheyden, J.L., Appelbaum, J.W., Bjorn, M.J., Fer, M.F., Wolf, S.B., Ratliff, B.A., Seiler, C.A., Foisie, D.C, Fisher, D.R., Schroff, R.W., Fritzberg, A.R. and Abrams, P.G. Clinical experience with rhenium-186-labeled monoclonal antibodies for radioimmunotherapy: Results of phase I trials. J. Nucl. Med.33:1099-1112, 1992.

22. Loevinger, R., Budinger, T.F. and Watson, E.E. MIRD Primer for Absorbed Dose Calculations. The Society of Nuclear Medicine, Inc., New York, 1991.

23. Humm, J.L. Dosimetric aspects of radiolabelled antibodies for tumor therapy. J. Nucl. Med.27:1490-1497, 1986.

24. Goddu, S.M., Rao, D.V. and Howells, R.W. Multicellular dosimetry for micrometastases. Dependence of self-dose versus cross-dose to cell nuclei on type and energy of radiation and subcellular distribution of radionuclides. J. Nucl. Med.
25:521-530, 1994.

25. Griffith, M.H., Yorke, E.D., Wessels, B.W., De Nardo, G.L. and Neacy, W.P. Direct dose confirmation of quantitative autoradiography with micro-TLD
measurements for radioimmunotherapy. J. Nucl. Med.29:1795-1809, 1988.

26. Daghighian, F., Pentlow, K.S., Larson, S.M., Graham, M.C, DiResta, G.R., Yeh, S.D., Macapinlac, H., Finn, R.D., Arbit, E. and Cheung, N.K. Development of a method to measure kinetics of radiolabelled monoclonal antibody in human tumour with applications to microdosimetry: Positron emission tomography studies of iodine-124 labelled 3F8 monoclonal antibody in glioma. Eur. J. Nucl. Med.20:402-409, 1993.

27. Rubin, S.C, Kairemo, K.J.A., Brownell, A.L., Daghighian, F., Federici, M.G., Pentlow, K.S., Finn, R.D., Lambrecht, R.M., Hoskins, W.J., Lewis, J.L.Jr. and Larson, S.M. High-resolution positron emission tomography of human ovarian cancer in nude rats using 124I-labeled monoclonal antibodies. Gynecol. Oncol.48:61-67, 1993.

28. Humm, J.L., Macklis, R.M., Lu, X.Q., Yang, Y., Bump, K., Beresford, B. and Chin, L.M. Spacial accuracy and cellular dose estimates obtained from 3D reconstructed serial tissue autoradiographs. Phys. Med. Biol.40:163-180, 1995.

29. Schneider-Gadicke, E., Humm, J.L., Lau, CC, Macklis, R.M., Bastert, G. and Knapp, R.C. Analysis of cytotoxicity of 131I-labeled OC125 F(ab')2 on human
epithelial ovarian cancer cell lines. Radiother. Oncol.22:150-159, 1992.


1. Tumor biopsy from patient 2 with a serous ovarian carcinoma following administration of radiolabeled mAb MX35 F(ab')2 (5 days earlier) by intravenous route. Indirect immunoperoxidase staining with mAb MX35 at high power (A) and low power (B). Adjacent tumor sections of digitized film image at low power (C) and storage phosphor screen image at low power (D). Note the strong immunoperoxidase staining of tumor cell clusters and co-localization of radiolabeled mAb to the tumor.

2. Para aortic lymph node from patient 5 with a poorly differentiated ovarian carcinoma following administration of radiolabeled mAb MX35 F(ab')2 (5 days earlier) by intraperitoneal route. Indirect immunoperoxidase staining with mAb MX35 at high power (A) and low power (B). Adjacent tumor sections of digitized film image (saturated) at low power (C) and storage phosphor screen image at low power (D). Note the distribution of immunostained tumor cells between the hematopoietic cells and stromal tissue and localization of radiolabeled mAb to the tumor cell area specifically.

3. Tumor cell clusters in ovary (upper panel) and Fallopian tube (lower panel) from patient 6 with an endometrioid carcinoma of the peritoneum following administration of radiolabeled mAb MX35 F(ab')2 (4 days earlier) by intravenous route. Indirect iinmunoperoxidase staining with mAb MX35 at high power (A, E) and low power (B, F). Adjacent tumor sections of digitized film images at low power (C, G) and storage phosphor screen images at low power (D, H). Note that both the
Fallopian tube epithelium and tumor cell clusters show strong immunoperoxidase staining, however, the radiolabeled mAb localizes to the tumor cell clusters on the film and phosphor screen images and not the adjacent normal tube epithelium.

4. Determination of specific activity of radionuclide within tumor specimens from patient 4 with an endometrioid ovarian carcinoma following administration of radiolabeled mAb MX35 F(ab')2 (1 day earlier) by intraperitoneal route. Three line profiles were drawn to traverse regions of micrometastatic tumor foci which impregnate the regions of normal stromal tissue (left panels). The location of the line profiles in the tumor biopsy are illustrated (right panel).